Commit | Line | Data |
---|---|---|
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 | 28 | #include <linux/errno.h> |
bc3b2d7f | 29 | #include <linux/export.h> |
1da177e4 LT |
30 | #include <linux/types.h> |
31 | #include <linux/times.h> | |
32 | #include <linux/socket.h> | |
33 | #include <linux/sockios.h> | |
34 | #include <linux/net.h> | |
35 | #include <linux/route.h> | |
36 | #include <linux/netdevice.h> | |
37 | #include <linux/in6.h> | |
7bc570c8 | 38 | #include <linux/mroute6.h> |
1da177e4 | 39 | #include <linux/init.h> |
1da177e4 | 40 | #include <linux/if_arp.h> |
1da177e4 LT |
41 | #include <linux/proc_fs.h> |
42 | #include <linux/seq_file.h> | |
5b7c931d | 43 | #include <linux/nsproxy.h> |
5a0e3ad6 | 44 | #include <linux/slab.h> |
457c4cbc | 45 | #include <net/net_namespace.h> |
1da177e4 LT |
46 | #include <net/snmp.h> |
47 | #include <net/ipv6.h> | |
48 | #include <net/ip6_fib.h> | |
49 | #include <net/ip6_route.h> | |
50 | #include <net/ndisc.h> | |
51 | #include <net/addrconf.h> | |
52 | #include <net/tcp.h> | |
53 | #include <linux/rtnetlink.h> | |
54 | #include <net/dst.h> | |
55 | #include <net/xfrm.h> | |
8d71740c | 56 | #include <net/netevent.h> |
21713ebc | 57 | #include <net/netlink.h> |
1da177e4 LT |
58 | |
59 | #include <asm/uaccess.h> | |
60 | ||
61 | #ifdef CONFIG_SYSCTL | |
62 | #include <linux/sysctl.h> | |
63 | #endif | |
64 | ||
65 | /* Set to 3 to get tracing. */ | |
66 | #define RT6_DEBUG 2 | |
67 | ||
68 | #if RT6_DEBUG >= 3 | |
69 | #define RDBG(x) printk x | |
70 | #define RT6_TRACE(x...) printk(KERN_DEBUG x) | |
71 | #else | |
72 | #define RDBG(x) | |
73 | #define RT6_TRACE(x...) do { ; } while (0) | |
74 | #endif | |
75 | ||
21efcfa0 ED |
76 | static struct rt6_info *ip6_rt_copy(const struct rt6_info *ort, |
77 | const struct in6_addr *dest); | |
1da177e4 | 78 | static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie); |
0dbaee3b | 79 | static unsigned int ip6_default_advmss(const struct dst_entry *dst); |
ebb762f2 | 80 | static unsigned int ip6_mtu(const struct dst_entry *dst); |
1da177e4 LT |
81 | static struct dst_entry *ip6_negative_advice(struct dst_entry *); |
82 | static void ip6_dst_destroy(struct dst_entry *); | |
83 | static void ip6_dst_ifdown(struct dst_entry *, | |
84 | struct net_device *dev, int how); | |
569d3645 | 85 | static int ip6_dst_gc(struct dst_ops *ops); |
1da177e4 LT |
86 | |
87 | static int ip6_pkt_discard(struct sk_buff *skb); | |
88 | static int ip6_pkt_discard_out(struct sk_buff *skb); | |
89 | static void ip6_link_failure(struct sk_buff *skb); | |
90 | static void ip6_rt_update_pmtu(struct dst_entry *dst, u32 mtu); | |
91 | ||
70ceb4f5 | 92 | #ifdef CONFIG_IPV6_ROUTE_INFO |
efa2cea0 | 93 | static struct rt6_info *rt6_add_route_info(struct net *net, |
b71d1d42 ED |
94 | const struct in6_addr *prefix, int prefixlen, |
95 | const struct in6_addr *gwaddr, int ifindex, | |
70ceb4f5 | 96 | unsigned pref); |
efa2cea0 | 97 | static struct rt6_info *rt6_get_route_info(struct net *net, |
b71d1d42 ED |
98 | const struct in6_addr *prefix, int prefixlen, |
99 | const struct in6_addr *gwaddr, int ifindex); | |
70ceb4f5 YH |
100 | #endif |
101 | ||
06582540 DM |
102 | static u32 *ipv6_cow_metrics(struct dst_entry *dst, unsigned long old) |
103 | { | |
104 | struct rt6_info *rt = (struct rt6_info *) dst; | |
105 | struct inet_peer *peer; | |
106 | u32 *p = NULL; | |
107 | ||
8e2ec639 YZ |
108 | if (!(rt->dst.flags & DST_HOST)) |
109 | return NULL; | |
110 | ||
06582540 DM |
111 | if (!rt->rt6i_peer) |
112 | rt6_bind_peer(rt, 1); | |
113 | ||
114 | peer = rt->rt6i_peer; | |
115 | if (peer) { | |
116 | u32 *old_p = __DST_METRICS_PTR(old); | |
117 | unsigned long prev, new; | |
118 | ||
119 | p = peer->metrics; | |
120 | if (inet_metrics_new(peer)) | |
121 | memcpy(p, old_p, sizeof(u32) * RTAX_MAX); | |
122 | ||
123 | new = (unsigned long) p; | |
124 | prev = cmpxchg(&dst->_metrics, old, new); | |
125 | ||
126 | if (prev != old) { | |
127 | p = __DST_METRICS_PTR(prev); | |
128 | if (prev & DST_METRICS_READ_ONLY) | |
129 | p = NULL; | |
130 | } | |
131 | } | |
132 | return p; | |
133 | } | |
134 | ||
d3aaeb38 DM |
135 | static struct neighbour *ip6_neigh_lookup(const struct dst_entry *dst, const void *daddr) |
136 | { | |
137 | return __neigh_lookup_errno(&nd_tbl, daddr, dst->dev); | |
138 | } | |
139 | ||
9a7ec3a9 | 140 | static struct dst_ops ip6_dst_ops_template = { |
1da177e4 | 141 | .family = AF_INET6, |
09640e63 | 142 | .protocol = cpu_to_be16(ETH_P_IPV6), |
1da177e4 LT |
143 | .gc = ip6_dst_gc, |
144 | .gc_thresh = 1024, | |
145 | .check = ip6_dst_check, | |
0dbaee3b | 146 | .default_advmss = ip6_default_advmss, |
ebb762f2 | 147 | .mtu = ip6_mtu, |
06582540 | 148 | .cow_metrics = ipv6_cow_metrics, |
1da177e4 LT |
149 | .destroy = ip6_dst_destroy, |
150 | .ifdown = ip6_dst_ifdown, | |
151 | .negative_advice = ip6_negative_advice, | |
152 | .link_failure = ip6_link_failure, | |
153 | .update_pmtu = ip6_rt_update_pmtu, | |
1ac06e03 | 154 | .local_out = __ip6_local_out, |
d3aaeb38 | 155 | .neigh_lookup = ip6_neigh_lookup, |
1da177e4 LT |
156 | }; |
157 | ||
ebb762f2 | 158 | static unsigned int ip6_blackhole_mtu(const struct dst_entry *dst) |
ec831ea7 | 159 | { |
618f9bc7 SK |
160 | unsigned int mtu = dst_metric_raw(dst, RTAX_MTU); |
161 | ||
162 | return mtu ? : dst->dev->mtu; | |
ec831ea7 RD |
163 | } |
164 | ||
14e50e57 DM |
165 | static void ip6_rt_blackhole_update_pmtu(struct dst_entry *dst, u32 mtu) |
166 | { | |
167 | } | |
168 | ||
0972ddb2 HB |
169 | static u32 *ip6_rt_blackhole_cow_metrics(struct dst_entry *dst, |
170 | unsigned long old) | |
171 | { | |
172 | return NULL; | |
173 | } | |
174 | ||
14e50e57 DM |
175 | static struct dst_ops ip6_dst_blackhole_ops = { |
176 | .family = AF_INET6, | |
09640e63 | 177 | .protocol = cpu_to_be16(ETH_P_IPV6), |
14e50e57 DM |
178 | .destroy = ip6_dst_destroy, |
179 | .check = ip6_dst_check, | |
ebb762f2 | 180 | .mtu = ip6_blackhole_mtu, |
214f45c9 | 181 | .default_advmss = ip6_default_advmss, |
14e50e57 | 182 | .update_pmtu = ip6_rt_blackhole_update_pmtu, |
0972ddb2 | 183 | .cow_metrics = ip6_rt_blackhole_cow_metrics, |
d3aaeb38 | 184 | .neigh_lookup = ip6_neigh_lookup, |
14e50e57 DM |
185 | }; |
186 | ||
62fa8a84 DM |
187 | static const u32 ip6_template_metrics[RTAX_MAX] = { |
188 | [RTAX_HOPLIMIT - 1] = 255, | |
189 | }; | |
190 | ||
bdb3289f | 191 | static struct rt6_info ip6_null_entry_template = { |
d8d1f30b CG |
192 | .dst = { |
193 | .__refcnt = ATOMIC_INIT(1), | |
194 | .__use = 1, | |
195 | .obsolete = -1, | |
196 | .error = -ENETUNREACH, | |
d8d1f30b CG |
197 | .input = ip6_pkt_discard, |
198 | .output = ip6_pkt_discard_out, | |
1da177e4 LT |
199 | }, |
200 | .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP), | |
4f724279 | 201 | .rt6i_protocol = RTPROT_KERNEL, |
1da177e4 LT |
202 | .rt6i_metric = ~(u32) 0, |
203 | .rt6i_ref = ATOMIC_INIT(1), | |
204 | }; | |
205 | ||
101367c2 TG |
206 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES |
207 | ||
6723ab54 DM |
208 | static int ip6_pkt_prohibit(struct sk_buff *skb); |
209 | static int ip6_pkt_prohibit_out(struct sk_buff *skb); | |
6723ab54 | 210 | |
280a34c8 | 211 | static struct rt6_info ip6_prohibit_entry_template = { |
d8d1f30b CG |
212 | .dst = { |
213 | .__refcnt = ATOMIC_INIT(1), | |
214 | .__use = 1, | |
215 | .obsolete = -1, | |
216 | .error = -EACCES, | |
d8d1f30b CG |
217 | .input = ip6_pkt_prohibit, |
218 | .output = ip6_pkt_prohibit_out, | |
101367c2 TG |
219 | }, |
220 | .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP), | |
4f724279 | 221 | .rt6i_protocol = RTPROT_KERNEL, |
101367c2 TG |
222 | .rt6i_metric = ~(u32) 0, |
223 | .rt6i_ref = ATOMIC_INIT(1), | |
224 | }; | |
225 | ||
bdb3289f | 226 | static struct rt6_info ip6_blk_hole_entry_template = { |
d8d1f30b CG |
227 | .dst = { |
228 | .__refcnt = ATOMIC_INIT(1), | |
229 | .__use = 1, | |
230 | .obsolete = -1, | |
231 | .error = -EINVAL, | |
d8d1f30b CG |
232 | .input = dst_discard, |
233 | .output = dst_discard, | |
101367c2 TG |
234 | }, |
235 | .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP), | |
4f724279 | 236 | .rt6i_protocol = RTPROT_KERNEL, |
101367c2 TG |
237 | .rt6i_metric = ~(u32) 0, |
238 | .rt6i_ref = ATOMIC_INIT(1), | |
239 | }; | |
240 | ||
241 | #endif | |
242 | ||
1da177e4 | 243 | /* allocate dst with ip6_dst_ops */ |
5c1e6aa3 | 244 | static inline struct rt6_info *ip6_dst_alloc(struct dst_ops *ops, |
957c665f DM |
245 | struct net_device *dev, |
246 | int flags) | |
1da177e4 | 247 | { |
957c665f | 248 | struct rt6_info *rt = dst_alloc(ops, dev, 0, 0, flags); |
cf911662 | 249 | |
38308473 | 250 | if (rt) |
fbe58186 | 251 | memset(&rt->rt6i_table, 0, |
38308473 | 252 | sizeof(*rt) - sizeof(struct dst_entry)); |
cf911662 DM |
253 | |
254 | return rt; | |
1da177e4 LT |
255 | } |
256 | ||
257 | static void ip6_dst_destroy(struct dst_entry *dst) | |
258 | { | |
259 | struct rt6_info *rt = (struct rt6_info *)dst; | |
260 | struct inet6_dev *idev = rt->rt6i_idev; | |
b3419363 | 261 | struct inet_peer *peer = rt->rt6i_peer; |
1da177e4 | 262 | |
8e2ec639 YZ |
263 | if (!(rt->dst.flags & DST_HOST)) |
264 | dst_destroy_metrics_generic(dst); | |
265 | ||
38308473 | 266 | if (idev) { |
1da177e4 LT |
267 | rt->rt6i_idev = NULL; |
268 | in6_dev_put(idev); | |
1ab1457c | 269 | } |
b3419363 | 270 | if (peer) { |
b3419363 DM |
271 | rt->rt6i_peer = NULL; |
272 | inet_putpeer(peer); | |
273 | } | |
274 | } | |
275 | ||
6431cbc2 DM |
276 | static atomic_t __rt6_peer_genid = ATOMIC_INIT(0); |
277 | ||
278 | static u32 rt6_peer_genid(void) | |
279 | { | |
280 | return atomic_read(&__rt6_peer_genid); | |
281 | } | |
282 | ||
b3419363 DM |
283 | void rt6_bind_peer(struct rt6_info *rt, int create) |
284 | { | |
285 | struct inet_peer *peer; | |
286 | ||
b3419363 DM |
287 | peer = inet_getpeer_v6(&rt->rt6i_dst.addr, create); |
288 | if (peer && cmpxchg(&rt->rt6i_peer, NULL, peer) != NULL) | |
289 | inet_putpeer(peer); | |
6431cbc2 DM |
290 | else |
291 | rt->rt6i_peer_genid = rt6_peer_genid(); | |
1da177e4 LT |
292 | } |
293 | ||
294 | static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev, | |
295 | int how) | |
296 | { | |
297 | struct rt6_info *rt = (struct rt6_info *)dst; | |
298 | struct inet6_dev *idev = rt->rt6i_idev; | |
5a3e55d6 | 299 | struct net_device *loopback_dev = |
c346dca1 | 300 | dev_net(dev)->loopback_dev; |
1da177e4 | 301 | |
38308473 | 302 | if (dev != loopback_dev && idev && idev->dev == dev) { |
5a3e55d6 DL |
303 | struct inet6_dev *loopback_idev = |
304 | in6_dev_get(loopback_dev); | |
38308473 | 305 | if (loopback_idev) { |
1da177e4 LT |
306 | rt->rt6i_idev = loopback_idev; |
307 | in6_dev_put(idev); | |
308 | } | |
309 | } | |
310 | } | |
311 | ||
312 | static __inline__ int rt6_check_expired(const struct rt6_info *rt) | |
313 | { | |
a02cec21 ED |
314 | return (rt->rt6i_flags & RTF_EXPIRES) && |
315 | time_after(jiffies, rt->rt6i_expires); | |
1da177e4 LT |
316 | } |
317 | ||
b71d1d42 | 318 | static inline int rt6_need_strict(const struct in6_addr *daddr) |
c71099ac | 319 | { |
a02cec21 ED |
320 | return ipv6_addr_type(daddr) & |
321 | (IPV6_ADDR_MULTICAST | IPV6_ADDR_LINKLOCAL | IPV6_ADDR_LOOPBACK); | |
c71099ac TG |
322 | } |
323 | ||
1da177e4 | 324 | /* |
c71099ac | 325 | * Route lookup. Any table->tb6_lock is implied. |
1da177e4 LT |
326 | */ |
327 | ||
8ed67789 DL |
328 | static inline struct rt6_info *rt6_device_match(struct net *net, |
329 | struct rt6_info *rt, | |
b71d1d42 | 330 | const struct in6_addr *saddr, |
1da177e4 | 331 | int oif, |
d420895e | 332 | int flags) |
1da177e4 LT |
333 | { |
334 | struct rt6_info *local = NULL; | |
335 | struct rt6_info *sprt; | |
336 | ||
dd3abc4e YH |
337 | if (!oif && ipv6_addr_any(saddr)) |
338 | goto out; | |
339 | ||
d8d1f30b | 340 | for (sprt = rt; sprt; sprt = sprt->dst.rt6_next) { |
dd3abc4e YH |
341 | struct net_device *dev = sprt->rt6i_dev; |
342 | ||
343 | if (oif) { | |
1da177e4 LT |
344 | if (dev->ifindex == oif) |
345 | return sprt; | |
346 | if (dev->flags & IFF_LOOPBACK) { | |
38308473 | 347 | if (!sprt->rt6i_idev || |
1da177e4 | 348 | sprt->rt6i_idev->dev->ifindex != oif) { |
d420895e | 349 | if (flags & RT6_LOOKUP_F_IFACE && oif) |
1da177e4 | 350 | continue; |
1ab1457c | 351 | if (local && (!oif || |
1da177e4 LT |
352 | local->rt6i_idev->dev->ifindex == oif)) |
353 | continue; | |
354 | } | |
355 | local = sprt; | |
356 | } | |
dd3abc4e YH |
357 | } else { |
358 | if (ipv6_chk_addr(net, saddr, dev, | |
359 | flags & RT6_LOOKUP_F_IFACE)) | |
360 | return sprt; | |
1da177e4 | 361 | } |
dd3abc4e | 362 | } |
1da177e4 | 363 | |
dd3abc4e | 364 | if (oif) { |
1da177e4 LT |
365 | if (local) |
366 | return local; | |
367 | ||
d420895e | 368 | if (flags & RT6_LOOKUP_F_IFACE) |
8ed67789 | 369 | return net->ipv6.ip6_null_entry; |
1da177e4 | 370 | } |
dd3abc4e | 371 | out: |
1da177e4 LT |
372 | return rt; |
373 | } | |
374 | ||
27097255 YH |
375 | #ifdef CONFIG_IPV6_ROUTER_PREF |
376 | static void rt6_probe(struct rt6_info *rt) | |
377 | { | |
f2c31e32 | 378 | struct neighbour *neigh; |
27097255 YH |
379 | /* |
380 | * Okay, this does not seem to be appropriate | |
381 | * for now, however, we need to check if it | |
382 | * is really so; aka Router Reachability Probing. | |
383 | * | |
384 | * Router Reachability Probe MUST be rate-limited | |
385 | * to no more than one per minute. | |
386 | */ | |
f2c31e32 | 387 | rcu_read_lock(); |
27217455 | 388 | neigh = rt ? dst_get_neighbour_noref(&rt->dst) : NULL; |
27097255 | 389 | if (!neigh || (neigh->nud_state & NUD_VALID)) |
f2c31e32 | 390 | goto out; |
27097255 YH |
391 | read_lock_bh(&neigh->lock); |
392 | if (!(neigh->nud_state & NUD_VALID) && | |
52e16356 | 393 | time_after(jiffies, neigh->updated + rt->rt6i_idev->cnf.rtr_probe_interval)) { |
27097255 YH |
394 | struct in6_addr mcaddr; |
395 | struct in6_addr *target; | |
396 | ||
397 | neigh->updated = jiffies; | |
398 | read_unlock_bh(&neigh->lock); | |
399 | ||
400 | target = (struct in6_addr *)&neigh->primary_key; | |
401 | addrconf_addr_solict_mult(target, &mcaddr); | |
402 | ndisc_send_ns(rt->rt6i_dev, NULL, target, &mcaddr, NULL); | |
f2c31e32 | 403 | } else { |
27097255 | 404 | read_unlock_bh(&neigh->lock); |
f2c31e32 ED |
405 | } |
406 | out: | |
407 | rcu_read_unlock(); | |
27097255 YH |
408 | } |
409 | #else | |
410 | static inline void rt6_probe(struct rt6_info *rt) | |
411 | { | |
27097255 YH |
412 | } |
