]> bbs.cooldavid.org Git - net-next-2.6.git/blame - net/ipv6/route.c
git://bbs.cooldavid.org / net-next-2.6.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
inet: Don't lookup the socket if there's a socket attached to the skb
[net-next-2.6.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
4fc268d2 27#include <linux/capability.h>
1da177e4
LT		 28#include <linux/errno.h>
29#include <linux/types.h>
30#include <linux/times.h>
31#include <linux/socket.h>
32#include <linux/sockios.h>
33#include <linux/net.h>
34#include <linux/route.h>
35#include <linux/netdevice.h>
36#include <linux/in6.h>
7bc570c8 37#include <linux/mroute6.h>
1da177e4 38#include <linux/init.h>
1da177e4 39#include <linux/if_arp.h>
1da177e4
LT		 40#include <linux/proc_fs.h>
41#include <linux/seq_file.h>
5b7c931d 42#include <linux/nsproxy.h>
457c4cbc 43#include <net/net_namespace.h>
1da177e4
LT		 44#include <net/snmp.h>
45#include <net/ipv6.h>
46#include <net/ip6_fib.h>
47#include <net/ip6_route.h>
48#include <net/ndisc.h>
49#include <net/addrconf.h>
50#include <net/tcp.h>
51#include <linux/rtnetlink.h>
52#include <net/dst.h>
53#include <net/xfrm.h>
8d71740c 54#include <net/netevent.h>
21713ebc 55#include <net/netlink.h>
1da177e4
LT		 56
57#include <asm/uaccess.h>
58
59#ifdef CONFIG_SYSCTL
60#include <linux/sysctl.h>
61#endif
62
63/* Set to 3 to get tracing. */
64#define RT6_DEBUG 2
65
66#if RT6_DEBUG >= 3
67#define RDBG(x) printk x
68#define RT6_TRACE(x...) printk(KERN_DEBUG x)
69#else
70#define RDBG(x)
71#define RT6_TRACE(x...) do { ; } while (0)
72#endif
73
519fbd87 74#define CLONE_OFFLINK_ROUTE 0
1da177e4 75
1da177e4
LT		 76static struct rt6_info * ip6_rt_copy(struct rt6_info *ort);
77static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie);
78static struct dst_entry *ip6_negative_advice(struct dst_entry *);
79static void ip6_dst_destroy(struct dst_entry *);
80static void ip6_dst_ifdown(struct dst_entry *,
81 struct net_device *dev, int how);
569d3645 82static int ip6_dst_gc(struct dst_ops *ops);
1da177e4
LT		 83
84static int ip6_pkt_discard(struct sk_buff *skb);
85static int ip6_pkt_discard_out(struct sk_buff *skb);
86static void ip6_link_failure(struct sk_buff *skb);
87static void ip6_rt_update_pmtu(struct dst_entry *dst, u32 mtu);
88
70ceb4f5 89#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0
DL		 90static struct rt6_info *rt6_add_route_info(struct net *net,
91 struct in6_addr *prefix, int prefixlen,
70ceb4f5
YH		 92 struct in6_addr *gwaddr, int ifindex,
93 unsigned pref);
efa2cea0
DL		 94static struct rt6_info *rt6_get_route_info(struct net *net,
95 struct in6_addr *prefix, int prefixlen,
70ceb4f5
YH		 96 struct in6_addr *gwaddr, int ifindex);
97#endif
98
9a7ec3a9 99static struct dst_ops ip6_dst_ops_template = {
1da177e4
LT		 100 .family = AF_INET6,
101 .protocol = __constant_htons(ETH_P_IPV6),
102 .gc = ip6_dst_gc,
103 .gc_thresh = 1024,
104 .check = ip6_dst_check,
105 .destroy = ip6_dst_destroy,
106 .ifdown = ip6_dst_ifdown,
107 .negative_advice = ip6_negative_advice,
108 .link_failure = ip6_link_failure,
109 .update_pmtu = ip6_rt_update_pmtu,
1ac06e03 110 .local_out = __ip6_local_out,
1da177e4 111 .entry_size = sizeof(struct rt6_info),
e2422970 112 .entries = ATOMIC_INIT(0),
1da177e4
LT		 113};
114
14e50e57
DM		 115static void ip6_rt_blackhole_update_pmtu(struct dst_entry *dst, u32 mtu)
116{
117}
118
119static struct dst_ops ip6_dst_blackhole_ops = {
120 .family = AF_INET6,
121 .protocol = __constant_htons(ETH_P_IPV6),
122 .destroy = ip6_dst_destroy,
123 .check = ip6_dst_check,
124 .update_pmtu = ip6_rt_blackhole_update_pmtu,
125 .entry_size = sizeof(struct rt6_info),
e2422970 126 .entries = ATOMIC_INIT(0),
14e50e57
DM		 127};
128
bdb3289f 129static struct rt6_info ip6_null_entry_template = {
1da177e4
LT		 130 .u = {
131 .dst = {
132 .__refcnt = ATOMIC_INIT(1),
133 .__use = 1,
1da177e4
LT		 134 .obsolete = -1,
135 .error = -ENETUNREACH,
136 .metrics = { [RTAX_HOPLIMIT - 1] = 255, },
137 .input = ip6_pkt_discard,
138 .output = ip6_pkt_discard_out,
1da177e4
LT		 139 }
140 },
141 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
142 .rt6i_metric = ~(u32) 0,
143 .rt6i_ref = ATOMIC_INIT(1),
144};
145
101367c2
TG		 146#ifdef CONFIG_IPV6_MULTIPLE_TABLES
147
6723ab54
DM		 148static int ip6_pkt_prohibit(struct sk_buff *skb);
149static int ip6_pkt_prohibit_out(struct sk_buff *skb);
6723ab54 150
280a34c8 151static struct rt6_info ip6_prohibit_entry_template = {
101367c2
TG		 152 .u = {
153 .dst = {
154 .__refcnt = ATOMIC_INIT(1),
155 .__use = 1,
101367c2
TG		 156 .obsolete = -1,
157 .error = -EACCES,
158 .metrics = { [RTAX_HOPLIMIT - 1] = 255, },
9ce8ade0
TG		 159 .input = ip6_pkt_prohibit,
160 .output = ip6_pkt_prohibit_out,
101367c2
TG		 161 }
162 },
163 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
164 .rt6i_metric = ~(u32) 0,
165 .rt6i_ref = ATOMIC_INIT(1),
166};
167
bdb3289f 168static struct rt6_info ip6_blk_hole_entry_template = {
101367c2
TG		 169 .u = {
170 .dst = {
171 .__refcnt = ATOMIC_INIT(1),
172 .__use = 1,
101367c2
TG		 173 .obsolete = -1,
174 .error = -EINVAL,
175 .metrics = { [RTAX_HOPLIMIT - 1] = 255, },
352e512c
HX		 176 .input = dst_discard,
177 .output = dst_discard,
101367c2
TG		 178 }
179 },
180 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
181 .rt6i_metric = ~(u32) 0,
182 .rt6i_ref = ATOMIC_INIT(1),
183};
184
185#endif
186
1da177e4 187/* allocate dst with ip6_dst_ops */
f2fc6a54 188static inline struct rt6_info *ip6_dst_alloc(struct dst_ops *ops)
1da177e4 189{
f2fc6a54 190 return (struct rt6_info *)dst_alloc(ops);
1da177e4
LT		 191}
192
193static void ip6_dst_destroy(struct dst_entry *dst)
194{
195 struct rt6_info *rt = (struct rt6_info *)dst;
196 struct inet6_dev *idev = rt->rt6i_idev;
197
198 if (idev != NULL) {
199 rt->rt6i_idev = NULL;
200 in6_dev_put(idev);
1ab1457c 201 }
1da177e4
LT		 202}
203
204static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
205 int how)
206{
207 struct rt6_info *rt = (struct rt6_info *)dst;
208 struct inet6_dev *idev = rt->rt6i_idev;
5a3e55d6 209 struct net_device *loopback_dev =
c346dca1 210 dev_net(dev)->loopback_dev;
1da177e4 211
5a3e55d6
DL		 212 if (dev != loopback_dev && idev != NULL && idev->dev == dev) {
213 struct inet6_dev *loopback_idev =
214 in6_dev_get(loopback_dev);
1da177e4
LT		 215 if (loopback_idev != NULL) {
216 rt->rt6i_idev = loopback_idev;
217 in6_dev_put(idev);
218 }
219 }
220}
221
222static __inline__ int rt6_check_expired(const struct rt6_info *rt)
223{
224 return (rt->rt6i_flags & RTF_EXPIRES &&
225 time_after(jiffies, rt->rt6i_expires));
226}
227
c71099ac
TG		 228static inline int rt6_need_strict(struct in6_addr *daddr)
229{
230 return (ipv6_addr_type(daddr) &
5ce83afa 231 (IPV6_ADDR_MULTICAST | IPV6_ADDR_LINKLOCAL | IPV6_ADDR_LOOPBACK));
c71099ac
TG		 232}
233
1da177e4 234/*
c71099ac 235 * Route lookup. Any table->tb6_lock is implied.
1da177e4
LT		 236 */
237
8ed67789
DL		 238static inline struct rt6_info *rt6_device_match(struct net *net,
239 struct rt6_info *rt,
dd3abc4e 240 struct in6_addr *saddr,
1da177e4 241 int oif,
d420895e 242 int flags)
1da177e4
LT		 243{
244 struct rt6_info *local = NULL;
245 struct rt6_info *sprt;
246
dd3abc4e
YH		 247 if (!oif && ipv6_addr_any(saddr))
248 goto out;
249
250 for (sprt = rt; sprt; sprt = sprt->u.dst.rt6_next) {
251 struct net_device *dev = sprt->rt6i_dev;
252
253 if (oif) {
1da177e4
LT		 254 if (dev->ifindex == oif)
255 return sprt;
256 if (dev->flags & IFF_LOOPBACK) {
257 if (sprt->rt6i_idev == NULL ||
258 sprt->rt6i_idev->dev->ifindex != oif) {
d420895e 259 if (flags & RT6_LOOKUP_F_IFACE && oif)
1da177e4 260 continue;
1ab1457c 261 if (local && (!oif ||
1da177e4
LT		 262 local->rt6i_idev->dev->ifindex == oif))
263 continue;
264 }
265 local = sprt;
266 }
dd3abc4e
YH		 267 } else {
268 if (ipv6_chk_addr(net, saddr, dev,
269 flags & RT6_LOOKUP_F_IFACE))
270 return sprt;
1da177e4 271 }
dd3abc4e 272 }
1da177e4 273
dd3abc4e 274 if (oif) {
1da177e4
LT		 275 if (local)
276 return local;
277
d420895e 278 if (flags & RT6_LOOKUP_F_IFACE)
8ed67789 279 return net->ipv6.ip6_null_entry;
1da177e4 280 }
dd3abc4e 281out:
1da177e4
LT		 282 return rt;
283}
284
27097255
YH		 285#ifdef CONFIG_IPV6_ROUTER_PREF
286static void rt6_probe(struct rt6_info *rt)
287{
288 struct neighbour *neigh = rt ? rt->rt6i_nexthop : NULL;
289 /*
290 * Okay, this does not seem to be appropriate
291 * for now, however, we need to check if it
292 * is really so; aka Router Reachability Probing.
293 *
294 * Router Reachability Probe MUST be rate-limited
295 * to no more than one per minute.
