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 | ||
f3213831 JP |
27 | #define pr_fmt(fmt) "IPv6: " fmt |
28 | ||
4fc268d2 | 29 | #include <linux/capability.h> |
1da177e4 | 30 | #include <linux/errno.h> |
bc3b2d7f | 31 | #include <linux/export.h> |
1da177e4 LT |
32 | #include <linux/types.h> |
33 | #include <linux/times.h> | |
34 | #include <linux/socket.h> | |
35 | #include <linux/sockios.h> | |
36 | #include <linux/net.h> | |
37 | #include <linux/route.h> | |
38 | #include <linux/netdevice.h> | |
39 | #include <linux/in6.h> | |
7bc570c8 | 40 | #include <linux/mroute6.h> |
1da177e4 | 41 | #include <linux/init.h> |
1da177e4 | 42 | #include <linux/if_arp.h> |
1da177e4 LT |
43 | #include <linux/proc_fs.h> |
44 | #include <linux/seq_file.h> | |
5b7c931d | 45 | #include <linux/nsproxy.h> |
5a0e3ad6 | 46 | #include <linux/slab.h> |
457c4cbc | 47 | #include <net/net_namespace.h> |
1da177e4 LT |
48 | #include <net/snmp.h> |
49 | #include <net/ipv6.h> | |
50 | #include <net/ip6_fib.h> | |
51 | #include <net/ip6_route.h> | |
52 | #include <net/ndisc.h> | |
53 | #include <net/addrconf.h> | |
54 | #include <net/tcp.h> | |
55 | #include <linux/rtnetlink.h> | |
56 | #include <net/dst.h> | |
57 | #include <net/xfrm.h> | |
8d71740c | 58 | #include <net/netevent.h> |
21713ebc | 59 | #include <net/netlink.h> |
51ebd318 | 60 | #include <net/nexthop.h> |
1da177e4 LT |
61 | |
62 | #include <asm/uaccess.h> | |
63 | ||
64 | #ifdef CONFIG_SYSCTL | |
65 | #include <linux/sysctl.h> | |
66 | #endif | |
67 | ||
afc154e9 | 68 | enum rt6_nud_state { |
7e980569 JB |
69 | RT6_NUD_FAIL_HARD = -3, |
70 | RT6_NUD_FAIL_PROBE = -2, | |
71 | RT6_NUD_FAIL_DO_RR = -1, | |
afc154e9 HFS |
72 | RT6_NUD_SUCCEED = 1 |
73 | }; | |
74 | ||
1716a961 | 75 | static struct rt6_info *ip6_rt_copy(struct rt6_info *ort, |
21efcfa0 | 76 | const struct in6_addr *dest); |
1da177e4 | 77 | static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie); |
0dbaee3b | 78 | static unsigned int ip6_default_advmss(const struct dst_entry *dst); |
ebb762f2 | 79 | static unsigned int ip6_mtu(const struct dst_entry *dst); |
1da177e4 LT |
80 | static struct dst_entry *ip6_negative_advice(struct dst_entry *); |
81 | static void ip6_dst_destroy(struct dst_entry *); | |
82 | static void ip6_dst_ifdown(struct dst_entry *, | |
83 | struct net_device *dev, int how); | |
569d3645 | 84 | static int ip6_dst_gc(struct dst_ops *ops); |
1da177e4 LT |
85 | |
86 | static int ip6_pkt_discard(struct sk_buff *skb); | |
aad88724 | 87 | static int ip6_pkt_discard_out(struct sock *sk, struct sk_buff *skb); |
7150aede | 88 | static int ip6_pkt_prohibit(struct sk_buff *skb); |
aad88724 | 89 | static int ip6_pkt_prohibit_out(struct sock *sk, struct sk_buff *skb); |
1da177e4 | 90 | static void ip6_link_failure(struct sk_buff *skb); |
6700c270 DM |
91 | static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk, |
92 | struct sk_buff *skb, u32 mtu); | |
93 | static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, | |
94 | struct sk_buff *skb); | |
52bd4c0c | 95 | static int rt6_score_route(struct rt6_info *rt, int oif, int strict); |
1da177e4 | 96 | |
70ceb4f5 | 97 | #ifdef CONFIG_IPV6_ROUTE_INFO |
efa2cea0 | 98 | static struct rt6_info *rt6_add_route_info(struct net *net, |
b71d1d42 ED |
99 | const struct in6_addr *prefix, int prefixlen, |
100 | const struct in6_addr *gwaddr, int ifindex, | |
95c96174 | 101 | unsigned int pref); |
efa2cea0 | 102 | static struct rt6_info *rt6_get_route_info(struct net *net, |
b71d1d42 ED |
103 | const struct in6_addr *prefix, int prefixlen, |
104 | const struct in6_addr *gwaddr, int ifindex); | |
70ceb4f5 YH |
105 | #endif |
106 | ||
e8243534 | 107 | static void rt6_bind_peer(struct rt6_info *rt, int create) |
108 | { | |
109 | struct inet_peer_base *base; | |
110 | struct inet_peer *peer; | |
111 | ||
112 | base = inetpeer_base_ptr(rt->_rt6i_peer); | |
113 | if (!base) | |
114 | return; | |
115 | ||
116 | peer = inet_getpeer_v6(base, &rt->rt6i_dst.addr, create); | |
117 | if (peer) { | |
118 | if (!rt6_set_peer(rt, peer)) | |
119 | inet_putpeer(peer); | |
120 | } | |
121 | } | |
122 | ||
123 | static struct inet_peer *__rt6_get_peer(struct rt6_info *rt, int create) | |
124 | { | |
125 | if (rt6_has_peer(rt)) | |
126 | return rt6_peer_ptr(rt); | |
127 | ||
128 | rt6_bind_peer(rt, create); | |
129 | return (rt6_has_peer(rt) ? rt6_peer_ptr(rt) : NULL); | |
130 | } | |
131 | ||
132 | static struct inet_peer *rt6_get_peer_create(struct rt6_info *rt) | |
133 | { | |
134 | return __rt6_get_peer(rt, 1); | |
135 | } | |
136 | ||
06582540 DM |
137 | static u32 *ipv6_cow_metrics(struct dst_entry *dst, unsigned long old) |
138 | { | |
139 | struct rt6_info *rt = (struct rt6_info *) dst; | |
140 | struct inet_peer *peer; | |
141 | u32 *p = NULL; | |
142 | ||
8e2ec639 | 143 | if (!(rt->dst.flags & DST_HOST)) |
3b471175 | 144 | return dst_cow_metrics_generic(dst, old); |
8e2ec639 | 145 | |
fbfe95a4 | 146 | peer = rt6_get_peer_create(rt); |
06582540 DM |
147 | if (peer) { |
148 | u32 *old_p = __DST_METRICS_PTR(old); | |
149 | unsigned long prev, new; | |
150 | ||
151 | p = peer->metrics; | |
e5fd387a MK |
152 | if (inet_metrics_new(peer) || |
153 | (old & DST_METRICS_FORCE_OVERWRITE)) | |
06582540 DM |
154 | memcpy(p, old_p, sizeof(u32) * RTAX_MAX); |
155 | ||
156 | new = (unsigned long) p; | |
157 | prev = cmpxchg(&dst->_metrics, old, new); | |
158 | ||
159 | if (prev != old) { | |
160 | p = __DST_METRICS_PTR(prev); | |
161 | if (prev & DST_METRICS_READ_ONLY) | |
162 | p = NULL; | |
163 | } | |
164 | } | |
165 | return p; | |
166 | } | |
167 | ||
f894cbf8 DM |
168 | static inline const void *choose_neigh_daddr(struct rt6_info *rt, |
169 | struct sk_buff *skb, | |
170 | const void *daddr) | |
39232973 DM |
171 | { |
172 | struct in6_addr *p = &rt->rt6i_gateway; | |
173 | ||
a7563f34 | 174 | if (!ipv6_addr_any(p)) |
39232973 | 175 | return (const void *) p; |
f894cbf8 DM |
176 | else if (skb) |
177 | return &ipv6_hdr(skb)->daddr; | |
39232973 DM |
178 | return daddr; |
179 | } | |
180 | ||
f894cbf8 DM |
181 | static struct neighbour *ip6_neigh_lookup(const struct dst_entry *dst, |
182 | struct sk_buff *skb, | |
183 | const void *daddr) | |
d3aaeb38 | 184 | { |
39232973 DM |
185 | struct rt6_info *rt = (struct rt6_info *) dst; |
186 | struct neighbour *n; | |
187 | ||
f894cbf8 | 188 | daddr = choose_neigh_daddr(rt, skb, daddr); |
8e022ee6 | 189 | n = __ipv6_neigh_lookup(dst->dev, daddr); |
f83c7790 DM |
190 | if (n) |
191 | return n; | |
192 | return neigh_create(&nd_tbl, daddr, dst->dev); | |
193 | } | |
194 | ||
9a7ec3a9 | 195 | static struct dst_ops ip6_dst_ops_template = { |
1da177e4 | 196 | .family = AF_INET6, |
1da177e4 LT |
197 | .gc = ip6_dst_gc, |
198 | .gc_thresh = 1024, | |
199 | .check = ip6_dst_check, | |
0dbaee3b | 200 | .default_advmss = ip6_default_advmss, |
ebb762f2 | 201 | .mtu = ip6_mtu, |
06582540 | 202 | .cow_metrics = ipv6_cow_metrics, |
1da177e4 LT |
203 | .destroy = ip6_dst_destroy, |
204 | .ifdown = ip6_dst_ifdown, | |
205 | .negative_advice = ip6_negative_advice, | |
206 | .link_failure = ip6_link_failure, | |
207 | .update_pmtu = ip6_rt_update_pmtu, | |
6e157b6a | 208 | .redirect = rt6_do_redirect, |
1ac06e03 | 209 | .local_out = __ip6_local_out, |
d3aaeb38 | 210 | .neigh_lookup = ip6_neigh_lookup, |
1da177e4 LT |
211 | }; |
212 | ||
ebb762f2 | 213 | static unsigned int ip6_blackhole_mtu(const struct dst_entry *dst) |
ec831ea7 | 214 | { |
618f9bc7 SK |
215 | unsigned int mtu = dst_metric_raw(dst, RTAX_MTU); |
216 | ||
217 | return mtu ? : dst->dev->mtu; | |
ec831ea7 RD |
218 | } |
219 | ||
6700c270 DM |
220 | static void ip6_rt_blackhole_update_pmtu(struct dst_entry *dst, struct sock *sk, |
221 | struct sk_buff *skb, u32 mtu) | |
14e50e57 DM |
222 | { |
223 | } | |
224 | ||
6700c270 DM |
225 | static void ip6_rt_blackhole_redirect(struct dst_entry *dst, struct sock *sk, |
226 | struct sk_buff *skb) | |
b587ee3b DM |
227 | { |
228 | } | |
229 | ||
0972ddb2 HB |
230 | static u32 *ip6_rt_blackhole_cow_metrics(struct dst_entry *dst, |
231 | unsigned long old) | |
232 | { | |
233 | return NULL; | |
234 | } | |
235 | ||
14e50e57 DM |
236 | static struct dst_ops ip6_dst_blackhole_ops = { |
237 | .family = AF_INET6, | |
14e50e57 DM |
238 | .destroy = ip6_dst_destroy, |
239 | .check = ip6_dst_check, | |
ebb762f2 | 240 | .mtu = ip6_blackhole_mtu, |
214f45c9 | 241 | .default_advmss = ip6_default_advmss, |
14e50e57 | 242 | .update_pmtu = ip6_rt_blackhole_update_pmtu, |
b587ee3b | 243 | .redirect = ip6_rt_blackhole_redirect, |
0972ddb2 | 244 | .cow_metrics = ip6_rt_blackhole_cow_metrics, |
d3aaeb38 | 245 | .neigh_lookup = ip6_neigh_lookup, |
14e50e57 DM |
246 | }; |
247 | ||
62fa8a84 | 248 | static const u32 ip6_template_metrics[RTAX_MAX] = { |
14edd87d | 249 | [RTAX_HOPLIMIT - 1] = 0, |
62fa8a84 DM |
250 | }; |
251 | ||
fb0af4c7 | 252 | static const struct rt6_info ip6_null_entry_template = { |
d8d1f30b CG |
253 | .dst = { |
254 | .__refcnt = ATOMIC_INIT(1), | |
255 | .__use = 1, | |
2c20cbd7 | 256 | .obsolete = DST_OBSOLETE_FORCE_CHK, |
d8d1f30b | 257 | .error = -ENETUNREACH, |
d8d1f30b CG |
258 | .input = ip6_pkt_discard, |
259 | .output = ip6_pkt_discard_out, | |
1da177e4 LT |
260 | }, |
261 | .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP), | |
4f724279 | 262 | .rt6i_protocol = RTPROT_KERNEL, |
1da177e4 LT |
263 | .rt6i_metric = ~(u32) 0, |
264 | .rt6i_ref = ATOMIC_INIT(1), | |
265 | }; | |
266 | ||
101367c2 TG |
267 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES |
268 | ||
fb0af4c7 | 269 | static const struct rt6_info ip6_prohibit_entry_template = { |
d8d1f30b CG |
270 | .dst = { |
271 | .__refcnt = ATOMIC_INIT(1), | |
272 | .__use = 1, | |
2c20cbd7 | 273 | .obsolete = DST_OBSOLETE_FORCE_CHK, |
d8d1f30b | 274 | .error = -EACCES, |
d8d1f30b CG |
275 | .input = ip6_pkt_prohibit, |
276 | .output = ip6_pkt_prohibit_out, | |
101367c2 TG |
277 | }, |
278 | .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP), | |
4f724279 | 279 | .rt6i_protocol = RTPROT_KERNEL, |
101367c2 TG |
280 | .rt6i_metric = ~(u32) 0, |
281 | .rt6i_ref = ATOMIC_INIT(1), | |
282 | }; | |
283 | ||
fb0af4c7 | 284 | static const struct rt6_info ip6_blk_hole_entry_template = { |
d8d1f30b CG |
285 | .dst = { |
286 | .__refcnt = ATOMIC_INIT(1), | |
287 | .__use = 1, | |
2c20cbd7 | 288 | .obsolete = DST_OBSOLETE_FORCE_CHK, |
d8d1f30b | 289 | .error = -EINVAL, |
d8d1f30b | 290 | .input = dst_discard, |
aad88724 | 291 | .output = dst_discard_sk, |
101367c2 TG |
292 | }, |
293 | .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP), | |
4f724279 | 294 | .rt6i_protocol = RTPROT_KERNEL, |
101367c2 TG |
295 | .rt6i_metric = ~(u32) 0, |
296 | .rt6i_ref = ATOMIC_INIT(1), | |
297 | }; | |
298 | ||
299 | #endif | |
300 | ||
1da177e4 | 301 | /* allocate dst with ip6_dst_ops */ |
97bab73f | 302 | static inline struct rt6_info *ip6_dst_alloc(struct net *net, |
957c665f | 303 | struct net_device *dev, |
8b96d22d DM |
304 | int flags, |
305 | struct fib6_table *table) | |
1da177e4 | 306 | { |
97bab73f | 307 | struct rt6_info *rt = dst_alloc(&net->ipv6.ip6_dst_ops, dev, |
6f3118b5 | 308 | 0, DST_OBSOLETE_FORCE_CHK, flags); |
cf911662 | 309 | |
97bab73f | 310 | if (rt) { |
8104891b SK |
311 | struct dst_entry *dst = &rt->dst; |
312 | ||
313 | memset(dst + 1, 0, sizeof(*rt) - sizeof(*dst)); | |
8b96d22d | 314 | rt6_init_peer(rt, table ? &table->tb6_peers : net->ipv6.peers); |
51ebd318 | 315 | INIT_LIST_HEAD(&rt->rt6i_siblings); |
97bab73f | 316 | } |
cf911662 | 317 | return rt; |
1da177e4 LT |
318 | } |
319 | ||
320 | static void ip6_dst_destroy(struct dst_entry *dst) | |
321 | { | |
322 | struct rt6_info *rt = (struct rt6_info *)dst; | |
323 | struct inet6_dev *idev = rt->rt6i_idev; | |
ecd98837 | 324 | struct dst_entry *from = dst->from; |
1da177e4 | 325 | |
8e2ec639 YZ |
326 | if (!(rt->dst.flags & DST_HOST)) |
327 | dst_destroy_metrics_generic(dst); | |
328 | ||
38308473 | 329 | if (idev) { |
1da177e4 LT |
330 | rt->rt6i_idev = NULL; |
331 | in6_dev_put(idev); | |
1ab1457c | 332 | } |
1716a961 | 333 | |
ecd98837 YH |
334 | dst->from = NULL; |
335 | dst_release(from); | |
1716a961 | 336 | |
97bab73f DM |
337 | if (rt6_has_peer(rt)) { |
338 | struct inet_peer *peer = rt6_peer_ptr(rt); | |
b3419363 DM |
339 | inet_putpeer(peer); |
340 | } | |
341 | } | |
342 | ||
1da177e4 LT |
343 | static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev, |
344 | int how) | |
345 | { | |
346 | struct rt6_info *rt = (struct rt6_info *)dst; | |
347 | struct inet6_dev *idev = rt->rt6i_idev; | |
5a3e55d6 | 348 | struct net_device *loopback_dev = |
c346dca1 | 349 | dev_net(dev)->loopback_dev; |
1da177e4 | 350 | |
97cac082 DM |
351 | if (dev != loopback_dev) { |
352 | if (idev && idev->dev == dev) { | |
353 | struct inet6_dev *loopback_idev = | |
354 | in6_dev_get(loopback_dev); | |
355 | if (loopback_idev) { | |
356 | rt->rt6i_idev = loopback_idev; | |
357 | in6_dev_put(idev); | |
358 | } | |
359 | } | |
1da177e4 LT |
360 | } |
361 | } | |
362 | ||
a50feda5 | 363 | static bool rt6_check_expired(const struct rt6_info *rt) |
1da177e4 | 364 | { |
1716a961 G |
365 | if (rt->rt6i_flags & RTF_EXPIRES) { |
366 | if (time_after(jiffies, rt->dst.expires)) | |
a50feda5 | 367 | return true; |
1716a961 | 368 | } else if (rt->dst.from) { |
3fd91fb3 | 369 | return rt6_check_expired((struct rt6_info *) rt->dst.from); |
1716a961 | 370 | } |
a50feda5 | 371 | return false; |
1da177e4 LT |
372 | } |
373 | ||
51ebd318 ND |
374 | /* Multipath route selection: |
375 | * Hash based function using packet header and flowlabel. | |
376 | * Adapted from fib_info_hashfn() | |
377 | */ | |
378 | static int rt6_info_hash_nhsfn(unsigned int candidate_count, | |
379 | const struct flowi6 *fl6) | |
380 | { | |
381 | unsigned int val = fl6->flowi6_proto; | |
382 | ||
c08977bb YH |
383 | val ^= ipv6_addr_hash(&fl6->daddr); |
384 | val ^= ipv6_addr_hash(&fl6->saddr); | |
51ebd318 ND |
385 | |
386 | /* Work only if this not encapsulated */ | |
387 | switch (fl6->flowi6_proto) { | |
388 | case IPPROTO_UDP: | |
389 | case IPPROTO_TCP: | |
390 | case IPPROTO_SCTP: | |
b3ce5ae1 ND |
391 | val ^= (__force u16)fl6->fl6_sport; |
392 | val ^= (__force u16)fl6->fl6_dport; | |
51ebd318 ND |
393 | break; |
394 | ||
395 | case IPPROTO_ICMPV6: | |
b3ce5ae1 ND |
396 | val ^= (__force u16)fl6->fl6_icmp_type; |
397 | val ^= (__force u16)fl6->fl6_icmp_code; | |
51ebd318 ND |
398 | break; |
399 | } | |
400 | /* RFC6438 recommands to use flowlabel */ | |
b3ce5ae1 | 401 | val ^= (__force u32)fl6->flowlabel; |
51ebd318 ND |
402 | |
403 | /* Perhaps, we need to tune, this function? */ | |
404 | val = val ^ (val >> 7) ^ (val >> 12); | |
405 | return val % candidate_count; | |
406 | } | |
407 | ||
408 | static struct rt6_info *rt6_multipath_select(struct rt6_info *match, | |
52bd4c0c ND |
409 | struct flowi6 *fl6, int oif, |
410 | int strict) | |
51ebd318 ND |
411 | { |
412 | struct rt6_info *sibling, *next_sibling; | |
413 | int route_choosen; | |
414 | ||
415 | route_choosen = rt6_info_hash_nhsfn(match->rt6i_nsiblings + 1, fl6); | |
416 | /* Don't change the route, if route_choosen == 0 | |
417 | * (siblings does not include ourself) | |
418 | */ | |
419 | if (route_choosen) | |
420 | list_for_each_entry_safe(sibling, next_sibling, | |
421 | &match->rt6i_siblings, rt6i_siblings) { | |
422 | route_choosen--; | |
423 | if (route_choosen == 0) { | |
52bd4c0c ND |
424 | if (rt6_score_route(sibling, oif, strict) < 0) |
425 | break; | |
51ebd318 ND |
426 | match = sibling; |
427 | break; | |
428 | } | |
429 | } | |
430 | return match; | |
431 | } | |
432 | ||
1da177e4 | 433 | /* |
c71099ac | 434 | * Route lookup. Any table->tb6_lock is implied. |
1da177e4 LT |
435 | */ |
