Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
2 | * Linux INET6 implementation | |
3 | * FIB front-end. | |
4 | * | |
5 | * Authors: | |
1ab1457c | 6 | * Pedro Roque <roque@di.fc.ul.pt> |
1da177e4 | 7 | * |
1da177e4 LT |
8 | * This program is free software; you can redistribute it and/or |
9 | * modify it under the terms of the GNU General Public License | |
10 | * as published by the Free Software Foundation; either version | |
11 | * 2 of the License, or (at your option) any later version. | |
12 | */ | |
13 | ||
14 | /* Changes: | |
15 | * | |
16 | * YOSHIFUJI Hideaki @USAGI | |
17 | * reworked default router selection. | |
18 | * - respect outgoing interface | |
19 | * - select from (probably) reachable routers (i.e. | |
20 | * routers in REACHABLE, STALE, DELAY or PROBE states). | |
21 | * - always select the same router if it is (probably) | |
22 | * reachable. otherwise, round-robin the list. | |
c0bece9f YH |
23 | * Ville Nuorvala |
24 | * Fixed routing subtrees. | |
1da177e4 LT |
25 | */ |
26 | ||
4fc268d2 | 27 | #include <linux/capability.h> |
1da177e4 LT |
28 | #include <linux/errno.h> |
29 | #include <linux/types.h> | |
30 | #include <linux/times.h> | |
31 | #include <linux/socket.h> | |
32 | #include <linux/sockios.h> | |
33 | #include <linux/net.h> | |
34 | #include <linux/route.h> | |
35 | #include <linux/netdevice.h> | |
36 | #include <linux/in6.h> | |
7bc570c8 | 37 | #include <linux/mroute6.h> |
1da177e4 | 38 | #include <linux/init.h> |
1da177e4 | 39 | #include <linux/if_arp.h> |
1da177e4 LT |
40 | #include <linux/proc_fs.h> |
41 | #include <linux/seq_file.h> | |
5b7c931d | 42 | #include <linux/nsproxy.h> |
5a0e3ad6 | 43 | #include <linux/slab.h> |
457c4cbc | 44 | #include <net/net_namespace.h> |
1da177e4 LT |
45 | #include <net/snmp.h> |
46 | #include <net/ipv6.h> | |
47 | #include <net/ip6_fib.h> | |
48 | #include <net/ip6_route.h> | |
49 | #include <net/ndisc.h> | |
50 | #include <net/addrconf.h> | |
51 | #include <net/tcp.h> | |
52 | #include <linux/rtnetlink.h> | |
53 | #include <net/dst.h> | |
54 | #include <net/xfrm.h> | |
8d71740c | 55 | #include <net/netevent.h> |
21713ebc | 56 | #include <net/netlink.h> |
1da177e4 LT |
57 | |
58 | #include <asm/uaccess.h> | |
59 | ||
60 | #ifdef CONFIG_SYSCTL | |
61 | #include <linux/sysctl.h> | |
62 | #endif | |
63 | ||
64 | /* Set to 3 to get tracing. */ | |
65 | #define RT6_DEBUG 2 | |
66 | ||
67 | #if RT6_DEBUG >= 3 | |
68 | #define RDBG(x) printk x | |
69 | #define RT6_TRACE(x...) printk(KERN_DEBUG x) | |
70 | #else | |
71 | #define RDBG(x) | |
72 | #define RT6_TRACE(x...) do { ; } while (0) | |
73 | #endif | |
74 | ||
519fbd87 | 75 | #define CLONE_OFFLINK_ROUTE 0 |
1da177e4 | 76 | |
1da177e4 LT |
77 | static struct rt6_info * ip6_rt_copy(struct rt6_info *ort); |
78 | static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie); | |
79 | static struct dst_entry *ip6_negative_advice(struct dst_entry *); | |
80 | static void ip6_dst_destroy(struct dst_entry *); | |
81 | static void ip6_dst_ifdown(struct dst_entry *, | |
82 | struct net_device *dev, int how); | |
569d3645 | 83 | static int ip6_dst_gc(struct dst_ops *ops); |
1da177e4 LT |
84 | |
85 | static int ip6_pkt_discard(struct sk_buff *skb); | |
86 | static int ip6_pkt_discard_out(struct sk_buff *skb); | |
87 | static void ip6_link_failure(struct sk_buff *skb); | |
88 | static void ip6_rt_update_pmtu(struct dst_entry *dst, u32 mtu); | |
89 | ||
70ceb4f5 | 90 | #ifdef CONFIG_IPV6_ROUTE_INFO |
efa2cea0 DL |
91 | static struct rt6_info *rt6_add_route_info(struct net *net, |
92 | struct in6_addr *prefix, int prefixlen, | |
70ceb4f5 YH |
93 | struct in6_addr *gwaddr, int ifindex, |
94 | unsigned pref); | |
efa2cea0 DL |
95 | static struct rt6_info *rt6_get_route_info(struct net *net, |
96 | struct in6_addr *prefix, int prefixlen, | |
70ceb4f5 YH |
97 | struct in6_addr *gwaddr, int ifindex); |
98 | #endif | |
99 | ||
9a7ec3a9 | 100 | static struct dst_ops ip6_dst_ops_template = { |
1da177e4 | 101 | .family = AF_INET6, |
09640e63 | 102 | .protocol = cpu_to_be16(ETH_P_IPV6), |
1da177e4 LT |
103 | .gc = ip6_dst_gc, |
104 | .gc_thresh = 1024, | |
105 | .check = ip6_dst_check, | |
106 | .destroy = ip6_dst_destroy, | |
107 | .ifdown = ip6_dst_ifdown, | |
108 | .negative_advice = ip6_negative_advice, | |
109 | .link_failure = ip6_link_failure, | |
110 | .update_pmtu = ip6_rt_update_pmtu, | |
1ac06e03 | 111 | .local_out = __ip6_local_out, |
1da177e4 LT |
112 | }; |
113 | ||
14e50e57 DM |
114 | static void ip6_rt_blackhole_update_pmtu(struct dst_entry *dst, u32 mtu) |
115 | { | |
116 | } | |
117 | ||
118 | static struct dst_ops ip6_dst_blackhole_ops = { | |
119 | .family = AF_INET6, | |
09640e63 | 120 | .protocol = cpu_to_be16(ETH_P_IPV6), |
14e50e57 DM |
121 | .destroy = ip6_dst_destroy, |
122 | .check = ip6_dst_check, | |
123 | .update_pmtu = ip6_rt_blackhole_update_pmtu, | |
14e50e57 DM |
124 | }; |
125 | ||
bdb3289f | 126 | static struct rt6_info ip6_null_entry_template = { |
d8d1f30b CG |
127 | .dst = { |
128 | .__refcnt = ATOMIC_INIT(1), | |
129 | .__use = 1, | |
130 | .obsolete = -1, | |
131 | .error = -ENETUNREACH, | |
d8d1f30b CG |
132 | .input = ip6_pkt_discard, |
133 | .output = ip6_pkt_discard_out, | |
1da177e4 LT |
134 | }, |
135 | .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP), | |
4f724279 | 136 | .rt6i_protocol = RTPROT_KERNEL, |
1da177e4 LT |
137 | .rt6i_metric = ~(u32) 0, |
138 | .rt6i_ref = ATOMIC_INIT(1), | |
139 | }; | |
140 | ||
101367c2 TG |
141 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES |
142 | ||
6723ab54 DM |
143 | static int ip6_pkt_prohibit(struct sk_buff *skb); |
144 | static int ip6_pkt_prohibit_out(struct sk_buff *skb); | |
6723ab54 | 145 | |
280a34c8 | 146 | static struct rt6_info ip6_prohibit_entry_template = { |
d8d1f30b CG |
147 | .dst = { |
148 | .__refcnt = ATOMIC_INIT(1), | |
149 | .__use = 1, | |
150 | .obsolete = -1, | |
151 | .error = -EACCES, | |
d8d1f30b CG |
152 | .input = ip6_pkt_prohibit, |
153 | .output = ip6_pkt_prohibit_out, | |
101367c2 TG |
154 | }, |
155 | .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP), | |
4f724279 | 156 | .rt6i_protocol = RTPROT_KERNEL, |
101367c2 TG |
157 | .rt6i_metric = ~(u32) 0, |
158 | .rt6i_ref = ATOMIC_INIT(1), | |
159 | }; | |
160 | ||
bdb3289f | 161 | static struct rt6_info ip6_blk_hole_entry_template = { |
d8d1f30b CG |
162 | .dst = { |
163 | .__refcnt = ATOMIC_INIT(1), | |
164 | .__use = 1, | |
165 | .obsolete = -1, | |
166 | .error = -EINVAL, | |
d8d1f30b CG |
167 | .input = dst_discard, |
168 | .output = dst_discard, | |
101367c2 TG |
169 | }, |
170 | .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP), | |
4f724279 | 171 | .rt6i_protocol = RTPROT_KERNEL, |
101367c2 TG |
172 | .rt6i_metric = ~(u32) 0, |
173 | .rt6i_ref = ATOMIC_INIT(1), | |
174 | }; | |
175 | ||
176 | #endif | |
177 | ||
1da177e4 | 178 | /* allocate dst with ip6_dst_ops */ |
f2fc6a54 | 179 | static inline struct rt6_info *ip6_dst_alloc(struct dst_ops *ops) |
1da177e4 | 180 | { |
f2fc6a54 | 181 | return (struct rt6_info *)dst_alloc(ops); |
1da177e4 LT |
182 | } |
183 | ||
184 | static void ip6_dst_destroy(struct dst_entry *dst) | |
185 | { | |
186 | struct rt6_info *rt = (struct rt6_info *)dst; | |
187 | struct inet6_dev *idev = rt->rt6i_idev; | |
b3419363 | 188 | struct inet_peer *peer = rt->rt6i_peer; |
1da177e4 LT |
189 | |
190 | if (idev != NULL) { | |
191 | rt->rt6i_idev = NULL; | |
192 | in6_dev_put(idev); | |
1ab1457c | 193 | } |
b3419363 DM |
194 | if (peer) { |
195 | BUG_ON(!(rt->rt6i_flags & RTF_CACHE)); | |
196 | rt->rt6i_peer = NULL; | |
197 | inet_putpeer(peer); | |
198 | } | |
199 | } | |
200 | ||
201 | void rt6_bind_peer(struct rt6_info *rt, int create) | |
202 | { | |
203 | struct inet_peer *peer; | |
204 | ||
205 | if (WARN_ON(!(rt->rt6i_flags & RTF_CACHE))) | |
206 | return; | |
207 | ||
208 | peer = inet_getpeer_v6(&rt->rt6i_dst.addr, create); | |
209 | if (peer && cmpxchg(&rt->rt6i_peer, NULL, peer) != NULL) | |
210 | inet_putpeer(peer); | |
1da177e4 LT |
211 | } |
212 | ||
213 | static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev, | |
214 | int how) | |
215 | { | |
216 | struct rt6_info *rt = (struct rt6_info *)dst; | |
217 | struct inet6_dev *idev = rt->rt6i_idev; | |
5a3e55d6 | 218 | struct net_device *loopback_dev = |
c346dca1 | 219 | dev_net(dev)->loopback_dev; |
1da177e4 | 220 | |
5a3e55d6 DL |
221 | if (dev != loopback_dev && idev != NULL && idev->dev == dev) { |
222 | struct inet6_dev *loopback_idev = | |
223 | in6_dev_get(loopback_dev); | |
1da177e4 LT |
224 | if (loopback_idev != NULL) { |
225 | rt->rt6i_idev = loopback_idev; | |
226 | in6_dev_put(idev); | |
227 | } | |
228 | } | |
229 | } | |
230 | ||
231 | static __inline__ int rt6_check_expired(const struct rt6_info *rt) | |
232 | { | |
a02cec21 ED |
233 | return (rt->rt6i_flags & RTF_EXPIRES) && |
234 | time_after(jiffies, rt->rt6i_expires); | |
1da177e4 LT |
235 | } |
236 | ||
c71099ac TG |
237 | static inline int rt6_need_strict(struct in6_addr *daddr) |
238 | { | |
a02cec21 ED |
239 | return ipv6_addr_type(daddr) & |
240 | (IPV6_ADDR_MULTICAST | IPV6_ADDR_LINKLOCAL | IPV6_ADDR_LOOPBACK); | |
c71099ac TG |
241 | } |
242 | ||
1da177e4 | 243 | /* |
c71099ac | 244 | * Route lookup. Any table->tb6_lock is implied. |
1da177e4 LT |
245 | */ |
246 | ||
8ed67789 DL |
247 | static inline struct rt6_info *rt6_device_match(struct net *net, |
248 | struct rt6_info *rt, | |
dd3abc4e | 249 | struct in6_addr *saddr, |
1da177e4 | 250 | int oif, |
d420895e | 251 | int flags) |
1da177e4 LT |
252 | { |
253 | struct rt6_info *local = NULL; | |
254 | struct rt6_info *sprt; | |
255 | ||
dd3abc4e YH |
256 | if (!oif && ipv6_addr_any(saddr)) |
257 | goto out; | |
258 | ||
d8d1f30b | 259 | for (sprt = rt; sprt; sprt = sprt->dst.rt6_next) { |
dd3abc4e YH |
260 | struct net_device *dev = sprt->rt6i_dev; |
261 | ||
262 | if (oif) { | |
1da177e4 LT |
263 | if (dev->ifindex == oif) |
264 | return sprt; | |
265 | if (dev->flags & IFF_LOOPBACK) { | |
266 | if (sprt->rt6i_idev == NULL || | |
267 | sprt->rt6i_idev->dev->ifindex != oif) { | |
d420895e | 268 | if (flags & RT6_LOOKUP_F_IFACE && oif) |
1da177e4 | 269 | continue; |
1ab1457c | 270 | if (local && (!oif || |
1da177e4 LT |
271 | local->rt6i_idev->dev->ifindex == oif)) |
272 | continue; | |
273 | } | |
274 | local = sprt; | |
275 | } | |
dd3abc4e YH |
276 | } else { |
277 | if (ipv6_chk_addr(net, saddr, dev, | |
278 | flags & RT6_LOOKUP_F_IFACE)) | |
279 | return sprt; | |
1da177e4 | 280 | } |
dd3abc4e | 281 | } |
1da177e4 | 282 | |
dd3abc4e | 283 | if (oif) { |
1da177e4 LT |
284 | if (local) |
285 | return local; | |
286 | ||
d420895e | 287 | if (flags & RT6_LOOKUP_F_IFACE) |
8ed67789 | 288 | return net->ipv6.ip6_null_entry; |
1da177e4 | 289 | } |
dd3abc4e | 290 | out: |
1da177e4 LT |
291 | return rt; |
292 | } | |
293 | ||
27097255 YH |
294 | #ifdef CONFIG_IPV6_ROUTER_PREF |
295 | static void rt6_probe(struct rt6_info *rt) | |
296 | { | |
297 | struct neighbour *neigh = rt ? rt->rt6i_nexthop : NULL; | |
298 | /* | |
299 | * Okay, this does not seem to be appropriate | |
300 | * for now, however, we need to check if it | |
301 | * is really so; aka Router Reachability Probing. | |
302 | * | |
303 | * Router Reachability Probe MUST be rate-limited | |
304 | * to no more than one per minute. | |
305 | */ | |
306 | if (!neigh || (neigh->nud_state & NUD_VALID)) | |
307 | return; | |
308 | read_lock_bh(&neigh->lock); | |
309 | if (!(neigh->nud_state & NUD_VALID) && | |
52e16356 | 310 | time_after(jiffies, neigh->updated + rt->rt6i_idev->cnf.rtr_probe_interval)) { |
27097255 YH |