413 | #endif | |
414 | ||
1da177e4 | 415 | /* |
554cfb7e | 416 | * Default Router Selection (RFC 2461 6.3.6) |
1da177e4 | 417 | */ |
b6f99a21 | 418 | static inline int rt6_check_dev(struct rt6_info *rt, int oif) |
554cfb7e YH |
419 | { |
420 | struct net_device *dev = rt->rt6i_dev; | |
161980f4 | 421 | if (!oif || dev->ifindex == oif) |
554cfb7e | 422 | return 2; |
161980f4 DM |
423 | if ((dev->flags & IFF_LOOPBACK) && |
424 | rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif) | |
425 | return 1; | |
426 | return 0; | |
554cfb7e | 427 | } |
1da177e4 | 428 | |
b6f99a21 | 429 | static inline int rt6_check_neigh(struct rt6_info *rt) |
1da177e4 | 430 | { |
f2c31e32 | 431 | struct neighbour *neigh; |
398bcbeb | 432 | int m; |
f2c31e32 ED |
433 | |
434 | rcu_read_lock(); | |
27217455 | 435 | neigh = dst_get_neighbour_noref(&rt->dst); |
4d0c5911 YH |
436 | if (rt->rt6i_flags & RTF_NONEXTHOP || |
437 | !(rt->rt6i_flags & RTF_GATEWAY)) | |
438 | m = 1; | |
439 | else if (neigh) { | |
554cfb7e YH |
440 | read_lock_bh(&neigh->lock); |
441 | if (neigh->nud_state & NUD_VALID) | |
4d0c5911 | 442 | m = 2; |
398bcbeb YH |
443 | #ifdef CONFIG_IPV6_ROUTER_PREF |
444 | else if (neigh->nud_state & NUD_FAILED) | |
445 | m = 0; | |
446 | #endif | |
447 | else | |
ea73ee23 | 448 | m = 1; |
554cfb7e | 449 | read_unlock_bh(&neigh->lock); |
398bcbeb YH |
450 | } else |
451 | m = 0; | |
f2c31e32 | 452 | rcu_read_unlock(); |
554cfb7e | 453 | return m; |
1da177e4 LT |
454 | } |
455 | ||
554cfb7e YH |
456 | static int rt6_score_route(struct rt6_info *rt, int oif, |
457 | int strict) | |
1da177e4 | 458 | { |
4d0c5911 | 459 | int m, n; |
1ab1457c | 460 | |
4d0c5911 | 461 | m = rt6_check_dev(rt, oif); |
77d16f45 | 462 | if (!m && (strict & RT6_LOOKUP_F_IFACE)) |
554cfb7e | 463 | return -1; |
ebacaaa0 YH |
464 | #ifdef CONFIG_IPV6_ROUTER_PREF |
465 | m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2; | |
466 | #endif | |
4d0c5911 | 467 | n = rt6_check_neigh(rt); |
557e92ef | 468 | if (!n && (strict & RT6_LOOKUP_F_REACHABLE)) |
554cfb7e YH |
469 | return -1; |
470 | return m; | |
471 | } | |
472 | ||
f11e6659 DM |
473 | static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict, |
474 | int *mpri, struct rt6_info *match) | |
554cfb7e | 475 | { |
f11e6659 DM |
476 | int m; |
477 | ||
478 | if (rt6_check_expired(rt)) | |
479 | goto out; | |
480 | ||
481 | m = rt6_score_route(rt, oif, strict); | |
482 | if (m < 0) | |
483 | goto out; | |
484 | ||
485 | if (m > *mpri) { | |
486 | if (strict & RT6_LOOKUP_F_REACHABLE) | |
487 | rt6_probe(match); | |
488 | *mpri = m; | |
489 | match = rt; | |
490 | } else if (strict & RT6_LOOKUP_F_REACHABLE) { | |
491 | rt6_probe(rt); | |
492 | } | |
493 | ||
494 | out: | |
495 | return match; | |
496 | } | |
497 | ||
498 | static struct rt6_info *find_rr_leaf(struct fib6_node *fn, | |
499 | struct rt6_info *rr_head, | |
500 | u32 metric, int oif, int strict) | |
501 | { | |
502 | struct rt6_info *rt, *match; | |
554cfb7e | 503 | int mpri = -1; |
1da177e4 | 504 | |
f11e6659 DM |
505 | match = NULL; |
506 | for (rt = rr_head; rt && rt->rt6i_metric == metric; | |
d8d1f30b | 507 | rt = rt->dst.rt6_next) |
f11e6659 DM |
508 | match = find_match(rt, oif, strict, &mpri, match); |
509 | for (rt = fn->leaf; rt && rt != rr_head && rt->rt6i_metric == metric; | |
d8d1f30b | 510 | rt = rt->dst.rt6_next) |
f11e6659 | 511 | match = find_match(rt, oif, strict, &mpri, match); |
1da177e4 | 512 | |
f11e6659 DM |
513 | return match; |
514 | } | |
1da177e4 | 515 | |
f11e6659 DM |
516 | static struct rt6_info *rt6_select(struct fib6_node *fn, int oif, int strict) |
517 | { | |
518 | struct rt6_info *match, *rt0; | |
8ed67789 | 519 | struct net *net; |
1da177e4 | 520 | |
f11e6659 | 521 | RT6_TRACE("%s(fn->leaf=%p, oif=%d)\n", |
0dc47877 | 522 | __func__, fn->leaf, oif); |
554cfb7e | 523 | |
f11e6659 DM |
524 | rt0 = fn->rr_ptr; |
525 | if (!rt0) | |
526 | fn->rr_ptr = rt0 = fn->leaf; | |
1da177e4 | 527 | |
f11e6659 | 528 | match = find_rr_leaf(fn, rt0, rt0->rt6i_metric, oif, strict); |
1da177e4 | 529 | |
554cfb7e | 530 | if (!match && |
f11e6659 | 531 | (strict & RT6_LOOKUP_F_REACHABLE)) { |
d8d1f30b | 532 | struct rt6_info *next = rt0->dst.rt6_next; |
f11e6659 | 533 | |
554cfb7e | 534 | /* no entries matched; do round-robin */ |
f11e6659 DM |
535 | if (!next || next->rt6i_metric != rt0->rt6i_metric) |
536 | next = fn->leaf; | |
537 | ||
538 | if (next != rt0) | |
539 | fn->rr_ptr = next; | |
1da177e4 | 540 | } |
1da177e4 | 541 | |
f11e6659 | 542 | RT6_TRACE("%s() => %p\n", |
0dc47877 | 543 | __func__, match); |
1da177e4 | 544 | |
c346dca1 | 545 | net = dev_net(rt0->rt6i_dev); |
a02cec21 | 546 | return match ? match : net->ipv6.ip6_null_entry; |
1da177e4 LT |
547 | } |
548 | ||
70ceb4f5 YH |
549 | #ifdef CONFIG_IPV6_ROUTE_INFO |
550 | int rt6_route_rcv(struct net_device *dev, u8 *opt, int len, | |
b71d1d42 | 551 | const struct in6_addr *gwaddr) |
70ceb4f5 | 552 | { |
c346dca1 | 553 | struct net *net = dev_net(dev); |
70ceb4f5 YH |
554 | struct route_info *rinfo = (struct route_info *) opt; |
555 | struct in6_addr prefix_buf, *prefix; | |
556 | unsigned int pref; | |
4bed72e4 | 557 | unsigned long lifetime; |
70ceb4f5 YH |
558 | struct rt6_info *rt; |
559 | ||
560 | if (len < sizeof(struct route_info)) { | |
561 | return -EINVAL; | |
562 | } | |
563 | ||
564 | /* Sanity check for prefix_len and length */ | |
565 | if (rinfo->length > 3) { | |
566 | return -EINVAL; | |
567 | } else if (rinfo->prefix_len > 128) { | |
568 | return -EINVAL; | |
569 | } else if (rinfo->prefix_len > 64) { | |
570 | if (rinfo->length < 2) { | |
571 | return -EINVAL; | |
572 | } | |
573 | } else if (rinfo->prefix_len > 0) { | |
574 | if (rinfo->length < 1) { | |
575 | return -EINVAL; | |
576 | } | |
577 | } | |
578 | ||
579 | pref = rinfo->route_pref; | |
580 | if (pref == ICMPV6_ROUTER_PREF_INVALID) | |
3933fc95 | 581 | return -EINVAL; |
70ceb4f5 | 582 | |
4bed72e4 | 583 | lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ); |
70ceb4f5 YH |
584 | |
585 | if (rinfo->length == 3) | |
586 | prefix = (struct in6_addr *)rinfo->prefix; | |
587 | else { | |
588 | /* this function is safe */ | |
589 | ipv6_addr_prefix(&prefix_buf, | |
590 | (struct in6_addr *)rinfo->prefix, | |
591 | rinfo->prefix_len); | |
592 | prefix = &prefix_buf; | |
593 | } | |
594 | ||
efa2cea0 DL |
595 | rt = rt6_get_route_info(net, prefix, rinfo->prefix_len, gwaddr, |
596 | dev->ifindex); | |
70ceb4f5 YH |
597 | |
598 | if (rt && !lifetime) { | |
e0a1ad73 | 599 | ip6_del_rt(rt); |
70ceb4f5 YH |
600 | rt = NULL; |
601 | } | |
602 | ||
603 | if (!rt && lifetime) | |
efa2cea0 | 604 | rt = rt6_add_route_info(net, prefix, rinfo->prefix_len, gwaddr, dev->ifindex, |
70ceb4f5 YH |
605 | pref); |
606 | else if (rt) | |
607 | rt->rt6i_flags = RTF_ROUTEINFO | | |
608 | (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref); | |
609 | ||
610 | if (rt) { | |
4bed72e4 | 611 | if (!addrconf_finite_timeout(lifetime)) { |
70ceb4f5 YH |
612 | rt->rt6i_flags &= ~RTF_EXPIRES; |
613 | } else { | |
614 | rt->rt6i_expires = jiffies + HZ * lifetime; | |
615 | rt->rt6i_flags |= RTF_EXPIRES; | |
616 | } | |
d8d1f30b | 617 | dst_release(&rt->dst); |
70ceb4f5 YH |
618 | } |
619 | return 0; | |
620 | } | |
621 | #endif | |
622 | ||
8ed67789 | 623 | #define BACKTRACK(__net, saddr) \ |
982f56f3 | 624 | do { \ |
8ed67789 | 625 | if (rt == __net->ipv6.ip6_null_entry) { \ |
982f56f3 | 626 | struct fib6_node *pn; \ |
e0eda7bb | 627 | while (1) { \ |
982f56f3 YH |
628 | if (fn->fn_flags & RTN_TL_ROOT) \ |
629 | goto out; \ | |
630 | pn = fn->parent; \ | |
631 | if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn) \ | |
8bce65b9 | 632 | fn = fib6_lookup(FIB6_SUBTREE(pn), NULL, saddr); \ |
982f56f3 YH |
633 | else \ |
634 | fn = pn; \ | |
635 | if (fn->fn_flags & RTN_RTINFO) \ | |
636 | goto restart; \ | |
c71099ac | 637 | } \ |
c71099ac | 638 | } \ |
38308473 | 639 | } while (0) |
c71099ac | 640 | |
8ed67789 DL |
641 | static struct rt6_info *ip6_pol_route_lookup(struct net *net, |
642 | struct fib6_table *table, | |
4c9483b2 | 643 | struct flowi6 *fl6, int flags) |
1da177e4 LT |
644 | { |
645 | struct fib6_node *fn; | |
646 | struct rt6_info *rt; | |
647 | ||
c71099ac | 648 | read_lock_bh(&table->tb6_lock); |
4c9483b2 | 649 | fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr); |
c71099ac TG |
650 | restart: |
651 | rt = fn->leaf; | |
4c9483b2 DM |
652 | rt = rt6_device_match(net, rt, &fl6->saddr, fl6->flowi6_oif, flags); |
653 | BACKTRACK(net, &fl6->saddr); | |
c71099ac | 654 | out: |
d8d1f30b | 655 | dst_use(&rt->dst, jiffies); |
c71099ac | 656 | read_unlock_bh(&table->tb6_lock); |
c71099ac TG |
657 | return rt; |
658 | ||
659 | } | |
660 | ||
ea6e574e FW |
661 | struct dst_entry * ip6_route_lookup(struct net *net, struct flowi6 *fl6, |
662 | int flags) | |
663 | { | |
664 | return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_lookup); | |
665 | } | |
666 | EXPORT_SYMBOL_GPL(ip6_route_lookup); | |
667 | ||
9acd9f3a YH |
668 | struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr, |
669 | const struct in6_addr *saddr, int oif, int strict) | |
c71099ac | 670 | { |
4c9483b2 DM |
671 | struct flowi6 fl6 = { |
672 | .flowi6_oif = oif, | |
673 | .daddr = *daddr, | |
c71099ac TG |
674 | }; |
675 | struct dst_entry *dst; | |
77d16f45 | 676 | int flags = strict ? RT6_LOOKUP_F_IFACE : 0; |
c71099ac | 677 | |
adaa70bb | 678 | if (saddr) { |
4c9483b2 | 679 | memcpy(&fl6.saddr, saddr, sizeof(*saddr)); |
adaa70bb TG |
680 | flags |= RT6_LOOKUP_F_HAS_SADDR; |
681 | } | |
682 | ||
4c9483b2 | 683 | dst = fib6_rule_lookup(net, &fl6, flags, ip6_pol_route_lookup); |
c71099ac TG |
684 | if (dst->error == 0) |
685 | return (struct rt6_info *) dst; | |
686 | ||
687 | dst_release(dst); | |
688 | ||
1da177e4 LT |
689 | return NULL; |
690 | } | |
691 | ||
7159039a YH |
692 | EXPORT_SYMBOL(rt6_lookup); |
693 | ||
c71099ac | 694 | /* ip6_ins_rt is called with FREE table->tb6_lock. |
1da177e4 LT |
695 | It takes new route entry, the addition fails by any reason the |
696 | route is freed. In any case, if caller does not hold it, it may | |
697 | be destroyed. | |
698 | */ | |
699 | ||
86872cb5 | 700 | static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info) |
1da177e4 LT |
701 | { |
702 | int err; | |
c71099ac | 703 | struct fib6_table *table; |
1da177e4 | 704 | |
c71099ac TG |
705 | table = rt->rt6i_table; |
706 | write_lock_bh(&table->tb6_lock); | |
86872cb5 | 707 | err = fib6_add(&table->tb6_root, rt, info); |
c71099ac | 708 | write_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
709 | |
710 | return err; | |
711 | } | |
712 | ||
40e22e8f TG |
713 | int ip6_ins_rt(struct rt6_info *rt) |
714 | { | |
4d1169c1 | 715 | struct nl_info info = { |
c346dca1 | 716 | .nl_net = dev_net(rt->rt6i_dev), |
4d1169c1 | 717 | }; |
528c4ceb | 718 | return __ip6_ins_rt(rt, &info); |
40e22e8f TG |
719 | } |
720 | ||
21efcfa0 ED |
721 | static struct rt6_info *rt6_alloc_cow(const struct rt6_info *ort, |
722 | const struct in6_addr *daddr, | |
b71d1d42 | 723 | const struct in6_addr *saddr) |
1da177e4 | 724 | { |
1da177e4 LT |
725 | struct rt6_info *rt; |
726 | ||
727 | /* | |
728 | * Clone the route. | |
729 | */ | |
730 | ||
21efcfa0 | 731 | rt = ip6_rt_copy(ort, daddr); |
1da177e4 LT |
732 | |
733 | if (rt) { | |
14deae41 DM |
734 | struct neighbour *neigh; |
735 | int attempts = !in_softirq(); | |
736 | ||
38308473 | 737 | if (!(rt->rt6i_flags & RTF_GATEWAY)) { |
bb3c3686 | 738 | if (ort->rt6i_dst.plen != 128 && |
21efcfa0 | 739 | ipv6_addr_equal(&ort->rt6i_dst.addr, daddr)) |
58c4fb86 | 740 | rt->rt6i_flags |= RTF_ANYCAST; |
4e3fd7a0 | 741 | rt->rt6i_gateway = *daddr; |
58c4fb86 | 742 | } |
1da177e4 | 743 | |
1da177e4 | 744 | rt->rt6i_flags |= RTF_CACHE; |
1da177e4 LT |
745 | |
746 | #ifdef CONFIG_IPV6_SUBTREES | |
747 | if (rt->rt6i_src.plen && saddr) { | |
4e3fd7a0 | 748 | rt->rt6i_src.addr = *saddr; |
1da177e4 LT |
749 | rt->rt6i_src.plen = 128; |
750 | } | |
751 | #endif | |
752 | ||
14deae41 | 753 | retry: |
04a6f441 DM |
754 | neigh = __neigh_lookup_errno(&nd_tbl, &rt->rt6i_gateway, |
755 | rt->rt6i_dev); | |
14deae41 DM |
756 | if (IS_ERR(neigh)) { |
757 | struct net *net = dev_net(rt->rt6i_dev); | |
758 | int saved_rt_min_interval = | |
759 | net->ipv6.sysctl.ip6_rt_gc_min_interval; | |
760 | int saved_rt_elasticity = | |
761 | net->ipv6.sysctl.ip6_rt_gc_elasticity; | |
762 | ||
763 | if (attempts-- > 0) { | |
764 | net->ipv6.sysctl.ip6_rt_gc_elasticity = 1; | |
765 | net->ipv6.sysctl.ip6_rt_gc_min_interval = 0; | |
766 | ||
86393e52 | 767 | ip6_dst_gc(&net->ipv6.ip6_dst_ops); |
14deae41 DM |
768 | |
769 | net->ipv6.sysctl.ip6_rt_gc_elasticity = | |
770 | saved_rt_elasticity; | |
771 | net->ipv6.sysctl.ip6_rt_gc_min_interval = | |
772 | saved_rt_min_interval; | |
773 | goto retry; | |
774 | } | |
775 | ||
776 | if (net_ratelimit()) | |
777 | printk(KERN_WARNING | |
7e1b33e5 | 778 | "ipv6: Neighbour table overflow.\n"); |
d8d1f30b | 779 | dst_free(&rt->dst); |
14deae41 DM |
780 | return NULL; |
781 | } | |
69cce1d1 | 782 | dst_set_neighbour(&rt->dst, neigh); |
1da177e4 | 783 | |
95a9a5ba | 784 | } |
1da177e4 | 785 | |
95a9a5ba YH |
786 | return rt; |
787 | } | |
1da177e4 | 788 | |
21efcfa0 ED |
789 | static struct rt6_info *rt6_alloc_clone(struct rt6_info *ort, |
790 | const struct in6_addr *daddr) | |
299d9939 | 791 | { |
21efcfa0 ED |
792 | struct rt6_info *rt = ip6_rt_copy(ort, daddr); |
793 | ||
299d9939 | 794 | if (rt) { |
299d9939 | 795 | rt->rt6i_flags |= RTF_CACHE; |
27217455 | 796 | dst_set_neighbour(&rt->dst, neigh_clone(dst_get_neighbour_noref_raw(&ort->dst))); |
299d9939 YH |
797 | } |
798 | return rt; | |
799 | } | |
800 | ||
8ed67789 | 801 | static struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table, int oif, |
4c9483b2 | 802 | struct flowi6 *fl6, int flags) |
1da177e4 LT |
803 | { |
804 | struct fib6_node *fn; | |
519fbd87 | 805 | struct rt6_info *rt, *nrt; |
c71099ac | 806 | int strict = 0; |
1da177e4 | 807 | int attempts = 3; |
519fbd87 | 808 | int err; |
53b7997f | 809 | int reachable = net->ipv6.devconf_all->forwarding ? 0 : RT6_LOOKUP_F_REACHABLE; |
1da177e4 | 810 | |
77d16f45 | 811 | strict |= flags & RT6_LOOKUP_F_IFACE; |
1da177e4 LT |
812 | |
813 | relookup: | |
c71099ac | 814 | read_lock_bh(&table->tb6_lock); |
1da177e4 | 815 | |
8238dd06 | 816 | restart_2: |
4c9483b2 | 817 | fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr); |