436 | ||
8ed67789 DL |
437 | static inline struct rt6_info *rt6_device_match(struct net *net, |
438 | struct rt6_info *rt, | |
b71d1d42 | 439 | const struct in6_addr *saddr, |
1da177e4 | 440 | int oif, |
d420895e | 441 | int flags) |
1da177e4 LT |
442 | { |
443 | struct rt6_info *local = NULL; | |
444 | struct rt6_info *sprt; | |
445 | ||
dd3abc4e YH |
446 | if (!oif && ipv6_addr_any(saddr)) |
447 | goto out; | |
448 | ||
d8d1f30b | 449 | for (sprt = rt; sprt; sprt = sprt->dst.rt6_next) { |
d1918542 | 450 | struct net_device *dev = sprt->dst.dev; |
dd3abc4e YH |
451 | |
452 | if (oif) { | |
1da177e4 LT |
453 | if (dev->ifindex == oif) |
454 | return sprt; | |
455 | if (dev->flags & IFF_LOOPBACK) { | |
38308473 | 456 | if (!sprt->rt6i_idev || |
1da177e4 | 457 | sprt->rt6i_idev->dev->ifindex != oif) { |
d420895e | 458 | if (flags & RT6_LOOKUP_F_IFACE && oif) |
1da177e4 | 459 | continue; |
1ab1457c | 460 | if (local && (!oif || |
1da177e4 LT |
461 | local->rt6i_idev->dev->ifindex == oif)) |
462 | continue; | |
463 | } | |
464 | local = sprt; | |
465 | } | |
dd3abc4e YH |
466 | } else { |
467 | if (ipv6_chk_addr(net, saddr, dev, | |
468 | flags & RT6_LOOKUP_F_IFACE)) | |
469 | return sprt; | |
1da177e4 | 470 | } |
dd3abc4e | 471 | } |
1da177e4 | 472 | |
dd3abc4e | 473 | if (oif) { |
1da177e4 LT |
474 | if (local) |
475 | return local; | |
476 | ||
d420895e | 477 | if (flags & RT6_LOOKUP_F_IFACE) |
8ed67789 | 478 | return net->ipv6.ip6_null_entry; |
1da177e4 | 479 | } |
dd3abc4e | 480 | out: |
1da177e4 LT |
481 | return rt; |
482 | } | |
483 | ||
27097255 | 484 | #ifdef CONFIG_IPV6_ROUTER_PREF |
c2f17e82 HFS |
485 | struct __rt6_probe_work { |
486 | struct work_struct work; | |
487 | struct in6_addr target; | |
488 | struct net_device *dev; | |
489 | }; | |
490 | ||
491 | static void rt6_probe_deferred(struct work_struct *w) | |
492 | { | |
493 | struct in6_addr mcaddr; | |
494 | struct __rt6_probe_work *work = | |
495 | container_of(w, struct __rt6_probe_work, work); | |
496 | ||
497 | addrconf_addr_solict_mult(&work->target, &mcaddr); | |
498 | ndisc_send_ns(work->dev, NULL, &work->target, &mcaddr, NULL); | |
499 | dev_put(work->dev); | |
662f5533 | 500 | kfree(work); |
c2f17e82 HFS |
501 | } |
502 | ||
27097255 YH |
503 | static void rt6_probe(struct rt6_info *rt) |
504 | { | |
f2c31e32 | 505 | struct neighbour *neigh; |
27097255 YH |
506 | /* |
507 | * Okay, this does not seem to be appropriate | |
508 | * for now, however, we need to check if it | |
509 | * is really so; aka Router Reachability Probing. | |
510 | * | |
511 | * Router Reachability Probe MUST be rate-limited | |
512 | * to no more than one per minute. | |
513 | */ | |
2152caea | 514 | if (!rt || !(rt->rt6i_flags & RTF_GATEWAY)) |
7ff74a59 | 515 | return; |
2152caea YH |
516 | rcu_read_lock_bh(); |
517 | neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway); | |
518 | if (neigh) { | |
519 | write_lock(&neigh->lock); | |
520 | if (neigh->nud_state & NUD_VALID) | |
521 | goto out; | |
7ff74a59 | 522 | } |
2152caea YH |
523 | |
524 | if (!neigh || | |
52e16356 | 525 | time_after(jiffies, neigh->updated + rt->rt6i_idev->cnf.rtr_probe_interval)) { |
c2f17e82 | 526 | struct __rt6_probe_work *work; |
27097255 | 527 | |
c2f17e82 HFS |
528 | work = kmalloc(sizeof(*work), GFP_ATOMIC); |
529 | ||
530 | if (neigh && work) | |
7e980569 | 531 | __neigh_set_probe_once(neigh); |
c2f17e82 HFS |
532 | |
533 | if (neigh) | |
2152caea YH |
534 | write_unlock(&neigh->lock); |
535 | ||
c2f17e82 HFS |
536 | if (work) { |
537 | INIT_WORK(&work->work, rt6_probe_deferred); | |
538 | work->target = rt->rt6i_gateway; | |
539 | dev_hold(rt->dst.dev); | |
540 | work->dev = rt->dst.dev; | |
541 | schedule_work(&work->work); | |
542 | } | |
f2c31e32 | 543 | } else { |
2152caea YH |
544 | out: |
545 | write_unlock(&neigh->lock); | |
f2c31e32 | 546 | } |
2152caea | 547 | rcu_read_unlock_bh(); |
27097255 YH |
548 | } |
549 | #else | |
550 | static inline void rt6_probe(struct rt6_info *rt) | |
551 | { | |
27097255 YH |
552 | } |
553 | #endif | |
554 | ||
1da177e4 | 555 | /* |
554cfb7e | 556 | * Default Router Selection (RFC 2461 6.3.6) |
1da177e4 | 557 | */ |
b6f99a21 | 558 | static inline int rt6_check_dev(struct rt6_info *rt, int oif) |
554cfb7e | 559 | { |
d1918542 | 560 | struct net_device *dev = rt->dst.dev; |
161980f4 | 561 | if (!oif || dev->ifindex == oif) |
554cfb7e | 562 | return 2; |
161980f4 DM |
563 | if ((dev->flags & IFF_LOOPBACK) && |
564 | rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif) | |
565 | return 1; | |
566 | return 0; | |
554cfb7e | 567 | } |
1da177e4 | 568 | |
afc154e9 | 569 | static inline enum rt6_nud_state rt6_check_neigh(struct rt6_info *rt) |
1da177e4 | 570 | { |
f2c31e32 | 571 | struct neighbour *neigh; |
afc154e9 | 572 | enum rt6_nud_state ret = RT6_NUD_FAIL_HARD; |
f2c31e32 | 573 | |
4d0c5911 YH |
574 | if (rt->rt6i_flags & RTF_NONEXTHOP || |
575 | !(rt->rt6i_flags & RTF_GATEWAY)) | |
afc154e9 | 576 | return RT6_NUD_SUCCEED; |
145a3621 YH |
577 | |
578 | rcu_read_lock_bh(); | |
579 | neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway); | |
580 | if (neigh) { | |
581 | read_lock(&neigh->lock); | |
554cfb7e | 582 | if (neigh->nud_state & NUD_VALID) |
afc154e9 | 583 | ret = RT6_NUD_SUCCEED; |
398bcbeb | 584 | #ifdef CONFIG_IPV6_ROUTER_PREF |
a5a81f0b | 585 | else if (!(neigh->nud_state & NUD_FAILED)) |
afc154e9 | 586 | ret = RT6_NUD_SUCCEED; |
7e980569 JB |
587 | else |
588 | ret = RT6_NUD_FAIL_PROBE; | |
398bcbeb | 589 | #endif |
145a3621 | 590 | read_unlock(&neigh->lock); |
afc154e9 HFS |
591 | } else { |
592 | ret = IS_ENABLED(CONFIG_IPV6_ROUTER_PREF) ? | |
7e980569 | 593 | RT6_NUD_SUCCEED : RT6_NUD_FAIL_DO_RR; |
a5a81f0b | 594 | } |
145a3621 YH |
595 | rcu_read_unlock_bh(); |
596 | ||
a5a81f0b | 597 | return ret; |
1da177e4 LT |
598 | } |
599 | ||
554cfb7e YH |
600 | static int rt6_score_route(struct rt6_info *rt, int oif, |
601 | int strict) | |
1da177e4 | 602 | { |
a5a81f0b | 603 | int m; |
1ab1457c | 604 | |
4d0c5911 | 605 | m = rt6_check_dev(rt, oif); |
77d16f45 | 606 | if (!m && (strict & RT6_LOOKUP_F_IFACE)) |
afc154e9 | 607 | return RT6_NUD_FAIL_HARD; |
ebacaaa0 YH |
608 | #ifdef CONFIG_IPV6_ROUTER_PREF |
609 | m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2; | |
610 | #endif | |
afc154e9 HFS |
611 | if (strict & RT6_LOOKUP_F_REACHABLE) { |
612 | int n = rt6_check_neigh(rt); | |
613 | if (n < 0) | |
614 | return n; | |
615 | } | |
554cfb7e YH |
616 | return m; |
617 | } | |
618 | ||
f11e6659 | 619 | static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict, |
afc154e9 HFS |
620 | int *mpri, struct rt6_info *match, |
621 | bool *do_rr) | |
554cfb7e | 622 | { |
f11e6659 | 623 | int m; |
afc154e9 | 624 | bool match_do_rr = false; |
f11e6659 DM |
625 | |
626 | if (rt6_check_expired(rt)) | |
627 | goto out; | |
628 | ||
629 | m = rt6_score_route(rt, oif, strict); | |
7e980569 | 630 | if (m == RT6_NUD_FAIL_DO_RR) { |
afc154e9 HFS |
631 | match_do_rr = true; |
632 | m = 0; /* lowest valid score */ | |
7e980569 | 633 | } else if (m == RT6_NUD_FAIL_HARD) { |
f11e6659 | 634 | goto out; |
afc154e9 HFS |
635 | } |
636 | ||
637 | if (strict & RT6_LOOKUP_F_REACHABLE) | |
638 | rt6_probe(rt); | |
f11e6659 | 639 | |
7e980569 | 640 | /* note that m can be RT6_NUD_FAIL_PROBE at this point */ |
f11e6659 | 641 | if (m > *mpri) { |
afc154e9 | 642 | *do_rr = match_do_rr; |
f11e6659 DM |
643 | *mpri = m; |
644 | match = rt; | |
f11e6659 | 645 | } |
f11e6659 DM |
646 | out: |
647 | return match; | |
648 | } | |
649 | ||
650 | static struct rt6_info *find_rr_leaf(struct fib6_node *fn, | |
651 | struct rt6_info *rr_head, | |
afc154e9 HFS |
652 | u32 metric, int oif, int strict, |
653 | bool *do_rr) | |
f11e6659 DM |
654 | { |
655 | struct rt6_info *rt, *match; | |
554cfb7e | 656 | int mpri = -1; |
1da177e4 | 657 | |
f11e6659 DM |
658 | match = NULL; |
659 | for (rt = rr_head; rt && rt->rt6i_metric == metric; | |
d8d1f30b | 660 | rt = rt->dst.rt6_next) |
afc154e9 | 661 | match = find_match(rt, oif, strict, &mpri, match, do_rr); |
f11e6659 | 662 | for (rt = fn->leaf; rt && rt != rr_head && rt->rt6i_metric == metric; |
d8d1f30b | 663 | rt = rt->dst.rt6_next) |
afc154e9 | 664 | match = find_match(rt, oif, strict, &mpri, match, do_rr); |
1da177e4 | 665 | |
f11e6659 DM |
666 | return match; |
667 | } | |
1da177e4 | 668 | |
f11e6659 DM |
669 | static struct rt6_info *rt6_select(struct fib6_node *fn, int oif, int strict) |
670 | { | |
671 | struct rt6_info *match, *rt0; | |
8ed67789 | 672 | struct net *net; |
afc154e9 | 673 | bool do_rr = false; |
1da177e4 | 674 | |
f11e6659 DM |
675 | rt0 = fn->rr_ptr; |
676 | if (!rt0) | |
677 | fn->rr_ptr = rt0 = fn->leaf; | |
1da177e4 | 678 | |
afc154e9 HFS |
679 | match = find_rr_leaf(fn, rt0, rt0->rt6i_metric, oif, strict, |
680 | &do_rr); | |
1da177e4 | 681 | |
afc154e9 | 682 | if (do_rr) { |
d8d1f30b | 683 | struct rt6_info *next = rt0->dst.rt6_next; |
f11e6659 | 684 | |
554cfb7e | 685 | /* no entries matched; do round-robin */ |
f11e6659 DM |
686 | if (!next || next->rt6i_metric != rt0->rt6i_metric) |
687 | next = fn->leaf; | |
688 | ||
689 | if (next != rt0) | |
690 | fn->rr_ptr = next; | |
1da177e4 | 691 | } |
1da177e4 | 692 | |
d1918542 | 693 | net = dev_net(rt0->dst.dev); |
a02cec21 | 694 | return match ? match : net->ipv6.ip6_null_entry; |
1da177e4 LT |
695 | } |
696 | ||
70ceb4f5 YH |
697 | #ifdef CONFIG_IPV6_ROUTE_INFO |
698 | int rt6_route_rcv(struct net_device *dev, u8 *opt, int len, | |
b71d1d42 | 699 | const struct in6_addr *gwaddr) |
70ceb4f5 | 700 | { |
c346dca1 | 701 | struct net *net = dev_net(dev); |
70ceb4f5 YH |
702 | struct route_info *rinfo = (struct route_info *) opt; |
703 | struct in6_addr prefix_buf, *prefix; | |
704 | unsigned int pref; | |
4bed72e4 | 705 | unsigned long lifetime; |
70ceb4f5 YH |
706 | struct rt6_info *rt; |
707 | ||
708 | if (len < sizeof(struct route_info)) { | |
709 | return -EINVAL; | |
710 | } | |
711 | ||
712 | /* Sanity check for prefix_len and length */ | |
713 | if (rinfo->length > 3) { | |
714 | return -EINVAL; | |
715 | } else if (rinfo->prefix_len > 128) { | |
716 | return -EINVAL; | |
717 | } else if (rinfo->prefix_len > 64) { | |
718 | if (rinfo->length < 2) { | |
719 | return -EINVAL; | |
720 | } | |
721 | } else if (rinfo->prefix_len > 0) { | |
722 | if (rinfo->length < 1) { | |
723 | return -EINVAL; | |
724 | } | |
725 | } | |
726 | ||
727 | pref = rinfo->route_pref; | |
728 | if (pref == ICMPV6_ROUTER_PREF_INVALID) | |
3933fc95 | 729 | return -EINVAL; |
70ceb4f5 | 730 | |
4bed72e4 | 731 | lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ); |
70ceb4f5 YH |
732 | |
733 | if (rinfo->length == 3) | |
734 | prefix = (struct in6_addr *)rinfo->prefix; | |
735 | else { | |
736 | /* this function is safe */ | |
737 | ipv6_addr_prefix(&prefix_buf, | |
738 | (struct in6_addr *)rinfo->prefix, | |
739 | rinfo->prefix_len); | |
740 | prefix = &prefix_buf; | |
741 | } | |
742 | ||
f104a567 DJ |
743 | if (rinfo->prefix_len == 0) |
744 | rt = rt6_get_dflt_router(gwaddr, dev); | |
745 | else | |
746 | rt = rt6_get_route_info(net, prefix, rinfo->prefix_len, | |
747 | gwaddr, dev->ifindex); | |
70ceb4f5 YH |
748 | |
749 | if (rt && !lifetime) { | |
e0a1ad73 | 750 | ip6_del_rt(rt); |
70ceb4f5 YH |
751 | rt = NULL; |
752 | } | |
753 | ||
754 | if (!rt && lifetime) | |
efa2cea0 | 755 | rt = rt6_add_route_info(net, prefix, rinfo->prefix_len, gwaddr, dev->ifindex, |
70ceb4f5 YH |
756 | pref); |
757 | else if (rt) | |
758 | rt->rt6i_flags = RTF_ROUTEINFO | | |
759 | (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref); | |
760 | ||
761 | if (rt) { | |
1716a961 G |
762 | if (!addrconf_finite_timeout(lifetime)) |
763 | rt6_clean_expires(rt); | |
764 | else | |
765 | rt6_set_expires(rt, jiffies + HZ * lifetime); | |
766 | ||
94e187c0 | 767 | ip6_rt_put(rt); |
70ceb4f5 YH |
768 | } |
769 | return 0; | |
770 | } | |
771 | #endif | |
772 | ||
a3c00e46 MKL |
773 | static struct fib6_node* fib6_backtrack(struct fib6_node *fn, |
774 | struct in6_addr *saddr) | |
775 | { | |
776 | struct fib6_node *pn; | |
777 | while (1) { | |
778 | if (fn->fn_flags & RTN_TL_ROOT) | |
779 | return NULL; | |
780 | pn = fn->parent; | |
781 | if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn) | |
782 | fn = fib6_lookup(FIB6_SUBTREE(pn), NULL, saddr); | |
783 | else | |
784 | fn = pn; | |
785 | if (fn->fn_flags & RTN_RTINFO) | |
786 | return fn; | |
787 | } | |
788 | } | |
c71099ac | 789 | |
8ed67789 DL |
790 | static struct rt6_info *ip6_pol_route_lookup(struct net *net, |
791 | struct fib6_table *table, | |
4c9483b2 | 792 | struct flowi6 *fl6, int flags) |
1da177e4 LT |
793 | { |
794 | struct fib6_node *fn; | |
795 | struct rt6_info *rt; | |
796 | ||
c71099ac | 797 | read_lock_bh(&table->tb6_lock); |
4c9483b2 | 798 | fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr); |
c71099ac TG |
799 | restart: |
800 | rt = fn->leaf; | |
4c9483b2 | 801 | rt = rt6_device_match(net, rt, &fl6->saddr, fl6->flowi6_oif, flags); |
51ebd318 | 802 | if (rt->rt6i_nsiblings && fl6->flowi6_oif == 0) |
52bd4c0c | 803 | rt = rt6_multipath_select(rt, fl6, fl6->flowi6_oif, flags); |
a3c00e46 MKL |
804 | if (rt == net->ipv6.ip6_null_entry) { |
805 | fn = fib6_backtrack(fn, &fl6->saddr); | |
806 | if (fn) | |
807 | goto restart; | |
808 | } | |
d8d1f30b | 809 | dst_use(&rt->dst, jiffies); |
c71099ac | 810 | read_unlock_bh(&table->tb6_lock); |
c71099ac TG |
811 | return rt; |
812 | ||
813 | } | |
814 | ||
67ba4152 | 815 | struct dst_entry *ip6_route_lookup(struct net *net, struct flowi6 *fl6, |
ea6e574e FW |
816 | int flags) |
817 | { | |
818 | return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_lookup); | |
819 | } | |
820 | EXPORT_SYMBOL_GPL(ip6_route_lookup); | |
821 | ||
9acd9f3a YH |
822 | struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr, |
823 | const struct in6_addr *saddr, int oif, int strict) | |
c71099ac | 824 | { |
4c9483b2 DM |
825 | struct flowi6 fl6 = { |
826 | .flowi6_oif = oif, | |
827 | .daddr = *daddr, | |
c71099ac TG |
828 | }; |
829 | struct dst_entry *dst; | |
77d16f45 | 830 | int flags = strict ? RT6_LOOKUP_F_IFACE : 0; |
c71099ac | 831 | |
adaa70bb | 832 | if (saddr) { |
4c9483b2 | 833 | memcpy(&fl6.saddr, saddr, sizeof(*saddr)); |
adaa70bb TG |
834 | flags |= RT6_LOOKUP_F_HAS_SADDR; |
835 | } | |
836 | ||
4c9483b2 | 837 | dst = fib6_rule_lookup(net, &fl6, flags, ip6_pol_route_lookup); |
c71099ac TG |
838 | if (dst->error == 0) |
839 | return (struct rt6_info *) dst; | |
840 | ||
841 | dst_release(dst); | |
842 | ||
1da177e4 LT |
843 | return NULL; |
844 | } | |
7159039a YH |
845 | EXPORT_SYMBOL(rt6_lookup); |
846 | ||
c71099ac | 847 | /* ip6_ins_rt is called with FREE table->tb6_lock. |
1da177e4 LT |
848 | It takes new route entry, the addition fails by any reason the |
849 | route is freed. In any case, if caller does not hold it, it may | |
850 | be destroyed. | |
851 | */ | |
852 | ||
e5fd387a | 853 | static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info, |
e715b6d3 | 854 | struct mx6_config *mxc) |
1da177e4 LT |
855 | { |
856 | int err; | |
c71099ac | 857 | struct fib6_table *table; |
1da177e4 | 858 | |
c71099ac TG |
859 | table = rt->rt6i_table; |
860 | write_lock_bh(&table->tb6_lock); | |
e715b6d3 | 861 | err = fib6_add(&table->tb6_root, rt, info, mxc); |
c71099ac | 862 | write_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
863 | |
864 | return err; | |
865 | } | |
866 | ||
40e22e8f TG |
867 | int ip6_ins_rt(struct rt6_info *rt) |
868 | { | |
e715b6d3 FW |
869 | struct nl_info info = { .nl_net = dev_net(rt->dst.dev), }; |
870 | struct mx6_config mxc = { .mx = NULL, }; | |
871 | ||
872 | return __ip6_ins_rt(rt, &info, &mxc); | |