311 | struct in6_addr mcaddr; |
312 | struct in6_addr *target; | |
313 | ||
314 | neigh->updated = jiffies; | |
315 | read_unlock_bh(&neigh->lock); | |
316 | ||
317 | target = (struct in6_addr *)&neigh->primary_key; | |
318 | addrconf_addr_solict_mult(target, &mcaddr); | |
319 | ndisc_send_ns(rt->rt6i_dev, NULL, target, &mcaddr, NULL); | |
320 | } else | |
321 | read_unlock_bh(&neigh->lock); | |
322 | } | |
323 | #else | |
324 | static inline void rt6_probe(struct rt6_info *rt) | |
325 | { | |
27097255 YH |
326 | } |
327 | #endif | |
328 | ||
1da177e4 | 329 | /* |
554cfb7e | 330 | * Default Router Selection (RFC 2461 6.3.6) |
1da177e4 | 331 | */ |
b6f99a21 | 332 | static inline int rt6_check_dev(struct rt6_info *rt, int oif) |
554cfb7e YH |
333 | { |
334 | struct net_device *dev = rt->rt6i_dev; | |
161980f4 | 335 | if (!oif || dev->ifindex == oif) |
554cfb7e | 336 | return 2; |
161980f4 DM |
337 | if ((dev->flags & IFF_LOOPBACK) && |
338 | rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif) | |
339 | return 1; | |
340 | return 0; | |
554cfb7e | 341 | } |
1da177e4 | 342 | |
b6f99a21 | 343 | static inline int rt6_check_neigh(struct rt6_info *rt) |
1da177e4 | 344 | { |
554cfb7e | 345 | struct neighbour *neigh = rt->rt6i_nexthop; |
398bcbeb | 346 | int m; |
4d0c5911 YH |
347 | if (rt->rt6i_flags & RTF_NONEXTHOP || |
348 | !(rt->rt6i_flags & RTF_GATEWAY)) | |
349 | m = 1; | |
350 | else if (neigh) { | |
554cfb7e YH |
351 | read_lock_bh(&neigh->lock); |
352 | if (neigh->nud_state & NUD_VALID) | |
4d0c5911 | 353 | m = 2; |
398bcbeb YH |
354 | #ifdef CONFIG_IPV6_ROUTER_PREF |
355 | else if (neigh->nud_state & NUD_FAILED) | |
356 | m = 0; | |
357 | #endif | |
358 | else | |
ea73ee23 | 359 | m = 1; |
554cfb7e | 360 | read_unlock_bh(&neigh->lock); |
398bcbeb YH |
361 | } else |
362 | m = 0; | |
554cfb7e | 363 | return m; |
1da177e4 LT |
364 | } |
365 | ||
554cfb7e YH |
366 | static int rt6_score_route(struct rt6_info *rt, int oif, |
367 | int strict) | |
1da177e4 | 368 | { |
4d0c5911 | 369 | int m, n; |
1ab1457c | 370 | |
4d0c5911 | 371 | m = rt6_check_dev(rt, oif); |
77d16f45 | 372 | if (!m && (strict & RT6_LOOKUP_F_IFACE)) |
554cfb7e | 373 | return -1; |
ebacaaa0 YH |
374 | #ifdef CONFIG_IPV6_ROUTER_PREF |
375 | m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2; | |
376 | #endif | |
4d0c5911 | 377 | n = rt6_check_neigh(rt); |
557e92ef | 378 | if (!n && (strict & RT6_LOOKUP_F_REACHABLE)) |
554cfb7e YH |
379 | return -1; |
380 | return m; | |
381 | } | |
382 | ||
f11e6659 DM |
383 | static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict, |
384 | int *mpri, struct rt6_info *match) | |
554cfb7e | 385 | { |
f11e6659 DM |
386 | int m; |
387 | ||
388 | if (rt6_check_expired(rt)) | |
389 | goto out; | |
390 | ||
391 | m = rt6_score_route(rt, oif, strict); | |
392 | if (m < 0) | |
393 | goto out; | |
394 | ||
395 | if (m > *mpri) { | |
396 | if (strict & RT6_LOOKUP_F_REACHABLE) | |
397 | rt6_probe(match); | |
398 | *mpri = m; | |
399 | match = rt; | |
400 | } else if (strict & RT6_LOOKUP_F_REACHABLE) { | |
401 | rt6_probe(rt); | |
402 | } | |
403 | ||
404 | out: | |
405 | return match; | |
406 | } | |
407 | ||
408 | static struct rt6_info *find_rr_leaf(struct fib6_node *fn, | |
409 | struct rt6_info *rr_head, | |
410 | u32 metric, int oif, int strict) | |
411 | { | |
412 | struct rt6_info *rt, *match; | |
554cfb7e | 413 | int mpri = -1; |
1da177e4 | 414 | |
f11e6659 DM |
415 | match = NULL; |
416 | for (rt = rr_head; rt && rt->rt6i_metric == metric; | |
d8d1f30b | 417 | rt = rt->dst.rt6_next) |
f11e6659 DM |
418 | match = find_match(rt, oif, strict, &mpri, match); |
419 | for (rt = fn->leaf; rt && rt != rr_head && rt->rt6i_metric == metric; | |
d8d1f30b | 420 | rt = rt->dst.rt6_next) |
f11e6659 | 421 | match = find_match(rt, oif, strict, &mpri, match); |
1da177e4 | 422 | |
f11e6659 DM |
423 | return match; |
424 | } | |
1da177e4 | 425 | |
f11e6659 DM |
426 | static struct rt6_info *rt6_select(struct fib6_node *fn, int oif, int strict) |
427 | { | |
428 | struct rt6_info *match, *rt0; | |
8ed67789 | 429 | struct net *net; |
1da177e4 | 430 | |
f11e6659 | 431 | RT6_TRACE("%s(fn->leaf=%p, oif=%d)\n", |
0dc47877 | 432 | __func__, fn->leaf, oif); |
554cfb7e | 433 | |
f11e6659 DM |
434 | rt0 = fn->rr_ptr; |
435 | if (!rt0) | |
436 | fn->rr_ptr = rt0 = fn->leaf; | |
1da177e4 | 437 | |
f11e6659 | 438 | match = find_rr_leaf(fn, rt0, rt0->rt6i_metric, oif, strict); |
1da177e4 | 439 | |
554cfb7e | 440 | if (!match && |
f11e6659 | 441 | (strict & RT6_LOOKUP_F_REACHABLE)) { |
d8d1f30b | 442 | struct rt6_info *next = rt0->dst.rt6_next; |
f11e6659 | 443 | |
554cfb7e | 444 | /* no entries matched; do round-robin */ |
f11e6659 DM |
445 | if (!next || next->rt6i_metric != rt0->rt6i_metric) |
446 | next = fn->leaf; | |
447 | ||
448 | if (next != rt0) | |
449 | fn->rr_ptr = next; | |
1da177e4 | 450 | } |
1da177e4 | 451 | |
f11e6659 | 452 | RT6_TRACE("%s() => %p\n", |
0dc47877 | 453 | __func__, match); |
1da177e4 | 454 | |
c346dca1 | 455 | net = dev_net(rt0->rt6i_dev); |
a02cec21 | 456 | return match ? match : net->ipv6.ip6_null_entry; |
1da177e4 LT |
457 | } |
458 | ||
70ceb4f5 YH |
459 | #ifdef CONFIG_IPV6_ROUTE_INFO |
460 | int rt6_route_rcv(struct net_device *dev, u8 *opt, int len, | |
461 | struct in6_addr *gwaddr) | |
462 | { | |
c346dca1 | 463 | struct net *net = dev_net(dev); |
70ceb4f5 YH |
464 | struct route_info *rinfo = (struct route_info *) opt; |
465 | struct in6_addr prefix_buf, *prefix; | |
466 | unsigned int pref; | |
4bed72e4 | 467 | unsigned long lifetime; |
70ceb4f5 YH |
468 | struct rt6_info *rt; |
469 | ||
470 | if (len < sizeof(struct route_info)) { | |
471 | return -EINVAL; | |
472 | } | |
473 | ||
474 | /* Sanity check for prefix_len and length */ | |
475 | if (rinfo->length > 3) { | |
476 | return -EINVAL; | |
477 | } else if (rinfo->prefix_len > 128) { | |
478 | return -EINVAL; | |
479 | } else if (rinfo->prefix_len > 64) { | |
480 | if (rinfo->length < 2) { | |
481 | return -EINVAL; | |
482 | } | |
483 | } else if (rinfo->prefix_len > 0) { | |
484 | if (rinfo->length < 1) { | |
485 | return -EINVAL; | |
486 | } | |
487 | } | |
488 | ||
489 | pref = rinfo->route_pref; | |
490 | if (pref == ICMPV6_ROUTER_PREF_INVALID) | |
3933fc95 | 491 | return -EINVAL; |
70ceb4f5 | 492 | |
4bed72e4 | 493 | lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ); |
70ceb4f5 YH |
494 | |
495 | if (rinfo->length == 3) | |
496 | prefix = (struct in6_addr *)rinfo->prefix; | |
497 | else { | |
498 | /* this function is safe */ | |
499 | ipv6_addr_prefix(&prefix_buf, | |
500 | (struct in6_addr *)rinfo->prefix, | |
501 | rinfo->prefix_len); | |
502 | prefix = &prefix_buf; | |
503 | } | |
504 | ||
efa2cea0 DL |
505 | rt = rt6_get_route_info(net, prefix, rinfo->prefix_len, gwaddr, |
506 | dev->ifindex); | |
70ceb4f5 YH |
507 | |
508 | if (rt && !lifetime) { | |
e0a1ad73 | 509 | ip6_del_rt(rt); |
70ceb4f5 YH |
510 | rt = NULL; |
511 | } | |
512 | ||
513 | if (!rt && lifetime) | |
efa2cea0 | 514 | rt = rt6_add_route_info(net, prefix, rinfo->prefix_len, gwaddr, dev->ifindex, |
70ceb4f5 YH |
515 | pref); |
516 | else if (rt) | |
517 | rt->rt6i_flags = RTF_ROUTEINFO | | |
518 | (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref); | |
519 | ||
520 | if (rt) { | |
4bed72e4 | 521 | if (!addrconf_finite_timeout(lifetime)) { |
70ceb4f5 YH |
522 | rt->rt6i_flags &= ~RTF_EXPIRES; |
523 | } else { | |
524 | rt->rt6i_expires = jiffies + HZ * lifetime; | |
525 | rt->rt6i_flags |= RTF_EXPIRES; | |
526 | } | |
d8d1f30b | 527 | dst_release(&rt->dst); |
70ceb4f5 YH |
528 | } |
529 | return 0; | |
530 | } | |
531 | #endif | |
532 | ||
8ed67789 | 533 | #define BACKTRACK(__net, saddr) \ |
982f56f3 | 534 | do { \ |
8ed67789 | 535 | if (rt == __net->ipv6.ip6_null_entry) { \ |
982f56f3 | 536 | struct fib6_node *pn; \ |
e0eda7bb | 537 | while (1) { \ |
982f56f3 YH |
538 | if (fn->fn_flags & RTN_TL_ROOT) \ |
539 | goto out; \ | |
540 | pn = fn->parent; \ | |
541 | if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn) \ | |
8bce65b9 | 542 | fn = fib6_lookup(FIB6_SUBTREE(pn), NULL, saddr); \ |
982f56f3 YH |
543 | else \ |
544 | fn = pn; \ | |
545 | if (fn->fn_flags & RTN_RTINFO) \ | |
546 | goto restart; \ | |
c71099ac | 547 | } \ |
c71099ac | 548 | } \ |
982f56f3 | 549 | } while(0) |
c71099ac | 550 | |
8ed67789 DL |
551 | static struct rt6_info *ip6_pol_route_lookup(struct net *net, |
552 | struct fib6_table *table, | |
c71099ac | 553 | struct flowi *fl, int flags) |
1da177e4 LT |
554 | { |
555 | struct fib6_node *fn; | |
556 | struct rt6_info *rt; | |
557 | ||
c71099ac TG |
558 | read_lock_bh(&table->tb6_lock); |
559 | fn = fib6_lookup(&table->tb6_root, &fl->fl6_dst, &fl->fl6_src); | |
560 | restart: | |
561 | rt = fn->leaf; | |
dd3abc4e | 562 | rt = rt6_device_match(net, rt, &fl->fl6_src, fl->oif, flags); |
8ed67789 | 563 | BACKTRACK(net, &fl->fl6_src); |
c71099ac | 564 | out: |
d8d1f30b | 565 | dst_use(&rt->dst, jiffies); |
c71099ac | 566 | read_unlock_bh(&table->tb6_lock); |
c71099ac TG |
567 | return rt; |
568 | ||
569 | } | |
570 | ||
9acd9f3a YH |
571 | struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr, |
572 | const struct in6_addr *saddr, int oif, int strict) | |
c71099ac TG |
573 | { |
574 | struct flowi fl = { | |
575 | .oif = oif, | |
5811662b | 576 | .fl6_dst = *daddr, |
c71099ac TG |
577 | }; |
578 | struct dst_entry *dst; | |
77d16f45 | 579 | int flags = strict ? RT6_LOOKUP_F_IFACE : 0; |
c71099ac | 580 | |
adaa70bb TG |
581 | if (saddr) { |
582 | memcpy(&fl.fl6_src, saddr, sizeof(*saddr)); | |
583 | flags |= RT6_LOOKUP_F_HAS_SADDR; | |
584 | } | |
585 | ||
606a2b48 | 586 | dst = fib6_rule_lookup(net, &fl, flags, ip6_pol_route_lookup); |
c71099ac TG |
587 | if (dst->error == 0) |
588 | return (struct rt6_info *) dst; | |
589 | ||
590 | dst_release(dst); | |
591 | ||
1da177e4 LT |
592 | return NULL; |
593 | } | |
594 | ||
7159039a YH |
595 | EXPORT_SYMBOL(rt6_lookup); |
596 | ||
c71099ac | 597 | /* ip6_ins_rt is called with FREE table->tb6_lock. |
1da177e4 LT |
598 | It takes new route entry, the addition fails by any reason the |
599 | route is freed. In any case, if caller does not hold it, it may | |
600 | be destroyed. | |
601 | */ | |
602 | ||
86872cb5 | 603 | static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info) |
1da177e4 LT |
604 | { |
605 | int err; | |
c71099ac | 606 | struct fib6_table *table; |
1da177e4 | 607 | |
c71099ac TG |
608 | table = rt->rt6i_table; |
609 | write_lock_bh(&table->tb6_lock); | |
86872cb5 | 610 | err = fib6_add(&table->tb6_root, rt, info); |
c71099ac | 611 | write_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
612 | |
613 | return err; | |
614 | } | |
615 | ||
40e22e8f TG |
616 | int ip6_ins_rt(struct rt6_info *rt) |
617 | { | |
4d1169c1 | 618 | struct nl_info info = { |
c346dca1 | 619 | .nl_net = dev_net(rt->rt6i_dev), |
4d1169c1 | 620 | }; |
528c4ceb | 621 | return __ip6_ins_rt(rt, &info); |
40e22e8f TG |
622 | } |
623 | ||
95a9a5ba YH |
624 | static struct rt6_info *rt6_alloc_cow(struct rt6_info *ort, struct in6_addr *daddr, |
625 | struct in6_addr *saddr) | |
1da177e4 | 626 | { |
1da177e4 LT |
627 | struct rt6_info *rt; |
628 | ||
629 | /* | |
630 | * Clone the route. | |
631 | */ | |
632 | ||
633 | rt = ip6_rt_copy(ort); | |
634 | ||
635 | if (rt) { | |
14deae41 DM |
636 | struct neighbour *neigh; |
637 | int attempts = !in_softirq(); | |
638 | ||
58c4fb86 YH |
639 | if (!(rt->rt6i_flags&RTF_GATEWAY)) { |
640 | if (rt->rt6i_dst.plen != 128 && | |
641 | ipv6_addr_equal(&rt->rt6i_dst.addr, daddr)) | |
642 | rt->rt6i_flags |= RTF_ANYCAST; | |
