]> 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
net: vrf: Flip IPv6 output path from FIB lookup hook to out hook
[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
ED		 104 const struct in6_addr *prefix, int prefixlen,
105 const struct in6_addr *gwaddr, int ifindex,
95c96174 106 unsigned int pref);
efa2cea0 107static struct rt6_info *rt6_get_route_info(struct net *net,
b71d1d42
ED		 108 const struct in6_addr *prefix, int prefixlen,
109 const struct in6_addr *gwaddr, int ifindex);
70ceb4f5
YH		 110#endif
111
8d0b94af
MKL		 112struct uncached_list {
113 spinlock_t lock;
114 struct list_head head;
115};
116
117static DEFINE_PER_CPU_ALIGNED(struct uncached_list, rt6_uncached_list);
118
119static void rt6_uncached_list_add(struct rt6_info *rt)
120{
121 struct uncached_list *ul = raw_cpu_ptr(&rt6_uncached_list);
122
123 rt->dst.flags |= DST_NOCACHE;
124 rt->rt6i_uncached_list = ul;
125
126 spin_lock_bh(&ul->lock);
127 list_add_tail(&rt->rt6i_uncached, &ul->head);
128 spin_unlock_bh(&ul->lock);
129}
130
131static void rt6_uncached_list_del(struct rt6_info *rt)
132{
133 if (!list_empty(&rt->rt6i_uncached)) {
134 struct uncached_list *ul = rt->rt6i_uncached_list;
135
136 spin_lock_bh(&ul->lock);
137 list_del(&rt->rt6i_uncached);
138 spin_unlock_bh(&ul->lock);
139 }
140}
141
142static void rt6_uncached_list_flush_dev(struct net *net, struct net_device *dev)
143{
144 struct net_device *loopback_dev = net->loopback_dev;
145 int cpu;
146
e332bc67
EB		 147 if (dev == loopback_dev)
148 return;
149
8d0b94af
MKL		 150 for_each_possible_cpu(cpu) {
151 struct uncached_list *ul = per_cpu_ptr(&rt6_uncached_list, cpu);
152 struct rt6_info *rt;
153
154 spin_lock_bh(&ul->lock);
155 list_for_each_entry(rt, &ul->head, rt6i_uncached) {
156 struct inet6_dev *rt_idev = rt->rt6i_idev;
157 struct net_device *rt_dev = rt->dst.dev;
158
e332bc67 159 if (rt_idev->dev == dev) {
8d0b94af
MKL		 160 rt->rt6i_idev = in6_dev_get(loopback_dev);
161 in6_dev_put(rt_idev);
162 }
163
e332bc67 164 if (rt_dev == dev) {
8d0b94af
MKL		 165 rt->dst.dev = loopback_dev;
166 dev_hold(rt->dst.dev);
167 dev_put(rt_dev);
168 }
169 }
170 spin_unlock_bh(&ul->lock);
171 }
172}
173
d52d3997
MKL		 174static u32 *rt6_pcpu_cow_metrics(struct rt6_info *rt)
175{
176 return dst_metrics_write_ptr(rt->dst.from);
177}
178
06582540
DM		 179static u32 *ipv6_cow_metrics(struct dst_entry *dst, unsigned long old)
180{
4b32b5ad 181 struct rt6_info *rt = (struct rt6_info *)dst;
06582540 182
d52d3997
MKL		 183 if (rt->rt6i_flags & RTF_PCPU)
184 return rt6_pcpu_cow_metrics(rt);
185 else if (rt->rt6i_flags & RTF_CACHE)
4b32b5ad
MKL		 186 return NULL;
187 else
3b471175 188 return dst_cow_metrics_generic(dst, old);
06582540
DM		 189}
190
f894cbf8
DM		 191static inline const void *choose_neigh_daddr(struct rt6_info *rt,
192 struct sk_buff *skb,
193 const void *daddr)
39232973
DM		 194{
195 struct in6_addr *p = &rt->rt6i_gateway;
196
a7563f34 197 if (!ipv6_addr_any(p))
39232973 198 return (const void *) p;
f894cbf8
DM		 199 else if (skb)
200 return &ipv6_hdr(skb)->daddr;
39232973
DM		 201 return daddr;
202}
203
f894cbf8
DM		 204static struct neighbour *ip6_neigh_lookup(const struct dst_entry *dst,
205 struct sk_buff *skb,
206 const void *daddr)
d3aaeb38 207{
39232973
DM		 208 struct rt6_info *rt = (struct rt6_info *) dst;
209 struct neighbour *n;
210
f894cbf8 211 daddr = choose_neigh_daddr(rt, skb, daddr);
8e022ee6 212 n = __ipv6_neigh_lookup(dst->dev, daddr);
f83c7790
DM		 213 if (n)
214 return n;
215 return neigh_create(&nd_tbl, daddr, dst->dev);
216}
217
9a7ec3a9 218static struct dst_ops ip6_dst_ops_template = {
1da177e4 219 .family = AF_INET6,
1da177e4
LT		 220 .gc = ip6_dst_gc,
221 .gc_thresh = 1024,
222 .check = ip6_dst_check,
0dbaee3b 223 .default_advmss = ip6_default_advmss,
ebb762f2 224 .mtu = ip6_mtu,
06582540 225 .cow_metrics = ipv6_cow_metrics,
1da177e4
LT		 226 .destroy = ip6_dst_destroy,
227 .ifdown = ip6_dst_ifdown,
228 .negative_advice = ip6_negative_advice,
229 .link_failure = ip6_link_failure,
230 .update_pmtu = ip6_rt_update_pmtu,
6e157b6a 231 .redirect = rt6_do_redirect,
9f8955cc 232 .local_out = __ip6_local_out,
d3aaeb38 233 .neigh_lookup = ip6_neigh_lookup,
1da177e4
LT		 234};
235
ebb762f2 236static unsigned int ip6_blackhole_mtu(const struct dst_entry *dst)
ec831ea7 237{
618f9bc7
SK		 238 unsigned int mtu = dst_metric_raw(dst, RTAX_MTU);
239
240 return mtu ? : dst->dev->mtu;
ec831ea7
RD		 241}
242
6700c270
DM		 243static void ip6_rt_blackhole_update_pmtu(struct dst_entry *dst, struct sock *sk,
244 struct sk_buff *skb, u32 mtu)
14e50e57
DM		 245{
246}
247
6700c270
DM		 248static void ip6_rt_blackhole_redirect(struct dst_entry *dst, struct sock *sk,
249 struct sk_buff *skb)
b587ee3b
DM		 250{
251}
252
14e50e57
DM		 253static struct dst_ops ip6_dst_blackhole_ops = {
254 .family = AF_INET6,
14e50e57
DM		 255 .destroy = ip6_dst_destroy,
256 .check = ip6_dst_check,
ebb762f2 257 .mtu = ip6_blackhole_mtu,
214f45c9 258 .default_advmss = ip6_default_advmss,
14e50e57 259 .update_pmtu = ip6_rt_blackhole_update_pmtu,
b587ee3b 260 .redirect = ip6_rt_blackhole_redirect,
0a1f5962 261 .cow_metrics = dst_cow_metrics_generic,
d3aaeb38 262 .neigh_lookup = ip6_neigh_lookup,
14e50e57
DM		 263};
264
62fa8a84 265static const u32 ip6_template_metrics[RTAX_MAX] = {
14edd87d 266 [RTAX_HOPLIMIT - 1] = 0,
62fa8a84
DM		 267};
268
fb0af4c7 269static const struct rt6_info ip6_null_entry_template = {
d8d1f30b
CG		 270 .dst = {
271 .__refcnt = ATOMIC_INIT(1),
272 .__use = 1,
2c20cbd7 273 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 274 .error = -ENETUNREACH,
d8d1f30b
CG		 275 .input = ip6_pkt_discard,
276 .output = ip6_pkt_discard_out,
1da177e4
LT		 277 },
278 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 279 .rt6i_protocol = RTPROT_KERNEL,
1da177e4
LT		 280 .rt6i_metric = ~(u32) 0,
281 .rt6i_ref = ATOMIC_INIT(1),
282};
283
101367c2
TG		 284#ifdef CONFIG_IPV6_MULTIPLE_TABLES
285
fb0af4c7 286static const struct rt6_info ip6_prohibit_entry_template = {
d8d1f30b
CG		 287 .dst = {
288 .__refcnt = ATOMIC_INIT(1),
289 .__use = 1,
2c20cbd7 290 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 291 .error = -EACCES,
d8d1f30b
CG		 292 .input = ip6_pkt_prohibit,
293 .output = ip6_pkt_prohibit_out,
101367c2
TG		 294 },
295 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 296 .rt6i_protocol = RTPROT_KERNEL,
101367c2
TG		 297 .rt6i_metric = ~(u32) 0,
298 .rt6i_ref = ATOMIC_INIT(1),
299};
300
fb0af4c7 301static const struct rt6_info ip6_blk_hole_entry_template = {
d8d1f30b
CG		 302 .dst = {
303 .__refcnt = ATOMIC_INIT(1),
304 .__use = 1,
2c20cbd7 305 .obsolete = DST_OBSOLETE_FORCE_CHK,
d8d1f30b 306 .error = -EINVAL,
d8d1f30b 307 .input = dst_discard,
ede2059d 308 .output = dst_discard_out,
101367c2
TG		 309 },
310 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
4f724279 311 .rt6i_protocol = RTPROT_KERNEL,
101367c2
TG		 312 .rt6i_metric = ~(u32) 0,
313 .rt6i_ref = ATOMIC_INIT(1),
314};
315
316#endif
317
ebfa45f0
MKL		 318static void rt6_info_init(struct rt6_info *rt)
319{
320 struct dst_entry *dst = &rt->dst;
321
322 memset(dst + 1, 0, sizeof(*rt) - sizeof(*dst));
323 INIT_LIST_HEAD(&rt->rt6i_siblings);
324 INIT_LIST_HEAD(&rt->rt6i_uncached);
325}
326
1da177e4 327/* allocate dst with ip6_dst_ops */
d52d3997
MKL		 328static struct rt6_info *__ip6_dst_alloc(struct net *net,
329 struct net_device *dev,
ad706862 330 int flags)
1da177e4 331{
97bab73f 332 struct rt6_info *rt = dst_alloc(&net->ipv6.ip6_dst_ops, dev,
6f3118b5 333 0, DST_OBSOLETE_FORCE_CHK, flags);
cf911662 334
ebfa45f0
MKL		 335 if (rt)
336 rt6_info_init(rt);
8104891b 337
cf911662 338 return rt;
1da177e4
LT		 339}
340
9ab179d8
DA		 341struct rt6_info *ip6_dst_alloc(struct net *net,
342 struct net_device *dev,
343 int flags)
d52d3997 344{
ad706862 345 struct rt6_info *rt = __ip6_dst_alloc(net, dev, flags);
d52d3997
MKL		 346
347 if (rt) {
348 rt->rt6i_pcpu = alloc_percpu_gfp(struct rt6_info *, GFP_ATOMIC);
349 if (rt->rt6i_pcpu) {
350 int cpu;
351
352 for_each_possible_cpu(cpu) {
353 struct rt6_info **p;
354
355 p = per_cpu_ptr(rt->rt6i_pcpu, cpu);
356 /* no one shares rt */
357 *p = NULL;
358 }
359 } else {
360 dst_destroy((struct dst_entry *)rt);
361 return NULL;
362 }
363 }
364
365 return rt;
366}
9ab179d8 367EXPORT_SYMBOL(ip6_dst_alloc);
d52d3997 368
1da177e4
LT		 369static void ip6_dst_destroy(struct dst_entry *dst)
370{
371 struct rt6_info *rt = (struct rt6_info *)dst;
ecd98837 372 struct dst_entry *from = dst->from;
8d0b94af 373 struct inet6_dev *idev;
1da177e4 374
4b32b5ad 375 dst_destroy_metrics_generic(dst);
87775312 376 free_percpu(rt->rt6i_pcpu);
8d0b94af
MKL		 377 rt6_uncached_list_del(rt);
378
379 idev = rt->rt6i_idev;
38308473 380 if (idev) {
1da177e4
LT		 381 rt->rt6i_idev = NULL;
382 in6_dev_put(idev);
1ab1457c 383 }
1716a961 384
ecd98837
YH		 385 dst->from = NULL;
386 dst_release(from);
b3419363
DM		 387}
388
1da177e4
LT		 389static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
390 int how)
391{
392 struct rt6_info *rt = (struct rt6_info *)dst;
393 struct inet6_dev *idev = rt->rt6i_idev;
5a3e55d6 394 struct net_device *loopback_dev =
c346dca1 395 dev_net(dev)->loopback_dev;
1da177e4 396
97cac082
DM		 397 if (dev != loopback_dev) {
398 if (idev && idev->dev == dev) {
399 struct inet6_dev *loopback_idev =
400 in6_dev_get(loopback_dev);
401 if (loopback_idev) {
402 rt->rt6i_idev = loopback_idev;
403 in6_dev_put(idev);
404 }
405 }
1da177e4
LT		 406 }
407}
408
5973fb1e
MKL		 409static bool __rt6_check_expired(const struct rt6_info *rt)
410{
411 if (rt->rt6i_flags & RTF_EXPIRES)
412 return time_after(jiffies, rt->dst.expires);
413 else
414 return false;
415}
416
a50feda5 417static bool rt6_check_expired(const struct rt6_info *rt)
1da177e4 418{
1716a961
G		 419 if (rt->rt6i_flags & RTF_EXPIRES) {
420 if (time_after(jiffies, rt->dst.expires))
a50feda5 421 return true;
1716a961 422 } else if (rt->dst.from) {
3fd91fb3 423 return rt6_check_expired((struct rt6_info *) rt->dst.from);
1716a961 424 }
a50feda5 425 return false;
1da177e4
LT		 426}
427
51ebd318
ND		 428/* Multipath route selection:
429 * Hash based function using packet header and flowlabel.
430 * Adapted from fib_info_hashfn()
431 */
432static int rt6_info_hash_nhsfn(unsigned int candidate_count,
433 const struct flowi6 *fl6)
434{
644d0e65 435 return get_hash_from_flowi6(fl6) % candidate_count;
51ebd318
ND		 436}
437
438static struct rt6_info *rt6_multipath_select(struct rt6_info *match,
52bd4c0c
ND		 439 struct flowi6 *fl6, int oif,
440 int strict)
51ebd318
ND		 441{
442 struct rt6_info *sibling, *next_sibling;
443 int route_choosen;
444
445 route_choosen = rt6_info_hash_nhsfn(match->rt6i_nsiblings + 1, fl6);
446 /* Don't change the route, if route_choosen == 0
447 * (siblings does not include ourself)
448 */
449 if (route_choosen)
450 list_for_each_entry_safe(sibling, next_sibling,
451 &match->rt6i_siblings, rt6i_siblings) {
452 route_choosen--;
453 if (route_choosen == 0) {
52bd4c0c
ND		 454 if (rt6_score_route(sibling, oif, strict) < 0)
455 break;
51ebd318
ND		 456 match = sibling;
457 break;
458 }
459 }
460 return match;
461}
462
1da177e4 463/*
c71099ac 464 * Route lookup. Any table->tb6_lock is implied.
1da177e4
LT		 465 */
466
8ed67789
DL		 467static inline struct rt6_info *rt6_device_match(struct net *net,
468 struct rt6_info *rt,
b71d1d42 469 const struct in6_addr *saddr,
1da177e4 470 int oif,
d420895e 471 int flags)
1da177e4
LT		 472{
473 struct rt6_info *local = NULL;
474 struct rt6_info *sprt;
475
dd3abc4e
YH		 476 if (!oif && ipv6_addr_any(saddr))
477 goto out;
478
d8d1f30b 479 for (sprt = rt; sprt; sprt = sprt->dst.rt6_next) {
d1918542 480 struct net_device *dev = sprt->dst.dev;
dd3abc4e
YH		 481
482 if (oif) {
1da177e4
LT		 483 if (dev->ifindex == oif)
484 return sprt;
485 if (dev->flags & IFF_LOOPBACK) {
38308473 486 if (!sprt->rt6i_idev ||
1da177e4 487 sprt->rt6i_idev->dev->ifindex != oif) {
17fb0b2b 488 if (flags & RT6_LOOKUP_F_IFACE)
1da177e4 489 continue;
17fb0b2b
DA		 490 if (local &&
491 local->rt6i_idev->dev->ifindex == oif)
1da177e4
LT		 492 continue;
493 }
494 local = sprt;
495 }
dd3abc4e
YH		 496 } else {
497 if (ipv6_chk_addr(net, saddr, dev,
498 flags & RT6_LOOKUP_F_IFACE))
499 return sprt;
1da177e4 500 }
dd3abc4e 501 }
1da177e4 502
dd3abc4e 503 if (oif) {
1da177e4
LT		 504 if (local)
505 return local;
506
d420895e 507 if (flags & RT6_LOOKUP_F_IFACE)
8ed67789 508 return net->ipv6.ip6_null_entry;
1da177e4 509 }
dd3abc4e 510out:
1da177e4
LT		 511 return rt;
512}
513
27097255 514#ifdef CONFIG_IPV6_ROUTER_PREF
c2f17e82
HFS		 515struct __rt6_probe_work {
516 struct work_struct work;
517 struct in6_addr target;
518 struct net_device *dev;
519};
520
521static void rt6_probe_deferred(struct work_struct *w)
522{
523 struct in6_addr mcaddr;
524 struct __rt6_probe_work *work =
525 container_of(w, struct __rt6_probe_work, work);
526
527 addrconf_addr_solict_mult(&work->target, &mcaddr);
304d888b 528 ndisc_send_ns(work->dev, &work->target, &mcaddr, NULL);
c2f17e82 529 dev_put(work->dev);
662f5533 530 kfree(work);
c2f17e82
HFS		 531}
532
27097255
YH		 533static void rt6_probe(struct rt6_info *rt)
534{
990edb42 535 struct __rt6_probe_work *work;
f2c31e32 536 struct neighbour *neigh;
27097255
YH		 537 /*
538 * Okay, this does not seem to be appropriate
539 * for now, however, we need to check if it
540 * is really so; aka Router Reachability Probing.
541 *
542 * Router Reachability Probe MUST be rate-limited
543 * to no more than one per minute.
544 */
2152caea 545 if (!rt || !(rt->rt6i_flags & RTF_GATEWAY))
7ff74a59 546 return;
2152caea
YH		 547 rcu_read_lock_bh();
548 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
549 if (neigh) {
8d6c31bf
MKL		 550 if (neigh->nud_state & NUD_VALID)
551 goto out;
552
990edb42 553 work = NULL;
2152caea 554 write_lock(&neigh->lock);
990edb42
MKL		 555 if (!(neigh->nud_state & NUD_VALID) &&
556 time_after(jiffies,
557 neigh->updated +
558 rt->rt6i_idev->cnf.rtr_probe_interval)) {
559 work = kmalloc(sizeof(*work), GFP_ATOMIC);
560 if (work)
561 __neigh_set_probe_once(neigh);
c2f17e82 562 }
2152caea 563 write_unlock(&neigh->lock);
990edb42
MKL		 564 } else {
565 work = kmalloc(sizeof(*work), GFP_ATOMIC);
f2c31e32 566 }
990edb42
MKL		 567
568 if (work) {
569 INIT_WORK(&work->work, rt6_probe_deferred);
570 work->target = rt->rt6i_gateway;
571 dev_hold(rt->dst.dev);
572 work->dev = rt->dst.dev;
573 schedule_work(&work->work);
574 }
575
8d6c31bf 576out:
2152caea 577 rcu_read_unlock_bh();
27097255
YH		 578}
579#else
580static inline void rt6_probe(struct rt6_info *rt)
581{
27097255
YH		 582}
583#endif
584
1da177e4 585/*
554cfb7e 586 * Default Router Selection (RFC 2461 6.3.6)
1da177e4 587 */
b6f99a21 588static inline int rt6_check_dev(struct rt6_info *rt, int oif)
554cfb7e 589{
d1918542 590 struct net_device *dev = rt->dst.dev;
161980f4 591 if (!oif || dev->ifindex == oif)
554cfb7e 592 return 2;
161980f4
DM		 593 if ((dev->flags & IFF_LOOPBACK) &&
594 rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif)
595 return 1;
596 return 0;
554cfb7e 597}
1da177e4 598
afc154e9 599static inline enum rt6_nud_state rt6_check_neigh(struct rt6_info *rt)
1da177e4 600{
f2c31e32 601 struct neighbour *neigh;
afc154e9 602 enum rt6_nud_state ret = RT6_NUD_FAIL_HARD;
f2c31e32 603
4d0c5911
YH		 604 if (rt->rt6i_flags & RTF_NONEXTHOP ||
605 !(rt->rt6i_flags & RTF_GATEWAY))
afc154e9 606 return RT6_NUD_SUCCEED;
145a3621
YH		 607
608 rcu_read_lock_bh();
609 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
610 if (neigh) {
611 read_lock(&neigh->lock);
554cfb7e 612 if (neigh->nud_state & NUD_VALID)
afc154e9 613 ret = RT6_NUD_SUCCEED;
398bcbeb 614#ifdef CONFIG_IPV6_ROUTER_PREF
a5a81f0b 615 else if (!(neigh->nud_state & NUD_FAILED))
afc154e9 616 ret = RT6_NUD_SUCCEED;
7e980569
JB		 617 else
618 ret = RT6_NUD_FAIL_PROBE;
398bcbeb 619#endif
145a3621 620 read_unlock(&neigh->lock);
afc154e9
HFS		 621 } else {
622 ret = IS_ENABLED(CONFIG_IPV6_ROUTER_PREF) ?