1da177e4 LT |
818 | |
819 | restart: | |
4acad72d | 820 | rt = rt6_select(fn, oif, strict | reachable); |
8ed67789 | 821 | |
4c9483b2 | 822 | BACKTRACK(net, &fl6->saddr); |
8ed67789 | 823 | if (rt == net->ipv6.ip6_null_entry || |
8238dd06 | 824 | rt->rt6i_flags & RTF_CACHE) |
1ddef044 | 825 | goto out; |
1da177e4 | 826 | |
d8d1f30b | 827 | dst_hold(&rt->dst); |
c71099ac | 828 | read_unlock_bh(&table->tb6_lock); |
fb9de91e | 829 | |
27217455 | 830 | if (!dst_get_neighbour_noref_raw(&rt->dst) && !(rt->rt6i_flags & RTF_NONEXTHOP)) |
4c9483b2 | 831 | nrt = rt6_alloc_cow(rt, &fl6->daddr, &fl6->saddr); |
7343ff31 | 832 | else if (!(rt->dst.flags & DST_HOST)) |
4c9483b2 | 833 | nrt = rt6_alloc_clone(rt, &fl6->daddr); |
7343ff31 DM |
834 | else |
835 | goto out2; | |
e40cf353 | 836 | |
d8d1f30b | 837 | dst_release(&rt->dst); |
8ed67789 | 838 | rt = nrt ? : net->ipv6.ip6_null_entry; |
1da177e4 | 839 | |
d8d1f30b | 840 | dst_hold(&rt->dst); |
519fbd87 | 841 | if (nrt) { |
40e22e8f | 842 | err = ip6_ins_rt(nrt); |
519fbd87 | 843 | if (!err) |
1da177e4 | 844 | goto out2; |
1da177e4 | 845 | } |
1da177e4 | 846 | |
519fbd87 YH |
847 | if (--attempts <= 0) |
848 | goto out2; | |
849 | ||
850 | /* | |
c71099ac | 851 | * Race condition! In the gap, when table->tb6_lock was |
519fbd87 YH |
852 | * released someone could insert this route. Relookup. |
853 | */ | |
d8d1f30b | 854 | dst_release(&rt->dst); |
519fbd87 YH |
855 | goto relookup; |
856 | ||
857 | out: | |
8238dd06 YH |
858 | if (reachable) { |
859 | reachable = 0; | |
860 | goto restart_2; | |
861 | } | |
d8d1f30b | 862 | dst_hold(&rt->dst); |
c71099ac | 863 | read_unlock_bh(&table->tb6_lock); |
1da177e4 | 864 | out2: |
d8d1f30b CG |
865 | rt->dst.lastuse = jiffies; |
866 | rt->dst.__use++; | |
c71099ac TG |
867 | |
868 | return rt; | |
1da177e4 LT |
869 | } |
870 | ||
8ed67789 | 871 | static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table, |
4c9483b2 | 872 | struct flowi6 *fl6, int flags) |
4acad72d | 873 | { |
4c9483b2 | 874 | return ip6_pol_route(net, table, fl6->flowi6_iif, fl6, flags); |
4acad72d PE |
875 | } |
876 | ||
c71099ac TG |
877 | void ip6_route_input(struct sk_buff *skb) |
878 | { | |
b71d1d42 | 879 | const struct ipv6hdr *iph = ipv6_hdr(skb); |
c346dca1 | 880 | struct net *net = dev_net(skb->dev); |
adaa70bb | 881 | int flags = RT6_LOOKUP_F_HAS_SADDR; |
4c9483b2 DM |
882 | struct flowi6 fl6 = { |
883 | .flowi6_iif = skb->dev->ifindex, | |
884 | .daddr = iph->daddr, | |
885 | .saddr = iph->saddr, | |
38308473 | 886 | .flowlabel = (* (__be32 *) iph) & IPV6_FLOWINFO_MASK, |
4c9483b2 DM |
887 | .flowi6_mark = skb->mark, |
888 | .flowi6_proto = iph->nexthdr, | |
c71099ac | 889 | }; |
adaa70bb | 890 | |
1d6e55f1 | 891 | if (rt6_need_strict(&iph->daddr) && skb->dev->type != ARPHRD_PIMREG) |
adaa70bb | 892 | flags |= RT6_LOOKUP_F_IFACE; |
c71099ac | 893 | |
4c9483b2 | 894 | skb_dst_set(skb, fib6_rule_lookup(net, &fl6, flags, ip6_pol_route_input)); |
c71099ac TG |
895 | } |
896 | ||
8ed67789 | 897 | static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table, |
4c9483b2 | 898 | struct flowi6 *fl6, int flags) |
1da177e4 | 899 | { |
4c9483b2 | 900 | return ip6_pol_route(net, table, fl6->flowi6_oif, fl6, flags); |
c71099ac TG |
901 | } |
902 | ||
9c7a4f9c | 903 | struct dst_entry * ip6_route_output(struct net *net, const struct sock *sk, |
4c9483b2 | 904 | struct flowi6 *fl6) |
c71099ac TG |
905 | { |
906 | int flags = 0; | |
907 | ||
4c9483b2 | 908 | if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl6->daddr)) |
77d16f45 | 909 | flags |= RT6_LOOKUP_F_IFACE; |
c71099ac | 910 | |
4c9483b2 | 911 | if (!ipv6_addr_any(&fl6->saddr)) |
adaa70bb | 912 | flags |= RT6_LOOKUP_F_HAS_SADDR; |
0c9a2ac1 YH |
913 | else if (sk) |
914 | flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs); | |
adaa70bb | 915 | |
4c9483b2 | 916 | return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_output); |
1da177e4 LT |
917 | } |
918 | ||
7159039a | 919 | EXPORT_SYMBOL(ip6_route_output); |
1da177e4 | 920 | |
2774c131 | 921 | struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_orig) |
14e50e57 | 922 | { |
5c1e6aa3 | 923 | struct rt6_info *rt, *ort = (struct rt6_info *) dst_orig; |
14e50e57 DM |
924 | struct dst_entry *new = NULL; |
925 | ||
5c1e6aa3 | 926 | rt = dst_alloc(&ip6_dst_blackhole_ops, ort->dst.dev, 1, 0, 0); |
14e50e57 | 927 | if (rt) { |
cf911662 DM |
928 | memset(&rt->rt6i_table, 0, sizeof(*rt) - sizeof(struct dst_entry)); |
929 | ||
d8d1f30b | 930 | new = &rt->dst; |
14e50e57 | 931 | |
14e50e57 | 932 | new->__use = 1; |
352e512c HX |
933 | new->input = dst_discard; |
934 | new->output = dst_discard; | |
14e50e57 | 935 | |
21efcfa0 ED |
936 | if (dst_metrics_read_only(&ort->dst)) |
937 | new->_metrics = ort->dst._metrics; | |
938 | else | |
939 | dst_copy_metrics(new, &ort->dst); | |
14e50e57 DM |
940 | rt->rt6i_idev = ort->rt6i_idev; |
941 | if (rt->rt6i_idev) | |
942 | in6_dev_hold(rt->rt6i_idev); | |
943 | rt->rt6i_expires = 0; | |
944 | ||
4e3fd7a0 | 945 | rt->rt6i_gateway = ort->rt6i_gateway; |
14e50e57 DM |
946 | rt->rt6i_flags = ort->rt6i_flags & ~RTF_EXPIRES; |
947 | rt->rt6i_metric = 0; | |
948 | ||
949 | memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key)); | |
950 | #ifdef CONFIG_IPV6_SUBTREES | |
951 | memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key)); | |
952 | #endif | |
953 | ||
954 | dst_free(new); | |
955 | } | |
956 | ||
69ead7af DM |
957 | dst_release(dst_orig); |
958 | return new ? new : ERR_PTR(-ENOMEM); | |
14e50e57 | 959 | } |
14e50e57 | 960 | |
1da177e4 LT |
961 | /* |
962 | * Destination cache support functions | |
963 | */ | |
964 | ||
965 | static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie) | |
966 | { | |
967 | struct rt6_info *rt; | |
968 | ||
969 | rt = (struct rt6_info *) dst; | |
970 | ||
6431cbc2 DM |
971 | if (rt->rt6i_node && (rt->rt6i_node->fn_sernum == cookie)) { |
972 | if (rt->rt6i_peer_genid != rt6_peer_genid()) { | |
973 | if (!rt->rt6i_peer) | |
974 | rt6_bind_peer(rt, 0); | |
975 | rt->rt6i_peer_genid = rt6_peer_genid(); | |
976 | } | |
1da177e4 | 977 | return dst; |
6431cbc2 | 978 | } |
1da177e4 LT |
979 | return NULL; |
980 | } | |
981 | ||
982 | static struct dst_entry *ip6_negative_advice(struct dst_entry *dst) | |
983 | { | |
984 | struct rt6_info *rt = (struct rt6_info *) dst; | |
985 | ||
986 | if (rt) { | |
54c1a859 YH |
987 | if (rt->rt6i_flags & RTF_CACHE) { |
988 | if (rt6_check_expired(rt)) { | |
989 | ip6_del_rt(rt); | |
990 | dst = NULL; | |
991 | } | |
992 | } else { | |
1da177e4 | 993 | dst_release(dst); |
54c1a859 YH |
994 | dst = NULL; |
995 | } | |
1da177e4 | 996 | } |
54c1a859 | 997 | return dst; |
1da177e4 LT |
998 | } |
999 | ||
1000 | static void ip6_link_failure(struct sk_buff *skb) | |
1001 | { | |
1002 | struct rt6_info *rt; | |
1003 | ||
3ffe533c | 1004 | icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0); |
1da177e4 | 1005 | |
adf30907 | 1006 | rt = (struct rt6_info *) skb_dst(skb); |
1da177e4 | 1007 | if (rt) { |
38308473 | 1008 | if (rt->rt6i_flags & RTF_CACHE) { |
d8d1f30b | 1009 | dst_set_expires(&rt->dst, 0); |
1da177e4 LT |
1010 | rt->rt6i_flags |= RTF_EXPIRES; |
1011 | } else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT)) | |
1012 | rt->rt6i_node->fn_sernum = -1; | |
1013 | } | |
1014 | } | |
1015 | ||
1016 | static void ip6_rt_update_pmtu(struct dst_entry *dst, u32 mtu) | |
1017 | { | |
1018 | struct rt6_info *rt6 = (struct rt6_info*)dst; | |
1019 | ||
1020 | if (mtu < dst_mtu(dst) && rt6->rt6i_dst.plen == 128) { | |
1021 | rt6->rt6i_flags |= RTF_MODIFIED; | |
1022 | if (mtu < IPV6_MIN_MTU) { | |
defb3519 | 1023 | u32 features = dst_metric(dst, RTAX_FEATURES); |
1da177e4 | 1024 | mtu = IPV6_MIN_MTU; |
defb3519 DM |
1025 | features |= RTAX_FEATURE_ALLFRAG; |
1026 | dst_metric_set(dst, RTAX_FEATURES, features); | |
1da177e4 | 1027 | } |
defb3519 | 1028 | dst_metric_set(dst, RTAX_MTU, mtu); |
1da177e4 LT |
1029 | } |
1030 | } | |
1031 | ||
0dbaee3b | 1032 | static unsigned int ip6_default_advmss(const struct dst_entry *dst) |
1da177e4 | 1033 | { |
0dbaee3b DM |
1034 | struct net_device *dev = dst->dev; |
1035 | unsigned int mtu = dst_mtu(dst); | |
1036 | struct net *net = dev_net(dev); | |
1037 | ||
1da177e4 LT |
1038 | mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr); |
1039 | ||
5578689a DL |
1040 | if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss) |
1041 | mtu = net->ipv6.sysctl.ip6_rt_min_advmss; | |
1da177e4 LT |
1042 | |
1043 | /* | |
1ab1457c YH |
1044 | * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and |
1045 | * corresponding MSS is IPV6_MAXPLEN - tcp_header_size. | |
1046 | * IPV6_MAXPLEN is also valid and means: "any MSS, | |
1da177e4 LT |
1047 | * rely only on pmtu discovery" |
1048 | */ | |
1049 | if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr)) | |
1050 | mtu = IPV6_MAXPLEN; | |
1051 | return mtu; | |
1052 | } | |
1053 | ||
ebb762f2 | 1054 | static unsigned int ip6_mtu(const struct dst_entry *dst) |
d33e4553 | 1055 | { |
d33e4553 | 1056 | struct inet6_dev *idev; |
618f9bc7 SK |
1057 | unsigned int mtu = dst_metric_raw(dst, RTAX_MTU); |
1058 | ||
1059 | if (mtu) | |
1060 | return mtu; | |
1061 | ||
1062 | mtu = IPV6_MIN_MTU; | |
d33e4553 DM |
1063 | |
1064 | rcu_read_lock(); | |
1065 | idev = __in6_dev_get(dst->dev); | |
1066 | if (idev) | |
1067 | mtu = idev->cnf.mtu6; | |
1068 | rcu_read_unlock(); | |
1069 | ||
1070 | return mtu; | |
1071 | } | |
1072 | ||
3b00944c YH |
1073 | static struct dst_entry *icmp6_dst_gc_list; |
1074 | static DEFINE_SPINLOCK(icmp6_dst_lock); | |
5d0bbeeb | 1075 | |
3b00944c | 1076 | struct dst_entry *icmp6_dst_alloc(struct net_device *dev, |
1da177e4 | 1077 | struct neighbour *neigh, |
87a11578 | 1078 | struct flowi6 *fl6) |
1da177e4 | 1079 | { |
87a11578 | 1080 | struct dst_entry *dst; |
1da177e4 LT |
1081 | struct rt6_info *rt; |
1082 | struct inet6_dev *idev = in6_dev_get(dev); | |
c346dca1 | 1083 | struct net *net = dev_net(dev); |
1da177e4 | 1084 | |
38308473 | 1085 | if (unlikely(!idev)) |
1da177e4 LT |
1086 | return NULL; |
1087 | ||
957c665f | 1088 | rt = ip6_dst_alloc(&net->ipv6.ip6_dst_ops, dev, 0); |
38308473 | 1089 | if (unlikely(!rt)) { |
1da177e4 | 1090 | in6_dev_put(idev); |
87a11578 | 1091 | dst = ERR_PTR(-ENOMEM); |
1da177e4 LT |
1092 | goto out; |
1093 | } | |
1094 | ||
1da177e4 LT |
1095 | if (neigh) |
1096 | neigh_hold(neigh); | |
14deae41 | 1097 | else { |
87a11578 | 1098 | neigh = __neigh_lookup_errno(&nd_tbl, &fl6->daddr, dev); |
b43faac6 DM |
1099 | if (IS_ERR(neigh)) { |
1100 | dst_free(&rt->dst); | |
1101 | return ERR_CAST(neigh); | |
1102 | } | |
14deae41 | 1103 | } |
1da177e4 | 1104 | |
8e2ec639 YZ |
1105 | rt->dst.flags |= DST_HOST; |
1106 | rt->dst.output = ip6_output; | |
69cce1d1 | 1107 | dst_set_neighbour(&rt->dst, neigh); |
d8d1f30b | 1108 | atomic_set(&rt->dst.__refcnt, 1); |
87a11578 | 1109 | rt->rt6i_dst.addr = fl6->daddr; |
8e2ec639 YZ |
1110 | rt->rt6i_dst.plen = 128; |
1111 | rt->rt6i_idev = idev; | |
7011687f | 1112 | dst_metric_set(&rt->dst, RTAX_HOPLIMIT, 255); |
1da177e4 | 1113 | |
3b00944c | 1114 | spin_lock_bh(&icmp6_dst_lock); |
d8d1f30b CG |
1115 | rt->dst.next = icmp6_dst_gc_list; |
1116 | icmp6_dst_gc_list = &rt->dst; | |
3b00944c | 1117 | spin_unlock_bh(&icmp6_dst_lock); |
1da177e4 | 1118 | |
5578689a | 1119 | fib6_force_start_gc(net); |
1da177e4 | 1120 | |
87a11578 DM |
1121 | dst = xfrm_lookup(net, &rt->dst, flowi6_to_flowi(fl6), NULL, 0); |
1122 | ||
1da177e4 | 1123 | out: |
87a11578 | 1124 | return dst; |
1da177e4 LT |
1125 | } |
1126 | ||
3d0f24a7 | 1127 | int icmp6_dst_gc(void) |
1da177e4 | 1128 | { |
e9476e95 | 1129 | struct dst_entry *dst, **pprev; |
3d0f24a7 | 1130 | int more = 0; |
1da177e4 | 1131 | |
3b00944c YH |
1132 | spin_lock_bh(&icmp6_dst_lock); |
1133 | pprev = &icmp6_dst_gc_list; | |
5d0bbeeb | 1134 | |
1da177e4 LT |
1135 | while ((dst = *pprev) != NULL) { |
1136 | if (!atomic_read(&dst->__refcnt)) { | |
1137 | *pprev = dst->next; | |
1138 | dst_free(dst); | |
1da177e4 LT |
1139 | } else { |
1140 | pprev = &dst->next; | |
3d0f24a7 | 1141 | ++more; |
1da177e4 LT |
1142 | } |
1143 | } | |
1144 | ||
3b00944c | 1145 | spin_unlock_bh(&icmp6_dst_lock); |
5d0bbeeb | 1146 | |
3d0f24a7 | 1147 | return more; |
1da177e4 LT |
1148 | } |
1149 | ||
1e493d19 DM |
1150 | static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg), |
1151 | void *arg) | |
1152 | { | |
1153 | struct dst_entry *dst, **pprev; | |
1154 | ||
1155 | spin_lock_bh(&icmp6_dst_lock); | |
1156 | pprev = &icmp6_dst_gc_list; | |
1157 | while ((dst = *pprev) != NULL) { | |
1158 | struct rt6_info *rt = (struct rt6_info *) dst; | |
1159 | if (func(rt, arg)) { | |
1160 | *pprev = dst->next; | |
1161 | dst_free(dst); | |
1162 | } else { | |
1163 | pprev = &dst->next; | |
1164 | } | |
1165 | } | |
1166 | spin_unlock_bh(&icmp6_dst_lock); | |
1167 | } | |
1168 | ||
569d3645 | 1169 | static int ip6_dst_gc(struct dst_ops *ops) |
1da177e4 | 1170 | { |
1da177e4 | 1171 | unsigned long now = jiffies; |
86393e52 | 1172 | struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops); |
7019b78e DL |
1173 | int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval; |
1174 | int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size; | |
1175 | int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity; | |
1176 | int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout; | |
1177 | unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc; | |
fc66f95c | 1178 | int entries; |
7019b78e | 1179 | |
fc66f95c | 1180 | entries = dst_entries_get_fast(ops); |
7019b78e | 1181 | if (time_after(rt_last_gc + rt_min_interval, now) && |
fc66f95c | 1182 | entries <= rt_max_size) |
1da177e4 LT |
1183 | goto out; |
1184 | ||
6891a346 BT |
1185 | net->ipv6.ip6_rt_gc_expire++; |
1186 | fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net); | |
1187 | net->ipv6.ip6_rt_last_gc = now; | |
fc66f95c ED |
1188 | entries = dst_entries_get_slow(ops); |
1189 | if (entries < ops->gc_thresh) | |
7019b78e | 1190 | net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1; |
1da177e4 | 1191 | out: |
7019b78e | 1192 | net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity; |
fc66f95c | 1193 | return entries > rt_max_size; |
1da177e4 LT |
1194 | } |
1195 | ||
1196 | /* Clean host part of a prefix. Not necessary in radix tree, | |
1197 | but results in cleaner routing tables. | |