1da177e4 | 643 | ipv6_addr_copy(&rt->rt6i_gateway, daddr); |
58c4fb86 | 644 | } |
1da177e4 | 645 | |
58c4fb86 | 646 | ipv6_addr_copy(&rt->rt6i_dst.addr, daddr); |
1da177e4 LT |
647 | rt->rt6i_dst.plen = 128; |
648 | rt->rt6i_flags |= RTF_CACHE; | |
d8d1f30b | 649 | rt->dst.flags |= DST_HOST; |
1da177e4 LT |
650 | |
651 | #ifdef CONFIG_IPV6_SUBTREES | |
652 | if (rt->rt6i_src.plen && saddr) { | |
653 | ipv6_addr_copy(&rt->rt6i_src.addr, saddr); | |
654 | rt->rt6i_src.plen = 128; | |
655 | } | |
656 | #endif | |
657 | ||
14deae41 DM |
658 | retry: |
659 | neigh = ndisc_get_neigh(rt->rt6i_dev, &rt->rt6i_gateway); | |
660 | if (IS_ERR(neigh)) { | |
661 | struct net *net = dev_net(rt->rt6i_dev); | |
662 | int saved_rt_min_interval = | |
663 | net->ipv6.sysctl.ip6_rt_gc_min_interval; | |
664 | int saved_rt_elasticity = | |
665 | net->ipv6.sysctl.ip6_rt_gc_elasticity; | |
666 | ||
667 | if (attempts-- > 0) { | |
668 | net->ipv6.sysctl.ip6_rt_gc_elasticity = 1; | |
669 | net->ipv6.sysctl.ip6_rt_gc_min_interval = 0; | |
670 | ||
86393e52 | 671 | ip6_dst_gc(&net->ipv6.ip6_dst_ops); |
14deae41 DM |
672 | |
673 | net->ipv6.sysctl.ip6_rt_gc_elasticity = | |
674 | saved_rt_elasticity; | |
675 | net->ipv6.sysctl.ip6_rt_gc_min_interval = | |
676 | saved_rt_min_interval; | |
677 | goto retry; | |
678 | } | |
679 | ||
680 | if (net_ratelimit()) | |
681 | printk(KERN_WARNING | |
7e1b33e5 | 682 | "ipv6: Neighbour table overflow.\n"); |
d8d1f30b | 683 | dst_free(&rt->dst); |
14deae41 DM |
684 | return NULL; |
685 | } | |
686 | rt->rt6i_nexthop = neigh; | |
1da177e4 | 687 | |
95a9a5ba | 688 | } |
1da177e4 | 689 | |
95a9a5ba YH |
690 | return rt; |
691 | } | |
1da177e4 | 692 | |
299d9939 YH |
693 | static struct rt6_info *rt6_alloc_clone(struct rt6_info *ort, struct in6_addr *daddr) |
694 | { | |
695 | struct rt6_info *rt = ip6_rt_copy(ort); | |
696 | if (rt) { | |
697 | ipv6_addr_copy(&rt->rt6i_dst.addr, daddr); | |
698 | rt->rt6i_dst.plen = 128; | |
699 | rt->rt6i_flags |= RTF_CACHE; | |
d8d1f30b | 700 | rt->dst.flags |= DST_HOST; |
299d9939 YH |
701 | rt->rt6i_nexthop = neigh_clone(ort->rt6i_nexthop); |
702 | } | |
703 | return rt; | |
704 | } | |
705 | ||
8ed67789 DL |
706 | static struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table, int oif, |
707 | struct flowi *fl, int flags) | |
1da177e4 LT |
708 | { |
709 | struct fib6_node *fn; | |
519fbd87 | 710 | struct rt6_info *rt, *nrt; |
c71099ac | 711 | int strict = 0; |
1da177e4 | 712 | int attempts = 3; |
519fbd87 | 713 | int err; |
53b7997f | 714 | int reachable = net->ipv6.devconf_all->forwarding ? 0 : RT6_LOOKUP_F_REACHABLE; |
1da177e4 | 715 | |
77d16f45 | 716 | strict |= flags & RT6_LOOKUP_F_IFACE; |
1da177e4 LT |
717 | |
718 | relookup: | |
c71099ac | 719 | read_lock_bh(&table->tb6_lock); |
1da177e4 | 720 | |
8238dd06 | 721 | restart_2: |
c71099ac | 722 | fn = fib6_lookup(&table->tb6_root, &fl->fl6_dst, &fl->fl6_src); |
1da177e4 LT |
723 | |
724 | restart: | |
4acad72d | 725 | rt = rt6_select(fn, oif, strict | reachable); |
8ed67789 DL |
726 | |
727 | BACKTRACK(net, &fl->fl6_src); | |
728 | if (rt == net->ipv6.ip6_null_entry || | |
8238dd06 | 729 | rt->rt6i_flags & RTF_CACHE) |
1ddef044 | 730 | goto out; |
1da177e4 | 731 | |
d8d1f30b | 732 | dst_hold(&rt->dst); |
c71099ac | 733 | read_unlock_bh(&table->tb6_lock); |
fb9de91e | 734 | |
519fbd87 | 735 | if (!rt->rt6i_nexthop && !(rt->rt6i_flags & RTF_NONEXTHOP)) |
c71099ac | 736 | nrt = rt6_alloc_cow(rt, &fl->fl6_dst, &fl->fl6_src); |
519fbd87 YH |
737 | else { |
738 | #if CLONE_OFFLINK_ROUTE | |
c71099ac | 739 | nrt = rt6_alloc_clone(rt, &fl->fl6_dst); |
519fbd87 YH |
740 | #else |
741 | goto out2; | |
742 | #endif | |
743 | } | |
e40cf353 | 744 | |
d8d1f30b | 745 | dst_release(&rt->dst); |
8ed67789 | 746 | rt = nrt ? : net->ipv6.ip6_null_entry; |
1da177e4 | 747 | |
d8d1f30b | 748 | dst_hold(&rt->dst); |
519fbd87 | 749 | if (nrt) { |
40e22e8f | 750 | err = ip6_ins_rt(nrt); |
519fbd87 | 751 | if (!err) |
1da177e4 | 752 | goto out2; |
1da177e4 | 753 | } |
1da177e4 | 754 | |
519fbd87 YH |
755 | if (--attempts <= 0) |
756 | goto out2; | |
757 | ||
758 | /* | |
c71099ac | 759 | * Race condition! In the gap, when table->tb6_lock was |
519fbd87 YH |
760 | * released someone could insert this route. Relookup. |
761 | */ | |
d8d1f30b | 762 | dst_release(&rt->dst); |
519fbd87 YH |
763 | goto relookup; |
764 | ||
765 | out: | |
8238dd06 YH |
766 | if (reachable) { |
767 | reachable = 0; | |
768 | goto restart_2; | |
769 | } | |
d8d1f30b | 770 | dst_hold(&rt->dst); |
c71099ac | 771 | read_unlock_bh(&table->tb6_lock); |
1da177e4 | 772 | out2: |
d8d1f30b CG |
773 | rt->dst.lastuse = jiffies; |
774 | rt->dst.__use++; | |
c71099ac TG |
775 | |
776 | return rt; | |
1da177e4 LT |
777 | } |
778 | ||
8ed67789 | 779 | static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table, |
4acad72d PE |
780 | struct flowi *fl, int flags) |
781 | { | |
8ed67789 | 782 | return ip6_pol_route(net, table, fl->iif, fl, flags); |
4acad72d PE |
783 | } |
784 | ||
c71099ac TG |
785 | void ip6_route_input(struct sk_buff *skb) |
786 | { | |
0660e03f | 787 | struct ipv6hdr *iph = ipv6_hdr(skb); |
c346dca1 | 788 | struct net *net = dev_net(skb->dev); |
adaa70bb | 789 | int flags = RT6_LOOKUP_F_HAS_SADDR; |
c71099ac TG |
790 | struct flowi fl = { |
791 | .iif = skb->dev->ifindex, | |
5811662b CG |
792 | .fl6_dst = iph->daddr, |
793 | .fl6_src = iph->saddr, | |
794 | .fl6_flowlabel = (* (__be32 *) iph)&IPV6_FLOWINFO_MASK, | |
1ab1457c | 795 | .mark = skb->mark, |
c71099ac TG |
796 | .proto = iph->nexthdr, |
797 | }; | |
adaa70bb | 798 | |
1d6e55f1 | 799 | if (rt6_need_strict(&iph->daddr) && skb->dev->type != ARPHRD_PIMREG) |
adaa70bb | 800 | flags |= RT6_LOOKUP_F_IFACE; |
c71099ac | 801 | |
adf30907 | 802 | skb_dst_set(skb, fib6_rule_lookup(net, &fl, flags, ip6_pol_route_input)); |
c71099ac TG |
803 | } |
804 | ||
8ed67789 | 805 | static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table, |
c71099ac | 806 | struct flowi *fl, int flags) |
1da177e4 | 807 | { |
8ed67789 | 808 | return ip6_pol_route(net, table, fl->oif, fl, flags); |
c71099ac TG |
809 | } |
810 | ||
4591db4f DL |
811 | struct dst_entry * ip6_route_output(struct net *net, struct sock *sk, |
812 | struct flowi *fl) | |
c71099ac TG |
813 | { |
814 | int flags = 0; | |
815 | ||
6057fd78 | 816 | if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl->fl6_dst)) |
77d16f45 | 817 | flags |= RT6_LOOKUP_F_IFACE; |
c71099ac | 818 | |
adaa70bb TG |
819 | if (!ipv6_addr_any(&fl->fl6_src)) |
820 | flags |= RT6_LOOKUP_F_HAS_SADDR; | |
0c9a2ac1 YH |
821 | else if (sk) |
822 | flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs); | |
adaa70bb | 823 | |
4591db4f | 824 | return fib6_rule_lookup(net, fl, flags, ip6_pol_route_output); |
1da177e4 LT |
825 | } |
826 | ||
7159039a | 827 | EXPORT_SYMBOL(ip6_route_output); |
1da177e4 | 828 | |
14e50e57 DM |
829 | int ip6_dst_blackhole(struct sock *sk, struct dst_entry **dstp, struct flowi *fl) |
830 | { | |
831 | struct rt6_info *ort = (struct rt6_info *) *dstp; | |
832 | struct rt6_info *rt = (struct rt6_info *) | |
833 | dst_alloc(&ip6_dst_blackhole_ops); | |
834 | struct dst_entry *new = NULL; | |
835 | ||
836 | if (rt) { | |
d8d1f30b | 837 | new = &rt->dst; |
14e50e57 DM |
838 | |
839 | atomic_set(&new->__refcnt, 1); | |
840 | new->__use = 1; | |
352e512c HX |
841 | new->input = dst_discard; |
842 | new->output = dst_discard; | |
14e50e57 | 843 | |
defb3519 | 844 | dst_copy_metrics(new, &ort->dst); |
d8d1f30b | 845 | new->dev = ort->dst.dev; |
14e50e57 DM |
846 | if (new->dev) |
847 | dev_hold(new->dev); | |
848 | rt->rt6i_idev = ort->rt6i_idev; | |
849 | if (rt->rt6i_idev) | |
850 | in6_dev_hold(rt->rt6i_idev); | |
851 | rt->rt6i_expires = 0; | |
852 | ||
853 | ipv6_addr_copy(&rt->rt6i_gateway, &ort->rt6i_gateway); | |
854 | rt->rt6i_flags = ort->rt6i_flags & ~RTF_EXPIRES; | |
855 | rt->rt6i_metric = 0; | |
856 | ||
857 | memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key)); | |
858 | #ifdef CONFIG_IPV6_SUBTREES | |
859 | memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key)); | |
860 | #endif | |
861 | ||
862 | dst_free(new); | |
863 | } | |
864 | ||
865 | dst_release(*dstp); | |
866 | *dstp = new; | |
a02cec21 | 867 | return new ? 0 : -ENOMEM; |
14e50e57 DM |
868 | } |
869 | EXPORT_SYMBOL_GPL(ip6_dst_blackhole); | |
870 | ||
1da177e4 LT |
871 | /* |
872 | * Destination cache support functions | |
873 | */ | |
874 | ||
875 | static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie) | |
876 | { | |
877 | struct rt6_info *rt; | |
878 | ||
879 | rt = (struct rt6_info *) dst; | |
880 | ||
10414444 | 881 | if (rt->rt6i_node && (rt->rt6i_node->fn_sernum == cookie)) |
1da177e4 LT |
882 | return dst; |
883 | ||
884 | return NULL; | |
885 | } | |
886 | ||
887 | static struct dst_entry *ip6_negative_advice(struct dst_entry *dst) | |
888 | { | |
889 | struct rt6_info *rt = (struct rt6_info *) dst; | |
890 | ||
891 | if (rt) { | |
54c1a859 YH |
892 | if (rt->rt6i_flags & RTF_CACHE) { |
893 | if (rt6_check_expired(rt)) { | |
894 | ip6_del_rt(rt); | |
895 | dst = NULL; | |
896 | } | |
897 | } else { | |
1da177e4 | 898 | dst_release(dst); |
54c1a859 YH |
899 | dst = NULL; |
900 | } | |
1da177e4 | 901 | } |
54c1a859 | 902 | return dst; |
1da177e4 LT |
903 | } |
904 | ||
905 | static void ip6_link_failure(struct sk_buff *skb) | |
906 | { | |
907 | struct rt6_info *rt; | |
908 | ||
3ffe533c | 909 | icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0); |
1da177e4 | 910 | |
adf30907 | 911 | rt = (struct rt6_info *) skb_dst(skb); |
1da177e4 LT |
912 | if (rt) { |
913 | if (rt->rt6i_flags&RTF_CACHE) { | |
d8d1f30b | 914 | dst_set_expires(&rt->dst, 0); |
1da177e4 LT |
915 | rt->rt6i_flags |= RTF_EXPIRES; |
916 | } else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT)) | |
917 | rt->rt6i_node->fn_sernum = -1; | |
918 | } | |
919 | } | |
920 | ||
921 | static void ip6_rt_update_pmtu(struct dst_entry *dst, u32 mtu) | |
922 | { | |
923 | struct rt6_info *rt6 = (struct rt6_info*)dst; | |
924 | ||
925 | if (mtu < dst_mtu(dst) && rt6->rt6i_dst.plen == 128) { | |
926 | rt6->rt6i_flags |= RTF_MODIFIED; | |
927 | if (mtu < IPV6_MIN_MTU) { | |
defb3519 | 928 | u32 features = dst_metric(dst, RTAX_FEATURES); |
1da177e4 | 929 | mtu = IPV6_MIN_MTU; |
defb3519 DM |
930 | features |= RTAX_FEATURE_ALLFRAG; |
931 | dst_metric_set(dst, RTAX_FEATURES, features); | |
1da177e4 | 932 | } |
defb3519 | 933 | dst_metric_set(dst, RTAX_MTU, mtu); |
8d71740c | 934 | call_netevent_notifiers(NETEVENT_PMTU_UPDATE, dst); |
1da177e4 LT |
935 | } |
936 | } | |
937 | ||
1da177e4 LT |
938 | static int ipv6_get_mtu(struct net_device *dev); |
939 | ||
5578689a | 940 | static inline unsigned int ipv6_advmss(struct net *net, unsigned int mtu) |
1da177e4 LT |
941 | { |
942 | mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr); | |
943 | ||
5578689a DL |
944 | if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss) |
945 | mtu = net->ipv6.sysctl.ip6_rt_min_advmss; | |
1da177e4 LT |
946 | |
947 | /* | |
1ab1457c YH |
948 | * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and |
949 | * corresponding MSS is IPV6_MAXPLEN - tcp_header_size. | |
950 | * IPV6_MAXPLEN is also valid and means: "any MSS, | |
1da177e4 LT |
951 | * rely only on pmtu discovery" |
952 | */ | |
953 | if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr)) | |
954 | mtu = IPV6_MAXPLEN; | |
955 | return mtu; | |
956 | } | |
957 | ||
3b00944c YH |
958 | static struct dst_entry *icmp6_dst_gc_list; |
959 | static DEFINE_SPINLOCK(icmp6_dst_lock); | |
5d0bbeeb | 960 | |
3b00944c | 961 | struct dst_entry *icmp6_dst_alloc(struct net_device *dev, |