7e980569 623 RT6_NUD_SUCCEED : RT6_NUD_FAIL_DO_RR;
a5a81f0b 624 }
145a3621
YH		 625 rcu_read_unlock_bh();
626
a5a81f0b 627 return ret;
1da177e4
LT		 628}
629
554cfb7e
YH		 630static int rt6_score_route(struct rt6_info *rt, int oif,
631 int strict)
1da177e4 632{
a5a81f0b 633 int m;
1ab1457c 634
4d0c5911 635 m = rt6_check_dev(rt, oif);
77d16f45 636 if (!m && (strict & RT6_LOOKUP_F_IFACE))
afc154e9 637 return RT6_NUD_FAIL_HARD;
ebacaaa0
YH		 638#ifdef CONFIG_IPV6_ROUTER_PREF
639 m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2;
640#endif
afc154e9
HFS		 641 if (strict & RT6_LOOKUP_F_REACHABLE) {
642 int n = rt6_check_neigh(rt);
643 if (n < 0)
644 return n;
645 }
554cfb7e
YH		 646 return m;
647}
648
f11e6659 649static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict,
afc154e9
HFS		 650 int *mpri, struct rt6_info *match,
651 bool *do_rr)
554cfb7e 652{
f11e6659 653 int m;
afc154e9 654 bool match_do_rr = false;
35103d11
AG		 655 struct inet6_dev *idev = rt->rt6i_idev;
656 struct net_device *dev = rt->dst.dev;
657
658 if (dev && !netif_carrier_ok(dev) &&
659 idev->cnf.ignore_routes_with_linkdown)
660 goto out;
f11e6659
DM		 661
662 if (rt6_check_expired(rt))
663 goto out;
664
665 m = rt6_score_route(rt, oif, strict);
7e980569 666 if (m == RT6_NUD_FAIL_DO_RR) {
afc154e9
HFS		 667 match_do_rr = true;
668 m = 0; /* lowest valid score */
7e980569 669 } else if (m == RT6_NUD_FAIL_HARD) {
f11e6659 670 goto out;
afc154e9
HFS		 671 }
672
673 if (strict & RT6_LOOKUP_F_REACHABLE)
674 rt6_probe(rt);
f11e6659 675
7e980569 676 /* note that m can be RT6_NUD_FAIL_PROBE at this point */
f11e6659 677 if (m > *mpri) {
afc154e9 678 *do_rr = match_do_rr;
f11e6659
DM		 679 *mpri = m;
680 match = rt;
f11e6659 681 }
f11e6659
DM		 682out:
683 return match;
684}
685
686static struct rt6_info *find_rr_leaf(struct fib6_node *fn,
687 struct rt6_info *rr_head,
afc154e9
HFS		 688 u32 metric, int oif, int strict,
689 bool *do_rr)
f11e6659 690{
9fbdcfaf 691 struct rt6_info *rt, *match, *cont;
554cfb7e 692 int mpri = -1;
1da177e4 693
f11e6659 694 match = NULL;
9fbdcfaf
SK		 695 cont = NULL;
696 for (rt = rr_head; rt; rt = rt->dst.rt6_next) {
697 if (rt->rt6i_metric != metric) {
698 cont = rt;
699 break;
700 }
701
702 match = find_match(rt, oif, strict, &mpri, match, do_rr);
703 }
704
705 for (rt = fn->leaf; rt && rt != rr_head; rt = rt->dst.rt6_next) {
706 if (rt->rt6i_metric != metric) {
707 cont = rt;
708 break;
709 }
710
afc154e9 711 match = find_match(rt, oif, strict, &mpri, match, do_rr);
9fbdcfaf
SK		 712 }
713
714 if (match || !cont)
715 return match;
716
717 for (rt = cont; rt; rt = rt->dst.rt6_next)
afc154e9 718 match = find_match(rt, oif, strict, &mpri, match, do_rr);
1da177e4 719
f11e6659
DM		 720 return match;
721}
1da177e4 722
f11e6659
DM		 723static struct rt6_info *rt6_select(struct fib6_node *fn, int oif, int strict)
724{
725 struct rt6_info *match, *rt0;
8ed67789 726 struct net *net;
afc154e9 727 bool do_rr = false;
1da177e4 728
f11e6659
DM		 729 rt0 = fn->rr_ptr;
730 if (!rt0)
731 fn->rr_ptr = rt0 = fn->leaf;
1da177e4 732
afc154e9
HFS		 733 match = find_rr_leaf(fn, rt0, rt0->rt6i_metric, oif, strict,
734 &do_rr);
1da177e4 735
afc154e9 736 if (do_rr) {
d8d1f30b 737 struct rt6_info *next = rt0->dst.rt6_next;
f11e6659 738
554cfb7e 739 /* no entries matched; do round-robin */
f11e6659
DM		 740 if (!next || next->rt6i_metric != rt0->rt6i_metric)
741 next = fn->leaf;
742
743 if (next != rt0)
744 fn->rr_ptr = next;
1da177e4 745 }
1da177e4 746
d1918542 747 net = dev_net(rt0->dst.dev);
a02cec21 748 return match ? match : net->ipv6.ip6_null_entry;
1da177e4
LT		 749}
750
8b9df265
MKL		 751static bool rt6_is_gw_or_nonexthop(const struct rt6_info *rt)
752{
753 return (rt->rt6i_flags & (RTF_NONEXTHOP | RTF_GATEWAY));
754}
755
70ceb4f5
YH		 756#ifdef CONFIG_IPV6_ROUTE_INFO
757int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
b71d1d42 758 const struct in6_addr *gwaddr)
70ceb4f5 759{
c346dca1 760 struct net *net = dev_net(dev);
70ceb4f5
YH		 761 struct route_info *rinfo = (struct route_info *) opt;
762 struct in6_addr prefix_buf, *prefix;
763 unsigned int pref;
4bed72e4 764 unsigned long lifetime;
70ceb4f5
YH		 765 struct rt6_info *rt;
766
767 if (len < sizeof(struct route_info)) {
768 return -EINVAL;
769 }
770
771 /* Sanity check for prefix_len and length */
772 if (rinfo->length > 3) {
773 return -EINVAL;
774 } else if (rinfo->prefix_len > 128) {
775 return -EINVAL;
776 } else if (rinfo->prefix_len > 64) {
777 if (rinfo->length < 2) {
778 return -EINVAL;
779 }
780 } else if (rinfo->prefix_len > 0) {
781 if (rinfo->length < 1) {
782 return -EINVAL;
783 }
784 }
785
786 pref = rinfo->route_pref;
787 if (pref == ICMPV6_ROUTER_PREF_INVALID)
3933fc95 788 return -EINVAL;
70ceb4f5 789
4bed72e4 790 lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ);
70ceb4f5
YH		 791
792 if (rinfo->length == 3)
793 prefix = (struct in6_addr *)rinfo->prefix;
794 else {
795 /* this function is safe */
796 ipv6_addr_prefix(&prefix_buf,
797 (struct in6_addr *)rinfo->prefix,
798 rinfo->prefix_len);
799 prefix = &prefix_buf;
800 }
801
f104a567
DJ		 802 if (rinfo->prefix_len == 0)
803 rt = rt6_get_dflt_router(gwaddr, dev);
804 else
805 rt = rt6_get_route_info(net, prefix, rinfo->prefix_len,
806 gwaddr, dev->ifindex);
70ceb4f5
YH		 807
808 if (rt && !lifetime) {
e0a1ad73 809 ip6_del_rt(rt);
70ceb4f5
YH		 810 rt = NULL;
811 }
812
813 if (!rt && lifetime)
efa2cea0 814 rt = rt6_add_route_info(net, prefix, rinfo->prefix_len, gwaddr, dev->ifindex,
70ceb4f5
YH		 815 pref);
816 else if (rt)
817 rt->rt6i_flags = RTF_ROUTEINFO |
818 (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref);
819
820 if (rt) {
1716a961
G		 821 if (!addrconf_finite_timeout(lifetime))
822 rt6_clean_expires(rt);
823 else
824 rt6_set_expires(rt, jiffies + HZ * lifetime);
825
94e187c0 826 ip6_rt_put(rt);
70ceb4f5
YH		 827 }
828 return 0;
829}
830#endif
831
a3c00e46
MKL		 832static struct fib6_node* fib6_backtrack(struct fib6_node *fn,
833 struct in6_addr *saddr)
834{
835 struct fib6_node *pn;
836 while (1) {
837 if (fn->fn_flags & RTN_TL_ROOT)
838 return NULL;
839 pn = fn->parent;
840 if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn)
841 fn = fib6_lookup(FIB6_SUBTREE(pn), NULL, saddr);
842 else
843 fn = pn;
844 if (fn->fn_flags & RTN_RTINFO)
845 return fn;
846 }
847}
c71099ac 848
8ed67789
DL		 849static struct rt6_info *ip6_pol_route_lookup(struct net *net,
850 struct fib6_table *table,
4c9483b2 851 struct flowi6 *fl6, int flags)
1da177e4
LT		 852{
853 struct fib6_node *fn;
854 struct rt6_info *rt;
855
c71099ac 856 read_lock_bh(&table->tb6_lock);
4c9483b2 857 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
c71099ac
TG		 858restart:
859 rt = fn->leaf;
4c9483b2 860 rt = rt6_device_match(net, rt, &fl6->saddr, fl6->flowi6_oif, flags);
51ebd318 861 if (rt->rt6i_nsiblings && fl6->flowi6_oif == 0)
52bd4c0c 862 rt = rt6_multipath_select(rt, fl6, fl6->flowi6_oif, flags);
a3c00e46
MKL		 863 if (rt == net->ipv6.ip6_null_entry) {
864 fn = fib6_backtrack(fn, &fl6->saddr);
865 if (fn)
866 goto restart;
867 }
d8d1f30b 868 dst_use(&rt->dst, jiffies);
c71099ac 869 read_unlock_bh(&table->tb6_lock);
b811580d
DA		 870
871 trace_fib6_table_lookup(net, rt, table->tb6_id, fl6);
872
c71099ac
TG		 873 return rt;
874
875}
876
67ba4152 877struct dst_entry *ip6_route_lookup(struct net *net, struct flowi6 *fl6,
ea6e574e
FW		 878 int flags)
879{
880 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_lookup);
881}
882EXPORT_SYMBOL_GPL(ip6_route_lookup);
883
9acd9f3a
YH		 884struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
885 const struct in6_addr *saddr, int oif, int strict)
c71099ac 886{
4c9483b2
DM		 887 struct flowi6 fl6 = {
888 .flowi6_oif = oif,
889 .daddr = *daddr,
c71099ac
TG		 890 };
891 struct dst_entry *dst;
77d16f45 892 int flags = strict ? RT6_LOOKUP_F_IFACE : 0;
c71099ac 893
adaa70bb 894 if (saddr) {
4c9483b2 895 memcpy(&fl6.saddr, saddr, sizeof(*saddr));
adaa70bb
TG		 896 flags |= RT6_LOOKUP_F_HAS_SADDR;
897 }
898
4c9483b2 899 dst = fib6_rule_lookup(net, &fl6, flags, ip6_pol_route_lookup);
c71099ac
TG		 900 if (dst->error == 0)
901 return (struct rt6_info *) dst;
902
903 dst_release(dst);
904
1da177e4
LT		 905 return NULL;
906}
7159039a
YH		 907EXPORT_SYMBOL(rt6_lookup);
908
c71099ac 909/* ip6_ins_rt is called with FREE table->tb6_lock.
1da177e4
LT		 910 It takes new route entry, the addition fails by any reason the
911 route is freed. In any case, if caller does not hold it, it may
912 be destroyed.
913 */
914
e5fd387a 915static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info,
e715b6d3 916 struct mx6_config *mxc)
1da177e4
LT		 917{
918 int err;
c71099ac 919 struct fib6_table *table;
1da177e4 920
c71099ac
TG		 921 table = rt->rt6i_table;
922 write_lock_bh(&table->tb6_lock);
e715b6d3 923 err = fib6_add(&table->tb6_root, rt, info, mxc);
c71099ac 924 write_unlock_bh(&table->tb6_lock);
1da177e4
LT		 925
926 return err;
927}
928
40e22e8f
TG		 929int ip6_ins_rt(struct rt6_info *rt)
930{
e715b6d3
FW		 931 struct nl_info info = { .nl_net = dev_net(rt->dst.dev), };
932 struct mx6_config mxc = { .mx = NULL, };
933
934 return __ip6_ins_rt(rt, &info, &mxc);
40e22e8f
TG		 935}
936
8b9df265
MKL		 937static struct rt6_info *ip6_rt_cache_alloc(struct rt6_info *ort,
938 const struct in6_addr *daddr,
939 const struct in6_addr *saddr)
1da177e4 940{
1da177e4
LT		 941 struct rt6_info *rt;
942
943 /*
944 * Clone the route.
945 */
946
d52d3997 947 if (ort->rt6i_flags & (RTF_CACHE | RTF_PCPU))
83a09abd 948 ort = (struct rt6_info *)ort->dst.from;
1da177e4 949
ad706862 950 rt = __ip6_dst_alloc(dev_net(ort->dst.dev), ort->dst.dev, 0);
83a09abd
MKL		 951
952 if (!rt)
953 return NULL;
954
955 ip6_rt_copy_init(rt, ort);
956 rt->rt6i_flags |= RTF_CACHE;
957 rt->rt6i_metric = 0;
958 rt->dst.flags |= DST_HOST;
959 rt->rt6i_dst.addr = *daddr;
960 rt->rt6i_dst.plen = 128;
1da177e4 961
83a09abd
MKL		 962 if (!rt6_is_gw_or_nonexthop(ort)) {
963 if (ort->rt6i_dst.plen != 128 &&
964 ipv6_addr_equal(&ort->rt6i_dst.addr, daddr))
965 rt->rt6i_flags |= RTF_ANYCAST;
1da177e4 966#ifdef CONFIG_IPV6_SUBTREES
83a09abd
MKL		 967 if (rt->rt6i_src.plen && saddr) {
968 rt->rt6i_src.addr = *saddr;
969 rt->rt6i_src.plen = 128;
8b9df265 970 }
83a09abd 971#endif
95a9a5ba 972 }
1da177e4 973
95a9a5ba
YH		 974 return rt;
975}
1da177e4 976
d52d3997
MKL		 977static struct rt6_info *ip6_rt_pcpu_alloc(struct rt6_info *rt)
978{
979 struct rt6_info *pcpu_rt;
980
981 pcpu_rt = __ip6_dst_alloc(dev_net(rt->dst.dev),
ad706862 982 rt->dst.dev, rt->dst.flags);
d52d3997
MKL		 983
984 if (!pcpu_rt)
985 return NULL;
986 ip6_rt_copy_init(pcpu_rt, rt);
987 pcpu_rt->rt6i_protocol = rt->rt6i_protocol;
988 pcpu_rt->rt6i_flags |= RTF_PCPU;
989 return pcpu_rt;
990}
991
992/* It should be called with read_lock_bh(&tb6_lock) acquired */
993static struct rt6_info *rt6_get_pcpu_route(struct rt6_info *rt)
994{
a73e4195 995 struct rt6_info *pcpu_rt, **p;
d52d3997
MKL		 996
997 p = this_cpu_ptr(rt->rt6i_pcpu);
998 pcpu_rt = *p;
999
a73e4195
MKL		 1000 if (pcpu_rt) {
1001 dst_hold(&pcpu_rt->dst);
1002 rt6_dst_from_metrics_check(pcpu_rt);
1003 }
1004 return pcpu_rt;
1005}
1006
1007static struct rt6_info *rt6_make_pcpu_route(struct rt6_info *rt)
1008{
9c7370a1 1009 struct fib6_table *table = rt->rt6i_table;
a73e4195 1010 struct rt6_info *pcpu_rt, *prev, **p;
d52d3997
MKL		 1011
1012 pcpu_rt = ip6_rt_pcpu_alloc(rt);
1013 if (!pcpu_rt) {
1014 struct net *net = dev_net(rt->dst.dev);
1015
9c7370a1
MKL		 1016 dst_hold(&net->ipv6.ip6_null_entry->dst);
1017 return net->ipv6.ip6_null_entry;
d52d3997
MKL		 1018 }
1019
9c7370a1
MKL		 1020 read_lock_bh(&table->tb6_lock);
1021 if (rt->rt6i_pcpu) {
1022 p = this_cpu_ptr(rt->rt6i_pcpu);
1023 prev = cmpxchg(p, NULL, pcpu_rt);
1024 if (prev) {
1025 /* If someone did it before us, return prev instead */
1026 dst_destroy(&pcpu_rt->dst);
1027 pcpu_rt = prev;
1028 }
1029 } else {
1030 /* rt has been removed from the fib6 tree
1031 * before we have a chance to acquire the read_lock.
1032 * In this case, don't brother to create a pcpu rt
1033 * since rt is going away anyway. The next
1034 * dst_check() will trigger a re-lookup.
1035 */
d52d3997 1036 dst_destroy(&pcpu_rt->dst);
9c7370a1 1037 pcpu_rt = rt;
d52d3997 1038 }
d52d3997
MKL		 1039 dst_hold(&pcpu_rt->dst);
1040 rt6_dst_from_metrics_check(pcpu_rt);
9c7370a1 1041 read_unlock_bh(&table->tb6_lock);
d52d3997
MKL		 1042 return pcpu_rt;
1043}
1044
9ff74384
DA		 1045struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table,
1046 int oif, struct flowi6 *fl6, int flags)
1da177e4 1047{
367efcb9 1048 struct fib6_node *fn, *saved_fn;
45e4fd26 1049 struct rt6_info *rt;
c71099ac 1050 int strict = 0;
1da177e4 1051
77d16f45 1052 strict |= flags & RT6_LOOKUP_F_IFACE;
367efcb9
MKL		 1053 if (net->ipv6.devconf_all->forwarding == 0)
1054 strict |= RT6_LOOKUP_F_REACHABLE;
1da177e4 1055
c71099ac 1056 read_lock_bh(&table->tb6_lock);
1da177e4 1057
4c9483b2 1058 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
367efcb9 1059 saved_fn = fn;
1da177e4 1060
ca254490
DA		 1061 if (fl6->flowi6_flags & FLOWI_FLAG_SKIP_NH_OIF)
1062 oif = 0;
1063
a3c00e46 1064redo_rt6_select:
367efcb9 1065 rt = rt6_select(fn, oif, strict);
52bd4c0c 1066 if (rt->rt6i_nsiblings)
367efcb9 1067 rt = rt6_multipath_select(rt, fl6, oif, strict);
a3c00e46
MKL		 1068 if (rt == net->ipv6.ip6_null_entry) {
1069 fn = fib6_backtrack(fn, &fl6->saddr);
1070 if (fn)
1071 goto redo_rt6_select;
367efcb9
MKL		 1072 else if (strict & RT6_LOOKUP_F_REACHABLE) {
1073 /* also consider unreachable route */
1074 strict &= ~RT6_LOOKUP_F_REACHABLE;
1075 fn = saved_fn;
1076 goto redo_rt6_select;
367efcb9 1077 }
a3c00e46
MKL		 1078 }
1079
fb9de91e 1080
3da59bd9 1081 if (rt == net->ipv6.ip6_null_entry || (rt->rt6i_flags & RTF_CACHE)) {
d52d3997
MKL		 1082 dst_use(&rt->dst, jiffies);
1083 read_unlock_bh(&table->tb6_lock);
1084
1085 rt6_dst_from_metrics_check(rt);
b811580d
DA		 1086
1087 trace_fib6_table_lookup(net, rt, table->tb6_id, fl6);
d52d3997 1088 return rt;
3da59bd9
MKL		 1089 } else if (unlikely((fl6->flowi6_flags & FLOWI_FLAG_KNOWN_NH) &&
1090 !(rt->rt6i_flags & RTF_GATEWAY))) {
1091 /* Create a RTF_CACHE clone which will not be
1092 * owned by the fib6 tree. It is for the special case where
1093 * the daddr in the skb during the neighbor look-up is different
1094 * from the fl6->daddr used to look-up route here.