40e22e8f TG |
873 | } |
874 | ||
1716a961 | 875 | static struct rt6_info *rt6_alloc_cow(struct rt6_info *ort, |
21efcfa0 | 876 | const struct in6_addr *daddr, |
b71d1d42 | 877 | const struct in6_addr *saddr) |
1da177e4 | 878 | { |
1da177e4 LT |
879 | struct rt6_info *rt; |
880 | ||
881 | /* | |
882 | * Clone the route. | |
883 | */ | |
884 | ||
21efcfa0 | 885 | rt = ip6_rt_copy(ort, daddr); |
1da177e4 LT |
886 | |
887 | if (rt) { | |
249a3630 DJ |
888 | if (ort->rt6i_dst.plen != 128 && |
889 | ipv6_addr_equal(&ort->rt6i_dst.addr, daddr)) | |
890 | rt->rt6i_flags |= RTF_ANYCAST; | |
1da177e4 | 891 | |
1da177e4 | 892 | rt->rt6i_flags |= RTF_CACHE; |
1da177e4 LT |
893 | |
894 | #ifdef CONFIG_IPV6_SUBTREES | |
895 | if (rt->rt6i_src.plen && saddr) { | |
4e3fd7a0 | 896 | rt->rt6i_src.addr = *saddr; |
1da177e4 LT |
897 | rt->rt6i_src.plen = 128; |
898 | } | |
899 | #endif | |
95a9a5ba | 900 | } |
1da177e4 | 901 | |
95a9a5ba YH |
902 | return rt; |
903 | } | |
1da177e4 | 904 | |
21efcfa0 ED |
905 | static struct rt6_info *rt6_alloc_clone(struct rt6_info *ort, |
906 | const struct in6_addr *daddr) | |
299d9939 | 907 | { |
21efcfa0 ED |
908 | struct rt6_info *rt = ip6_rt_copy(ort, daddr); |
909 | ||
887c95cc | 910 | if (rt) |
299d9939 | 911 | rt->rt6i_flags |= RTF_CACHE; |
299d9939 YH |
912 | return rt; |
913 | } | |
914 | ||
8ed67789 | 915 | static struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table, int oif, |
4c9483b2 | 916 | struct flowi6 *fl6, int flags) |
1da177e4 | 917 | { |
367efcb9 | 918 | struct fib6_node *fn, *saved_fn; |
519fbd87 | 919 | struct rt6_info *rt, *nrt; |
c71099ac | 920 | int strict = 0; |
1da177e4 | 921 | int attempts = 3; |
519fbd87 | 922 | int err; |
1da177e4 | 923 | |
77d16f45 | 924 | strict |= flags & RT6_LOOKUP_F_IFACE; |
367efcb9 MKL |
925 | if (net->ipv6.devconf_all->forwarding == 0) |
926 | strict |= RT6_LOOKUP_F_REACHABLE; | |
1da177e4 | 927 | |
a3c00e46 | 928 | redo_fib6_lookup_lock: |
c71099ac | 929 | read_lock_bh(&table->tb6_lock); |
1da177e4 | 930 | |
4c9483b2 | 931 | fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr); |
367efcb9 | 932 | saved_fn = fn; |
1da177e4 | 933 | |
a3c00e46 | 934 | redo_rt6_select: |
367efcb9 | 935 | rt = rt6_select(fn, oif, strict); |
52bd4c0c | 936 | if (rt->rt6i_nsiblings) |
367efcb9 | 937 | rt = rt6_multipath_select(rt, fl6, oif, strict); |
a3c00e46 MKL |
938 | if (rt == net->ipv6.ip6_null_entry) { |
939 | fn = fib6_backtrack(fn, &fl6->saddr); | |
940 | if (fn) | |
941 | goto redo_rt6_select; | |
367efcb9 MKL |
942 | else if (strict & RT6_LOOKUP_F_REACHABLE) { |
943 | /* also consider unreachable route */ | |
944 | strict &= ~RT6_LOOKUP_F_REACHABLE; | |
945 | fn = saved_fn; | |
946 | goto redo_rt6_select; | |
947 | } else { | |
948 | dst_hold(&rt->dst); | |
949 | read_unlock_bh(&table->tb6_lock); | |
950 | goto out2; | |
951 | } | |
a3c00e46 MKL |
952 | } |
953 | ||
d8d1f30b | 954 | dst_hold(&rt->dst); |
c71099ac | 955 | read_unlock_bh(&table->tb6_lock); |
fb9de91e | 956 | |
94c77bb4 MKL |
957 | if (rt->rt6i_flags & RTF_CACHE) |
958 | goto out2; | |
959 | ||
c440f160 | 960 | if (!(rt->rt6i_flags & (RTF_NONEXTHOP | RTF_GATEWAY))) |
4c9483b2 | 961 | nrt = rt6_alloc_cow(rt, &fl6->daddr, &fl6->saddr); |
7343ff31 | 962 | else if (!(rt->dst.flags & DST_HOST)) |
4c9483b2 | 963 | nrt = rt6_alloc_clone(rt, &fl6->daddr); |
7343ff31 DM |
964 | else |
965 | goto out2; | |
e40cf353 | 966 | |
94e187c0 | 967 | ip6_rt_put(rt); |
8ed67789 | 968 | rt = nrt ? : net->ipv6.ip6_null_entry; |
1da177e4 | 969 | |
d8d1f30b | 970 | dst_hold(&rt->dst); |
519fbd87 | 971 | if (nrt) { |
40e22e8f | 972 | err = ip6_ins_rt(nrt); |
519fbd87 | 973 | if (!err) |
1da177e4 | 974 | goto out2; |
1da177e4 | 975 | } |
1da177e4 | 976 | |
519fbd87 YH |
977 | if (--attempts <= 0) |
978 | goto out2; | |
979 | ||
980 | /* | |
c71099ac | 981 | * Race condition! In the gap, when table->tb6_lock was |
519fbd87 YH |
982 | * released someone could insert this route. Relookup. |
983 | */ | |
94e187c0 | 984 | ip6_rt_put(rt); |
a3c00e46 | 985 | goto redo_fib6_lookup_lock; |
519fbd87 | 986 | |
1da177e4 | 987 | out2: |
d8d1f30b CG |
988 | rt->dst.lastuse = jiffies; |
989 | rt->dst.__use++; | |
c71099ac TG |
990 | |
991 | return rt; | |
1da177e4 LT |
992 | } |
993 | ||
8ed67789 | 994 | static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table, |
4c9483b2 | 995 | struct flowi6 *fl6, int flags) |
4acad72d | 996 | { |
4c9483b2 | 997 | return ip6_pol_route(net, table, fl6->flowi6_iif, fl6, flags); |
4acad72d PE |
998 | } |
999 | ||
72331bc0 SL |
1000 | static struct dst_entry *ip6_route_input_lookup(struct net *net, |
1001 | struct net_device *dev, | |
1002 | struct flowi6 *fl6, int flags) | |
1003 | { | |
1004 | if (rt6_need_strict(&fl6->daddr) && dev->type != ARPHRD_PIMREG) | |
1005 | flags |= RT6_LOOKUP_F_IFACE; | |
1006 | ||
1007 | return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_input); | |
1008 | } | |
1009 | ||
c71099ac TG |
1010 | void ip6_route_input(struct sk_buff *skb) |
1011 | { | |
b71d1d42 | 1012 | const struct ipv6hdr *iph = ipv6_hdr(skb); |
c346dca1 | 1013 | struct net *net = dev_net(skb->dev); |
adaa70bb | 1014 | int flags = RT6_LOOKUP_F_HAS_SADDR; |
4c9483b2 DM |
1015 | struct flowi6 fl6 = { |
1016 | .flowi6_iif = skb->dev->ifindex, | |
1017 | .daddr = iph->daddr, | |
1018 | .saddr = iph->saddr, | |
6502ca52 | 1019 | .flowlabel = ip6_flowinfo(iph), |
4c9483b2 DM |
1020 | .flowi6_mark = skb->mark, |
1021 | .flowi6_proto = iph->nexthdr, | |
c71099ac | 1022 | }; |
adaa70bb | 1023 | |
72331bc0 | 1024 | skb_dst_set(skb, ip6_route_input_lookup(net, skb->dev, &fl6, flags)); |
c71099ac TG |
1025 | } |
1026 | ||
8ed67789 | 1027 | static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table, |
4c9483b2 | 1028 | struct flowi6 *fl6, int flags) |
1da177e4 | 1029 | { |
4c9483b2 | 1030 | return ip6_pol_route(net, table, fl6->flowi6_oif, fl6, flags); |
c71099ac TG |
1031 | } |
1032 | ||
67ba4152 | 1033 | struct dst_entry *ip6_route_output(struct net *net, const struct sock *sk, |
4c9483b2 | 1034 | struct flowi6 *fl6) |
c71099ac TG |
1035 | { |
1036 | int flags = 0; | |
1037 | ||
1fb9489b | 1038 | fl6->flowi6_iif = LOOPBACK_IFINDEX; |
4dc27d1c | 1039 | |
4c9483b2 | 1040 | if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl6->daddr)) |
77d16f45 | 1041 | flags |= RT6_LOOKUP_F_IFACE; |
c71099ac | 1042 | |
4c9483b2 | 1043 | if (!ipv6_addr_any(&fl6->saddr)) |
adaa70bb | 1044 | flags |= RT6_LOOKUP_F_HAS_SADDR; |
0c9a2ac1 YH |
1045 | else if (sk) |
1046 | flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs); | |
adaa70bb | 1047 | |
4c9483b2 | 1048 | return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_output); |
1da177e4 | 1049 | } |
7159039a | 1050 | EXPORT_SYMBOL(ip6_route_output); |
1da177e4 | 1051 | |
2774c131 | 1052 | struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_orig) |
14e50e57 | 1053 | { |
5c1e6aa3 | 1054 | struct rt6_info *rt, *ort = (struct rt6_info *) dst_orig; |
14e50e57 DM |
1055 | struct dst_entry *new = NULL; |
1056 | ||
f5b0a874 | 1057 | rt = dst_alloc(&ip6_dst_blackhole_ops, ort->dst.dev, 1, DST_OBSOLETE_NONE, 0); |
14e50e57 | 1058 | if (rt) { |
d8d1f30b | 1059 | new = &rt->dst; |
14e50e57 | 1060 | |
8104891b SK |
1061 | memset(new + 1, 0, sizeof(*rt) - sizeof(*new)); |
1062 | rt6_init_peer(rt, net->ipv6.peers); | |
1063 | ||
14e50e57 | 1064 | new->__use = 1; |
352e512c | 1065 | new->input = dst_discard; |
aad88724 | 1066 | new->output = dst_discard_sk; |
14e50e57 | 1067 | |
21efcfa0 ED |
1068 | if (dst_metrics_read_only(&ort->dst)) |
1069 | new->_metrics = ort->dst._metrics; | |
1070 | else | |
1071 | dst_copy_metrics(new, &ort->dst); | |
14e50e57 DM |
1072 | rt->rt6i_idev = ort->rt6i_idev; |
1073 | if (rt->rt6i_idev) | |
1074 | in6_dev_hold(rt->rt6i_idev); | |
14e50e57 | 1075 | |
4e3fd7a0 | 1076 | rt->rt6i_gateway = ort->rt6i_gateway; |
1716a961 | 1077 | rt->rt6i_flags = ort->rt6i_flags; |
14e50e57 DM |
1078 | rt->rt6i_metric = 0; |
1079 | ||
1080 | memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key)); | |
1081 | #ifdef CONFIG_IPV6_SUBTREES | |
1082 | memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key)); | |
1083 | #endif | |
1084 | ||
1085 | dst_free(new); | |
1086 | } | |
1087 | ||
69ead7af DM |
1088 | dst_release(dst_orig); |
1089 | return new ? new : ERR_PTR(-ENOMEM); | |
14e50e57 | 1090 | } |
14e50e57 | 1091 | |
1da177e4 LT |
1092 | /* |
1093 | * Destination cache support functions | |
1094 | */ | |
1095 | ||
1096 | static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie) | |
1097 | { | |
1098 | struct rt6_info *rt; | |
1099 | ||
1100 | rt = (struct rt6_info *) dst; | |
1101 | ||
6f3118b5 ND |
1102 | /* All IPV6 dsts are created with ->obsolete set to the value |
1103 | * DST_OBSOLETE_FORCE_CHK which forces validation calls down | |
1104 | * into this function always. | |
1105 | */ | |
e3bc10bd HFS |
1106 | if (!rt->rt6i_node || (rt->rt6i_node->fn_sernum != cookie)) |
1107 | return NULL; | |
a4477c4d | 1108 | |
e3bc10bd HFS |
1109 | if (rt6_check_expired(rt)) |
1110 | return NULL; | |
1111 | ||
1112 | return dst; | |
1da177e4 LT |
1113 | } |
1114 | ||
1115 | static struct dst_entry *ip6_negative_advice(struct dst_entry *dst) | |
1116 | { | |
1117 | struct rt6_info *rt = (struct rt6_info *) dst; | |
1118 | ||
1119 | if (rt) { | |
54c1a859 YH |
1120 | if (rt->rt6i_flags & RTF_CACHE) { |
1121 | if (rt6_check_expired(rt)) { | |
1122 | ip6_del_rt(rt); | |
1123 | dst = NULL; | |
1124 | } | |
1125 | } else { | |
1da177e4 | 1126 | dst_release(dst); |
54c1a859 YH |
1127 | dst = NULL; |
1128 | } | |
1da177e4 | 1129 | } |
54c1a859 | 1130 | return dst; |
1da177e4 LT |
1131 | } |
1132 | ||
1133 | static void ip6_link_failure(struct sk_buff *skb) | |
1134 | { | |
1135 | struct rt6_info *rt; | |
1136 | ||
3ffe533c | 1137 | icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0); |
1da177e4 | 1138 | |
adf30907 | 1139 | rt = (struct rt6_info *) skb_dst(skb); |
1da177e4 | 1140 | if (rt) { |
1eb4f758 HFS |
1141 | if (rt->rt6i_flags & RTF_CACHE) { |
1142 | dst_hold(&rt->dst); | |
1143 | if (ip6_del_rt(rt)) | |
1144 | dst_free(&rt->dst); | |
1145 | } else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT)) { | |
1da177e4 | 1146 | rt->rt6i_node->fn_sernum = -1; |
1eb4f758 | 1147 | } |
1da177e4 LT |
1148 | } |
1149 | } | |
1150 | ||
6700c270 DM |
1151 | static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk, |
1152 | struct sk_buff *skb, u32 mtu) | |
1da177e4 | 1153 | { |
67ba4152 | 1154 | struct rt6_info *rt6 = (struct rt6_info *)dst; |
1da177e4 | 1155 | |
81aded24 | 1156 | dst_confirm(dst); |
1da177e4 | 1157 | if (mtu < dst_mtu(dst) && rt6->rt6i_dst.plen == 128) { |
81aded24 DM |
1158 | struct net *net = dev_net(dst->dev); |
1159 | ||
1da177e4 | 1160 | rt6->rt6i_flags |= RTF_MODIFIED; |
9d289715 | 1161 | if (mtu < IPV6_MIN_MTU) |
1da177e4 | 1162 | mtu = IPV6_MIN_MTU; |
9d289715 | 1163 | |
defb3519 | 1164 | dst_metric_set(dst, RTAX_MTU, mtu); |
81aded24 | 1165 | rt6_update_expires(rt6, net->ipv6.sysctl.ip6_rt_mtu_expires); |
1da177e4 LT |
1166 | } |
1167 | } | |
1168 | ||
42ae66c8 DM |
1169 | void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu, |
1170 | int oif, u32 mark) | |
81aded24 DM |
1171 | { |
1172 | const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data; | |
1173 | struct dst_entry *dst; | |
1174 | struct flowi6 fl6; | |
1175 | ||
1176 | memset(&fl6, 0, sizeof(fl6)); | |
1177 | fl6.flowi6_oif = oif; | |
1b3c61dc | 1178 | fl6.flowi6_mark = mark ? mark : IP6_REPLY_MARK(net, skb->mark); |
81aded24 DM |
1179 | fl6.daddr = iph->daddr; |
1180 | fl6.saddr = iph->saddr; | |
6502ca52 | 1181 | fl6.flowlabel = ip6_flowinfo(iph); |
81aded24 DM |
1182 | |
1183 | dst = ip6_route_output(net, NULL, &fl6); | |
1184 | if (!dst->error) | |
6700c270 | 1185 | ip6_rt_update_pmtu(dst, NULL, skb, ntohl(mtu)); |
81aded24 DM |
1186 | dst_release(dst); |
1187 | } | |
1188 | EXPORT_SYMBOL_GPL(ip6_update_pmtu); | |
1189 | ||
1190 | void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu) | |
1191 | { | |
1192 | ip6_update_pmtu(skb, sock_net(sk), mtu, | |
1193 | sk->sk_bound_dev_if, sk->sk_mark); | |
1194 | } | |
1195 | EXPORT_SYMBOL_GPL(ip6_sk_update_pmtu); | |
1196 | ||
b55b76b2 DJ |
1197 | /* Handle redirects */ |
1198 | struct ip6rd_flowi { | |
1199 | struct flowi6 fl6; | |
1200 | struct in6_addr gateway; | |
1201 | }; | |
1202 | ||
1203 | static struct rt6_info *__ip6_route_redirect(struct net *net, | |
1204 | struct fib6_table *table, | |
1205 | struct flowi6 *fl6, | |
1206 | int flags) | |
1207 | { | |
1208 | struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl6; | |
1209 | struct rt6_info *rt; | |
1210 | struct fib6_node *fn; | |
1211 | ||
1212 | /* Get the "current" route for this destination and | |
1213 | * check if the redirect has come from approriate router. | |
1214 | * | |
1215 | * RFC 4861 specifies that redirects should only be | |
1216 | * accepted if they come from the nexthop to the target. | |
1217 | * Due to the way the routes are chosen, this notion | |
1218 | * is a bit fuzzy and one might need to check all possible | |
1219 | * routes. | |
1220 | */ | |
1221 | ||
1222 | read_lock_bh(&table->tb6_lock); | |
1223 | fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr); | |
1224 | restart: | |
1225 | for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) { | |
1226 | if (rt6_check_expired(rt)) | |
1227 | continue; | |
1228 | if (rt->dst.error) | |
1229 | break; | |
1230 | if (!(rt->rt6i_flags & RTF_GATEWAY)) | |
1231 | continue; | |
1232 | if (fl6->flowi6_oif != rt->dst.dev->ifindex) | |
1233 | continue; | |
1234 | if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway)) | |
1235 | continue; | |
1236 | break; | |
1237 | } | |
1238 | ||
1239 | if (!rt) | |
1240 | rt = net->ipv6.ip6_null_entry; | |
1241 | else if (rt->dst.error) { | |
1242 | rt = net->ipv6.ip6_null_entry; | |
b0a1ba59 MKL |
1243 | goto out; |
1244 | } | |
1245 | ||
1246 | if (rt == net->ipv6.ip6_null_entry) { | |
a3c00e46 MKL |
1247 | fn = fib6_backtrack(fn, &fl6->saddr); |
1248 | if (fn) | |
1249 | goto restart; | |
b55b76b2 | 1250 | } |
a3c00e46 | 1251 | |
b0a1ba59 | 1252 | out: |
b55b76b2 DJ |
1253 | dst_hold(&rt->dst); |
1254 | ||
1255 | read_unlock_bh(&table->tb6_lock); | |
1256 | ||
1257 | return rt; | |
1258 | }; | |
1259 | ||
1260 | static struct dst_entry *ip6_route_redirect(struct net *net, | |
1261 | const struct flowi6 *fl6, | |
1262 | const struct in6_addr *gateway) | |
1263 | { | |
1264 | int flags = RT6_LOOKUP_F_HAS_SADDR; | |
1265 | struct ip6rd_flowi rdfl; | |
1266 | ||
1267 | rdfl.fl6 = *fl6; | |
1268 | rdfl.gateway = *gateway; | |
1269 | ||
1270 | return fib6_rule_lookup(net, &rdfl.fl6, | |
1271 | flags, __ip6_route_redirect); | |
1272 | } | |
1273 | ||
3a5ad2ee DM |
1274 | void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark) |
1275 | { | |
1276 | const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data; | |
1277 | struct dst_entry *dst; | |
1278 | struct flowi6 fl6; | |
1279 | ||
1280 | memset(&fl6, 0, sizeof(fl6)); | |
e374c618 | 1281 | fl6.flowi6_iif = LOOPBACK_IFINDEX; |
3a5ad2ee DM |
1282 | fl6.flowi6_oif = oif; |
1283 | fl6.flowi6_mark = mark; | |
3a5ad2ee DM |
1284 | fl6.daddr = iph->daddr; |
1285 | fl6.saddr = iph->saddr; | |
6502ca52 | 1286 | fl6.flowlabel = ip6_flowinfo(iph); |
3a5ad2ee | 1287 | |
b55b76b2 DJ |
1288 | dst = ip6_route_redirect(net, &fl6, &ipv6_hdr(skb)->saddr); |
1289 | rt6_do_redirect(dst, NULL, skb); | |
3a5ad2ee DM |
1290 | dst_release(dst); |
1291 | } | |
1292 | EXPORT_SYMBOL_GPL(ip6_redirect); | |
1293 | ||
c92a59ec DJ |
1294 | void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif, |
1295 | u32 mark) | |
1296 | { | |
1297 | const struct ipv6hdr *iph = ipv6_hdr(skb); | |