296 */
297 if (!neigh || (neigh->nud_state & NUD_VALID))
298 return;
299 read_lock_bh(&neigh->lock);
300 if (!(neigh->nud_state & NUD_VALID) &&
52e16356 301 time_after(jiffies, neigh->updated + rt->rt6i_idev->cnf.rtr_probe_interval)) {
27097255
YH		 302 struct in6_addr mcaddr;
303 struct in6_addr *target;
304
305 neigh->updated = jiffies;
306 read_unlock_bh(&neigh->lock);
307
308 target = (struct in6_addr *)&neigh->primary_key;
309 addrconf_addr_solict_mult(target, &mcaddr);
310 ndisc_send_ns(rt->rt6i_dev, NULL, target, &mcaddr, NULL);
311 } else
312 read_unlock_bh(&neigh->lock);
313}
314#else
315static inline void rt6_probe(struct rt6_info *rt)
316{
317 return;
318}
319#endif
320
1da177e4 321/*
554cfb7e 322 * Default Router Selection (RFC 2461 6.3.6)
1da177e4 323 */
b6f99a21 324static inline int rt6_check_dev(struct rt6_info *rt, int oif)
554cfb7e
YH		 325{
326 struct net_device *dev = rt->rt6i_dev;
161980f4 327 if (!oif || dev->ifindex == oif)
554cfb7e 328 return 2;
161980f4
DM		 329 if ((dev->flags & IFF_LOOPBACK) &&
330 rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif)
331 return 1;
332 return 0;
554cfb7e 333}
1da177e4 334
b6f99a21 335static inline int rt6_check_neigh(struct rt6_info *rt)
1da177e4 336{
554cfb7e 337 struct neighbour *neigh = rt->rt6i_nexthop;
398bcbeb 338 int m;
4d0c5911
YH		 339 if (rt->rt6i_flags & RTF_NONEXTHOP ||
340 !(rt->rt6i_flags & RTF_GATEWAY))
341 m = 1;
342 else if (neigh) {
554cfb7e
YH		 343 read_lock_bh(&neigh->lock);
344 if (neigh->nud_state & NUD_VALID)
4d0c5911 345 m = 2;
398bcbeb
YH		 346#ifdef CONFIG_IPV6_ROUTER_PREF
347 else if (neigh->nud_state & NUD_FAILED)
348 m = 0;
349#endif
350 else
ea73ee23 351 m = 1;
554cfb7e 352 read_unlock_bh(&neigh->lock);
398bcbeb
YH		 353 } else
354 m = 0;
554cfb7e 355 return m;
1da177e4
LT		 356}
357
554cfb7e
YH		 358static int rt6_score_route(struct rt6_info *rt, int oif,
359 int strict)
1da177e4 360{
4d0c5911 361 int m, n;
1ab1457c 362
4d0c5911 363 m = rt6_check_dev(rt, oif);
77d16f45 364 if (!m && (strict & RT6_LOOKUP_F_IFACE))
554cfb7e 365 return -1;
ebacaaa0
YH		 366#ifdef CONFIG_IPV6_ROUTER_PREF
367 m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2;
368#endif
4d0c5911 369 n = rt6_check_neigh(rt);
557e92ef 370 if (!n && (strict & RT6_LOOKUP_F_REACHABLE))
554cfb7e
YH		 371 return -1;
372 return m;
373}
374
f11e6659
DM		 375static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict,
376 int *mpri, struct rt6_info *match)
554cfb7e 377{
f11e6659
DM		 378 int m;
379
380 if (rt6_check_expired(rt))
381 goto out;
382
383 m = rt6_score_route(rt, oif, strict);
384 if (m < 0)
385 goto out;
386
387 if (m > *mpri) {
388 if (strict & RT6_LOOKUP_F_REACHABLE)
389 rt6_probe(match);
390 *mpri = m;
391 match = rt;
392 } else if (strict & RT6_LOOKUP_F_REACHABLE) {
393 rt6_probe(rt);
394 }
395
396out:
397 return match;
398}
399
400static struct rt6_info *find_rr_leaf(struct fib6_node *fn,
401 struct rt6_info *rr_head,
402 u32 metric, int oif, int strict)
403{
404 struct rt6_info *rt, *match;
554cfb7e 405 int mpri = -1;
1da177e4 406
f11e6659
DM		 407 match = NULL;
408 for (rt = rr_head; rt && rt->rt6i_metric == metric;
409 rt = rt->u.dst.rt6_next)
410 match = find_match(rt, oif, strict, &mpri, match);
411 for (rt = fn->leaf; rt && rt != rr_head && rt->rt6i_metric == metric;
412 rt = rt->u.dst.rt6_next)
413 match = find_match(rt, oif, strict, &mpri, match);
1da177e4 414
f11e6659
DM		 415 return match;
416}
1da177e4 417
f11e6659
DM		 418static struct rt6_info *rt6_select(struct fib6_node *fn, int oif, int strict)
419{
420 struct rt6_info *match, *rt0;
8ed67789 421 struct net *net;
1da177e4 422
f11e6659 423 RT6_TRACE("%s(fn->leaf=%p, oif=%d)\n",
0dc47877 424 __func__, fn->leaf, oif);
554cfb7e 425
f11e6659
DM		 426 rt0 = fn->rr_ptr;
427 if (!rt0)
428 fn->rr_ptr = rt0 = fn->leaf;
1da177e4 429
f11e6659 430 match = find_rr_leaf(fn, rt0, rt0->rt6i_metric, oif, strict);
1da177e4 431
554cfb7e 432 if (!match &&
f11e6659
DM		 433 (strict & RT6_LOOKUP_F_REACHABLE)) {
434 struct rt6_info *next = rt0->u.dst.rt6_next;
435
554cfb7e 436 /* no entries matched; do round-robin */
f11e6659
DM		 437 if (!next || next->rt6i_metric != rt0->rt6i_metric)
438 next = fn->leaf;
439
440 if (next != rt0)
441 fn->rr_ptr = next;
1da177e4 442 }
1da177e4 443
f11e6659 444 RT6_TRACE("%s() => %p\n",
0dc47877 445 __func__, match);
1da177e4 446
c346dca1 447 net = dev_net(rt0->rt6i_dev);
8ed67789 448 return (match ? match : net->ipv6.ip6_null_entry);
1da177e4
LT		 449}
450
70ceb4f5
YH		 451#ifdef CONFIG_IPV6_ROUTE_INFO
452int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
453 struct in6_addr *gwaddr)
454{
c346dca1 455 struct net *net = dev_net(dev);
70ceb4f5
YH		 456 struct route_info *rinfo = (struct route_info *) opt;
457 struct in6_addr prefix_buf, *prefix;
458 unsigned int pref;
4bed72e4 459 unsigned long lifetime;
70ceb4f5
YH		 460 struct rt6_info *rt;
461
462 if (len < sizeof(struct route_info)) {
463 return -EINVAL;
464 }
465
466 /* Sanity check for prefix_len and length */
467 if (rinfo->length > 3) {
468 return -EINVAL;
469 } else if (rinfo->prefix_len > 128) {
470 return -EINVAL;
471 } else if (rinfo->prefix_len > 64) {
472 if (rinfo->length < 2) {
473 return -EINVAL;
474 }
475 } else if (rinfo->prefix_len > 0) {
476 if (rinfo->length < 1) {
477 return -EINVAL;
478 }
479 }
480
481 pref = rinfo->route_pref;
482 if (pref == ICMPV6_ROUTER_PREF_INVALID)
483 pref = ICMPV6_ROUTER_PREF_MEDIUM;
484
4bed72e4 485 lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ);
70ceb4f5
YH		 486
487 if (rinfo->length == 3)
488 prefix = (struct in6_addr *)rinfo->prefix;
489 else {
490 /* this function is safe */
491 ipv6_addr_prefix(&prefix_buf,
492 (struct in6_addr *)rinfo->prefix,
493 rinfo->prefix_len);
494 prefix = &prefix_buf;
495 }
496
efa2cea0
DL		 497 rt = rt6_get_route_info(net, prefix, rinfo->prefix_len, gwaddr,
498 dev->ifindex);
70ceb4f5
YH		 499
500 if (rt && !lifetime) {
e0a1ad73 501 ip6_del_rt(rt);
70ceb4f5
YH		 502 rt = NULL;
503 }
504
505 if (!rt && lifetime)
efa2cea0 506 rt = rt6_add_route_info(net, prefix, rinfo->prefix_len, gwaddr, dev->ifindex,
70ceb4f5
YH		 507 pref);
508 else if (rt)
509 rt->rt6i_flags = RTF_ROUTEINFO |
510 (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref);
511
512 if (rt) {
4bed72e4 513 if (!addrconf_finite_timeout(lifetime)) {
70ceb4f5
YH		 514 rt->rt6i_flags &= ~RTF_EXPIRES;
515 } else {
516 rt->rt6i_expires = jiffies + HZ * lifetime;
517 rt->rt6i_flags |= RTF_EXPIRES;
518 }
519 dst_release(&rt->u.dst);
520 }
521 return 0;
522}
523#endif
524
8ed67789 525#define BACKTRACK(__net, saddr) \
982f56f3 526do { \
8ed67789 527 if (rt == __net->ipv6.ip6_null_entry) { \
982f56f3 528 struct fib6_node *pn; \
e0eda7bb 529 while (1) { \
982f56f3
YH		 530 if (fn->fn_flags & RTN_TL_ROOT) \
531 goto out; \
532 pn = fn->parent; \
533 if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn) \
8bce65b9 534 fn = fib6_lookup(FIB6_SUBTREE(pn), NULL, saddr); \
982f56f3
YH		 535 else \
536 fn = pn; \
537 if (fn->fn_flags & RTN_RTINFO) \
538 goto restart; \
c71099ac 539 } \
c71099ac 540 } \
982f56f3 541} while(0)
c71099ac 542
8ed67789
DL		 543static struct rt6_info *ip6_pol_route_lookup(struct net *net,
544 struct fib6_table *table,
c71099ac 545 struct flowi *fl, int flags)
1da177e4
LT		 546{
547 struct fib6_node *fn;
548 struct rt6_info *rt;
549
c71099ac
TG		 550 read_lock_bh(&table->tb6_lock);
551 fn = fib6_lookup(&table->tb6_root, &fl->fl6_dst, &fl->fl6_src);
552restart:
553 rt = fn->leaf;
dd3abc4e 554 rt = rt6_device_match(net, rt, &fl->fl6_src, fl->oif, flags);
8ed67789 555 BACKTRACK(net, &fl->fl6_src);
c71099ac 556out:
03f49f34 557 dst_use(&rt->u.dst, jiffies);
c71099ac 558 read_unlock_bh(&table->tb6_lock);
c71099ac
TG		 559 return rt;
560
561}
562
9acd9f3a
YH		 563struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
564 const struct in6_addr *saddr, int oif, int strict)
c71099ac
TG		 565{
566 struct flowi fl = {
567 .oif = oif,
568 .nl_u = {
569 .ip6_u = {
570 .daddr = *daddr,
c71099ac
TG		 571 },
572 },
573 };
574 struct dst_entry *dst;
77d16f45 575 int flags = strict ? RT6_LOOKUP_F_IFACE : 0;
c71099ac 576
adaa70bb
TG		 577 if (saddr) {
578 memcpy(&fl.fl6_src, saddr, sizeof(*saddr));
579 flags |= RT6_LOOKUP_F_HAS_SADDR;
580 }
581
606a2b48 582 dst = fib6_rule_lookup(net, &fl, flags, ip6_pol_route_lookup);
c71099ac
TG		 583 if (dst->error == 0)
584 return (struct rt6_info *) dst;
585
586 dst_release(dst);
587
1da177e4
LT		 588 return NULL;
589}
590
7159039a
YH		 591EXPORT_SYMBOL(rt6_lookup);
592
c71099ac 593/* ip6_ins_rt is called with FREE table->tb6_lock.
1da177e4
LT		 594 It takes new route entry, the addition fails by any reason the
595 route is freed. In any case, if caller does not hold it, it may
596 be destroyed.
597 */
598
86872cb5 599static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 600{
601 int err;
c71099ac 602 struct fib6_table *table;
1da177e4 603
c71099ac
TG		 604 table = rt->rt6i_table;
605 write_lock_bh(&table->tb6_lock);
86872cb5 606 err = fib6_add(&table->tb6_root, rt, info);
c71099ac 607 write_unlock_bh(&table->tb6_lock);
1da177e4
LT		 608
609 return err;
610}
611
40e22e8f
TG		 612int ip6_ins_rt(struct rt6_info *rt)
613{
4d1169c1 614 struct nl_info info = {
c346dca1 615 .nl_net = dev_net(rt->rt6i_dev),
4d1169c1 616 };
528c4ceb 617 return __ip6_ins_rt(rt, &info);
40e22e8f
TG		 618}
619
95a9a5ba
YH		 620static struct rt6_info *rt6_alloc_cow(struct rt6_info *ort, struct in6_addr *daddr,
621 struct in6_addr *saddr)
1da177e4 622{
1da177e4
LT		 623 struct rt6_info *rt;
624
625 /*
626 * Clone the route.
627 */
628
629 rt = ip6_rt_copy(ort);
630
631 if (rt) {
58c4fb86
YH		 632 if (!(rt->rt6i_flags&RTF_GATEWAY)) {
633 if (rt->rt6i_dst.plen != 128 &&
634 ipv6_addr_equal(&rt->rt6i_dst.addr, daddr))
635 rt->rt6i_flags |= RTF_ANYCAST;
1da177e4 636 ipv6_addr_copy(&rt->rt6i_gateway, daddr);
58c4fb86 637 }
1da177e4 638
58c4fb86 639 ipv6_addr_copy(&rt->rt6i_dst.addr, daddr);
1da177e4
LT		 640 rt->rt6i_dst.plen = 128;
641 rt->rt6i_flags |= RTF_CACHE;
642 rt->u.dst.flags |= DST_HOST;
643
644#ifdef CONFIG_IPV6_SUBTREES
645 if (rt->rt6i_src.plen && saddr) {
646 ipv6_addr_copy(&rt->rt6i_src.addr, saddr);
647 rt->rt6i_src.plen = 128;
648 }
649#endif
650
651 rt->rt6i_nexthop = ndisc_get_neigh(rt->rt6i_dev, &rt->rt6i_gateway);
652
95a9a5ba 653 }
1da177e4 654
95a9a5ba
YH		 655 return rt;
656}
1da177e4 657
299d9939
YH		 658static struct rt6_info *rt6_alloc_clone(struct rt6_info *ort, struct in6_addr *daddr)
659{
660 struct rt6_info *rt = ip6_rt_copy(ort);
661 if (rt) {
662 ipv6_addr_copy(&rt->rt6i_dst.addr, daddr);
663 rt->rt6i_dst.plen = 128;
664 rt->rt6i_flags |= RTF_CACHE;
299d9939
YH		 665 rt->u.dst.flags |= DST_HOST;
666 rt->rt6i_nexthop = neigh_clone(ort->rt6i_nexthop);
667 }
668 return rt;
669}
670
8ed67789
DL		 671static struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table, int oif,
672 struct flowi *fl, int flags)
1da177e4
LT		 673{
674 struct fib6_node *fn;
519fbd87 675 struct rt6_info *rt, *nrt;
c71099ac 676 int strict = 0;
1da177e4 677 int attempts = 3;
519fbd87 678 int err;
53b7997f 679 int reachable = net->ipv6.devconf_all->forwarding ? 0 : RT6_LOOKUP_F_REACHABLE;
1da177e4 680
77d16f45 681 strict |= flags & RT6_LOOKUP_F_IFACE;
1da177e4
LT		 682
683relookup:
c71099ac 684 read_lock_bh(&table->tb6_lock);
1da177e4 685
8238dd06 686restart_2:
c71099ac 687 fn = fib6_lookup(&table->tb6_root, &fl->fl6_dst, &fl->fl6_src);
1da177e4
LT		 688
689restart:
4acad72d 690 rt = rt6_select(fn, oif, strict | reachable);
8ed67789
DL		 691
692 BACKTRACK(net, &fl->fl6_src);
693 if (rt == net->ipv6.ip6_null_entry ||
8238dd06 694 rt->rt6i_flags & RTF_CACHE)
1ddef044 695 goto out;
1da177e4 696
fb9de91e 697 dst_hold(&rt->u.dst);
c71099ac 698 read_unlock_bh(&table->tb6_lock);
fb9de91e 699
519fbd87 700 if (!rt->rt6i_nexthop && !(rt->rt6i_flags & RTF_NONEXTHOP))
c71099ac 701 nrt = rt6_alloc_cow(rt, &fl->fl6_dst, &fl->fl6_src);
519fbd87
YH		 702 else {
703#if CLONE_OFFLINK_ROUTE
c71099ac 704 nrt = rt6_alloc_clone(rt, &fl->fl6_dst);
519fbd87
YH		 705#else
706 goto out2;
707#endif
708 }
e40cf353 709
519fbd87 710 dst_release(&rt->u.dst);
8ed67789 711 rt = nrt ? : net->ipv6.ip6_null_entry;
1da177e4 712
519fbd87
YH		 713 dst_hold(&rt->u.dst);
714 if (nrt) {
40e22e8f 715 err = ip6_ins_rt(nrt);
519fbd87 716 if (!err)
1da177e4 717 goto out2;
1da177e4 718 }
1da177e4 719
519fbd87
YH		 720 if (--attempts <= 0)
721 goto out2;
722
723 /*
c71099ac 724 * Race condition! In the gap, when table->tb6_lock was
519fbd87
YH		 725 * released someone could insert this route. Relookup.