1095 */
1096
1097 struct rt6_info *uncached_rt;
1098
d52d3997
MKL		 1099 dst_use(&rt->dst, jiffies);
1100 read_unlock_bh(&table->tb6_lock);
1101
3da59bd9
MKL		 1102 uncached_rt = ip6_rt_cache_alloc(rt, &fl6->daddr, NULL);
1103 dst_release(&rt->dst);
c71099ac 1104
3da59bd9 1105 if (uncached_rt)
8d0b94af 1106 rt6_uncached_list_add(uncached_rt);
3da59bd9
MKL		 1107 else
1108 uncached_rt = net->ipv6.ip6_null_entry;
d52d3997 1109
3da59bd9 1110 dst_hold(&uncached_rt->dst);
b811580d
DA		 1111
1112 trace_fib6_table_lookup(net, uncached_rt, table->tb6_id, fl6);
3da59bd9 1113 return uncached_rt;
3da59bd9 1114
d52d3997
MKL		 1115 } else {
1116 /* Get a percpu copy */
1117
1118 struct rt6_info *pcpu_rt;
1119
1120 rt->dst.lastuse = jiffies;
1121 rt->dst.__use++;
1122 pcpu_rt = rt6_get_pcpu_route(rt);
d52d3997 1123
9c7370a1
MKL		 1124 if (pcpu_rt) {
1125 read_unlock_bh(&table->tb6_lock);
1126 } else {
1127 /* We have to do the read_unlock first
1128 * because rt6_make_pcpu_route() may trigger
1129 * ip6_dst_gc() which will take the write_lock.
1130 */
1131 dst_hold(&rt->dst);
1132 read_unlock_bh(&table->tb6_lock);
a73e4195 1133 pcpu_rt = rt6_make_pcpu_route(rt);
9c7370a1
MKL		 1134 dst_release(&rt->dst);
1135 }
d52d3997 1136
b811580d 1137 trace_fib6_table_lookup(net, pcpu_rt, table->tb6_id, fl6);
d52d3997 1138 return pcpu_rt;
9c7370a1 1139
d52d3997 1140 }
1da177e4 1141}
9ff74384 1142EXPORT_SYMBOL_GPL(ip6_pol_route);
1da177e4 1143
8ed67789 1144static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table,
4c9483b2 1145 struct flowi6 *fl6, int flags)
4acad72d 1146{
4c9483b2 1147 return ip6_pol_route(net, table, fl6->flowi6_iif, fl6, flags);
4acad72d
PE		 1148}
1149
72331bc0
SL		 1150static struct dst_entry *ip6_route_input_lookup(struct net *net,
1151 struct net_device *dev,
1152 struct flowi6 *fl6, int flags)
1153{
1154 if (rt6_need_strict(&fl6->daddr) && dev->type != ARPHRD_PIMREG)
1155 flags |= RT6_LOOKUP_F_IFACE;
1156
1157 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_input);
1158}
1159
c71099ac
TG		 1160void ip6_route_input(struct sk_buff *skb)
1161{
b71d1d42 1162 const struct ipv6hdr *iph = ipv6_hdr(skb);
c346dca1 1163 struct net *net = dev_net(skb->dev);
adaa70bb 1164 int flags = RT6_LOOKUP_F_HAS_SADDR;
904af04d 1165 struct ip_tunnel_info *tun_info;
4c9483b2 1166 struct flowi6 fl6 = {
ca254490 1167 .flowi6_iif = l3mdev_fib_oif(skb->dev),
4c9483b2
DM		 1168 .daddr = iph->daddr,
1169 .saddr = iph->saddr,
6502ca52 1170 .flowlabel = ip6_flowinfo(iph),
4c9483b2
DM		 1171 .flowi6_mark = skb->mark,
1172 .flowi6_proto = iph->nexthdr,
c71099ac 1173 };
adaa70bb 1174
904af04d 1175 tun_info = skb_tunnel_info(skb);
46fa062a 1176 if (tun_info && !(tun_info->mode & IP_TUNNEL_INFO_TX))
904af04d 1177 fl6.flowi6_tun_key.tun_id = tun_info->key.tun_id;
06e9d040 1178 skb_dst_drop(skb);
72331bc0 1179 skb_dst_set(skb, ip6_route_input_lookup(net, skb->dev, &fl6, flags));
c71099ac
TG		 1180}
1181
8ed67789 1182static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table,
4c9483b2 1183 struct flowi6 *fl6, int flags)
1da177e4 1184{
4c9483b2 1185 return ip6_pol_route(net, table, fl6->flowi6_oif, fl6, flags);
c71099ac
TG		 1186}
1187
6f21c96a
PA		 1188struct dst_entry *ip6_route_output_flags(struct net *net, const struct sock *sk,
1189 struct flowi6 *fl6, int flags)
c71099ac 1190{
d46a9d67 1191 bool any_src;
c71099ac 1192
4c1feac5
DA		 1193 if (rt6_need_strict(&fl6->daddr)) {
1194 struct dst_entry *dst;
1195
1196 dst = l3mdev_link_scope_lookup(net, fl6);
1197 if (dst)
1198 return dst;
1199 }
ca254490 1200
1fb9489b 1201 fl6->flowi6_iif = LOOPBACK_IFINDEX;
4dc27d1c 1202
d46a9d67 1203 any_src = ipv6_addr_any(&fl6->saddr);
741a11d9 1204 if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl6->daddr) ||
d46a9d67 1205 (fl6->flowi6_oif && any_src))
77d16f45 1206 flags |= RT6_LOOKUP_F_IFACE;
c71099ac 1207
d46a9d67 1208 if (!any_src)
adaa70bb 1209 flags |= RT6_LOOKUP_F_HAS_SADDR;
0c9a2ac1
YH		 1210 else if (sk)
1211 flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs);
adaa70bb 1212
4c9483b2 1213 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_output);
1da177e4 1214}
6f21c96a 1215EXPORT_SYMBOL_GPL(ip6_route_output_flags);
1da177e4 1216
2774c131 1217struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_orig)
14e50e57 1218{
5c1e6aa3 1219 struct rt6_info *rt, *ort = (struct rt6_info *) dst_orig;
14e50e57
DM		 1220 struct dst_entry *new = NULL;
1221
f5b0a874 1222 rt = dst_alloc(&ip6_dst_blackhole_ops, ort->dst.dev, 1, DST_OBSOLETE_NONE, 0);
14e50e57 1223 if (rt) {
0a1f5962 1224 rt6_info_init(rt);
8104891b 1225
0a1f5962 1226 new = &rt->dst;
14e50e57 1227 new->__use = 1;
352e512c 1228 new->input = dst_discard;
ede2059d 1229 new->output = dst_discard_out;
14e50e57 1230
0a1f5962 1231 dst_copy_metrics(new, &ort->dst);
14e50e57
DM		 1232 rt->rt6i_idev = ort->rt6i_idev;
1233 if (rt->rt6i_idev)
1234 in6_dev_hold(rt->rt6i_idev);
14e50e57 1235
4e3fd7a0 1236 rt->rt6i_gateway = ort->rt6i_gateway;
0a1f5962 1237 rt->rt6i_flags = ort->rt6i_flags & ~RTF_PCPU;
14e50e57
DM		 1238 rt->rt6i_metric = 0;
1239
1240 memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key));
1241#ifdef CONFIG_IPV6_SUBTREES
1242 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1243#endif
1244
1245 dst_free(new);
1246 }
1247
69ead7af
DM		 1248 dst_release(dst_orig);
1249 return new ? new : ERR_PTR(-ENOMEM);
14e50e57 1250}
14e50e57 1251
1da177e4
LT		 1252/*
1253 * Destination cache support functions
1254 */
1255
4b32b5ad
MKL		 1256static void rt6_dst_from_metrics_check(struct rt6_info *rt)
1257{
1258 if (rt->dst.from &&
1259 dst_metrics_ptr(&rt->dst) != dst_metrics_ptr(rt->dst.from))
1260 dst_init_metrics(&rt->dst, dst_metrics_ptr(rt->dst.from), true);
1261}
1262
3da59bd9
MKL		 1263static struct dst_entry *rt6_check(struct rt6_info *rt, u32 cookie)
1264{
1265 if (!rt->rt6i_node || (rt->rt6i_node->fn_sernum != cookie))
1266 return NULL;
1267
1268 if (rt6_check_expired(rt))
1269 return NULL;
1270
1271 return &rt->dst;
1272}
1273
1274static struct dst_entry *rt6_dst_from_check(struct rt6_info *rt, u32 cookie)
1275{
5973fb1e
MKL		 1276 if (!__rt6_check_expired(rt) &&
1277 rt->dst.obsolete == DST_OBSOLETE_FORCE_CHK &&
3da59bd9
MKL		 1278 rt6_check((struct rt6_info *)(rt->dst.from), cookie))
1279 return &rt->dst;
1280 else
1281 return NULL;
1282}
1283
1da177e4
LT		 1284static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie)
1285{
1286 struct rt6_info *rt;
1287
1288 rt = (struct rt6_info *) dst;
1289
6f3118b5
ND		 1290 /* All IPV6 dsts are created with ->obsolete set to the value
1291 * DST_OBSOLETE_FORCE_CHK which forces validation calls down
1292 * into this function always.
1293 */
e3bc10bd 1294
4b32b5ad
MKL		 1295 rt6_dst_from_metrics_check(rt);
1296
02bcf4e0
MKL		 1297 if (rt->rt6i_flags & RTF_PCPU ||
1298 (unlikely(dst->flags & DST_NOCACHE) && rt->dst.from))
3da59bd9
MKL		 1299 return rt6_dst_from_check(rt, cookie);
1300 else
1301 return rt6_check(rt, cookie);
1da177e4
LT		 1302}
1303
1304static struct dst_entry *ip6_negative_advice(struct dst_entry *dst)
1305{
1306 struct rt6_info *rt = (struct rt6_info *) dst;
1307
1308 if (rt) {
54c1a859
YH		 1309 if (rt->rt6i_flags & RTF_CACHE) {
1310 if (rt6_check_expired(rt)) {
1311 ip6_del_rt(rt);
1312 dst = NULL;
1313 }
1314 } else {
1da177e4 1315 dst_release(dst);
54c1a859
YH		 1316 dst = NULL;
1317 }
1da177e4 1318 }
54c1a859 1319 return dst;
1da177e4
LT		 1320}
1321
1322static void ip6_link_failure(struct sk_buff *skb)
1323{
1324 struct rt6_info *rt;
1325
3ffe533c 1326 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0);
1da177e4 1327
adf30907 1328 rt = (struct rt6_info *) skb_dst(skb);
1da177e4 1329 if (rt) {
1eb4f758
HFS		 1330 if (rt->rt6i_flags & RTF_CACHE) {
1331 dst_hold(&rt->dst);
8e3d5be7 1332 ip6_del_rt(rt);
1eb4f758 1333 } else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT)) {
1da177e4 1334 rt->rt6i_node->fn_sernum = -1;
1eb4f758 1335 }
1da177e4
LT		 1336 }
1337}
1338
45e4fd26
MKL		 1339static void rt6_do_update_pmtu(struct rt6_info *rt, u32 mtu)
1340{
1341 struct net *net = dev_net(rt->dst.dev);
1342
1343 rt->rt6i_flags |= RTF_MODIFIED;
1344 rt->rt6i_pmtu = mtu;
1345 rt6_update_expires(rt, net->ipv6.sysctl.ip6_rt_mtu_expires);
1346}
1347
0d3f6d29
MKL		 1348static bool rt6_cache_allowed_for_pmtu(const struct rt6_info *rt)
1349{
1350 return !(rt->rt6i_flags & RTF_CACHE) &&
1351 (rt->rt6i_flags & RTF_PCPU || rt->rt6i_node);
1352}
1353
45e4fd26
MKL		 1354static void __ip6_rt_update_pmtu(struct dst_entry *dst, const struct sock *sk,
1355 const struct ipv6hdr *iph, u32 mtu)
1da177e4 1356{
67ba4152 1357 struct rt6_info *rt6 = (struct rt6_info *)dst;
1da177e4 1358
45e4fd26
MKL		 1359 if (rt6->rt6i_flags & RTF_LOCAL)
1360 return;
81aded24 1361
45e4fd26
MKL		 1362 dst_confirm(dst);
1363 mtu = max_t(u32, mtu, IPV6_MIN_MTU);
1364 if (mtu >= dst_mtu(dst))
1365 return;
9d289715 1366
0d3f6d29 1367 if (!rt6_cache_allowed_for_pmtu(rt6)) {
45e4fd26
MKL		 1368 rt6_do_update_pmtu(rt6, mtu);
1369 } else {
1370 const struct in6_addr *daddr, *saddr;
1371 struct rt6_info *nrt6;
1372
1373 if (iph) {
1374 daddr = &iph->daddr;
1375 saddr = &iph->saddr;
1376 } else if (sk) {
1377 daddr = &sk->sk_v6_daddr;
1378 saddr = &inet6_sk(sk)->saddr;
1379 } else {
1380 return;
1381 }
1382 nrt6 = ip6_rt_cache_alloc(rt6, daddr, saddr);
1383 if (nrt6) {
1384 rt6_do_update_pmtu(nrt6, mtu);
1385
1386 /* ip6_ins_rt(nrt6) will bump the
1387 * rt6->rt6i_node->fn_sernum
1388 * which will fail the next rt6_check() and
1389 * invalidate the sk->sk_dst_cache.
1390 */
1391 ip6_ins_rt(nrt6);
1392 }
1da177e4
LT		 1393 }
1394}
1395
45e4fd26
MKL		 1396static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
1397 struct sk_buff *skb, u32 mtu)
1398{
1399 __ip6_rt_update_pmtu(dst, sk, skb ? ipv6_hdr(skb) : NULL, mtu);
1400}
1401
42ae66c8
DM		 1402void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu,
1403 int oif, u32 mark)
81aded24
DM		 1404{
1405 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1406 struct dst_entry *dst;
1407 struct flowi6 fl6;
1408
1409 memset(&fl6, 0, sizeof(fl6));
1410 fl6.flowi6_oif = oif;
1b3c61dc 1411 fl6.flowi6_mark = mark ? mark : IP6_REPLY_MARK(net, skb->mark);
81aded24
DM		 1412 fl6.daddr = iph->daddr;
1413 fl6.saddr = iph->saddr;
6502ca52 1414 fl6.flowlabel = ip6_flowinfo(iph);
81aded24
DM		 1415
1416 dst = ip6_route_output(net, NULL, &fl6);
1417 if (!dst->error)
45e4fd26 1418 __ip6_rt_update_pmtu(dst, NULL, iph, ntohl(mtu));
81aded24
DM		 1419 dst_release(dst);
1420}
1421EXPORT_SYMBOL_GPL(ip6_update_pmtu);
1422
1423void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu)
1424{
33c162a9
MKL		 1425 struct dst_entry *dst;
1426
81aded24
DM		 1427 ip6_update_pmtu(skb, sock_net(sk), mtu,
1428 sk->sk_bound_dev_if, sk->sk_mark);
33c162a9
MKL		 1429
1430 dst = __sk_dst_get(sk);
1431 if (!dst || !dst->obsolete ||
1432 dst->ops->check(dst, inet6_sk(sk)->dst_cookie))
1433 return;
1434
1435 bh_lock_sock(sk);
1436 if (!sock_owned_by_user(sk) && !ipv6_addr_v4mapped(&sk->sk_v6_daddr))
1437 ip6_datagram_dst_update(sk, false);
1438 bh_unlock_sock(sk);
81aded24
DM		 1439}
1440EXPORT_SYMBOL_GPL(ip6_sk_update_pmtu);
1441
b55b76b2
DJ		 1442/* Handle redirects */
1443struct ip6rd_flowi {
1444 struct flowi6 fl6;
1445 struct in6_addr gateway;
1446};
1447
1448static struct rt6_info *__ip6_route_redirect(struct net *net,
1449 struct fib6_table *table,
1450 struct flowi6 *fl6,
1451 int flags)
1452{
1453 struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl6;
1454 struct rt6_info *rt;
1455 struct fib6_node *fn;
1456
1457 /* Get the "current" route for this destination and
1458 * check if the redirect has come from approriate router.
1459 *
1460 * RFC 4861 specifies that redirects should only be
1461 * accepted if they come from the nexthop to the target.
1462 * Due to the way the routes are chosen, this notion
1463 * is a bit fuzzy and one might need to check all possible
1464 * routes.
1465 */
1466
1467 read_lock_bh(&table->tb6_lock);
1468 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
1469restart:
1470 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1471 if (rt6_check_expired(rt))
1472 continue;
1473 if (rt->dst.error)
1474 break;
1475 if (!(rt->rt6i_flags & RTF_GATEWAY))
1476 continue;
1477 if (fl6->flowi6_oif != rt->dst.dev->ifindex)
1478 continue;
1479 if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway))
1480 continue;
1481 break;
1482 }
1483
1484 if (!rt)
1485 rt = net->ipv6.ip6_null_entry;
1486 else if (rt->dst.error) {
1487 rt = net->ipv6.ip6_null_entry;
b0a1ba59
MKL		 1488 goto out;
1489 }
1490
1491 if (rt == net->ipv6.ip6_null_entry) {
a3c00e46
MKL		 1492 fn = fib6_backtrack(fn, &fl6->saddr);
1493 if (fn)
1494 goto restart;
b55b76b2 1495 }
a3c00e46 1496
b0a1ba59 1497out:
b55b76b2
DJ		 1498 dst_hold(&rt->dst);
1499
1500 read_unlock_bh(&table->tb6_lock);
1501
b811580d 1502 trace_fib6_table_lookup(net, rt, table->tb6_id, fl6);
b55b76b2
DJ		 1503 return rt;
1504};
1505
1506static struct dst_entry *ip6_route_redirect(struct net *net,
1507 const struct flowi6 *fl6,
1508 const struct in6_addr *gateway)
1509{
1510 int flags = RT6_LOOKUP_F_HAS_SADDR;
1511 struct ip6rd_flowi rdfl;
1512
1513 rdfl.fl6 = *fl6;
1514 rdfl.gateway = *gateway;
1515
1516 return fib6_rule_lookup(net, &rdfl.fl6,
1517 flags, __ip6_route_redirect);
1518}
1519
3a5ad2ee
DM		 1520void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark)
1521{
1522 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1523 struct dst_entry *dst;
1524 struct flowi6 fl6;
1525
1526 memset(&fl6, 0, sizeof(fl6));
e374c618 1527 fl6.flowi6_iif = LOOPBACK_IFINDEX;
3a5ad2ee
DM		 1528 fl6.flowi6_oif = oif;
1529 fl6.flowi6_mark = mark;
3a5ad2ee
DM		 1530 fl6.daddr = iph->daddr;
1531 fl6.saddr = iph->saddr;
6502ca52 1532 fl6.flowlabel = ip6_flowinfo(iph);
3a5ad2ee 1533
b55b76b2
DJ		 1534 dst = ip6_route_redirect(net, &fl6, &ipv6_hdr(skb)->saddr);
1535 rt6_do_redirect(dst, NULL, skb);
3a5ad2ee
DM		 1536 dst_release(dst);
1537}
1538EXPORT_SYMBOL_GPL(ip6_redirect);
1539
c92a59ec
DJ		 1540void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif,
1541 u32 mark)
1542{
1543 const struct ipv6hdr *iph = ipv6_hdr(skb);
1544 const struct rd_msg *msg = (struct rd_msg *)icmp6_hdr(skb);
1545 struct dst_entry *dst;
1546 struct flowi6 fl6;
1547
1548 memset(&fl6, 0, sizeof(fl6));
e374c618 1549 fl6.flowi6_iif = LOOPBACK_IFINDEX;
c92a59ec
DJ		 1550 fl6.flowi6_oif = oif;
1551 fl6.flowi6_mark = mark;
c92a59ec
DJ		 1552 fl6.daddr = msg->dest;
1553 fl6.saddr = iph->daddr;
1554
b55b76b2
DJ		 1555 dst = ip6_route_redirect(net, &fl6, &iph->saddr);
1556 rt6_do_redirect(dst, NULL, skb);
c92a59ec
DJ		 1557 dst_release(dst);
1558}
1559
3a5ad2ee
DM		 1560void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk)
1561{
1562 ip6_redirect(skb, sock_net(sk), sk->sk_bound_dev_if, sk->sk_mark);
1563}
1564EXPORT_SYMBOL_GPL(ip6_sk_redirect);
1565
0dbaee3b 1566static unsigned int ip6_default_advmss(const struct dst_entry *dst)
1da177e4 1567{
0dbaee3b
DM		 1568 struct net_device *dev = dst->dev;
1569 unsigned int mtu = dst_mtu(dst);
1570 struct net *net = dev_net(dev);
1571
1da177e4
LT		 1572 mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr);
1573
5578689a
DL		 1574 if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss)
1575 mtu = net->ipv6.sysctl.ip6_rt_min_advmss;
1da177e4
LT		 1576
1577 /*
1ab1457c
YH		 1578 * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and
1579 * corresponding MSS is IPV6_MAXPLEN - tcp_header_size.