1198 | ||
1199 | Remove it only when all the things will work! | |
1200 | */ | |
1201 | ||
6b75d090 | 1202 | int ip6_dst_hoplimit(struct dst_entry *dst) |
1da177e4 | 1203 | { |
5170ae82 | 1204 | int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT); |
a02e4b7d | 1205 | if (hoplimit == 0) { |
6b75d090 | 1206 | struct net_device *dev = dst->dev; |
c68f24cc ED |
1207 | struct inet6_dev *idev; |
1208 | ||
1209 | rcu_read_lock(); | |
1210 | idev = __in6_dev_get(dev); | |
1211 | if (idev) | |
6b75d090 | 1212 | hoplimit = idev->cnf.hop_limit; |
c68f24cc | 1213 | else |
53b7997f | 1214 | hoplimit = dev_net(dev)->ipv6.devconf_all->hop_limit; |
c68f24cc | 1215 | rcu_read_unlock(); |
1da177e4 LT |
1216 | } |
1217 | return hoplimit; | |
1218 | } | |
abbf46ae | 1219 | EXPORT_SYMBOL(ip6_dst_hoplimit); |
1da177e4 LT |
1220 | |
1221 | /* | |
1222 | * | |
1223 | */ | |
1224 | ||
86872cb5 | 1225 | int ip6_route_add(struct fib6_config *cfg) |
1da177e4 LT |
1226 | { |
1227 | int err; | |
5578689a | 1228 | struct net *net = cfg->fc_nlinfo.nl_net; |
1da177e4 LT |
1229 | struct rt6_info *rt = NULL; |
1230 | struct net_device *dev = NULL; | |
1231 | struct inet6_dev *idev = NULL; | |
c71099ac | 1232 | struct fib6_table *table; |
1da177e4 LT |
1233 | int addr_type; |
1234 | ||
86872cb5 | 1235 | if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128) |
1da177e4 LT |
1236 | return -EINVAL; |
1237 | #ifndef CONFIG_IPV6_SUBTREES | |
86872cb5 | 1238 | if (cfg->fc_src_len) |
1da177e4 LT |
1239 | return -EINVAL; |
1240 | #endif | |
86872cb5 | 1241 | if (cfg->fc_ifindex) { |
1da177e4 | 1242 | err = -ENODEV; |
5578689a | 1243 | dev = dev_get_by_index(net, cfg->fc_ifindex); |
1da177e4 LT |
1244 | if (!dev) |
1245 | goto out; | |
1246 | idev = in6_dev_get(dev); | |
1247 | if (!idev) | |
1248 | goto out; | |
1249 | } | |
1250 | ||
86872cb5 TG |
1251 | if (cfg->fc_metric == 0) |
1252 | cfg->fc_metric = IP6_RT_PRIO_USER; | |
1da177e4 | 1253 | |
d71314b4 | 1254 | err = -ENOBUFS; |
38308473 DM |
1255 | if (cfg->fc_nlinfo.nlh && |
1256 | !(cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_CREATE)) { | |
d71314b4 | 1257 | table = fib6_get_table(net, cfg->fc_table); |
38308473 | 1258 | if (!table) { |
d71314b4 MV |
1259 | printk(KERN_WARNING "IPv6: NLM_F_CREATE should be specified when creating new route\n"); |
1260 | table = fib6_new_table(net, cfg->fc_table); | |
1261 | } | |
1262 | } else { | |
1263 | table = fib6_new_table(net, cfg->fc_table); | |
1264 | } | |
38308473 DM |
1265 | |
1266 | if (!table) | |
c71099ac | 1267 | goto out; |
c71099ac | 1268 | |
957c665f | 1269 | rt = ip6_dst_alloc(&net->ipv6.ip6_dst_ops, NULL, DST_NOCOUNT); |
1da177e4 | 1270 | |
38308473 | 1271 | if (!rt) { |
1da177e4 LT |
1272 | err = -ENOMEM; |
1273 | goto out; | |
1274 | } | |
1275 | ||
d8d1f30b | 1276 | rt->dst.obsolete = -1; |
6f704992 YH |
1277 | rt->rt6i_expires = (cfg->fc_flags & RTF_EXPIRES) ? |
1278 | jiffies + clock_t_to_jiffies(cfg->fc_expires) : | |
1279 | 0; | |
1da177e4 | 1280 | |
86872cb5 TG |
1281 | if (cfg->fc_protocol == RTPROT_UNSPEC) |
1282 | cfg->fc_protocol = RTPROT_BOOT; | |
1283 | rt->rt6i_protocol = cfg->fc_protocol; | |
1284 | ||
1285 | addr_type = ipv6_addr_type(&cfg->fc_dst); | |
1da177e4 LT |
1286 | |
1287 | if (addr_type & IPV6_ADDR_MULTICAST) | |
d8d1f30b | 1288 | rt->dst.input = ip6_mc_input; |
ab79ad14 MŻ |
1289 | else if (cfg->fc_flags & RTF_LOCAL) |
1290 | rt->dst.input = ip6_input; | |
1da177e4 | 1291 | else |
d8d1f30b | 1292 | rt->dst.input = ip6_forward; |
1da177e4 | 1293 | |
d8d1f30b | 1294 | rt->dst.output = ip6_output; |
1da177e4 | 1295 | |
86872cb5 TG |
1296 | ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len); |
1297 | rt->rt6i_dst.plen = cfg->fc_dst_len; | |
1da177e4 | 1298 | if (rt->rt6i_dst.plen == 128) |
11d53b49 | 1299 | rt->dst.flags |= DST_HOST; |
1da177e4 | 1300 | |
8e2ec639 YZ |
1301 | if (!(rt->dst.flags & DST_HOST) && cfg->fc_mx) { |
1302 | u32 *metrics = kzalloc(sizeof(u32) * RTAX_MAX, GFP_KERNEL); | |
1303 | if (!metrics) { | |
1304 | err = -ENOMEM; | |
1305 | goto out; | |
1306 | } | |
1307 | dst_init_metrics(&rt->dst, metrics, 0); | |
1308 | } | |
1da177e4 | 1309 | #ifdef CONFIG_IPV6_SUBTREES |
86872cb5 TG |
1310 | ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len); |
1311 | rt->rt6i_src.plen = cfg->fc_src_len; | |
1da177e4 LT |
1312 | #endif |
1313 | ||
86872cb5 | 1314 | rt->rt6i_metric = cfg->fc_metric; |
1da177e4 LT |
1315 | |
1316 | /* We cannot add true routes via loopback here, | |
1317 | they would result in kernel looping; promote them to reject routes | |
1318 | */ | |
86872cb5 | 1319 | if ((cfg->fc_flags & RTF_REJECT) || |
38308473 DM |
1320 | (dev && (dev->flags & IFF_LOOPBACK) && |
1321 | !(addr_type & IPV6_ADDR_LOOPBACK) && | |
1322 | !(cfg->fc_flags & RTF_LOCAL))) { | |
1da177e4 | 1323 | /* hold loopback dev/idev if we haven't done so. */ |
5578689a | 1324 | if (dev != net->loopback_dev) { |
1da177e4 LT |
1325 | if (dev) { |
1326 | dev_put(dev); | |
1327 | in6_dev_put(idev); | |
1328 | } | |
5578689a | 1329 | dev = net->loopback_dev; |
1da177e4 LT |
1330 | dev_hold(dev); |
1331 | idev = in6_dev_get(dev); | |
1332 | if (!idev) { | |
1333 | err = -ENODEV; | |
1334 | goto out; | |
1335 | } | |
1336 | } | |
d8d1f30b CG |
1337 | rt->dst.output = ip6_pkt_discard_out; |
1338 | rt->dst.input = ip6_pkt_discard; | |
1339 | rt->dst.error = -ENETUNREACH; | |
1da177e4 LT |
1340 | rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP; |
1341 | goto install_route; | |
1342 | } | |
1343 | ||
86872cb5 | 1344 | if (cfg->fc_flags & RTF_GATEWAY) { |
b71d1d42 | 1345 | const struct in6_addr *gw_addr; |
1da177e4 LT |
1346 | int gwa_type; |
1347 | ||
86872cb5 | 1348 | gw_addr = &cfg->fc_gateway; |
4e3fd7a0 | 1349 | rt->rt6i_gateway = *gw_addr; |
1da177e4 LT |
1350 | gwa_type = ipv6_addr_type(gw_addr); |
1351 | ||
1352 | if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) { | |
1353 | struct rt6_info *grt; | |
1354 | ||
1355 | /* IPv6 strictly inhibits using not link-local | |
1356 | addresses as nexthop address. | |
1357 | Otherwise, router will not able to send redirects. | |
1358 | It is very good, but in some (rare!) circumstances | |
1359 | (SIT, PtP, NBMA NOARP links) it is handy to allow | |
1360 | some exceptions. --ANK | |
1361 | */ | |
1362 | err = -EINVAL; | |
38308473 | 1363 | if (!(gwa_type & IPV6_ADDR_UNICAST)) |
1da177e4 LT |
1364 | goto out; |
1365 | ||
5578689a | 1366 | grt = rt6_lookup(net, gw_addr, NULL, cfg->fc_ifindex, 1); |
1da177e4 LT |
1367 | |
1368 | err = -EHOSTUNREACH; | |
38308473 | 1369 | if (!grt) |
1da177e4 LT |
1370 | goto out; |
1371 | if (dev) { | |
1372 | if (dev != grt->rt6i_dev) { | |
d8d1f30b | 1373 | dst_release(&grt->dst); |
1da177e4 LT |
1374 | goto out; |
1375 | } | |
1376 | } else { | |
1377 | dev = grt->rt6i_dev; | |
1378 | idev = grt->rt6i_idev; | |
1379 | dev_hold(dev); | |
1380 | in6_dev_hold(grt->rt6i_idev); | |
1381 | } | |
38308473 | 1382 | if (!(grt->rt6i_flags & RTF_GATEWAY)) |
1da177e4 | 1383 | err = 0; |
d8d1f30b | 1384 | dst_release(&grt->dst); |
1da177e4 LT |
1385 | |
1386 | if (err) | |
1387 | goto out; | |
1388 | } | |
1389 | err = -EINVAL; | |
38308473 | 1390 | if (!dev || (dev->flags & IFF_LOOPBACK)) |
1da177e4 LT |
1391 | goto out; |
1392 | } | |
1393 | ||
1394 | err = -ENODEV; | |
38308473 | 1395 | if (!dev) |
1da177e4 LT |
1396 | goto out; |
1397 | ||
c3968a85 DW |
1398 | if (!ipv6_addr_any(&cfg->fc_prefsrc)) { |
1399 | if (!ipv6_chk_addr(net, &cfg->fc_prefsrc, dev, 0)) { | |
1400 | err = -EINVAL; | |
1401 | goto out; | |
1402 | } | |
4e3fd7a0 | 1403 | rt->rt6i_prefsrc.addr = cfg->fc_prefsrc; |
c3968a85 DW |
1404 | rt->rt6i_prefsrc.plen = 128; |
1405 | } else | |
1406 | rt->rt6i_prefsrc.plen = 0; | |
1407 | ||
86872cb5 | 1408 | if (cfg->fc_flags & (RTF_GATEWAY | RTF_NONEXTHOP)) { |
69cce1d1 DM |
1409 | struct neighbour *n = __neigh_lookup_errno(&nd_tbl, &rt->rt6i_gateway, dev); |
1410 | if (IS_ERR(n)) { | |
1411 | err = PTR_ERR(n); | |
1da177e4 LT |
1412 | goto out; |
1413 | } | |
69cce1d1 | 1414 | dst_set_neighbour(&rt->dst, n); |
1da177e4 LT |
1415 | } |
1416 | ||
86872cb5 | 1417 | rt->rt6i_flags = cfg->fc_flags; |
1da177e4 LT |
1418 | |
1419 | install_route: | |
86872cb5 TG |
1420 | if (cfg->fc_mx) { |
1421 | struct nlattr *nla; | |
1422 | int remaining; | |
1423 | ||
1424 | nla_for_each_attr(nla, cfg->fc_mx, cfg->fc_mx_len, remaining) { | |
8f4c1f9b | 1425 | int type = nla_type(nla); |
86872cb5 TG |
1426 | |
1427 | if (type) { | |
1428 | if (type > RTAX_MAX) { | |
1da177e4 LT |
1429 | err = -EINVAL; |
1430 | goto out; | |
1431 | } | |
86872cb5 | 1432 | |
defb3519 | 1433 | dst_metric_set(&rt->dst, type, nla_get_u32(nla)); |
1da177e4 | 1434 | } |
1da177e4 LT |
1435 | } |
1436 | } | |
1437 | ||
d8d1f30b | 1438 | rt->dst.dev = dev; |
1da177e4 | 1439 | rt->rt6i_idev = idev; |
c71099ac | 1440 | rt->rt6i_table = table; |
63152fc0 | 1441 | |
c346dca1 | 1442 | cfg->fc_nlinfo.nl_net = dev_net(dev); |
63152fc0 | 1443 | |
86872cb5 | 1444 | return __ip6_ins_rt(rt, &cfg->fc_nlinfo); |
1da177e4 LT |
1445 | |
1446 | out: | |
1447 | if (dev) | |
1448 | dev_put(dev); | |
1449 | if (idev) | |
1450 | in6_dev_put(idev); | |
1451 | if (rt) | |
d8d1f30b | 1452 | dst_free(&rt->dst); |
1da177e4 LT |
1453 | return err; |
1454 | } | |
1455 | ||
86872cb5 | 1456 | static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info) |
1da177e4 LT |
1457 | { |
1458 | int err; | |
c71099ac | 1459 | struct fib6_table *table; |
c346dca1 | 1460 | struct net *net = dev_net(rt->rt6i_dev); |
1da177e4 | 1461 | |
8ed67789 | 1462 | if (rt == net->ipv6.ip6_null_entry) |
6c813a72 PM |
1463 | return -ENOENT; |
1464 | ||
c71099ac TG |
1465 | table = rt->rt6i_table; |
1466 | write_lock_bh(&table->tb6_lock); | |
1da177e4 | 1467 | |
86872cb5 | 1468 | err = fib6_del(rt, info); |
d8d1f30b | 1469 | dst_release(&rt->dst); |
1da177e4 | 1470 | |
c71099ac | 1471 | write_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
1472 | |
1473 | return err; | |
1474 | } | |
1475 | ||
e0a1ad73 TG |
1476 | int ip6_del_rt(struct rt6_info *rt) |
1477 | { | |
4d1169c1 | 1478 | struct nl_info info = { |
c346dca1 | 1479 | .nl_net = dev_net(rt->rt6i_dev), |
4d1169c1 | 1480 | }; |
528c4ceb | 1481 | return __ip6_del_rt(rt, &info); |
e0a1ad73 TG |
1482 | } |
1483 | ||
86872cb5 | 1484 | static int ip6_route_del(struct fib6_config *cfg) |
1da177e4 | 1485 | { |
c71099ac | 1486 | struct fib6_table *table; |
1da177e4 LT |
1487 | struct fib6_node *fn; |
1488 | struct rt6_info *rt; | |
1489 | int err = -ESRCH; | |
1490 | ||
5578689a | 1491 | table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table); |
38308473 | 1492 | if (!table) |
c71099ac TG |
1493 | return err; |
1494 | ||
1495 | read_lock_bh(&table->tb6_lock); | |
1da177e4 | 1496 | |
c71099ac | 1497 | fn = fib6_locate(&table->tb6_root, |
86872cb5 TG |
1498 | &cfg->fc_dst, cfg->fc_dst_len, |
1499 | &cfg->fc_src, cfg->fc_src_len); | |
1ab1457c | 1500 | |
1da177e4 | 1501 | if (fn) { |
d8d1f30b | 1502 | for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) { |
86872cb5 | 1503 | if (cfg->fc_ifindex && |
38308473 | 1504 | (!rt->rt6i_dev || |
86872cb5 | 1505 | rt->rt6i_dev->ifindex != cfg->fc_ifindex)) |
1da177e4 | 1506 | continue; |
86872cb5 TG |
1507 | if (cfg->fc_flags & RTF_GATEWAY && |
1508 | !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway)) | |
1da177e4 | 1509 | continue; |
86872cb5 | 1510 | if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric) |
1da177e4 | 1511 | continue; |
d8d1f30b | 1512 | dst_hold(&rt->dst); |
c71099ac | 1513 | read_unlock_bh(&table->tb6_lock); |
1da177e4 | 1514 | |
86872cb5 | 1515 | return __ip6_del_rt(rt, &cfg->fc_nlinfo); |
1da177e4 LT |
1516 | } |
1517 | } | |
c71099ac | 1518 | read_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
1519 | |
1520 | return err; | |
1521 | } | |
1522 | ||
1523 | /* | |
1524 | * Handle redirects | |
1525 | */ | |
a6279458 | 1526 | struct ip6rd_flowi { |
4c9483b2 | 1527 | struct flowi6 fl6; |
a6279458 YH |
1528 | struct in6_addr gateway; |
1529 | }; | |
1530 | ||
8ed67789 DL |
1531 | static struct rt6_info *__ip6_route_redirect(struct net *net, |
1532 | struct fib6_table *table, | |
4c9483b2 | 1533 | struct flowi6 *fl6, |
a6279458 | 1534 | int flags) |
1da177e4 | 1535 | { |
4c9483b2 | 1536 | struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl6; |
a6279458 | 1537 | struct rt6_info *rt; |
e843b9e1 | 1538 | struct fib6_node *fn; |
c71099ac | 1539 | |
1da177e4 | 1540 | /* |
e843b9e1 YH |
1541 | * Get the "current" route for this destination and |
1542 | * check if the redirect has come from approriate router. | |
1543 | * | |
1544 | * RFC 2461 specifies that redirects should only be | |
1545 | * accepted if they come from the nexthop to the target. | |
1546 | * Due to the way the routes are chosen, this notion | |
1547 | * is a bit fuzzy and one might need to check all possible | |
1548 | * routes. | |
1da177e4 | 1549 | */ |
1da177e4 | 1550 | |
c71099ac | 1551 | read_lock_bh(&table->tb6_lock); |
4c9483b2 | 1552 | fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr); |
e843b9e1 | 1553 | restart: |
d8d1f30b | 1554 | for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) { |
e843b9e1 YH |
1555 | /* |
1556 | * Current route is on-link; redirect is always invalid. | |
1557 | * | |
1558 | * Seems, previous statement is not true. It could | |
1559 | * be node, which looks for us as on-link (f.e. proxy ndisc) | |
1560 | * But then router serving it might decide, that we should | |
1561 | * know truth 8)8) --ANK (980726). | |
1562 | */ | |
1563 | if (rt6_check_expired(rt)) | |
1564 | continue; | |
1565 | if (!(rt->rt6i_flags & RTF_GATEWAY)) | |
1566 | continue; | |
4c9483b2 | 1567 | if (fl6->flowi6_oif != rt->rt6i_dev->ifindex) |
e843b9e1 | 1568 | continue; |
a6279458 | 1569 | if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway)) |
e843b9e1 YH |
1570 | continue; |
1571 | break; | |
1572 | } | |
a6279458 | 1573 | |
cb15d9c2 | 1574 | if (!rt) |
8ed67789 | 1575 | rt = net->ipv6.ip6_null_entry; |
4c9483b2 | 1576 | BACKTRACK(net, &fl6->saddr); |
cb15d9c2 | 1577 | out: |
d8d1f30b | 1578 | dst_hold(&rt->dst); |
a6279458 | 1579 | |
c71099ac | 1580 | read_unlock_bh(&table->tb6_lock); |
e843b9e1 | 1581 | |
a6279458 YH |
1582 | return rt; |
1583 | }; | |
1584 | ||
b71d1d42 ED |
1585 | static struct rt6_info *ip6_route_redirect(const struct in6_addr *dest, |
1586 | const struct in6_addr *src, | |