1da177e4 | 962 | struct neighbour *neigh, |
9acd9f3a | 963 | const struct in6_addr *addr) |
1da177e4 LT |
964 | { |
965 | struct rt6_info *rt; | |
966 | struct inet6_dev *idev = in6_dev_get(dev); | |
c346dca1 | 967 | struct net *net = dev_net(dev); |
1da177e4 LT |
968 | |
969 | if (unlikely(idev == NULL)) | |
970 | return NULL; | |
971 | ||
86393e52 | 972 | rt = ip6_dst_alloc(&net->ipv6.ip6_dst_ops); |
1da177e4 LT |
973 | if (unlikely(rt == NULL)) { |
974 | in6_dev_put(idev); | |
975 | goto out; | |
976 | } | |
977 | ||
978 | dev_hold(dev); | |
979 | if (neigh) | |
980 | neigh_hold(neigh); | |
14deae41 | 981 | else { |
1da177e4 | 982 | neigh = ndisc_get_neigh(dev, addr); |
14deae41 DM |
983 | if (IS_ERR(neigh)) |
984 | neigh = NULL; | |
985 | } | |
1da177e4 LT |
986 | |
987 | rt->rt6i_dev = dev; | |
988 | rt->rt6i_idev = idev; | |
989 | rt->rt6i_nexthop = neigh; | |
d8d1f30b | 990 | atomic_set(&rt->dst.__refcnt, 1); |
defb3519 DM |
991 | dst_metric_set(&rt->dst, RTAX_HOPLIMIT, 255); |
992 | dst_metric_set(&rt->dst, RTAX_MTU, ipv6_get_mtu(rt->rt6i_dev)); | |
993 | dst_metric_set(&rt->dst, RTAX_ADVMSS, ipv6_advmss(net, dst_mtu(&rt->dst))); | |
d8d1f30b | 994 | rt->dst.output = ip6_output; |
1da177e4 LT |
995 | |
996 | #if 0 /* there's no chance to use these for ndisc */ | |
d8d1f30b | 997 | rt->dst.flags = ipv6_addr_type(addr) & IPV6_ADDR_UNICAST |
1ab1457c | 998 | ? DST_HOST |
1da177e4 LT |
999 | : 0; |
1000 | ipv6_addr_copy(&rt->rt6i_dst.addr, addr); | |
1001 | rt->rt6i_dst.plen = 128; | |
1002 | #endif | |
1003 | ||
3b00944c | 1004 | spin_lock_bh(&icmp6_dst_lock); |
d8d1f30b CG |
1005 | rt->dst.next = icmp6_dst_gc_list; |
1006 | icmp6_dst_gc_list = &rt->dst; | |
3b00944c | 1007 | spin_unlock_bh(&icmp6_dst_lock); |
1da177e4 | 1008 | |
5578689a | 1009 | fib6_force_start_gc(net); |
1da177e4 LT |
1010 | |
1011 | out: | |
d8d1f30b | 1012 | return &rt->dst; |
1da177e4 LT |
1013 | } |
1014 | ||
3d0f24a7 | 1015 | int icmp6_dst_gc(void) |
1da177e4 LT |
1016 | { |
1017 | struct dst_entry *dst, *next, **pprev; | |
3d0f24a7 | 1018 | int more = 0; |
1da177e4 LT |
1019 | |
1020 | next = NULL; | |
5d0bbeeb | 1021 | |
3b00944c YH |
1022 | spin_lock_bh(&icmp6_dst_lock); |
1023 | pprev = &icmp6_dst_gc_list; | |
5d0bbeeb | 1024 | |
1da177e4 LT |
1025 | while ((dst = *pprev) != NULL) { |
1026 | if (!atomic_read(&dst->__refcnt)) { | |
1027 | *pprev = dst->next; | |
1028 | dst_free(dst); | |
1da177e4 LT |
1029 | } else { |
1030 | pprev = &dst->next; | |
3d0f24a7 | 1031 | ++more; |
1da177e4 LT |
1032 | } |
1033 | } | |
1034 | ||
3b00944c | 1035 | spin_unlock_bh(&icmp6_dst_lock); |
5d0bbeeb | 1036 | |
3d0f24a7 | 1037 | return more; |
1da177e4 LT |
1038 | } |
1039 | ||
1e493d19 DM |
1040 | static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg), |
1041 | void *arg) | |
1042 | { | |
1043 | struct dst_entry *dst, **pprev; | |
1044 | ||
1045 | spin_lock_bh(&icmp6_dst_lock); | |
1046 | pprev = &icmp6_dst_gc_list; | |
1047 | while ((dst = *pprev) != NULL) { | |
1048 | struct rt6_info *rt = (struct rt6_info *) dst; | |
1049 | if (func(rt, arg)) { | |
1050 | *pprev = dst->next; | |
1051 | dst_free(dst); | |
1052 | } else { | |
1053 | pprev = &dst->next; | |
1054 | } | |
1055 | } | |
1056 | spin_unlock_bh(&icmp6_dst_lock); | |
1057 | } | |
1058 | ||
569d3645 | 1059 | static int ip6_dst_gc(struct dst_ops *ops) |
1da177e4 | 1060 | { |
1da177e4 | 1061 | unsigned long now = jiffies; |
86393e52 | 1062 | struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops); |
7019b78e DL |
1063 | int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval; |
1064 | int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size; | |
1065 | int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity; | |
1066 | int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout; | |
1067 | unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc; | |
fc66f95c | 1068 | int entries; |
7019b78e | 1069 | |
fc66f95c | 1070 | entries = dst_entries_get_fast(ops); |
7019b78e | 1071 | if (time_after(rt_last_gc + rt_min_interval, now) && |
fc66f95c | 1072 | entries <= rt_max_size) |
1da177e4 LT |
1073 | goto out; |
1074 | ||
6891a346 BT |
1075 | net->ipv6.ip6_rt_gc_expire++; |
1076 | fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net); | |
1077 | net->ipv6.ip6_rt_last_gc = now; | |
fc66f95c ED |
1078 | entries = dst_entries_get_slow(ops); |
1079 | if (entries < ops->gc_thresh) | |
7019b78e | 1080 | net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1; |
1da177e4 | 1081 | out: |
7019b78e | 1082 | net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity; |
fc66f95c | 1083 | return entries > rt_max_size; |
1da177e4 LT |
1084 | } |
1085 | ||
1086 | /* Clean host part of a prefix. Not necessary in radix tree, | |
1087 | but results in cleaner routing tables. | |
1088 | ||
1089 | Remove it only when all the things will work! | |
1090 | */ | |
1091 | ||
1092 | static int ipv6_get_mtu(struct net_device *dev) | |
1093 | { | |
1094 | int mtu = IPV6_MIN_MTU; | |
1095 | struct inet6_dev *idev; | |
1096 | ||
c68f24cc ED |
1097 | rcu_read_lock(); |
1098 | idev = __in6_dev_get(dev); | |
1099 | if (idev) | |
1da177e4 | 1100 | mtu = idev->cnf.mtu6; |
c68f24cc | 1101 | rcu_read_unlock(); |
1da177e4 LT |
1102 | return mtu; |
1103 | } | |
1104 | ||
6b75d090 | 1105 | int ip6_dst_hoplimit(struct dst_entry *dst) |
1da177e4 | 1106 | { |
5170ae82 | 1107 | int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT); |
a02e4b7d | 1108 | if (hoplimit == 0) { |
6b75d090 | 1109 | struct net_device *dev = dst->dev; |
c68f24cc ED |
1110 | struct inet6_dev *idev; |
1111 | ||
1112 | rcu_read_lock(); | |
1113 | idev = __in6_dev_get(dev); | |
1114 | if (idev) | |
6b75d090 | 1115 | hoplimit = idev->cnf.hop_limit; |
c68f24cc | 1116 | else |
53b7997f | 1117 | hoplimit = dev_net(dev)->ipv6.devconf_all->hop_limit; |
c68f24cc | 1118 | rcu_read_unlock(); |
1da177e4 LT |
1119 | } |
1120 | return hoplimit; | |
1121 | } | |
abbf46ae | 1122 | EXPORT_SYMBOL(ip6_dst_hoplimit); |
1da177e4 LT |
1123 | |
1124 | /* | |
1125 | * | |
1126 | */ | |
1127 | ||
86872cb5 | 1128 | int ip6_route_add(struct fib6_config *cfg) |
1da177e4 LT |
1129 | { |
1130 | int err; | |
5578689a | 1131 | struct net *net = cfg->fc_nlinfo.nl_net; |
1da177e4 LT |
1132 | struct rt6_info *rt = NULL; |
1133 | struct net_device *dev = NULL; | |
1134 | struct inet6_dev *idev = NULL; | |
c71099ac | 1135 | struct fib6_table *table; |
1da177e4 LT |
1136 | int addr_type; |
1137 | ||
86872cb5 | 1138 | if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128) |
1da177e4 LT |
1139 | return -EINVAL; |
1140 | #ifndef CONFIG_IPV6_SUBTREES | |
86872cb5 | 1141 | if (cfg->fc_src_len) |
1da177e4 LT |
1142 | return -EINVAL; |
1143 | #endif | |
86872cb5 | 1144 | if (cfg->fc_ifindex) { |
1da177e4 | 1145 | err = -ENODEV; |
5578689a | 1146 | dev = dev_get_by_index(net, cfg->fc_ifindex); |
1da177e4 LT |
1147 | if (!dev) |
1148 | goto out; | |
1149 | idev = in6_dev_get(dev); | |
1150 | if (!idev) | |
1151 | goto out; | |
1152 | } | |
1153 | ||
86872cb5 TG |
1154 | if (cfg->fc_metric == 0) |
1155 | cfg->fc_metric = IP6_RT_PRIO_USER; | |
1da177e4 | 1156 | |
5578689a | 1157 | table = fib6_new_table(net, cfg->fc_table); |
c71099ac TG |
1158 | if (table == NULL) { |
1159 | err = -ENOBUFS; | |
1160 | goto out; | |
1161 | } | |
1162 | ||
86393e52 | 1163 | rt = ip6_dst_alloc(&net->ipv6.ip6_dst_ops); |
1da177e4 LT |
1164 | |
1165 | if (rt == NULL) { | |
1166 | err = -ENOMEM; | |
1167 | goto out; | |
1168 | } | |
1169 | ||
d8d1f30b | 1170 | rt->dst.obsolete = -1; |
6f704992 YH |
1171 | rt->rt6i_expires = (cfg->fc_flags & RTF_EXPIRES) ? |
1172 | jiffies + clock_t_to_jiffies(cfg->fc_expires) : | |
1173 | 0; | |
1da177e4 | 1174 | |
86872cb5 TG |
1175 | if (cfg->fc_protocol == RTPROT_UNSPEC) |
1176 | cfg->fc_protocol = RTPROT_BOOT; | |
1177 | rt->rt6i_protocol = cfg->fc_protocol; | |
1178 | ||
1179 | addr_type = ipv6_addr_type(&cfg->fc_dst); | |
1da177e4 LT |
1180 | |
1181 | if (addr_type & IPV6_ADDR_MULTICAST) | |
d8d1f30b | 1182 | rt->dst.input = ip6_mc_input; |
ab79ad14 MŻ |
1183 | else if (cfg->fc_flags & RTF_LOCAL) |
1184 | rt->dst.input = ip6_input; | |
1da177e4 | 1185 | else |
d8d1f30b | 1186 | rt->dst.input = ip6_forward; |
1da177e4 | 1187 | |
d8d1f30b | 1188 | rt->dst.output = ip6_output; |
1da177e4 | 1189 | |
86872cb5 TG |
1190 | ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len); |
1191 | rt->rt6i_dst.plen = cfg->fc_dst_len; | |
1da177e4 | 1192 | if (rt->rt6i_dst.plen == 128) |
d8d1f30b | 1193 | rt->dst.flags = DST_HOST; |
1da177e4 LT |
1194 | |
1195 | #ifdef CONFIG_IPV6_SUBTREES | |
86872cb5 TG |
1196 | ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len); |
1197 | rt->rt6i_src.plen = cfg->fc_src_len; | |
1da177e4 LT |
1198 | #endif |
1199 | ||
86872cb5 | 1200 | rt->rt6i_metric = cfg->fc_metric; |
1da177e4 LT |
1201 | |
1202 | /* We cannot add true routes via loopback here, | |
1203 | they would result in kernel looping; promote them to reject routes | |
1204 | */ | |
86872cb5 | 1205 | if ((cfg->fc_flags & RTF_REJECT) || |
ab79ad14 MŻ |
1206 | (dev && (dev->flags&IFF_LOOPBACK) && !(addr_type&IPV6_ADDR_LOOPBACK) |
1207 | && !(cfg->fc_flags&RTF_LOCAL))) { | |
1da177e4 | 1208 | /* hold loopback dev/idev if we haven't done so. */ |
5578689a | 1209 | if (dev != net->loopback_dev) { |
1da177e4 LT |
1210 | if (dev) { |
1211 | dev_put(dev); | |
1212 | in6_dev_put(idev); | |
1213 | } | |
5578689a | 1214 | dev = net->loopback_dev; |
1da177e4 LT |
1215 | dev_hold(dev); |
1216 | idev = in6_dev_get(dev); | |
1217 | if (!idev) { | |
1218 | err = -ENODEV; | |
1219 | goto out; | |
1220 | } | |
1221 | } | |
d8d1f30b CG |
1222 | rt->dst.output = ip6_pkt_discard_out; |
1223 | rt->dst.input = ip6_pkt_discard; | |
1224 | rt->dst.error = -ENETUNREACH; | |
1da177e4 LT |
1225 | rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP; |
1226 | goto install_route; | |
1227 | } | |
1228 | ||
86872cb5 | 1229 | if (cfg->fc_flags & RTF_GATEWAY) { |
1da177e4 LT |
1230 | struct in6_addr *gw_addr; |
1231 | int gwa_type; | |
1232 | ||
86872cb5 TG |
1233 | gw_addr = &cfg->fc_gateway; |
1234 | ipv6_addr_copy(&rt->rt6i_gateway, gw_addr); | |
1da177e4 LT |
1235 | gwa_type = ipv6_addr_type(gw_addr); |
1236 | ||
1237 | if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) { | |
1238 | struct rt6_info *grt; | |
1239 | ||
1240 | /* IPv6 strictly inhibits using not link-local | |
1241 | addresses as nexthop address. | |
1242 | Otherwise, router will not able to send redirects. | |
1243 | It is very good, but in some (rare!) circumstances | |
1244 | (SIT, PtP, NBMA NOARP links) it is handy to allow | |
1245 | some exceptions. --ANK | |
1246 | */ | |
1247 | err = -EINVAL; | |
1248 | if (!(gwa_type&IPV6_ADDR_UNICAST)) | |
1249 | goto out; | |
1250 | ||
5578689a | 1251 | grt = rt6_lookup(net, gw_addr, NULL, cfg->fc_ifindex, 1); |
1da177e4 LT |
1252 | |
1253 | err = -EHOSTUNREACH; | |
1254 | if (grt == NULL) | |
1255 | goto out; | |
1256 | if (dev) { | |
1257 | if (dev != grt->rt6i_dev) { | |
d8d1f30b | 1258 | dst_release(&grt->dst); |
1da177e4 LT |
1259 | goto out; |
1260 | } | |
1261 | } else { | |
1262 | dev = grt->rt6i_dev; | |
1263 | idev = grt->rt6i_idev; | |
1264 | dev_hold(dev); | |
1265 | in6_dev_hold(grt->rt6i_idev); | |
1266 | } | |
1267 | if (!(grt->rt6i_flags&RTF_GATEWAY)) | |
1268 | err = 0; | |
d8d1f30b | 1269 | dst_release(&grt->dst); |
1da177e4 LT |
1270 | |
1271 | if (err) | |
1272 | goto out; | |
1273 | } | |
1274 | err = -EINVAL; | |
1275 | if (dev == NULL || (dev->flags&IFF_LOOPBACK)) | |
1276 | goto out; | |
1277 | } | |
1278 | ||