1580 * IPV6_MAXPLEN is also valid and means: "any MSS,
1da177e4
LT		 1581 * rely only on pmtu discovery"
1582 */
1583 if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr))
1584 mtu = IPV6_MAXPLEN;
1585 return mtu;
1586}
1587
ebb762f2 1588static unsigned int ip6_mtu(const struct dst_entry *dst)
d33e4553 1589{
4b32b5ad
MKL		 1590 const struct rt6_info *rt = (const struct rt6_info *)dst;
1591 unsigned int mtu = rt->rt6i_pmtu;
d33e4553 1592 struct inet6_dev *idev;
618f9bc7 1593
4b32b5ad
MKL		 1594 if (mtu)
1595 goto out;
1596
1597 mtu = dst_metric_raw(dst, RTAX_MTU);
618f9bc7 1598 if (mtu)
30f78d8e 1599 goto out;
618f9bc7
SK		 1600
1601 mtu = IPV6_MIN_MTU;
d33e4553
DM		 1602
1603 rcu_read_lock();
1604 idev = __in6_dev_get(dst->dev);
1605 if (idev)
1606 mtu = idev->cnf.mtu6;
1607 rcu_read_unlock();
1608
30f78d8e 1609out:
14972cbd
RP		 1610 mtu = min_t(unsigned int, mtu, IP6_MAX_MTU);
1611
1612 return mtu - lwtunnel_headroom(dst->lwtstate, mtu);
d33e4553
DM		 1613}
1614
3b00944c
YH		 1615static struct dst_entry *icmp6_dst_gc_list;
1616static DEFINE_SPINLOCK(icmp6_dst_lock);
5d0bbeeb 1617
3b00944c 1618struct dst_entry *icmp6_dst_alloc(struct net_device *dev,
87a11578 1619 struct flowi6 *fl6)
1da177e4 1620{
87a11578 1621 struct dst_entry *dst;
1da177e4
LT		 1622 struct rt6_info *rt;
1623 struct inet6_dev *idev = in6_dev_get(dev);
c346dca1 1624 struct net *net = dev_net(dev);
1da177e4 1625
38308473 1626 if (unlikely(!idev))
122bdf67 1627 return ERR_PTR(-ENODEV);
1da177e4 1628
ad706862 1629 rt = ip6_dst_alloc(net, dev, 0);
38308473 1630 if (unlikely(!rt)) {
1da177e4 1631 in6_dev_put(idev);
87a11578 1632 dst = ERR_PTR(-ENOMEM);
1da177e4
LT		 1633 goto out;
1634 }
1635
8e2ec639
YZ		 1636 rt->dst.flags |= DST_HOST;
1637 rt->dst.output = ip6_output;
d8d1f30b 1638 atomic_set(&rt->dst.__refcnt, 1);
550bab42 1639 rt->rt6i_gateway = fl6->daddr;
87a11578 1640 rt->rt6i_dst.addr = fl6->daddr;
8e2ec639
YZ		 1641 rt->rt6i_dst.plen = 128;
1642 rt->rt6i_idev = idev;
14edd87d 1643 dst_metric_set(&rt->dst, RTAX_HOPLIMIT, 0);
1da177e4 1644
3b00944c 1645 spin_lock_bh(&icmp6_dst_lock);
d8d1f30b
CG		 1646 rt->dst.next = icmp6_dst_gc_list;
1647 icmp6_dst_gc_list = &rt->dst;
3b00944c 1648 spin_unlock_bh(&icmp6_dst_lock);
1da177e4 1649
5578689a 1650 fib6_force_start_gc(net);
1da177e4 1651
87a11578
DM		 1652 dst = xfrm_lookup(net, &rt->dst, flowi6_to_flowi(fl6), NULL, 0);
1653
1da177e4 1654out:
87a11578 1655 return dst;
1da177e4
LT		 1656}
1657
3d0f24a7 1658int icmp6_dst_gc(void)
1da177e4 1659{
e9476e95 1660 struct dst_entry *dst, **pprev;
3d0f24a7 1661 int more = 0;
1da177e4 1662
3b00944c
YH		 1663 spin_lock_bh(&icmp6_dst_lock);
1664 pprev = &icmp6_dst_gc_list;
5d0bbeeb 1665
1da177e4
LT		 1666 while ((dst = *pprev) != NULL) {
1667 if (!atomic_read(&dst->__refcnt)) {
1668 *pprev = dst->next;
1669 dst_free(dst);
1da177e4
LT		 1670 } else {
1671 pprev = &dst->next;
3d0f24a7 1672 ++more;
1da177e4
LT		 1673 }
1674 }
1675
3b00944c 1676 spin_unlock_bh(&icmp6_dst_lock);
5d0bbeeb 1677
3d0f24a7 1678 return more;
1da177e4
LT		 1679}
1680
1e493d19
DM		 1681static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg),
1682 void *arg)
1683{
1684 struct dst_entry *dst, **pprev;
1685
1686 spin_lock_bh(&icmp6_dst_lock);
1687 pprev = &icmp6_dst_gc_list;
1688 while ((dst = *pprev) != NULL) {
1689 struct rt6_info *rt = (struct rt6_info *) dst;
1690 if (func(rt, arg)) {
1691 *pprev = dst->next;
1692 dst_free(dst);
1693 } else {
1694 pprev = &dst->next;
1695 }
1696 }
1697 spin_unlock_bh(&icmp6_dst_lock);
1698}
1699
569d3645 1700static int ip6_dst_gc(struct dst_ops *ops)
1da177e4 1701{
86393e52 1702 struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops);
7019b78e
DL		 1703 int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval;
1704 int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size;
1705 int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity;
1706 int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout;
1707 unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc;
fc66f95c 1708 int entries;
7019b78e 1709
fc66f95c 1710 entries = dst_entries_get_fast(ops);
49a18d86 1711 if (time_after(rt_last_gc + rt_min_interval, jiffies) &&
fc66f95c 1712 entries <= rt_max_size)
1da177e4
LT		 1713 goto out;
1714
6891a346 1715 net->ipv6.ip6_rt_gc_expire++;
14956643 1716 fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net, true);
fc66f95c
ED		 1717 entries = dst_entries_get_slow(ops);
1718 if (entries < ops->gc_thresh)
7019b78e 1719 net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1;
1da177e4 1720out:
7019b78e 1721 net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity;
fc66f95c 1722 return entries > rt_max_size;
1da177e4
LT		 1723}
1724
e715b6d3
FW		 1725static int ip6_convert_metrics(struct mx6_config *mxc,
1726 const struct fib6_config *cfg)
1727{
c3a8d947 1728 bool ecn_ca = false;
e715b6d3
FW		 1729 struct nlattr *nla;
1730 int remaining;
1731 u32 *mp;
1732
63159f29 1733 if (!cfg->fc_mx)
e715b6d3
FW		 1734 return 0;
1735
1736 mp = kzalloc(sizeof(u32) * RTAX_MAX, GFP_KERNEL);
1737 if (unlikely(!mp))
1738 return -ENOMEM;
1739
1740 nla_for_each_attr(nla, cfg->fc_mx, cfg->fc_mx_len, remaining) {
1741 int type = nla_type(nla);
1bb14807 1742 u32 val;
e715b6d3 1743
1bb14807
DB		 1744 if (!type)
1745 continue;
1746 if (unlikely(type > RTAX_MAX))
1747 goto err;
ea697639 1748
1bb14807
DB		 1749 if (type == RTAX_CC_ALGO) {
1750 char tmp[TCP_CA_NAME_MAX];
e715b6d3 1751
1bb14807 1752 nla_strlcpy(tmp, nla, sizeof(tmp));
c3a8d947 1753 val = tcp_ca_get_key_by_name(tmp, &ecn_ca);
1bb14807
DB		 1754 if (val == TCP_CA_UNSPEC)
1755 goto err;
1756 } else {
1757 val = nla_get_u32(nla);
e715b6d3 1758 }
626abd59
PA		 1759 if (type == RTAX_HOPLIMIT && val > 255)
1760 val = 255;
b8d3e416
DB		 1761 if (type == RTAX_FEATURES && (val & ~RTAX_FEATURE_MASK))
1762 goto err;
1bb14807
DB		 1763
1764 mp[type - 1] = val;
1765 __set_bit(type - 1, mxc->mx_valid);
e715b6d3
FW		 1766 }
1767
c3a8d947
DB		 1768 if (ecn_ca) {
1769 __set_bit(RTAX_FEATURES - 1, mxc->mx_valid);
1770 mp[RTAX_FEATURES - 1] |= DST_FEATURE_ECN_CA;
1771 }
e715b6d3 1772
c3a8d947 1773 mxc->mx = mp;
e715b6d3
FW		 1774 return 0;
1775 err:
1776 kfree(mp);
1777 return -EINVAL;
1778}
1da177e4 1779
8c14586f
DA		 1780static struct rt6_info *ip6_nh_lookup_table(struct net *net,
1781 struct fib6_config *cfg,
1782 const struct in6_addr *gw_addr)
1783{
1784 struct flowi6 fl6 = {
1785 .flowi6_oif = cfg->fc_ifindex,
1786 .daddr = *gw_addr,
1787 .saddr = cfg->fc_prefsrc,
1788 };
1789 struct fib6_table *table;
1790 struct rt6_info *rt;
48f1dcb5 1791 int flags = RT6_LOOKUP_F_IFACE;
8c14586f
DA		 1792
1793 table = fib6_get_table(net, cfg->fc_table);
1794 if (!table)
1795 return NULL;
1796
1797 if (!ipv6_addr_any(&cfg->fc_prefsrc))
1798 flags |= RT6_LOOKUP_F_HAS_SADDR;
1799
1800 rt = ip6_pol_route(net, table, cfg->fc_ifindex, &fl6, flags);
1801
1802 /* if table lookup failed, fall back to full lookup */
1803 if (rt == net->ipv6.ip6_null_entry) {
1804 ip6_rt_put(rt);
1805 rt = NULL;
1806 }
1807
1808 return rt;
1809}
1810
8c5b83f0 1811static struct rt6_info *ip6_route_info_create(struct fib6_config *cfg)
1da177e4 1812{
5578689a 1813 struct net *net = cfg->fc_nlinfo.nl_net;
1da177e4
LT		 1814 struct rt6_info *rt = NULL;
1815 struct net_device *dev = NULL;
1816 struct inet6_dev *idev = NULL;
c71099ac 1817 struct fib6_table *table;
1da177e4 1818 int addr_type;
8c5b83f0 1819 int err = -EINVAL;
1da177e4 1820
86872cb5 1821 if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128)
8c5b83f0 1822 goto out;
1da177e4 1823#ifndef CONFIG_IPV6_SUBTREES
86872cb5 1824 if (cfg->fc_src_len)
8c5b83f0 1825 goto out;
1da177e4 1826#endif
86872cb5 1827 if (cfg->fc_ifindex) {
1da177e4 1828 err = -ENODEV;
5578689a 1829 dev = dev_get_by_index(net, cfg->fc_ifindex);
1da177e4
LT		 1830 if (!dev)
1831 goto out;
1832 idev = in6_dev_get(dev);
1833 if (!idev)
1834 goto out;
1835 }
1836
86872cb5
TG		 1837 if (cfg->fc_metric == 0)
1838 cfg->fc_metric = IP6_RT_PRIO_USER;
1da177e4 1839
d71314b4 1840 err = -ENOBUFS;
38308473
DM		 1841 if (cfg->fc_nlinfo.nlh &&
1842 !(cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_CREATE)) {
d71314b4 1843 table = fib6_get_table(net, cfg->fc_table);
38308473 1844 if (!table) {
f3213831 1845 pr_warn("NLM_F_CREATE should be specified when creating new route\n");
d71314b4
MV		 1846 table = fib6_new_table(net, cfg->fc_table);
1847 }
1848 } else {
1849 table = fib6_new_table(net, cfg->fc_table);
1850 }
38308473
DM		 1851
1852 if (!table)
c71099ac 1853 goto out;
c71099ac 1854
ad706862
MKL		 1855 rt = ip6_dst_alloc(net, NULL,
1856 (cfg->fc_flags & RTF_ADDRCONF) ? 0 : DST_NOCOUNT);
1da177e4 1857
38308473 1858 if (!rt) {
1da177e4
LT		 1859 err = -ENOMEM;
1860 goto out;
1861 }
1862
1716a961
G		 1863 if (cfg->fc_flags & RTF_EXPIRES)
1864 rt6_set_expires(rt, jiffies +
1865 clock_t_to_jiffies(cfg->fc_expires));
1866 else
1867 rt6_clean_expires(rt);
1da177e4 1868
86872cb5
TG		 1869 if (cfg->fc_protocol == RTPROT_UNSPEC)
1870 cfg->fc_protocol = RTPROT_BOOT;
1871 rt->rt6i_protocol = cfg->fc_protocol;
1872
1873 addr_type = ipv6_addr_type(&cfg->fc_dst);
1da177e4
LT		 1874
1875 if (addr_type & IPV6_ADDR_MULTICAST)
d8d1f30b 1876 rt->dst.input = ip6_mc_input;
ab79ad14
MŻ		 1877 else if (cfg->fc_flags & RTF_LOCAL)
1878 rt->dst.input = ip6_input;
1da177e4 1879 else
d8d1f30b 1880 rt->dst.input = ip6_forward;
1da177e4 1881
d8d1f30b 1882 rt->dst.output = ip6_output;
1da177e4 1883
19e42e45
RP		 1884 if (cfg->fc_encap) {
1885 struct lwtunnel_state *lwtstate;
1886
1887 err = lwtunnel_build_state(dev, cfg->fc_encap_type,
127eb7cd
TH		 1888 cfg->fc_encap, AF_INET6, cfg,
1889 &lwtstate);
19e42e45
RP		 1890 if (err)
1891 goto out;
61adedf3
JB		 1892 rt->dst.lwtstate = lwtstate_get(lwtstate);
1893 if (lwtunnel_output_redirect(rt->dst.lwtstate)) {
1894 rt->dst.lwtstate->orig_output = rt->dst.output;
1895 rt->dst.output = lwtunnel_output;
25368623 1896 }
61adedf3
JB		 1897 if (lwtunnel_input_redirect(rt->dst.lwtstate)) {
1898 rt->dst.lwtstate->orig_input = rt->dst.input;
1899 rt->dst.input = lwtunnel_input;
25368623 1900 }
19e42e45
RP		 1901 }
1902
86872cb5
TG		 1903 ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len);
1904 rt->rt6i_dst.plen = cfg->fc_dst_len;
afc4eef8 1905 if (rt->rt6i_dst.plen == 128)
e5fd387a 1906 rt->dst.flags |= DST_HOST;
e5fd387a 1907
1da177e4 1908#ifdef CONFIG_IPV6_SUBTREES
86872cb5
TG		 1909 ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len);
1910 rt->rt6i_src.plen = cfg->fc_src_len;
1da177e4
LT		 1911#endif
1912
86872cb5 1913 rt->rt6i_metric = cfg->fc_metric;
1da177e4
LT		 1914
1915 /* We cannot add true routes via loopback here,
1916 they would result in kernel looping; promote them to reject routes
1917 */
86872cb5 1918 if ((cfg->fc_flags & RTF_REJECT) ||
38308473
DM		 1919 (dev && (dev->flags & IFF_LOOPBACK) &&
1920 !(addr_type & IPV6_ADDR_LOOPBACK) &&
1921 !(cfg->fc_flags & RTF_LOCAL))) {
1da177e4 1922 /* hold loopback dev/idev if we haven't done so. */
5578689a 1923 if (dev != net->loopback_dev) {
1da177e4
LT		 1924 if (dev) {
1925 dev_put(dev);
1926 in6_dev_put(idev);
1927 }
5578689a 1928 dev = net->loopback_dev;
1da177e4
LT		 1929 dev_hold(dev);
1930 idev = in6_dev_get(dev);
1931 if (!idev) {
1932 err = -ENODEV;
1933 goto out;
1934 }
1935 }
1da177e4 1936 rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP;
ef2c7d7b
ND		 1937 switch (cfg->fc_type) {
1938 case RTN_BLACKHOLE:
1939 rt->dst.error = -EINVAL;
ede2059d 1940 rt->dst.output = dst_discard_out;
7150aede 1941 rt->dst.input = dst_discard;
ef2c7d7b
ND		 1942 break;
1943 case RTN_PROHIBIT:
1944 rt->dst.error = -EACCES;
7150aede
K		 1945 rt->dst.output = ip6_pkt_prohibit_out;
1946 rt->dst.input = ip6_pkt_prohibit;
ef2c7d7b 1947 break;
b4949ab2 1948 case RTN_THROW:
0315e382 1949 case RTN_UNREACHABLE:
ef2c7d7b 1950 default:
7150aede 1951 rt->dst.error = (cfg->fc_type == RTN_THROW) ? -EAGAIN
0315e382
NF		 1952 : (cfg->fc_type == RTN_UNREACHABLE)
1953 ? -EHOSTUNREACH : -ENETUNREACH;
7150aede
K		 1954 rt->dst.output = ip6_pkt_discard_out;
1955 rt->dst.input = ip6_pkt_discard;
ef2c7d7b
ND		 1956 break;
1957 }
1da177e4
LT		 1958 goto install_route;
1959 }
1960
86872cb5 1961 if (cfg->fc_flags & RTF_GATEWAY) {
b71d1d42 1962 const struct in6_addr *gw_addr;
1da177e4
LT		 1963 int gwa_type;
1964
86872cb5 1965 gw_addr = &cfg->fc_gateway;
330567b7 1966 gwa_type = ipv6_addr_type(gw_addr);
48ed7b26
FW		 1967
1968 /* if gw_addr is local we will fail to detect this in case
1969 * address is still TENTATIVE (DAD in progress). rt6_lookup()
1970 * will return already-added prefix route via interface that
1971 * prefix route was assigned to, which might be non-loopback.
1972 */
1973 err = -EINVAL;
330567b7
FW		 1974 if (ipv6_chk_addr_and_flags(net, gw_addr,
1975 gwa_type & IPV6_ADDR_LINKLOCAL ?
1976 dev : NULL, 0, 0))
48ed7b26
FW		 1977 goto out;
1978
4e3fd7a0 1979 rt->rt6i_gateway = *gw_addr;
1da177e4
LT		 1980
1981 if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) {
8c14586f 1982 struct rt6_info *grt = NULL;
1da177e4
LT		 1983
1984 /* IPv6 strictly inhibits using not link-local
1985 addresses as nexthop address.