726 */
727 dst_release(&rt->u.dst);
728 goto relookup;
729
730out:
8238dd06
YH		 731 if (reachable) {
732 reachable = 0;
733 goto restart_2;
734 }
519fbd87 735 dst_hold(&rt->u.dst);
c71099ac 736 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 737out2:
738 rt->u.dst.lastuse = jiffies;
739 rt->u.dst.__use++;
c71099ac
TG		 740
741 return rt;
1da177e4
LT		 742}
743
8ed67789 744static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table,
4acad72d
PE		 745 struct flowi *fl, int flags)
746{
8ed67789 747 return ip6_pol_route(net, table, fl->iif, fl, flags);
4acad72d
PE		 748}
749
c71099ac
TG		 750void ip6_route_input(struct sk_buff *skb)
751{
0660e03f 752 struct ipv6hdr *iph = ipv6_hdr(skb);
c346dca1 753 struct net *net = dev_net(skb->dev);
adaa70bb 754 int flags = RT6_LOOKUP_F_HAS_SADDR;
c71099ac
TG		 755 struct flowi fl = {
756 .iif = skb->dev->ifindex,
757 .nl_u = {
758 .ip6_u = {
759 .daddr = iph->daddr,
760 .saddr = iph->saddr,
90bcaf7b 761 .flowlabel = (* (__be32 *) iph)&IPV6_FLOWINFO_MASK,
c71099ac
TG		 762 },
763 },
1ab1457c 764 .mark = skb->mark,
c71099ac
TG		 765 .proto = iph->nexthdr,
766 };
adaa70bb
TG		 767
768 if (rt6_need_strict(&iph->daddr))
769 flags |= RT6_LOOKUP_F_IFACE;
c71099ac 770
5578689a 771 skb->dst = fib6_rule_lookup(net, &fl, flags, ip6_pol_route_input);
c71099ac
TG		 772}
773
8ed67789 774static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table,
c71099ac 775 struct flowi *fl, int flags)
1da177e4 776{
8ed67789 777 return ip6_pol_route(net, table, fl->oif, fl, flags);
c71099ac
TG		 778}
779
4591db4f
DL		 780struct dst_entry * ip6_route_output(struct net *net, struct sock *sk,
781 struct flowi *fl)
c71099ac
TG		 782{
783 int flags = 0;
784
785 if (rt6_need_strict(&fl->fl6_dst))
77d16f45 786 flags |= RT6_LOOKUP_F_IFACE;
c71099ac 787
adaa70bb
TG		 788 if (!ipv6_addr_any(&fl->fl6_src))
789 flags |= RT6_LOOKUP_F_HAS_SADDR;
7cbca67c
YH		 790 else if (sk) {
791 unsigned int prefs = inet6_sk(sk)->srcprefs;
792 if (prefs & IPV6_PREFER_SRC_TMP)
793 flags |= RT6_LOOKUP_F_SRCPREF_TMP;
794 if (prefs & IPV6_PREFER_SRC_PUBLIC)
795 flags |= RT6_LOOKUP_F_SRCPREF_PUBLIC;
796 if (prefs & IPV6_PREFER_SRC_COA)
797 flags |= RT6_LOOKUP_F_SRCPREF_COA;
798 }
adaa70bb 799
4591db4f 800 return fib6_rule_lookup(net, fl, flags, ip6_pol_route_output);
1da177e4
LT		 801}
802
7159039a 803EXPORT_SYMBOL(ip6_route_output);
1da177e4 804
14e50e57
DM		 805int ip6_dst_blackhole(struct sock *sk, struct dst_entry **dstp, struct flowi *fl)
806{
807 struct rt6_info *ort = (struct rt6_info *) *dstp;
808 struct rt6_info *rt = (struct rt6_info *)
809 dst_alloc(&ip6_dst_blackhole_ops);
810 struct dst_entry *new = NULL;
811
812 if (rt) {
813 new = &rt->u.dst;
814
815 atomic_set(&new->__refcnt, 1);
816 new->__use = 1;
352e512c
HX		 817 new->input = dst_discard;
818 new->output = dst_discard;
14e50e57
DM		 819
820 memcpy(new->metrics, ort->u.dst.metrics, RTAX_MAX*sizeof(u32));
821 new->dev = ort->u.dst.dev;
822 if (new->dev)
823 dev_hold(new->dev);
824 rt->rt6i_idev = ort->rt6i_idev;
825 if (rt->rt6i_idev)
826 in6_dev_hold(rt->rt6i_idev);
827 rt->rt6i_expires = 0;
828
829 ipv6_addr_copy(&rt->rt6i_gateway, &ort->rt6i_gateway);
830 rt->rt6i_flags = ort->rt6i_flags & ~RTF_EXPIRES;
831 rt->rt6i_metric = 0;
832
833 memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key));
834#ifdef CONFIG_IPV6_SUBTREES
835 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
836#endif
837
838 dst_free(new);
839 }
840
841 dst_release(*dstp);
842 *dstp = new;
843 return (new ? 0 : -ENOMEM);
844}
845EXPORT_SYMBOL_GPL(ip6_dst_blackhole);
846
1da177e4
LT		 847/*
848 * Destination cache support functions
849 */
850
851static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie)
852{
853 struct rt6_info *rt;
854
855 rt = (struct rt6_info *) dst;
856
857 if (rt && rt->rt6i_node && (rt->rt6i_node->fn_sernum == cookie))
858 return dst;
859
860 return NULL;
861}
862
863static struct dst_entry *ip6_negative_advice(struct dst_entry *dst)
864{
865 struct rt6_info *rt = (struct rt6_info *) dst;
866
867 if (rt) {
868 if (rt->rt6i_flags & RTF_CACHE)
e0a1ad73 869 ip6_del_rt(rt);
1da177e4
LT		 870 else
871 dst_release(dst);
872 }
873 return NULL;
874}
875
876static void ip6_link_failure(struct sk_buff *skb)
877{
878 struct rt6_info *rt;
879
880 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0, skb->dev);
881
882 rt = (struct rt6_info *) skb->dst;
883 if (rt) {
884 if (rt->rt6i_flags&RTF_CACHE) {
885 dst_set_expires(&rt->u.dst, 0);
886 rt->rt6i_flags |= RTF_EXPIRES;
887 } else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT))
888 rt->rt6i_node->fn_sernum = -1;
889 }
890}
891
892static void ip6_rt_update_pmtu(struct dst_entry *dst, u32 mtu)
893{
894 struct rt6_info *rt6 = (struct rt6_info*)dst;
895
896 if (mtu < dst_mtu(dst) && rt6->rt6i_dst.plen == 128) {
897 rt6->rt6i_flags |= RTF_MODIFIED;
898 if (mtu < IPV6_MIN_MTU) {
899 mtu = IPV6_MIN_MTU;
900 dst->metrics[RTAX_FEATURES-1] |= RTAX_FEATURE_ALLFRAG;
901 }
902 dst->metrics[RTAX_MTU-1] = mtu;
8d71740c 903 call_netevent_notifiers(NETEVENT_PMTU_UPDATE, dst);
1da177e4
LT		 904 }
905}
906
1da177e4
LT		 907static int ipv6_get_mtu(struct net_device *dev);
908
5578689a 909static inline unsigned int ipv6_advmss(struct net *net, unsigned int mtu)
1da177e4
LT		 910{
911 mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr);
912
5578689a
DL		 913 if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss)
914 mtu = net->ipv6.sysctl.ip6_rt_min_advmss;
1da177e4
LT		 915
916 /*
1ab1457c
YH		 917 * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and
918 * corresponding MSS is IPV6_MAXPLEN - tcp_header_size.
919 * IPV6_MAXPLEN is also valid and means: "any MSS,
1da177e4
LT		 920 * rely only on pmtu discovery"
921 */
922 if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr))
923 mtu = IPV6_MAXPLEN;
924 return mtu;
925}
926
3b00944c
YH		 927static struct dst_entry *icmp6_dst_gc_list;
928static DEFINE_SPINLOCK(icmp6_dst_lock);
5d0bbeeb 929
3b00944c 930struct dst_entry *icmp6_dst_alloc(struct net_device *dev,
1da177e4 931 struct neighbour *neigh,
9acd9f3a 932 const struct in6_addr *addr)
1da177e4
LT		 933{
934 struct rt6_info *rt;
935 struct inet6_dev *idev = in6_dev_get(dev);
c346dca1 936 struct net *net = dev_net(dev);
1da177e4
LT		 937
938 if (unlikely(idev == NULL))
939 return NULL;
940
f2fc6a54 941 rt = ip6_dst_alloc(net->ipv6.ip6_dst_ops);
1da177e4
LT		 942 if (unlikely(rt == NULL)) {
943 in6_dev_put(idev);
944 goto out;
945 }
946
947 dev_hold(dev);
948 if (neigh)
949 neigh_hold(neigh);
950 else
951 neigh = ndisc_get_neigh(dev, addr);
952
953 rt->rt6i_dev = dev;
954 rt->rt6i_idev = idev;
955 rt->rt6i_nexthop = neigh;
956 atomic_set(&rt->u.dst.__refcnt, 1);
957 rt->u.dst.metrics[RTAX_HOPLIMIT-1] = 255;
958 rt->u.dst.metrics[RTAX_MTU-1] = ipv6_get_mtu(rt->rt6i_dev);
5578689a 959 rt->u.dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(net, dst_mtu(&rt->u.dst));
3b00944c 960 rt->u.dst.output = ip6_output;
1da177e4
LT		 961
962#if 0 /* there's no chance to use these for ndisc */
1ab1457c
YH		 963 rt->u.dst.flags = ipv6_addr_type(addr) & IPV6_ADDR_UNICAST
964 ? DST_HOST
1da177e4
LT		 965 : 0;
966 ipv6_addr_copy(&rt->rt6i_dst.addr, addr);
967 rt->rt6i_dst.plen = 128;
968#endif
969
3b00944c
YH		 970 spin_lock_bh(&icmp6_dst_lock);
971 rt->u.dst.next = icmp6_dst_gc_list;
972 icmp6_dst_gc_list = &rt->u.dst;
973 spin_unlock_bh(&icmp6_dst_lock);
1da177e4 974
5578689a 975 fib6_force_start_gc(net);
1da177e4
LT		 976
977out:
40aa7b90 978 return &rt->u.dst;
1da177e4
LT		 979}
980
3d0f24a7 981int icmp6_dst_gc(void)
1da177e4
LT		 982{
983 struct dst_entry *dst, *next, **pprev;
3d0f24a7 984 int more = 0;
1da177e4
LT		 985
986 next = NULL;
5d0bbeeb 987
3b00944c
YH		 988 spin_lock_bh(&icmp6_dst_lock);
989 pprev = &icmp6_dst_gc_list;
5d0bbeeb 990
1da177e4
LT		 991 while ((dst = *pprev) != NULL) {
992 if (!atomic_read(&dst->__refcnt)) {
993 *pprev = dst->next;
994 dst_free(dst);
1da177e4
LT		 995 } else {
996 pprev = &dst->next;
3d0f24a7 997 ++more;
1da177e4
LT		 998 }
999 }
1000
3b00944c 1001 spin_unlock_bh(&icmp6_dst_lock);
5d0bbeeb 1002
3d0f24a7 1003 return more;
1da177e4
LT		 1004}
1005
1e493d19
DM		 1006static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg),
1007 void *arg)
1008{
1009 struct dst_entry *dst, **pprev;
1010
1011 spin_lock_bh(&icmp6_dst_lock);
1012 pprev = &icmp6_dst_gc_list;
1013 while ((dst = *pprev) != NULL) {
1014 struct rt6_info *rt = (struct rt6_info *) dst;
1015 if (func(rt, arg)) {
1016 *pprev = dst->next;
1017 dst_free(dst);
1018 } else {
1019 pprev = &dst->next;
1020 }
1021 }
1022 spin_unlock_bh(&icmp6_dst_lock);
1023}
1024
569d3645 1025static int ip6_dst_gc(struct dst_ops *ops)
1da177e4 1026{
1da177e4 1027 unsigned long now = jiffies;
7019b78e
DL		 1028 struct net *net = ops->dst_net;
1029 int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval;
1030 int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size;
1031 int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity;
1032 int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout;
1033 unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc;
1034
1035 if (time_after(rt_last_gc + rt_min_interval, now) &&
1036 atomic_read(&ops->entries) <= rt_max_size)
1da177e4
LT		 1037 goto out;
1038
6891a346
BT		 1039 net->ipv6.ip6_rt_gc_expire++;
1040 fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net);
1041 net->ipv6.ip6_rt_last_gc = now;
7019b78e
DL		 1042 if (atomic_read(&ops->entries) < ops->gc_thresh)
1043 net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1;
1da177e4 1044out:
7019b78e
DL		 1045 net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity;
1046 return (atomic_read(&ops->entries) > rt_max_size);
1da177e4
LT		 1047}
1048
1049/* Clean host part of a prefix. Not necessary in radix tree,
1050 but results in cleaner routing tables.
1051
1052 Remove it only when all the things will work!
1053 */
1054
1055static int ipv6_get_mtu(struct net_device *dev)
1056{
1057 int mtu = IPV6_MIN_MTU;
1058 struct inet6_dev *idev;
1059
1060 idev = in6_dev_get(dev);
1061 if (idev) {
1062 mtu = idev->cnf.mtu6;
1063 in6_dev_put(idev);
1064 }
1065 return mtu;
1066}
1067
6b75d090 1068int ip6_dst_hoplimit(struct dst_entry *dst)
1da177e4 1069{
6b75d090
YH		 1070 int hoplimit = dst_metric(dst, RTAX_HOPLIMIT);
1071 if (hoplimit < 0) {
1072 struct net_device *dev = dst->dev;
1073 struct inet6_dev *idev = in6_dev_get(dev);
1074 if (idev) {
1075 hoplimit = idev->cnf.hop_limit;
1076 in6_dev_put(idev);
1077 } else
53b7997f 1078 hoplimit = dev_net(dev)->ipv6.devconf_all->hop_limit;
1da177e4
LT		 1079 }
1080 return hoplimit;
1081}
1082
1083/*
1084 *
1085 */
1086
86872cb5 1087int ip6_route_add(struct fib6_config *cfg)
1da177e4
LT		 1088{
1089 int err;
5578689a 1090 struct net *net = cfg->fc_nlinfo.nl_net;
1da177e4
LT		 1091 struct rt6_info *rt = NULL;
1092 struct net_device *dev = NULL;
1093 struct inet6_dev *idev = NULL;
c71099ac 1094 struct fib6_table *table;
1da177e4
LT		 1095 int addr_type;
1096
86872cb5 1097 if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128)
1da177e4
LT		 1098 return -EINVAL;
1099#ifndef CONFIG_IPV6_SUBTREES
86872cb5 1100 if (cfg->fc_src_len)
1da177e4
LT		 1101 return -EINVAL;
1102#endif
86872cb5 1103 if (cfg->fc_ifindex) {
1da177e4 1104 err = -ENODEV;
5578689a 1105 dev = dev_get_by_index(net, cfg->fc_ifindex);
1da177e4
LT		 1106 if (!dev)
1107 goto out;
1108 idev = in6_dev_get(dev);
1109 if (!idev)
1110 goto out;
1111 }
1112
86872cb5
TG		 1113 if (cfg->fc_metric == 0)
1114 cfg->fc_metric = IP6_RT_PRIO_USER;
1da177e4 1115
5578689a 1116 table = fib6_new_table(net, cfg->fc_table);
c71099ac
TG		 1117 if (table == NULL) {
1118 err = -ENOBUFS;
1119 goto out;
1120 }
1121
f2fc6a54 1122 rt = ip6_dst_alloc(net->ipv6.ip6_dst_ops);
1da177e4
LT		 1123
1124 if (rt == NULL) {
1125 err = -ENOMEM;
1126 goto out;
1127 }
1128
1129 rt->u.dst.obsolete = -1;
6f704992
YH		 1130 rt->rt6i_expires = (cfg->fc_flags & RTF_EXPIRES) ?