1279 | err = -ENODEV; | |
1280 | if (dev == NULL) | |
1281 | goto out; | |
1282 | ||
86872cb5 | 1283 | if (cfg->fc_flags & (RTF_GATEWAY | RTF_NONEXTHOP)) { |
1da177e4 LT |
1284 | rt->rt6i_nexthop = __neigh_lookup_errno(&nd_tbl, &rt->rt6i_gateway, dev); |
1285 | if (IS_ERR(rt->rt6i_nexthop)) { | |
1286 | err = PTR_ERR(rt->rt6i_nexthop); | |
1287 | rt->rt6i_nexthop = NULL; | |
1288 | goto out; | |
1289 | } | |
1290 | } | |
1291 | ||
86872cb5 | 1292 | rt->rt6i_flags = cfg->fc_flags; |
1da177e4 LT |
1293 | |
1294 | install_route: | |
86872cb5 TG |
1295 | if (cfg->fc_mx) { |
1296 | struct nlattr *nla; | |
1297 | int remaining; | |
1298 | ||
1299 | nla_for_each_attr(nla, cfg->fc_mx, cfg->fc_mx_len, remaining) { | |
8f4c1f9b | 1300 | int type = nla_type(nla); |
86872cb5 TG |
1301 | |
1302 | if (type) { | |
1303 | if (type > RTAX_MAX) { | |
1da177e4 LT |
1304 | err = -EINVAL; |
1305 | goto out; | |
1306 | } | |
86872cb5 | 1307 | |
defb3519 | 1308 | dst_metric_set(&rt->dst, type, nla_get_u32(nla)); |
1da177e4 | 1309 | } |
1da177e4 LT |
1310 | } |
1311 | } | |
1312 | ||
d8d1f30b | 1313 | if (!dst_mtu(&rt->dst)) |
defb3519 | 1314 | dst_metric_set(&rt->dst, RTAX_MTU, ipv6_get_mtu(dev)); |
d8d1f30b | 1315 | if (!dst_metric(&rt->dst, RTAX_ADVMSS)) |
defb3519 | 1316 | dst_metric_set(&rt->dst, RTAX_ADVMSS, ipv6_advmss(net, dst_mtu(&rt->dst))); |
d8d1f30b | 1317 | rt->dst.dev = dev; |
1da177e4 | 1318 | rt->rt6i_idev = idev; |
c71099ac | 1319 | rt->rt6i_table = table; |
63152fc0 | 1320 | |
c346dca1 | 1321 | cfg->fc_nlinfo.nl_net = dev_net(dev); |
63152fc0 | 1322 | |
86872cb5 | 1323 | return __ip6_ins_rt(rt, &cfg->fc_nlinfo); |
1da177e4 LT |
1324 | |
1325 | out: | |
1326 | if (dev) | |
1327 | dev_put(dev); | |
1328 | if (idev) | |
1329 | in6_dev_put(idev); | |
1330 | if (rt) | |
d8d1f30b | 1331 | dst_free(&rt->dst); |
1da177e4 LT |
1332 | return err; |
1333 | } | |
1334 | ||
86872cb5 | 1335 | static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info) |
1da177e4 LT |
1336 | { |
1337 | int err; | |
c71099ac | 1338 | struct fib6_table *table; |
c346dca1 | 1339 | struct net *net = dev_net(rt->rt6i_dev); |
1da177e4 | 1340 | |
8ed67789 | 1341 | if (rt == net->ipv6.ip6_null_entry) |
6c813a72 PM |
1342 | return -ENOENT; |
1343 | ||
c71099ac TG |
1344 | table = rt->rt6i_table; |
1345 | write_lock_bh(&table->tb6_lock); | |
1da177e4 | 1346 | |
86872cb5 | 1347 | err = fib6_del(rt, info); |
d8d1f30b | 1348 | dst_release(&rt->dst); |
1da177e4 | 1349 | |
c71099ac | 1350 | write_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
1351 | |
1352 | return err; | |
1353 | } | |
1354 | ||
e0a1ad73 TG |
1355 | int ip6_del_rt(struct rt6_info *rt) |
1356 | { | |
4d1169c1 | 1357 | struct nl_info info = { |
c346dca1 | 1358 | .nl_net = dev_net(rt->rt6i_dev), |
4d1169c1 | 1359 | }; |
528c4ceb | 1360 | return __ip6_del_rt(rt, &info); |
e0a1ad73 TG |
1361 | } |
1362 | ||
86872cb5 | 1363 | static int ip6_route_del(struct fib6_config *cfg) |
1da177e4 | 1364 | { |
c71099ac | 1365 | struct fib6_table *table; |
1da177e4 LT |
1366 | struct fib6_node *fn; |
1367 | struct rt6_info *rt; | |
1368 | int err = -ESRCH; | |
1369 | ||
5578689a | 1370 | table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table); |
c71099ac TG |
1371 | if (table == NULL) |
1372 | return err; | |
1373 | ||
1374 | read_lock_bh(&table->tb6_lock); | |
1da177e4 | 1375 | |
c71099ac | 1376 | fn = fib6_locate(&table->tb6_root, |
86872cb5 TG |
1377 | &cfg->fc_dst, cfg->fc_dst_len, |
1378 | &cfg->fc_src, cfg->fc_src_len); | |
1ab1457c | 1379 | |
1da177e4 | 1380 | if (fn) { |
d8d1f30b | 1381 | for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) { |
86872cb5 | 1382 | if (cfg->fc_ifindex && |
1da177e4 | 1383 | (rt->rt6i_dev == NULL || |
86872cb5 | 1384 | rt->rt6i_dev->ifindex != cfg->fc_ifindex)) |
1da177e4 | 1385 | continue; |
86872cb5 TG |
1386 | if (cfg->fc_flags & RTF_GATEWAY && |
1387 | !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway)) | |
1da177e4 | 1388 | continue; |
86872cb5 | 1389 | if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric) |
1da177e4 | 1390 | continue; |
d8d1f30b | 1391 | dst_hold(&rt->dst); |
c71099ac | 1392 | read_unlock_bh(&table->tb6_lock); |
1da177e4 | 1393 | |
86872cb5 | 1394 | return __ip6_del_rt(rt, &cfg->fc_nlinfo); |
1da177e4 LT |
1395 | } |
1396 | } | |
c71099ac | 1397 | read_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
1398 | |
1399 | return err; | |
1400 | } | |
1401 | ||
1402 | /* | |
1403 | * Handle redirects | |
1404 | */ | |
a6279458 YH |
1405 | struct ip6rd_flowi { |
1406 | struct flowi fl; | |
1407 | struct in6_addr gateway; | |
1408 | }; | |
1409 | ||
8ed67789 DL |
1410 | static struct rt6_info *__ip6_route_redirect(struct net *net, |
1411 | struct fib6_table *table, | |
a6279458 YH |
1412 | struct flowi *fl, |
1413 | int flags) | |
1da177e4 | 1414 | { |
a6279458 YH |
1415 | struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl; |
1416 | struct rt6_info *rt; | |
e843b9e1 | 1417 | struct fib6_node *fn; |
c71099ac | 1418 | |
1da177e4 | 1419 | /* |
e843b9e1 YH |
1420 | * Get the "current" route for this destination and |
1421 | * check if the redirect has come from approriate router. | |
1422 | * | |
1423 | * RFC 2461 specifies that redirects should only be | |
1424 | * accepted if they come from the nexthop to the target. | |
1425 | * Due to the way the routes are chosen, this notion | |
1426 | * is a bit fuzzy and one might need to check all possible | |
1427 | * routes. | |
1da177e4 | 1428 | */ |
1da177e4 | 1429 | |
c71099ac | 1430 | read_lock_bh(&table->tb6_lock); |
a6279458 | 1431 | fn = fib6_lookup(&table->tb6_root, &fl->fl6_dst, &fl->fl6_src); |
e843b9e1 | 1432 | restart: |
d8d1f30b | 1433 | for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) { |
e843b9e1 YH |
1434 | /* |
1435 | * Current route is on-link; redirect is always invalid. | |
1436 | * | |
1437 | * Seems, previous statement is not true. It could | |
1438 | * be node, which looks for us as on-link (f.e. proxy ndisc) | |
1439 | * But then router serving it might decide, that we should | |
1440 | * know truth 8)8) --ANK (980726). | |
1441 | */ | |
1442 | if (rt6_check_expired(rt)) | |
1443 | continue; | |
1444 | if (!(rt->rt6i_flags & RTF_GATEWAY)) | |
1445 | continue; | |
a6279458 | 1446 | if (fl->oif != rt->rt6i_dev->ifindex) |
e843b9e1 | 1447 | continue; |
a6279458 | 1448 | if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway)) |
e843b9e1 YH |
1449 | continue; |
1450 | break; | |
1451 | } | |
a6279458 | 1452 | |
cb15d9c2 | 1453 | if (!rt) |
8ed67789 DL |
1454 | rt = net->ipv6.ip6_null_entry; |
1455 | BACKTRACK(net, &fl->fl6_src); | |
cb15d9c2 | 1456 | out: |
d8d1f30b | 1457 | dst_hold(&rt->dst); |
a6279458 | 1458 | |
c71099ac | 1459 | read_unlock_bh(&table->tb6_lock); |
e843b9e1 | 1460 | |
a6279458 YH |
1461 | return rt; |
1462 | }; | |
1463 | ||
1464 | static struct rt6_info *ip6_route_redirect(struct in6_addr *dest, | |
1465 | struct in6_addr *src, | |
1466 | struct in6_addr *gateway, | |
1467 | struct net_device *dev) | |
1468 | { | |
adaa70bb | 1469 | int flags = RT6_LOOKUP_F_HAS_SADDR; |
c346dca1 | 1470 | struct net *net = dev_net(dev); |
a6279458 YH |
1471 | struct ip6rd_flowi rdfl = { |
1472 | .fl = { | |
1473 | .oif = dev->ifindex, | |
5811662b CG |
1474 | .fl6_dst = *dest, |
1475 | .fl6_src = *src, | |
a6279458 | 1476 | }, |
a6279458 | 1477 | }; |
adaa70bb | 1478 | |
86c36ce4 BH |
1479 | ipv6_addr_copy(&rdfl.gateway, gateway); |
1480 | ||
adaa70bb TG |
1481 | if (rt6_need_strict(dest)) |
1482 | flags |= RT6_LOOKUP_F_IFACE; | |
a6279458 | 1483 | |
5578689a | 1484 | return (struct rt6_info *)fib6_rule_lookup(net, (struct flowi *)&rdfl, |
58f09b78 | 1485 | flags, __ip6_route_redirect); |
a6279458 YH |
1486 | } |
1487 | ||
1488 | void rt6_redirect(struct in6_addr *dest, struct in6_addr *src, | |
1489 | struct in6_addr *saddr, | |
1490 | struct neighbour *neigh, u8 *lladdr, int on_link) | |
1491 | { | |
1492 | struct rt6_info *rt, *nrt = NULL; | |
1493 | struct netevent_redirect netevent; | |
c346dca1 | 1494 | struct net *net = dev_net(neigh->dev); |
a6279458 YH |
1495 | |
1496 | rt = ip6_route_redirect(dest, src, saddr, neigh->dev); | |
1497 | ||
8ed67789 | 1498 | if (rt == net->ipv6.ip6_null_entry) { |
1da177e4 LT |
1499 | if (net_ratelimit()) |
1500 | printk(KERN_DEBUG "rt6_redirect: source isn't a valid nexthop " | |
1501 | "for redirect target\n"); | |
a6279458 | 1502 | goto out; |
1da177e4 LT |
1503 | } |
1504 | ||
1da177e4 LT |
1505 | /* |
1506 | * We have finally decided to accept it. | |
1507 | */ | |
1508 | ||
1ab1457c | 1509 | neigh_update(neigh, lladdr, NUD_STALE, |
1da177e4 LT |
1510 | NEIGH_UPDATE_F_WEAK_OVERRIDE| |
1511 | NEIGH_UPDATE_F_OVERRIDE| | |
1512 | (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER| | |
1513 | NEIGH_UPDATE_F_ISROUTER)) | |
1514 | ); | |
1515 | ||
1516 | /* | |
1517 | * Redirect received -> path was valid. | |
1518 | * Look, redirects are sent only in response to data packets, | |
1519 | * so that this nexthop apparently is reachable. --ANK | |
1520 | */ | |
d8d1f30b | 1521 | dst_confirm(&rt->dst); |
1da177e4 LT |
1522 | |
1523 | /* Duplicate redirect: silently ignore. */ | |
d8d1f30b | 1524 | if (neigh == rt->dst.neighbour) |
1da177e4 LT |
1525 | goto out; |
1526 | ||
1527 | nrt = ip6_rt_copy(rt); | |
1528 | if (nrt == NULL) | |
1529 | goto out; | |
1530 | ||
1531 | nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE; | |
1532 | if (on_link) | |
1533 | nrt->rt6i_flags &= ~RTF_GATEWAY; | |
1534 | ||
1535 | ipv6_addr_copy(&nrt->rt6i_dst.addr, dest); | |
1536 | nrt->rt6i_dst.plen = 128; | |
d8d1f30b | 1537 | nrt->dst.flags |= DST_HOST; |
1da177e4 LT |
1538 | |
1539 | ipv6_addr_copy(&nrt->rt6i_gateway, (struct in6_addr*)neigh->primary_key); | |
1540 | nrt->rt6i_nexthop = neigh_clone(neigh); | |
1541 | /* Reset pmtu, it may be better */ | |
defb3519 DM |
1542 | dst_metric_set(&nrt->dst, RTAX_MTU, ipv6_get_mtu(neigh->dev)); |
1543 | dst_metric_set(&nrt->dst, RTAX_ADVMSS, ipv6_advmss(dev_net(neigh->dev), | |
1544 | dst_mtu(&nrt->dst))); | |
1da177e4 | 1545 | |
40e22e8f | 1546 | if (ip6_ins_rt(nrt)) |
1da177e4 LT |
1547 | goto out; |
1548 | ||
d8d1f30b CG |
1549 | netevent.old = &rt->dst; |
1550 | netevent.new = &nrt->dst; | |
8d71740c TT |
1551 | call_netevent_notifiers(NETEVENT_REDIRECT, &netevent); |
1552 | ||
1da177e4 | 1553 | if (rt->rt6i_flags&RTF_CACHE) { |
e0a1ad73 | 1554 | ip6_del_rt(rt); |
1da177e4 LT |
1555 | return; |
1556 | } | |
1557 | ||
1558 | out: | |
d8d1f30b | 1559 | dst_release(&rt->dst); |
1da177e4 LT |
1560 | } |
1561 | ||
1562 | /* | |
1563 | * Handle ICMP "packet too big" messages | |
1564 | * i.e. Path MTU discovery | |
1565 | */ | |
1566 | ||
ae878ae2 MŻ |
1567 | static void rt6_do_pmtu_disc(struct in6_addr *daddr, struct in6_addr *saddr, |
1568 | struct net *net, u32 pmtu, int ifindex) | |
1da177e4 LT |
1569 | { |
1570 | struct rt6_info *rt, *nrt; | |
1571 | int allfrag = 0; | |
1572 | ||
ae878ae2 | 1573 | rt = rt6_lookup(net, daddr, saddr, ifindex, 0); |
1da177e4 LT |
1574 | if (rt == NULL) |
1575 | return; | |
1576 | ||
d8d1f30b | 1577 | if (pmtu >= dst_mtu(&rt->dst)) |
1da177e4 LT |
1578 | goto out; |
1579 | ||
1580 | if (pmtu < IPV6_MIN_MTU) { | |
1581 | /* | |
1ab1457c | 1582 | * According to RFC2460, PMTU is set to the IPv6 Minimum Link |
1da177e4 LT |
1583 | * MTU (1280) and a fragment header should always be included |
1584 | * after a node receiving Too Big message reporting PMTU is | |
1585 | * less than the IPv6 Minimum Link MTU. | |
1586 | */ | |
1587 | pmtu = IPV6_MIN_MTU; | |
1588 | allfrag = 1; | |
1589 | } | |
1590 | ||