1986 Otherwise, router will not able to send redirects.
1987 It is very good, but in some (rare!) circumstances
1988 (SIT, PtP, NBMA NOARP links) it is handy to allow
1989 some exceptions. --ANK
1990 */
38308473 1991 if (!(gwa_type & IPV6_ADDR_UNICAST))
1da177e4
LT		 1992 goto out;
1993
8c14586f
DA		 1994 if (cfg->fc_table)
1995 grt = ip6_nh_lookup_table(net, cfg, gw_addr);
1996
1997 if (!grt)
1998 grt = rt6_lookup(net, gw_addr, NULL,
1999 cfg->fc_ifindex, 1);
1da177e4
LT		 2000
2001 err = -EHOSTUNREACH;
38308473 2002 if (!grt)
1da177e4
LT		 2003 goto out;
2004 if (dev) {
d1918542 2005 if (dev != grt->dst.dev) {
94e187c0 2006 ip6_rt_put(grt);
1da177e4
LT		 2007 goto out;
2008 }
2009 } else {
d1918542 2010 dev = grt->dst.dev;
1da177e4
LT		 2011 idev = grt->rt6i_idev;
2012 dev_hold(dev);
2013 in6_dev_hold(grt->rt6i_idev);
2014 }
38308473 2015 if (!(grt->rt6i_flags & RTF_GATEWAY))
1da177e4 2016 err = 0;
94e187c0 2017 ip6_rt_put(grt);
1da177e4
LT		 2018
2019 if (err)
2020 goto out;
2021 }
2022 err = -EINVAL;
38308473 2023 if (!dev || (dev->flags & IFF_LOOPBACK))
1da177e4
LT		 2024 goto out;
2025 }
2026
2027 err = -ENODEV;
38308473 2028 if (!dev)
1da177e4
LT		 2029 goto out;
2030
c3968a85
DW		 2031 if (!ipv6_addr_any(&cfg->fc_prefsrc)) {
2032 if (!ipv6_chk_addr(net, &cfg->fc_prefsrc, dev, 0)) {
2033 err = -EINVAL;
2034 goto out;
2035 }
4e3fd7a0 2036 rt->rt6i_prefsrc.addr = cfg->fc_prefsrc;
c3968a85
DW		 2037 rt->rt6i_prefsrc.plen = 128;
2038 } else
2039 rt->rt6i_prefsrc.plen = 0;
2040
86872cb5 2041 rt->rt6i_flags = cfg->fc_flags;
1da177e4
LT		 2042
2043install_route:
d8d1f30b 2044 rt->dst.dev = dev;
1da177e4 2045 rt->rt6i_idev = idev;
c71099ac 2046 rt->rt6i_table = table;
63152fc0 2047
c346dca1 2048 cfg->fc_nlinfo.nl_net = dev_net(dev);
63152fc0 2049
8c5b83f0 2050 return rt;
6b9ea5a6
RP		 2051out:
2052 if (dev)
2053 dev_put(dev);
2054 if (idev)
2055 in6_dev_put(idev);
2056 if (rt)
2057 dst_free(&rt->dst);
2058
8c5b83f0 2059 return ERR_PTR(err);
6b9ea5a6
RP		 2060}
2061
2062int ip6_route_add(struct fib6_config *cfg)
2063{
2064 struct mx6_config mxc = { .mx = NULL, };
8c5b83f0 2065 struct rt6_info *rt;
6b9ea5a6
RP		 2066 int err;
2067
8c5b83f0
RP		 2068 rt = ip6_route_info_create(cfg);
2069 if (IS_ERR(rt)) {
2070 err = PTR_ERR(rt);
2071 rt = NULL;
6b9ea5a6 2072 goto out;
8c5b83f0 2073 }
6b9ea5a6 2074
e715b6d3
FW		 2075 err = ip6_convert_metrics(&mxc, cfg);
2076 if (err)
2077 goto out;
1da177e4 2078
e715b6d3
FW		 2079 err = __ip6_ins_rt(rt, &cfg->fc_nlinfo, &mxc);
2080
2081 kfree(mxc.mx);
6b9ea5a6 2082
e715b6d3 2083 return err;
1da177e4 2084out:
1da177e4 2085 if (rt)
d8d1f30b 2086 dst_free(&rt->dst);
6b9ea5a6 2087
1da177e4
LT		 2088 return err;
2089}
2090
86872cb5 2091static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 2092{
2093 int err;
c71099ac 2094 struct fib6_table *table;
d1918542 2095 struct net *net = dev_net(rt->dst.dev);
1da177e4 2096
8e3d5be7
MKL		 2097 if (rt == net->ipv6.ip6_null_entry ||
2098 rt->dst.flags & DST_NOCACHE) {
6825a26c
G		 2099 err = -ENOENT;
2100 goto out;
2101 }
6c813a72 2102
c71099ac
TG		 2103 table = rt->rt6i_table;
2104 write_lock_bh(&table->tb6_lock);
86872cb5 2105 err = fib6_del(rt, info);
c71099ac 2106 write_unlock_bh(&table->tb6_lock);
1da177e4 2107
6825a26c 2108out:
94e187c0 2109 ip6_rt_put(rt);
1da177e4
LT		 2110 return err;
2111}
2112
e0a1ad73
TG		 2113int ip6_del_rt(struct rt6_info *rt)
2114{
4d1169c1 2115 struct nl_info info = {
d1918542 2116 .nl_net = dev_net(rt->dst.dev),
4d1169c1 2117 };
528c4ceb 2118 return __ip6_del_rt(rt, &info);
e0a1ad73
TG		 2119}
2120
86872cb5 2121static int ip6_route_del(struct fib6_config *cfg)
1da177e4 2122{
c71099ac 2123 struct fib6_table *table;
1da177e4
LT		 2124 struct fib6_node *fn;
2125 struct rt6_info *rt;
2126 int err = -ESRCH;
2127
5578689a 2128 table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table);
38308473 2129 if (!table)
c71099ac
TG		 2130 return err;
2131
2132 read_lock_bh(&table->tb6_lock);
1da177e4 2133
c71099ac 2134 fn = fib6_locate(&table->tb6_root,
86872cb5
TG		 2135 &cfg->fc_dst, cfg->fc_dst_len,
2136 &cfg->fc_src, cfg->fc_src_len);
1ab1457c 2137
1da177e4 2138 if (fn) {
d8d1f30b 2139 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1f56a01f
MKL		 2140 if ((rt->rt6i_flags & RTF_CACHE) &&
2141 !(cfg->fc_flags & RTF_CACHE))
2142 continue;
86872cb5 2143 if (cfg->fc_ifindex &&
d1918542
DM		 2144 (!rt->dst.dev ||
2145 rt->dst.dev->ifindex != cfg->fc_ifindex))
1da177e4 2146 continue;
86872cb5
TG		 2147 if (cfg->fc_flags & RTF_GATEWAY &&
2148 !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway))
1da177e4 2149 continue;
86872cb5 2150 if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric)
1da177e4 2151 continue;
d8d1f30b 2152 dst_hold(&rt->dst);
c71099ac 2153 read_unlock_bh(&table->tb6_lock);
1da177e4 2154
86872cb5 2155 return __ip6_del_rt(rt, &cfg->fc_nlinfo);
1da177e4
LT		 2156 }
2157 }
c71099ac 2158 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2159
2160 return err;
2161}
2162
6700c270 2163static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_buff *skb)
a6279458 2164{
a6279458 2165 struct netevent_redirect netevent;
e8599ff4 2166 struct rt6_info *rt, *nrt = NULL;
e8599ff4
DM		 2167 struct ndisc_options ndopts;
2168 struct inet6_dev *in6_dev;
2169 struct neighbour *neigh;
71bcdba0 2170 struct rd_msg *msg;
6e157b6a
DM		 2171 int optlen, on_link;
2172 u8 *lladdr;
e8599ff4 2173
29a3cad5 2174 optlen = skb_tail_pointer(skb) - skb_transport_header(skb);
71bcdba0 2175 optlen -= sizeof(*msg);
e8599ff4
DM		 2176
2177 if (optlen < 0) {
6e157b6a 2178 net_dbg_ratelimited("rt6_do_redirect: packet too short\n");
e8599ff4
DM		 2179 return;
2180 }
2181
71bcdba0 2182 msg = (struct rd_msg *)icmp6_hdr(skb);
e8599ff4 2183
71bcdba0 2184 if (ipv6_addr_is_multicast(&msg->dest)) {
6e157b6a 2185 net_dbg_ratelimited("rt6_do_redirect: destination address is multicast\n");
e8599ff4
DM		 2186 return;
2187 }
2188
6e157b6a 2189 on_link = 0;
71bcdba0 2190 if (ipv6_addr_equal(&msg->dest, &msg->target)) {
e8599ff4 2191 on_link = 1;
71bcdba0 2192 } else if (ipv6_addr_type(&msg->target) !=
e8599ff4 2193 (IPV6_ADDR_UNICAST|IPV6_ADDR_LINKLOCAL)) {
6e157b6a 2194 net_dbg_ratelimited("rt6_do_redirect: target address is not link-local unicast\n");
e8599ff4
DM		 2195 return;
2196 }
2197
2198 in6_dev = __in6_dev_get(skb->dev);
2199 if (!in6_dev)
2200 return;
2201 if (in6_dev->cnf.forwarding || !in6_dev->cnf.accept_redirects)
2202 return;
2203
2204 /* RFC2461 8.1:
2205 * The IP source address of the Redirect MUST be the same as the current
2206 * first-hop router for the specified ICMP Destination Address.
2207 */
2208
f997c55c 2209 if (!ndisc_parse_options(skb->dev, msg->opt, optlen, &ndopts)) {
e8599ff4
DM		 2210 net_dbg_ratelimited("rt6_redirect: invalid ND options\n");
2211 return;
2212 }
6e157b6a
DM		 2213
2214 lladdr = NULL;
e8599ff4
DM		 2215 if (ndopts.nd_opts_tgt_lladdr) {
2216 lladdr = ndisc_opt_addr_data(ndopts.nd_opts_tgt_lladdr,
2217 skb->dev);
2218 if (!lladdr) {
2219 net_dbg_ratelimited("rt6_redirect: invalid link-layer address length\n");
2220 return;
2221 }
2222 }
2223
6e157b6a 2224 rt = (struct rt6_info *) dst;
ec13ad1d 2225 if (rt->rt6i_flags & RTF_REJECT) {
6e157b6a 2226 net_dbg_ratelimited("rt6_redirect: source isn't a valid nexthop for redirect target\n");
e8599ff4 2227 return;
6e157b6a 2228 }
e8599ff4 2229
6e157b6a
DM		 2230 /* Redirect received -> path was valid.
2231 * Look, redirects are sent only in response to data packets,
2232 * so that this nexthop apparently is reachable. --ANK
2233 */
2234 dst_confirm(&rt->dst);
a6279458 2235
71bcdba0 2236 neigh = __neigh_lookup(&nd_tbl, &msg->target, skb->dev, 1);
6e157b6a
DM		 2237 if (!neigh)
2238 return;
a6279458 2239
1da177e4
LT		 2240 /*
2241 * We have finally decided to accept it.
2242 */
2243
f997c55c 2244 ndisc_update(skb->dev, neigh, lladdr, NUD_STALE,
1da177e4
LT		 2245 NEIGH_UPDATE_F_WEAK_OVERRIDE|
2246 NEIGH_UPDATE_F_OVERRIDE|
2247 (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER|
f997c55c
AA		 2248 NEIGH_UPDATE_F_ISROUTER)),
2249 NDISC_REDIRECT, &ndopts);
1da177e4 2250
83a09abd 2251 nrt = ip6_rt_cache_alloc(rt, &msg->dest, NULL);
38308473 2252 if (!nrt)
1da177e4
LT		 2253 goto out;
2254
2255 nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE;
2256 if (on_link)
2257 nrt->rt6i_flags &= ~RTF_GATEWAY;
2258
4e3fd7a0 2259 nrt->rt6i_gateway = *(struct in6_addr *)neigh->primary_key;
1da177e4 2260
40e22e8f 2261 if (ip6_ins_rt(nrt))
1da177e4
LT		 2262 goto out;
2263
d8d1f30b
CG		 2264 netevent.old = &rt->dst;
2265 netevent.new = &nrt->dst;
71bcdba0 2266 netevent.daddr = &msg->dest;
60592833 2267 netevent.neigh = neigh;
8d71740c
TT		 2268 call_netevent_notifiers(NETEVENT_REDIRECT, &netevent);
2269
38308473 2270 if (rt->rt6i_flags & RTF_CACHE) {
6e157b6a 2271 rt = (struct rt6_info *) dst_clone(&rt->dst);
e0a1ad73 2272 ip6_del_rt(rt);
1da177e4
LT		 2273 }
2274
2275out:
e8599ff4 2276 neigh_release(neigh);
6e157b6a
DM		 2277}
2278
1da177e4
LT		 2279/*
2280 * Misc support functions
2281 */
2282
4b32b5ad
MKL		 2283static void rt6_set_from(struct rt6_info *rt, struct rt6_info *from)
2284{
2285 BUG_ON(from->dst.from);
2286
2287 rt->rt6i_flags &= ~RTF_EXPIRES;
2288 dst_hold(&from->dst);
2289 rt->dst.from = &from->dst;
2290 dst_init_metrics(&rt->dst, dst_metrics_ptr(&from->dst), true);
2291}
2292
83a09abd
MKL		 2293static void ip6_rt_copy_init(struct rt6_info *rt, struct rt6_info *ort)
2294{
2295 rt->dst.input = ort->dst.input;
2296 rt->dst.output = ort->dst.output;
2297 rt->rt6i_dst = ort->rt6i_dst;
2298 rt->dst.error = ort->dst.error;
2299 rt->rt6i_idev = ort->rt6i_idev;
2300 if (rt->rt6i_idev)
2301 in6_dev_hold(rt->rt6i_idev);
2302 rt->dst.lastuse = jiffies;
2303 rt->rt6i_gateway = ort->rt6i_gateway;
2304 rt->rt6i_flags = ort->rt6i_flags;
2305 rt6_set_from(rt, ort);
2306 rt->rt6i_metric = ort->rt6i_metric;
1da177e4 2307#ifdef CONFIG_IPV6_SUBTREES
83a09abd 2308 rt->rt6i_src = ort->rt6i_src;
1da177e4 2309#endif
83a09abd
MKL		 2310 rt->rt6i_prefsrc = ort->rt6i_prefsrc;
2311 rt->rt6i_table = ort->rt6i_table;
61adedf3 2312 rt->dst.lwtstate = lwtstate_get(ort->dst.lwtstate);
1da177e4
LT		 2313}
2314
70ceb4f5 2315#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0 2316static struct rt6_info *rt6_get_route_info(struct net *net,
b71d1d42
ED		 2317 const struct in6_addr *prefix, int prefixlen,
2318 const struct in6_addr *gwaddr, int ifindex)
70ceb4f5
YH		 2319{
2320 struct fib6_node *fn;
2321 struct rt6_info *rt = NULL;
c71099ac
TG		 2322 struct fib6_table *table;
2323
efa2cea0 2324 table = fib6_get_table(net, RT6_TABLE_INFO);
38308473 2325 if (!table)
c71099ac 2326 return NULL;
70ceb4f5 2327
5744dd9b 2328 read_lock_bh(&table->tb6_lock);
67ba4152 2329 fn = fib6_locate(&table->tb6_root, prefix, prefixlen, NULL, 0);
70ceb4f5
YH		 2330 if (!fn)
2331 goto out;
2332
d8d1f30b 2333 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
d1918542 2334 if (rt->dst.dev->ifindex != ifindex)
70ceb4f5
YH		 2335 continue;
2336 if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY))
2337 continue;
2338 if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr))
2339 continue;
d8d1f30b 2340 dst_hold(&rt->dst);
70ceb4f5
YH		 2341 break;
2342 }
2343out:
5744dd9b 2344 read_unlock_bh(&table->tb6_lock);
70ceb4f5
YH		 2345 return rt;
2346}
2347
efa2cea0 2348static struct rt6_info *rt6_add_route_info(struct net *net,
b71d1d42
ED		 2349 const struct in6_addr *prefix, int prefixlen,
2350 const struct in6_addr *gwaddr, int ifindex,
95c96174 2351 unsigned int pref)
70ceb4f5 2352{
86872cb5 2353 struct fib6_config cfg = {
238fc7ea 2354 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 2355 .fc_ifindex = ifindex,
2356 .fc_dst_len = prefixlen,
2357 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO |
2358 RTF_UP | RTF_PREF(pref),
15e47304 2359 .fc_nlinfo.portid = 0,
efa2cea0
DL		 2360 .fc_nlinfo.nlh = NULL,
2361 .fc_nlinfo.nl_net = net,
86872cb5
TG		 2362 };
2363
ca254490 2364 cfg.fc_table = l3mdev_fib_table_by_index(net, ifindex) ? : RT6_TABLE_INFO;
4e3fd7a0
AD		 2365 cfg.fc_dst = *prefix;
2366 cfg.fc_gateway = *gwaddr;
70ceb4f5 2367
e317da96
YH		 2368 /* We should treat it as a default route if prefix length is 0. */
2369 if (!prefixlen)
86872cb5 2370 cfg.fc_flags |= RTF_DEFAULT;
70ceb4f5 2371
86872cb5 2372 ip6_route_add(&cfg);
70ceb4f5 2373
efa2cea0 2374 return rt6_get_route_info(net, prefix, prefixlen, gwaddr, ifindex);
70ceb4f5
YH		 2375}
2376#endif
2377
b71d1d42 2378struct rt6_info *rt6_get_dflt_router(const struct in6_addr *addr, struct net_device *dev)
1ab1457c 2379{
1da177e4 2380 struct rt6_info *rt;
c71099ac 2381 struct fib6_table *table;
1da177e4 2382
c346dca1 2383 table = fib6_get_table(dev_net(dev), RT6_TABLE_DFLT);
38308473 2384 if (!table)
c71099ac 2385 return NULL;
1da177e4 2386
5744dd9b 2387 read_lock_bh(&table->tb6_lock);
67ba4152 2388 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
d1918542 2389 if (dev == rt->dst.dev &&
045927ff 2390 ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) &&
1da177e4
LT		 2391 ipv6_addr_equal(&rt->rt6i_gateway, addr))
2392 break;
2393 }
2394 if (rt)
d8d1f30b 2395 dst_hold(&rt->dst);
5744dd9b 2396 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2397 return rt;
2398}
2399
b71d1d42 2400struct rt6_info *rt6_add_dflt_router(const struct in6_addr *gwaddr,
ebacaaa0
YH		 2401 struct net_device *dev,
2402 unsigned int pref)
1da177e4 2403{
86872cb5 2404 struct fib6_config cfg = {
ca254490 2405 .fc_table = l3mdev_fib_table(dev) ? : RT6_TABLE_DFLT,
238fc7ea 2406 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 2407 .fc_ifindex = dev->ifindex,
2408 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT |
2409 RTF_UP | RTF_EXPIRES | RTF_PREF(pref),
15e47304 2410 .fc_nlinfo.portid = 0,
5578689a 2411 .fc_nlinfo.nlh = NULL,
c346dca1 2412 .fc_nlinfo.nl_net = dev_net(dev),
86872cb5 2413 };
1da177e4 2414
4e3fd7a0 2415 cfg.fc_gateway = *gwaddr;
1da177e4 2416
86872cb5 2417 ip6_route_add(&cfg);
1da177e4 2418
1da177e4
LT		 2419 return rt6_get_dflt_router(gwaddr, dev);
2420}
2421
7b4da532 2422void rt6_purge_dflt_routers(struct net *net)
1da177e4
LT		 2423{
2424 struct rt6_info *rt;
c71099ac
TG		 2425 struct fib6_table *table;
2426
2427 /* NOTE: Keep consistent with rt6_get_dflt_router */
7b4da532 2428 table = fib6_get_table(net, RT6_TABLE_DFLT);
38308473 2429 if (!table)
c71099ac 2430 return;
1da177e4
LT		 2431
2432restart:
c71099ac 2433 read_lock_bh(&table->tb6_lock);
d8d1f30b 2434 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
3e8b0ac3
LC		 2435 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF) &&
2436 (!rt->rt6i_idev || rt->rt6i_idev->cnf.accept_ra != 2)) {
d8d1f30b 2437 dst_hold(&rt->dst);
c71099ac 2438 read_unlock_bh(&table->tb6_lock);
e0a1ad73 2439 ip6_del_rt(rt);
1da177e4
LT		 2440 goto restart;
2441 }
2442 }
c71099ac 2443 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 2444}
2445
5578689a
DL		 2446static void rtmsg_to_fib6_config(struct net *net,
2447 struct in6_rtmsg *rtmsg,
86872cb5
TG		 2448 struct fib6_config *cfg)
2449{
2450 memset(cfg, 0, sizeof(*cfg));
2451
ca254490
DA		 2452 cfg->fc_table = l3mdev_fib_table_by_index(net, rtmsg->rtmsg_ifindex) ?