1587 | const struct in6_addr *gateway, | |
a6279458 YH |
1588 | struct net_device *dev) |
1589 | { | |
adaa70bb | 1590 | int flags = RT6_LOOKUP_F_HAS_SADDR; |
c346dca1 | 1591 | struct net *net = dev_net(dev); |
a6279458 | 1592 | struct ip6rd_flowi rdfl = { |
4c9483b2 DM |
1593 | .fl6 = { |
1594 | .flowi6_oif = dev->ifindex, | |
1595 | .daddr = *dest, | |
1596 | .saddr = *src, | |
a6279458 | 1597 | }, |
a6279458 | 1598 | }; |
adaa70bb | 1599 | |
4e3fd7a0 | 1600 | rdfl.gateway = *gateway; |
86c36ce4 | 1601 | |
adaa70bb TG |
1602 | if (rt6_need_strict(dest)) |
1603 | flags |= RT6_LOOKUP_F_IFACE; | |
a6279458 | 1604 | |
4c9483b2 | 1605 | return (struct rt6_info *)fib6_rule_lookup(net, &rdfl.fl6, |
58f09b78 | 1606 | flags, __ip6_route_redirect); |
a6279458 YH |
1607 | } |
1608 | ||
b71d1d42 ED |
1609 | void rt6_redirect(const struct in6_addr *dest, const struct in6_addr *src, |
1610 | const struct in6_addr *saddr, | |
a6279458 YH |
1611 | struct neighbour *neigh, u8 *lladdr, int on_link) |
1612 | { | |
1613 | struct rt6_info *rt, *nrt = NULL; | |
1614 | struct netevent_redirect netevent; | |
c346dca1 | 1615 | struct net *net = dev_net(neigh->dev); |
a6279458 YH |
1616 | |
1617 | rt = ip6_route_redirect(dest, src, saddr, neigh->dev); | |
1618 | ||
8ed67789 | 1619 | if (rt == net->ipv6.ip6_null_entry) { |
1da177e4 LT |
1620 | if (net_ratelimit()) |
1621 | printk(KERN_DEBUG "rt6_redirect: source isn't a valid nexthop " | |
1622 | "for redirect target\n"); | |
a6279458 | 1623 | goto out; |
1da177e4 LT |
1624 | } |
1625 | ||
1da177e4 LT |
1626 | /* |
1627 | * We have finally decided to accept it. | |
1628 | */ | |
1629 | ||
1ab1457c | 1630 | neigh_update(neigh, lladdr, NUD_STALE, |
1da177e4 LT |
1631 | NEIGH_UPDATE_F_WEAK_OVERRIDE| |
1632 | NEIGH_UPDATE_F_OVERRIDE| | |
1633 | (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER| | |
1634 | NEIGH_UPDATE_F_ISROUTER)) | |
1635 | ); | |
1636 | ||
1637 | /* | |
1638 | * Redirect received -> path was valid. | |
1639 | * Look, redirects are sent only in response to data packets, | |
1640 | * so that this nexthop apparently is reachable. --ANK | |
1641 | */ | |
d8d1f30b | 1642 | dst_confirm(&rt->dst); |
1da177e4 LT |
1643 | |
1644 | /* Duplicate redirect: silently ignore. */ | |
27217455 | 1645 | if (neigh == dst_get_neighbour_noref_raw(&rt->dst)) |
1da177e4 LT |
1646 | goto out; |
1647 | ||
21efcfa0 | 1648 | nrt = ip6_rt_copy(rt, dest); |
38308473 | 1649 | if (!nrt) |
1da177e4 LT |
1650 | goto out; |
1651 | ||
1652 | nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE; | |
1653 | if (on_link) | |
1654 | nrt->rt6i_flags &= ~RTF_GATEWAY; | |
1655 | ||
4e3fd7a0 | 1656 | nrt->rt6i_gateway = *(struct in6_addr *)neigh->primary_key; |
69cce1d1 | 1657 | dst_set_neighbour(&nrt->dst, neigh_clone(neigh)); |
1da177e4 | 1658 | |
40e22e8f | 1659 | if (ip6_ins_rt(nrt)) |
1da177e4 LT |
1660 | goto out; |
1661 | ||
d8d1f30b CG |
1662 | netevent.old = &rt->dst; |
1663 | netevent.new = &nrt->dst; | |
8d71740c TT |
1664 | call_netevent_notifiers(NETEVENT_REDIRECT, &netevent); |
1665 | ||
38308473 | 1666 | if (rt->rt6i_flags & RTF_CACHE) { |
e0a1ad73 | 1667 | ip6_del_rt(rt); |
1da177e4 LT |
1668 | return; |
1669 | } | |
1670 | ||
1671 | out: | |
d8d1f30b | 1672 | dst_release(&rt->dst); |
1da177e4 LT |
1673 | } |
1674 | ||
1675 | /* | |
1676 | * Handle ICMP "packet too big" messages | |
1677 | * i.e. Path MTU discovery | |
1678 | */ | |
1679 | ||
b71d1d42 | 1680 | static void rt6_do_pmtu_disc(const struct in6_addr *daddr, const struct in6_addr *saddr, |
ae878ae2 | 1681 | struct net *net, u32 pmtu, int ifindex) |
1da177e4 LT |
1682 | { |
1683 | struct rt6_info *rt, *nrt; | |
1684 | int allfrag = 0; | |
d3052b55 | 1685 | again: |
ae878ae2 | 1686 | rt = rt6_lookup(net, daddr, saddr, ifindex, 0); |
38308473 | 1687 | if (!rt) |
1da177e4 LT |
1688 | return; |
1689 | ||
d3052b55 AV |
1690 | if (rt6_check_expired(rt)) { |
1691 | ip6_del_rt(rt); | |
1692 | goto again; | |
1693 | } | |
1694 | ||
d8d1f30b | 1695 | if (pmtu >= dst_mtu(&rt->dst)) |
1da177e4 LT |
1696 | goto out; |
1697 | ||
1698 | if (pmtu < IPV6_MIN_MTU) { | |
1699 | /* | |
1ab1457c | 1700 | * According to RFC2460, PMTU is set to the IPv6 Minimum Link |
1da177e4 LT |
1701 | * MTU (1280) and a fragment header should always be included |
1702 | * after a node receiving Too Big message reporting PMTU is | |
1703 | * less than the IPv6 Minimum Link MTU. | |
1704 | */ | |
1705 | pmtu = IPV6_MIN_MTU; | |
1706 | allfrag = 1; | |
1707 | } | |
1708 | ||
1709 | /* New mtu received -> path was valid. | |
1710 | They are sent only in response to data packets, | |
1711 | so that this nexthop apparently is reachable. --ANK | |
1712 | */ | |
d8d1f30b | 1713 | dst_confirm(&rt->dst); |
1da177e4 LT |
1714 | |
1715 | /* Host route. If it is static, it would be better | |
1716 | not to override it, but add new one, so that | |
1717 | when cache entry will expire old pmtu | |
1718 | would return automatically. | |
1719 | */ | |
1720 | if (rt->rt6i_flags & RTF_CACHE) { | |
defb3519 DM |
1721 | dst_metric_set(&rt->dst, RTAX_MTU, pmtu); |
1722 | if (allfrag) { | |
1723 | u32 features = dst_metric(&rt->dst, RTAX_FEATURES); | |
1724 | features |= RTAX_FEATURE_ALLFRAG; | |
1725 | dst_metric_set(&rt->dst, RTAX_FEATURES, features); | |
1726 | } | |
d8d1f30b | 1727 | dst_set_expires(&rt->dst, net->ipv6.sysctl.ip6_rt_mtu_expires); |
1da177e4 LT |
1728 | rt->rt6i_flags |= RTF_MODIFIED|RTF_EXPIRES; |
1729 | goto out; | |
1730 | } | |
1731 | ||
1732 | /* Network route. | |
1733 | Two cases are possible: | |
1734 | 1. It is connected route. Action: COW | |
1735 | 2. It is gatewayed route or NONEXTHOP route. Action: clone it. | |
1736 | */ | |
27217455 | 1737 | if (!dst_get_neighbour_noref_raw(&rt->dst) && !(rt->rt6i_flags & RTF_NONEXTHOP)) |
a1e78363 | 1738 | nrt = rt6_alloc_cow(rt, daddr, saddr); |
d5315b50 YH |
1739 | else |
1740 | nrt = rt6_alloc_clone(rt, daddr); | |
a1e78363 | 1741 | |
d5315b50 | 1742 | if (nrt) { |
defb3519 DM |
1743 | dst_metric_set(&nrt->dst, RTAX_MTU, pmtu); |
1744 | if (allfrag) { | |
1745 | u32 features = dst_metric(&nrt->dst, RTAX_FEATURES); | |
1746 | features |= RTAX_FEATURE_ALLFRAG; | |
1747 | dst_metric_set(&nrt->dst, RTAX_FEATURES, features); | |
1748 | } | |
a1e78363 YH |
1749 | |
1750 | /* According to RFC 1981, detecting PMTU increase shouldn't be | |
1751 | * happened within 5 mins, the recommended timer is 10 mins. | |
1752 | * Here this route expiration time is set to ip6_rt_mtu_expires | |
1753 | * which is 10 mins. After 10 mins the decreased pmtu is expired | |
1754 | * and detecting PMTU increase will be automatically happened. | |
1755 | */ | |
d8d1f30b | 1756 | dst_set_expires(&nrt->dst, net->ipv6.sysctl.ip6_rt_mtu_expires); |
a1e78363 YH |
1757 | nrt->rt6i_flags |= RTF_DYNAMIC|RTF_EXPIRES; |
1758 | ||
40e22e8f | 1759 | ip6_ins_rt(nrt); |
1da177e4 | 1760 | } |
1da177e4 | 1761 | out: |
d8d1f30b | 1762 | dst_release(&rt->dst); |
1da177e4 LT |
1763 | } |
1764 | ||
b71d1d42 | 1765 | void rt6_pmtu_discovery(const struct in6_addr *daddr, const struct in6_addr *saddr, |
ae878ae2 MŻ |
1766 | struct net_device *dev, u32 pmtu) |
1767 | { | |
1768 | struct net *net = dev_net(dev); | |
1769 | ||
1770 | /* | |
1771 | * RFC 1981 states that a node "MUST reduce the size of the packets it | |
1772 | * is sending along the path" that caused the Packet Too Big message. | |
1773 | * Since it's not possible in the general case to determine which | |
1774 | * interface was used to send the original packet, we update the MTU | |
1775 | * on the interface that will be used to send future packets. We also | |
1776 | * update the MTU on the interface that received the Packet Too Big in | |
1777 | * case the original packet was forced out that interface with | |
1778 | * SO_BINDTODEVICE or similar. This is the next best thing to the | |
1779 | * correct behaviour, which would be to update the MTU on all | |
1780 | * interfaces. | |
1781 | */ | |
1782 | rt6_do_pmtu_disc(daddr, saddr, net, pmtu, 0); | |
1783 | rt6_do_pmtu_disc(daddr, saddr, net, pmtu, dev->ifindex); | |
1784 | } | |
1785 | ||
1da177e4 LT |
1786 | /* |
1787 | * Misc support functions | |
1788 | */ | |
1789 | ||
21efcfa0 ED |
1790 | static struct rt6_info *ip6_rt_copy(const struct rt6_info *ort, |
1791 | const struct in6_addr *dest) | |
1da177e4 | 1792 | { |
c346dca1 | 1793 | struct net *net = dev_net(ort->rt6i_dev); |
5c1e6aa3 | 1794 | struct rt6_info *rt = ip6_dst_alloc(&net->ipv6.ip6_dst_ops, |
957c665f | 1795 | ort->dst.dev, 0); |
1da177e4 LT |
1796 | |
1797 | if (rt) { | |
d8d1f30b CG |
1798 | rt->dst.input = ort->dst.input; |
1799 | rt->dst.output = ort->dst.output; | |
8e2ec639 | 1800 | rt->dst.flags |= DST_HOST; |
d8d1f30b | 1801 | |
4e3fd7a0 | 1802 | rt->rt6i_dst.addr = *dest; |
8e2ec639 | 1803 | rt->rt6i_dst.plen = 128; |
defb3519 | 1804 | dst_copy_metrics(&rt->dst, &ort->dst); |
d8d1f30b | 1805 | rt->dst.error = ort->dst.error; |
1da177e4 LT |
1806 | rt->rt6i_idev = ort->rt6i_idev; |
1807 | if (rt->rt6i_idev) | |
1808 | in6_dev_hold(rt->rt6i_idev); | |
d8d1f30b | 1809 | rt->dst.lastuse = jiffies; |
1da177e4 LT |
1810 | rt->rt6i_expires = 0; |
1811 | ||
4e3fd7a0 | 1812 | rt->rt6i_gateway = ort->rt6i_gateway; |
1da177e4 LT |
1813 | rt->rt6i_flags = ort->rt6i_flags & ~RTF_EXPIRES; |
1814 | rt->rt6i_metric = 0; | |
1815 | ||
1da177e4 LT |
1816 | #ifdef CONFIG_IPV6_SUBTREES |
1817 | memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key)); | |
1818 | #endif | |
0f6c6392 | 1819 | memcpy(&rt->rt6i_prefsrc, &ort->rt6i_prefsrc, sizeof(struct rt6key)); |
c71099ac | 1820 | rt->rt6i_table = ort->rt6i_table; |
1da177e4 LT |
1821 | } |
1822 | return rt; | |
1823 | } | |
1824 | ||
70ceb4f5 | 1825 | #ifdef CONFIG_IPV6_ROUTE_INFO |
efa2cea0 | 1826 | static struct rt6_info *rt6_get_route_info(struct net *net, |
b71d1d42 ED |
1827 | const struct in6_addr *prefix, int prefixlen, |
1828 | const struct in6_addr *gwaddr, int ifindex) | |
70ceb4f5 YH |
1829 | { |
1830 | struct fib6_node *fn; | |
1831 | struct rt6_info *rt = NULL; | |
c71099ac TG |
1832 | struct fib6_table *table; |
1833 | ||
efa2cea0 | 1834 | table = fib6_get_table(net, RT6_TABLE_INFO); |
38308473 | 1835 | if (!table) |
c71099ac | 1836 | return NULL; |
70ceb4f5 | 1837 | |
c71099ac TG |
1838 | write_lock_bh(&table->tb6_lock); |
1839 | fn = fib6_locate(&table->tb6_root, prefix ,prefixlen, NULL, 0); | |
70ceb4f5 YH |
1840 | if (!fn) |
1841 | goto out; | |
1842 | ||
d8d1f30b | 1843 | for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) { |
70ceb4f5 YH |
1844 | if (rt->rt6i_dev->ifindex != ifindex) |
1845 | continue; | |
1846 | if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY)) | |
1847 | continue; | |
1848 | if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr)) | |
1849 | continue; | |
d8d1f30b | 1850 | dst_hold(&rt->dst); |
70ceb4f5 YH |
1851 | break; |
1852 | } | |
1853 | out: | |
c71099ac | 1854 | write_unlock_bh(&table->tb6_lock); |
70ceb4f5 YH |
1855 | return rt; |
1856 | } | |
1857 | ||
efa2cea0 | 1858 | static struct rt6_info *rt6_add_route_info(struct net *net, |
b71d1d42 ED |
1859 | const struct in6_addr *prefix, int prefixlen, |
1860 | const struct in6_addr *gwaddr, int ifindex, | |
70ceb4f5 YH |
1861 | unsigned pref) |
1862 | { | |
86872cb5 TG |
1863 | struct fib6_config cfg = { |
1864 | .fc_table = RT6_TABLE_INFO, | |
238fc7ea | 1865 | .fc_metric = IP6_RT_PRIO_USER, |
86872cb5 TG |
1866 | .fc_ifindex = ifindex, |
1867 | .fc_dst_len = prefixlen, | |
1868 | .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO | | |
1869 | RTF_UP | RTF_PREF(pref), | |
efa2cea0 DL |
1870 | .fc_nlinfo.pid = 0, |
1871 | .fc_nlinfo.nlh = NULL, | |
1872 | .fc_nlinfo.nl_net = net, | |
86872cb5 TG |
1873 | }; |
1874 | ||
4e3fd7a0 AD |
1875 | cfg.fc_dst = *prefix; |
1876 | cfg.fc_gateway = *gwaddr; | |
70ceb4f5 | 1877 | |
e317da96 YH |
1878 | /* We should treat it as a default route if prefix length is 0. */ |
1879 | if (!prefixlen) | |
86872cb5 | 1880 | cfg.fc_flags |= RTF_DEFAULT; |
70ceb4f5 | 1881 | |
86872cb5 | 1882 | ip6_route_add(&cfg); |
70ceb4f5 | 1883 | |
efa2cea0 | 1884 | return rt6_get_route_info(net, prefix, prefixlen, gwaddr, ifindex); |
70ceb4f5 YH |
1885 | } |
1886 | #endif | |
1887 | ||
b71d1d42 | 1888 | struct rt6_info *rt6_get_dflt_router(const struct in6_addr *addr, struct net_device *dev) |
1ab1457c | 1889 | { |
1da177e4 | 1890 | struct rt6_info *rt; |
c71099ac | 1891 | struct fib6_table *table; |
1da177e4 | 1892 | |
c346dca1 | 1893 | table = fib6_get_table(dev_net(dev), RT6_TABLE_DFLT); |
38308473 | 1894 | if (!table) |
c71099ac | 1895 | return NULL; |
1da177e4 | 1896 | |
c71099ac | 1897 | write_lock_bh(&table->tb6_lock); |
d8d1f30b | 1898 | for (rt = table->tb6_root.leaf; rt; rt=rt->dst.rt6_next) { |
1da177e4 | 1899 | if (dev == rt->rt6i_dev && |
045927ff | 1900 | ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) && |
1da177e4 LT |
1901 | ipv6_addr_equal(&rt->rt6i_gateway, addr)) |
1902 | break; | |
1903 | } | |
1904 | if (rt) | |
d8d1f30b | 1905 | dst_hold(&rt->dst); |
c71099ac | 1906 | write_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
1907 | return rt; |
1908 | } | |
1909 | ||
b71d1d42 | 1910 | struct rt6_info *rt6_add_dflt_router(const struct in6_addr *gwaddr, |
ebacaaa0 YH |
1911 | struct net_device *dev, |
1912 | unsigned int pref) | |
1da177e4 | 1913 | { |
86872cb5 TG |
1914 | struct fib6_config cfg = { |
1915 | .fc_table = RT6_TABLE_DFLT, | |
238fc7ea | 1916 | .fc_metric = IP6_RT_PRIO_USER, |
86872cb5 TG |
1917 | .fc_ifindex = dev->ifindex, |
1918 | .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT | | |
1919 | RTF_UP | RTF_EXPIRES | RTF_PREF(pref), | |
5578689a DL |
1920 | .fc_nlinfo.pid = 0, |
1921 | .fc_nlinfo.nlh = NULL, | |
c346dca1 | 1922 | .fc_nlinfo.nl_net = dev_net(dev), |
86872cb5 | 1923 | }; |
1da177e4 | 1924 | |
4e3fd7a0 | 1925 | cfg.fc_gateway = *gwaddr; |
1da177e4 | 1926 | |
86872cb5 | 1927 | ip6_route_add(&cfg); |
1da177e4 | 1928 | |
1da177e4 LT |
1929 | return rt6_get_dflt_router(gwaddr, dev); |
1930 | } | |
1931 | ||
7b4da532 | 1932 | void rt6_purge_dflt_routers(struct net *net) |
1da177e4 LT |
1933 | { |
1934 | struct rt6_info *rt; | |
c71099ac TG |
1935 | struct fib6_table *table; |
1936 | ||
1937 | /* NOTE: Keep consistent with rt6_get_dflt_router */ | |
7b4da532 | 1938 | table = fib6_get_table(net, RT6_TABLE_DFLT); |
38308473 | 1939 | if (!table) |
c71099ac | 1940 | return; |
1da177e4 LT |
1941 | |