1131 jiffies + clock_t_to_jiffies(cfg->fc_expires) :
1132 0;
1da177e4 1133
86872cb5
TG		 1134 if (cfg->fc_protocol == RTPROT_UNSPEC)
1135 cfg->fc_protocol = RTPROT_BOOT;
1136 rt->rt6i_protocol = cfg->fc_protocol;
1137
1138 addr_type = ipv6_addr_type(&cfg->fc_dst);
1da177e4
LT		 1139
1140 if (addr_type & IPV6_ADDR_MULTICAST)
1141 rt->u.dst.input = ip6_mc_input;
1142 else
1143 rt->u.dst.input = ip6_forward;
1144
1145 rt->u.dst.output = ip6_output;
1146
86872cb5
TG		 1147 ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len);
1148 rt->rt6i_dst.plen = cfg->fc_dst_len;
1da177e4
LT		 1149 if (rt->rt6i_dst.plen == 128)
1150 rt->u.dst.flags = DST_HOST;
1151
1152#ifdef CONFIG_IPV6_SUBTREES
86872cb5
TG		 1153 ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len);
1154 rt->rt6i_src.plen = cfg->fc_src_len;
1da177e4
LT		 1155#endif
1156
86872cb5 1157 rt->rt6i_metric = cfg->fc_metric;
1da177e4
LT		 1158
1159 /* We cannot add true routes via loopback here,
1160 they would result in kernel looping; promote them to reject routes
1161 */
86872cb5 1162 if ((cfg->fc_flags & RTF_REJECT) ||
1da177e4
LT		 1163 (dev && (dev->flags&IFF_LOOPBACK) && !(addr_type&IPV6_ADDR_LOOPBACK))) {
1164 /* hold loopback dev/idev if we haven't done so. */
5578689a 1165 if (dev != net->loopback_dev) {
1da177e4
LT		 1166 if (dev) {
1167 dev_put(dev);
1168 in6_dev_put(idev);
1169 }
5578689a 1170 dev = net->loopback_dev;
1da177e4
LT		 1171 dev_hold(dev);
1172 idev = in6_dev_get(dev);
1173 if (!idev) {
1174 err = -ENODEV;
1175 goto out;
1176 }
1177 }
1178 rt->u.dst.output = ip6_pkt_discard_out;
1179 rt->u.dst.input = ip6_pkt_discard;
1180 rt->u.dst.error = -ENETUNREACH;
1181 rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP;
1182 goto install_route;
1183 }
1184
86872cb5 1185 if (cfg->fc_flags & RTF_GATEWAY) {
1da177e4
LT		 1186 struct in6_addr *gw_addr;
1187 int gwa_type;
1188
86872cb5
TG		 1189 gw_addr = &cfg->fc_gateway;
1190 ipv6_addr_copy(&rt->rt6i_gateway, gw_addr);
1da177e4
LT		 1191 gwa_type = ipv6_addr_type(gw_addr);
1192
1193 if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) {
1194 struct rt6_info *grt;
1195
1196 /* IPv6 strictly inhibits using not link-local
1197 addresses as nexthop address.
1198 Otherwise, router will not able to send redirects.
1199 It is very good, but in some (rare!) circumstances
1200 (SIT, PtP, NBMA NOARP links) it is handy to allow
1201 some exceptions. --ANK
1202 */
1203 err = -EINVAL;
1204 if (!(gwa_type&IPV6_ADDR_UNICAST))
1205 goto out;
1206
5578689a 1207 grt = rt6_lookup(net, gw_addr, NULL, cfg->fc_ifindex, 1);
1da177e4
LT		 1208
1209 err = -EHOSTUNREACH;
1210 if (grt == NULL)
1211 goto out;
1212 if (dev) {
1213 if (dev != grt->rt6i_dev) {
1214 dst_release(&grt->u.dst);
1215 goto out;
1216 }
1217 } else {
1218 dev = grt->rt6i_dev;
1219 idev = grt->rt6i_idev;
1220 dev_hold(dev);
1221 in6_dev_hold(grt->rt6i_idev);
1222 }
1223 if (!(grt->rt6i_flags&RTF_GATEWAY))
1224 err = 0;
1225 dst_release(&grt->u.dst);
1226
1227 if (err)
1228 goto out;
1229 }
1230 err = -EINVAL;
1231 if (dev == NULL || (dev->flags&IFF_LOOPBACK))
1232 goto out;
1233 }
1234
1235 err = -ENODEV;
1236 if (dev == NULL)
1237 goto out;
1238
86872cb5 1239 if (cfg->fc_flags & (RTF_GATEWAY | RTF_NONEXTHOP)) {
1da177e4
LT		 1240 rt->rt6i_nexthop = __neigh_lookup_errno(&nd_tbl, &rt->rt6i_gateway, dev);
1241 if (IS_ERR(rt->rt6i_nexthop)) {
1242 err = PTR_ERR(rt->rt6i_nexthop);
1243 rt->rt6i_nexthop = NULL;
1244 goto out;
1245 }
1246 }
1247
86872cb5 1248 rt->rt6i_flags = cfg->fc_flags;
1da177e4
LT		 1249
1250install_route:
86872cb5
TG		 1251 if (cfg->fc_mx) {
1252 struct nlattr *nla;
1253 int remaining;
1254
1255 nla_for_each_attr(nla, cfg->fc_mx, cfg->fc_mx_len, remaining) {
8f4c1f9b 1256 int type = nla_type(nla);
86872cb5
TG		 1257
1258 if (type) {
1259 if (type > RTAX_MAX) {
1da177e4
LT		 1260 err = -EINVAL;
1261 goto out;
1262 }
86872cb5
TG		 1263
1264 rt->u.dst.metrics[type - 1] = nla_get_u32(nla);
1da177e4 1265 }
1da177e4
LT		 1266 }
1267 }
1268
5ffc02a1 1269 if (dst_metric(&rt->u.dst, RTAX_HOPLIMIT) == 0)
1da177e4 1270 rt->u.dst.metrics[RTAX_HOPLIMIT-1] = -1;
1ca615fb 1271 if (!dst_mtu(&rt->u.dst))
1da177e4 1272 rt->u.dst.metrics[RTAX_MTU-1] = ipv6_get_mtu(dev);
5ffc02a1 1273 if (!dst_metric(&rt->u.dst, RTAX_ADVMSS))
5578689a 1274 rt->u.dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(net, dst_mtu(&rt->u.dst));
1da177e4
LT		 1275 rt->u.dst.dev = dev;
1276 rt->rt6i_idev = idev;
c71099ac 1277 rt->rt6i_table = table;
63152fc0 1278
c346dca1 1279 cfg->fc_nlinfo.nl_net = dev_net(dev);
63152fc0 1280
86872cb5 1281 return __ip6_ins_rt(rt, &cfg->fc_nlinfo);
1da177e4
LT		 1282
1283out:
1284 if (dev)
1285 dev_put(dev);
1286 if (idev)
1287 in6_dev_put(idev);
1288 if (rt)
40aa7b90 1289 dst_free(&rt->u.dst);
1da177e4
LT		 1290 return err;
1291}
1292
86872cb5 1293static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 1294{
1295 int err;
c71099ac 1296 struct fib6_table *table;
c346dca1 1297 struct net *net = dev_net(rt->rt6i_dev);
1da177e4 1298
8ed67789 1299 if (rt == net->ipv6.ip6_null_entry)
6c813a72
PM		 1300 return -ENOENT;
1301
c71099ac
TG		 1302 table = rt->rt6i_table;
1303 write_lock_bh(&table->tb6_lock);
1da177e4 1304
86872cb5 1305 err = fib6_del(rt, info);
1da177e4
LT		 1306 dst_release(&rt->u.dst);
1307
c71099ac 1308 write_unlock_bh(&table->tb6_lock);
1da177e4
LT		 1309
1310 return err;
1311}
1312
e0a1ad73
TG		 1313int ip6_del_rt(struct rt6_info *rt)
1314{
4d1169c1 1315 struct nl_info info = {
c346dca1 1316 .nl_net = dev_net(rt->rt6i_dev),
4d1169c1 1317 };
528c4ceb 1318 return __ip6_del_rt(rt, &info);
e0a1ad73
TG		 1319}
1320
86872cb5 1321static int ip6_route_del(struct fib6_config *cfg)
1da177e4 1322{
c71099ac 1323 struct fib6_table *table;
1da177e4
LT		 1324 struct fib6_node *fn;
1325 struct rt6_info *rt;
1326 int err = -ESRCH;
1327
5578689a 1328 table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table);
c71099ac
TG		 1329 if (table == NULL)
1330 return err;
1331
1332 read_lock_bh(&table->tb6_lock);
1da177e4 1333
c71099ac 1334 fn = fib6_locate(&table->tb6_root,
86872cb5
TG		 1335 &cfg->fc_dst, cfg->fc_dst_len,
1336 &cfg->fc_src, cfg->fc_src_len);
1ab1457c 1337
1da177e4 1338 if (fn) {
7cc48263 1339 for (rt = fn->leaf; rt; rt = rt->u.dst.rt6_next) {
86872cb5 1340 if (cfg->fc_ifindex &&
1da177e4 1341 (rt->rt6i_dev == NULL ||
86872cb5 1342 rt->rt6i_dev->ifindex != cfg->fc_ifindex))
1da177e4 1343 continue;
86872cb5
TG		 1344 if (cfg->fc_flags & RTF_GATEWAY &&
1345 !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway))
1da177e4 1346 continue;
86872cb5 1347 if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric)
1da177e4
LT		 1348 continue;
1349 dst_hold(&rt->u.dst);
c71099ac 1350 read_unlock_bh(&table->tb6_lock);
1da177e4 1351
86872cb5 1352 return __ip6_del_rt(rt, &cfg->fc_nlinfo);
1da177e4
LT		 1353 }
1354 }
c71099ac 1355 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 1356
1357 return err;
1358}
1359
1360/*
1361 * Handle redirects
1362 */
a6279458
YH		 1363struct ip6rd_flowi {
1364 struct flowi fl;
1365 struct in6_addr gateway;
1366};
1367
8ed67789
DL		 1368static struct rt6_info *__ip6_route_redirect(struct net *net,
1369 struct fib6_table *table,
a6279458
YH		 1370 struct flowi *fl,
1371 int flags)
1da177e4 1372{
a6279458
YH		 1373 struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl;
1374 struct rt6_info *rt;
e843b9e1 1375 struct fib6_node *fn;
c71099ac 1376
1da177e4 1377 /*
e843b9e1
YH		 1378 * Get the "current" route for this destination and
1379 * check if the redirect has come from approriate router.
1380 *
1381 * RFC 2461 specifies that redirects should only be
1382 * accepted if they come from the nexthop to the target.
1383 * Due to the way the routes are chosen, this notion
1384 * is a bit fuzzy and one might need to check all possible
1385 * routes.
1da177e4 1386 */
1da177e4 1387
c71099ac 1388 read_lock_bh(&table->tb6_lock);
a6279458 1389 fn = fib6_lookup(&table->tb6_root, &fl->fl6_dst, &fl->fl6_src);
e843b9e1 1390restart:
7cc48263 1391 for (rt = fn->leaf; rt; rt = rt->u.dst.rt6_next) {
e843b9e1
YH		 1392 /*
1393 * Current route is on-link; redirect is always invalid.
1394 *
1395 * Seems, previous statement is not true. It could
1396 * be node, which looks for us as on-link (f.e. proxy ndisc)
1397 * But then router serving it might decide, that we should
1398 * know truth 8)8) --ANK (980726).
1399 */
1400 if (rt6_check_expired(rt))
1401 continue;
1402 if (!(rt->rt6i_flags & RTF_GATEWAY))
1403 continue;
a6279458 1404 if (fl->oif != rt->rt6i_dev->ifindex)
e843b9e1 1405 continue;
a6279458 1406 if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway))
e843b9e1
YH		 1407 continue;
1408 break;
1409 }
a6279458 1410
cb15d9c2 1411 if (!rt)
8ed67789
DL		 1412 rt = net->ipv6.ip6_null_entry;
1413 BACKTRACK(net, &fl->fl6_src);
cb15d9c2 1414out:
a6279458
YH		 1415 dst_hold(&rt->u.dst);
1416
c71099ac 1417 read_unlock_bh(&table->tb6_lock);
e843b9e1 1418
a6279458
YH		 1419 return rt;
1420};
1421
1422static struct rt6_info *ip6_route_redirect(struct in6_addr *dest,
1423 struct in6_addr *src,
1424 struct in6_addr *gateway,
1425 struct net_device *dev)
1426{
adaa70bb 1427 int flags = RT6_LOOKUP_F_HAS_SADDR;
c346dca1 1428 struct net *net = dev_net(dev);
a6279458
YH		 1429 struct ip6rd_flowi rdfl = {
1430 .fl = {
1431 .oif = dev->ifindex,
1432 .nl_u = {
1433 .ip6_u = {
1434 .daddr = *dest,
1435 .saddr = *src,
1436 },
1437 },
1438 },
1439 .gateway = *gateway,
1440 };
adaa70bb
TG		 1441
1442 if (rt6_need_strict(dest))
1443 flags |= RT6_LOOKUP_F_IFACE;
a6279458 1444
5578689a 1445 return (struct rt6_info *)fib6_rule_lookup(net, (struct flowi *)&rdfl,
58f09b78 1446 flags, __ip6_route_redirect);
a6279458
YH		 1447}
1448
1449void rt6_redirect(struct in6_addr *dest, struct in6_addr *src,
1450 struct in6_addr *saddr,
1451 struct neighbour *neigh, u8 *lladdr, int on_link)
1452{
1453 struct rt6_info *rt, *nrt = NULL;
1454 struct netevent_redirect netevent;
c346dca1 1455 struct net *net = dev_net(neigh->dev);
a6279458
YH		 1456
1457 rt = ip6_route_redirect(dest, src, saddr, neigh->dev);
1458
8ed67789 1459 if (rt == net->ipv6.ip6_null_entry) {
1da177e4
LT		 1460 if (net_ratelimit())
1461 printk(KERN_DEBUG "rt6_redirect: source isn't a valid nexthop "
1462 "for redirect target\n");
a6279458 1463 goto out;
1da177e4
LT		 1464 }
1465
1da177e4
LT		 1466 /*
1467 * We have finally decided to accept it.