2453 : RT6_TABLE_MAIN;
86872cb5
TG		 2454 cfg->fc_ifindex = rtmsg->rtmsg_ifindex;
2455 cfg->fc_metric = rtmsg->rtmsg_metric;
2456 cfg->fc_expires = rtmsg->rtmsg_info;
2457 cfg->fc_dst_len = rtmsg->rtmsg_dst_len;
2458 cfg->fc_src_len = rtmsg->rtmsg_src_len;
2459 cfg->fc_flags = rtmsg->rtmsg_flags;
2460
5578689a 2461 cfg->fc_nlinfo.nl_net = net;
f1243c2d 2462
4e3fd7a0
AD		 2463 cfg->fc_dst = rtmsg->rtmsg_dst;
2464 cfg->fc_src = rtmsg->rtmsg_src;
2465 cfg->fc_gateway = rtmsg->rtmsg_gateway;
86872cb5
TG		 2466}
2467
5578689a 2468int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg)
1da177e4 2469{
86872cb5 2470 struct fib6_config cfg;
1da177e4
LT		 2471 struct in6_rtmsg rtmsg;
2472 int err;
2473
67ba4152 2474 switch (cmd) {
1da177e4
LT		 2475 case SIOCADDRT: /* Add a route */
2476 case SIOCDELRT: /* Delete a route */
af31f412 2477 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
1da177e4
LT		 2478 return -EPERM;
2479 err = copy_from_user(&rtmsg, arg,
2480 sizeof(struct in6_rtmsg));
2481 if (err)
2482 return -EFAULT;
86872cb5 2483
5578689a 2484 rtmsg_to_fib6_config(net, &rtmsg, &cfg);
86872cb5 2485
1da177e4
LT		 2486 rtnl_lock();
2487 switch (cmd) {
2488 case SIOCADDRT:
86872cb5 2489 err = ip6_route_add(&cfg);
1da177e4
LT		 2490 break;
2491 case SIOCDELRT:
86872cb5 2492 err = ip6_route_del(&cfg);
1da177e4
LT		 2493 break;
2494 default:
2495 err = -EINVAL;
2496 }
2497 rtnl_unlock();
2498
2499 return err;
3ff50b79 2500 }
1da177e4
LT		 2501
2502 return -EINVAL;
2503}
2504
2505/*
2506 * Drop the packet on the floor
2507 */
2508
d5fdd6ba 2509static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes)
1da177e4 2510{
612f09e8 2511 int type;
adf30907 2512 struct dst_entry *dst = skb_dst(skb);
612f09e8
YH		 2513 switch (ipstats_mib_noroutes) {
2514 case IPSTATS_MIB_INNOROUTES:
0660e03f 2515 type = ipv6_addr_type(&ipv6_hdr(skb)->daddr);
45bb0060 2516 if (type == IPV6_ADDR_ANY) {
3bd653c8
DL		 2517 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2518 IPSTATS_MIB_INADDRERRORS);
612f09e8
YH		 2519 break;
2520 }
2521 /* FALLTHROUGH */
2522 case IPSTATS_MIB_OUTNOROUTES:
3bd653c8
DL		 2523 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2524 ipstats_mib_noroutes);
612f09e8
YH		 2525 break;
2526 }
3ffe533c 2527 icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0);
1da177e4
LT		 2528 kfree_skb(skb);
2529 return 0;
2530}
2531
9ce8ade0
TG		 2532static int ip6_pkt_discard(struct sk_buff *skb)
2533{
612f09e8 2534 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2535}
2536
ede2059d 2537static int ip6_pkt_discard_out(struct net *net, struct sock *sk, struct sk_buff *skb)
1da177e4 2538{
adf30907 2539 skb->dev = skb_dst(skb)->dev;
612f09e8 2540 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES);
1da177e4
LT		 2541}
2542
9ce8ade0
TG		 2543static int ip6_pkt_prohibit(struct sk_buff *skb)
2544{
612f09e8 2545 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 2546}
2547
ede2059d 2548static int ip6_pkt_prohibit_out(struct net *net, struct sock *sk, struct sk_buff *skb)
9ce8ade0 2549{
adf30907 2550 skb->dev = skb_dst(skb)->dev;
612f09e8 2551 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES);
9ce8ade0
TG		 2552}
2553
1da177e4
LT		 2554/*
2555 * Allocate a dst for local (unicast / anycast) address.
2556 */
2557
2558struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
2559 const struct in6_addr *addr,
8f031519 2560 bool anycast)
1da177e4 2561{
ca254490 2562 u32 tb_id;
c346dca1 2563 struct net *net = dev_net(idev->dev);
5f02ce24
DA		 2564 struct net_device *dev = net->loopback_dev;
2565 struct rt6_info *rt;
2566
2567 /* use L3 Master device as loopback for host routes if device
2568 * is enslaved and address is not link local or multicast
2569 */
2570 if (!rt6_need_strict(addr))
2571 dev = l3mdev_master_dev_rcu(idev->dev) ? : dev;
2572
2573 rt = ip6_dst_alloc(net, dev, DST_NOCOUNT);
a3300ef4 2574 if (!rt)
1da177e4
LT		 2575 return ERR_PTR(-ENOMEM);
2576
1da177e4
LT		 2577 in6_dev_hold(idev);
2578
11d53b49 2579 rt->dst.flags |= DST_HOST;
d8d1f30b
CG		 2580 rt->dst.input = ip6_input;
2581 rt->dst.output = ip6_output;
1da177e4 2582 rt->rt6i_idev = idev;
1da177e4
LT		 2583
2584 rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP;
58c4fb86
YH		 2585 if (anycast)
2586 rt->rt6i_flags |= RTF_ANYCAST;
2587 else
1da177e4 2588 rt->rt6i_flags |= RTF_LOCAL;
1da177e4 2589
550bab42 2590 rt->rt6i_gateway = *addr;
4e3fd7a0 2591 rt->rt6i_dst.addr = *addr;
1da177e4 2592 rt->rt6i_dst.plen = 128;
ca254490
DA		 2593 tb_id = l3mdev_fib_table(idev->dev) ? : RT6_TABLE_LOCAL;
2594 rt->rt6i_table = fib6_get_table(net, tb_id);
8e3d5be7 2595 rt->dst.flags |= DST_NOCACHE;
1da177e4 2596
d8d1f30b 2597 atomic_set(&rt->dst.__refcnt, 1);
1da177e4
LT		 2598
2599 return rt;
2600}
2601
c3968a85
DW		 2602/* remove deleted ip from prefsrc entries */
2603struct arg_dev_net_ip {
2604 struct net_device *dev;
2605 struct net *net;
2606 struct in6_addr *addr;
2607};
2608
2609static int fib6_remove_prefsrc(struct rt6_info *rt, void *arg)
2610{
2611 struct net_device *dev = ((struct arg_dev_net_ip *)arg)->dev;
2612 struct net *net = ((struct arg_dev_net_ip *)arg)->net;
2613 struct in6_addr *addr = ((struct arg_dev_net_ip *)arg)->addr;
2614
d1918542 2615 if (((void *)rt->dst.dev == dev || !dev) &&
c3968a85
DW		 2616 rt != net->ipv6.ip6_null_entry &&
2617 ipv6_addr_equal(addr, &rt->rt6i_prefsrc.addr)) {
2618 /* remove prefsrc entry */
2619 rt->rt6i_prefsrc.plen = 0;
2620 }
2621 return 0;
2622}
2623
2624void rt6_remove_prefsrc(struct inet6_ifaddr *ifp)
2625{
2626 struct net *net = dev_net(ifp->idev->dev);
2627 struct arg_dev_net_ip adni = {
2628 .dev = ifp->idev->dev,
2629 .net = net,
2630 .addr = &ifp->addr,
2631 };
0c3584d5 2632 fib6_clean_all(net, fib6_remove_prefsrc, &adni);
c3968a85
DW		 2633}
2634
be7a010d
DJ		 2635#define RTF_RA_ROUTER (RTF_ADDRCONF | RTF_DEFAULT | RTF_GATEWAY)
2636#define RTF_CACHE_GATEWAY (RTF_GATEWAY | RTF_CACHE)
2637
2638/* Remove routers and update dst entries when gateway turn into host. */
2639static int fib6_clean_tohost(struct rt6_info *rt, void *arg)
2640{
2641 struct in6_addr *gateway = (struct in6_addr *)arg;
2642
2643 if ((((rt->rt6i_flags & RTF_RA_ROUTER) == RTF_RA_ROUTER) ||
2644 ((rt->rt6i_flags & RTF_CACHE_GATEWAY) == RTF_CACHE_GATEWAY)) &&
2645 ipv6_addr_equal(gateway, &rt->rt6i_gateway)) {
2646 return -1;
2647 }
2648 return 0;
2649}
2650
2651void rt6_clean_tohost(struct net *net, struct in6_addr *gateway)
2652{
2653 fib6_clean_all(net, fib6_clean_tohost, gateway);
2654}
2655
8ed67789
DL		 2656struct arg_dev_net {
2657 struct net_device *dev;
2658 struct net *net;
2659};
2660
1da177e4
LT		 2661static int fib6_ifdown(struct rt6_info *rt, void *arg)
2662{
bc3ef660 2663 const struct arg_dev_net *adn = arg;
2664 const struct net_device *dev = adn->dev;
8ed67789 2665
d1918542 2666 if ((rt->dst.dev == dev || !dev) &&
c159d30c 2667 rt != adn->net->ipv6.ip6_null_entry)
1da177e4 2668 return -1;
c159d30c 2669
1da177e4
LT		 2670 return 0;
2671}
2672
f3db4851 2673void rt6_ifdown(struct net *net, struct net_device *dev)
1da177e4 2674{
8ed67789
DL		 2675 struct arg_dev_net adn = {
2676 .dev = dev,
2677 .net = net,
2678 };
2679
0c3584d5 2680 fib6_clean_all(net, fib6_ifdown, &adn);
1e493d19 2681 icmp6_clean_all(fib6_ifdown, &adn);
e332bc67
EB		 2682 if (dev)
2683 rt6_uncached_list_flush_dev(net, dev);
1da177e4
LT		 2684}
2685
95c96174 2686struct rt6_mtu_change_arg {
1da177e4 2687 struct net_device *dev;
95c96174 2688 unsigned int mtu;
1da177e4
LT		 2689};
2690
2691static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg)
2692{
2693 struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg;
2694 struct inet6_dev *idev;
2695
2696 /* In IPv6 pmtu discovery is not optional,
2697 so that RTAX_MTU lock cannot disable it.
2698 We still use this lock to block changes
2699 caused by addrconf/ndisc.
2700 */
2701
2702 idev = __in6_dev_get(arg->dev);
38308473 2703 if (!idev)
1da177e4
LT		 2704 return 0;
2705
2706 /* For administrative MTU increase, there is no way to discover
2707 IPv6 PMTU increase, so PMTU increase should be updated here.
2708 Since RFC 1981 doesn't include administrative MTU increase
2709 update PMTU increase is a MUST. (i.e. jumbo frame)
2710 */
2711 /*
2712 If new MTU is less than route PMTU, this new MTU will be the
2713 lowest MTU in the path, update the route PMTU to reflect PMTU
2714 decreases; if new MTU is greater than route PMTU, and the
2715 old MTU is the lowest MTU in the path, update the route PMTU
2716 to reflect the increase. In this case if the other nodes' MTU
2717 also have the lowest MTU, TOO BIG MESSAGE will be lead to
2718 PMTU discouvery.
2719 */
d1918542 2720 if (rt->dst.dev == arg->dev &&
4b32b5ad
MKL		 2721 !dst_metric_locked(&rt->dst, RTAX_MTU)) {
2722 if (rt->rt6i_flags & RTF_CACHE) {
2723 /* For RTF_CACHE with rt6i_pmtu == 0
2724 * (i.e. a redirected route),
2725 * the metrics of its rt->dst.from has already
2726 * been updated.
2727 */
2728 if (rt->rt6i_pmtu && rt->rt6i_pmtu > arg->mtu)
2729 rt->rt6i_pmtu = arg->mtu;
2730 } else if (dst_mtu(&rt->dst) >= arg->mtu ||
2731 (dst_mtu(&rt->dst) < arg->mtu &&
2732 dst_mtu(&rt->dst) == idev->cnf.mtu6)) {
2733 dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu);
2734 }
566cfd8f 2735 }
1da177e4
LT		 2736 return 0;
2737}
2738
95c96174 2739void rt6_mtu_change(struct net_device *dev, unsigned int mtu)
1da177e4 2740{
c71099ac
TG		 2741 struct rt6_mtu_change_arg arg = {
2742 .dev = dev,
2743 .mtu = mtu,
2744 };
1da177e4 2745
0c3584d5 2746 fib6_clean_all(dev_net(dev), rt6_mtu_change_route, &arg);
1da177e4
LT		 2747}
2748
ef7c79ed 2749static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = {
5176f91e 2750 [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) },
86872cb5 2751 [RTA_OIF] = { .type = NLA_U32 },
ab364a6f 2752 [RTA_IIF] = { .type = NLA_U32 },
86872cb5
TG		 2753 [RTA_PRIORITY] = { .type = NLA_U32 },
2754 [RTA_METRICS] = { .type = NLA_NESTED },
51ebd318 2755 [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) },
c78ba6d6 2756 [RTA_PREF] = { .type = NLA_U8 },
19e42e45
RP		 2757 [RTA_ENCAP_TYPE] = { .type = NLA_U16 },
2758 [RTA_ENCAP] = { .type = NLA_NESTED },
32bc201e 2759 [RTA_EXPIRES] = { .type = NLA_U32 },
86872cb5
TG		 2760};
2761
2762static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh,
2763 struct fib6_config *cfg)
1da177e4 2764{
86872cb5
TG		 2765 struct rtmsg *rtm;
2766 struct nlattr *tb[RTA_MAX+1];
c78ba6d6 2767 unsigned int pref;
86872cb5 2768 int err;
1da177e4 2769
86872cb5
TG		 2770 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2771 if (err < 0)
2772 goto errout;
1da177e4 2773
86872cb5
TG		 2774 err = -EINVAL;
2775 rtm = nlmsg_data(nlh);
2776 memset(cfg, 0, sizeof(*cfg));
2777
2778 cfg->fc_table = rtm->rtm_table;
2779 cfg->fc_dst_len = rtm->rtm_dst_len;
2780 cfg->fc_src_len = rtm->rtm_src_len;
2781 cfg->fc_flags = RTF_UP;
2782 cfg->fc_protocol = rtm->rtm_protocol;
ef2c7d7b 2783 cfg->fc_type = rtm->rtm_type;
86872cb5 2784
ef2c7d7b
ND		 2785 if (rtm->rtm_type == RTN_UNREACHABLE ||
2786 rtm->rtm_type == RTN_BLACKHOLE ||
b4949ab2
ND		 2787 rtm->rtm_type == RTN_PROHIBIT ||
2788 rtm->rtm_type == RTN_THROW)
86872cb5
TG		 2789 cfg->fc_flags |= RTF_REJECT;
2790
ab79ad14
MŻ		 2791 if (rtm->rtm_type == RTN_LOCAL)
2792 cfg->fc_flags |= RTF_LOCAL;
2793
1f56a01f
MKL		 2794 if (rtm->rtm_flags & RTM_F_CLONED)
2795 cfg->fc_flags |= RTF_CACHE;
2796
15e47304 2797 cfg->fc_nlinfo.portid = NETLINK_CB(skb).portid;
86872cb5 2798 cfg->fc_nlinfo.nlh = nlh;
3b1e0a65 2799 cfg->fc_nlinfo.nl_net = sock_net(skb->sk);
86872cb5
TG		 2800
2801 if (tb[RTA_GATEWAY]) {
67b61f6c 2802 cfg->fc_gateway = nla_get_in6_addr(tb[RTA_GATEWAY]);
86872cb5 2803 cfg->fc_flags |= RTF_GATEWAY;
1da177e4 2804 }
86872cb5
TG		 2805
2806 if (tb[RTA_DST]) {
2807 int plen = (rtm->rtm_dst_len + 7) >> 3;
2808
2809 if (nla_len(tb[RTA_DST]) < plen)
2810 goto errout;
2811
2812 nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen);
1da177e4 2813 }
86872cb5
TG		 2814
2815 if (tb[RTA_SRC]) {
2816 int plen = (rtm->rtm_src_len + 7) >> 3;
2817
2818 if (nla_len(tb[RTA_SRC]) < plen)
2819 goto errout;
2820
2821 nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen);
1da177e4 2822 }
86872cb5 2823
c3968a85 2824 if (tb[RTA_PREFSRC])
67b61f6c 2825 cfg->fc_prefsrc = nla_get_in6_addr(tb[RTA_PREFSRC]);
c3968a85 2826
86872cb5
TG		 2827 if (tb[RTA_OIF])
2828 cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]);
2829
2830 if (tb[RTA_PRIORITY])
2831 cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]);
2832
2833 if (tb[RTA_METRICS]) {
2834 cfg->fc_mx = nla_data(tb[RTA_METRICS]);
2835 cfg->fc_mx_len = nla_len(tb[RTA_METRICS]);
1da177e4 2836 }
86872cb5
TG		 2837
2838 if (tb[RTA_TABLE])
2839 cfg->fc_table = nla_get_u32(tb[RTA_TABLE]);
2840
51ebd318
ND		 2841 if (tb[RTA_MULTIPATH]) {
2842 cfg->fc_mp = nla_data(tb[RTA_MULTIPATH]);
2843 cfg->fc_mp_len = nla_len(tb[RTA_MULTIPATH]);
2844 }
2845
c78ba6d6
LR		 2846 if (tb[RTA_PREF]) {
2847 pref = nla_get_u8(tb[RTA_PREF]);
2848 if (pref != ICMPV6_ROUTER_PREF_LOW &&
2849 pref != ICMPV6_ROUTER_PREF_HIGH)
2850 pref = ICMPV6_ROUTER_PREF_MEDIUM;
2851 cfg->fc_flags |= RTF_PREF(pref);
2852 }
2853
19e42e45
RP		 2854 if (tb[RTA_ENCAP])
2855 cfg->fc_encap = tb[RTA_ENCAP];
2856
2857 if (tb[RTA_ENCAP_TYPE])
2858 cfg->fc_encap_type = nla_get_u16(tb[RTA_ENCAP_TYPE]);
2859
32bc201e
XL		 2860 if (tb[RTA_EXPIRES]) {
2861 unsigned long timeout = addrconf_timeout_fixup(nla_get_u32(tb[RTA_EXPIRES]), HZ);
2862
2863 if (addrconf_finite_timeout(timeout)) {
2864 cfg->fc_expires = jiffies_to_clock_t(timeout * HZ);
2865 cfg->fc_flags |= RTF_EXPIRES;
2866 }
2867 }
2868
86872cb5
TG		 2869 err = 0;
2870errout:
2871 return err;
1da177e4
LT		 2872}
2873
6b9ea5a6
RP		 2874struct rt6_nh {
2875 struct rt6_info *rt6_info;
2876 struct fib6_config r_cfg;
2877 struct mx6_config mxc;
2878 struct list_head next;
2879};
2880
2881static void ip6_print_replace_route_err(struct list_head *rt6_nh_list)
2882{
2883 struct rt6_nh *nh;
2884
2885 list_for_each_entry(nh, rt6_nh_list, next) {
2886 pr_warn("IPV6: multipath route replace failed (check consistency of installed routes): %pI6 nexthop %pI6 ifi %d\n",
2887 &nh->r_cfg.fc_dst, &nh->r_cfg.fc_gateway,
2888 nh->r_cfg.fc_ifindex);
2889 }
2890}
2891
2892static int ip6_route_info_append(struct list_head *rt6_nh_list,
2893 struct rt6_info *rt, struct fib6_config *r_cfg)
2894{
2895 struct rt6_nh *nh;
2896 struct rt6_info *rtnh;
2897 int err = -EEXIST;
2898
2899 list_for_each_entry(nh, rt6_nh_list, next) {
2900 /* check if rt6_info already exists */
2901 rtnh = nh->rt6_info;
2902
2903 if (rtnh->dst.dev == rt->dst.dev &&
2904 rtnh->rt6i_idev == rt->rt6i_idev &&
2905 ipv6_addr_equal(&rtnh->rt6i_gateway,
2906 &rt->rt6i_gateway))
2907 return err;
2908 }
2909
2910 nh = kzalloc(sizeof(*nh), GFP_KERNEL);
2911 if (!nh)
2912 return -ENOMEM;
2913 nh->rt6_info = rt;
2914 err = ip6_convert_metrics(&nh->mxc, r_cfg);
2915 if (err) {
2916 kfree(nh);
2917 return err;
2918 }
2919 memcpy(&nh->r_cfg, r_cfg, sizeof(*r_cfg));
2920 list_add_tail(&nh->next, rt6_nh_list);
2921
2922 return 0;
2923}
2924
2925static int ip6_route_multipath_add(struct fib6_config *cfg)
51ebd318
ND		 2926{
2927 struct fib6_config r_cfg;
2928 struct rtnexthop *rtnh;
6b9ea5a6
RP		 2929 struct rt6_info *rt;
2930 struct rt6_nh *err_nh;
2931 struct rt6_nh *nh, *nh_safe;
51ebd318
ND		 2932 int remaining;
2933 int attrlen;
6b9ea5a6
RP		 2934 int err = 1;
2935 int nhn = 0;
2936 int replace = (cfg->fc_nlinfo.nlh &&
2937 (cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_REPLACE));
2938 LIST_HEAD(rt6_nh_list);
51ebd318 2939
35f1b4e9 2940 remaining = cfg->fc_mp_len;
51ebd318 2941 rtnh = (struct rtnexthop *)cfg->fc_mp;
51ebd318 2942
6b9ea5a6
RP		 2943 /* Parse a Multipath Entry and build a list (rt6_nh_list) of
2944 * rt6_info structs per nexthop
2945 */
51ebd318
ND		 2946 while (rtnh_ok(rtnh, remaining)) {
2947 memcpy(&r_cfg, cfg, sizeof(*cfg));
2948 if (rtnh->rtnh_ifindex)
2949 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
2950
2951 attrlen = rtnh_attrlen(rtnh);
2952 if (attrlen > 0) {
2953 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
2954
2955 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
2956 if (nla) {
67b61f6c 2957 r_cfg.fc_gateway = nla_get_in6_addr(nla);
51ebd318
ND		 2958 r_cfg.fc_flags |= RTF_GATEWAY;
2959 }
19e42e45
RP		 2960 r_cfg.fc_encap = nla_find(attrs, attrlen, RTA_ENCAP);
2961 nla = nla_find(attrs, attrlen, RTA_ENCAP_TYPE);
2962 if (nla)
2963 r_cfg.fc_encap_type = nla_get_u16(nla);
51ebd318 2964 }
6b9ea5a6 2965
8c5b83f0
RP		 2966 rt = ip6_route_info_create(&r_cfg);
2967 if (IS_ERR(rt)) {
2968 err = PTR_ERR(rt);
2969 rt = NULL;
6b9ea5a6 2970 goto cleanup;
8c5b83f0 2971 }
6b9ea5a6
RP		 2972
2973 err = ip6_route_info_append(&rt6_nh_list, rt, &r_cfg);
51ebd318 2974 if (err) {
6b9ea5a6
RP		 2975 dst_free(&rt->dst);
2976 goto cleanup;
2977 }
2978
2979 rtnh = rtnh_next(rtnh, &remaining);
2980 }
2981
2982 err_nh = NULL;
2983 list_for_each_entry(nh, &rt6_nh_list, next) {
2984 err = __ip6_ins_rt(nh->rt6_info, &cfg->fc_nlinfo, &nh->mxc);
2985 /* nh->rt6_info is used or freed at this point, reset to NULL*/
2986 nh->rt6_info = NULL;
2987 if (err) {
2988 if (replace && nhn)
2989 ip6_print_replace_route_err(&rt6_nh_list);
2990 err_nh = nh;
2991 goto add_errout;
51ebd318 2992 }
6b9ea5a6 2993
1a72418b 2994 /* Because each route is added like a single route we remove
27596472
MK		 2995 * these flags after the first nexthop: if there is a collision,
2996 * we have already failed to add the first nexthop:
2997 * fib6_add_rt2node() has rejected it; when replacing, old
2998 * nexthops have been replaced by first new, the rest should
2999 * be added to it.