1942 | restart: | |
c71099ac | 1943 | read_lock_bh(&table->tb6_lock); |
d8d1f30b | 1944 | for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) { |
1da177e4 | 1945 | if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF)) { |
d8d1f30b | 1946 | dst_hold(&rt->dst); |
c71099ac | 1947 | read_unlock_bh(&table->tb6_lock); |
e0a1ad73 | 1948 | ip6_del_rt(rt); |
1da177e4 LT |
1949 | goto restart; |
1950 | } | |
1951 | } | |
c71099ac | 1952 | read_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
1953 | } |
1954 | ||
5578689a DL |
1955 | static void rtmsg_to_fib6_config(struct net *net, |
1956 | struct in6_rtmsg *rtmsg, | |
86872cb5 TG |
1957 | struct fib6_config *cfg) |
1958 | { | |
1959 | memset(cfg, 0, sizeof(*cfg)); | |
1960 | ||
1961 | cfg->fc_table = RT6_TABLE_MAIN; | |
1962 | cfg->fc_ifindex = rtmsg->rtmsg_ifindex; | |
1963 | cfg->fc_metric = rtmsg->rtmsg_metric; | |
1964 | cfg->fc_expires = rtmsg->rtmsg_info; | |
1965 | cfg->fc_dst_len = rtmsg->rtmsg_dst_len; | |
1966 | cfg->fc_src_len = rtmsg->rtmsg_src_len; | |
1967 | cfg->fc_flags = rtmsg->rtmsg_flags; | |
1968 | ||
5578689a | 1969 | cfg->fc_nlinfo.nl_net = net; |
f1243c2d | 1970 | |
4e3fd7a0 AD |
1971 | cfg->fc_dst = rtmsg->rtmsg_dst; |
1972 | cfg->fc_src = rtmsg->rtmsg_src; | |
1973 | cfg->fc_gateway = rtmsg->rtmsg_gateway; | |
86872cb5 TG |
1974 | } |
1975 | ||
5578689a | 1976 | int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg) |
1da177e4 | 1977 | { |
86872cb5 | 1978 | struct fib6_config cfg; |
1da177e4 LT |
1979 | struct in6_rtmsg rtmsg; |
1980 | int err; | |
1981 | ||
1982 | switch(cmd) { | |
1983 | case SIOCADDRT: /* Add a route */ | |
1984 | case SIOCDELRT: /* Delete a route */ | |
1985 | if (!capable(CAP_NET_ADMIN)) | |
1986 | return -EPERM; | |
1987 | err = copy_from_user(&rtmsg, arg, | |
1988 | sizeof(struct in6_rtmsg)); | |
1989 | if (err) | |
1990 | return -EFAULT; | |
86872cb5 | 1991 | |
5578689a | 1992 | rtmsg_to_fib6_config(net, &rtmsg, &cfg); |
86872cb5 | 1993 | |
1da177e4 LT |
1994 | rtnl_lock(); |
1995 | switch (cmd) { | |
1996 | case SIOCADDRT: | |
86872cb5 | 1997 | err = ip6_route_add(&cfg); |
1da177e4 LT |
1998 | break; |
1999 | case SIOCDELRT: | |
86872cb5 | 2000 | err = ip6_route_del(&cfg); |
1da177e4 LT |
2001 | break; |
2002 | default: | |
2003 | err = -EINVAL; | |
2004 | } | |
2005 | rtnl_unlock(); | |
2006 | ||
2007 | return err; | |
3ff50b79 | 2008 | } |
1da177e4 LT |
2009 | |
2010 | return -EINVAL; | |
2011 | } | |
2012 | ||
2013 | /* | |
2014 | * Drop the packet on the floor | |
2015 | */ | |
2016 | ||
d5fdd6ba | 2017 | static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes) |
1da177e4 | 2018 | { |
612f09e8 | 2019 | int type; |
adf30907 | 2020 | struct dst_entry *dst = skb_dst(skb); |
612f09e8 YH |
2021 | switch (ipstats_mib_noroutes) { |
2022 | case IPSTATS_MIB_INNOROUTES: | |
0660e03f | 2023 | type = ipv6_addr_type(&ipv6_hdr(skb)->daddr); |
45bb0060 | 2024 | if (type == IPV6_ADDR_ANY) { |
3bd653c8 DL |
2025 | IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst), |
2026 | IPSTATS_MIB_INADDRERRORS); | |
612f09e8 YH |
2027 | break; |
2028 | } | |
2029 | /* FALLTHROUGH */ | |
2030 | case IPSTATS_MIB_OUTNOROUTES: | |
3bd653c8 DL |
2031 | IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst), |
2032 | ipstats_mib_noroutes); | |
612f09e8 YH |
2033 | break; |
2034 | } | |
3ffe533c | 2035 | icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0); |
1da177e4 LT |
2036 | kfree_skb(skb); |
2037 | return 0; | |
2038 | } | |
2039 | ||
9ce8ade0 TG |
2040 | static int ip6_pkt_discard(struct sk_buff *skb) |
2041 | { | |
612f09e8 | 2042 | return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES); |
9ce8ade0 TG |
2043 | } |
2044 | ||
20380731 | 2045 | static int ip6_pkt_discard_out(struct sk_buff *skb) |
1da177e4 | 2046 | { |
adf30907 | 2047 | skb->dev = skb_dst(skb)->dev; |
612f09e8 | 2048 | return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES); |
1da177e4 LT |
2049 | } |
2050 | ||
6723ab54 DM |
2051 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES |
2052 | ||
9ce8ade0 TG |
2053 | static int ip6_pkt_prohibit(struct sk_buff *skb) |
2054 | { | |
612f09e8 | 2055 | return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES); |
9ce8ade0 TG |
2056 | } |
2057 | ||
2058 | static int ip6_pkt_prohibit_out(struct sk_buff *skb) | |
2059 | { | |
adf30907 | 2060 | skb->dev = skb_dst(skb)->dev; |
612f09e8 | 2061 | return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES); |
9ce8ade0 TG |
2062 | } |
2063 | ||
6723ab54 DM |
2064 | #endif |
2065 | ||
1da177e4 LT |
2066 | /* |
2067 | * Allocate a dst for local (unicast / anycast) address. | |
2068 | */ | |
2069 | ||
2070 | struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev, | |
2071 | const struct in6_addr *addr, | |
8f031519 | 2072 | bool anycast) |
1da177e4 | 2073 | { |
c346dca1 | 2074 | struct net *net = dev_net(idev->dev); |
5c1e6aa3 | 2075 | struct rt6_info *rt = ip6_dst_alloc(&net->ipv6.ip6_dst_ops, |
957c665f | 2076 | net->loopback_dev, 0); |
14deae41 | 2077 | struct neighbour *neigh; |
1da177e4 | 2078 | |
38308473 | 2079 | if (!rt) { |
40385653 BG |
2080 | if (net_ratelimit()) |
2081 | pr_warning("IPv6: Maximum number of routes reached," | |
2082 | " consider increasing route/max_size.\n"); | |
1da177e4 | 2083 | return ERR_PTR(-ENOMEM); |
40385653 | 2084 | } |
1da177e4 | 2085 | |
1da177e4 LT |
2086 | in6_dev_hold(idev); |
2087 | ||
11d53b49 | 2088 | rt->dst.flags |= DST_HOST; |
d8d1f30b CG |
2089 | rt->dst.input = ip6_input; |
2090 | rt->dst.output = ip6_output; | |
1da177e4 | 2091 | rt->rt6i_idev = idev; |
d8d1f30b | 2092 | rt->dst.obsolete = -1; |
1da177e4 LT |
2093 | |
2094 | rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP; | |
58c4fb86 YH |
2095 | if (anycast) |
2096 | rt->rt6i_flags |= RTF_ANYCAST; | |
2097 | else | |
1da177e4 | 2098 | rt->rt6i_flags |= RTF_LOCAL; |
04a6f441 | 2099 | neigh = __neigh_lookup_errno(&nd_tbl, &rt->rt6i_gateway, rt->rt6i_dev); |
14deae41 | 2100 | if (IS_ERR(neigh)) { |
d8d1f30b | 2101 | dst_free(&rt->dst); |
14deae41 | 2102 | |
29546a64 | 2103 | return ERR_CAST(neigh); |
1da177e4 | 2104 | } |
69cce1d1 | 2105 | dst_set_neighbour(&rt->dst, neigh); |
1da177e4 | 2106 | |
4e3fd7a0 | 2107 | rt->rt6i_dst.addr = *addr; |
1da177e4 | 2108 | rt->rt6i_dst.plen = 128; |
5578689a | 2109 | rt->rt6i_table = fib6_get_table(net, RT6_TABLE_LOCAL); |
1da177e4 | 2110 | |
d8d1f30b | 2111 | atomic_set(&rt->dst.__refcnt, 1); |
1da177e4 LT |
2112 | |
2113 | return rt; | |
2114 | } | |
2115 | ||
c3968a85 DW |
2116 | int ip6_route_get_saddr(struct net *net, |
2117 | struct rt6_info *rt, | |
b71d1d42 | 2118 | const struct in6_addr *daddr, |
c3968a85 DW |
2119 | unsigned int prefs, |
2120 | struct in6_addr *saddr) | |
2121 | { | |
2122 | struct inet6_dev *idev = ip6_dst_idev((struct dst_entry*)rt); | |
2123 | int err = 0; | |
2124 | if (rt->rt6i_prefsrc.plen) | |
4e3fd7a0 | 2125 | *saddr = rt->rt6i_prefsrc.addr; |
c3968a85 DW |
2126 | else |
2127 | err = ipv6_dev_get_saddr(net, idev ? idev->dev : NULL, | |
2128 | daddr, prefs, saddr); | |
2129 | return err; | |
2130 | } | |
2131 | ||
2132 | /* remove deleted ip from prefsrc entries */ | |
2133 | struct arg_dev_net_ip { | |
2134 | struct net_device *dev; | |
2135 | struct net *net; | |
2136 | struct in6_addr *addr; | |
2137 | }; | |
2138 | ||
2139 | static int fib6_remove_prefsrc(struct rt6_info *rt, void *arg) | |
2140 | { | |
2141 | struct net_device *dev = ((struct arg_dev_net_ip *)arg)->dev; | |
2142 | struct net *net = ((struct arg_dev_net_ip *)arg)->net; | |
2143 | struct in6_addr *addr = ((struct arg_dev_net_ip *)arg)->addr; | |
2144 | ||
38308473 | 2145 | if (((void *)rt->rt6i_dev == dev || !dev) && |
c3968a85 DW |
2146 | rt != net->ipv6.ip6_null_entry && |
2147 | ipv6_addr_equal(addr, &rt->rt6i_prefsrc.addr)) { | |
2148 | /* remove prefsrc entry */ | |
2149 | rt->rt6i_prefsrc.plen = 0; | |
2150 | } | |
2151 | return 0; | |
2152 | } | |
2153 | ||
2154 | void rt6_remove_prefsrc(struct inet6_ifaddr *ifp) | |
2155 | { | |
2156 | struct net *net = dev_net(ifp->idev->dev); | |
2157 | struct arg_dev_net_ip adni = { | |
2158 | .dev = ifp->idev->dev, | |
2159 | .net = net, | |
2160 | .addr = &ifp->addr, | |
2161 | }; | |
2162 | fib6_clean_all(net, fib6_remove_prefsrc, 0, &adni); | |
2163 | } | |
2164 | ||
8ed67789 DL |
2165 | struct arg_dev_net { |
2166 | struct net_device *dev; | |
2167 | struct net *net; | |
2168 | }; | |
2169 | ||
1da177e4 LT |
2170 | static int fib6_ifdown(struct rt6_info *rt, void *arg) |
2171 | { | |
bc3ef660 | 2172 | const struct arg_dev_net *adn = arg; |
2173 | const struct net_device *dev = adn->dev; | |
8ed67789 | 2174 | |
38308473 | 2175 | if ((rt->rt6i_dev == dev || !dev) && |
bc3ef660 | 2176 | rt != adn->net->ipv6.ip6_null_entry) { |
1da177e4 LT |
2177 | RT6_TRACE("deleted by ifdown %p\n", rt); |
2178 | return -1; | |
2179 | } | |
2180 | return 0; | |
2181 | } | |
2182 | ||
f3db4851 | 2183 | void rt6_ifdown(struct net *net, struct net_device *dev) |
1da177e4 | 2184 | { |
8ed67789 DL |
2185 | struct arg_dev_net adn = { |
2186 | .dev = dev, | |
2187 | .net = net, | |
2188 | }; | |
2189 | ||
2190 | fib6_clean_all(net, fib6_ifdown, 0, &adn); | |
1e493d19 | 2191 | icmp6_clean_all(fib6_ifdown, &adn); |
1da177e4 LT |
2192 | } |
2193 | ||
2194 | struct rt6_mtu_change_arg | |
2195 | { | |
2196 | struct net_device *dev; | |
2197 | unsigned mtu; | |
2198 | }; | |
2199 | ||
2200 | static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg) | |
2201 | { | |
2202 | struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg; | |
2203 | struct inet6_dev *idev; | |
2204 | ||
2205 | /* In IPv6 pmtu discovery is not optional, | |
2206 | so that RTAX_MTU lock cannot disable it. | |
2207 | We still use this lock to block changes | |
2208 | caused by addrconf/ndisc. | |
2209 | */ | |
2210 | ||
2211 | idev = __in6_dev_get(arg->dev); | |
38308473 | 2212 | if (!idev) |
1da177e4 LT |
2213 | return 0; |
2214 | ||
2215 | /* For administrative MTU increase, there is no way to discover | |
2216 | IPv6 PMTU increase, so PMTU increase should be updated here. | |
2217 | Since RFC 1981 doesn't include administrative MTU increase | |
2218 | update PMTU increase is a MUST. (i.e. jumbo frame) | |
2219 | */ | |
2220 | /* | |
2221 | If new MTU is less than route PMTU, this new MTU will be the | |
2222 | lowest MTU in the path, update the route PMTU to reflect PMTU | |
2223 | decreases; if new MTU is greater than route PMTU, and the | |
2224 | old MTU is the lowest MTU in the path, update the route PMTU | |
2225 | to reflect the increase. In this case if the other nodes' MTU | |
2226 | also have the lowest MTU, TOO BIG MESSAGE will be lead to | |
2227 | PMTU discouvery. | |
2228 | */ | |
2229 | if (rt->rt6i_dev == arg->dev && | |
d8d1f30b CG |
2230 | !dst_metric_locked(&rt->dst, RTAX_MTU) && |
2231 | (dst_mtu(&rt->dst) >= arg->mtu || | |
2232 | (dst_mtu(&rt->dst) < arg->mtu && | |
2233 | dst_mtu(&rt->dst) == idev->cnf.mtu6))) { | |
defb3519 | 2234 | dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu); |
566cfd8f | 2235 | } |
1da177e4 LT |
2236 | return 0; |
2237 | } | |
2238 | ||
2239 | void rt6_mtu_change(struct net_device *dev, unsigned mtu) | |
2240 | { | |
c71099ac TG |
2241 | struct rt6_mtu_change_arg arg = { |
2242 | .dev = dev, | |
2243 | .mtu = mtu, | |
2244 | }; | |
1da177e4 | 2245 | |
c346dca1 | 2246 | fib6_clean_all(dev_net(dev), rt6_mtu_change_route, 0, &arg); |
1da177e4 LT |
2247 | } |
2248 | ||
ef7c79ed | 2249 | static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = { |
5176f91e | 2250 | [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) }, |
86872cb5 | 2251 | [RTA_OIF] = { .type = NLA_U32 }, |
ab364a6f | 2252 | [RTA_IIF] = { .type = NLA_U32 }, |
86872cb5 TG |
2253 | [RTA_PRIORITY] = { .type = NLA_U32 }, |
2254 | [RTA_METRICS] = { .type = NLA_NESTED }, | |
2255 | }; | |
2256 | ||
2257 | static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh, | |
2258 | struct fib6_config *cfg) | |
1da177e4 | 2259 | { |
86872cb5 TG |
2260 | struct rtmsg *rtm; |
2261 | struct nlattr *tb[RTA_MAX+1]; | |
2262 | int err; | |
1da177e4 | 2263 | |
86872cb5 TG |
2264 | err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy); |
2265 | if (err < 0) | |
2266 | goto errout; | |
1da177e4 | 2267 | |
86872cb5 TG |
2268 | err = -EINVAL; |
2269 | rtm = nlmsg_data(nlh); | |
2270 | memset(cfg, 0, sizeof(*cfg)); | |
2271 | ||
2272 | cfg->fc_table = rtm->rtm_table; | |
2273 | cfg->fc_dst_len = rtm->rtm_dst_len; | |
2274 | cfg->fc_src_len = rtm->rtm_src_len; | |
2275 | cfg->fc_flags = RTF_UP; | |
2276 | cfg->fc_protocol = rtm->rtm_protocol; | |
2277 | ||
2278 | if (rtm->rtm_type == RTN_UNREACHABLE) | |
2279 | cfg->fc_flags |= RTF_REJECT; | |
2280 | ||
ab79ad14 MŻ |
2281 | if (rtm->rtm_type == RTN_LOCAL) |
2282 | cfg->fc_flags |= RTF_LOCAL; | |
2283 | ||
86872cb5 TG |
2284 | cfg->fc_nlinfo.pid = NETLINK_CB(skb).pid; |
2285 | cfg->fc_nlinfo.nlh = nlh; | |
3b1e0a65 | 2286 | cfg->fc_nlinfo.nl_net = sock_net(skb->sk); |
86872cb5 TG |
2287 | |
2288 | if (tb[RTA_GATEWAY]) { | |
2289 | nla_memcpy(&cfg->fc_gateway, tb[RTA_GATEWAY], 16); | |
2290 | cfg->fc_flags |= RTF_GATEWAY; | |
1da177e4 | 2291 | } |
86872cb5 TG |
2292 | |
2293 | if (tb[RTA_DST]) { | |
2294 | int plen = (rtm->rtm_dst_len + 7) >> 3; | |
2295 | ||
2296 | if (nla_len(tb[RTA_DST]) < plen) | |
2297 | goto errout; | |
2298 | ||
2299 | nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen); | |
1da177e4 | 2300 | } |
86872cb5 TG |
2301 | |
2302 | if (tb[RTA_SRC]) { | |
2303 | int plen = (rtm->rtm_src_len + 7) >> 3; | |
2304 | ||
2305 | if (nla_len(tb[RTA_SRC]) < plen) | |
2306 | goto errout; | |
2307 | ||
2308 | nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen); | |
1da177e4 | 2309 | } |
86872cb5 | 2310 | |
c3968a85 DW |
2311 | if (tb[RTA_PREFSRC]) |
2312 | nla_memcpy(&cfg->fc_prefsrc, tb[RTA_PREFSRC], 16); | |
2313 | ||
86872cb5 TG |
2314 | if (tb[RTA_OIF]) |
2315 | cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]); | |
2316 | ||
2317 | if (tb[RTA_PRIORITY]) | |
2318 | cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]); | |
2319 | ||
2320 | if (tb[RTA_METRICS]) { | |
2321 | cfg->fc_mx = nla_data(tb[RTA_METRICS]); | |
2322 | cfg->fc_mx_len = nla_len(tb[RTA_METRICS]); | |
1da177e4 | 2323 | } |
86872cb5 TG |
2324 | |
2325 | if (tb[RTA_TABLE]) | |
2326 | cfg->fc_table = nla_get_u32(tb[RTA_TABLE]); | |
2327 | ||
2328 | err = 0; | |
2329 | errout: | |