1468 */
1469
1ab1457c 1470 neigh_update(neigh, lladdr, NUD_STALE,
1da177e4
LT		 1471 NEIGH_UPDATE_F_WEAK_OVERRIDE|
1472 NEIGH_UPDATE_F_OVERRIDE|
1473 (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER|
1474 NEIGH_UPDATE_F_ISROUTER))
1475 );
1476
1477 /*
1478 * Redirect received -> path was valid.
1479 * Look, redirects are sent only in response to data packets,
1480 * so that this nexthop apparently is reachable. --ANK
1481 */
1482 dst_confirm(&rt->u.dst);
1483
1484 /* Duplicate redirect: silently ignore. */
1485 if (neigh == rt->u.dst.neighbour)
1486 goto out;
1487
1488 nrt = ip6_rt_copy(rt);
1489 if (nrt == NULL)
1490 goto out;
1491
1492 nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE;
1493 if (on_link)
1494 nrt->rt6i_flags &= ~RTF_GATEWAY;
1495
1496 ipv6_addr_copy(&nrt->rt6i_dst.addr, dest);
1497 nrt->rt6i_dst.plen = 128;
1498 nrt->u.dst.flags |= DST_HOST;
1499
1500 ipv6_addr_copy(&nrt->rt6i_gateway, (struct in6_addr*)neigh->primary_key);
1501 nrt->rt6i_nexthop = neigh_clone(neigh);
1502 /* Reset pmtu, it may be better */
1503 nrt->u.dst.metrics[RTAX_MTU-1] = ipv6_get_mtu(neigh->dev);
c346dca1 1504 nrt->u.dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(dev_net(neigh->dev),
5578689a 1505 dst_mtu(&nrt->u.dst));
1da177e4 1506
40e22e8f 1507 if (ip6_ins_rt(nrt))
1da177e4
LT		 1508 goto out;
1509
8d71740c
TT		 1510 netevent.old = &rt->u.dst;
1511 netevent.new = &nrt->u.dst;
1512 call_netevent_notifiers(NETEVENT_REDIRECT, &netevent);
1513
1da177e4 1514 if (rt->rt6i_flags&RTF_CACHE) {
e0a1ad73 1515 ip6_del_rt(rt);
1da177e4
LT		 1516 return;
1517 }
1518
1519out:
1ab1457c 1520 dst_release(&rt->u.dst);
1da177e4
LT		 1521 return;
1522}
1523
1524/*
1525 * Handle ICMP "packet too big" messages
1526 * i.e. Path MTU discovery
1527 */
1528
1529void rt6_pmtu_discovery(struct in6_addr *daddr, struct in6_addr *saddr,
1530 struct net_device *dev, u32 pmtu)
1531{
1532 struct rt6_info *rt, *nrt;
c346dca1 1533 struct net *net = dev_net(dev);
1da177e4
LT		 1534 int allfrag = 0;
1535
5578689a 1536 rt = rt6_lookup(net, daddr, saddr, dev->ifindex, 0);
1da177e4
LT		 1537 if (rt == NULL)
1538 return;
1539
1540 if (pmtu >= dst_mtu(&rt->u.dst))
1541 goto out;
1542
1543 if (pmtu < IPV6_MIN_MTU) {
1544 /*
1ab1457c 1545 * According to RFC2460, PMTU is set to the IPv6 Minimum Link
1da177e4
LT		 1546 * MTU (1280) and a fragment header should always be included
1547 * after a node receiving Too Big message reporting PMTU is
1548 * less than the IPv6 Minimum Link MTU.
1549 */
1550 pmtu = IPV6_MIN_MTU;
1551 allfrag = 1;
1552 }
1553
1554 /* New mtu received -> path was valid.
1555 They are sent only in response to data packets,
1556 so that this nexthop apparently is reachable. --ANK
1557 */
1558 dst_confirm(&rt->u.dst);
1559
1560 /* Host route. If it is static, it would be better
1561 not to override it, but add new one, so that
1562 when cache entry will expire old pmtu
1563 would return automatically.
1564 */
1565 if (rt->rt6i_flags & RTF_CACHE) {
1566 rt->u.dst.metrics[RTAX_MTU-1] = pmtu;
1567 if (allfrag)
1568 rt->u.dst.metrics[RTAX_FEATURES-1] |= RTAX_FEATURE_ALLFRAG;
5578689a 1569 dst_set_expires(&rt->u.dst, net->ipv6.sysctl.ip6_rt_mtu_expires);
1da177e4
LT		 1570 rt->rt6i_flags |= RTF_MODIFIED|RTF_EXPIRES;
1571 goto out;
1572 }
1573
1574 /* Network route.
1575 Two cases are possible:
1576 1. It is connected route. Action: COW
1577 2. It is gatewayed route or NONEXTHOP route. Action: clone it.
1578 */
d5315b50 1579 if (!rt->rt6i_nexthop && !(rt->rt6i_flags & RTF_NONEXTHOP))
a1e78363 1580 nrt = rt6_alloc_cow(rt, daddr, saddr);
d5315b50
YH		 1581 else
1582 nrt = rt6_alloc_clone(rt, daddr);
a1e78363 1583
d5315b50 1584 if (nrt) {
a1e78363
YH		 1585 nrt->u.dst.metrics[RTAX_MTU-1] = pmtu;
1586 if (allfrag)
1587 nrt->u.dst.metrics[RTAX_FEATURES-1] |= RTAX_FEATURE_ALLFRAG;
1588
1589 /* According to RFC 1981, detecting PMTU increase shouldn't be
1590 * happened within 5 mins, the recommended timer is 10 mins.
1591 * Here this route expiration time is set to ip6_rt_mtu_expires
1592 * which is 10 mins. After 10 mins the decreased pmtu is expired
1593 * and detecting PMTU increase will be automatically happened.
1594 */
5578689a 1595 dst_set_expires(&nrt->u.dst, net->ipv6.sysctl.ip6_rt_mtu_expires);
a1e78363
YH		 1596 nrt->rt6i_flags |= RTF_DYNAMIC|RTF_EXPIRES;
1597
40e22e8f 1598 ip6_ins_rt(nrt);
1da177e4 1599 }
1da177e4
LT		 1600out:
1601 dst_release(&rt->u.dst);
1602}
1603
1604/*
1605 * Misc support functions
1606 */
1607
1608static struct rt6_info * ip6_rt_copy(struct rt6_info *ort)
1609{
c346dca1 1610 struct net *net = dev_net(ort->rt6i_dev);
f2fc6a54 1611 struct rt6_info *rt = ip6_dst_alloc(net->ipv6.ip6_dst_ops);
1da177e4
LT		 1612
1613 if (rt) {
1614 rt->u.dst.input = ort->u.dst.input;
1615 rt->u.dst.output = ort->u.dst.output;
1616
1617 memcpy(rt->u.dst.metrics, ort->u.dst.metrics, RTAX_MAX*sizeof(u32));
22e1e4d8 1618 rt->u.dst.error = ort->u.dst.error;
1da177e4
LT		 1619 rt->u.dst.dev = ort->u.dst.dev;
1620 if (rt->u.dst.dev)
1621 dev_hold(rt->u.dst.dev);
1622 rt->rt6i_idev = ort->rt6i_idev;
1623 if (rt->rt6i_idev)
1624 in6_dev_hold(rt->rt6i_idev);
1625 rt->u.dst.lastuse = jiffies;
1626 rt->rt6i_expires = 0;
1627
1628 ipv6_addr_copy(&rt->rt6i_gateway, &ort->rt6i_gateway);
1629 rt->rt6i_flags = ort->rt6i_flags & ~RTF_EXPIRES;
1630 rt->rt6i_metric = 0;
1631
1632 memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key));
1633#ifdef CONFIG_IPV6_SUBTREES
1634 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1635#endif
c71099ac 1636 rt->rt6i_table = ort->rt6i_table;
1da177e4
LT		 1637 }
1638 return rt;
1639}
1640
70ceb4f5 1641#ifdef CONFIG_IPV6_ROUTE_INFO
efa2cea0
DL		 1642static struct rt6_info *rt6_get_route_info(struct net *net,
1643 struct in6_addr *prefix, int prefixlen,
70ceb4f5
YH		 1644 struct in6_addr *gwaddr, int ifindex)
1645{
1646 struct fib6_node *fn;
1647 struct rt6_info *rt = NULL;
c71099ac
TG		 1648 struct fib6_table *table;
1649
efa2cea0 1650 table = fib6_get_table(net, RT6_TABLE_INFO);
c71099ac
TG		 1651 if (table == NULL)
1652 return NULL;
70ceb4f5 1653
c71099ac
TG		 1654 write_lock_bh(&table->tb6_lock);
1655 fn = fib6_locate(&table->tb6_root, prefix ,prefixlen, NULL, 0);
70ceb4f5
YH		 1656 if (!fn)
1657 goto out;
1658
7cc48263 1659 for (rt = fn->leaf; rt; rt = rt->u.dst.rt6_next) {
70ceb4f5
YH		 1660 if (rt->rt6i_dev->ifindex != ifindex)
1661 continue;
1662 if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY))
1663 continue;
1664 if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr))
1665 continue;
1666 dst_hold(&rt->u.dst);
1667 break;
1668 }
1669out:
c71099ac 1670 write_unlock_bh(&table->tb6_lock);
70ceb4f5
YH		 1671 return rt;
1672}
1673
efa2cea0
DL		 1674static struct rt6_info *rt6_add_route_info(struct net *net,
1675 struct in6_addr *prefix, int prefixlen,
70ceb4f5
YH		 1676 struct in6_addr *gwaddr, int ifindex,
1677 unsigned pref)
1678{
86872cb5
TG		 1679 struct fib6_config cfg = {
1680 .fc_table = RT6_TABLE_INFO,
238fc7ea 1681 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 1682 .fc_ifindex = ifindex,
1683 .fc_dst_len = prefixlen,
1684 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO |
1685 RTF_UP | RTF_PREF(pref),
efa2cea0
DL		 1686 .fc_nlinfo.pid = 0,
1687 .fc_nlinfo.nlh = NULL,
1688 .fc_nlinfo.nl_net = net,
86872cb5
TG		 1689 };
1690
1691 ipv6_addr_copy(&cfg.fc_dst, prefix);
1692 ipv6_addr_copy(&cfg.fc_gateway, gwaddr);
70ceb4f5 1693
e317da96
YH		 1694 /* We should treat it as a default route if prefix length is 0. */
1695 if (!prefixlen)
86872cb5 1696 cfg.fc_flags |= RTF_DEFAULT;
70ceb4f5 1697
86872cb5 1698 ip6_route_add(&cfg);
70ceb4f5 1699
efa2cea0 1700 return rt6_get_route_info(net, prefix, prefixlen, gwaddr, ifindex);
70ceb4f5
YH		 1701}
1702#endif
1703
1da177e4 1704struct rt6_info *rt6_get_dflt_router(struct in6_addr *addr, struct net_device *dev)
1ab1457c 1705{
1da177e4 1706 struct rt6_info *rt;
c71099ac 1707 struct fib6_table *table;
1da177e4 1708
c346dca1 1709 table = fib6_get_table(dev_net(dev), RT6_TABLE_DFLT);
c71099ac
TG		 1710 if (table == NULL)
1711 return NULL;
1da177e4 1712
c71099ac 1713 write_lock_bh(&table->tb6_lock);
7cc48263 1714 for (rt = table->tb6_root.leaf; rt; rt=rt->u.dst.rt6_next) {
1da177e4 1715 if (dev == rt->rt6i_dev &&
045927ff 1716 ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) &&
1da177e4
LT		 1717 ipv6_addr_equal(&rt->rt6i_gateway, addr))
1718 break;
1719 }
1720 if (rt)
1721 dst_hold(&rt->u.dst);
c71099ac 1722 write_unlock_bh(&table->tb6_lock);
1da177e4
LT		 1723 return rt;
1724}
1725
1726struct rt6_info *rt6_add_dflt_router(struct in6_addr *gwaddr,
ebacaaa0
YH		 1727 struct net_device *dev,
1728 unsigned int pref)
1da177e4 1729{
86872cb5
TG		 1730 struct fib6_config cfg = {
1731 .fc_table = RT6_TABLE_DFLT,
238fc7ea 1732 .fc_metric = IP6_RT_PRIO_USER,
86872cb5
TG		 1733 .fc_ifindex = dev->ifindex,
1734 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT |
1735 RTF_UP | RTF_EXPIRES | RTF_PREF(pref),
5578689a
DL		 1736 .fc_nlinfo.pid = 0,
1737 .fc_nlinfo.nlh = NULL,
c346dca1 1738 .fc_nlinfo.nl_net = dev_net(dev),
86872cb5 1739 };
1da177e4 1740
86872cb5 1741 ipv6_addr_copy(&cfg.fc_gateway, gwaddr);
1da177e4 1742
86872cb5 1743 ip6_route_add(&cfg);
1da177e4 1744
1da177e4
LT		 1745 return rt6_get_dflt_router(gwaddr, dev);
1746}
1747
7b4da532 1748void rt6_purge_dflt_routers(struct net *net)
1da177e4
LT		 1749{
1750 struct rt6_info *rt;
c71099ac
TG		 1751 struct fib6_table *table;
1752
1753 /* NOTE: Keep consistent with rt6_get_dflt_router */
7b4da532 1754 table = fib6_get_table(net, RT6_TABLE_DFLT);
c71099ac
TG		 1755 if (table == NULL)
1756 return;
1da177e4
LT		 1757
1758restart:
c71099ac 1759 read_lock_bh(&table->tb6_lock);
7cc48263 1760 for (rt = table->tb6_root.leaf; rt; rt = rt->u.dst.rt6_next) {
1da177e4
LT		 1761 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF)) {
1762 dst_hold(&rt->u.dst);
c71099ac 1763 read_unlock_bh(&table->tb6_lock);
e0a1ad73 1764 ip6_del_rt(rt);
1da177e4
LT		 1765 goto restart;
1766 }
1767 }
c71099ac 1768 read_unlock_bh(&table->tb6_lock);
1da177e4
LT		 1769}
1770
5578689a
DL		 1771static void rtmsg_to_fib6_config(struct net *net,
1772 struct in6_rtmsg *rtmsg,
86872cb5
TG		 1773 struct fib6_config *cfg)
1774{
1775 memset(cfg, 0, sizeof(*cfg));
1776
1777 cfg->fc_table = RT6_TABLE_MAIN;
1778 cfg->fc_ifindex = rtmsg->rtmsg_ifindex;
1779 cfg->fc_metric = rtmsg->rtmsg_metric;
1780 cfg->fc_expires = rtmsg->rtmsg_info;
1781 cfg->fc_dst_len = rtmsg->rtmsg_dst_len;
1782 cfg->fc_src_len = rtmsg->rtmsg_src_len;
1783 cfg->fc_flags = rtmsg->rtmsg_flags;
1784
5578689a 1785 cfg->fc_nlinfo.nl_net = net;
f1243c2d 1786
86872cb5
TG		 1787 ipv6_addr_copy(&cfg->fc_dst, &rtmsg->rtmsg_dst);
1788 ipv6_addr_copy(&cfg->fc_src, &rtmsg->rtmsg_src);
1789 ipv6_addr_copy(&cfg->fc_gateway, &rtmsg->rtmsg_gateway);
1790}
1791
5578689a 1792int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg)
1da177e4 1793{
86872cb5 1794 struct fib6_config cfg;
1da177e4
LT		 1795 struct in6_rtmsg rtmsg;
1796 int err;
1797
1798 switch(cmd) {
1799 case SIOCADDRT: /* Add a route */
1800 case SIOCDELRT: /* Delete a route */
1801 if (!capable(CAP_NET_ADMIN))
1802 return -EPERM;
1803 err = copy_from_user(&rtmsg, arg,
1804 sizeof(struct in6_rtmsg));
1805 if (err)
1806 return -EFAULT;
86872cb5 1807
5578689a 1808 rtmsg_to_fib6_config(net, &rtmsg, &cfg);
86872cb5 1809
1da177e4
LT		 1810 rtnl_lock();
1811 switch (cmd) {
1812 case SIOCADDRT:
86872cb5 1813 err = ip6_route_add(&cfg);
1da177e4
LT		 1814 break;
1815 case SIOCDELRT:
86872cb5 1816 err = ip6_route_del(&cfg);
1da177e4
LT		 1817 break;
1818 default:
1819 err = -EINVAL;
1820 }
1821 rtnl_unlock();
1822
1823 return err;
3ff50b79 1824 }
1da177e4
LT		 1825
1826 return -EINVAL;
1827}
1828
1829/*
1830 * Drop the packet on the floor
1831 */
1832
50eb431d 1833static int ip6_pkt_drop(struct sk_buff *skb, int code, int ipstats_mib_noroutes)
1da177e4 1834{
612f09e8
YH		 1835 int type;
1836 switch (ipstats_mib_noroutes) {
1837 case IPSTATS_MIB_INNOROUTES:
0660e03f 1838 type = ipv6_addr_type(&ipv6_hdr(skb)->daddr);
612f09e8
YH		 1839 if (type == IPV6_ADDR_ANY || type == IPV6_ADDR_RESERVED) {
1840 IP6_INC_STATS(ip6_dst_idev(skb->dst), IPSTATS_MIB_INADDRERRORS);
1841 break;
1842 }
1843 /* FALLTHROUGH */
1844 case IPSTATS_MIB_OUTNOROUTES:
1845 IP6_INC_STATS(ip6_dst_idev(skb->dst), ipstats_mib_noroutes);
1846 break;
1847 }
9ce8ade0 1848 icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0, skb->dev);
1da177e4
LT		 1849 kfree_skb(skb);
1850 return 0;
1851}
1852
9ce8ade0
TG		 1853static int ip6_pkt_discard(struct sk_buff *skb)
1854{
612f09e8 1855 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 1856}
1857
20380731 1858static int ip6_pkt_discard_out(struct sk_buff *skb)
1da177e4
LT		 1859{
1860 skb->dev = skb->dst->dev;
612f09e8 1861 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES);
1da177e4
LT		 1862}
1863
6723ab54
DM		 1864#ifdef CONFIG_IPV6_MULTIPLE_TABLES
1865
9ce8ade0
TG		 1866static int ip6_pkt_prohibit(struct sk_buff *skb)
1867{
612f09e8 1868 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES);
9ce8ade0
TG		 1869}
1870
1871static int ip6_pkt_prohibit_out(struct sk_buff *skb)
1872{
1873 skb->dev = skb->dst->dev;
612f09e8 1874 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES);
9ce8ade0
TG		 1875}
1876
6723ab54
DM		 1877#endif
1878
1da177e4
LT		 1879/*
1880 * Allocate a dst for local (unicast / anycast) address.