1a72418b 3000 */
27596472
MK		 3001 cfg->fc_nlinfo.nlh->nlmsg_flags &= ~(NLM_F_EXCL |
3002 NLM_F_REPLACE);
6b9ea5a6
RP		 3003 nhn++;
3004 }
3005
3006 goto cleanup;
3007
3008add_errout:
3009 /* Delete routes that were already added */
3010 list_for_each_entry(nh, &rt6_nh_list, next) {
3011 if (err_nh == nh)
3012 break;
3013 ip6_route_del(&nh->r_cfg);
3014 }
3015
3016cleanup:
3017 list_for_each_entry_safe(nh, nh_safe, &rt6_nh_list, next) {
3018 if (nh->rt6_info)
3019 dst_free(&nh->rt6_info->dst);
52fe51f8 3020 kfree(nh->mxc.mx);
6b9ea5a6
RP		 3021 list_del(&nh->next);
3022 kfree(nh);
3023 }
3024
3025 return err;
3026}
3027
3028static int ip6_route_multipath_del(struct fib6_config *cfg)
3029{
3030 struct fib6_config r_cfg;
3031 struct rtnexthop *rtnh;
3032 int remaining;
3033 int attrlen;
3034 int err = 1, last_err = 0;
3035
3036 remaining = cfg->fc_mp_len;
3037 rtnh = (struct rtnexthop *)cfg->fc_mp;
3038
3039 /* Parse a Multipath Entry */
3040 while (rtnh_ok(rtnh, remaining)) {
3041 memcpy(&r_cfg, cfg, sizeof(*cfg));
3042 if (rtnh->rtnh_ifindex)
3043 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
3044
3045 attrlen = rtnh_attrlen(rtnh);
3046 if (attrlen > 0) {
3047 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
3048
3049 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
3050 if (nla) {
3051 nla_memcpy(&r_cfg.fc_gateway, nla, 16);
3052 r_cfg.fc_flags |= RTF_GATEWAY;
3053 }
3054 }
3055 err = ip6_route_del(&r_cfg);
3056 if (err)
3057 last_err = err;
3058
51ebd318
ND		 3059 rtnh = rtnh_next(rtnh, &remaining);
3060 }
3061
3062 return last_err;
3063}
3064
67ba4152 3065static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr *nlh)
1da177e4 3066{
86872cb5
TG		 3067 struct fib6_config cfg;
3068 int err;
1da177e4 3069
86872cb5
TG		 3070 err = rtm_to_fib6_config(skb, nlh, &cfg);
3071 if (err < 0)
3072 return err;
3073
51ebd318 3074 if (cfg.fc_mp)
6b9ea5a6 3075 return ip6_route_multipath_del(&cfg);
51ebd318
ND		 3076 else
3077 return ip6_route_del(&cfg);
1da177e4
LT		 3078}
3079
67ba4152 3080static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr *nlh)
1da177e4 3081{
86872cb5
TG		 3082 struct fib6_config cfg;
3083 int err;
1da177e4 3084
86872cb5
TG		 3085 err = rtm_to_fib6_config(skb, nlh, &cfg);
3086 if (err < 0)
3087 return err;
3088
51ebd318 3089 if (cfg.fc_mp)
6b9ea5a6 3090 return ip6_route_multipath_add(&cfg);
51ebd318
ND		 3091 else
3092 return ip6_route_add(&cfg);
1da177e4
LT		 3093}
3094
19e42e45 3095static inline size_t rt6_nlmsg_size(struct rt6_info *rt)
339bf98f
TG		 3096{
3097 return NLMSG_ALIGN(sizeof(struct rtmsg))
3098 + nla_total_size(16) /* RTA_SRC */
3099 + nla_total_size(16) /* RTA_DST */
3100 + nla_total_size(16) /* RTA_GATEWAY */
3101 + nla_total_size(16) /* RTA_PREFSRC */
3102 + nla_total_size(4) /* RTA_TABLE */
3103 + nla_total_size(4) /* RTA_IIF */
3104 + nla_total_size(4) /* RTA_OIF */
3105 + nla_total_size(4) /* RTA_PRIORITY */
6a2b9ce0 3106 + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */
ea697639 3107 + nla_total_size(sizeof(struct rta_cacheinfo))
c78ba6d6 3108 + nla_total_size(TCP_CA_NAME_MAX) /* RTAX_CC_ALGO */
19e42e45 3109 + nla_total_size(1) /* RTA_PREF */
61adedf3 3110 + lwtunnel_get_encap_size(rt->dst.lwtstate);
339bf98f
TG		 3111}
3112
191cd582
BH		 3113static int rt6_fill_node(struct net *net,
3114 struct sk_buff *skb, struct rt6_info *rt,
0d51aa80 3115 struct in6_addr *dst, struct in6_addr *src,
15e47304 3116 int iif, int type, u32 portid, u32 seq,
7bc570c8 3117 int prefix, int nowait, unsigned int flags)
1da177e4 3118{
4b32b5ad 3119 u32 metrics[RTAX_MAX];
1da177e4 3120 struct rtmsg *rtm;
2d7202bf 3121 struct nlmsghdr *nlh;
e3703b3d 3122 long expires;
9e762a4a 3123 u32 table;
1da177e4
LT		 3124
3125 if (prefix) { /* user wants prefix routes only */
3126 if (!(rt->rt6i_flags & RTF_PREFIX_RT)) {
3127 /* success since this is not a prefix route */
3128 return 1;
3129 }
3130 }
3131
15e47304 3132 nlh = nlmsg_put(skb, portid, seq, type, sizeof(*rtm), flags);
38308473 3133 if (!nlh)
26932566 3134 return -EMSGSIZE;
2d7202bf
TG		 3135
3136 rtm = nlmsg_data(nlh);
1da177e4
LT		 3137 rtm->rtm_family = AF_INET6;
3138 rtm->rtm_dst_len = rt->rt6i_dst.plen;
3139 rtm->rtm_src_len = rt->rt6i_src.plen;
3140 rtm->rtm_tos = 0;
c71099ac 3141 if (rt->rt6i_table)
9e762a4a 3142 table = rt->rt6i_table->tb6_id;
c71099ac 3143 else
9e762a4a
PM		 3144 table = RT6_TABLE_UNSPEC;
3145 rtm->rtm_table = table;
c78679e8
DM		 3146 if (nla_put_u32(skb, RTA_TABLE, table))
3147 goto nla_put_failure;
ef2c7d7b
ND		 3148 if (rt->rt6i_flags & RTF_REJECT) {
3149 switch (rt->dst.error) {
3150 case -EINVAL:
3151 rtm->rtm_type = RTN_BLACKHOLE;
3152 break;
3153 case -EACCES:
3154 rtm->rtm_type = RTN_PROHIBIT;
3155 break;
b4949ab2
ND		 3156 case -EAGAIN:
3157 rtm->rtm_type = RTN_THROW;
3158 break;
ef2c7d7b
ND		 3159 default:
3160 rtm->rtm_type = RTN_UNREACHABLE;
3161 break;
3162 }
3163 }
38308473 3164 else if (rt->rt6i_flags & RTF_LOCAL)
ab79ad14 3165 rtm->rtm_type = RTN_LOCAL;
d1918542 3166 else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK))
1da177e4
LT		 3167 rtm->rtm_type = RTN_LOCAL;
3168 else
3169 rtm->rtm_type = RTN_UNICAST;
3170 rtm->rtm_flags = 0;
35103d11 3171 if (!netif_carrier_ok(rt->dst.dev)) {
cea45e20 3172 rtm->rtm_flags |= RTNH_F_LINKDOWN;
35103d11
AG		 3173 if (rt->rt6i_idev->cnf.ignore_routes_with_linkdown)
3174 rtm->rtm_flags |= RTNH_F_DEAD;
3175 }
1da177e4
LT		 3176 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
3177 rtm->rtm_protocol = rt->rt6i_protocol;
38308473 3178 if (rt->rt6i_flags & RTF_DYNAMIC)
1da177e4 3179 rtm->rtm_protocol = RTPROT_REDIRECT;
f0396f60
DO		 3180 else if (rt->rt6i_flags & RTF_ADDRCONF) {
3181 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ROUTEINFO))
3182 rtm->rtm_protocol = RTPROT_RA;
3183 else
3184 rtm->rtm_protocol = RTPROT_KERNEL;
3185 }
1da177e4 3186
38308473 3187 if (rt->rt6i_flags & RTF_CACHE)
1da177e4
LT		 3188 rtm->rtm_flags |= RTM_F_CLONED;
3189
3190 if (dst) {
930345ea 3191 if (nla_put_in6_addr(skb, RTA_DST, dst))
c78679e8 3192 goto nla_put_failure;
1ab1457c 3193 rtm->rtm_dst_len = 128;
1da177e4 3194 } else if (rtm->rtm_dst_len)
930345ea 3195 if (nla_put_in6_addr(skb, RTA_DST, &rt->rt6i_dst.addr))
c78679e8 3196 goto nla_put_failure;
1da177e4
LT		 3197#ifdef CONFIG_IPV6_SUBTREES
3198 if (src) {
930345ea 3199 if (nla_put_in6_addr(skb, RTA_SRC, src))
c78679e8 3200 goto nla_put_failure;
1ab1457c 3201 rtm->rtm_src_len = 128;
c78679e8 3202 } else if (rtm->rtm_src_len &&
930345ea 3203 nla_put_in6_addr(skb, RTA_SRC, &rt->rt6i_src.addr))
c78679e8 3204 goto nla_put_failure;
1da177e4 3205#endif
7bc570c8
YH		 3206 if (iif) {
3207#ifdef CONFIG_IPV6_MROUTE
3208 if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) {
8229efda 3209 int err = ip6mr_get_route(net, skb, rtm, nowait);
7bc570c8
YH		 3210 if (err <= 0) {
3211 if (!nowait) {
3212 if (err == 0)
3213 return 0;
3214 goto nla_put_failure;
3215 } else {
3216 if (err == -EMSGSIZE)
3217 goto nla_put_failure;
3218 }
3219 }
3220 } else
3221#endif
c78679e8
DM		 3222 if (nla_put_u32(skb, RTA_IIF, iif))
3223 goto nla_put_failure;
7bc570c8 3224 } else if (dst) {
1da177e4 3225 struct in6_addr saddr_buf;
c78679e8 3226 if (ip6_route_get_saddr(net, rt, dst, 0, &saddr_buf) == 0 &&
930345ea 3227 nla_put_in6_addr(skb, RTA_PREFSRC, &saddr_buf))
c78679e8 3228 goto nla_put_failure;
1da177e4 3229 }
2d7202bf 3230
c3968a85
DW		 3231 if (rt->rt6i_prefsrc.plen) {
3232 struct in6_addr saddr_buf;
4e3fd7a0 3233 saddr_buf = rt->rt6i_prefsrc.addr;
930345ea 3234 if (nla_put_in6_addr(skb, RTA_PREFSRC, &saddr_buf))
c78679e8 3235 goto nla_put_failure;
c3968a85
DW		 3236 }
3237
4b32b5ad
MKL		 3238 memcpy(metrics, dst_metrics_ptr(&rt->dst), sizeof(metrics));
3239 if (rt->rt6i_pmtu)
3240 metrics[RTAX_MTU - 1] = rt->rt6i_pmtu;
3241 if (rtnetlink_put_metrics(skb, metrics) < 0)
2d7202bf
TG		 3242 goto nla_put_failure;
3243
dd0cbf29 3244 if (rt->rt6i_flags & RTF_GATEWAY) {
930345ea 3245 if (nla_put_in6_addr(skb, RTA_GATEWAY, &rt->rt6i_gateway) < 0)
94f826b8 3246 goto nla_put_failure;
94f826b8 3247 }
2d7202bf 3248
c78679e8
DM		 3249 if (rt->dst.dev &&
3250 nla_put_u32(skb, RTA_OIF, rt->dst.dev->ifindex))
3251 goto nla_put_failure;
3252 if (nla_put_u32(skb, RTA_PRIORITY, rt->rt6i_metric))
3253 goto nla_put_failure;
8253947e
LW		 3254
3255 expires = (rt->rt6i_flags & RTF_EXPIRES) ? rt->dst.expires - jiffies : 0;
69cdf8f9 3256
87a50699 3257 if (rtnl_put_cacheinfo(skb, &rt->dst, 0, expires, rt->dst.error) < 0)
e3703b3d 3258 goto nla_put_failure;
2d7202bf 3259
c78ba6d6
LR		 3260 if (nla_put_u8(skb, RTA_PREF, IPV6_EXTRACT_PREF(rt->rt6i_flags)))
3261 goto nla_put_failure;
3262
61adedf3 3263 lwtunnel_fill_encap(skb, rt->dst.lwtstate);
19e42e45 3264
053c095a
JB		 3265 nlmsg_end(skb, nlh);
3266 return 0;
2d7202bf
TG		 3267
3268nla_put_failure:
26932566
PM		 3269 nlmsg_cancel(skb, nlh);
3270 return -EMSGSIZE;
1da177e4
LT		 3271}
3272
1b43af54 3273int rt6_dump_route(struct rt6_info *rt, void *p_arg)
1da177e4
LT		 3274{
3275 struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg;
3276 int prefix;
3277
2d7202bf
TG		 3278 if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) {
3279 struct rtmsg *rtm = nlmsg_data(arg->cb->nlh);
1da177e4
LT		 3280 prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0;
3281 } else
3282 prefix = 0;
3283
191cd582
BH		 3284 return rt6_fill_node(arg->net,
3285 arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE,
15e47304 3286 NETLINK_CB(arg->cb->skb).portid, arg->cb->nlh->nlmsg_seq,
7bc570c8 3287 prefix, 0, NLM_F_MULTI);
1da177e4
LT		 3288}
3289
67ba4152 3290static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh)
1da177e4 3291{
3b1e0a65 3292 struct net *net = sock_net(in_skb->sk);
ab364a6f
TG		 3293 struct nlattr *tb[RTA_MAX+1];
3294 struct rt6_info *rt;
1da177e4 3295 struct sk_buff *skb;
ab364a6f 3296 struct rtmsg *rtm;
4c9483b2 3297 struct flowi6 fl6;
72331bc0 3298 int err, iif = 0, oif = 0;
1da177e4 3299
ab364a6f
TG		 3300 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
3301 if (err < 0)
3302 goto errout;
1da177e4 3303
ab364a6f 3304 err = -EINVAL;
4c9483b2 3305 memset(&fl6, 0, sizeof(fl6));
38b7097b
HFS		 3306 rtm = nlmsg_data(nlh);
3307 fl6.flowlabel = ip6_make_flowinfo(rtm->rtm_tos, 0);
1da177e4 3308
ab364a6f
TG		 3309 if (tb[RTA_SRC]) {
3310 if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr))
3311 goto errout;
3312
4e3fd7a0 3313 fl6.saddr = *(struct in6_addr *)nla_data(tb[RTA_SRC]);
ab364a6f
TG		 3314 }
3315
3316 if (tb[RTA_DST]) {
3317 if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr))
3318 goto errout;
3319
4e3fd7a0 3320 fl6.daddr = *(struct in6_addr *)nla_data(tb[RTA_DST]);
ab364a6f
TG		 3321 }
3322
3323 if (tb[RTA_IIF])
3324 iif = nla_get_u32(tb[RTA_IIF]);
3325
3326 if (tb[RTA_OIF])
72331bc0 3327 oif = nla_get_u32(tb[RTA_OIF]);
1da177e4 3328
2e47b291
LC		 3329 if (tb[RTA_MARK])
3330 fl6.flowi6_mark = nla_get_u32(tb[RTA_MARK]);
3331
1da177e4
LT		 3332 if (iif) {
3333 struct net_device *dev;
72331bc0
SL		 3334 int flags = 0;
3335
5578689a 3336 dev = __dev_get_by_index(net, iif);
1da177e4
LT		 3337 if (!dev) {
3338 err = -ENODEV;
ab364a6f 3339 goto errout;
1da177e4 3340 }
72331bc0
SL		 3341
3342 fl6.flowi6_iif = iif;
3343
3344 if (!ipv6_addr_any(&fl6.saddr))
3345 flags |= RT6_LOOKUP_F_HAS_SADDR;
3346
3347 rt = (struct rt6_info *)ip6_route_input_lookup(net, dev, &fl6,
3348 flags);
3349 } else {
3350 fl6.flowi6_oif = oif;
3351
ca254490
DA		 3352 if (netif_index_is_l3_master(net, oif)) {
3353 fl6.flowi6_flags = FLOWI_FLAG_L3MDEV_SRC |
3354 FLOWI_FLAG_SKIP_NH_OIF;
3355 }
3356
72331bc0 3357 rt = (struct rt6_info *)ip6_route_output(net, NULL, &fl6);
1da177e4
LT		 3358 }
3359
ab364a6f 3360 skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
38308473 3361 if (!skb) {
94e187c0 3362 ip6_rt_put(rt);
ab364a6f
TG		 3363 err = -ENOBUFS;
3364 goto errout;
3365 }
1da177e4 3366
ab364a6f
TG		 3367 /* Reserve room for dummy headers, this skb can pass
3368 through good chunk of routing engine.