1298 | const struct rd_msg *msg = (struct rd_msg *)icmp6_hdr(skb); | |
1299 | struct dst_entry *dst; | |
1300 | struct flowi6 fl6; | |
1301 | ||
1302 | memset(&fl6, 0, sizeof(fl6)); | |
e374c618 | 1303 | fl6.flowi6_iif = LOOPBACK_IFINDEX; |
c92a59ec DJ |
1304 | fl6.flowi6_oif = oif; |
1305 | fl6.flowi6_mark = mark; | |
c92a59ec DJ |
1306 | fl6.daddr = msg->dest; |
1307 | fl6.saddr = iph->daddr; | |
1308 | ||
b55b76b2 DJ |
1309 | dst = ip6_route_redirect(net, &fl6, &iph->saddr); |
1310 | rt6_do_redirect(dst, NULL, skb); | |
c92a59ec DJ |
1311 | dst_release(dst); |
1312 | } | |
1313 | ||
3a5ad2ee DM |
1314 | void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk) |
1315 | { | |
1316 | ip6_redirect(skb, sock_net(sk), sk->sk_bound_dev_if, sk->sk_mark); | |
1317 | } | |
1318 | EXPORT_SYMBOL_GPL(ip6_sk_redirect); | |
1319 | ||
0dbaee3b | 1320 | static unsigned int ip6_default_advmss(const struct dst_entry *dst) |
1da177e4 | 1321 | { |
0dbaee3b DM |
1322 | struct net_device *dev = dst->dev; |
1323 | unsigned int mtu = dst_mtu(dst); | |
1324 | struct net *net = dev_net(dev); | |
1325 | ||
1da177e4 LT |
1326 | mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr); |
1327 | ||
5578689a DL |
1328 | if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss) |
1329 | mtu = net->ipv6.sysctl.ip6_rt_min_advmss; | |
1da177e4 LT |
1330 | |
1331 | /* | |
1ab1457c YH |
1332 | * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and |
1333 | * corresponding MSS is IPV6_MAXPLEN - tcp_header_size. | |
1334 | * IPV6_MAXPLEN is also valid and means: "any MSS, | |
1da177e4 LT |
1335 | * rely only on pmtu discovery" |
1336 | */ | |
1337 | if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr)) | |
1338 | mtu = IPV6_MAXPLEN; | |
1339 | return mtu; | |
1340 | } | |
1341 | ||
ebb762f2 | 1342 | static unsigned int ip6_mtu(const struct dst_entry *dst) |
d33e4553 | 1343 | { |
d33e4553 | 1344 | struct inet6_dev *idev; |
618f9bc7 SK |
1345 | unsigned int mtu = dst_metric_raw(dst, RTAX_MTU); |
1346 | ||
1347 | if (mtu) | |
30f78d8e | 1348 | goto out; |
618f9bc7 SK |
1349 | |
1350 | mtu = IPV6_MIN_MTU; | |
d33e4553 DM |
1351 | |
1352 | rcu_read_lock(); | |
1353 | idev = __in6_dev_get(dst->dev); | |
1354 | if (idev) | |
1355 | mtu = idev->cnf.mtu6; | |
1356 | rcu_read_unlock(); | |
1357 | ||
30f78d8e ED |
1358 | out: |
1359 | return min_t(unsigned int, mtu, IP6_MAX_MTU); | |
d33e4553 DM |
1360 | } |
1361 | ||
3b00944c YH |
1362 | static struct dst_entry *icmp6_dst_gc_list; |
1363 | static DEFINE_SPINLOCK(icmp6_dst_lock); | |
5d0bbeeb | 1364 | |
3b00944c | 1365 | struct dst_entry *icmp6_dst_alloc(struct net_device *dev, |
87a11578 | 1366 | struct flowi6 *fl6) |
1da177e4 | 1367 | { |
87a11578 | 1368 | struct dst_entry *dst; |
1da177e4 LT |
1369 | struct rt6_info *rt; |
1370 | struct inet6_dev *idev = in6_dev_get(dev); | |
c346dca1 | 1371 | struct net *net = dev_net(dev); |
1da177e4 | 1372 | |
38308473 | 1373 | if (unlikely(!idev)) |
122bdf67 | 1374 | return ERR_PTR(-ENODEV); |
1da177e4 | 1375 | |
8b96d22d | 1376 | rt = ip6_dst_alloc(net, dev, 0, NULL); |
38308473 | 1377 | if (unlikely(!rt)) { |
1da177e4 | 1378 | in6_dev_put(idev); |
87a11578 | 1379 | dst = ERR_PTR(-ENOMEM); |
1da177e4 LT |
1380 | goto out; |
1381 | } | |
1382 | ||
8e2ec639 YZ |
1383 | rt->dst.flags |= DST_HOST; |
1384 | rt->dst.output = ip6_output; | |
d8d1f30b | 1385 | atomic_set(&rt->dst.__refcnt, 1); |
550bab42 | 1386 | rt->rt6i_gateway = fl6->daddr; |
87a11578 | 1387 | rt->rt6i_dst.addr = fl6->daddr; |
8e2ec639 YZ |
1388 | rt->rt6i_dst.plen = 128; |
1389 | rt->rt6i_idev = idev; | |
14edd87d | 1390 | dst_metric_set(&rt->dst, RTAX_HOPLIMIT, 0); |
1da177e4 | 1391 | |
3b00944c | 1392 | spin_lock_bh(&icmp6_dst_lock); |
d8d1f30b CG |
1393 | rt->dst.next = icmp6_dst_gc_list; |
1394 | icmp6_dst_gc_list = &rt->dst; | |
3b00944c | 1395 | spin_unlock_bh(&icmp6_dst_lock); |
1da177e4 | 1396 | |
5578689a | 1397 | fib6_force_start_gc(net); |
1da177e4 | 1398 | |
87a11578 DM |
1399 | dst = xfrm_lookup(net, &rt->dst, flowi6_to_flowi(fl6), NULL, 0); |
1400 | ||
1da177e4 | 1401 | out: |
87a11578 | 1402 | return dst; |
1da177e4 LT |
1403 | } |
1404 | ||
3d0f24a7 | 1405 | int icmp6_dst_gc(void) |
1da177e4 | 1406 | { |
e9476e95 | 1407 | struct dst_entry *dst, **pprev; |
3d0f24a7 | 1408 | int more = 0; |
1da177e4 | 1409 | |
3b00944c YH |
1410 | spin_lock_bh(&icmp6_dst_lock); |
1411 | pprev = &icmp6_dst_gc_list; | |
5d0bbeeb | 1412 | |
1da177e4 LT |
1413 | while ((dst = *pprev) != NULL) { |
1414 | if (!atomic_read(&dst->__refcnt)) { | |
1415 | *pprev = dst->next; | |
1416 | dst_free(dst); | |
1da177e4 LT |
1417 | } else { |
1418 | pprev = &dst->next; | |
3d0f24a7 | 1419 | ++more; |
1da177e4 LT |
1420 | } |
1421 | } | |
1422 | ||
3b00944c | 1423 | spin_unlock_bh(&icmp6_dst_lock); |
5d0bbeeb | 1424 | |
3d0f24a7 | 1425 | return more; |
1da177e4 LT |
1426 | } |
1427 | ||
1e493d19 DM |
1428 | static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg), |
1429 | void *arg) | |
1430 | { | |
1431 | struct dst_entry *dst, **pprev; | |
1432 | ||
1433 | spin_lock_bh(&icmp6_dst_lock); | |
1434 | pprev = &icmp6_dst_gc_list; | |
1435 | while ((dst = *pprev) != NULL) { | |
1436 | struct rt6_info *rt = (struct rt6_info *) dst; | |
1437 | if (func(rt, arg)) { | |
1438 | *pprev = dst->next; | |
1439 | dst_free(dst); | |
1440 | } else { | |
1441 | pprev = &dst->next; | |
1442 | } | |
1443 | } | |
1444 | spin_unlock_bh(&icmp6_dst_lock); | |
1445 | } | |
1446 | ||
569d3645 | 1447 | static int ip6_dst_gc(struct dst_ops *ops) |
1da177e4 | 1448 | { |
86393e52 | 1449 | struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops); |
7019b78e DL |
1450 | int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval; |
1451 | int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size; | |
1452 | int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity; | |
1453 | int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout; | |
1454 | unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc; | |
fc66f95c | 1455 | int entries; |
7019b78e | 1456 | |
fc66f95c | 1457 | entries = dst_entries_get_fast(ops); |
49a18d86 | 1458 | if (time_after(rt_last_gc + rt_min_interval, jiffies) && |
fc66f95c | 1459 | entries <= rt_max_size) |
1da177e4 LT |
1460 | goto out; |
1461 | ||
6891a346 | 1462 | net->ipv6.ip6_rt_gc_expire++; |
14956643 | 1463 | fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net, true); |
fc66f95c ED |
1464 | entries = dst_entries_get_slow(ops); |
1465 | if (entries < ops->gc_thresh) | |
7019b78e | 1466 | net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1; |
1da177e4 | 1467 | out: |
7019b78e | 1468 | net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity; |
fc66f95c | 1469 | return entries > rt_max_size; |
1da177e4 LT |
1470 | } |
1471 | ||
e715b6d3 FW |
1472 | static int ip6_convert_metrics(struct mx6_config *mxc, |
1473 | const struct fib6_config *cfg) | |
1474 | { | |
1475 | struct nlattr *nla; | |
1476 | int remaining; | |
1477 | u32 *mp; | |
1478 | ||
63159f29 | 1479 | if (!cfg->fc_mx) |
e715b6d3 FW |
1480 | return 0; |
1481 | ||
1482 | mp = kzalloc(sizeof(u32) * RTAX_MAX, GFP_KERNEL); | |
1483 | if (unlikely(!mp)) | |
1484 | return -ENOMEM; | |
1485 | ||
1486 | nla_for_each_attr(nla, cfg->fc_mx, cfg->fc_mx_len, remaining) { | |
1487 | int type = nla_type(nla); | |
1488 | ||
1489 | if (type) { | |
ea697639 DB |
1490 | u32 val; |
1491 | ||
e715b6d3 FW |
1492 | if (unlikely(type > RTAX_MAX)) |
1493 | goto err; | |
ea697639 DB |
1494 | if (type == RTAX_CC_ALGO) { |
1495 | char tmp[TCP_CA_NAME_MAX]; | |
1496 | ||
1497 | nla_strlcpy(tmp, nla, sizeof(tmp)); | |
1498 | val = tcp_ca_get_key_by_name(tmp); | |
1499 | if (val == TCP_CA_UNSPEC) | |
1500 | goto err; | |
1501 | } else { | |
1502 | val = nla_get_u32(nla); | |
1503 | } | |
e715b6d3 | 1504 | |
ea697639 | 1505 | mp[type - 1] = val; |
e715b6d3 FW |
1506 | __set_bit(type - 1, mxc->mx_valid); |
1507 | } | |
1508 | } | |
1509 | ||
1510 | mxc->mx = mp; | |
1511 | ||
1512 | return 0; | |
1513 | err: | |
1514 | kfree(mp); | |
1515 | return -EINVAL; | |
1516 | } | |
1da177e4 | 1517 | |
86872cb5 | 1518 | int ip6_route_add(struct fib6_config *cfg) |
1da177e4 LT |
1519 | { |
1520 | int err; | |
5578689a | 1521 | struct net *net = cfg->fc_nlinfo.nl_net; |
1da177e4 LT |
1522 | struct rt6_info *rt = NULL; |
1523 | struct net_device *dev = NULL; | |
1524 | struct inet6_dev *idev = NULL; | |
c71099ac | 1525 | struct fib6_table *table; |
e715b6d3 | 1526 | struct mx6_config mxc = { .mx = NULL, }; |
1da177e4 LT |
1527 | int addr_type; |
1528 | ||
86872cb5 | 1529 | if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128) |
1da177e4 LT |
1530 | return -EINVAL; |
1531 | #ifndef CONFIG_IPV6_SUBTREES | |
86872cb5 | 1532 | if (cfg->fc_src_len) |
1da177e4 LT |
1533 | return -EINVAL; |
1534 | #endif | |
86872cb5 | 1535 | if (cfg->fc_ifindex) { |
1da177e4 | 1536 | err = -ENODEV; |
5578689a | 1537 | dev = dev_get_by_index(net, cfg->fc_ifindex); |
1da177e4 LT |
1538 | if (!dev) |
1539 | goto out; | |
1540 | idev = in6_dev_get(dev); | |
1541 | if (!idev) | |
1542 | goto out; | |
1543 | } | |
1544 | ||
86872cb5 TG |
1545 | if (cfg->fc_metric == 0) |
1546 | cfg->fc_metric = IP6_RT_PRIO_USER; | |
1da177e4 | 1547 | |
d71314b4 | 1548 | err = -ENOBUFS; |
38308473 DM |
1549 | if (cfg->fc_nlinfo.nlh && |
1550 | !(cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_CREATE)) { | |
d71314b4 | 1551 | table = fib6_get_table(net, cfg->fc_table); |
38308473 | 1552 | if (!table) { |
f3213831 | 1553 | pr_warn("NLM_F_CREATE should be specified when creating new route\n"); |
d71314b4 MV |
1554 | table = fib6_new_table(net, cfg->fc_table); |
1555 | } | |
1556 | } else { | |
1557 | table = fib6_new_table(net, cfg->fc_table); | |
1558 | } | |
38308473 DM |
1559 | |
1560 | if (!table) | |
c71099ac | 1561 | goto out; |
c71099ac | 1562 | |
c88507fb | 1563 | rt = ip6_dst_alloc(net, NULL, (cfg->fc_flags & RTF_ADDRCONF) ? 0 : DST_NOCOUNT, table); |
1da177e4 | 1564 | |
38308473 | 1565 | if (!rt) { |
1da177e4 LT |
1566 | err = -ENOMEM; |
1567 | goto out; | |
1568 | } | |
1569 | ||
1716a961 G |
1570 | if (cfg->fc_flags & RTF_EXPIRES) |
1571 | rt6_set_expires(rt, jiffies + | |
1572 | clock_t_to_jiffies(cfg->fc_expires)); | |
1573 | else | |
1574 | rt6_clean_expires(rt); | |
1da177e4 | 1575 | |
86872cb5 TG |
1576 | if (cfg->fc_protocol == RTPROT_UNSPEC) |
1577 | cfg->fc_protocol = RTPROT_BOOT; | |
1578 | rt->rt6i_protocol = cfg->fc_protocol; | |
1579 | ||
1580 | addr_type = ipv6_addr_type(&cfg->fc_dst); | |
1da177e4 LT |
1581 | |
1582 | if (addr_type & IPV6_ADDR_MULTICAST) | |
d8d1f30b | 1583 | rt->dst.input = ip6_mc_input; |
ab79ad14 MŻ |
1584 | else if (cfg->fc_flags & RTF_LOCAL) |
1585 | rt->dst.input = ip6_input; | |
1da177e4 | 1586 | else |
d8d1f30b | 1587 | rt->dst.input = ip6_forward; |
1da177e4 | 1588 | |
d8d1f30b | 1589 | rt->dst.output = ip6_output; |
1da177e4 | 1590 | |
86872cb5 TG |
1591 | ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len); |
1592 | rt->rt6i_dst.plen = cfg->fc_dst_len; | |
e5fd387a MK |
1593 | if (rt->rt6i_dst.plen == 128) { |
1594 | rt->dst.flags |= DST_HOST; | |
1595 | dst_metrics_set_force_overwrite(&rt->dst); | |
8e2ec639 | 1596 | } |
e5fd387a | 1597 | |
1da177e4 | 1598 | #ifdef CONFIG_IPV6_SUBTREES |
86872cb5 TG |
1599 | ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len); |
1600 | rt->rt6i_src.plen = cfg->fc_src_len; | |
1da177e4 LT |
1601 | #endif |
1602 | ||
86872cb5 | 1603 | rt->rt6i_metric = cfg->fc_metric; |
1da177e4 LT |
1604 | |
1605 | /* We cannot add true routes via loopback here, | |
1606 | they would result in kernel looping; promote them to reject routes | |
1607 | */ | |
86872cb5 | 1608 | if ((cfg->fc_flags & RTF_REJECT) || |
38308473 DM |
1609 | (dev && (dev->flags & IFF_LOOPBACK) && |
1610 | !(addr_type & IPV6_ADDR_LOOPBACK) && | |
1611 | !(cfg->fc_flags & RTF_LOCAL))) { | |
1da177e4 | 1612 | /* hold loopback dev/idev if we haven't done so. */ |
5578689a | 1613 | if (dev != net->loopback_dev) { |
1da177e4 LT |
1614 | if (dev) { |
1615 | dev_put(dev); | |
1616 | in6_dev_put(idev); | |
1617 | } | |
5578689a | 1618 | dev = net->loopback_dev; |
1da177e4 LT |
1619 | dev_hold(dev); |
1620 | idev = in6_dev_get(dev); | |
1621 | if (!idev) { | |
1622 | err = -ENODEV; | |
1623 | goto out; | |
1624 | } | |
1625 | } | |
1da177e4 | 1626 | rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP; |
ef2c7d7b ND |
1627 | switch (cfg->fc_type) { |
1628 | case RTN_BLACKHOLE: | |
1629 | rt->dst.error = -EINVAL; | |
aad88724 | 1630 | rt->dst.output = dst_discard_sk; |
7150aede | 1631 | rt->dst.input = dst_discard; |
ef2c7d7b ND |
1632 | break; |
1633 | case RTN_PROHIBIT: | |
1634 | rt->dst.error = -EACCES; | |
7150aede K |
1635 | rt->dst.output = ip6_pkt_prohibit_out; |
1636 | rt->dst.input = ip6_pkt_prohibit; | |
ef2c7d7b | 1637 | break; |
b4949ab2 | 1638 | case RTN_THROW: |
ef2c7d7b | 1639 | default: |
7150aede K |
1640 | rt->dst.error = (cfg->fc_type == RTN_THROW) ? -EAGAIN |
1641 | : -ENETUNREACH; | |
1642 | rt->dst.output = ip6_pkt_discard_out; | |
1643 | rt->dst.input = ip6_pkt_discard; | |
ef2c7d7b ND |
1644 | break; |
1645 | } | |
1da177e4 LT |
1646 | goto install_route; |
1647 | } | |
1648 | ||
86872cb5 | 1649 | if (cfg->fc_flags & RTF_GATEWAY) { |
b71d1d42 | 1650 | const struct in6_addr *gw_addr; |
1da177e4 LT |
1651 | int gwa_type; |
1652 | ||
86872cb5 | 1653 | gw_addr = &cfg->fc_gateway; |
4e3fd7a0 | 1654 | rt->rt6i_gateway = *gw_addr; |
1da177e4 LT |
1655 | gwa_type = ipv6_addr_type(gw_addr); |
1656 | ||
1657 | if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) { | |
1658 | struct rt6_info *grt; | |
1659 | ||
1660 | /* IPv6 strictly inhibits using not link-local | |
1661 | addresses as nexthop address. | |
1662 | Otherwise, router will not able to send redirects. | |
1663 | It is very good, but in some (rare!) circumstances | |
1664 | (SIT, PtP, NBMA NOARP links) it is handy to allow | |
1665 | some exceptions. --ANK | |
1666 | */ | |
1667 | err = -EINVAL; | |
38308473 | 1668 | if (!(gwa_type & IPV6_ADDR_UNICAST)) |
1da177e4 LT |
1669 | goto out; |
1670 | ||
5578689a | 1671 | grt = rt6_lookup(net, gw_addr, NULL, cfg->fc_ifindex, 1); |
1da177e4 LT |
1672 | |
1673 | err = -EHOSTUNREACH; | |
38308473 | 1674 | if (!grt) |
1da177e4 LT |
1675 | goto out; |
1676 | if (dev) { | |
d1918542 | 1677 | if (dev != grt->dst.dev) { |
94e187c0 | 1678 | ip6_rt_put(grt); |
1da177e4 LT |
1679 | goto out; |
1680 | } | |
1681 | } else { | |
d1918542 | 1682 | dev = grt->dst.dev; |
1da177e4 LT |
1683 | idev = grt->rt6i_idev; |
1684 | dev_hold(dev); | |
1685 | in6_dev_hold(grt->rt6i_idev); | |
1686 | } | |
38308473 | 1687 | if (!(grt->rt6i_flags & RTF_GATEWAY)) |
1da177e4 | 1688 | err = 0; |
94e187c0 | 1689 | ip6_rt_put(grt); |
1da177e4 LT |
1690 | |
1691 | if (err) | |
1692 | goto out; | |
1693 | } | |
1694 | err = -EINVAL; | |
38308473 | 1695 | if (!dev || (dev->flags & IFF_LOOPBACK)) |
1da177e4 LT |
1696 | goto out; |
1697 | } | |
1698 | ||
1699 | err = -ENODEV; | |
38308473 | 1700 | if (!dev) |
1da177e4 LT |
1701 | goto out; |
1702 | ||
c3968a85 DW |
1703 | if (!ipv6_addr_any(&cfg->fc_prefsrc)) { |
1704 | if (!ipv6_chk_addr(net, &cfg->fc_prefsrc, dev, 0)) { | |
1705 | err = -EINVAL; | |
1706 | goto out; | |
1707 | } | |
4e3fd7a0 | 1708 | rt->rt6i_prefsrc.addr = cfg->fc_prefsrc; |
c3968a85 DW |
1709 | rt->rt6i_prefsrc.plen = 128; |
1710 | } else | |
1711 | rt->rt6i_prefsrc.plen = 0; | |
1712 | ||
86872cb5 | 1713 | rt->rt6i_flags = cfg->fc_flags; |
1da177e4 LT |
1714 | |
1715 | install_route: | |
d8d1f30b | 1716 | rt->dst.dev = dev; |
1da177e4 | 1717 | rt->rt6i_idev = idev; |
c71099ac | 1718 | rt->rt6i_table = table; |
63152fc0 | 1719 | |