1881 */
1882
1883struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
1884 const struct in6_addr *addr,
1885 int anycast)
1886{
c346dca1 1887 struct net *net = dev_net(idev->dev);
f2fc6a54 1888 struct rt6_info *rt = ip6_dst_alloc(net->ipv6.ip6_dst_ops);
1da177e4
LT		 1889
1890 if (rt == NULL)
1891 return ERR_PTR(-ENOMEM);
1892
5578689a 1893 dev_hold(net->loopback_dev);
1da177e4
LT		 1894 in6_dev_hold(idev);
1895
1896 rt->u.dst.flags = DST_HOST;
1897 rt->u.dst.input = ip6_input;
1898 rt->u.dst.output = ip6_output;
5578689a 1899 rt->rt6i_dev = net->loopback_dev;
1da177e4
LT		 1900 rt->rt6i_idev = idev;
1901 rt->u.dst.metrics[RTAX_MTU-1] = ipv6_get_mtu(rt->rt6i_dev);
5578689a 1902 rt->u.dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(net, dst_mtu(&rt->u.dst));
1da177e4
LT		 1903 rt->u.dst.metrics[RTAX_HOPLIMIT-1] = -1;
1904 rt->u.dst.obsolete = -1;
1905
1906 rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP;
58c4fb86
YH		 1907 if (anycast)
1908 rt->rt6i_flags |= RTF_ANYCAST;
1909 else
1da177e4
LT		 1910 rt->rt6i_flags |= RTF_LOCAL;
1911 rt->rt6i_nexthop = ndisc_get_neigh(rt->rt6i_dev, &rt->rt6i_gateway);
1912 if (rt->rt6i_nexthop == NULL) {
40aa7b90 1913 dst_free(&rt->u.dst);
1da177e4
LT		 1914 return ERR_PTR(-ENOMEM);
1915 }
1916
1917 ipv6_addr_copy(&rt->rt6i_dst.addr, addr);
1918 rt->rt6i_dst.plen = 128;
5578689a 1919 rt->rt6i_table = fib6_get_table(net, RT6_TABLE_LOCAL);
1da177e4
LT		 1920
1921 atomic_set(&rt->u.dst.__refcnt, 1);
1922
1923 return rt;
1924}
1925
8ed67789
DL		 1926struct arg_dev_net {
1927 struct net_device *dev;
1928 struct net *net;
1929};
1930
1da177e4
LT		 1931static int fib6_ifdown(struct rt6_info *rt, void *arg)
1932{
8ed67789
DL		 1933 struct net_device *dev = ((struct arg_dev_net *)arg)->dev;
1934 struct net *net = ((struct arg_dev_net *)arg)->net;
1935
1936 if (((void *)rt->rt6i_dev == dev || dev == NULL) &&
1937 rt != net->ipv6.ip6_null_entry) {
1da177e4
LT		 1938 RT6_TRACE("deleted by ifdown %p\n", rt);
1939 return -1;
1940 }
1941 return 0;
1942}
1943
f3db4851 1944void rt6_ifdown(struct net *net, struct net_device *dev)
1da177e4 1945{
8ed67789
DL		 1946 struct arg_dev_net adn = {
1947 .dev = dev,
1948 .net = net,
1949 };
1950
1951 fib6_clean_all(net, fib6_ifdown, 0, &adn);
1e493d19 1952 icmp6_clean_all(fib6_ifdown, &adn);
1da177e4
LT		 1953}
1954
1955struct rt6_mtu_change_arg
1956{
1957 struct net_device *dev;
1958 unsigned mtu;
1959};
1960
1961static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg)
1962{
1963 struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg;
1964 struct inet6_dev *idev;
c346dca1 1965 struct net *net = dev_net(arg->dev);
1da177e4
LT		 1966
1967 /* In IPv6 pmtu discovery is not optional,
1968 so that RTAX_MTU lock cannot disable it.
1969 We still use this lock to block changes
1970 caused by addrconf/ndisc.
1971 */
1972
1973 idev = __in6_dev_get(arg->dev);
1974 if (idev == NULL)
1975 return 0;
1976
1977 /* For administrative MTU increase, there is no way to discover
1978 IPv6 PMTU increase, so PMTU increase should be updated here.
1979 Since RFC 1981 doesn't include administrative MTU increase
1980 update PMTU increase is a MUST. (i.e. jumbo frame)
1981 */
1982 /*
1983 If new MTU is less than route PMTU, this new MTU will be the
1984 lowest MTU in the path, update the route PMTU to reflect PMTU
1985 decreases; if new MTU is greater than route PMTU, and the
1986 old MTU is the lowest MTU in the path, update the route PMTU
1987 to reflect the increase. In this case if the other nodes' MTU
1988 also have the lowest MTU, TOO BIG MESSAGE will be lead to
1989 PMTU discouvery.
1990 */
1991 if (rt->rt6i_dev == arg->dev &&
1992 !dst_metric_locked(&rt->u.dst, RTAX_MTU) &&
23717795 1993 (dst_mtu(&rt->u.dst) >= arg->mtu ||
1ab1457c 1994 (dst_mtu(&rt->u.dst) < arg->mtu &&
566cfd8f 1995 dst_mtu(&rt->u.dst) == idev->cnf.mtu6))) {
1da177e4 1996 rt->u.dst.metrics[RTAX_MTU-1] = arg->mtu;
5578689a 1997 rt->u.dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(net, arg->mtu);
566cfd8f 1998 }
1da177e4
LT		 1999 return 0;
2000}
2001
2002void rt6_mtu_change(struct net_device *dev, unsigned mtu)
2003{
c71099ac
TG		 2004 struct rt6_mtu_change_arg arg = {
2005 .dev = dev,
2006 .mtu = mtu,
2007 };
1da177e4 2008
c346dca1 2009 fib6_clean_all(dev_net(dev), rt6_mtu_change_route, 0, &arg);
1da177e4
LT		 2010}
2011
ef7c79ed 2012static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = {
5176f91e 2013 [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) },
86872cb5 2014 [RTA_OIF] = { .type = NLA_U32 },
ab364a6f 2015 [RTA_IIF] = { .type = NLA_U32 },
86872cb5
TG		 2016 [RTA_PRIORITY] = { .type = NLA_U32 },
2017 [RTA_METRICS] = { .type = NLA_NESTED },
2018};
2019
2020static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh,
2021 struct fib6_config *cfg)
1da177e4 2022{
86872cb5
TG		 2023 struct rtmsg *rtm;
2024 struct nlattr *tb[RTA_MAX+1];
2025 int err;
1da177e4 2026
86872cb5
TG		 2027 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2028 if (err < 0)
2029 goto errout;
1da177e4 2030
86872cb5
TG		 2031 err = -EINVAL;
2032 rtm = nlmsg_data(nlh);
2033 memset(cfg, 0, sizeof(*cfg));
2034
2035 cfg->fc_table = rtm->rtm_table;
2036 cfg->fc_dst_len = rtm->rtm_dst_len;
2037 cfg->fc_src_len = rtm->rtm_src_len;
2038 cfg->fc_flags = RTF_UP;
2039 cfg->fc_protocol = rtm->rtm_protocol;
2040
2041 if (rtm->rtm_type == RTN_UNREACHABLE)
2042 cfg->fc_flags |= RTF_REJECT;
2043
2044 cfg->fc_nlinfo.pid = NETLINK_CB(skb).pid;
2045 cfg->fc_nlinfo.nlh = nlh;
3b1e0a65 2046 cfg->fc_nlinfo.nl_net = sock_net(skb->sk);
86872cb5
TG		 2047
2048 if (tb[RTA_GATEWAY]) {
2049 nla_memcpy(&cfg->fc_gateway, tb[RTA_GATEWAY], 16);
2050 cfg->fc_flags |= RTF_GATEWAY;
1da177e4 2051 }
86872cb5
TG		 2052
2053 if (tb[RTA_DST]) {
2054 int plen = (rtm->rtm_dst_len + 7) >> 3;
2055
2056 if (nla_len(tb[RTA_DST]) < plen)
2057 goto errout;
2058
2059 nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen);
1da177e4 2060 }
86872cb5
TG		 2061
2062 if (tb[RTA_SRC]) {
2063 int plen = (rtm->rtm_src_len + 7) >> 3;
2064
2065 if (nla_len(tb[RTA_SRC]) < plen)
2066 goto errout;
2067
2068 nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen);
1da177e4 2069 }
86872cb5
TG		 2070
2071 if (tb[RTA_OIF])
2072 cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]);
2073
2074 if (tb[RTA_PRIORITY])
2075 cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]);
2076
2077 if (tb[RTA_METRICS]) {
2078 cfg->fc_mx = nla_data(tb[RTA_METRICS]);
2079 cfg->fc_mx_len = nla_len(tb[RTA_METRICS]);
1da177e4 2080 }
86872cb5
TG		 2081
2082 if (tb[RTA_TABLE])
2083 cfg->fc_table = nla_get_u32(tb[RTA_TABLE]);
2084
2085 err = 0;
2086errout:
2087 return err;
1da177e4
LT		 2088}
2089
c127ea2c 2090static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg)
1da177e4 2091{
86872cb5
TG		 2092 struct fib6_config cfg;
2093 int err;
1da177e4 2094
86872cb5
TG		 2095 err = rtm_to_fib6_config(skb, nlh, &cfg);
2096 if (err < 0)
2097 return err;
2098
2099 return ip6_route_del(&cfg);
1da177e4
LT		 2100}
2101
c127ea2c 2102static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg)
1da177e4 2103{
86872cb5
TG		 2104 struct fib6_config cfg;
2105 int err;
1da177e4 2106
86872cb5
TG		 2107 err = rtm_to_fib6_config(skb, nlh, &cfg);
2108 if (err < 0)
2109 return err;
2110
2111 return ip6_route_add(&cfg);
1da177e4
LT		 2112}
2113
339bf98f
TG		 2114static inline size_t rt6_nlmsg_size(void)
2115{
2116 return NLMSG_ALIGN(sizeof(struct rtmsg))
2117 + nla_total_size(16) /* RTA_SRC */
2118 + nla_total_size(16) /* RTA_DST */
2119 + nla_total_size(16) /* RTA_GATEWAY */
2120 + nla_total_size(16) /* RTA_PREFSRC */
2121 + nla_total_size(4) /* RTA_TABLE */
2122 + nla_total_size(4) /* RTA_IIF */
2123 + nla_total_size(4) /* RTA_OIF */
2124 + nla_total_size(4) /* RTA_PRIORITY */
6a2b9ce0 2125 + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */
339bf98f
TG		 2126 + nla_total_size(sizeof(struct rta_cacheinfo));
2127}
2128
191cd582
BH		 2129static int rt6_fill_node(struct net *net,
2130 struct sk_buff *skb, struct rt6_info *rt,
0d51aa80
JHS		 2131 struct in6_addr *dst, struct in6_addr *src,
2132 int iif, int type, u32 pid, u32 seq,
7bc570c8 2133 int prefix, int nowait, unsigned int flags)
1da177e4
LT		 2134{
2135 struct rtmsg *rtm;
2d7202bf 2136 struct nlmsghdr *nlh;
e3703b3d 2137 long expires;
9e762a4a 2138 u32 table;
1da177e4
LT		 2139
2140 if (prefix) { /* user wants prefix routes only */
2141 if (!(rt->rt6i_flags & RTF_PREFIX_RT)) {
2142 /* success since this is not a prefix route */
2143 return 1;
2144 }
2145 }
2146
2d7202bf
TG		 2147 nlh = nlmsg_put(skb, pid, seq, type, sizeof(*rtm), flags);
2148 if (nlh == NULL)
26932566 2149 return -EMSGSIZE;
2d7202bf
TG		 2150
2151 rtm = nlmsg_data(nlh);
1da177e4
LT		 2152 rtm->rtm_family = AF_INET6;
2153 rtm->rtm_dst_len = rt->rt6i_dst.plen;
2154 rtm->rtm_src_len = rt->rt6i_src.plen;
2155 rtm->rtm_tos = 0;
c71099ac 2156 if (rt->rt6i_table)
9e762a4a 2157 table = rt->rt6i_table->tb6_id;
c71099ac 2158 else
9e762a4a
PM		 2159 table = RT6_TABLE_UNSPEC;
2160 rtm->rtm_table = table;
2d7202bf 2161 NLA_PUT_U32(skb, RTA_TABLE, table);
1da177e4
LT		 2162 if (rt->rt6i_flags&RTF_REJECT)
2163 rtm->rtm_type = RTN_UNREACHABLE;
2164 else if (rt->rt6i_dev && (rt->rt6i_dev->flags&IFF_LOOPBACK))
2165 rtm->rtm_type = RTN_LOCAL;
2166 else
2167 rtm->rtm_type = RTN_UNICAST;
2168 rtm->rtm_flags = 0;
2169 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
2170 rtm->rtm_protocol = rt->rt6i_protocol;
2171 if (rt->rt6i_flags&RTF_DYNAMIC)
2172 rtm->rtm_protocol = RTPROT_REDIRECT;
2173 else if (rt->rt6i_flags & RTF_ADDRCONF)
2174 rtm->rtm_protocol = RTPROT_KERNEL;
2175 else if (rt->rt6i_flags&RTF_DEFAULT)
2176 rtm->rtm_protocol = RTPROT_RA;
2177
2178 if (rt->rt6i_flags&RTF_CACHE)
2179 rtm->rtm_flags |= RTM_F_CLONED;
2180
2181 if (dst) {
2d7202bf 2182 NLA_PUT(skb, RTA_DST, 16, dst);
1ab1457c 2183 rtm->rtm_dst_len = 128;
1da177e4 2184 } else if (rtm->rtm_dst_len)
2d7202bf 2185 NLA_PUT(skb, RTA_DST, 16, &rt->rt6i_dst.addr);
1da177e4
LT		 2186#ifdef CONFIG_IPV6_SUBTREES