2330 | return err; | |
1da177e4 LT |
2331 | } |
2332 | ||
c127ea2c | 2333 | static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg) |
1da177e4 | 2334 | { |
86872cb5 TG |
2335 | struct fib6_config cfg; |
2336 | int err; | |
1da177e4 | 2337 | |
86872cb5 TG |
2338 | err = rtm_to_fib6_config(skb, nlh, &cfg); |
2339 | if (err < 0) | |
2340 | return err; | |
2341 | ||
2342 | return ip6_route_del(&cfg); | |
1da177e4 LT |
2343 | } |
2344 | ||
c127ea2c | 2345 | static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg) |
1da177e4 | 2346 | { |
86872cb5 TG |
2347 | struct fib6_config cfg; |
2348 | int err; | |
1da177e4 | 2349 | |
86872cb5 TG |
2350 | err = rtm_to_fib6_config(skb, nlh, &cfg); |
2351 | if (err < 0) | |
2352 | return err; | |
2353 | ||
2354 | return ip6_route_add(&cfg); | |
1da177e4 LT |
2355 | } |
2356 | ||
339bf98f TG |
2357 | static inline size_t rt6_nlmsg_size(void) |
2358 | { | |
2359 | return NLMSG_ALIGN(sizeof(struct rtmsg)) | |
2360 | + nla_total_size(16) /* RTA_SRC */ | |
2361 | + nla_total_size(16) /* RTA_DST */ | |
2362 | + nla_total_size(16) /* RTA_GATEWAY */ | |
2363 | + nla_total_size(16) /* RTA_PREFSRC */ | |
2364 | + nla_total_size(4) /* RTA_TABLE */ | |
2365 | + nla_total_size(4) /* RTA_IIF */ | |
2366 | + nla_total_size(4) /* RTA_OIF */ | |
2367 | + nla_total_size(4) /* RTA_PRIORITY */ | |
6a2b9ce0 | 2368 | + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */ |
339bf98f TG |
2369 | + nla_total_size(sizeof(struct rta_cacheinfo)); |
2370 | } | |
2371 | ||
191cd582 BH |
2372 | static int rt6_fill_node(struct net *net, |
2373 | struct sk_buff *skb, struct rt6_info *rt, | |
0d51aa80 JHS |
2374 | struct in6_addr *dst, struct in6_addr *src, |
2375 | int iif, int type, u32 pid, u32 seq, | |
7bc570c8 | 2376 | int prefix, int nowait, unsigned int flags) |
1da177e4 LT |
2377 | { |
2378 | struct rtmsg *rtm; | |
2d7202bf | 2379 | struct nlmsghdr *nlh; |
e3703b3d | 2380 | long expires; |
9e762a4a | 2381 | u32 table; |
f2c31e32 | 2382 | struct neighbour *n; |
1da177e4 LT |
2383 | |
2384 | if (prefix) { /* user wants prefix routes only */ | |
2385 | if (!(rt->rt6i_flags & RTF_PREFIX_RT)) { | |
2386 | /* success since this is not a prefix route */ | |
2387 | return 1; | |
2388 | } | |
2389 | } | |
2390 | ||
2d7202bf | 2391 | nlh = nlmsg_put(skb, pid, seq, type, sizeof(*rtm), flags); |
38308473 | 2392 | if (!nlh) |
26932566 | 2393 | return -EMSGSIZE; |
2d7202bf TG |
2394 | |
2395 | rtm = nlmsg_data(nlh); | |
1da177e4 LT |
2396 | rtm->rtm_family = AF_INET6; |
2397 | rtm->rtm_dst_len = rt->rt6i_dst.plen; | |
2398 | rtm->rtm_src_len = rt->rt6i_src.plen; | |
2399 | rtm->rtm_tos = 0; | |
c71099ac | 2400 | if (rt->rt6i_table) |
9e762a4a | 2401 | table = rt->rt6i_table->tb6_id; |
c71099ac | 2402 | else |
9e762a4a PM |
2403 | table = RT6_TABLE_UNSPEC; |
2404 | rtm->rtm_table = table; | |
2d7202bf | 2405 | NLA_PUT_U32(skb, RTA_TABLE, table); |
38308473 | 2406 | if (rt->rt6i_flags & RTF_REJECT) |
1da177e4 | 2407 | rtm->rtm_type = RTN_UNREACHABLE; |
38308473 | 2408 | else if (rt->rt6i_flags & RTF_LOCAL) |
ab79ad14 | 2409 | rtm->rtm_type = RTN_LOCAL; |
38308473 | 2410 | else if (rt->rt6i_dev && (rt->rt6i_dev->flags & IFF_LOOPBACK)) |
1da177e4 LT |
2411 | rtm->rtm_type = RTN_LOCAL; |
2412 | else | |
2413 | rtm->rtm_type = RTN_UNICAST; | |
2414 | rtm->rtm_flags = 0; | |
2415 | rtm->rtm_scope = RT_SCOPE_UNIVERSE; | |
2416 | rtm->rtm_protocol = rt->rt6i_protocol; | |
38308473 | 2417 | if (rt->rt6i_flags & RTF_DYNAMIC) |
1da177e4 LT |
2418 | rtm->rtm_protocol = RTPROT_REDIRECT; |
2419 | else if (rt->rt6i_flags & RTF_ADDRCONF) | |
2420 | rtm->rtm_protocol = RTPROT_KERNEL; | |
38308473 | 2421 | else if (rt->rt6i_flags & RTF_DEFAULT) |
1da177e4 LT |
2422 | rtm->rtm_protocol = RTPROT_RA; |
2423 | ||
38308473 | 2424 | if (rt->rt6i_flags & RTF_CACHE) |
1da177e4 LT |
2425 | rtm->rtm_flags |= RTM_F_CLONED; |
2426 | ||
2427 | if (dst) { | |
2d7202bf | 2428 | NLA_PUT(skb, RTA_DST, 16, dst); |
1ab1457c | 2429 | rtm->rtm_dst_len = 128; |
1da177e4 | 2430 | } else if (rtm->rtm_dst_len) |
2d7202bf | 2431 | NLA_PUT(skb, RTA_DST, 16, &rt->rt6i_dst.addr); |
1da177e4 LT |
2432 | #ifdef CONFIG_IPV6_SUBTREES |
2433 | if (src) { | |
2d7202bf | 2434 | NLA_PUT(skb, RTA_SRC, 16, src); |
1ab1457c | 2435 | rtm->rtm_src_len = 128; |
1da177e4 | 2436 | } else if (rtm->rtm_src_len) |
2d7202bf | 2437 | NLA_PUT(skb, RTA_SRC, 16, &rt->rt6i_src.addr); |
1da177e4 | 2438 | #endif |
7bc570c8 YH |
2439 | if (iif) { |
2440 | #ifdef CONFIG_IPV6_MROUTE | |
2441 | if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) { | |
8229efda | 2442 | int err = ip6mr_get_route(net, skb, rtm, nowait); |
7bc570c8 YH |
2443 | if (err <= 0) { |
2444 | if (!nowait) { | |
2445 | if (err == 0) | |
2446 | return 0; | |
2447 | goto nla_put_failure; | |
2448 | } else { | |
2449 | if (err == -EMSGSIZE) | |
2450 | goto nla_put_failure; | |
2451 | } | |
2452 | } | |
2453 | } else | |
2454 | #endif | |
2455 | NLA_PUT_U32(skb, RTA_IIF, iif); | |
2456 | } else if (dst) { | |
1da177e4 | 2457 | struct in6_addr saddr_buf; |
c3968a85 | 2458 | if (ip6_route_get_saddr(net, rt, dst, 0, &saddr_buf) == 0) |
2d7202bf | 2459 | NLA_PUT(skb, RTA_PREFSRC, 16, &saddr_buf); |
1da177e4 | 2460 | } |
2d7202bf | 2461 | |
c3968a85 DW |
2462 | if (rt->rt6i_prefsrc.plen) { |
2463 | struct in6_addr saddr_buf; | |
4e3fd7a0 | 2464 | saddr_buf = rt->rt6i_prefsrc.addr; |
c3968a85 DW |
2465 | NLA_PUT(skb, RTA_PREFSRC, 16, &saddr_buf); |
2466 | } | |
2467 | ||
defb3519 | 2468 | if (rtnetlink_put_metrics(skb, dst_metrics_ptr(&rt->dst)) < 0) |
2d7202bf TG |
2469 | goto nla_put_failure; |
2470 | ||
f2c31e32 | 2471 | rcu_read_lock(); |
27217455 | 2472 | n = dst_get_neighbour_noref(&rt->dst); |
f2c31e32 ED |
2473 | if (n) |
2474 | NLA_PUT(skb, RTA_GATEWAY, 16, &n->primary_key); | |
2475 | rcu_read_unlock(); | |
2d7202bf | 2476 | |
d8d1f30b | 2477 | if (rt->dst.dev) |
2d7202bf TG |
2478 | NLA_PUT_U32(skb, RTA_OIF, rt->rt6i_dev->ifindex); |
2479 | ||
2480 | NLA_PUT_U32(skb, RTA_PRIORITY, rt->rt6i_metric); | |
e3703b3d | 2481 | |
36e3deae YH |
2482 | if (!(rt->rt6i_flags & RTF_EXPIRES)) |
2483 | expires = 0; | |
2484 | else if (rt->rt6i_expires - jiffies < INT_MAX) | |
2485 | expires = rt->rt6i_expires - jiffies; | |
2486 | else | |
2487 | expires = INT_MAX; | |
69cdf8f9 | 2488 | |
d8d1f30b CG |
2489 | if (rtnl_put_cacheinfo(skb, &rt->dst, 0, 0, 0, |
2490 | expires, rt->dst.error) < 0) | |
e3703b3d | 2491 | goto nla_put_failure; |
2d7202bf TG |
2492 | |
2493 | return nlmsg_end(skb, nlh); | |
2494 | ||
2495 | nla_put_failure: | |
26932566 PM |
2496 | nlmsg_cancel(skb, nlh); |
2497 | return -EMSGSIZE; | |
1da177e4 LT |
2498 | } |
2499 | ||
1b43af54 | 2500 | int rt6_dump_route(struct rt6_info *rt, void *p_arg) |
1da177e4 LT |
2501 | { |
2502 | struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg; | |
2503 | int prefix; | |
2504 | ||
2d7202bf TG |
2505 | if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) { |
2506 | struct rtmsg *rtm = nlmsg_data(arg->cb->nlh); | |
1da177e4 LT |
2507 | prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0; |
2508 | } else | |
2509 | prefix = 0; | |
2510 | ||
191cd582 BH |
2511 | return rt6_fill_node(arg->net, |
2512 | arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE, | |
1da177e4 | 2513 | NETLINK_CB(arg->cb->skb).pid, arg->cb->nlh->nlmsg_seq, |
7bc570c8 | 2514 | prefix, 0, NLM_F_MULTI); |
1da177e4 LT |
2515 | } |
2516 | ||
c127ea2c | 2517 | static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr* nlh, void *arg) |
1da177e4 | 2518 | { |
3b1e0a65 | 2519 | struct net *net = sock_net(in_skb->sk); |
ab364a6f TG |
2520 | struct nlattr *tb[RTA_MAX+1]; |
2521 | struct rt6_info *rt; | |
1da177e4 | 2522 | struct sk_buff *skb; |
ab364a6f | 2523 | struct rtmsg *rtm; |
4c9483b2 | 2524 | struct flowi6 fl6; |
ab364a6f | 2525 | int err, iif = 0; |
1da177e4 | 2526 | |
ab364a6f TG |
2527 | err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy); |
2528 | if (err < 0) | |
2529 | goto errout; | |
1da177e4 | 2530 | |
ab364a6f | 2531 | err = -EINVAL; |
4c9483b2 | 2532 | memset(&fl6, 0, sizeof(fl6)); |
1da177e4 | 2533 | |
ab364a6f TG |
2534 | if (tb[RTA_SRC]) { |
2535 | if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr)) | |
2536 | goto errout; | |
2537 | ||
4e3fd7a0 | 2538 | fl6.saddr = *(struct in6_addr *)nla_data(tb[RTA_SRC]); |
ab364a6f TG |
2539 | } |
2540 | ||
2541 | if (tb[RTA_DST]) { | |
2542 | if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr)) | |
2543 | goto errout; | |
2544 | ||
4e3fd7a0 | 2545 | fl6.daddr = *(struct in6_addr *)nla_data(tb[RTA_DST]); |
ab364a6f TG |
2546 | } |
2547 | ||
2548 | if (tb[RTA_IIF]) | |
2549 | iif = nla_get_u32(tb[RTA_IIF]); | |
2550 | ||
2551 | if (tb[RTA_OIF]) | |
4c9483b2 | 2552 | fl6.flowi6_oif = nla_get_u32(tb[RTA_OIF]); |
1da177e4 LT |
2553 | |
2554 | if (iif) { | |
2555 | struct net_device *dev; | |
5578689a | 2556 | dev = __dev_get_by_index(net, iif); |
1da177e4 LT |
2557 | if (!dev) { |
2558 | err = -ENODEV; | |
ab364a6f | 2559 | goto errout; |
1da177e4 LT |
2560 | } |
2561 | } | |
2562 | ||
ab364a6f | 2563 | skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL); |
38308473 | 2564 | if (!skb) { |
ab364a6f TG |
2565 | err = -ENOBUFS; |
2566 | goto errout; | |
2567 | } | |
1da177e4 | 2568 | |
ab364a6f TG |
2569 | /* Reserve room for dummy headers, this skb can pass |
2570 | through good chunk of routing engine. | |
2571 | */ | |
459a98ed | 2572 | skb_reset_mac_header(skb); |
ab364a6f | 2573 | skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr)); |
1da177e4 | 2574 | |
4c9483b2 | 2575 | rt = (struct rt6_info*) ip6_route_output(net, NULL, &fl6); |
d8d1f30b | 2576 | skb_dst_set(skb, &rt->dst); |
1da177e4 | 2577 | |
4c9483b2 | 2578 | err = rt6_fill_node(net, skb, rt, &fl6.daddr, &fl6.saddr, iif, |
1da177e4 | 2579 | RTM_NEWROUTE, NETLINK_CB(in_skb).pid, |
7bc570c8 | 2580 | nlh->nlmsg_seq, 0, 0, 0); |
1da177e4 | 2581 | if (err < 0) { |
ab364a6f TG |
2582 | kfree_skb(skb); |
2583 | goto errout; | |
1da177e4 LT |
2584 | } |
2585 | ||
5578689a | 2586 | err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).pid); |
ab364a6f | 2587 | errout: |
1da177e4 | 2588 | return err; |
1da177e4 LT |
2589 | } |
2590 | ||
86872cb5 | 2591 | void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info) |
1da177e4 LT |
2592 | { |
2593 | struct sk_buff *skb; | |
5578689a | 2594 | struct net *net = info->nl_net; |
528c4ceb DL |
2595 | u32 seq; |
2596 | int err; | |
2597 | ||
2598 | err = -ENOBUFS; | |
38308473 | 2599 | seq = info->nlh ? info->nlh->nlmsg_seq : 0; |
86872cb5 | 2600 | |
339bf98f | 2601 | skb = nlmsg_new(rt6_nlmsg_size(), gfp_any()); |
38308473 | 2602 | if (!skb) |
21713ebc TG |
2603 | goto errout; |
2604 | ||
191cd582 | 2605 | err = rt6_fill_node(net, skb, rt, NULL, NULL, 0, |
7bc570c8 | 2606 | event, info->pid, seq, 0, 0, 0); |
26932566 PM |
2607 | if (err < 0) { |
2608 | /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */ | |
2609 | WARN_ON(err == -EMSGSIZE); | |
2610 | kfree_skb(skb); | |
2611 | goto errout; | |
2612 | } | |
1ce85fe4 PNA |
2613 | rtnl_notify(skb, net, info->pid, RTNLGRP_IPV6_ROUTE, |
2614 | info->nlh, gfp_any()); | |
2615 | return; | |
21713ebc TG |
2616 | errout: |
2617 | if (err < 0) | |
5578689a | 2618 | rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err); |
1da177e4 LT |
2619 | } |
2620 | ||
8ed67789 DL |
2621 | static int ip6_route_dev_notify(struct notifier_block *this, |
2622 | unsigned long event, void *data) | |
2623 | { | |
2624 | struct net_device *dev = (struct net_device *)data; | |
c346dca1 | 2625 | struct net *net = dev_net(dev); |
8ed67789 DL |
2626 | |
2627 | if (event == NETDEV_REGISTER && (dev->flags & IFF_LOOPBACK)) { | |
d8d1f30b | 2628 | net->ipv6.ip6_null_entry->dst.dev = dev; |
8ed67789 DL |
2629 | net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev); |
2630 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES | |
d8d1f30b | 2631 | net->ipv6.ip6_prohibit_entry->dst.dev = dev; |
8ed67789 | 2632 | net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev); |
d8d1f30b | 2633 | net->ipv6.ip6_blk_hole_entry->dst.dev = dev; |
8ed67789 DL |
2634 | net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev); |
2635 | #endif | |
2636 | } | |
2637 | ||
2638 | return NOTIFY_OK; | |
2639 | } | |
2640 | ||
1da177e4 LT |
2641 | /* |
2642 | * /proc | |
2643 | */ | |
2644 | ||
2645 | #ifdef CONFIG_PROC_FS | |
2646 | ||
1da177e4 LT |
2647 | struct rt6_proc_arg |
2648 | { | |
2649 | char *buffer; | |
2650 | int offset; | |
2651 | int length; | |
2652 | int skip; | |
2653 | int len; | |
2654 | }; | |
2655 | ||
2656 | static int rt6_info_route(struct rt6_info *rt, void *p_arg) | |
2657 | { | |
33120b30 | 2658 | struct seq_file *m = p_arg; |
69cce1d1 | 2659 | struct neighbour *n; |
1da177e4 | 2660 | |
4b7a4274 | 2661 | seq_printf(m, "%pi6 %02x ", &rt->rt6i_dst.addr, rt->rt6i_dst.plen); |
1da177e4 LT |
2662 | |
2663 | #ifdef CONFIG_IPV6_SUBTREES | |
4b7a4274 | 2664 | seq_printf(m, "%pi6 %02x ", &rt->rt6i_src.addr, rt->rt6i_src.plen); |
1da177e4 | 2665 | #else |
33120b30 | 2666 | seq_puts(m, "00000000000000000000000000000000 00 "); |
1da177e4 | 2667 | #endif |
f2c31e32 | 2668 | rcu_read_lock(); |
27217455 | 2669 | n = dst_get_neighbour_noref(&rt->dst); |
69cce1d1 DM |
2670 | if (n) { |
2671 | seq_printf(m, "%pi6", n->primary_key); | |
1da177e4 | 2672 | } else { |
33120b30 | 2673 | seq_puts(m, "00000000000000000000000000000000"); |
1da177e4 | 2674 | } |
f2c31e32 | 2675 | rcu_read_unlock(); |
33120b30 | 2676 | seq_printf(m, " %08x %08x %08x %08x %8s\n", |
d8d1f30b CG |
2677 | rt->rt6i_metric, atomic_read(&rt->dst.__refcnt), |
2678 | rt->dst.__use, rt->rt6i_flags, | |
33120b30 | 2679 | rt->rt6i_dev ? rt->rt6i_dev->name : ""); |
1da177e4 LT |
2680 | return 0; |
2681 | } | |
2682 | ||
33120b30 | 2683 | static int ipv6_route_show(struct seq_file *m, void *v) |
1da177e4 | 2684 | { |
f3db4851 DL |
2685 | struct net *net = (struct net *)m->private; |
2686 | fib6_clean_all(net, rt6_info_route, 0, m); | |
33120b30 AD |
2687 | return 0; |
2688 | } | |
1da177e4 | 2689 | |
33120b30 AD |
2690 | static int ipv6_route_open(struct inode *inode, struct file *file) |
2691 | { | |
de05c557 | 2692 | return single_open_net(inode, file, ipv6_route_show); |
f3db4851 DL |
2693 | } |
2694 | ||
33120b30 AD |
2695 | static const struct file_operations ipv6_route_proc_fops = { |
2696 | .owner = THIS_MODULE, | |
2697 | .open = ipv6_route_open, | |