c346dca1 | 1720 | cfg->fc_nlinfo.nl_net = dev_net(dev); |
63152fc0 | 1721 | |
e715b6d3 FW |
1722 | err = ip6_convert_metrics(&mxc, cfg); |
1723 | if (err) | |
1724 | goto out; | |
1da177e4 | 1725 | |
e715b6d3 FW |
1726 | err = __ip6_ins_rt(rt, &cfg->fc_nlinfo, &mxc); |
1727 | ||
1728 | kfree(mxc.mx); | |
1729 | return err; | |
1da177e4 LT |
1730 | out: |
1731 | if (dev) | |
1732 | dev_put(dev); | |
1733 | if (idev) | |
1734 | in6_dev_put(idev); | |
1735 | if (rt) | |
d8d1f30b | 1736 | dst_free(&rt->dst); |
1da177e4 LT |
1737 | return err; |
1738 | } | |
1739 | ||
86872cb5 | 1740 | static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info) |
1da177e4 LT |
1741 | { |
1742 | int err; | |
c71099ac | 1743 | struct fib6_table *table; |
d1918542 | 1744 | struct net *net = dev_net(rt->dst.dev); |
1da177e4 | 1745 | |
6825a26c G |
1746 | if (rt == net->ipv6.ip6_null_entry) { |
1747 | err = -ENOENT; | |
1748 | goto out; | |
1749 | } | |
6c813a72 | 1750 | |
c71099ac TG |
1751 | table = rt->rt6i_table; |
1752 | write_lock_bh(&table->tb6_lock); | |
86872cb5 | 1753 | err = fib6_del(rt, info); |
c71099ac | 1754 | write_unlock_bh(&table->tb6_lock); |
1da177e4 | 1755 | |
6825a26c | 1756 | out: |
94e187c0 | 1757 | ip6_rt_put(rt); |
1da177e4 LT |
1758 | return err; |
1759 | } | |
1760 | ||
e0a1ad73 TG |
1761 | int ip6_del_rt(struct rt6_info *rt) |
1762 | { | |
4d1169c1 | 1763 | struct nl_info info = { |
d1918542 | 1764 | .nl_net = dev_net(rt->dst.dev), |
4d1169c1 | 1765 | }; |
528c4ceb | 1766 | return __ip6_del_rt(rt, &info); |
e0a1ad73 TG |
1767 | } |
1768 | ||
86872cb5 | 1769 | static int ip6_route_del(struct fib6_config *cfg) |
1da177e4 | 1770 | { |
c71099ac | 1771 | struct fib6_table *table; |
1da177e4 LT |
1772 | struct fib6_node *fn; |
1773 | struct rt6_info *rt; | |
1774 | int err = -ESRCH; | |
1775 | ||
5578689a | 1776 | table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table); |
38308473 | 1777 | if (!table) |
c71099ac TG |
1778 | return err; |
1779 | ||
1780 | read_lock_bh(&table->tb6_lock); | |
1da177e4 | 1781 | |
c71099ac | 1782 | fn = fib6_locate(&table->tb6_root, |
86872cb5 TG |
1783 | &cfg->fc_dst, cfg->fc_dst_len, |
1784 | &cfg->fc_src, cfg->fc_src_len); | |
1ab1457c | 1785 | |
1da177e4 | 1786 | if (fn) { |
d8d1f30b | 1787 | for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) { |
86872cb5 | 1788 | if (cfg->fc_ifindex && |
d1918542 DM |
1789 | (!rt->dst.dev || |
1790 | rt->dst.dev->ifindex != cfg->fc_ifindex)) | |
1da177e4 | 1791 | continue; |
86872cb5 TG |
1792 | if (cfg->fc_flags & RTF_GATEWAY && |
1793 | !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway)) | |
1da177e4 | 1794 | continue; |
86872cb5 | 1795 | if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric) |
1da177e4 | 1796 | continue; |
d8d1f30b | 1797 | dst_hold(&rt->dst); |
c71099ac | 1798 | read_unlock_bh(&table->tb6_lock); |
1da177e4 | 1799 | |
86872cb5 | 1800 | return __ip6_del_rt(rt, &cfg->fc_nlinfo); |
1da177e4 LT |
1801 | } |
1802 | } | |
c71099ac | 1803 | read_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
1804 | |
1805 | return err; | |
1806 | } | |
1807 | ||
6700c270 | 1808 | static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_buff *skb) |
a6279458 | 1809 | { |
e8599ff4 | 1810 | struct net *net = dev_net(skb->dev); |
a6279458 | 1811 | struct netevent_redirect netevent; |
e8599ff4 | 1812 | struct rt6_info *rt, *nrt = NULL; |
e8599ff4 DM |
1813 | struct ndisc_options ndopts; |
1814 | struct inet6_dev *in6_dev; | |
1815 | struct neighbour *neigh; | |
71bcdba0 | 1816 | struct rd_msg *msg; |
6e157b6a DM |
1817 | int optlen, on_link; |
1818 | u8 *lladdr; | |
e8599ff4 | 1819 | |
29a3cad5 | 1820 | optlen = skb_tail_pointer(skb) - skb_transport_header(skb); |
71bcdba0 | 1821 | optlen -= sizeof(*msg); |
e8599ff4 DM |
1822 | |
1823 | if (optlen < 0) { | |
6e157b6a | 1824 | net_dbg_ratelimited("rt6_do_redirect: packet too short\n"); |
e8599ff4 DM |
1825 | return; |
1826 | } | |
1827 | ||
71bcdba0 | 1828 | msg = (struct rd_msg *)icmp6_hdr(skb); |
e8599ff4 | 1829 | |
71bcdba0 | 1830 | if (ipv6_addr_is_multicast(&msg->dest)) { |
6e157b6a | 1831 | net_dbg_ratelimited("rt6_do_redirect: destination address is multicast\n"); |
e8599ff4 DM |
1832 | return; |
1833 | } | |
1834 | ||
6e157b6a | 1835 | on_link = 0; |
71bcdba0 | 1836 | if (ipv6_addr_equal(&msg->dest, &msg->target)) { |
e8599ff4 | 1837 | on_link = 1; |
71bcdba0 | 1838 | } else if (ipv6_addr_type(&msg->target) != |
e8599ff4 | 1839 | (IPV6_ADDR_UNICAST|IPV6_ADDR_LINKLOCAL)) { |
6e157b6a | 1840 | net_dbg_ratelimited("rt6_do_redirect: target address is not link-local unicast\n"); |
e8599ff4 DM |
1841 | return; |
1842 | } | |
1843 | ||
1844 | in6_dev = __in6_dev_get(skb->dev); | |
1845 | if (!in6_dev) | |
1846 | return; | |
1847 | if (in6_dev->cnf.forwarding || !in6_dev->cnf.accept_redirects) | |
1848 | return; | |
1849 | ||
1850 | /* RFC2461 8.1: | |
1851 | * The IP source address of the Redirect MUST be the same as the current | |
1852 | * first-hop router for the specified ICMP Destination Address. | |
1853 | */ | |
1854 | ||
71bcdba0 | 1855 | if (!ndisc_parse_options(msg->opt, optlen, &ndopts)) { |
e8599ff4 DM |
1856 | net_dbg_ratelimited("rt6_redirect: invalid ND options\n"); |
1857 | return; | |
1858 | } | |
6e157b6a DM |
1859 | |
1860 | lladdr = NULL; | |
e8599ff4 DM |
1861 | if (ndopts.nd_opts_tgt_lladdr) { |
1862 | lladdr = ndisc_opt_addr_data(ndopts.nd_opts_tgt_lladdr, | |
1863 | skb->dev); | |
1864 | if (!lladdr) { | |
1865 | net_dbg_ratelimited("rt6_redirect: invalid link-layer address length\n"); | |
1866 | return; | |
1867 | } | |
1868 | } | |
1869 | ||
6e157b6a DM |
1870 | rt = (struct rt6_info *) dst; |
1871 | if (rt == net->ipv6.ip6_null_entry) { | |
1872 | net_dbg_ratelimited("rt6_redirect: source isn't a valid nexthop for redirect target\n"); | |
e8599ff4 | 1873 | return; |
6e157b6a | 1874 | } |
e8599ff4 | 1875 | |
6e157b6a DM |
1876 | /* Redirect received -> path was valid. |
1877 | * Look, redirects are sent only in response to data packets, | |
1878 | * so that this nexthop apparently is reachable. --ANK | |
1879 | */ | |
1880 | dst_confirm(&rt->dst); | |
a6279458 | 1881 | |
71bcdba0 | 1882 | neigh = __neigh_lookup(&nd_tbl, &msg->target, skb->dev, 1); |
6e157b6a DM |
1883 | if (!neigh) |
1884 | return; | |
a6279458 | 1885 | |
1da177e4 LT |
1886 | /* |
1887 | * We have finally decided to accept it. | |
1888 | */ | |
1889 | ||
1ab1457c | 1890 | neigh_update(neigh, lladdr, NUD_STALE, |
1da177e4 LT |
1891 | NEIGH_UPDATE_F_WEAK_OVERRIDE| |
1892 | NEIGH_UPDATE_F_OVERRIDE| | |
1893 | (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER| | |
1894 | NEIGH_UPDATE_F_ISROUTER)) | |
1895 | ); | |
1896 | ||
71bcdba0 | 1897 | nrt = ip6_rt_copy(rt, &msg->dest); |
38308473 | 1898 | if (!nrt) |
1da177e4 LT |
1899 | goto out; |
1900 | ||
1901 | nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE; | |
1902 | if (on_link) | |
1903 | nrt->rt6i_flags &= ~RTF_GATEWAY; | |
1904 | ||
4e3fd7a0 | 1905 | nrt->rt6i_gateway = *(struct in6_addr *)neigh->primary_key; |
1da177e4 | 1906 | |
40e22e8f | 1907 | if (ip6_ins_rt(nrt)) |
1da177e4 LT |
1908 | goto out; |
1909 | ||
d8d1f30b CG |
1910 | netevent.old = &rt->dst; |
1911 | netevent.new = &nrt->dst; | |
71bcdba0 | 1912 | netevent.daddr = &msg->dest; |
60592833 | 1913 | netevent.neigh = neigh; |
8d71740c TT |
1914 | call_netevent_notifiers(NETEVENT_REDIRECT, &netevent); |
1915 | ||
38308473 | 1916 | if (rt->rt6i_flags & RTF_CACHE) { |
6e157b6a | 1917 | rt = (struct rt6_info *) dst_clone(&rt->dst); |
e0a1ad73 | 1918 | ip6_del_rt(rt); |
1da177e4 LT |
1919 | } |
1920 | ||
1921 | out: | |
e8599ff4 | 1922 | neigh_release(neigh); |
6e157b6a DM |
1923 | } |
1924 | ||
1da177e4 LT |
1925 | /* |
1926 | * Misc support functions | |
1927 | */ | |
1928 | ||
1716a961 | 1929 | static struct rt6_info *ip6_rt_copy(struct rt6_info *ort, |
21efcfa0 | 1930 | const struct in6_addr *dest) |
1da177e4 | 1931 | { |
d1918542 | 1932 | struct net *net = dev_net(ort->dst.dev); |
8b96d22d DM |
1933 | struct rt6_info *rt = ip6_dst_alloc(net, ort->dst.dev, 0, |
1934 | ort->rt6i_table); | |
1da177e4 LT |
1935 | |
1936 | if (rt) { | |
d8d1f30b CG |
1937 | rt->dst.input = ort->dst.input; |
1938 | rt->dst.output = ort->dst.output; | |
8e2ec639 | 1939 | rt->dst.flags |= DST_HOST; |
d8d1f30b | 1940 | |
4e3fd7a0 | 1941 | rt->rt6i_dst.addr = *dest; |
8e2ec639 | 1942 | rt->rt6i_dst.plen = 128; |
defb3519 | 1943 | dst_copy_metrics(&rt->dst, &ort->dst); |
d8d1f30b | 1944 | rt->dst.error = ort->dst.error; |
1da177e4 LT |
1945 | rt->rt6i_idev = ort->rt6i_idev; |
1946 | if (rt->rt6i_idev) | |
1947 | in6_dev_hold(rt->rt6i_idev); | |
d8d1f30b | 1948 | rt->dst.lastuse = jiffies; |
1da177e4 | 1949 | |
550bab42 JA |
1950 | if (ort->rt6i_flags & RTF_GATEWAY) |
1951 | rt->rt6i_gateway = ort->rt6i_gateway; | |
1952 | else | |
1953 | rt->rt6i_gateway = *dest; | |
1716a961 | 1954 | rt->rt6i_flags = ort->rt6i_flags; |
24f5b855 | 1955 | rt6_set_from(rt, ort); |
1da177e4 LT |
1956 | rt->rt6i_metric = 0; |
1957 | ||
1da177e4 LT |
1958 | #ifdef CONFIG_IPV6_SUBTREES |
1959 | memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key)); | |
1960 | #endif | |
0f6c6392 | 1961 | memcpy(&rt->rt6i_prefsrc, &ort->rt6i_prefsrc, sizeof(struct rt6key)); |
c71099ac | 1962 | rt->rt6i_table = ort->rt6i_table; |
1da177e4 LT |
1963 | } |
1964 | return rt; | |
1965 | } | |
1966 | ||
70ceb4f5 | 1967 | #ifdef CONFIG_IPV6_ROUTE_INFO |
efa2cea0 | 1968 | static struct rt6_info *rt6_get_route_info(struct net *net, |
b71d1d42 ED |
1969 | const struct in6_addr *prefix, int prefixlen, |
1970 | const struct in6_addr *gwaddr, int ifindex) | |
70ceb4f5 YH |
1971 | { |
1972 | struct fib6_node *fn; | |
1973 | struct rt6_info *rt = NULL; | |
c71099ac TG |
1974 | struct fib6_table *table; |
1975 | ||
efa2cea0 | 1976 | table = fib6_get_table(net, RT6_TABLE_INFO); |
38308473 | 1977 | if (!table) |
c71099ac | 1978 | return NULL; |
70ceb4f5 | 1979 | |
5744dd9b | 1980 | read_lock_bh(&table->tb6_lock); |
67ba4152 | 1981 | fn = fib6_locate(&table->tb6_root, prefix, prefixlen, NULL, 0); |
70ceb4f5 YH |
1982 | if (!fn) |
1983 | goto out; | |
1984 | ||
d8d1f30b | 1985 | for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) { |
d1918542 | 1986 | if (rt->dst.dev->ifindex != ifindex) |
70ceb4f5 YH |
1987 | continue; |
1988 | if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY)) | |
1989 | continue; | |
1990 | if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr)) | |
1991 | continue; | |
d8d1f30b | 1992 | dst_hold(&rt->dst); |
70ceb4f5 YH |
1993 | break; |
1994 | } | |
1995 | out: | |
5744dd9b | 1996 | read_unlock_bh(&table->tb6_lock); |
70ceb4f5 YH |
1997 | return rt; |
1998 | } | |
1999 | ||
efa2cea0 | 2000 | static struct rt6_info *rt6_add_route_info(struct net *net, |
b71d1d42 ED |
2001 | const struct in6_addr *prefix, int prefixlen, |
2002 | const struct in6_addr *gwaddr, int ifindex, | |
95c96174 | 2003 | unsigned int pref) |
70ceb4f5 | 2004 | { |
86872cb5 TG |
2005 | struct fib6_config cfg = { |
2006 | .fc_table = RT6_TABLE_INFO, | |
238fc7ea | 2007 | .fc_metric = IP6_RT_PRIO_USER, |
86872cb5 TG |
2008 | .fc_ifindex = ifindex, |
2009 | .fc_dst_len = prefixlen, | |
2010 | .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO | | |
2011 | RTF_UP | RTF_PREF(pref), | |
15e47304 | 2012 | .fc_nlinfo.portid = 0, |
efa2cea0 DL |
2013 | .fc_nlinfo.nlh = NULL, |
2014 | .fc_nlinfo.nl_net = net, | |
86872cb5 TG |
2015 | }; |
2016 | ||
4e3fd7a0 AD |
2017 | cfg.fc_dst = *prefix; |
2018 | cfg.fc_gateway = *gwaddr; | |
70ceb4f5 | 2019 | |
e317da96 YH |
2020 | /* We should treat it as a default route if prefix length is 0. */ |
2021 | if (!prefixlen) | |
86872cb5 | 2022 | cfg.fc_flags |= RTF_DEFAULT; |
70ceb4f5 | 2023 | |
86872cb5 | 2024 | ip6_route_add(&cfg); |
70ceb4f5 | 2025 | |
efa2cea0 | 2026 | return rt6_get_route_info(net, prefix, prefixlen, gwaddr, ifindex); |
70ceb4f5 YH |
2027 | } |
2028 | #endif | |
2029 | ||
b71d1d42 | 2030 | struct rt6_info *rt6_get_dflt_router(const struct in6_addr *addr, struct net_device *dev) |
1ab1457c | 2031 | { |
1da177e4 | 2032 | struct rt6_info *rt; |
c71099ac | 2033 | struct fib6_table *table; |
1da177e4 | 2034 | |
c346dca1 | 2035 | table = fib6_get_table(dev_net(dev), RT6_TABLE_DFLT); |
38308473 | 2036 | if (!table) |
c71099ac | 2037 | return NULL; |
1da177e4 | 2038 | |
5744dd9b | 2039 | read_lock_bh(&table->tb6_lock); |
67ba4152 | 2040 | for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) { |
d1918542 | 2041 | if (dev == rt->dst.dev && |
045927ff | 2042 | ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) && |
1da177e4 LT |
2043 | ipv6_addr_equal(&rt->rt6i_gateway, addr)) |
2044 | break; | |
2045 | } | |
2046 | if (rt) | |
d8d1f30b | 2047 | dst_hold(&rt->dst); |
5744dd9b | 2048 | read_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
2049 | return rt; |
2050 | } | |
2051 | ||
b71d1d42 | 2052 | struct rt6_info *rt6_add_dflt_router(const struct in6_addr *gwaddr, |
ebacaaa0 YH |
2053 | struct net_device *dev, |
2054 | unsigned int pref) | |
1da177e4 | 2055 | { |
86872cb5 TG |
2056 | struct fib6_config cfg = { |
2057 | .fc_table = RT6_TABLE_DFLT, | |
238fc7ea | 2058 | .fc_metric = IP6_RT_PRIO_USER, |
86872cb5 TG |
2059 | .fc_ifindex = dev->ifindex, |
2060 | .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT | | |
2061 | RTF_UP | RTF_EXPIRES | RTF_PREF(pref), | |
15e47304 | 2062 | .fc_nlinfo.portid = 0, |
5578689a | 2063 | .fc_nlinfo.nlh = NULL, |
c346dca1 | 2064 | .fc_nlinfo.nl_net = dev_net(dev), |
86872cb5 | 2065 | }; |
1da177e4 | 2066 | |
4e3fd7a0 | 2067 | cfg.fc_gateway = *gwaddr; |
1da177e4 | 2068 | |
86872cb5 | 2069 | ip6_route_add(&cfg); |
1da177e4 | 2070 | |
1da177e4 LT |
2071 | return rt6_get_dflt_router(gwaddr, dev); |
2072 | } | |
2073 | ||
7b4da532 | 2074 | void rt6_purge_dflt_routers(struct net *net) |
1da177e4 LT |
2075 | { |
2076 | struct rt6_info *rt; | |
c71099ac TG |
2077 | struct fib6_table *table; |
2078 | ||
2079 | /* NOTE: Keep consistent with rt6_get_dflt_router */ | |
7b4da532 | 2080 | table = fib6_get_table(net, RT6_TABLE_DFLT); |
38308473 | 2081 | if (!table) |
c71099ac | 2082 | return; |
1da177e4 LT |
2083 | |
2084 | restart: | |
c71099ac | 2085 | read_lock_bh(&table->tb6_lock); |
d8d1f30b | 2086 | for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) { |
3e8b0ac3 LC |
2087 | if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF) && |
2088 | (!rt->rt6i_idev || rt->rt6i_idev->cnf.accept_ra != 2)) { | |
d8d1f30b | 2089 | dst_hold(&rt->dst); |
c71099ac | 2090 | read_unlock_bh(&table->tb6_lock); |
e0a1ad73 | 2091 | ip6_del_rt(rt); |
1da177e4 LT |
2092 | goto restart; |
2093 | } | |
2094 | } | |
c71099ac | 2095 | read_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
2096 | } |
2097 | ||
5578689a DL |
2098 | static void rtmsg_to_fib6_config(struct net *net, |
2099 | struct in6_rtmsg *rtmsg, | |
86872cb5 TG |
2100 | struct fib6_config *cfg) |
2101 | { | |
2102 | memset(cfg, 0, sizeof(*cfg)); | |
2103 | ||
2104 | cfg->fc_table = RT6_TABLE_MAIN; | |
2105 | cfg->fc_ifindex = rtmsg->rtmsg_ifindex; | |
2106 | cfg->fc_metric = rtmsg->rtmsg_metric; | |
2107 | cfg->fc_expires = rtmsg->rtmsg_info; | |
2108 | cfg->fc_dst_len = rtmsg->rtmsg_dst_len; | |
2109 | cfg->fc_src_len = rtmsg->rtmsg_src_len; | |
2110 | cfg->fc_flags = rtmsg->rtmsg_flags; | |
2111 | ||
5578689a | 2112 | cfg->fc_nlinfo.nl_net = net; |
f1243c2d | 2113 | |
4e3fd7a0 AD |
2114 | cfg->fc_dst = rtmsg->rtmsg_dst; |
2115 | cfg->fc_src = rtmsg->rtmsg_src; | |
2116 | cfg->fc_gateway = rtmsg->rtmsg_gateway; | |
86872cb5 TG |
2117 | } |
2118 | ||
5578689a | 2119 | int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg) |
1da177e4 | 2120 | { |
86872cb5 | 2121 | struct fib6_config cfg; |