2187 if (src) {
2d7202bf 2188 NLA_PUT(skb, RTA_SRC, 16, src);
1ab1457c 2189 rtm->rtm_src_len = 128;
1da177e4 2190 } else if (rtm->rtm_src_len)
2d7202bf 2191 NLA_PUT(skb, RTA_SRC, 16, &rt->rt6i_src.addr);
1da177e4 2192#endif
7bc570c8
YH		 2193 if (iif) {
2194#ifdef CONFIG_IPV6_MROUTE
2195 if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) {
2196 int err = ip6mr_get_route(skb, rtm, nowait);
2197 if (err <= 0) {
2198 if (!nowait) {
2199 if (err == 0)
2200 return 0;
2201 goto nla_put_failure;
2202 } else {
2203 if (err == -EMSGSIZE)
2204 goto nla_put_failure;
2205 }
2206 }
2207 } else
2208#endif
2209 NLA_PUT_U32(skb, RTA_IIF, iif);
2210 } else if (dst) {
5e0115e5 2211 struct inet6_dev *idev = ip6_dst_idev(&rt->u.dst);
1da177e4 2212 struct in6_addr saddr_buf;
191cd582 2213 if (ipv6_dev_get_saddr(net, idev ? idev->dev : NULL,
7cbca67c 2214 dst, 0, &saddr_buf) == 0)
2d7202bf 2215 NLA_PUT(skb, RTA_PREFSRC, 16, &saddr_buf);
1da177e4 2216 }
2d7202bf 2217
1da177e4 2218 if (rtnetlink_put_metrics(skb, rt->u.dst.metrics) < 0)
2d7202bf
TG		 2219 goto nla_put_failure;
2220
1da177e4 2221 if (rt->u.dst.neighbour)
2d7202bf
TG		 2222 NLA_PUT(skb, RTA_GATEWAY, 16, &rt->u.dst.neighbour->primary_key);
2223
1da177e4 2224 if (rt->u.dst.dev)
2d7202bf
TG		 2225 NLA_PUT_U32(skb, RTA_OIF, rt->rt6i_dev->ifindex);
2226
2227 NLA_PUT_U32(skb, RTA_PRIORITY, rt->rt6i_metric);
e3703b3d 2228
36e3deae
YH		 2229 if (!(rt->rt6i_flags & RTF_EXPIRES))
2230 expires = 0;
2231 else if (rt->rt6i_expires - jiffies < INT_MAX)
2232 expires = rt->rt6i_expires - jiffies;
2233 else
2234 expires = INT_MAX;
69cdf8f9 2235
e3703b3d
TG		 2236 if (rtnl_put_cacheinfo(skb, &rt->u.dst, 0, 0, 0,
2237 expires, rt->u.dst.error) < 0)
2238 goto nla_put_failure;
2d7202bf
TG		 2239
2240 return nlmsg_end(skb, nlh);
2241
2242nla_put_failure:
26932566
PM		 2243 nlmsg_cancel(skb, nlh);
2244 return -EMSGSIZE;
1da177e4
LT		 2245}
2246
1b43af54 2247int rt6_dump_route(struct rt6_info *rt, void *p_arg)
1da177e4
LT		 2248{
2249 struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg;
2250 int prefix;
2251
2d7202bf
TG		 2252 if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) {
2253 struct rtmsg *rtm = nlmsg_data(arg->cb->nlh);
1da177e4
LT		 2254 prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0;
2255 } else
2256 prefix = 0;
2257
191cd582
BH		 2258 return rt6_fill_node(arg->net,
2259 arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE,
1da177e4 2260 NETLINK_CB(arg->cb->skb).pid, arg->cb->nlh->nlmsg_seq,
7bc570c8 2261 prefix, 0, NLM_F_MULTI);
1da177e4
LT		 2262}
2263
c127ea2c 2264static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr* nlh, void *arg)
1da177e4 2265{
3b1e0a65 2266 struct net *net = sock_net(in_skb->sk);
ab364a6f
TG		 2267 struct nlattr *tb[RTA_MAX+1];
2268 struct rt6_info *rt;
1da177e4 2269 struct sk_buff *skb;
ab364a6f 2270 struct rtmsg *rtm;
1da177e4 2271 struct flowi fl;
ab364a6f 2272 int err, iif = 0;
1da177e4 2273
ab364a6f
TG		 2274 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2275 if (err < 0)
2276 goto errout;
1da177e4 2277
ab364a6f 2278 err = -EINVAL;
1da177e4 2279 memset(&fl, 0, sizeof(fl));
1da177e4 2280
ab364a6f
TG		 2281 if (tb[RTA_SRC]) {
2282 if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr))
2283 goto errout;
2284
2285 ipv6_addr_copy(&fl.fl6_src, nla_data(tb[RTA_SRC]));
2286 }
2287
2288 if (tb[RTA_DST]) {
2289 if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr))
2290 goto errout;
2291
2292 ipv6_addr_copy(&fl.fl6_dst, nla_data(tb[RTA_DST]));
2293 }
2294
2295 if (tb[RTA_IIF])
2296 iif = nla_get_u32(tb[RTA_IIF]);
2297
2298 if (tb[RTA_OIF])
2299 fl.oif = nla_get_u32(tb[RTA_OIF]);
1da177e4
LT		 2300
2301 if (iif) {
2302 struct net_device *dev;
5578689a 2303 dev = __dev_get_by_index(net, iif);
1da177e4
LT		 2304 if (!dev) {
2305 err = -ENODEV;
ab364a6f 2306 goto errout;
1da177e4
LT		 2307 }
2308 }
2309
ab364a6f
TG		 2310 skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
2311 if (skb == NULL) {
2312 err = -ENOBUFS;
2313 goto errout;
2314 }
1da177e4 2315
ab364a6f
TG		 2316 /* Reserve room for dummy headers, this skb can pass
2317 through good chunk of routing engine.
2318 */
459a98ed 2319 skb_reset_mac_header(skb);
ab364a6f 2320 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr));
1da177e4 2321
8a3edd80 2322 rt = (struct rt6_info*) ip6_route_output(net, NULL, &fl);
1da177e4
LT		 2323 skb->dst = &rt->u.dst;
2324
191cd582 2325 err = rt6_fill_node(net, skb, rt, &fl.fl6_dst, &fl.fl6_src, iif,
1da177e4 2326 RTM_NEWROUTE, NETLINK_CB(in_skb).pid,
7bc570c8 2327 nlh->nlmsg_seq, 0, 0, 0);
1da177e4 2328 if (err < 0) {
ab364a6f
TG		 2329 kfree_skb(skb);
2330 goto errout;
1da177e4
LT		 2331 }
2332
5578689a 2333 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).pid);
ab364a6f 2334errout:
1da177e4 2335 return err;
1da177e4
LT		 2336}
2337
86872cb5 2338void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info)
1da177e4
LT		 2339{
2340 struct sk_buff *skb;
5578689a 2341 struct net *net = info->nl_net;
528c4ceb
DL		 2342 u32 seq;
2343 int err;
2344
2345 err = -ENOBUFS;
2346 seq = info->nlh != NULL ? info->nlh->nlmsg_seq : 0;
86872cb5 2347
339bf98f 2348 skb = nlmsg_new(rt6_nlmsg_size(), gfp_any());
21713ebc
TG		 2349 if (skb == NULL)
2350 goto errout;
2351
191cd582 2352 err = rt6_fill_node(net, skb, rt, NULL, NULL, 0,
7bc570c8 2353 event, info->pid, seq, 0, 0, 0);
26932566
PM		 2354 if (err < 0) {
2355 /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */
2356 WARN_ON(err == -EMSGSIZE);
2357 kfree_skb(skb);
2358 goto errout;
2359 }
5578689a
DL		 2360 err = rtnl_notify(skb, net, info->pid, RTNLGRP_IPV6_ROUTE,
2361 info->nlh, gfp_any());
21713ebc
TG		 2362errout:
2363 if (err < 0)
5578689a 2364 rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err);
1da177e4
LT		 2365}
2366
8ed67789
DL		 2367static int ip6_route_dev_notify(struct notifier_block *this,
2368 unsigned long event, void *data)
2369{
2370 struct net_device *dev = (struct net_device *)data;
c346dca1 2371 struct net *net = dev_net(dev);
8ed67789
DL		 2372
2373 if (event == NETDEV_REGISTER && (dev->flags & IFF_LOOPBACK)) {
2374 net->ipv6.ip6_null_entry->u.dst.dev = dev;
2375 net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev);
2376#ifdef CONFIG_IPV6_MULTIPLE_TABLES
2377 net->ipv6.ip6_prohibit_entry->u.dst.dev = dev;
2378 net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev);
2379 net->ipv6.ip6_blk_hole_entry->u.dst.dev = dev;
2380 net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev);
2381#endif
2382 }
2383
2384 return NOTIFY_OK;
2385}
2386
1da177e4
LT		 2387/*
2388 * /proc
2389 */
2390
2391#ifdef CONFIG_PROC_FS
2392
2393#define RT6_INFO_LEN (32 + 4 + 32 + 4 + 32 + 40 + 5 + 1)
2394
2395struct rt6_proc_arg
2396{
2397 char *buffer;
2398 int offset;
2399 int length;
2400 int skip;
2401 int len;
2402};
2403
2404static int rt6_info_route(struct rt6_info *rt, void *p_arg)
2405{
33120b30 2406 struct seq_file *m = p_arg;
1da177e4 2407
33120b30
AD		 2408 seq_printf(m, NIP6_SEQFMT " %02x ", NIP6(rt->rt6i_dst.addr),
2409 rt->rt6i_dst.plen);
1da177e4
LT		 2410
2411#ifdef CONFIG_IPV6_SUBTREES
33120b30
AD		 2412 seq_printf(m, NIP6_SEQFMT " %02x ", NIP6(rt->rt6i_src.addr),
2413 rt->rt6i_src.plen);
1da177e4 2414#else
33120b30 2415 seq_puts(m, "00000000000000000000000000000000 00 ");
1da177e4
LT		 2416#endif
2417
2418 if (rt->rt6i_nexthop) {
33120b30
AD		 2419 seq_printf(m, NIP6_SEQFMT,
2420 NIP6(*((struct in6_addr *)rt->rt6i_nexthop->primary_key)));
1da177e4 2421 } else {
33120b30 2422 seq_puts(m, "00000000000000000000000000000000");
1da177e4 2423 }
33120b30
AD		 2424 seq_printf(m, " %08x %08x %08x %08x %8s\n",
2425 rt->rt6i_metric, atomic_read(&rt->u.dst.__refcnt),
2426 rt->u.dst.__use, rt->rt6i_flags,
2427 rt->rt6i_dev ? rt->rt6i_dev->name : "");
1da177e4
LT		 2428 return 0;
2429}
2430
33120b30 2431static int ipv6_route_show(struct seq_file *m, void *v)
1da177e4 2432{
f3db4851
DL		 2433 struct net *net = (struct net *)m->private;
2434 fib6_clean_all(net, rt6_info_route, 0, m);
33120b30
AD		 2435 return 0;
2436}
1da177e4 2437
33120b30
AD		 2438static int ipv6_route_open(struct inode *inode, struct file *file)
2439{
de05c557 2440 return single_open_net(inode, file, ipv6_route_show);
f3db4851
DL		 2441}
2442
33120b30
AD		 2443static const struct file_operations ipv6_route_proc_fops = {
2444 .owner = THIS_MODULE,
2445 .open = ipv6_route_open,
2446 .read = seq_read,
2447 .llseek = seq_lseek,
b6fcbdb4 2448 .release = single_release_net,
33120b30
AD		 2449};
2450
1da177e4
LT		 2451static int rt6_stats_seq_show(struct seq_file *seq, void *v)
2452{
69ddb805 2453 struct net *net = (struct net *)seq->private;
1da177e4 2454 seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n",
69ddb805
DL		 2455 net->ipv6.rt6_stats->fib_nodes,
2456 net->ipv6.rt6_stats->fib_route_nodes,
2457 net->ipv6.rt6_stats->fib_rt_alloc,
2458 net->ipv6.rt6_stats->fib_rt_entries,
2459 net->ipv6.rt6_stats->fib_rt_cache,
f2fc6a54 2460 atomic_read(&net->ipv6.ip6_dst_ops->entries),
69ddb805 2461 net->ipv6.rt6_stats->fib_discarded_routes);
1da177e4
LT		 2462
2463 return 0;
2464}
2465
2466static int rt6_stats_seq_open(struct inode *inode, struct file *file)
2467{
de05c557 2468 return single_open_net(inode, file, rt6_stats_seq_show);
69ddb805
DL		 2469}
2470
9a32144e 2471static const struct file_operations rt6_stats_seq_fops = {
1da177e4
LT		 2472 .owner = THIS_MODULE,
2473 .open = rt6_stats_seq_open,
2474 .read = seq_read,
2475 .llseek = seq_lseek,
b6fcbdb4 2476 .release = single_release_net,
1da177e4
LT		 2477};
2478#endif /* CONFIG_PROC_FS */
2479
2480#ifdef CONFIG_SYSCTL
2481
1da177e4
LT		 2482static
2483int ipv6_sysctl_rtcache_flush(ctl_table *ctl, int write, struct file * filp,
2484 void __user *buffer, size_t *lenp, loff_t *ppos)
2485{
5b7c931d
DL		 2486 struct net *net = current->nsproxy->net_ns;
2487 int delay = net->ipv6.sysctl.flush_delay;
1da177e4
LT		 2488 if (write) {
2489 proc_dointvec(ctl, write, filp, buffer, lenp, ppos);
5b7c931d 2490 fib6_run_gc(delay <= 0 ? ~0UL : (unsigned long)delay, net);