3369 */
459a98ed 3370 skb_reset_mac_header(skb);
ab364a6f 3371 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr));
1da177e4 3372
d8d1f30b 3373 skb_dst_set(skb, &rt->dst);
1da177e4 3374
4c9483b2 3375 err = rt6_fill_node(net, skb, rt, &fl6.daddr, &fl6.saddr, iif,
15e47304 3376 RTM_NEWROUTE, NETLINK_CB(in_skb).portid,
7bc570c8 3377 nlh->nlmsg_seq, 0, 0, 0);
1da177e4 3378 if (err < 0) {
ab364a6f
TG		 3379 kfree_skb(skb);
3380 goto errout;
1da177e4
LT		 3381 }
3382
15e47304 3383 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
ab364a6f 3384errout:
1da177e4 3385 return err;
1da177e4
LT		 3386}
3387
37a1d361
RP		 3388void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info,
3389 unsigned int nlm_flags)
1da177e4
LT		 3390{
3391 struct sk_buff *skb;
5578689a 3392 struct net *net = info->nl_net;
528c4ceb
DL		 3393 u32 seq;
3394 int err;
3395
3396 err = -ENOBUFS;
38308473 3397 seq = info->nlh ? info->nlh->nlmsg_seq : 0;
86872cb5 3398
19e42e45 3399 skb = nlmsg_new(rt6_nlmsg_size(rt), gfp_any());
38308473 3400 if (!skb)
21713ebc
TG		 3401 goto errout;
3402
191cd582 3403 err = rt6_fill_node(net, skb, rt, NULL, NULL, 0,
37a1d361 3404 event, info->portid, seq, 0, 0, nlm_flags);
26932566
PM		 3405 if (err < 0) {
3406 /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */
3407 WARN_ON(err == -EMSGSIZE);
3408 kfree_skb(skb);
3409 goto errout;
3410 }
15e47304 3411 rtnl_notify(skb, net, info->portid, RTNLGRP_IPV6_ROUTE,
1ce85fe4
PNA		 3412 info->nlh, gfp_any());
3413 return;
21713ebc
TG		 3414errout:
3415 if (err < 0)
5578689a 3416 rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err);
1da177e4
LT		 3417}
3418
8ed67789 3419static int ip6_route_dev_notify(struct notifier_block *this,
351638e7 3420 unsigned long event, void *ptr)
8ed67789 3421{
351638e7 3422 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
c346dca1 3423 struct net *net = dev_net(dev);
8ed67789
DL		 3424
3425 if (event == NETDEV_REGISTER && (dev->flags & IFF_LOOPBACK)) {
d8d1f30b 3426 net->ipv6.ip6_null_entry->dst.dev = dev;
8ed67789
DL		 3427 net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev);
3428#ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 3429 net->ipv6.ip6_prohibit_entry->dst.dev = dev;
8ed67789 3430 net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev);
d8d1f30b 3431 net->ipv6.ip6_blk_hole_entry->dst.dev = dev;
8ed67789
DL		 3432 net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev);
3433#endif
3434 }
3435
3436 return NOTIFY_OK;
3437}
3438
1da177e4
LT		 3439/*
3440 * /proc
3441 */
3442
3443#ifdef CONFIG_PROC_FS
3444
33120b30
AD		 3445static const struct file_operations ipv6_route_proc_fops = {
3446 .owner = THIS_MODULE,
3447 .open = ipv6_route_open,
3448 .read = seq_read,
3449 .llseek = seq_lseek,
8d2ca1d7 3450 .release = seq_release_net,
33120b30
AD		 3451};
3452
1da177e4
LT		 3453static int rt6_stats_seq_show(struct seq_file *seq, void *v)
3454{
69ddb805 3455 struct net *net = (struct net *)seq->private;
1da177e4 3456 seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n",
69ddb805
DL		 3457 net->ipv6.rt6_stats->fib_nodes,
3458 net->ipv6.rt6_stats->fib_route_nodes,
3459 net->ipv6.rt6_stats->fib_rt_alloc,
3460 net->ipv6.rt6_stats->fib_rt_entries,
3461 net->ipv6.rt6_stats->fib_rt_cache,
fc66f95c 3462 dst_entries_get_slow(&net->ipv6.ip6_dst_ops),
69ddb805 3463 net->ipv6.rt6_stats->fib_discarded_routes);
1da177e4
LT		 3464
3465 return 0;
3466}
3467
3468static int rt6_stats_seq_open(struct inode *inode, struct file *file)
3469{
de05c557 3470 return single_open_net(inode, file, rt6_stats_seq_show);
69ddb805
DL		 3471}
3472
9a32144e 3473static const struct file_operations rt6_stats_seq_fops = {
1da177e4
LT		 3474 .owner = THIS_MODULE,
3475 .open = rt6_stats_seq_open,
3476 .read = seq_read,
3477 .llseek = seq_lseek,
b6fcbdb4 3478 .release = single_release_net,
1da177e4
LT		 3479};
3480#endif /* CONFIG_PROC_FS */
3481
3482#ifdef CONFIG_SYSCTL
3483
1da177e4 3484static
fe2c6338 3485int ipv6_sysctl_rtcache_flush(struct ctl_table *ctl, int write,
1da177e4
LT		 3486 void __user *buffer, size_t *lenp, loff_t *ppos)
3487{
c486da34
LAG		 3488 struct net *net;
3489 int delay;
3490 if (!write)
1da177e4 3491 return -EINVAL;
c486da34
LAG		 3492
3493 net = (struct net *)ctl->extra1;
3494 delay = net->ipv6.sysctl.flush_delay;
3495 proc_dointvec(ctl, write, buffer, lenp, ppos);
2ac3ac8f 3496 fib6_run_gc(delay <= 0 ? 0 : (unsigned long)delay, net, delay > 0);
c486da34 3497 return 0;
1da177e4
LT		 3498}
3499
fe2c6338 3500struct ctl_table ipv6_route_table_template[] = {
1ab1457c 3501 {
1da177e4 3502 .procname = "flush",
4990509f 3503 .data = &init_net.ipv6.sysctl.flush_delay,
1da177e4 3504 .maxlen = sizeof(int),
89c8b3a1 3505 .mode = 0200,
6d9f239a 3506 .proc_handler = ipv6_sysctl_rtcache_flush
1da177e4
LT		 3507 },
3508 {
1da177e4 3509 .procname = "gc_thresh",
9a7ec3a9 3510 .data = &ip6_dst_ops_template.gc_thresh,
1da177e4
LT		 3511 .maxlen = sizeof(int),
3512 .mode = 0644,
6d9f239a 3513 .proc_handler = proc_dointvec,
1da177e4
LT		 3514 },
3515 {
1da177e4 3516 .procname = "max_size",
4990509f 3517 .data = &init_net.ipv6.sysctl.ip6_rt_max_size,
1da177e4
LT		 3518 .maxlen = sizeof(int),
3519 .mode = 0644,
6d9f239a 3520 .proc_handler = proc_dointvec,
1da177e4
LT		 3521 },
3522 {
1da177e4 3523 .procname = "gc_min_interval",
4990509f 3524 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 3525 .maxlen = sizeof(int),
3526 .mode = 0644,
6d9f239a 3527 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3528 },
3529 {
1da177e4 3530 .procname = "gc_timeout",
4990509f 3531 .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout,
1da177e4
LT		 3532 .maxlen = sizeof(int),
3533 .mode = 0644,
6d9f239a 3534 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3535 },
3536 {
1da177e4 3537 .procname = "gc_interval",
4990509f 3538 .data = &init_net.ipv6.sysctl.ip6_rt_gc_interval,
1da177e4
LT		 3539 .maxlen = sizeof(int),
3540 .mode = 0644,
6d9f239a 3541 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3542 },
3543 {
1da177e4 3544 .procname = "gc_elasticity",
4990509f 3545 .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity,
1da177e4
LT		 3546 .maxlen = sizeof(int),
3547 .mode = 0644,
f3d3f616 3548 .proc_handler = proc_dointvec,
1da177e4
LT		 3549 },
3550 {
1da177e4 3551 .procname = "mtu_expires",
4990509f 3552 .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires,
1da177e4
LT		 3553 .maxlen = sizeof(int),
3554 .mode = 0644,
6d9f239a 3555 .proc_handler = proc_dointvec_jiffies,
1da177e4
LT		 3556 },
3557 {
1da177e4 3558 .procname = "min_adv_mss",
4990509f 3559 .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss,
1da177e4
LT		 3560 .maxlen = sizeof(int),
3561 .mode = 0644,
f3d3f616 3562 .proc_handler = proc_dointvec,
1da177e4
LT		 3563 },
3564 {
1da177e4 3565 .procname = "gc_min_interval_ms",
4990509f 3566 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 3567 .maxlen = sizeof(int),
3568 .mode = 0644,
6d9f239a 3569 .proc_handler = proc_dointvec_ms_jiffies,
1da177e4 3570 },
f8572d8f 3571 { }
1da177e4
LT		 3572};
3573
2c8c1e72 3574struct ctl_table * __net_init ipv6_route_sysctl_init(struct net *net)
760f2d01
DL		 3575{
3576 struct ctl_table *table;
3577
3578 table = kmemdup(ipv6_route_table_template,
3579 sizeof(ipv6_route_table_template),
3580 GFP_KERNEL);
5ee09105
YH		 3581
3582 if (table) {
3583 table[0].data = &net->ipv6.sysctl.flush_delay;
c486da34 3584 table[0].extra1 = net;
86393e52 3585 table[1].data = &net->ipv6.ip6_dst_ops.gc_thresh;
5ee09105
YH		 3586 table[2].data = &net->ipv6.sysctl.ip6_rt_max_size;
3587 table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
3588 table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout;
3589 table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval;
3590 table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity;
3591 table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires;
3592 table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss;
9c69fabe 3593 table[9].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
464dc801
EB		 3594
3595 /* Don't export sysctls to unprivileged users */
3596 if (net->user_ns != &init_user_ns)
3597 table[0].procname = NULL;
5ee09105
YH		 3598 }
3599
760f2d01
DL		 3600 return table;
3601}
1da177e4
LT		 3602#endif
3603
2c8c1e72 3604static int __net_init ip6_route_net_init(struct net *net)
cdb18761 3605{
633d424b 3606 int ret = -ENOMEM;
8ed67789 3607
86393e52
AD		 3608 memcpy(&net->ipv6.ip6_dst_ops, &ip6_dst_ops_template,
3609 sizeof(net->ipv6.ip6_dst_ops));
f2fc6a54 3610
fc66f95c
ED		 3611 if (dst_entries_init(&net->ipv6.ip6_dst_ops) < 0)
3612 goto out_ip6_dst_ops;
3613
8ed67789
DL		 3614 net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template,
3615 sizeof(*net->ipv6.ip6_null_entry),
3616 GFP_KERNEL);
3617 if (!net->ipv6.ip6_null_entry)
fc66f95c 3618 goto out_ip6_dst_entries;
d8d1f30b 3619 net->ipv6.ip6_null_entry->dst.path =
8ed67789 3620 (struct dst_entry *)net->ipv6.ip6_null_entry;
d8d1f30b 3621 net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3622 dst_init_metrics(&net->ipv6.ip6_null_entry->dst,
3623 ip6_template_metrics, true);
8ed67789
DL		 3624
3625#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3626 net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template,
3627 sizeof(*net->ipv6.ip6_prohibit_entry),
3628 GFP_KERNEL);
68fffc67
PZ		 3629 if (!net->ipv6.ip6_prohibit_entry)
3630 goto out_ip6_null_entry;
d8d1f30b 3631 net->ipv6.ip6_prohibit_entry->dst.path =
8ed67789 3632 (struct dst_entry *)net->ipv6.ip6_prohibit_entry;
d8d1f30b 3633 net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3634 dst_init_metrics(&net->ipv6.ip6_prohibit_entry->dst,
3635 ip6_template_metrics, true);
8ed67789
DL		 3636
3637 net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template,
3638 sizeof(*net->ipv6.ip6_blk_hole_entry),
3639 GFP_KERNEL);
68fffc67
PZ		 3640 if (!net->ipv6.ip6_blk_hole_entry)
3641 goto out_ip6_prohibit_entry;
d8d1f30b 3642 net->ipv6.ip6_blk_hole_entry->dst.path =
8ed67789 3643 (struct dst_entry *)net->ipv6.ip6_blk_hole_entry;
d8d1f30b 3644 net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops;
62fa8a84
DM		 3645 dst_init_metrics(&net->ipv6.ip6_blk_hole_entry->dst,
3646 ip6_template_metrics, true);
8ed67789
DL		 3647#endif
3648
b339a47c
PZ		 3649 net->ipv6.sysctl.flush_delay = 0;
3650 net->ipv6.sysctl.ip6_rt_max_size = 4096;
3651 net->ipv6.sysctl.ip6_rt_gc_min_interval = HZ / 2;
3652 net->ipv6.sysctl.ip6_rt_gc_timeout = 60*HZ;
3653 net->ipv6.sysctl.ip6_rt_gc_interval = 30*HZ;
3654 net->ipv6.sysctl.ip6_rt_gc_elasticity = 9;
3655 net->ipv6.sysctl.ip6_rt_mtu_expires = 10*60*HZ;
3656 net->ipv6.sysctl.ip6_rt_min_advmss = IPV6_MIN_MTU - 20 - 40;
3657
6891a346
BT		 3658 net->ipv6.ip6_rt_gc_expire = 30*HZ;
3659
8ed67789
DL		 3660 ret = 0;
3661out:
3662 return ret;
f2fc6a54 3663
68fffc67
PZ		 3664#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3665out_ip6_prohibit_entry:
3666 kfree(net->ipv6.ip6_prohibit_entry);
3667out_ip6_null_entry:
3668 kfree(net->ipv6.ip6_null_entry);
3669#endif
fc66f95c
ED		 3670out_ip6_dst_entries:
3671 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
f2fc6a54 3672out_ip6_dst_ops:
f2fc6a54 3673 goto out;
cdb18761
DL		 3674}
3675
2c8c1e72 3676static void __net_exit ip6_route_net_exit(struct net *net)
cdb18761 3677{
8ed67789
DL		 3678 kfree(net->ipv6.ip6_null_entry);
3679#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3680 kfree(net->ipv6.ip6_prohibit_entry);
3681 kfree(net->ipv6.ip6_blk_hole_entry);
3682#endif
41bb78b4 3683 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
cdb18761
DL		 3684}
3685
d189634e
TG		 3686static int __net_init ip6_route_net_init_late(struct net *net)
3687{
3688#ifdef CONFIG_PROC_FS
d4beaa66
G		 3689 proc_create("ipv6_route", 0, net->proc_net, &ipv6_route_proc_fops);
3690 proc_create("rt6_stats", S_IRUGO, net->proc_net, &rt6_stats_seq_fops);
d189634e
TG		 3691#endif
3692 return 0;
3693}
3694
3695static void __net_exit ip6_route_net_exit_late(struct net *net)
3696{
3697#ifdef CONFIG_PROC_FS
ece31ffd
G		 3698 remove_proc_entry("ipv6_route", net->proc_net);
3699 remove_proc_entry("rt6_stats", net->proc_net);
d189634e
TG		 3700#endif
3701}
3702
cdb18761
DL		 3703static struct pernet_operations ip6_route_net_ops = {
3704 .init = ip6_route_net_init,
3705 .exit = ip6_route_net_exit,
3706};
3707
c3426b47
DM		 3708static int __net_init ipv6_inetpeer_init(struct net *net)
3709{
3710 struct inet_peer_base *bp = kmalloc(sizeof(*bp), GFP_KERNEL);
3711
3712 if (!bp)
3713 return -ENOMEM;
3714 inet_peer_base_init(bp);
3715 net->ipv6.peers = bp;
3716 return 0;
3717}
3718
3719static void __net_exit ipv6_inetpeer_exit(struct net *net)
3720{
3721 struct inet_peer_base *bp = net->ipv6.peers;
3722
3723 net->ipv6.peers = NULL;
56a6b248 3724 inetpeer_invalidate_tree(bp);
c3426b47
DM		 3725 kfree(bp);
3726}
3727
2b823f72 3728static struct pernet_operations ipv6_inetpeer_ops = {
c3426b47
DM		 3729 .init = ipv6_inetpeer_init,
3730 .exit = ipv6_inetpeer_exit,
3731};
3732
d189634e
TG		 3733static struct pernet_operations ip6_route_net_late_ops = {
3734 .init = ip6_route_net_init_late,
3735 .exit = ip6_route_net_exit_late,
3736};
3737
8ed67789
DL		 3738static struct notifier_block ip6_route_dev_notifier = {
3739 .notifier_call = ip6_route_dev_notify,
3740 .priority = 0,
3741};
3742
433d49c3 3743int __init ip6_route_init(void)
1da177e4 3744{
433d49c3 3745 int ret;
8d0b94af 3746 int cpu;
433d49c3 3747
9a7ec3a9
DL		 3748 ret = -ENOMEM;
3749 ip6_dst_ops_template.kmem_cachep =
e5d679f3 3750 kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0,
f845ab6b 3751 SLAB_HWCACHE_ALIGN, NULL);
9a7ec3a9 3752 if (!ip6_dst_ops_template.kmem_cachep)
c19a28e1 3753 goto out;
14e50e57 3754
fc66f95c 3755 ret = dst_entries_init(&ip6_dst_blackhole_ops);
8ed67789 3756 if (ret)
bdb3289f 3757 goto out_kmem_cache;
bdb3289f 3758
c3426b47
DM		 3759 ret = register_pernet_subsys(&ipv6_inetpeer_ops);
3760 if (ret)
e8803b6c 3761 goto out_dst_entries;
2a0c451a 3762
7e52b33b
DM		 3763 ret = register_pernet_subsys(&ip6_route_net_ops);
3764 if (ret)
3765 goto out_register_inetpeer;
c3426b47 3766
5dc121e9
AE		 3767 ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep;
3768
8ed67789
DL		 3769 /* Registering of the loopback is done before this portion of code,
3770 * the loopback reference in rt6_info will not be taken, do it
3771 * manually for init_net */
d8d1f30b 3772 init_net.ipv6.ip6_null_entry->dst.dev = init_net.loopback_dev;
8ed67789
DL		 3773 init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3774 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
d8d1f30b 3775 init_net.ipv6.ip6_prohibit_entry->dst.dev = init_net.loopback_dev;
8ed67789 3776 init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
d8d1f30b 3777 init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev;
8ed67789
DL		 3778 init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3779 #endif
e8803b6c 3780 ret = fib6_init();
433d49c3 3781 if (ret)
8ed67789 3782 goto out_register_subsys;
433d49c3 3783
433d49c3
DL		 3784 ret = xfrm6_init();
3785 if (ret)
e8803b6c 3786 goto out_fib6_init;
c35b7e72 3787
433d49c3
DL		 3788 ret = fib6_rules_init();
3789 if (ret)
3790 goto xfrm6_init;
7e5449c2 3791
d189634e
TG		 3792 ret = register_pernet_subsys(&ip6_route_net_late_ops);
3793 if (ret)
3794 goto fib6_rules_init;
3795
433d49c3 3796 ret = -ENOBUFS;
c7ac8679
GR		 3797 if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL, NULL) ||
3798 __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL, NULL) ||
3799 __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL, NULL))
d189634e 3800 goto out_register_late_subsys;
c127ea2c 3801
8ed67789 3802 ret = register_netdevice_notifier(&ip6_route_dev_notifier);
cdb18761 3803 if (ret)
d189634e 3804 goto out_register_late_subsys;
8ed67789 3805
8d0b94af
MKL		 3806 for_each_possible_cpu(cpu) {
3807 struct uncached_list *ul = per_cpu_ptr(&rt6_uncached_list, cpu);
3808
3809 INIT_LIST_HEAD(&ul->head);
3810 spin_lock_init(&ul->lock);
3811 }
3812
433d49c3
DL		 3813out:
3814 return ret;
3815
d189634e
TG		 3816out_register_late_subsys:
3817 unregister_pernet_subsys(&ip6_route_net_late_ops);
433d49c3 3818fib6_rules_init:
433d49c3
DL		 3819 fib6_rules_cleanup();
3820xfrm6_init:
433d49c3 3821 xfrm6_fini();
2a0c451a
TG		 3822out_fib6_init:
3823 fib6_gc_cleanup();
8ed67789
DL		 3824out_register_subsys:
3825 unregister_pernet_subsys(&ip6_route_net_ops);
7e52b33b
DM		 3826out_register_inetpeer:
3827 unregister_pernet_subsys(&ipv6_inetpeer_ops);
fc66f95c
ED		 3828out_dst_entries:
3829 dst_entries_destroy(&ip6_dst_blackhole_ops);
433d49c3 3830out_kmem_cache:
f2fc6a54 3831 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
433d49c3 3832 goto out;
1da177e4
LT		 3833}
3834
3835void ip6_route_cleanup(void)
3836{
8ed67789 3837 unregister_netdevice_notifier(&ip6_route_dev_notifier);
d189634e 3838 unregister_pernet_subsys(&ip6_route_net_late_ops);
101367c2 3839 fib6_rules_cleanup();
1da177e4 3840 xfrm6_fini();
1da177e4 3841 fib6_gc_cleanup();
c3426b47 3842 unregister_pernet_subsys(&ipv6_inetpeer_ops);
8ed67789 3843 unregister_pernet_subsys(&ip6_route_net_ops);
41bb78b4 3844 dst_entries_destroy(&ip6_dst_blackhole_ops);
f2fc6a54 3845 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
1da177e4 3846}
Ubuntu Artful kernel mirror