1da177e4 LT |
2122 | struct in6_rtmsg rtmsg; |
2123 | int err; | |
2124 | ||
67ba4152 | 2125 | switch (cmd) { |
1da177e4 LT |
2126 | case SIOCADDRT: /* Add a route */ |
2127 | case SIOCDELRT: /* Delete a route */ | |
af31f412 | 2128 | if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) |
1da177e4 LT |
2129 | return -EPERM; |
2130 | err = copy_from_user(&rtmsg, arg, | |
2131 | sizeof(struct in6_rtmsg)); | |
2132 | if (err) | |
2133 | return -EFAULT; | |
86872cb5 | 2134 | |
5578689a | 2135 | rtmsg_to_fib6_config(net, &rtmsg, &cfg); |
86872cb5 | 2136 | |
1da177e4 LT |
2137 | rtnl_lock(); |
2138 | switch (cmd) { | |
2139 | case SIOCADDRT: | |
86872cb5 | 2140 | err = ip6_route_add(&cfg); |
1da177e4 LT |
2141 | break; |
2142 | case SIOCDELRT: | |
86872cb5 | 2143 | err = ip6_route_del(&cfg); |
1da177e4 LT |
2144 | break; |
2145 | default: | |
2146 | err = -EINVAL; | |
2147 | } | |
2148 | rtnl_unlock(); | |
2149 | ||
2150 | return err; | |
3ff50b79 | 2151 | } |
1da177e4 LT |
2152 | |
2153 | return -EINVAL; | |
2154 | } | |
2155 | ||
2156 | /* | |
2157 | * Drop the packet on the floor | |
2158 | */ | |
2159 | ||
d5fdd6ba | 2160 | static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes) |
1da177e4 | 2161 | { |
612f09e8 | 2162 | int type; |
adf30907 | 2163 | struct dst_entry *dst = skb_dst(skb); |
612f09e8 YH |
2164 | switch (ipstats_mib_noroutes) { |
2165 | case IPSTATS_MIB_INNOROUTES: | |
0660e03f | 2166 | type = ipv6_addr_type(&ipv6_hdr(skb)->daddr); |
45bb0060 | 2167 | if (type == IPV6_ADDR_ANY) { |
3bd653c8 DL |
2168 | IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst), |
2169 | IPSTATS_MIB_INADDRERRORS); | |
612f09e8 YH |
2170 | break; |
2171 | } | |
2172 | /* FALLTHROUGH */ | |
2173 | case IPSTATS_MIB_OUTNOROUTES: | |
3bd653c8 DL |
2174 | IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst), |
2175 | ipstats_mib_noroutes); | |
612f09e8 YH |
2176 | break; |
2177 | } | |
3ffe533c | 2178 | icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0); |
1da177e4 LT |
2179 | kfree_skb(skb); |
2180 | return 0; | |
2181 | } | |
2182 | ||
9ce8ade0 TG |
2183 | static int ip6_pkt_discard(struct sk_buff *skb) |
2184 | { | |
612f09e8 | 2185 | return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES); |
9ce8ade0 TG |
2186 | } |
2187 | ||
aad88724 | 2188 | static int ip6_pkt_discard_out(struct sock *sk, struct sk_buff *skb) |
1da177e4 | 2189 | { |
adf30907 | 2190 | skb->dev = skb_dst(skb)->dev; |
612f09e8 | 2191 | return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES); |
1da177e4 LT |
2192 | } |
2193 | ||
9ce8ade0 TG |
2194 | static int ip6_pkt_prohibit(struct sk_buff *skb) |
2195 | { | |
612f09e8 | 2196 | return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES); |
9ce8ade0 TG |
2197 | } |
2198 | ||
aad88724 | 2199 | static int ip6_pkt_prohibit_out(struct sock *sk, struct sk_buff *skb) |
9ce8ade0 | 2200 | { |
adf30907 | 2201 | skb->dev = skb_dst(skb)->dev; |
612f09e8 | 2202 | return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES); |
9ce8ade0 TG |
2203 | } |
2204 | ||
1da177e4 LT |
2205 | /* |
2206 | * Allocate a dst for local (unicast / anycast) address. | |
2207 | */ | |
2208 | ||
2209 | struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev, | |
2210 | const struct in6_addr *addr, | |
8f031519 | 2211 | bool anycast) |
1da177e4 | 2212 | { |
c346dca1 | 2213 | struct net *net = dev_net(idev->dev); |
a3300ef4 HFS |
2214 | struct rt6_info *rt = ip6_dst_alloc(net, net->loopback_dev, |
2215 | DST_NOCOUNT, NULL); | |
2216 | if (!rt) | |
1da177e4 LT |
2217 | return ERR_PTR(-ENOMEM); |
2218 | ||
1da177e4 LT |
2219 | in6_dev_hold(idev); |
2220 | ||
11d53b49 | 2221 | rt->dst.flags |= DST_HOST; |
d8d1f30b CG |
2222 | rt->dst.input = ip6_input; |
2223 | rt->dst.output = ip6_output; | |
1da177e4 | 2224 | rt->rt6i_idev = idev; |
1da177e4 LT |
2225 | |
2226 | rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP; | |
58c4fb86 YH |
2227 | if (anycast) |
2228 | rt->rt6i_flags |= RTF_ANYCAST; | |
2229 | else | |
1da177e4 | 2230 | rt->rt6i_flags |= RTF_LOCAL; |
1da177e4 | 2231 | |
550bab42 | 2232 | rt->rt6i_gateway = *addr; |
4e3fd7a0 | 2233 | rt->rt6i_dst.addr = *addr; |
1da177e4 | 2234 | rt->rt6i_dst.plen = 128; |
5578689a | 2235 | rt->rt6i_table = fib6_get_table(net, RT6_TABLE_LOCAL); |
1da177e4 | 2236 | |
d8d1f30b | 2237 | atomic_set(&rt->dst.__refcnt, 1); |
1da177e4 LT |
2238 | |
2239 | return rt; | |
2240 | } | |
2241 | ||
c3968a85 DW |
2242 | int ip6_route_get_saddr(struct net *net, |
2243 | struct rt6_info *rt, | |
b71d1d42 | 2244 | const struct in6_addr *daddr, |
c3968a85 DW |
2245 | unsigned int prefs, |
2246 | struct in6_addr *saddr) | |
2247 | { | |
67ba4152 | 2248 | struct inet6_dev *idev = ip6_dst_idev((struct dst_entry *)rt); |
c3968a85 DW |
2249 | int err = 0; |
2250 | if (rt->rt6i_prefsrc.plen) | |
4e3fd7a0 | 2251 | *saddr = rt->rt6i_prefsrc.addr; |
c3968a85 DW |
2252 | else |
2253 | err = ipv6_dev_get_saddr(net, idev ? idev->dev : NULL, | |
2254 | daddr, prefs, saddr); | |
2255 | return err; | |
2256 | } | |
2257 | ||
2258 | /* remove deleted ip from prefsrc entries */ | |
2259 | struct arg_dev_net_ip { | |
2260 | struct net_device *dev; | |
2261 | struct net *net; | |
2262 | struct in6_addr *addr; | |
2263 | }; | |
2264 | ||
2265 | static int fib6_remove_prefsrc(struct rt6_info *rt, void *arg) | |
2266 | { | |
2267 | struct net_device *dev = ((struct arg_dev_net_ip *)arg)->dev; | |
2268 | struct net *net = ((struct arg_dev_net_ip *)arg)->net; | |
2269 | struct in6_addr *addr = ((struct arg_dev_net_ip *)arg)->addr; | |
2270 | ||
d1918542 | 2271 | if (((void *)rt->dst.dev == dev || !dev) && |
c3968a85 DW |
2272 | rt != net->ipv6.ip6_null_entry && |
2273 | ipv6_addr_equal(addr, &rt->rt6i_prefsrc.addr)) { | |
2274 | /* remove prefsrc entry */ | |
2275 | rt->rt6i_prefsrc.plen = 0; | |
2276 | } | |
2277 | return 0; | |
2278 | } | |
2279 | ||
2280 | void rt6_remove_prefsrc(struct inet6_ifaddr *ifp) | |
2281 | { | |
2282 | struct net *net = dev_net(ifp->idev->dev); | |
2283 | struct arg_dev_net_ip adni = { | |
2284 | .dev = ifp->idev->dev, | |
2285 | .net = net, | |
2286 | .addr = &ifp->addr, | |
2287 | }; | |
0c3584d5 | 2288 | fib6_clean_all(net, fib6_remove_prefsrc, &adni); |
c3968a85 DW |
2289 | } |
2290 | ||
be7a010d DJ |
2291 | #define RTF_RA_ROUTER (RTF_ADDRCONF | RTF_DEFAULT | RTF_GATEWAY) |
2292 | #define RTF_CACHE_GATEWAY (RTF_GATEWAY | RTF_CACHE) | |
2293 | ||
2294 | /* Remove routers and update dst entries when gateway turn into host. */ | |
2295 | static int fib6_clean_tohost(struct rt6_info *rt, void *arg) | |
2296 | { | |
2297 | struct in6_addr *gateway = (struct in6_addr *)arg; | |
2298 | ||
2299 | if ((((rt->rt6i_flags & RTF_RA_ROUTER) == RTF_RA_ROUTER) || | |
2300 | ((rt->rt6i_flags & RTF_CACHE_GATEWAY) == RTF_CACHE_GATEWAY)) && | |
2301 | ipv6_addr_equal(gateway, &rt->rt6i_gateway)) { | |
2302 | return -1; | |
2303 | } | |
2304 | return 0; | |
2305 | } | |
2306 | ||
2307 | void rt6_clean_tohost(struct net *net, struct in6_addr *gateway) | |
2308 | { | |
2309 | fib6_clean_all(net, fib6_clean_tohost, gateway); | |
2310 | } | |
2311 | ||
8ed67789 DL |
2312 | struct arg_dev_net { |
2313 | struct net_device *dev; | |
2314 | struct net *net; | |
2315 | }; | |
2316 | ||
1da177e4 LT |
2317 | static int fib6_ifdown(struct rt6_info *rt, void *arg) |
2318 | { | |
bc3ef660 | 2319 | const struct arg_dev_net *adn = arg; |
2320 | const struct net_device *dev = adn->dev; | |
8ed67789 | 2321 | |
d1918542 | 2322 | if ((rt->dst.dev == dev || !dev) && |
c159d30c | 2323 | rt != adn->net->ipv6.ip6_null_entry) |
1da177e4 | 2324 | return -1; |
c159d30c | 2325 | |
1da177e4 LT |
2326 | return 0; |
2327 | } | |
2328 | ||
f3db4851 | 2329 | void rt6_ifdown(struct net *net, struct net_device *dev) |
1da177e4 | 2330 | { |
8ed67789 DL |
2331 | struct arg_dev_net adn = { |
2332 | .dev = dev, | |
2333 | .net = net, | |
2334 | }; | |
2335 | ||
0c3584d5 | 2336 | fib6_clean_all(net, fib6_ifdown, &adn); |
1e493d19 | 2337 | icmp6_clean_all(fib6_ifdown, &adn); |
1da177e4 LT |
2338 | } |
2339 | ||
95c96174 | 2340 | struct rt6_mtu_change_arg { |
1da177e4 | 2341 | struct net_device *dev; |
95c96174 | 2342 | unsigned int mtu; |
1da177e4 LT |
2343 | }; |
2344 | ||
2345 | static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg) | |
2346 | { | |
2347 | struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg; | |
2348 | struct inet6_dev *idev; | |
2349 | ||
2350 | /* In IPv6 pmtu discovery is not optional, | |
2351 | so that RTAX_MTU lock cannot disable it. | |
2352 | We still use this lock to block changes | |
2353 | caused by addrconf/ndisc. | |
2354 | */ | |
2355 | ||
2356 | idev = __in6_dev_get(arg->dev); | |
38308473 | 2357 | if (!idev) |
1da177e4 LT |
2358 | return 0; |
2359 | ||
2360 | /* For administrative MTU increase, there is no way to discover | |
2361 | IPv6 PMTU increase, so PMTU increase should be updated here. | |
2362 | Since RFC 1981 doesn't include administrative MTU increase | |
2363 | update PMTU increase is a MUST. (i.e. jumbo frame) | |
2364 | */ | |
2365 | /* | |
2366 | If new MTU is less than route PMTU, this new MTU will be the | |
2367 | lowest MTU in the path, update the route PMTU to reflect PMTU | |
2368 | decreases; if new MTU is greater than route PMTU, and the | |
2369 | old MTU is the lowest MTU in the path, update the route PMTU | |
2370 | to reflect the increase. In this case if the other nodes' MTU | |
2371 | also have the lowest MTU, TOO BIG MESSAGE will be lead to | |
2372 | PMTU discouvery. | |
2373 | */ | |
d1918542 | 2374 | if (rt->dst.dev == arg->dev && |
d8d1f30b CG |
2375 | !dst_metric_locked(&rt->dst, RTAX_MTU) && |
2376 | (dst_mtu(&rt->dst) >= arg->mtu || | |
2377 | (dst_mtu(&rt->dst) < arg->mtu && | |
2378 | dst_mtu(&rt->dst) == idev->cnf.mtu6))) { | |
defb3519 | 2379 | dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu); |
566cfd8f | 2380 | } |
1da177e4 LT |
2381 | return 0; |
2382 | } | |
2383 | ||
95c96174 | 2384 | void rt6_mtu_change(struct net_device *dev, unsigned int mtu) |
1da177e4 | 2385 | { |
c71099ac TG |
2386 | struct rt6_mtu_change_arg arg = { |
2387 | .dev = dev, | |
2388 | .mtu = mtu, | |
2389 | }; | |
1da177e4 | 2390 | |
0c3584d5 | 2391 | fib6_clean_all(dev_net(dev), rt6_mtu_change_route, &arg); |
1da177e4 LT |
2392 | } |
2393 | ||
ef7c79ed | 2394 | static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = { |
5176f91e | 2395 | [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) }, |
86872cb5 | 2396 | [RTA_OIF] = { .type = NLA_U32 }, |
ab364a6f | 2397 | [RTA_IIF] = { .type = NLA_U32 }, |
86872cb5 TG |
2398 | [RTA_PRIORITY] = { .type = NLA_U32 }, |
2399 | [RTA_METRICS] = { .type = NLA_NESTED }, | |
51ebd318 | 2400 | [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) }, |
c78ba6d6 | 2401 | [RTA_PREF] = { .type = NLA_U8 }, |
86872cb5 TG |
2402 | }; |
2403 | ||
2404 | static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh, | |
2405 | struct fib6_config *cfg) | |
1da177e4 | 2406 | { |
86872cb5 TG |
2407 | struct rtmsg *rtm; |
2408 | struct nlattr *tb[RTA_MAX+1]; | |
c78ba6d6 | 2409 | unsigned int pref; |
86872cb5 | 2410 | int err; |
1da177e4 | 2411 | |
86872cb5 TG |
2412 | err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy); |
2413 | if (err < 0) | |
2414 | goto errout; | |
1da177e4 | 2415 | |
86872cb5 TG |
2416 | err = -EINVAL; |
2417 | rtm = nlmsg_data(nlh); | |
2418 | memset(cfg, 0, sizeof(*cfg)); | |
2419 | ||
2420 | cfg->fc_table = rtm->rtm_table; | |
2421 | cfg->fc_dst_len = rtm->rtm_dst_len; | |
2422 | cfg->fc_src_len = rtm->rtm_src_len; | |
2423 | cfg->fc_flags = RTF_UP; | |
2424 | cfg->fc_protocol = rtm->rtm_protocol; | |
ef2c7d7b | 2425 | cfg->fc_type = rtm->rtm_type; |
86872cb5 | 2426 | |
ef2c7d7b ND |
2427 | if (rtm->rtm_type == RTN_UNREACHABLE || |
2428 | rtm->rtm_type == RTN_BLACKHOLE || | |
b4949ab2 ND |
2429 | rtm->rtm_type == RTN_PROHIBIT || |
2430 | rtm->rtm_type == RTN_THROW) | |
86872cb5 TG |
2431 | cfg->fc_flags |= RTF_REJECT; |
2432 | ||
ab79ad14 MŻ |
2433 | if (rtm->rtm_type == RTN_LOCAL) |
2434 | cfg->fc_flags |= RTF_LOCAL; | |
2435 | ||
15e47304 | 2436 | cfg->fc_nlinfo.portid = NETLINK_CB(skb).portid; |
86872cb5 | 2437 | cfg->fc_nlinfo.nlh = nlh; |
3b1e0a65 | 2438 | cfg->fc_nlinfo.nl_net = sock_net(skb->sk); |
86872cb5 TG |
2439 | |
2440 | if (tb[RTA_GATEWAY]) { | |
67b61f6c | 2441 | cfg->fc_gateway = nla_get_in6_addr(tb[RTA_GATEWAY]); |
86872cb5 | 2442 | cfg->fc_flags |= RTF_GATEWAY; |
1da177e4 | 2443 | } |
86872cb5 TG |
2444 | |
2445 | if (tb[RTA_DST]) { | |
2446 | int plen = (rtm->rtm_dst_len + 7) >> 3; | |
2447 | ||
2448 | if (nla_len(tb[RTA_DST]) < plen) | |
2449 | goto errout; | |
2450 | ||
2451 | nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen); | |
1da177e4 | 2452 | } |
86872cb5 TG |
2453 | |
2454 | if (tb[RTA_SRC]) { | |
2455 | int plen = (rtm->rtm_src_len + 7) >> 3; | |
2456 | ||
2457 | if (nla_len(tb[RTA_SRC]) < plen) | |
2458 | goto errout; | |
2459 | ||
2460 | nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen); | |
1da177e4 | 2461 | } |
86872cb5 | 2462 | |
c3968a85 | 2463 | if (tb[RTA_PREFSRC]) |
67b61f6c | 2464 | cfg->fc_prefsrc = nla_get_in6_addr(tb[RTA_PREFSRC]); |
c3968a85 | 2465 | |
86872cb5 TG |
2466 | if (tb[RTA_OIF]) |
2467 | cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]); | |
2468 | ||
2469 | if (tb[RTA_PRIORITY]) | |
2470 | cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]); | |
2471 | ||
2472 | if (tb[RTA_METRICS]) { | |
2473 | cfg->fc_mx = nla_data(tb[RTA_METRICS]); | |
2474 | cfg->fc_mx_len = nla_len(tb[RTA_METRICS]); | |
1da177e4 | 2475 | } |
86872cb5 TG |
2476 | |
2477 | if (tb[RTA_TABLE]) | |
2478 | cfg->fc_table = nla_get_u32(tb[RTA_TABLE]); | |
2479 | ||
51ebd318 ND |
2480 | if (tb[RTA_MULTIPATH]) { |
2481 | cfg->fc_mp = nla_data(tb[RTA_MULTIPATH]); | |
2482 | cfg->fc_mp_len = nla_len(tb[RTA_MULTIPATH]); | |
2483 | } | |
2484 | ||
c78ba6d6 LR |
2485 | if (tb[RTA_PREF]) { |
2486 | pref = nla_get_u8(tb[RTA_PREF]); | |
2487 | if (pref != ICMPV6_ROUTER_PREF_LOW && | |
2488 | pref != ICMPV6_ROUTER_PREF_HIGH) | |
2489 | pref = ICMPV6_ROUTER_PREF_MEDIUM; | |
2490 | cfg->fc_flags |= RTF_PREF(pref); | |
2491 | } | |
2492 | ||
86872cb5 TG |
2493 | err = 0; |
2494 | errout: | |
2495 | return err; | |
1da177e4 LT |
2496 | } |
2497 | ||
51ebd318 ND |
2498 | static int ip6_route_multipath(struct fib6_config *cfg, int add) |
2499 | { | |
2500 | struct fib6_config r_cfg; | |
2501 | struct rtnexthop *rtnh; | |
2502 | int remaining; | |
2503 | int attrlen; | |
2504 | int err = 0, last_err = 0; | |
2505 | ||
2506 | beginning: | |
2507 | rtnh = (struct rtnexthop *)cfg->fc_mp; | |
2508 | remaining = cfg->fc_mp_len; | |
2509 | ||
2510 | /* Parse a Multipath Entry */ | |
2511 | while (rtnh_ok(rtnh, remaining)) { | |
2512 | memcpy(&r_cfg, cfg, sizeof(*cfg)); | |
2513 | if (rtnh->rtnh_ifindex) | |
2514 | r_cfg.fc_ifindex = rtnh->rtnh_ifindex; | |
2515 | ||
2516 | attrlen = rtnh_attrlen(rtnh); | |
2517 | if (attrlen > 0) { | |
2518 | struct nlattr *nla, *attrs = rtnh_attrs(rtnh); | |
2519 | ||
2520 | nla = nla_find(attrs, attrlen, RTA_GATEWAY); | |
2521 | if (nla) { | |
67b61f6c | 2522 | r_cfg.fc_gateway = nla_get_in6_addr(nla); |
51ebd318 ND |
2523 | r_cfg.fc_flags |= RTF_GATEWAY; |
2524 | } | |
2525 | } | |
2526 | err = add ? ip6_route_add(&r_cfg) : ip6_route_del(&r_cfg); | |
2527 | if (err) { | |
2528 | last_err = err; | |
2529 | /* If we are trying to remove a route, do not stop the | |
2530 | * loop when ip6_route_del() fails (because next hop is | |
2531 | * already gone), we should try to remove all next hops. | |
2532 | */ | |
2533 | if (add) { | |
2534 | /* If add fails, we should try to delete all | |
2535 | * next hops that have been already added. | |