1da177e4
LT		 2491 return 0;
2492 } else
2493 return -EINVAL;
2494}
2495
760f2d01 2496ctl_table ipv6_route_table_template[] = {
1ab1457c 2497 {
1da177e4 2498 .procname = "flush",
4990509f 2499 .data = &init_net.ipv6.sysctl.flush_delay,
1da177e4 2500 .maxlen = sizeof(int),
89c8b3a1 2501 .mode = 0200,
1ab1457c 2502 .proc_handler = &ipv6_sysctl_rtcache_flush
1da177e4
LT		 2503 },
2504 {
2505 .ctl_name = NET_IPV6_ROUTE_GC_THRESH,
2506 .procname = "gc_thresh",
9a7ec3a9 2507 .data = &ip6_dst_ops_template.gc_thresh,
1da177e4
LT		 2508 .maxlen = sizeof(int),
2509 .mode = 0644,
1ab1457c 2510 .proc_handler = &proc_dointvec,
1da177e4
LT		 2511 },
2512 {
2513 .ctl_name = NET_IPV6_ROUTE_MAX_SIZE,
2514 .procname = "max_size",
4990509f 2515 .data = &init_net.ipv6.sysctl.ip6_rt_max_size,
1da177e4
LT		 2516 .maxlen = sizeof(int),
2517 .mode = 0644,
1ab1457c 2518 .proc_handler = &proc_dointvec,
1da177e4
LT		 2519 },
2520 {
2521 .ctl_name = NET_IPV6_ROUTE_GC_MIN_INTERVAL,
2522 .procname = "gc_min_interval",
4990509f 2523 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 2524 .maxlen = sizeof(int),
2525 .mode = 0644,
1ab1457c 2526 .proc_handler = &proc_dointvec_jiffies,
1da177e4
LT		 2527 .strategy = &sysctl_jiffies,
2528 },
2529 {
2530 .ctl_name = NET_IPV6_ROUTE_GC_TIMEOUT,
2531 .procname = "gc_timeout",
4990509f 2532 .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout,
1da177e4
LT		 2533 .maxlen = sizeof(int),
2534 .mode = 0644,
1ab1457c 2535 .proc_handler = &proc_dointvec_jiffies,
1da177e4
LT		 2536 .strategy = &sysctl_jiffies,
2537 },
2538 {
2539 .ctl_name = NET_IPV6_ROUTE_GC_INTERVAL,
2540 .procname = "gc_interval",
4990509f 2541 .data = &init_net.ipv6.sysctl.ip6_rt_gc_interval,
1da177e4
LT		 2542 .maxlen = sizeof(int),
2543 .mode = 0644,
1ab1457c 2544 .proc_handler = &proc_dointvec_jiffies,
1da177e4
LT		 2545 .strategy = &sysctl_jiffies,
2546 },
2547 {
2548 .ctl_name = NET_IPV6_ROUTE_GC_ELASTICITY,
2549 .procname = "gc_elasticity",
4990509f 2550 .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity,
1da177e4
LT		 2551 .maxlen = sizeof(int),
2552 .mode = 0644,
1ab1457c 2553 .proc_handler = &proc_dointvec_jiffies,
1da177e4
LT		 2554 .strategy = &sysctl_jiffies,
2555 },
2556 {
2557 .ctl_name = NET_IPV6_ROUTE_MTU_EXPIRES,
2558 .procname = "mtu_expires",
4990509f 2559 .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires,
1da177e4
LT		 2560 .maxlen = sizeof(int),
2561 .mode = 0644,
1ab1457c 2562 .proc_handler = &proc_dointvec_jiffies,
1da177e4
LT		 2563 .strategy = &sysctl_jiffies,
2564 },
2565 {
2566 .ctl_name = NET_IPV6_ROUTE_MIN_ADVMSS,
2567 .procname = "min_adv_mss",
4990509f 2568 .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss,
1da177e4
LT		 2569 .maxlen = sizeof(int),
2570 .mode = 0644,
1ab1457c 2571 .proc_handler = &proc_dointvec_jiffies,
1da177e4
LT		 2572 .strategy = &sysctl_jiffies,
2573 },
2574 {
2575 .ctl_name = NET_IPV6_ROUTE_GC_MIN_INTERVAL_MS,
2576 .procname = "gc_min_interval_ms",
4990509f 2577 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
1da177e4
LT		 2578 .maxlen = sizeof(int),
2579 .mode = 0644,
1ab1457c 2580 .proc_handler = &proc_dointvec_ms_jiffies,
1da177e4
LT		 2581 .strategy = &sysctl_ms_jiffies,
2582 },
2583 { .ctl_name = 0 }
2584};
2585
760f2d01
DL		 2586struct ctl_table *ipv6_route_sysctl_init(struct net *net)
2587{
2588 struct ctl_table *table;
2589
2590 table = kmemdup(ipv6_route_table_template,
2591 sizeof(ipv6_route_table_template),
2592 GFP_KERNEL);
5ee09105
YH		 2593
2594 if (table) {
2595 table[0].data = &net->ipv6.sysctl.flush_delay;
f2fc6a54 2596 table[1].data = &net->ipv6.ip6_dst_ops->gc_thresh;
5ee09105
YH		 2597 table[2].data = &net->ipv6.sysctl.ip6_rt_max_size;
2598 table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
2599 table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout;
2600 table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval;
2601 table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity;
2602 table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires;
2603 table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss;
2604 }
2605
760f2d01
DL		 2606 return table;
2607}
1da177e4
LT		 2608#endif
2609
cdb18761
DL		 2610static int ip6_route_net_init(struct net *net)
2611{
633d424b 2612 int ret = -ENOMEM;
8ed67789 2613
f2fc6a54
BT		 2614 net->ipv6.ip6_dst_ops = kmemdup(&ip6_dst_ops_template,
2615 sizeof(*net->ipv6.ip6_dst_ops),
2616 GFP_KERNEL);
2617 if (!net->ipv6.ip6_dst_ops)
2618 goto out;
48115bec 2619 net->ipv6.ip6_dst_ops->dst_net = hold_net(net);
f2fc6a54 2620
8ed67789
DL		 2621 net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template,
2622 sizeof(*net->ipv6.ip6_null_entry),
2623 GFP_KERNEL);
2624 if (!net->ipv6.ip6_null_entry)
f2fc6a54 2625 goto out_ip6_dst_ops;
8ed67789
DL		 2626 net->ipv6.ip6_null_entry->u.dst.path =
2627 (struct dst_entry *)net->ipv6.ip6_null_entry;
f2fc6a54 2628 net->ipv6.ip6_null_entry->u.dst.ops = net->ipv6.ip6_dst_ops;
8ed67789
DL		 2629
2630#ifdef CONFIG_IPV6_MULTIPLE_TABLES
2631 net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template,
2632 sizeof(*net->ipv6.ip6_prohibit_entry),
2633 GFP_KERNEL);
2634 if (!net->ipv6.ip6_prohibit_entry) {
2635 kfree(net->ipv6.ip6_null_entry);
2636 goto out;
2637 }
2638 net->ipv6.ip6_prohibit_entry->u.dst.path =
2639 (struct dst_entry *)net->ipv6.ip6_prohibit_entry;
f2fc6a54 2640 net->ipv6.ip6_prohibit_entry->u.dst.ops = net->ipv6.ip6_dst_ops;
8ed67789
DL		 2641
2642 net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template,
2643 sizeof(*net->ipv6.ip6_blk_hole_entry),
2644 GFP_KERNEL);
2645 if (!net->ipv6.ip6_blk_hole_entry) {
2646 kfree(net->ipv6.ip6_null_entry);
2647 kfree(net->ipv6.ip6_prohibit_entry);
2648 goto out;
2649 }
2650 net->ipv6.ip6_blk_hole_entry->u.dst.path =
2651 (struct dst_entry *)net->ipv6.ip6_blk_hole_entry;
f2fc6a54 2652 net->ipv6.ip6_blk_hole_entry->u.dst.ops = net->ipv6.ip6_dst_ops;
8ed67789
DL		 2653#endif
2654
cdb18761
DL		 2655#ifdef CONFIG_PROC_FS
2656 proc_net_fops_create(net, "ipv6_route", 0, &ipv6_route_proc_fops);
2657 proc_net_fops_create(net, "rt6_stats", S_IRUGO, &rt6_stats_seq_fops);
2658#endif
6891a346
BT		 2659 net->ipv6.ip6_rt_gc_expire = 30*HZ;
2660
8ed67789
DL		 2661 ret = 0;
2662out:
2663 return ret;
f2fc6a54
BT		 2664
2665out_ip6_dst_ops:
48115bec 2666 release_net(net->ipv6.ip6_dst_ops->dst_net);
f2fc6a54
BT		 2667 kfree(net->ipv6.ip6_dst_ops);
2668 goto out;
cdb18761
DL		 2669}
2670
2671static void ip6_route_net_exit(struct net *net)
2672{
2673#ifdef CONFIG_PROC_FS
2674 proc_net_remove(net, "ipv6_route");
2675 proc_net_remove(net, "rt6_stats");
2676#endif
8ed67789
DL		 2677 kfree(net->ipv6.ip6_null_entry);
2678#ifdef CONFIG_IPV6_MULTIPLE_TABLES
2679 kfree(net->ipv6.ip6_prohibit_entry);
2680 kfree(net->ipv6.ip6_blk_hole_entry);
2681#endif
48115bec 2682 release_net(net->ipv6.ip6_dst_ops->dst_net);
f2fc6a54 2683 kfree(net->ipv6.ip6_dst_ops);
cdb18761
DL		 2684}
2685
2686static struct pernet_operations ip6_route_net_ops = {
2687 .init = ip6_route_net_init,
2688 .exit = ip6_route_net_exit,
2689};
2690
8ed67789
DL		 2691static struct notifier_block ip6_route_dev_notifier = {
2692 .notifier_call = ip6_route_dev_notify,
2693 .priority = 0,
2694};
2695
433d49c3 2696int __init ip6_route_init(void)
1da177e4 2697{
433d49c3
DL		 2698 int ret;
2699
9a7ec3a9
DL		 2700 ret = -ENOMEM;
2701 ip6_dst_ops_template.kmem_cachep =
e5d679f3 2702 kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0,
f845ab6b 2703 SLAB_HWCACHE_ALIGN, NULL);
9a7ec3a9 2704 if (!ip6_dst_ops_template.kmem_cachep)
f2fc6a54 2705 goto out;;
14e50e57 2706
8ed67789
DL		 2707 ret = register_pernet_subsys(&ip6_route_net_ops);
2708 if (ret)
bdb3289f 2709 goto out_kmem_cache;
bdb3289f 2710
5dc121e9
AE		 2711 ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep;
2712
8ed67789
DL		 2713 /* Registering of the loopback is done before this portion of code,
2714 * the loopback reference in rt6_info will not be taken, do it
2715 * manually for init_net */
2716 init_net.ipv6.ip6_null_entry->u.dst.dev = init_net.loopback_dev;
2717 init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
2718 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
2719 init_net.ipv6.ip6_prohibit_entry->u.dst.dev = init_net.loopback_dev;
2720 init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
2721 init_net.ipv6.ip6_blk_hole_entry->u.dst.dev = init_net.loopback_dev;
2722 init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
2723 #endif
433d49c3
DL		 2724 ret = fib6_init();
2725 if (ret)
8ed67789 2726 goto out_register_subsys;
433d49c3 2727
433d49c3
DL		 2728 ret = xfrm6_init();
2729 if (ret)
cdb18761 2730 goto out_fib6_init;
c35b7e72 2731
433d49c3
DL		 2732 ret = fib6_rules_init();
2733 if (ret)
2734 goto xfrm6_init;
7e5449c2 2735
433d49c3
DL		 2736 ret = -ENOBUFS;
2737 if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL) ||
2738 __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL) ||
2739 __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL))
2740 goto fib6_rules_init;
c127ea2c 2741
8ed67789 2742 ret = register_netdevice_notifier(&ip6_route_dev_notifier);
cdb18761
DL		 2743 if (ret)
2744 goto fib6_rules_init;
8ed67789 2745
433d49c3
DL		 2746out:
2747 return ret;
2748
2749fib6_rules_init:
433d49c3
DL		 2750 fib6_rules_cleanup();
2751xfrm6_init:
433d49c3 2752 xfrm6_fini();
433d49c3 2753out_fib6_init:
433d49c3 2754 fib6_gc_cleanup();
8ed67789
DL		 2755out_register_subsys:
2756 unregister_pernet_subsys(&ip6_route_net_ops);
433d49c3 2757out_kmem_cache:
f2fc6a54 2758 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
433d49c3 2759 goto out;
1da177e4
LT		 2760}
2761
2762void ip6_route_cleanup(void)
2763{
8ed67789 2764 unregister_netdevice_notifier(&ip6_route_dev_notifier);
101367c2 2765 fib6_rules_cleanup();
1da177e4 2766 xfrm6_fini();
1da177e4 2767 fib6_gc_cleanup();
8ed67789 2768 unregister_pernet_subsys(&ip6_route_net_ops);
f2fc6a54 2769 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
1da177e4 2770}
Clone of davem's net-next-2.6 tree