1591 | /* New mtu received -> path was valid. | |
1592 | They are sent only in response to data packets, | |
1593 | so that this nexthop apparently is reachable. --ANK | |
1594 | */ | |
d8d1f30b | 1595 | dst_confirm(&rt->dst); |
1da177e4 LT |
1596 | |
1597 | /* Host route. If it is static, it would be better | |
1598 | not to override it, but add new one, so that | |
1599 | when cache entry will expire old pmtu | |
1600 | would return automatically. | |
1601 | */ | |
1602 | if (rt->rt6i_flags & RTF_CACHE) { | |
defb3519 DM |
1603 | dst_metric_set(&rt->dst, RTAX_MTU, pmtu); |
1604 | if (allfrag) { | |
1605 | u32 features = dst_metric(&rt->dst, RTAX_FEATURES); | |
1606 | features |= RTAX_FEATURE_ALLFRAG; | |
1607 | dst_metric_set(&rt->dst, RTAX_FEATURES, features); | |
1608 | } | |
d8d1f30b | 1609 | dst_set_expires(&rt->dst, net->ipv6.sysctl.ip6_rt_mtu_expires); |
1da177e4 LT |
1610 | rt->rt6i_flags |= RTF_MODIFIED|RTF_EXPIRES; |
1611 | goto out; | |
1612 | } | |
1613 | ||
1614 | /* Network route. | |
1615 | Two cases are possible: | |
1616 | 1. It is connected route. Action: COW | |
1617 | 2. It is gatewayed route or NONEXTHOP route. Action: clone it. | |
1618 | */ | |
d5315b50 | 1619 | if (!rt->rt6i_nexthop && !(rt->rt6i_flags & RTF_NONEXTHOP)) |
a1e78363 | 1620 | nrt = rt6_alloc_cow(rt, daddr, saddr); |
d5315b50 YH |
1621 | else |
1622 | nrt = rt6_alloc_clone(rt, daddr); | |
a1e78363 | 1623 | |
d5315b50 | 1624 | if (nrt) { |
defb3519 DM |
1625 | dst_metric_set(&nrt->dst, RTAX_MTU, pmtu); |
1626 | if (allfrag) { | |
1627 | u32 features = dst_metric(&nrt->dst, RTAX_FEATURES); | |
1628 | features |= RTAX_FEATURE_ALLFRAG; | |
1629 | dst_metric_set(&nrt->dst, RTAX_FEATURES, features); | |
1630 | } | |
a1e78363 YH |
1631 | |
1632 | /* According to RFC 1981, detecting PMTU increase shouldn't be | |
1633 | * happened within 5 mins, the recommended timer is 10 mins. | |
1634 | * Here this route expiration time is set to ip6_rt_mtu_expires | |
1635 | * which is 10 mins. After 10 mins the decreased pmtu is expired | |
1636 | * and detecting PMTU increase will be automatically happened. | |
1637 | */ | |
d8d1f30b | 1638 | dst_set_expires(&nrt->dst, net->ipv6.sysctl.ip6_rt_mtu_expires); |
a1e78363 YH |
1639 | nrt->rt6i_flags |= RTF_DYNAMIC|RTF_EXPIRES; |
1640 | ||
40e22e8f | 1641 | ip6_ins_rt(nrt); |
1da177e4 | 1642 | } |
1da177e4 | 1643 | out: |
d8d1f30b | 1644 | dst_release(&rt->dst); |
1da177e4 LT |
1645 | } |
1646 | ||
ae878ae2 MŻ |
1647 | void rt6_pmtu_discovery(struct in6_addr *daddr, struct in6_addr *saddr, |
1648 | struct net_device *dev, u32 pmtu) | |
1649 | { | |
1650 | struct net *net = dev_net(dev); | |
1651 | ||
1652 | /* | |
1653 | * RFC 1981 states that a node "MUST reduce the size of the packets it | |
1654 | * is sending along the path" that caused the Packet Too Big message. | |
1655 | * Since it's not possible in the general case to determine which | |
1656 | * interface was used to send the original packet, we update the MTU | |
1657 | * on the interface that will be used to send future packets. We also | |
1658 | * update the MTU on the interface that received the Packet Too Big in | |
1659 | * case the original packet was forced out that interface with | |
1660 | * SO_BINDTODEVICE or similar. This is the next best thing to the | |
1661 | * correct behaviour, which would be to update the MTU on all | |
1662 | * interfaces. | |
1663 | */ | |
1664 | rt6_do_pmtu_disc(daddr, saddr, net, pmtu, 0); | |
1665 | rt6_do_pmtu_disc(daddr, saddr, net, pmtu, dev->ifindex); | |
1666 | } | |
1667 | ||
1da177e4 LT |
1668 | /* |
1669 | * Misc support functions | |
1670 | */ | |
1671 | ||
1672 | static struct rt6_info * ip6_rt_copy(struct rt6_info *ort) | |
1673 | { | |
c346dca1 | 1674 | struct net *net = dev_net(ort->rt6i_dev); |
86393e52 | 1675 | struct rt6_info *rt = ip6_dst_alloc(&net->ipv6.ip6_dst_ops); |
1da177e4 LT |
1676 | |
1677 | if (rt) { | |
d8d1f30b CG |
1678 | rt->dst.input = ort->dst.input; |
1679 | rt->dst.output = ort->dst.output; | |
1680 | ||
defb3519 | 1681 | dst_copy_metrics(&rt->dst, &ort->dst); |
d8d1f30b CG |
1682 | rt->dst.error = ort->dst.error; |
1683 | rt->dst.dev = ort->dst.dev; | |
1684 | if (rt->dst.dev) | |
1685 | dev_hold(rt->dst.dev); | |
1da177e4 LT |
1686 | rt->rt6i_idev = ort->rt6i_idev; |
1687 | if (rt->rt6i_idev) | |
1688 | in6_dev_hold(rt->rt6i_idev); | |
d8d1f30b | 1689 | rt->dst.lastuse = jiffies; |
1da177e4 LT |
1690 | rt->rt6i_expires = 0; |
1691 | ||
1692 | ipv6_addr_copy(&rt->rt6i_gateway, &ort->rt6i_gateway); | |
1693 | rt->rt6i_flags = ort->rt6i_flags & ~RTF_EXPIRES; | |
1694 | rt->rt6i_metric = 0; | |
1695 | ||
1696 | memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key)); | |
1697 | #ifdef CONFIG_IPV6_SUBTREES | |
1698 | memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key)); | |
1699 | #endif | |
c71099ac | 1700 | rt->rt6i_table = ort->rt6i_table; |
1da177e4 LT |
1701 | } |
1702 | return rt; | |
1703 | } | |
1704 | ||
70ceb4f5 | 1705 | #ifdef CONFIG_IPV6_ROUTE_INFO |
efa2cea0 DL |
1706 | static struct rt6_info *rt6_get_route_info(struct net *net, |
1707 | struct in6_addr *prefix, int prefixlen, | |
70ceb4f5 YH |
1708 | struct in6_addr *gwaddr, int ifindex) |
1709 | { | |
1710 | struct fib6_node *fn; | |
1711 | struct rt6_info *rt = NULL; | |
c71099ac TG |
1712 | struct fib6_table *table; |
1713 | ||
efa2cea0 | 1714 | table = fib6_get_table(net, RT6_TABLE_INFO); |
c71099ac TG |
1715 | if (table == NULL) |
1716 | return NULL; | |
70ceb4f5 | 1717 | |
c71099ac TG |
1718 | write_lock_bh(&table->tb6_lock); |
1719 | fn = fib6_locate(&table->tb6_root, prefix ,prefixlen, NULL, 0); | |
70ceb4f5 YH |
1720 | if (!fn) |
1721 | goto out; | |
1722 | ||
d8d1f30b | 1723 | for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) { |
70ceb4f5 YH |
1724 | if (rt->rt6i_dev->ifindex != ifindex) |
1725 | continue; | |
1726 | if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY)) | |
1727 | continue; | |
1728 | if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr)) | |
1729 | continue; | |
d8d1f30b | 1730 | dst_hold(&rt->dst); |
70ceb4f5 YH |
1731 | break; |
1732 | } | |
1733 | out: | |
c71099ac | 1734 | write_unlock_bh(&table->tb6_lock); |
70ceb4f5 YH |
1735 | return rt; |
1736 | } | |
1737 | ||
efa2cea0 DL |
1738 | static struct rt6_info *rt6_add_route_info(struct net *net, |
1739 | struct in6_addr *prefix, int prefixlen, | |
70ceb4f5 YH |
1740 | struct in6_addr *gwaddr, int ifindex, |
1741 | unsigned pref) | |
1742 | { | |
86872cb5 TG |
1743 | struct fib6_config cfg = { |
1744 | .fc_table = RT6_TABLE_INFO, | |
238fc7ea | 1745 | .fc_metric = IP6_RT_PRIO_USER, |
86872cb5 TG |
1746 | .fc_ifindex = ifindex, |
1747 | .fc_dst_len = prefixlen, | |
1748 | .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO | | |
1749 | RTF_UP | RTF_PREF(pref), | |
efa2cea0 DL |
1750 | .fc_nlinfo.pid = 0, |
1751 | .fc_nlinfo.nlh = NULL, | |
1752 | .fc_nlinfo.nl_net = net, | |
86872cb5 TG |
1753 | }; |
1754 | ||
1755 | ipv6_addr_copy(&cfg.fc_dst, prefix); | |
1756 | ipv6_addr_copy(&cfg.fc_gateway, gwaddr); | |
70ceb4f5 | 1757 | |
e317da96 YH |
1758 | /* We should treat it as a default route if prefix length is 0. */ |
1759 | if (!prefixlen) | |
86872cb5 | 1760 | cfg.fc_flags |= RTF_DEFAULT; |
70ceb4f5 | 1761 | |
86872cb5 | 1762 | ip6_route_add(&cfg); |
70ceb4f5 | 1763 | |
efa2cea0 | 1764 | return rt6_get_route_info(net, prefix, prefixlen, gwaddr, ifindex); |
70ceb4f5 YH |
1765 | } |
1766 | #endif | |
1767 | ||
1da177e4 | 1768 | struct rt6_info *rt6_get_dflt_router(struct in6_addr *addr, struct net_device *dev) |
1ab1457c | 1769 | { |
1da177e4 | 1770 | struct rt6_info *rt; |
c71099ac | 1771 | struct fib6_table *table; |
1da177e4 | 1772 | |
c346dca1 | 1773 | table = fib6_get_table(dev_net(dev), RT6_TABLE_DFLT); |
c71099ac TG |
1774 | if (table == NULL) |
1775 | return NULL; | |
1da177e4 | 1776 | |
c71099ac | 1777 | write_lock_bh(&table->tb6_lock); |
d8d1f30b | 1778 | for (rt = table->tb6_root.leaf; rt; rt=rt->dst.rt6_next) { |
1da177e4 | 1779 | if (dev == rt->rt6i_dev && |
045927ff | 1780 | ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) && |
1da177e4 LT |
1781 | ipv6_addr_equal(&rt->rt6i_gateway, addr)) |
1782 | break; | |
1783 | } | |
1784 | if (rt) | |
d8d1f30b | 1785 | dst_hold(&rt->dst); |
c71099ac | 1786 | write_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
1787 | return rt; |
1788 | } | |
1789 | ||
1790 | struct rt6_info *rt6_add_dflt_router(struct in6_addr *gwaddr, | |
ebacaaa0 YH |
1791 | struct net_device *dev, |
1792 | unsigned int pref) | |
1da177e4 | 1793 | { |
86872cb5 TG |
1794 | struct fib6_config cfg = { |
1795 | .fc_table = RT6_TABLE_DFLT, | |
238fc7ea | 1796 | .fc_metric = IP6_RT_PRIO_USER, |
86872cb5 TG |
1797 | .fc_ifindex = dev->ifindex, |
1798 | .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT | | |
1799 | RTF_UP | RTF_EXPIRES | RTF_PREF(pref), | |
5578689a DL |
1800 | .fc_nlinfo.pid = 0, |
1801 | .fc_nlinfo.nlh = NULL, | |
c346dca1 | 1802 | .fc_nlinfo.nl_net = dev_net(dev), |
86872cb5 | 1803 | }; |
1da177e4 | 1804 | |
86872cb5 | 1805 | ipv6_addr_copy(&cfg.fc_gateway, gwaddr); |
1da177e4 | 1806 | |
86872cb5 | 1807 | ip6_route_add(&cfg); |
1da177e4 | 1808 | |
1da177e4 LT |
1809 | return rt6_get_dflt_router(gwaddr, dev); |
1810 | } | |
1811 | ||
7b4da532 | 1812 | void rt6_purge_dflt_routers(struct net *net) |
1da177e4 LT |
1813 | { |
1814 | struct rt6_info *rt; | |
c71099ac TG |
1815 | struct fib6_table *table; |
1816 | ||
1817 | /* NOTE: Keep consistent with rt6_get_dflt_router */ | |
7b4da532 | 1818 | table = fib6_get_table(net, RT6_TABLE_DFLT); |
c71099ac TG |
1819 | if (table == NULL) |
1820 | return; | |
1da177e4 LT |
1821 | |
1822 | restart: | |
c71099ac | 1823 | read_lock_bh(&table->tb6_lock); |
d8d1f30b | 1824 | for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) { |
1da177e4 | 1825 | if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF)) { |
d8d1f30b | 1826 | dst_hold(&rt->dst); |
c71099ac | 1827 | read_unlock_bh(&table->tb6_lock); |
e0a1ad73 | 1828 | ip6_del_rt(rt); |
1da177e4 LT |
1829 | goto restart; |
1830 | } | |
1831 | } | |
c71099ac | 1832 | read_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
1833 | } |
1834 | ||
5578689a DL |
1835 | static void rtmsg_to_fib6_config(struct net *net, |
1836 | struct in6_rtmsg *rtmsg, | |
86872cb5 TG |
1837 | struct fib6_config *cfg) |
1838 | { | |
1839 | memset(cfg, 0, sizeof(*cfg)); | |
1840 | ||
1841 | cfg->fc_table = RT6_TABLE_MAIN; | |
1842 | cfg->fc_ifindex = rtmsg->rtmsg_ifindex; | |
1843 | cfg->fc_metric = rtmsg->rtmsg_metric; | |
1844 | cfg->fc_expires = rtmsg->rtmsg_info; | |
1845 | cfg->fc_dst_len = rtmsg->rtmsg_dst_len; | |
1846 | cfg->fc_src_len = rtmsg->rtmsg_src_len; | |
1847 | cfg->fc_flags = rtmsg->rtmsg_flags; | |
1848 | ||
5578689a | 1849 | cfg->fc_nlinfo.nl_net = net; |
f1243c2d | 1850 | |
86872cb5 TG |
1851 | ipv6_addr_copy(&cfg->fc_dst, &rtmsg->rtmsg_dst); |
1852 | ipv6_addr_copy(&cfg->fc_src, &rtmsg->rtmsg_src); | |
1853 | ipv6_addr_copy(&cfg->fc_gateway, &rtmsg->rtmsg_gateway); | |
1854 | } | |
1855 | ||
5578689a | 1856 | int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg) |
1da177e4 | 1857 | { |
86872cb5 | 1858 | struct fib6_config cfg; |
1da177e4 LT |
1859 | struct in6_rtmsg rtmsg; |
1860 | int err; | |
1861 | ||
1862 | switch(cmd) { | |
1863 | case SIOCADDRT: /* Add a route */ | |
1864 | case SIOCDELRT: /* Delete a route */ | |
1865 | if (!capable(CAP_NET_ADMIN)) | |
1866 | return -EPERM; | |
1867 | err = copy_from_user(&rtmsg, arg, | |
1868 | sizeof(struct in6_rtmsg)); | |
1869 | if (err) | |
1870 | return -EFAULT; | |
86872cb5 | 1871 | |