2536 | */ | |
2537 | add = 0; | |
2538 | goto beginning; | |
2539 | } | |
2540 | } | |
1a72418b ND |
2541 | /* Because each route is added like a single route we remove |
2542 | * this flag after the first nexthop (if there is a collision, | |
2543 | * we have already fail to add the first nexthop: | |
2544 | * fib6_add_rt2node() has reject it). | |
2545 | */ | |
2546 | cfg->fc_nlinfo.nlh->nlmsg_flags &= ~NLM_F_EXCL; | |
51ebd318 ND |
2547 | rtnh = rtnh_next(rtnh, &remaining); |
2548 | } | |
2549 | ||
2550 | return last_err; | |
2551 | } | |
2552 | ||
67ba4152 | 2553 | static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr *nlh) |
1da177e4 | 2554 | { |
86872cb5 TG |
2555 | struct fib6_config cfg; |
2556 | int err; | |
1da177e4 | 2557 | |
86872cb5 TG |
2558 | err = rtm_to_fib6_config(skb, nlh, &cfg); |
2559 | if (err < 0) | |
2560 | return err; | |
2561 | ||
51ebd318 ND |
2562 | if (cfg.fc_mp) |
2563 | return ip6_route_multipath(&cfg, 0); | |
2564 | else | |
2565 | return ip6_route_del(&cfg); | |
1da177e4 LT |
2566 | } |
2567 | ||
67ba4152 | 2568 | static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr *nlh) |
1da177e4 | 2569 | { |
86872cb5 TG |
2570 | struct fib6_config cfg; |
2571 | int err; | |
1da177e4 | 2572 | |
86872cb5 TG |
2573 | err = rtm_to_fib6_config(skb, nlh, &cfg); |
2574 | if (err < 0) | |
2575 | return err; | |
2576 | ||
51ebd318 ND |
2577 | if (cfg.fc_mp) |
2578 | return ip6_route_multipath(&cfg, 1); | |
2579 | else | |
2580 | return ip6_route_add(&cfg); | |
1da177e4 LT |
2581 | } |
2582 | ||
339bf98f TG |
2583 | static inline size_t rt6_nlmsg_size(void) |
2584 | { | |
2585 | return NLMSG_ALIGN(sizeof(struct rtmsg)) | |
2586 | + nla_total_size(16) /* RTA_SRC */ | |
2587 | + nla_total_size(16) /* RTA_DST */ | |
2588 | + nla_total_size(16) /* RTA_GATEWAY */ | |
2589 | + nla_total_size(16) /* RTA_PREFSRC */ | |
2590 | + nla_total_size(4) /* RTA_TABLE */ | |
2591 | + nla_total_size(4) /* RTA_IIF */ | |
2592 | + nla_total_size(4) /* RTA_OIF */ | |
2593 | + nla_total_size(4) /* RTA_PRIORITY */ | |
6a2b9ce0 | 2594 | + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */ |
ea697639 | 2595 | + nla_total_size(sizeof(struct rta_cacheinfo)) |
c78ba6d6 LR |
2596 | + nla_total_size(TCP_CA_NAME_MAX) /* RTAX_CC_ALGO */ |
2597 | + nla_total_size(1); /* RTA_PREF */ | |
339bf98f TG |
2598 | } |
2599 | ||
191cd582 BH |
2600 | static int rt6_fill_node(struct net *net, |
2601 | struct sk_buff *skb, struct rt6_info *rt, | |
0d51aa80 | 2602 | struct in6_addr *dst, struct in6_addr *src, |
15e47304 | 2603 | int iif, int type, u32 portid, u32 seq, |
7bc570c8 | 2604 | int prefix, int nowait, unsigned int flags) |
1da177e4 LT |
2605 | { |
2606 | struct rtmsg *rtm; | |
2d7202bf | 2607 | struct nlmsghdr *nlh; |
e3703b3d | 2608 | long expires; |
9e762a4a | 2609 | u32 table; |
1da177e4 LT |
2610 | |
2611 | if (prefix) { /* user wants prefix routes only */ | |
2612 | if (!(rt->rt6i_flags & RTF_PREFIX_RT)) { | |
2613 | /* success since this is not a prefix route */ | |
2614 | return 1; | |
2615 | } | |
2616 | } | |
2617 | ||
15e47304 | 2618 | nlh = nlmsg_put(skb, portid, seq, type, sizeof(*rtm), flags); |
38308473 | 2619 | if (!nlh) |
26932566 | 2620 | return -EMSGSIZE; |
2d7202bf TG |
2621 | |
2622 | rtm = nlmsg_data(nlh); | |
1da177e4 LT |
2623 | rtm->rtm_family = AF_INET6; |
2624 | rtm->rtm_dst_len = rt->rt6i_dst.plen; | |
2625 | rtm->rtm_src_len = rt->rt6i_src.plen; | |
2626 | rtm->rtm_tos = 0; | |
c71099ac | 2627 | if (rt->rt6i_table) |
9e762a4a | 2628 | table = rt->rt6i_table->tb6_id; |
c71099ac | 2629 | else |
9e762a4a PM |
2630 | table = RT6_TABLE_UNSPEC; |
2631 | rtm->rtm_table = table; | |
c78679e8 DM |
2632 | if (nla_put_u32(skb, RTA_TABLE, table)) |
2633 | goto nla_put_failure; | |
ef2c7d7b ND |
2634 | if (rt->rt6i_flags & RTF_REJECT) { |
2635 | switch (rt->dst.error) { | |
2636 | case -EINVAL: | |
2637 | rtm->rtm_type = RTN_BLACKHOLE; | |
2638 | break; | |
2639 | case -EACCES: | |
2640 | rtm->rtm_type = RTN_PROHIBIT; | |
2641 | break; | |
b4949ab2 ND |
2642 | case -EAGAIN: |
2643 | rtm->rtm_type = RTN_THROW; | |
2644 | break; | |
ef2c7d7b ND |
2645 | default: |
2646 | rtm->rtm_type = RTN_UNREACHABLE; | |
2647 | break; | |
2648 | } | |
2649 | } | |
38308473 | 2650 | else if (rt->rt6i_flags & RTF_LOCAL) |
ab79ad14 | 2651 | rtm->rtm_type = RTN_LOCAL; |
d1918542 | 2652 | else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK)) |
1da177e4 LT |
2653 | rtm->rtm_type = RTN_LOCAL; |
2654 | else | |
2655 | rtm->rtm_type = RTN_UNICAST; | |
2656 | rtm->rtm_flags = 0; | |
2657 | rtm->rtm_scope = RT_SCOPE_UNIVERSE; | |
2658 | rtm->rtm_protocol = rt->rt6i_protocol; | |
38308473 | 2659 | if (rt->rt6i_flags & RTF_DYNAMIC) |
1da177e4 | 2660 | rtm->rtm_protocol = RTPROT_REDIRECT; |
f0396f60 DO |
2661 | else if (rt->rt6i_flags & RTF_ADDRCONF) { |
2662 | if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ROUTEINFO)) | |
2663 | rtm->rtm_protocol = RTPROT_RA; | |
2664 | else | |
2665 | rtm->rtm_protocol = RTPROT_KERNEL; | |
2666 | } | |
1da177e4 | 2667 | |
38308473 | 2668 | if (rt->rt6i_flags & RTF_CACHE) |
1da177e4 LT |
2669 | rtm->rtm_flags |= RTM_F_CLONED; |
2670 | ||
2671 | if (dst) { | |
930345ea | 2672 | if (nla_put_in6_addr(skb, RTA_DST, dst)) |
c78679e8 | 2673 | goto nla_put_failure; |
1ab1457c | 2674 | rtm->rtm_dst_len = 128; |
1da177e4 | 2675 | } else if (rtm->rtm_dst_len) |
930345ea | 2676 | if (nla_put_in6_addr(skb, RTA_DST, &rt->rt6i_dst.addr)) |
c78679e8 | 2677 | goto nla_put_failure; |
1da177e4 LT |
2678 | #ifdef CONFIG_IPV6_SUBTREES |
2679 | if (src) { | |
930345ea | 2680 | if (nla_put_in6_addr(skb, RTA_SRC, src)) |
c78679e8 | 2681 | goto nla_put_failure; |
1ab1457c | 2682 | rtm->rtm_src_len = 128; |
c78679e8 | 2683 | } else if (rtm->rtm_src_len && |
930345ea | 2684 | nla_put_in6_addr(skb, RTA_SRC, &rt->rt6i_src.addr)) |
c78679e8 | 2685 | goto nla_put_failure; |
1da177e4 | 2686 | #endif |
7bc570c8 YH |
2687 | if (iif) { |
2688 | #ifdef CONFIG_IPV6_MROUTE | |
2689 | if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) { | |
8229efda | 2690 | int err = ip6mr_get_route(net, skb, rtm, nowait); |
7bc570c8 YH |
2691 | if (err <= 0) { |
2692 | if (!nowait) { | |
2693 | if (err == 0) | |
2694 | return 0; | |
2695 | goto nla_put_failure; | |
2696 | } else { | |
2697 | if (err == -EMSGSIZE) | |
2698 | goto nla_put_failure; | |
2699 | } | |
2700 | } | |
2701 | } else | |
2702 | #endif | |
c78679e8 DM |
2703 | if (nla_put_u32(skb, RTA_IIF, iif)) |
2704 | goto nla_put_failure; | |
7bc570c8 | 2705 | } else if (dst) { |
1da177e4 | 2706 | struct in6_addr saddr_buf; |
c78679e8 | 2707 | if (ip6_route_get_saddr(net, rt, dst, 0, &saddr_buf) == 0 && |
930345ea | 2708 | nla_put_in6_addr(skb, RTA_PREFSRC, &saddr_buf)) |
c78679e8 | 2709 | goto nla_put_failure; |
1da177e4 | 2710 | } |
2d7202bf | 2711 | |
c3968a85 DW |
2712 | if (rt->rt6i_prefsrc.plen) { |
2713 | struct in6_addr saddr_buf; | |
4e3fd7a0 | 2714 | saddr_buf = rt->rt6i_prefsrc.addr; |
930345ea | 2715 | if (nla_put_in6_addr(skb, RTA_PREFSRC, &saddr_buf)) |
c78679e8 | 2716 | goto nla_put_failure; |
c3968a85 DW |
2717 | } |
2718 | ||
defb3519 | 2719 | if (rtnetlink_put_metrics(skb, dst_metrics_ptr(&rt->dst)) < 0) |
2d7202bf TG |
2720 | goto nla_put_failure; |
2721 | ||
dd0cbf29 | 2722 | if (rt->rt6i_flags & RTF_GATEWAY) { |
930345ea | 2723 | if (nla_put_in6_addr(skb, RTA_GATEWAY, &rt->rt6i_gateway) < 0) |
94f826b8 | 2724 | goto nla_put_failure; |
94f826b8 | 2725 | } |
2d7202bf | 2726 | |
c78679e8 DM |
2727 | if (rt->dst.dev && |
2728 | nla_put_u32(skb, RTA_OIF, rt->dst.dev->ifindex)) | |
2729 | goto nla_put_failure; | |
2730 | if (nla_put_u32(skb, RTA_PRIORITY, rt->rt6i_metric)) | |
2731 | goto nla_put_failure; | |
8253947e LW |
2732 | |
2733 | expires = (rt->rt6i_flags & RTF_EXPIRES) ? rt->dst.expires - jiffies : 0; | |
69cdf8f9 | 2734 | |
87a50699 | 2735 | if (rtnl_put_cacheinfo(skb, &rt->dst, 0, expires, rt->dst.error) < 0) |
e3703b3d | 2736 | goto nla_put_failure; |
2d7202bf | 2737 | |
c78ba6d6 LR |
2738 | if (nla_put_u8(skb, RTA_PREF, IPV6_EXTRACT_PREF(rt->rt6i_flags))) |
2739 | goto nla_put_failure; | |
2740 | ||
053c095a JB |
2741 | nlmsg_end(skb, nlh); |
2742 | return 0; | |
2d7202bf TG |
2743 | |
2744 | nla_put_failure: | |
26932566 PM |
2745 | nlmsg_cancel(skb, nlh); |
2746 | return -EMSGSIZE; | |
1da177e4 LT |
2747 | } |
2748 | ||
1b43af54 | 2749 | int rt6_dump_route(struct rt6_info *rt, void *p_arg) |
1da177e4 LT |
2750 | { |
2751 | struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg; | |
2752 | int prefix; | |
2753 | ||
2d7202bf TG |
2754 | if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) { |
2755 | struct rtmsg *rtm = nlmsg_data(arg->cb->nlh); | |
1da177e4 LT |
2756 | prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0; |
2757 | } else | |
2758 | prefix = 0; | |
2759 | ||
191cd582 BH |
2760 | return rt6_fill_node(arg->net, |
2761 | arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE, | |
15e47304 | 2762 | NETLINK_CB(arg->cb->skb).portid, arg->cb->nlh->nlmsg_seq, |
7bc570c8 | 2763 | prefix, 0, NLM_F_MULTI); |
1da177e4 LT |
2764 | } |
2765 | ||
67ba4152 | 2766 | static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh) |
1da177e4 | 2767 | { |
3b1e0a65 | 2768 | struct net *net = sock_net(in_skb->sk); |
ab364a6f TG |
2769 | struct nlattr *tb[RTA_MAX+1]; |
2770 | struct rt6_info *rt; | |
1da177e4 | 2771 | struct sk_buff *skb; |
ab364a6f | 2772 | struct rtmsg *rtm; |
4c9483b2 | 2773 | struct flowi6 fl6; |
72331bc0 | 2774 | int err, iif = 0, oif = 0; |
1da177e4 | 2775 | |
ab364a6f TG |
2776 | err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy); |
2777 | if (err < 0) | |
2778 | goto errout; | |
1da177e4 | 2779 | |
ab364a6f | 2780 | err = -EINVAL; |
4c9483b2 | 2781 | memset(&fl6, 0, sizeof(fl6)); |
1da177e4 | 2782 | |
ab364a6f TG |
2783 | if (tb[RTA_SRC]) { |
2784 | if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr)) | |
2785 | goto errout; | |
2786 | ||
4e3fd7a0 | 2787 | fl6.saddr = *(struct in6_addr *)nla_data(tb[RTA_SRC]); |
ab364a6f TG |
2788 | } |
2789 | ||
2790 | if (tb[RTA_DST]) { | |
2791 | if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr)) | |
2792 | goto errout; | |
2793 | ||
4e3fd7a0 | 2794 | fl6.daddr = *(struct in6_addr *)nla_data(tb[RTA_DST]); |
ab364a6f TG |
2795 | } |
2796 | ||
2797 | if (tb[RTA_IIF]) | |
2798 | iif = nla_get_u32(tb[RTA_IIF]); | |
2799 | ||
2800 | if (tb[RTA_OIF]) | |
72331bc0 | 2801 | oif = nla_get_u32(tb[RTA_OIF]); |
1da177e4 | 2802 | |
2e47b291 LC |
2803 | if (tb[RTA_MARK]) |
2804 | fl6.flowi6_mark = nla_get_u32(tb[RTA_MARK]); | |
2805 | ||
1da177e4 LT |
2806 | if (iif) { |
2807 | struct net_device *dev; | |
72331bc0 SL |
2808 | int flags = 0; |
2809 | ||
5578689a | 2810 | dev = __dev_get_by_index(net, iif); |
1da177e4 LT |
2811 | if (!dev) { |
2812 | err = -ENODEV; | |
ab364a6f | 2813 | goto errout; |
1da177e4 | 2814 | } |
72331bc0 SL |
2815 | |
2816 | fl6.flowi6_iif = iif; | |
2817 | ||
2818 | if (!ipv6_addr_any(&fl6.saddr)) | |
2819 | flags |= RT6_LOOKUP_F_HAS_SADDR; | |
2820 | ||
2821 | rt = (struct rt6_info *)ip6_route_input_lookup(net, dev, &fl6, | |
2822 | flags); | |
2823 | } else { | |
2824 | fl6.flowi6_oif = oif; | |
2825 | ||
2826 | rt = (struct rt6_info *)ip6_route_output(net, NULL, &fl6); | |
1da177e4 LT |
2827 | } |
2828 | ||
ab364a6f | 2829 | skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL); |
38308473 | 2830 | if (!skb) { |
94e187c0 | 2831 | ip6_rt_put(rt); |
ab364a6f TG |
2832 | err = -ENOBUFS; |
2833 | goto errout; | |
2834 | } | |
1da177e4 | 2835 | |
ab364a6f TG |
2836 | /* Reserve room for dummy headers, this skb can pass |
2837 | through good chunk of routing engine. | |
2838 | */ | |
459a98ed | 2839 | skb_reset_mac_header(skb); |
ab364a6f | 2840 | skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr)); |
1da177e4 | 2841 | |
d8d1f30b | 2842 | skb_dst_set(skb, &rt->dst); |
1da177e4 | 2843 | |
4c9483b2 | 2844 | err = rt6_fill_node(net, skb, rt, &fl6.daddr, &fl6.saddr, iif, |
15e47304 | 2845 | RTM_NEWROUTE, NETLINK_CB(in_skb).portid, |
7bc570c8 | 2846 | nlh->nlmsg_seq, 0, 0, 0); |
1da177e4 | 2847 | if (err < 0) { |
ab364a6f TG |
2848 | kfree_skb(skb); |
2849 | goto errout; | |
1da177e4 LT |
2850 | } |
2851 | ||
15e47304 | 2852 | err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid); |
ab364a6f | 2853 | errout: |
1da177e4 | 2854 | return err; |
1da177e4 LT |
2855 | } |
2856 | ||
86872cb5 | 2857 | void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info) |
1da177e4 LT |
2858 | { |
2859 | struct sk_buff *skb; | |
5578689a | 2860 | struct net *net = info->nl_net; |
528c4ceb DL |
2861 | u32 seq; |
2862 | int err; | |
2863 | ||
2864 | err = -ENOBUFS; | |
38308473 | 2865 | seq = info->nlh ? info->nlh->nlmsg_seq : 0; |
86872cb5 | 2866 | |
339bf98f | 2867 | skb = nlmsg_new(rt6_nlmsg_size(), gfp_any()); |
38308473 | 2868 | if (!skb) |
21713ebc TG |
2869 | goto errout; |
2870 | ||
191cd582 | 2871 | err = rt6_fill_node(net, skb, rt, NULL, NULL, 0, |
15e47304 | 2872 | event, info->portid, seq, 0, 0, 0); |
26932566 PM |
2873 | if (err < 0) { |
2874 | /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */ | |
2875 | WARN_ON(err == -EMSGSIZE); | |
2876 | kfree_skb(skb); | |
2877 | goto errout; | |
2878 | } | |
15e47304 | 2879 | rtnl_notify(skb, net, info->portid, RTNLGRP_IPV6_ROUTE, |
1ce85fe4 PNA |
2880 | info->nlh, gfp_any()); |
2881 | return; | |
21713ebc TG |
2882 | errout: |
2883 | if (err < 0) | |
5578689a | 2884 | rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err); |
1da177e4 LT |
2885 | } |
2886 | ||
8ed67789 | 2887 | static int ip6_route_dev_notify(struct notifier_block *this, |
351638e7 | 2888 | unsigned long event, void *ptr) |
8ed67789 | 2889 | { |
351638e7 | 2890 | struct net_device *dev = netdev_notifier_info_to_dev(ptr); |
c346dca1 | 2891 | struct net *net = dev_net(dev); |
8ed67789 DL |
2892 | |
2893 | if (event == NETDEV_REGISTER && (dev->flags & IFF_LOOPBACK)) { | |
d8d1f30b | 2894 | net->ipv6.ip6_null_entry->dst.dev = dev; |
8ed67789 DL |
2895 | net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev); |
2896 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES | |
d8d1f30b | 2897 | net->ipv6.ip6_prohibit_entry->dst.dev = dev; |
8ed67789 | 2898 | net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev); |
d8d1f30b | 2899 | net->ipv6.ip6_blk_hole_entry->dst.dev = dev; |
8ed67789 DL |
2900 | net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev); |
2901 | #endif | |
2902 | } | |
2903 | ||
2904 | return NOTIFY_OK; | |
2905 | } | |
2906 | ||
1da177e4 LT |
2907 | /* |
2908 | * /proc | |
2909 | */ | |
2910 | ||
2911 | #ifdef CONFIG_PROC_FS | |
2912 | ||
33120b30 AD |
2913 | static const struct file_operations ipv6_route_proc_fops = { |
2914 | .owner = THIS_MODULE, | |
2915 | .open = ipv6_route_open, | |
2916 | .read = seq_read, | |
2917 | .llseek = seq_lseek, | |
8d2ca1d7 | 2918 | .release = seq_release_net, |
33120b30 AD |
2919 | }; |
2920 | ||
1da177e4 LT |
2921 | static int rt6_stats_seq_show(struct seq_file *seq, void *v) |
2922 | { | |
69ddb805 | 2923 | struct net *net = (struct net *)seq->private; |
1da177e4 | 2924 | seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n", |
69ddb805 DL |
2925 | net->ipv6.rt6_stats->fib_nodes, |
2926 | net->ipv6.rt6_stats->fib_route_nodes, | |
2927 | net->ipv6.rt6_stats->fib_rt_alloc, | |
2928 | net->ipv6.rt6_stats->fib_rt_entries, | |
2929 | net->ipv6.rt6_stats->fib_rt_cache, | |
fc66f95c | 2930 | dst_entries_get_slow(&net->ipv6.ip6_dst_ops), |
69ddb805 | 2931 | net->ipv6.rt6_stats->fib_discarded_routes); |
1da177e4 LT |
2932 | |
2933 | return 0; | |
2934 | } | |
2935 | ||
2936 | static int rt6_stats_seq_open(struct inode *inode, struct file *file) | |
2937 | { | |