2698 | .read = seq_read, | |
2699 | .llseek = seq_lseek, | |
b6fcbdb4 | 2700 | .release = single_release_net, |
33120b30 AD |
2701 | }; |
2702 | ||
1da177e4 LT |
2703 | static int rt6_stats_seq_show(struct seq_file *seq, void *v) |
2704 | { | |
69ddb805 | 2705 | struct net *net = (struct net *)seq->private; |
1da177e4 | 2706 | seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n", |
69ddb805 DL |
2707 | net->ipv6.rt6_stats->fib_nodes, |
2708 | net->ipv6.rt6_stats->fib_route_nodes, | |
2709 | net->ipv6.rt6_stats->fib_rt_alloc, | |
2710 | net->ipv6.rt6_stats->fib_rt_entries, | |
2711 | net->ipv6.rt6_stats->fib_rt_cache, | |
fc66f95c | 2712 | dst_entries_get_slow(&net->ipv6.ip6_dst_ops), |
69ddb805 | 2713 | net->ipv6.rt6_stats->fib_discarded_routes); |
1da177e4 LT |
2714 | |
2715 | return 0; | |
2716 | } | |
2717 | ||
2718 | static int rt6_stats_seq_open(struct inode *inode, struct file *file) | |
2719 | { | |
de05c557 | 2720 | return single_open_net(inode, file, rt6_stats_seq_show); |
69ddb805 DL |
2721 | } |
2722 | ||
9a32144e | 2723 | static const struct file_operations rt6_stats_seq_fops = { |
1da177e4 LT |
2724 | .owner = THIS_MODULE, |
2725 | .open = rt6_stats_seq_open, | |
2726 | .read = seq_read, | |
2727 | .llseek = seq_lseek, | |
b6fcbdb4 | 2728 | .release = single_release_net, |
1da177e4 LT |
2729 | }; |
2730 | #endif /* CONFIG_PROC_FS */ | |
2731 | ||
2732 | #ifdef CONFIG_SYSCTL | |
2733 | ||
1da177e4 | 2734 | static |
8d65af78 | 2735 | int ipv6_sysctl_rtcache_flush(ctl_table *ctl, int write, |
1da177e4 LT |
2736 | void __user *buffer, size_t *lenp, loff_t *ppos) |
2737 | { | |
c486da34 LAG |
2738 | struct net *net; |
2739 | int delay; | |
2740 | if (!write) | |
1da177e4 | 2741 | return -EINVAL; |
c486da34 LAG |
2742 | |
2743 | net = (struct net *)ctl->extra1; | |
2744 | delay = net->ipv6.sysctl.flush_delay; | |
2745 | proc_dointvec(ctl, write, buffer, lenp, ppos); | |
2746 | fib6_run_gc(delay <= 0 ? ~0UL : (unsigned long)delay, net); | |
2747 | return 0; | |
1da177e4 LT |
2748 | } |
2749 | ||
760f2d01 | 2750 | ctl_table ipv6_route_table_template[] = { |
1ab1457c | 2751 | { |
1da177e4 | 2752 | .procname = "flush", |
4990509f | 2753 | .data = &init_net.ipv6.sysctl.flush_delay, |
1da177e4 | 2754 | .maxlen = sizeof(int), |
89c8b3a1 | 2755 | .mode = 0200, |
6d9f239a | 2756 | .proc_handler = ipv6_sysctl_rtcache_flush |
1da177e4 LT |
2757 | }, |
2758 | { | |
1da177e4 | 2759 | .procname = "gc_thresh", |
9a7ec3a9 | 2760 | .data = &ip6_dst_ops_template.gc_thresh, |
1da177e4 LT |
2761 | .maxlen = sizeof(int), |
2762 | .mode = 0644, | |
6d9f239a | 2763 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2764 | }, |
2765 | { | |
1da177e4 | 2766 | .procname = "max_size", |
4990509f | 2767 | .data = &init_net.ipv6.sysctl.ip6_rt_max_size, |
1da177e4 LT |
2768 | .maxlen = sizeof(int), |
2769 | .mode = 0644, | |
6d9f239a | 2770 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2771 | }, |
2772 | { | |
1da177e4 | 2773 | .procname = "gc_min_interval", |
4990509f | 2774 | .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval, |
1da177e4 LT |
2775 | .maxlen = sizeof(int), |
2776 | .mode = 0644, | |
6d9f239a | 2777 | .proc_handler = proc_dointvec_jiffies, |
1da177e4 LT |
2778 | }, |
2779 | { | |
1da177e4 | 2780 | .procname = "gc_timeout", |
4990509f | 2781 | .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout, |
1da177e4 LT |
2782 | .maxlen = sizeof(int), |
2783 | .mode = 0644, | |
6d9f239a | 2784 | .proc_handler = proc_dointvec_jiffies, |
1da177e4 LT |
2785 | }, |
2786 | { | |
1da177e4 | 2787 | .procname = "gc_interval", |
4990509f | 2788 | .data = &init_net.ipv6.sysctl.ip6_rt_gc_interval, |
1da177e4 LT |
2789 | .maxlen = sizeof(int), |
2790 | .mode = 0644, | |
6d9f239a | 2791 | .proc_handler = proc_dointvec_jiffies, |
1da177e4 LT |
2792 | }, |
2793 | { | |
1da177e4 | 2794 | .procname = "gc_elasticity", |
4990509f | 2795 | .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity, |
1da177e4 LT |
2796 | .maxlen = sizeof(int), |
2797 | .mode = 0644, | |
f3d3f616 | 2798 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2799 | }, |
2800 | { | |
1da177e4 | 2801 | .procname = "mtu_expires", |
4990509f | 2802 | .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires, |
1da177e4 LT |
2803 | .maxlen = sizeof(int), |
2804 | .mode = 0644, | |
6d9f239a | 2805 | .proc_handler = proc_dointvec_jiffies, |
1da177e4 LT |
2806 | }, |
2807 | { | |
1da177e4 | 2808 | .procname = "min_adv_mss", |
4990509f | 2809 | .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss, |
1da177e4 LT |
2810 | .maxlen = sizeof(int), |
2811 | .mode = 0644, | |
f3d3f616 | 2812 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2813 | }, |
2814 | { | |
1da177e4 | 2815 | .procname = "gc_min_interval_ms", |
4990509f | 2816 | .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval, |
1da177e4 LT |
2817 | .maxlen = sizeof(int), |
2818 | .mode = 0644, | |
6d9f239a | 2819 | .proc_handler = proc_dointvec_ms_jiffies, |
1da177e4 | 2820 | }, |
f8572d8f | 2821 | { } |
1da177e4 LT |
2822 | }; |
2823 | ||
2c8c1e72 | 2824 | struct ctl_table * __net_init ipv6_route_sysctl_init(struct net *net) |
760f2d01 DL |
2825 | { |
2826 | struct ctl_table *table; | |
2827 | ||
2828 | table = kmemdup(ipv6_route_table_template, | |
2829 | sizeof(ipv6_route_table_template), | |
2830 | GFP_KERNEL); | |
5ee09105 YH |
2831 | |
2832 | if (table) { | |
2833 | table[0].data = &net->ipv6.sysctl.flush_delay; | |
c486da34 | 2834 | table[0].extra1 = net; |
86393e52 | 2835 | table[1].data = &net->ipv6.ip6_dst_ops.gc_thresh; |
5ee09105 YH |
2836 | table[2].data = &net->ipv6.sysctl.ip6_rt_max_size; |
2837 | table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval; | |
2838 | table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout; | |
2839 | table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval; | |
2840 | table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity; | |
2841 | table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires; | |
2842 | table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss; | |
9c69fabe | 2843 | table[9].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval; |
5ee09105 YH |
2844 | } |
2845 | ||
760f2d01 DL |
2846 | return table; |
2847 | } | |
1da177e4 LT |
2848 | #endif |
2849 | ||
2c8c1e72 | 2850 | static int __net_init ip6_route_net_init(struct net *net) |
cdb18761 | 2851 | { |
633d424b | 2852 | int ret = -ENOMEM; |
8ed67789 | 2853 | |
86393e52 AD |
2854 | memcpy(&net->ipv6.ip6_dst_ops, &ip6_dst_ops_template, |
2855 | sizeof(net->ipv6.ip6_dst_ops)); | |
f2fc6a54 | 2856 | |
fc66f95c ED |
2857 | if (dst_entries_init(&net->ipv6.ip6_dst_ops) < 0) |
2858 | goto out_ip6_dst_ops; | |
2859 | ||
8ed67789 DL |
2860 | net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template, |
2861 | sizeof(*net->ipv6.ip6_null_entry), | |
2862 | GFP_KERNEL); | |
2863 | if (!net->ipv6.ip6_null_entry) | |
fc66f95c | 2864 | goto out_ip6_dst_entries; |
d8d1f30b | 2865 | net->ipv6.ip6_null_entry->dst.path = |
8ed67789 | 2866 | (struct dst_entry *)net->ipv6.ip6_null_entry; |
d8d1f30b | 2867 | net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops; |
62fa8a84 DM |
2868 | dst_init_metrics(&net->ipv6.ip6_null_entry->dst, |
2869 | ip6_template_metrics, true); | |
8ed67789 DL |
2870 | |
2871 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES | |
2872 | net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template, | |
2873 | sizeof(*net->ipv6.ip6_prohibit_entry), | |
2874 | GFP_KERNEL); | |
68fffc67 PZ |
2875 | if (!net->ipv6.ip6_prohibit_entry) |
2876 | goto out_ip6_null_entry; | |
d8d1f30b | 2877 | net->ipv6.ip6_prohibit_entry->dst.path = |
8ed67789 | 2878 | (struct dst_entry *)net->ipv6.ip6_prohibit_entry; |
d8d1f30b | 2879 | net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops; |
62fa8a84 DM |
2880 | dst_init_metrics(&net->ipv6.ip6_prohibit_entry->dst, |
2881 | ip6_template_metrics, true); | |
8ed67789 DL |
2882 | |
2883 | net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template, | |
2884 | sizeof(*net->ipv6.ip6_blk_hole_entry), | |
2885 | GFP_KERNEL); | |
68fffc67 PZ |
2886 | if (!net->ipv6.ip6_blk_hole_entry) |
2887 | goto out_ip6_prohibit_entry; | |
d8d1f30b | 2888 | net->ipv6.ip6_blk_hole_entry->dst.path = |
8ed67789 | 2889 | (struct dst_entry *)net->ipv6.ip6_blk_hole_entry; |
d8d1f30b | 2890 | net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops; |
62fa8a84 DM |
2891 | dst_init_metrics(&net->ipv6.ip6_blk_hole_entry->dst, |
2892 | ip6_template_metrics, true); | |
8ed67789 DL |
2893 | #endif |
2894 | ||
b339a47c PZ |
2895 | net->ipv6.sysctl.flush_delay = 0; |
2896 | net->ipv6.sysctl.ip6_rt_max_size = 4096; | |
2897 | net->ipv6.sysctl.ip6_rt_gc_min_interval = HZ / 2; | |
2898 | net->ipv6.sysctl.ip6_rt_gc_timeout = 60*HZ; | |
2899 | net->ipv6.sysctl.ip6_rt_gc_interval = 30*HZ; | |
2900 | net->ipv6.sysctl.ip6_rt_gc_elasticity = 9; | |
2901 | net->ipv6.sysctl.ip6_rt_mtu_expires = 10*60*HZ; | |
2902 | net->ipv6.sysctl.ip6_rt_min_advmss = IPV6_MIN_MTU - 20 - 40; | |
2903 | ||
cdb18761 DL |
2904 | #ifdef CONFIG_PROC_FS |
2905 | proc_net_fops_create(net, "ipv6_route", 0, &ipv6_route_proc_fops); | |
2906 | proc_net_fops_create(net, "rt6_stats", S_IRUGO, &rt6_stats_seq_fops); | |
2907 | #endif | |
6891a346 BT |
2908 | net->ipv6.ip6_rt_gc_expire = 30*HZ; |
2909 | ||
8ed67789 DL |
2910 | ret = 0; |
2911 | out: | |
2912 | return ret; | |
f2fc6a54 | 2913 | |
68fffc67 PZ |
2914 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES |
2915 | out_ip6_prohibit_entry: | |
2916 | kfree(net->ipv6.ip6_prohibit_entry); | |
2917 | out_ip6_null_entry: | |
2918 | kfree(net->ipv6.ip6_null_entry); | |
2919 | #endif | |
fc66f95c ED |
2920 | out_ip6_dst_entries: |
2921 | dst_entries_destroy(&net->ipv6.ip6_dst_ops); | |
f2fc6a54 | 2922 | out_ip6_dst_ops: |
f2fc6a54 | 2923 | goto out; |
cdb18761 DL |
2924 | } |
2925 | ||
2c8c1e72 | 2926 | static void __net_exit ip6_route_net_exit(struct net *net) |
cdb18761 DL |
2927 | { |
2928 | #ifdef CONFIG_PROC_FS | |
2929 | proc_net_remove(net, "ipv6_route"); | |
2930 | proc_net_remove(net, "rt6_stats"); | |
2931 | #endif | |
8ed67789 DL |
2932 | kfree(net->ipv6.ip6_null_entry); |
2933 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES | |
2934 | kfree(net->ipv6.ip6_prohibit_entry); | |
2935 | kfree(net->ipv6.ip6_blk_hole_entry); | |
2936 | #endif | |
41bb78b4 | 2937 | dst_entries_destroy(&net->ipv6.ip6_dst_ops); |
cdb18761 DL |
2938 | } |
2939 | ||
2940 | static struct pernet_operations ip6_route_net_ops = { | |
2941 | .init = ip6_route_net_init, | |
2942 | .exit = ip6_route_net_exit, | |
2943 | }; | |
2944 | ||
8ed67789 DL |
2945 | static struct notifier_block ip6_route_dev_notifier = { |
2946 | .notifier_call = ip6_route_dev_notify, | |
2947 | .priority = 0, | |
2948 | }; | |
2949 | ||
433d49c3 | 2950 | int __init ip6_route_init(void) |
1da177e4 | 2951 | { |
433d49c3 DL |
2952 | int ret; |
2953 | ||
9a7ec3a9 DL |
2954 | ret = -ENOMEM; |
2955 | ip6_dst_ops_template.kmem_cachep = | |
e5d679f3 | 2956 | kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0, |
f845ab6b | 2957 | SLAB_HWCACHE_ALIGN, NULL); |
9a7ec3a9 | 2958 | if (!ip6_dst_ops_template.kmem_cachep) |
c19a28e1 | 2959 | goto out; |
14e50e57 | 2960 | |
fc66f95c | 2961 | ret = dst_entries_init(&ip6_dst_blackhole_ops); |
8ed67789 | 2962 | if (ret) |
bdb3289f | 2963 | goto out_kmem_cache; |
bdb3289f | 2964 | |
fc66f95c ED |
2965 | ret = register_pernet_subsys(&ip6_route_net_ops); |
2966 | if (ret) | |
2967 | goto out_dst_entries; | |
2968 | ||
5dc121e9 AE |
2969 | ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep; |
2970 | ||
8ed67789 DL |
2971 | /* Registering of the loopback is done before this portion of code, |
2972 | * the loopback reference in rt6_info will not be taken, do it | |
2973 | * manually for init_net */ | |
d8d1f30b | 2974 | init_net.ipv6.ip6_null_entry->dst.dev = init_net.loopback_dev; |
8ed67789 DL |
2975 | init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev); |
2976 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES | |
d8d1f30b | 2977 | init_net.ipv6.ip6_prohibit_entry->dst.dev = init_net.loopback_dev; |
8ed67789 | 2978 | init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev); |
d8d1f30b | 2979 | init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev; |
8ed67789 DL |
2980 | init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev); |
2981 | #endif | |
433d49c3 DL |
2982 | ret = fib6_init(); |
2983 | if (ret) | |
8ed67789 | 2984 | goto out_register_subsys; |
433d49c3 | 2985 | |
433d49c3 DL |
2986 | ret = xfrm6_init(); |
2987 | if (ret) | |
cdb18761 | 2988 | goto out_fib6_init; |
c35b7e72 | 2989 | |
433d49c3 DL |
2990 | ret = fib6_rules_init(); |
2991 | if (ret) | |
2992 | goto xfrm6_init; | |
7e5449c2 | 2993 | |
433d49c3 | 2994 | ret = -ENOBUFS; |
c7ac8679 GR |
2995 | if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL, NULL) || |
2996 | __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL, NULL) || | |
2997 | __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL, NULL)) | |
433d49c3 | 2998 | goto fib6_rules_init; |
c127ea2c | 2999 | |
8ed67789 | 3000 | ret = register_netdevice_notifier(&ip6_route_dev_notifier); |
cdb18761 DL |
3001 | if (ret) |
3002 | goto fib6_rules_init; | |
8ed67789 | 3003 | |
433d49c3 DL |
3004 | out: |
3005 | return ret; | |
3006 | ||
3007 | fib6_rules_init: | |
433d49c3 DL |
3008 | fib6_rules_cleanup(); |
3009 | xfrm6_init: | |
433d49c3 | 3010 | xfrm6_fini(); |
433d49c3 | 3011 | out_fib6_init: |
433d49c3 | 3012 | fib6_gc_cleanup(); |
8ed67789 DL |
3013 | out_register_subsys: |
3014 | unregister_pernet_subsys(&ip6_route_net_ops); | |
fc66f95c ED |
3015 | out_dst_entries: |
3016 | dst_entries_destroy(&ip6_dst_blackhole_ops); | |
433d49c3 | 3017 | out_kmem_cache: |
f2fc6a54 | 3018 | kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep); |
433d49c3 | 3019 | goto out; |
1da177e4 LT |
3020 | } |
3021 | ||
3022 | void ip6_route_cleanup(void) | |
3023 | { | |
8ed67789 | 3024 | unregister_netdevice_notifier(&ip6_route_dev_notifier); |
101367c2 | 3025 | fib6_rules_cleanup(); |
1da177e4 | 3026 | xfrm6_fini(); |
1da177e4 | 3027 | fib6_gc_cleanup(); |
8ed67789 | 3028 | unregister_pernet_subsys(&ip6_route_net_ops); |
41bb78b4 | 3029 | dst_entries_destroy(&ip6_dst_blackhole_ops); |
f2fc6a54 | 3030 | kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep); |
1da177e4 | 3031 | } |