5578689a | 1872 | rtmsg_to_fib6_config(net, &rtmsg, &cfg); |
86872cb5 | 1873 | |
1da177e4 LT |
1874 | rtnl_lock(); |
1875 | switch (cmd) { | |
1876 | case SIOCADDRT: | |
86872cb5 | 1877 | err = ip6_route_add(&cfg); |
1da177e4 LT |
1878 | break; |
1879 | case SIOCDELRT: | |
86872cb5 | 1880 | err = ip6_route_del(&cfg); |
1da177e4 LT |
1881 | break; |
1882 | default: | |
1883 | err = -EINVAL; | |
1884 | } | |
1885 | rtnl_unlock(); | |
1886 | ||
1887 | return err; | |
3ff50b79 | 1888 | } |
1da177e4 LT |
1889 | |
1890 | return -EINVAL; | |
1891 | } | |
1892 | ||
1893 | /* | |
1894 | * Drop the packet on the floor | |
1895 | */ | |
1896 | ||
d5fdd6ba | 1897 | static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes) |
1da177e4 | 1898 | { |
612f09e8 | 1899 | int type; |
adf30907 | 1900 | struct dst_entry *dst = skb_dst(skb); |
612f09e8 YH |
1901 | switch (ipstats_mib_noroutes) { |
1902 | case IPSTATS_MIB_INNOROUTES: | |
0660e03f | 1903 | type = ipv6_addr_type(&ipv6_hdr(skb)->daddr); |
45bb0060 | 1904 | if (type == IPV6_ADDR_ANY) { |
3bd653c8 DL |
1905 | IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst), |
1906 | IPSTATS_MIB_INADDRERRORS); | |
612f09e8 YH |
1907 | break; |
1908 | } | |
1909 | /* FALLTHROUGH */ | |
1910 | case IPSTATS_MIB_OUTNOROUTES: | |
3bd653c8 DL |
1911 | IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst), |
1912 | ipstats_mib_noroutes); | |
612f09e8 YH |
1913 | break; |
1914 | } | |
3ffe533c | 1915 | icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0); |
1da177e4 LT |
1916 | kfree_skb(skb); |
1917 | return 0; | |
1918 | } | |
1919 | ||
9ce8ade0 TG |
1920 | static int ip6_pkt_discard(struct sk_buff *skb) |
1921 | { | |
612f09e8 | 1922 | return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES); |
9ce8ade0 TG |
1923 | } |
1924 | ||
20380731 | 1925 | static int ip6_pkt_discard_out(struct sk_buff *skb) |
1da177e4 | 1926 | { |
adf30907 | 1927 | skb->dev = skb_dst(skb)->dev; |
612f09e8 | 1928 | return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES); |
1da177e4 LT |
1929 | } |
1930 | ||
6723ab54 DM |
1931 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES |
1932 | ||
9ce8ade0 TG |
1933 | static int ip6_pkt_prohibit(struct sk_buff *skb) |
1934 | { | |
612f09e8 | 1935 | return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES); |
9ce8ade0 TG |
1936 | } |
1937 | ||
1938 | static int ip6_pkt_prohibit_out(struct sk_buff *skb) | |
1939 | { | |
adf30907 | 1940 | skb->dev = skb_dst(skb)->dev; |
612f09e8 | 1941 | return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES); |
9ce8ade0 TG |
1942 | } |
1943 | ||
6723ab54 DM |
1944 | #endif |
1945 | ||
1da177e4 LT |
1946 | /* |
1947 | * Allocate a dst for local (unicast / anycast) address. | |
1948 | */ | |
1949 | ||
1950 | struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev, | |
1951 | const struct in6_addr *addr, | |
1952 | int anycast) | |
1953 | { | |
c346dca1 | 1954 | struct net *net = dev_net(idev->dev); |
86393e52 | 1955 | struct rt6_info *rt = ip6_dst_alloc(&net->ipv6.ip6_dst_ops); |
14deae41 | 1956 | struct neighbour *neigh; |
1da177e4 | 1957 | |
40385653 BG |
1958 | if (rt == NULL) { |
1959 | if (net_ratelimit()) | |
1960 | pr_warning("IPv6: Maximum number of routes reached," | |
1961 | " consider increasing route/max_size.\n"); | |
1da177e4 | 1962 | return ERR_PTR(-ENOMEM); |
40385653 | 1963 | } |
1da177e4 | 1964 | |
5578689a | 1965 | dev_hold(net->loopback_dev); |
1da177e4 LT |
1966 | in6_dev_hold(idev); |
1967 | ||
d8d1f30b CG |
1968 | rt->dst.flags = DST_HOST; |
1969 | rt->dst.input = ip6_input; | |
1970 | rt->dst.output = ip6_output; | |
5578689a | 1971 | rt->rt6i_dev = net->loopback_dev; |
1da177e4 | 1972 | rt->rt6i_idev = idev; |
defb3519 DM |
1973 | dst_metric_set(&rt->dst, RTAX_MTU, ipv6_get_mtu(rt->rt6i_dev)); |
1974 | dst_metric_set(&rt->dst, RTAX_ADVMSS, ipv6_advmss(net, dst_mtu(&rt->dst))); | |
1975 | dst_metric_set(&rt->dst, RTAX_HOPLIMIT, -1); | |
d8d1f30b | 1976 | rt->dst.obsolete = -1; |
1da177e4 LT |
1977 | |
1978 | rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP; | |
58c4fb86 YH |
1979 | if (anycast) |
1980 | rt->rt6i_flags |= RTF_ANYCAST; | |
1981 | else | |
1da177e4 | 1982 | rt->rt6i_flags |= RTF_LOCAL; |
14deae41 DM |
1983 | neigh = ndisc_get_neigh(rt->rt6i_dev, &rt->rt6i_gateway); |
1984 | if (IS_ERR(neigh)) { | |
d8d1f30b | 1985 | dst_free(&rt->dst); |
14deae41 DM |
1986 | |
1987 | /* We are casting this because that is the return | |
1988 | * value type. But an errno encoded pointer is the | |
1989 | * same regardless of the underlying pointer type, | |
1990 | * and that's what we are returning. So this is OK. | |
1991 | */ | |
1992 | return (struct rt6_info *) neigh; | |
1da177e4 | 1993 | } |
14deae41 | 1994 | rt->rt6i_nexthop = neigh; |
1da177e4 LT |
1995 | |
1996 | ipv6_addr_copy(&rt->rt6i_dst.addr, addr); | |
1997 | rt->rt6i_dst.plen = 128; | |
5578689a | 1998 | rt->rt6i_table = fib6_get_table(net, RT6_TABLE_LOCAL); |
1da177e4 | 1999 | |
d8d1f30b | 2000 | atomic_set(&rt->dst.__refcnt, 1); |
1da177e4 LT |
2001 | |
2002 | return rt; | |
2003 | } | |
2004 | ||
8ed67789 DL |
2005 | struct arg_dev_net { |
2006 | struct net_device *dev; | |
2007 | struct net *net; | |
2008 | }; | |
2009 | ||
1da177e4 LT |
2010 | static int fib6_ifdown(struct rt6_info *rt, void *arg) |
2011 | { | |
8ed67789 DL |
2012 | struct net_device *dev = ((struct arg_dev_net *)arg)->dev; |
2013 | struct net *net = ((struct arg_dev_net *)arg)->net; | |
2014 | ||
2015 | if (((void *)rt->rt6i_dev == dev || dev == NULL) && | |
2016 | rt != net->ipv6.ip6_null_entry) { | |
1da177e4 LT |
2017 | RT6_TRACE("deleted by ifdown %p\n", rt); |
2018 | return -1; | |
2019 | } | |
2020 | return 0; | |
2021 | } | |
2022 | ||
f3db4851 | 2023 | void rt6_ifdown(struct net *net, struct net_device *dev) |
1da177e4 | 2024 | { |
8ed67789 DL |
2025 | struct arg_dev_net adn = { |
2026 | .dev = dev, | |
2027 | .net = net, | |
2028 | }; | |
2029 | ||
2030 | fib6_clean_all(net, fib6_ifdown, 0, &adn); | |
1e493d19 | 2031 | icmp6_clean_all(fib6_ifdown, &adn); |
1da177e4 LT |
2032 | } |
2033 | ||
2034 | struct rt6_mtu_change_arg | |
2035 | { | |
2036 | struct net_device *dev; | |
2037 | unsigned mtu; | |
2038 | }; | |
2039 | ||
2040 | static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg) | |
2041 | { | |
2042 | struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg; | |
2043 | struct inet6_dev *idev; | |
c346dca1 | 2044 | struct net *net = dev_net(arg->dev); |
1da177e4 LT |
2045 | |
2046 | /* In IPv6 pmtu discovery is not optional, | |
2047 | so that RTAX_MTU lock cannot disable it. | |
2048 | We still use this lock to block changes | |
2049 | caused by addrconf/ndisc. | |
2050 | */ | |
2051 | ||
2052 | idev = __in6_dev_get(arg->dev); | |
2053 | if (idev == NULL) | |
2054 | return 0; | |
2055 | ||
2056 | /* For administrative MTU increase, there is no way to discover | |
2057 | IPv6 PMTU increase, so PMTU increase should be updated here. | |
2058 | Since RFC 1981 doesn't include administrative MTU increase | |
2059 | update PMTU increase is a MUST. (i.e. jumbo frame) | |
2060 | */ | |
2061 | /* | |
2062 | If new MTU is less than route PMTU, this new MTU will be the | |
2063 | lowest MTU in the path, update the route PMTU to reflect PMTU | |
2064 | decreases; if new MTU is greater than route PMTU, and the | |
2065 | old MTU is the lowest MTU in the path, update the route PMTU | |
2066 | to reflect the increase. In this case if the other nodes' MTU | |
2067 | also have the lowest MTU, TOO BIG MESSAGE will be lead to | |
2068 | PMTU discouvery. | |
2069 | */ | |
2070 | if (rt->rt6i_dev == arg->dev && | |
d8d1f30b CG |
2071 | !dst_metric_locked(&rt->dst, RTAX_MTU) && |
2072 | (dst_mtu(&rt->dst) >= arg->mtu || | |
2073 | (dst_mtu(&rt->dst) < arg->mtu && | |
2074 | dst_mtu(&rt->dst) == idev->cnf.mtu6))) { | |
defb3519 DM |
2075 | dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu); |
2076 | dst_metric_set(&rt->dst, RTAX_ADVMSS, ipv6_advmss(net, arg->mtu)); | |
566cfd8f | 2077 | } |
1da177e4 LT |
2078 | return 0; |
2079 | } | |
2080 | ||
2081 | void rt6_mtu_change(struct net_device *dev, unsigned mtu) | |
2082 | { | |
c71099ac TG |
2083 | struct rt6_mtu_change_arg arg = { |
2084 | .dev = dev, | |
2085 | .mtu = mtu, | |
2086 | }; | |
1da177e4 | 2087 | |
c346dca1 | 2088 | fib6_clean_all(dev_net(dev), rt6_mtu_change_route, 0, &arg); |
1da177e4 LT |
2089 | } |
2090 | ||
ef7c79ed | 2091 | static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = { |
5176f91e | 2092 | [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) }, |
86872cb5 | 2093 | [RTA_OIF] = { .type = NLA_U32 }, |
ab364a6f | 2094 | [RTA_IIF] = { .type = NLA_U32 }, |
86872cb5 TG |
2095 | [RTA_PRIORITY] = { .type = NLA_U32 }, |
2096 | [RTA_METRICS] = { .type = NLA_NESTED }, | |
2097 | }; | |
2098 | ||
2099 | static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh, | |
2100 | struct fib6_config *cfg) | |
1da177e4 | 2101 | { |
86872cb5 TG |
2102 | struct rtmsg *rtm; |
2103 | struct nlattr *tb[RTA_MAX+1]; | |
2104 | int err; | |
1da177e4 | 2105 | |
86872cb5 TG |
2106 | err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy); |
2107 | if (err < 0) | |
2108 | goto errout; | |
1da177e4 | 2109 | |
86872cb5 TG |
2110 | err = -EINVAL; |
2111 | rtm = nlmsg_data(nlh); | |
2112 | memset(cfg, 0, sizeof(*cfg)); | |
2113 | ||
2114 | cfg->fc_table = rtm->rtm_table; | |
2115 | cfg->fc_dst_len = rtm->rtm_dst_len; | |
2116 | cfg->fc_src_len = rtm->rtm_src_len; | |
2117 | cfg->fc_flags = RTF_UP; | |
2118 | cfg->fc_protocol = rtm->rtm_protocol; | |
2119 | ||
2120 | if (rtm->rtm_type == RTN_UNREACHABLE) | |
2121 | cfg->fc_flags |= RTF_REJECT; | |
2122 | ||
ab79ad14 MŻ |
2123 | if (rtm->rtm_type == RTN_LOCAL) |
2124 | cfg->fc_flags |= RTF_LOCAL; | |
2125 | ||
86872cb5 TG |
2126 | cfg->fc_nlinfo.pid = NETLINK_CB(skb).pid; |
2127 | cfg->fc_nlinfo.nlh = nlh; | |
3b1e0a65 | 2128 | cfg->fc_nlinfo.nl_net = sock_net(skb->sk); |
86872cb5 TG |
2129 | |
2130 | if (tb[RTA_GATEWAY]) { | |
2131 | nla_memcpy(&cfg->fc_gateway, tb[RTA_GATEWAY], 16); | |
2132 | cfg->fc_flags |= RTF_GATEWAY; | |
1da177e4 | 2133 | } |
86872cb5 TG |
2134 | |
2135 | if (tb[RTA_DST]) { | |
2136 | int plen = (rtm->rtm_dst_len + 7) >> 3; | |
2137 | ||
2138 | if (nla_len(tb[RTA_DST]) < plen) | |
2139 | goto errout; | |
2140 | ||
2141 | nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen); | |
1da177e4 | 2142 | } |
86872cb5 TG |
2143 | |
2144 | if (tb[RTA_SRC]) { | |
2145 | int plen = (rtm->rtm_src_len + 7) >> 3; | |
2146 | ||
2147 | if (nla_len(tb[RTA_SRC]) < plen) | |
2148 | goto errout; | |
2149 | ||
2150 | nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen); | |
1da177e4 | 2151 | } |
86872cb5 TG |
2152 | |
2153 | if (tb[RTA_OIF]) | |
2154 | cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]); | |
2155 | ||
2156 | if (tb[RTA_PRIORITY]) | |
2157 | cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]); | |
2158 | ||
2159 | if (tb[RTA_METRICS]) { | |
2160 | cfg->fc_mx = nla_data(tb[RTA_METRICS]); | |
2161 | cfg->fc_mx_len = nla_len(tb[RTA_METRICS]); | |
1da177e4 | 2162 | } |
86872cb5 TG |
2163 | |
2164 | if (tb[RTA_TABLE]) | |
2165 | cfg->fc_table = nla_get_u32(tb[RTA_TABLE]); | |
2166 | ||
2167 | err = 0; | |
2168 | errout: | |
2169 | return err; | |
1da177e4 LT |
2170 | } |
2171 | ||
c127ea2c | 2172 | static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg) |
1da177e4 | 2173 | { |
86872cb5 TG |
2174 | struct fib6_config cfg; |
2175 | int err; | |
1da177e4 | 2176 | |
86872cb5 TG |
2177 | err = rtm_to_fib6_config(skb, nlh, &cfg); |
2178 | if (err < 0) | |
2179 | return err; | |
2180 | ||