de05c557 | 2938 | return single_open_net(inode, file, rt6_stats_seq_show); |
69ddb805 DL |
2939 | } |
2940 | ||
9a32144e | 2941 | static const struct file_operations rt6_stats_seq_fops = { |
1da177e4 LT |
2942 | .owner = THIS_MODULE, |
2943 | .open = rt6_stats_seq_open, | |
2944 | .read = seq_read, | |
2945 | .llseek = seq_lseek, | |
b6fcbdb4 | 2946 | .release = single_release_net, |
1da177e4 LT |
2947 | }; |
2948 | #endif /* CONFIG_PROC_FS */ | |
2949 | ||
2950 | #ifdef CONFIG_SYSCTL | |
2951 | ||
1da177e4 | 2952 | static |
fe2c6338 | 2953 | int ipv6_sysctl_rtcache_flush(struct ctl_table *ctl, int write, |
1da177e4 LT |
2954 | void __user *buffer, size_t *lenp, loff_t *ppos) |
2955 | { | |
c486da34 LAG |
2956 | struct net *net; |
2957 | int delay; | |
2958 | if (!write) | |
1da177e4 | 2959 | return -EINVAL; |
c486da34 LAG |
2960 | |
2961 | net = (struct net *)ctl->extra1; | |
2962 | delay = net->ipv6.sysctl.flush_delay; | |
2963 | proc_dointvec(ctl, write, buffer, lenp, ppos); | |
2ac3ac8f | 2964 | fib6_run_gc(delay <= 0 ? 0 : (unsigned long)delay, net, delay > 0); |
c486da34 | 2965 | return 0; |
1da177e4 LT |
2966 | } |
2967 | ||
fe2c6338 | 2968 | struct ctl_table ipv6_route_table_template[] = { |
1ab1457c | 2969 | { |
1da177e4 | 2970 | .procname = "flush", |
4990509f | 2971 | .data = &init_net.ipv6.sysctl.flush_delay, |
1da177e4 | 2972 | .maxlen = sizeof(int), |
89c8b3a1 | 2973 | .mode = 0200, |
6d9f239a | 2974 | .proc_handler = ipv6_sysctl_rtcache_flush |
1da177e4 LT |
2975 | }, |
2976 | { | |
1da177e4 | 2977 | .procname = "gc_thresh", |
9a7ec3a9 | 2978 | .data = &ip6_dst_ops_template.gc_thresh, |
1da177e4 LT |
2979 | .maxlen = sizeof(int), |
2980 | .mode = 0644, | |
6d9f239a | 2981 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2982 | }, |
2983 | { | |
1da177e4 | 2984 | .procname = "max_size", |
4990509f | 2985 | .data = &init_net.ipv6.sysctl.ip6_rt_max_size, |
1da177e4 LT |
2986 | .maxlen = sizeof(int), |
2987 | .mode = 0644, | |
6d9f239a | 2988 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2989 | }, |
2990 | { | |
1da177e4 | 2991 | .procname = "gc_min_interval", |
4990509f | 2992 | .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval, |
1da177e4 LT |
2993 | .maxlen = sizeof(int), |
2994 | .mode = 0644, | |
6d9f239a | 2995 | .proc_handler = proc_dointvec_jiffies, |
1da177e4 LT |
2996 | }, |
2997 | { | |
1da177e4 | 2998 | .procname = "gc_timeout", |
4990509f | 2999 | .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout, |
1da177e4 LT |
3000 | .maxlen = sizeof(int), |
3001 | .mode = 0644, | |
6d9f239a | 3002 | .proc_handler = proc_dointvec_jiffies, |
1da177e4 LT |
3003 | }, |
3004 | { | |
1da177e4 | 3005 | .procname = "gc_interval", |
4990509f | 3006 | .data = &init_net.ipv6.sysctl.ip6_rt_gc_interval, |
1da177e4 LT |
3007 | .maxlen = sizeof(int), |
3008 | .mode = 0644, | |
6d9f239a | 3009 | .proc_handler = proc_dointvec_jiffies, |
1da177e4 LT |
3010 | }, |
3011 | { | |
1da177e4 | 3012 | .procname = "gc_elasticity", |
4990509f | 3013 | .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity, |
1da177e4 LT |
3014 | .maxlen = sizeof(int), |
3015 | .mode = 0644, | |
f3d3f616 | 3016 | .proc_handler = proc_dointvec, |
1da177e4 LT |
3017 | }, |
3018 | { | |
1da177e4 | 3019 | .procname = "mtu_expires", |
4990509f | 3020 | .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires, |
1da177e4 LT |
3021 | .maxlen = sizeof(int), |
3022 | .mode = 0644, | |
6d9f239a | 3023 | .proc_handler = proc_dointvec_jiffies, |
1da177e4 LT |
3024 | }, |
3025 | { | |
1da177e4 | 3026 | .procname = "min_adv_mss", |
4990509f | 3027 | .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss, |
1da177e4 LT |
3028 | .maxlen = sizeof(int), |
3029 | .mode = 0644, | |
f3d3f616 | 3030 | .proc_handler = proc_dointvec, |
1da177e4 LT |
3031 | }, |
3032 | { | |
1da177e4 | 3033 | .procname = "gc_min_interval_ms", |
4990509f | 3034 | .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval, |
1da177e4 LT |
3035 | .maxlen = sizeof(int), |
3036 | .mode = 0644, | |
6d9f239a | 3037 | .proc_handler = proc_dointvec_ms_jiffies, |
1da177e4 | 3038 | }, |
f8572d8f | 3039 | { } |
1da177e4 LT |
3040 | }; |
3041 | ||
2c8c1e72 | 3042 | struct ctl_table * __net_init ipv6_route_sysctl_init(struct net *net) |
760f2d01 DL |
3043 | { |
3044 | struct ctl_table *table; | |
3045 | ||
3046 | table = kmemdup(ipv6_route_table_template, | |
3047 | sizeof(ipv6_route_table_template), | |
3048 | GFP_KERNEL); | |
5ee09105 YH |
3049 | |
3050 | if (table) { | |
3051 | table[0].data = &net->ipv6.sysctl.flush_delay; | |
c486da34 | 3052 | table[0].extra1 = net; |
86393e52 | 3053 | table[1].data = &net->ipv6.ip6_dst_ops.gc_thresh; |
5ee09105 YH |
3054 | table[2].data = &net->ipv6.sysctl.ip6_rt_max_size; |
3055 | table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval; | |
3056 | table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout; | |
3057 | table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval; | |
3058 | table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity; | |
3059 | table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires; | |
3060 | table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss; | |
9c69fabe | 3061 | table[9].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval; |
464dc801 EB |
3062 | |
3063 | /* Don't export sysctls to unprivileged users */ | |
3064 | if (net->user_ns != &init_user_ns) | |
3065 | table[0].procname = NULL; | |
5ee09105 YH |
3066 | } |
3067 | ||
760f2d01 DL |
3068 | return table; |
3069 | } | |
1da177e4 LT |
3070 | #endif |
3071 | ||
2c8c1e72 | 3072 | static int __net_init ip6_route_net_init(struct net *net) |
cdb18761 | 3073 | { |
633d424b | 3074 | int ret = -ENOMEM; |
8ed67789 | 3075 | |
86393e52 AD |
3076 | memcpy(&net->ipv6.ip6_dst_ops, &ip6_dst_ops_template, |
3077 | sizeof(net->ipv6.ip6_dst_ops)); | |
f2fc6a54 | 3078 | |
fc66f95c ED |
3079 | if (dst_entries_init(&net->ipv6.ip6_dst_ops) < 0) |
3080 | goto out_ip6_dst_ops; | |
3081 | ||
8ed67789 DL |
3082 | net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template, |
3083 | sizeof(*net->ipv6.ip6_null_entry), | |
3084 | GFP_KERNEL); | |
3085 | if (!net->ipv6.ip6_null_entry) | |
fc66f95c | 3086 | goto out_ip6_dst_entries; |
d8d1f30b | 3087 | net->ipv6.ip6_null_entry->dst.path = |
8ed67789 | 3088 | (struct dst_entry *)net->ipv6.ip6_null_entry; |
d8d1f30b | 3089 | net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops; |
62fa8a84 DM |
3090 | dst_init_metrics(&net->ipv6.ip6_null_entry->dst, |
3091 | ip6_template_metrics, true); | |
8ed67789 DL |
3092 | |
3093 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES | |
3094 | net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template, | |
3095 | sizeof(*net->ipv6.ip6_prohibit_entry), | |
3096 | GFP_KERNEL); | |
68fffc67 PZ |
3097 | if (!net->ipv6.ip6_prohibit_entry) |
3098 | goto out_ip6_null_entry; | |
d8d1f30b | 3099 | net->ipv6.ip6_prohibit_entry->dst.path = |
8ed67789 | 3100 | (struct dst_entry *)net->ipv6.ip6_prohibit_entry; |
d8d1f30b | 3101 | net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops; |
62fa8a84 DM |
3102 | dst_init_metrics(&net->ipv6.ip6_prohibit_entry->dst, |
3103 | ip6_template_metrics, true); | |
8ed67789 DL |
3104 | |
3105 | net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template, | |
3106 | sizeof(*net->ipv6.ip6_blk_hole_entry), | |
3107 | GFP_KERNEL); | |
68fffc67 PZ |
3108 | if (!net->ipv6.ip6_blk_hole_entry) |
3109 | goto out_ip6_prohibit_entry; | |
d8d1f30b | 3110 | net->ipv6.ip6_blk_hole_entry->dst.path = |
8ed67789 | 3111 | (struct dst_entry *)net->ipv6.ip6_blk_hole_entry; |
d8d1f30b | 3112 | net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops; |
62fa8a84 DM |
3113 | dst_init_metrics(&net->ipv6.ip6_blk_hole_entry->dst, |
3114 | ip6_template_metrics, true); | |
8ed67789 DL |
3115 | #endif |
3116 | ||
b339a47c PZ |
3117 | net->ipv6.sysctl.flush_delay = 0; |
3118 | net->ipv6.sysctl.ip6_rt_max_size = 4096; | |
3119 | net->ipv6.sysctl.ip6_rt_gc_min_interval = HZ / 2; | |
3120 | net->ipv6.sysctl.ip6_rt_gc_timeout = 60*HZ; | |
3121 | net->ipv6.sysctl.ip6_rt_gc_interval = 30*HZ; | |
3122 | net->ipv6.sysctl.ip6_rt_gc_elasticity = 9; | |
3123 | net->ipv6.sysctl.ip6_rt_mtu_expires = 10*60*HZ; | |
3124 | net->ipv6.sysctl.ip6_rt_min_advmss = IPV6_MIN_MTU - 20 - 40; | |
3125 | ||
6891a346 BT |
3126 | net->ipv6.ip6_rt_gc_expire = 30*HZ; |
3127 | ||
8ed67789 DL |
3128 | ret = 0; |
3129 | out: | |
3130 | return ret; | |
f2fc6a54 | 3131 | |
68fffc67 PZ |
3132 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES |
3133 | out_ip6_prohibit_entry: | |
3134 | kfree(net->ipv6.ip6_prohibit_entry); | |
3135 | out_ip6_null_entry: | |
3136 | kfree(net->ipv6.ip6_null_entry); | |
3137 | #endif | |
fc66f95c ED |
3138 | out_ip6_dst_entries: |
3139 | dst_entries_destroy(&net->ipv6.ip6_dst_ops); | |
f2fc6a54 | 3140 | out_ip6_dst_ops: |
f2fc6a54 | 3141 | goto out; |
cdb18761 DL |
3142 | } |
3143 | ||
2c8c1e72 | 3144 | static void __net_exit ip6_route_net_exit(struct net *net) |
cdb18761 | 3145 | { |
8ed67789 DL |
3146 | kfree(net->ipv6.ip6_null_entry); |
3147 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES | |
3148 | kfree(net->ipv6.ip6_prohibit_entry); | |
3149 | kfree(net->ipv6.ip6_blk_hole_entry); | |
3150 | #endif | |
41bb78b4 | 3151 | dst_entries_destroy(&net->ipv6.ip6_dst_ops); |
cdb18761 DL |
3152 | } |
3153 | ||
d189634e TG |
3154 | static int __net_init ip6_route_net_init_late(struct net *net) |
3155 | { | |
3156 | #ifdef CONFIG_PROC_FS | |
d4beaa66 G |
3157 | proc_create("ipv6_route", 0, net->proc_net, &ipv6_route_proc_fops); |
3158 | proc_create("rt6_stats", S_IRUGO, net->proc_net, &rt6_stats_seq_fops); | |
d189634e TG |
3159 | #endif |
3160 | return 0; | |
3161 | } | |
3162 | ||
3163 | static void __net_exit ip6_route_net_exit_late(struct net *net) | |
3164 | { | |
3165 | #ifdef CONFIG_PROC_FS | |
ece31ffd G |
3166 | remove_proc_entry("ipv6_route", net->proc_net); |
3167 | remove_proc_entry("rt6_stats", net->proc_net); | |
d189634e TG |
3168 | #endif |
3169 | } | |
3170 | ||
cdb18761 DL |
3171 | static struct pernet_operations ip6_route_net_ops = { |
3172 | .init = ip6_route_net_init, | |
3173 | .exit = ip6_route_net_exit, | |
3174 | }; | |
3175 | ||
c3426b47 DM |
3176 | static int __net_init ipv6_inetpeer_init(struct net *net) |
3177 | { | |
3178 | struct inet_peer_base *bp = kmalloc(sizeof(*bp), GFP_KERNEL); | |
3179 | ||
3180 | if (!bp) | |
3181 | return -ENOMEM; | |
3182 | inet_peer_base_init(bp); | |
3183 | net->ipv6.peers = bp; | |
3184 | return 0; | |
3185 | } | |
3186 | ||
3187 | static void __net_exit ipv6_inetpeer_exit(struct net *net) | |
3188 | { | |
3189 | struct inet_peer_base *bp = net->ipv6.peers; | |
3190 | ||
3191 | net->ipv6.peers = NULL; | |
56a6b248 | 3192 | inetpeer_invalidate_tree(bp); |
c3426b47 DM |
3193 | kfree(bp); |
3194 | } | |
3195 | ||
2b823f72 | 3196 | static struct pernet_operations ipv6_inetpeer_ops = { |
c3426b47 DM |
3197 | .init = ipv6_inetpeer_init, |
3198 | .exit = ipv6_inetpeer_exit, | |
3199 | }; | |
3200 | ||
d189634e TG |
3201 | static struct pernet_operations ip6_route_net_late_ops = { |
3202 | .init = ip6_route_net_init_late, | |
3203 | .exit = ip6_route_net_exit_late, | |
3204 | }; | |
3205 | ||
8ed67789 DL |
3206 | static struct notifier_block ip6_route_dev_notifier = { |
3207 | .notifier_call = ip6_route_dev_notify, | |
3208 | .priority = 0, | |
3209 | }; | |
3210 | ||
433d49c3 | 3211 | int __init ip6_route_init(void) |
1da177e4 | 3212 | { |
433d49c3 DL |
3213 | int ret; |
3214 | ||
9a7ec3a9 DL |
3215 | ret = -ENOMEM; |
3216 | ip6_dst_ops_template.kmem_cachep = | |
e5d679f3 | 3217 | kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0, |
f845ab6b | 3218 | SLAB_HWCACHE_ALIGN, NULL); |
9a7ec3a9 | 3219 | if (!ip6_dst_ops_template.kmem_cachep) |
c19a28e1 | 3220 | goto out; |
14e50e57 | 3221 | |
fc66f95c | 3222 | ret = dst_entries_init(&ip6_dst_blackhole_ops); |
8ed67789 | 3223 | if (ret) |
bdb3289f | 3224 | goto out_kmem_cache; |
bdb3289f | 3225 | |
c3426b47 DM |
3226 | ret = register_pernet_subsys(&ipv6_inetpeer_ops); |
3227 | if (ret) | |
e8803b6c | 3228 | goto out_dst_entries; |
2a0c451a | 3229 | |
7e52b33b DM |
3230 | ret = register_pernet_subsys(&ip6_route_net_ops); |
3231 | if (ret) | |
3232 | goto out_register_inetpeer; | |
c3426b47 | 3233 | |
5dc121e9 AE |
3234 | ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep; |
3235 | ||
8ed67789 DL |
3236 | /* Registering of the loopback is done before this portion of code, |
3237 | * the loopback reference in rt6_info will not be taken, do it | |
3238 | * manually for init_net */ | |
d8d1f30b | 3239 | init_net.ipv6.ip6_null_entry->dst.dev = init_net.loopback_dev; |
8ed67789 DL |
3240 | init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev); |
3241 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES | |
d8d1f30b | 3242 | init_net.ipv6.ip6_prohibit_entry->dst.dev = init_net.loopback_dev; |
8ed67789 | 3243 | init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev); |
d8d1f30b | 3244 | init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev; |
8ed67789 DL |
3245 | init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev); |
3246 | #endif | |
e8803b6c | 3247 | ret = fib6_init(); |
433d49c3 | 3248 | if (ret) |
8ed67789 | 3249 | goto out_register_subsys; |
433d49c3 | 3250 | |
433d49c3 DL |
3251 | ret = xfrm6_init(); |
3252 | if (ret) | |
e8803b6c | 3253 | goto out_fib6_init; |
c35b7e72 | 3254 | |
433d49c3 DL |
3255 | ret = fib6_rules_init(); |
3256 | if (ret) | |
3257 | goto xfrm6_init; | |
7e5449c2 | 3258 | |
d189634e TG |
3259 | ret = register_pernet_subsys(&ip6_route_net_late_ops); |
3260 | if (ret) | |
3261 | goto fib6_rules_init; | |
3262 | ||
433d49c3 | 3263 | ret = -ENOBUFS; |
c7ac8679 GR |
3264 | if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL, NULL) || |
3265 | __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL, NULL) || | |
3266 | __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL, NULL)) | |
d189634e | 3267 | goto out_register_late_subsys; |
c127ea2c | 3268 | |
8ed67789 | 3269 | ret = register_netdevice_notifier(&ip6_route_dev_notifier); |
cdb18761 | 3270 | if (ret) |
d189634e | 3271 | goto out_register_late_subsys; |
8ed67789 | 3272 | |
433d49c3 DL |
3273 | out: |
3274 | return ret; | |
3275 | ||
d189634e TG |
3276 | out_register_late_subsys: |
3277 | unregister_pernet_subsys(&ip6_route_net_late_ops); | |
433d49c3 | 3278 | fib6_rules_init: |
433d49c3 DL |
3279 | fib6_rules_cleanup(); |
3280 | xfrm6_init: | |
433d49c3 | 3281 | xfrm6_fini(); |
2a0c451a TG |
3282 | out_fib6_init: |
3283 | fib6_gc_cleanup(); | |
8ed67789 DL |
3284 | out_register_subsys: |
3285 | unregister_pernet_subsys(&ip6_route_net_ops); | |
7e52b33b DM |
3286 | out_register_inetpeer: |
3287 | unregister_pernet_subsys(&ipv6_inetpeer_ops); | |
fc66f95c ED |
3288 | out_dst_entries: |
3289 | dst_entries_destroy(&ip6_dst_blackhole_ops); | |
433d49c3 | 3290 | out_kmem_cache: |
f2fc6a54 | 3291 | kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep); |
433d49c3 | 3292 | goto out; |
1da177e4 LT |
3293 | } |
3294 | ||
3295 | void ip6_route_cleanup(void) | |
3296 | { | |
8ed67789 | 3297 | unregister_netdevice_notifier(&ip6_route_dev_notifier); |
d189634e | 3298 | unregister_pernet_subsys(&ip6_route_net_late_ops); |
101367c2 | 3299 | fib6_rules_cleanup(); |
1da177e4 | 3300 | xfrm6_fini(); |
1da177e4 | 3301 | fib6_gc_cleanup(); |
c3426b47 | 3302 | unregister_pernet_subsys(&ipv6_inetpeer_ops); |
8ed67789 | 3303 | unregister_pernet_subsys(&ip6_route_net_ops); |
41bb78b4 | 3304 | dst_entries_destroy(&ip6_dst_blackhole_ops); |
f2fc6a54 | 3305 | kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep); |
1da177e4 | 3306 | } |