2181 | return ip6_route_del(&cfg); | |
1da177e4 LT |
2182 | } |
2183 | ||
c127ea2c | 2184 | static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg) |
1da177e4 | 2185 | { |
86872cb5 TG |
2186 | struct fib6_config cfg; |
2187 | int err; | |
1da177e4 | 2188 | |
86872cb5 TG |
2189 | err = rtm_to_fib6_config(skb, nlh, &cfg); |
2190 | if (err < 0) | |
2191 | return err; | |
2192 | ||
2193 | return ip6_route_add(&cfg); | |
1da177e4 LT |
2194 | } |
2195 | ||
339bf98f TG |
2196 | static inline size_t rt6_nlmsg_size(void) |
2197 | { | |
2198 | return NLMSG_ALIGN(sizeof(struct rtmsg)) | |
2199 | + nla_total_size(16) /* RTA_SRC */ | |
2200 | + nla_total_size(16) /* RTA_DST */ | |
2201 | + nla_total_size(16) /* RTA_GATEWAY */ | |
2202 | + nla_total_size(16) /* RTA_PREFSRC */ | |
2203 | + nla_total_size(4) /* RTA_TABLE */ | |
2204 | + nla_total_size(4) /* RTA_IIF */ | |
2205 | + nla_total_size(4) /* RTA_OIF */ | |
2206 | + nla_total_size(4) /* RTA_PRIORITY */ | |
6a2b9ce0 | 2207 | + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */ |
339bf98f TG |
2208 | + nla_total_size(sizeof(struct rta_cacheinfo)); |
2209 | } | |
2210 | ||
191cd582 BH |
2211 | static int rt6_fill_node(struct net *net, |
2212 | struct sk_buff *skb, struct rt6_info *rt, | |
0d51aa80 JHS |
2213 | struct in6_addr *dst, struct in6_addr *src, |
2214 | int iif, int type, u32 pid, u32 seq, | |
7bc570c8 | 2215 | int prefix, int nowait, unsigned int flags) |
1da177e4 LT |
2216 | { |
2217 | struct rtmsg *rtm; | |
2d7202bf | 2218 | struct nlmsghdr *nlh; |
e3703b3d | 2219 | long expires; |
9e762a4a | 2220 | u32 table; |
1da177e4 LT |
2221 | |
2222 | if (prefix) { /* user wants prefix routes only */ | |
2223 | if (!(rt->rt6i_flags & RTF_PREFIX_RT)) { | |
2224 | /* success since this is not a prefix route */ | |
2225 | return 1; | |
2226 | } | |
2227 | } | |
2228 | ||
2d7202bf TG |
2229 | nlh = nlmsg_put(skb, pid, seq, type, sizeof(*rtm), flags); |
2230 | if (nlh == NULL) | |
26932566 | 2231 | return -EMSGSIZE; |
2d7202bf TG |
2232 | |
2233 | rtm = nlmsg_data(nlh); | |
1da177e4 LT |
2234 | rtm->rtm_family = AF_INET6; |
2235 | rtm->rtm_dst_len = rt->rt6i_dst.plen; | |
2236 | rtm->rtm_src_len = rt->rt6i_src.plen; | |
2237 | rtm->rtm_tos = 0; | |
c71099ac | 2238 | if (rt->rt6i_table) |
9e762a4a | 2239 | table = rt->rt6i_table->tb6_id; |
c71099ac | 2240 | else |
9e762a4a PM |
2241 | table = RT6_TABLE_UNSPEC; |
2242 | rtm->rtm_table = table; | |
2d7202bf | 2243 | NLA_PUT_U32(skb, RTA_TABLE, table); |
1da177e4 LT |
2244 | if (rt->rt6i_flags&RTF_REJECT) |
2245 | rtm->rtm_type = RTN_UNREACHABLE; | |
ab79ad14 MŻ |
2246 | else if (rt->rt6i_flags&RTF_LOCAL) |
2247 | rtm->rtm_type = RTN_LOCAL; | |
1da177e4 LT |
2248 | else if (rt->rt6i_dev && (rt->rt6i_dev->flags&IFF_LOOPBACK)) |
2249 | rtm->rtm_type = RTN_LOCAL; | |
2250 | else | |
2251 | rtm->rtm_type = RTN_UNICAST; | |
2252 | rtm->rtm_flags = 0; | |
2253 | rtm->rtm_scope = RT_SCOPE_UNIVERSE; | |
2254 | rtm->rtm_protocol = rt->rt6i_protocol; | |
2255 | if (rt->rt6i_flags&RTF_DYNAMIC) | |
2256 | rtm->rtm_protocol = RTPROT_REDIRECT; | |
2257 | else if (rt->rt6i_flags & RTF_ADDRCONF) | |
2258 | rtm->rtm_protocol = RTPROT_KERNEL; | |
2259 | else if (rt->rt6i_flags&RTF_DEFAULT) | |
2260 | rtm->rtm_protocol = RTPROT_RA; | |
2261 | ||
2262 | if (rt->rt6i_flags&RTF_CACHE) | |
2263 | rtm->rtm_flags |= RTM_F_CLONED; | |
2264 | ||
2265 | if (dst) { | |
2d7202bf | 2266 | NLA_PUT(skb, RTA_DST, 16, dst); |
1ab1457c | 2267 | rtm->rtm_dst_len = 128; |
1da177e4 | 2268 | } else if (rtm->rtm_dst_len) |
2d7202bf | 2269 | NLA_PUT(skb, RTA_DST, 16, &rt->rt6i_dst.addr); |
1da177e4 LT |
2270 | #ifdef CONFIG_IPV6_SUBTREES |
2271 | if (src) { | |
2d7202bf | 2272 | NLA_PUT(skb, RTA_SRC, 16, src); |
1ab1457c | 2273 | rtm->rtm_src_len = 128; |
1da177e4 | 2274 | } else if (rtm->rtm_src_len) |
2d7202bf | 2275 | NLA_PUT(skb, RTA_SRC, 16, &rt->rt6i_src.addr); |
1da177e4 | 2276 | #endif |
7bc570c8 YH |
2277 | if (iif) { |
2278 | #ifdef CONFIG_IPV6_MROUTE | |
2279 | if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) { | |
8229efda | 2280 | int err = ip6mr_get_route(net, skb, rtm, nowait); |
7bc570c8 YH |
2281 | if (err <= 0) { |
2282 | if (!nowait) { | |
2283 | if (err == 0) | |
2284 | return 0; | |
2285 | goto nla_put_failure; | |
2286 | } else { | |
2287 | if (err == -EMSGSIZE) | |
2288 | goto nla_put_failure; | |
2289 | } | |
2290 | } | |
2291 | } else | |
2292 | #endif | |
2293 | NLA_PUT_U32(skb, RTA_IIF, iif); | |
2294 | } else if (dst) { | |
d8d1f30b | 2295 | struct inet6_dev *idev = ip6_dst_idev(&rt->dst); |
1da177e4 | 2296 | struct in6_addr saddr_buf; |
191cd582 | 2297 | if (ipv6_dev_get_saddr(net, idev ? idev->dev : NULL, |
7cbca67c | 2298 | dst, 0, &saddr_buf) == 0) |
2d7202bf | 2299 | NLA_PUT(skb, RTA_PREFSRC, 16, &saddr_buf); |
1da177e4 | 2300 | } |
2d7202bf | 2301 | |
defb3519 | 2302 | if (rtnetlink_put_metrics(skb, dst_metrics_ptr(&rt->dst)) < 0) |
2d7202bf TG |
2303 | goto nla_put_failure; |
2304 | ||
d8d1f30b CG |
2305 | if (rt->dst.neighbour) |
2306 | NLA_PUT(skb, RTA_GATEWAY, 16, &rt->dst.neighbour->primary_key); | |
2d7202bf | 2307 | |
d8d1f30b | 2308 | if (rt->dst.dev) |
2d7202bf TG |
2309 | NLA_PUT_U32(skb, RTA_OIF, rt->rt6i_dev->ifindex); |
2310 | ||
2311 | NLA_PUT_U32(skb, RTA_PRIORITY, rt->rt6i_metric); | |
e3703b3d | 2312 | |
36e3deae YH |
2313 | if (!(rt->rt6i_flags & RTF_EXPIRES)) |
2314 | expires = 0; | |
2315 | else if (rt->rt6i_expires - jiffies < INT_MAX) | |
2316 | expires = rt->rt6i_expires - jiffies; | |
2317 | else | |
2318 | expires = INT_MAX; | |
69cdf8f9 | 2319 | |
d8d1f30b CG |
2320 | if (rtnl_put_cacheinfo(skb, &rt->dst, 0, 0, 0, |
2321 | expires, rt->dst.error) < 0) | |
e3703b3d | 2322 | goto nla_put_failure; |
2d7202bf TG |
2323 | |
2324 | return nlmsg_end(skb, nlh); | |
2325 | ||
2326 | nla_put_failure: | |
26932566 PM |
2327 | nlmsg_cancel(skb, nlh); |
2328 | return -EMSGSIZE; | |
1da177e4 LT |
2329 | } |
2330 | ||
1b43af54 | 2331 | int rt6_dump_route(struct rt6_info *rt, void *p_arg) |
1da177e4 LT |
2332 | { |
2333 | struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg; | |
2334 | int prefix; | |
2335 | ||
2d7202bf TG |
2336 | if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) { |
2337 | struct rtmsg *rtm = nlmsg_data(arg->cb->nlh); | |
1da177e4 LT |
2338 | prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0; |
2339 | } else | |
2340 | prefix = 0; | |
2341 | ||
191cd582 BH |
2342 | return rt6_fill_node(arg->net, |
2343 | arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE, | |
1da177e4 | 2344 | NETLINK_CB(arg->cb->skb).pid, arg->cb->nlh->nlmsg_seq, |
7bc570c8 | 2345 | prefix, 0, NLM_F_MULTI); |
1da177e4 LT |
2346 | } |
2347 | ||
c127ea2c | 2348 | static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr* nlh, void *arg) |
1da177e4 | 2349 | { |
3b1e0a65 | 2350 | struct net *net = sock_net(in_skb->sk); |
ab364a6f TG |
2351 | struct nlattr *tb[RTA_MAX+1]; |
2352 | struct rt6_info *rt; | |
1da177e4 | 2353 | struct sk_buff *skb; |
ab364a6f | 2354 | struct rtmsg *rtm; |
1da177e4 | 2355 | struct flowi fl; |
ab364a6f | 2356 | int err, iif = 0; |
1da177e4 | 2357 | |
ab364a6f TG |
2358 | err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy); |
2359 | if (err < 0) | |
2360 | goto errout; | |
1da177e4 | 2361 | |
ab364a6f | 2362 | err = -EINVAL; |
1da177e4 | 2363 | memset(&fl, 0, sizeof(fl)); |
1da177e4 | 2364 | |
ab364a6f TG |
2365 | if (tb[RTA_SRC]) { |
2366 | if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr)) | |
2367 | goto errout; | |
2368 | ||
2369 | ipv6_addr_copy(&fl.fl6_src, nla_data(tb[RTA_SRC])); | |
2370 | } | |
2371 | ||
2372 | if (tb[RTA_DST]) { | |
2373 | if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr)) | |
2374 | goto errout; | |
2375 | ||
2376 | ipv6_addr_copy(&fl.fl6_dst, nla_data(tb[RTA_DST])); | |
2377 | } | |
2378 | ||
2379 | if (tb[RTA_IIF]) | |
2380 | iif = nla_get_u32(tb[RTA_IIF]); | |
2381 | ||
2382 | if (tb[RTA_OIF]) | |
2383 | fl.oif = nla_get_u32(tb[RTA_OIF]); | |
1da177e4 LT |
2384 | |
2385 | if (iif) { | |
2386 | struct net_device *dev; | |
5578689a | 2387 | dev = __dev_get_by_index(net, iif); |
1da177e4 LT |
2388 | if (!dev) { |
2389 | err = -ENODEV; | |
ab364a6f | 2390 | goto errout; |
1da177e4 LT |
2391 | } |
2392 | } | |
2393 | ||
ab364a6f TG |
2394 | skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL); |
2395 | if (skb == NULL) { | |
2396 | err = -ENOBUFS; | |
2397 | goto errout; | |
2398 | } | |
1da177e4 | 2399 | |
ab364a6f TG |
2400 | /* Reserve room for dummy headers, this skb can pass |
2401 | through good chunk of routing engine. | |
2402 | */ | |
459a98ed | 2403 | skb_reset_mac_header(skb); |
ab364a6f | 2404 | skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr)); |
1da177e4 | 2405 | |
8a3edd80 | 2406 | rt = (struct rt6_info*) ip6_route_output(net, NULL, &fl); |
d8d1f30b | 2407 | skb_dst_set(skb, &rt->dst); |
1da177e4 | 2408 | |
191cd582 | 2409 | err = rt6_fill_node(net, skb, rt, &fl.fl6_dst, &fl.fl6_src, iif, |
1da177e4 | 2410 | RTM_NEWROUTE, NETLINK_CB(in_skb).pid, |
7bc570c8 | 2411 | nlh->nlmsg_seq, 0, 0, 0); |
1da177e4 | 2412 | if (err < 0) { |
ab364a6f TG |
2413 | kfree_skb(skb); |
2414 | goto errout; | |
1da177e4 LT |
2415 | } |
2416 | ||
5578689a | 2417 | err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).pid); |
ab364a6f | 2418 | errout: |
1da177e4 | 2419 | return err; |
1da177e4 LT |
2420 | } |
2421 | ||
86872cb5 | 2422 | void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info) |
1da177e4 LT |
2423 | { |
2424 | struct sk_buff *skb; | |
5578689a | 2425 | struct net *net = info->nl_net; |
528c4ceb DL |
2426 | u32 seq; |
2427 | int err; | |
2428 | ||
2429 | err = -ENOBUFS; | |
2430 | seq = info->nlh != NULL ? info->nlh->nlmsg_seq : 0; | |
86872cb5 | 2431 | |
339bf98f | 2432 | skb = nlmsg_new(rt6_nlmsg_size(), gfp_any()); |
21713ebc TG |
2433 | if (skb == NULL) |
2434 | goto errout; | |
2435 | ||
191cd582 | 2436 | err = rt6_fill_node(net, skb, rt, NULL, NULL, 0, |
7bc570c8 | 2437 | event, info->pid, seq, 0, 0, 0); |
26932566 PM |
2438 | if (err < 0) { |
2439 | /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */ | |
2440 | WARN_ON(err == -EMSGSIZE); | |
2441 | kfree_skb(skb); | |
2442 | goto errout; | |
2443 | } | |
1ce85fe4 PNA |
2444 | rtnl_notify(skb, net, info->pid, RTNLGRP_IPV6_ROUTE, |
2445 | info->nlh, gfp_any()); | |
2446 | return; | |
21713ebc TG |
2447 | errout: |
2448 | if (err < 0) | |
5578689a | 2449 | rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err); |
1da177e4 LT |
2450 | } |
2451 | ||
8ed67789 DL |
2452 | static int ip6_route_dev_notify(struct notifier_block *this, |
2453 | unsigned long event, void *data) | |
2454 | { | |
2455 | struct net_device *dev = (struct net_device *)data; | |
c346dca1 | 2456 | struct net *net = dev_net(dev); |
8ed67789 DL |
2457 | |
2458 | if (event == NETDEV_REGISTER && (dev->flags & IFF_LOOPBACK)) { | |
d8d1f30b | 2459 | net->ipv6.ip6_null_entry->dst.dev = dev; |
8ed67789 DL |
2460 | net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev); |
2461 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES | |
d8d1f30b | 2462 | net->ipv6.ip6_prohibit_entry->dst.dev = dev; |
8ed67789 | 2463 | net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev); |
d8d1f30b | 2464 | net->ipv6.ip6_blk_hole_entry->dst.dev = dev; |
8ed67789 DL |
2465 | net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev); |
2466 | #endif | |
2467 | } | |
2468 | ||
2469 | return NOTIFY_OK; | |
2470 | } | |
2471 | ||
1da177e4 LT |
2472 | /* |
2473 | * /proc | |
2474 | */ | |
2475 | ||
2476 | #ifdef CONFIG_PROC_FS | |
2477 | ||
1da177e4 LT |
2478 | struct rt6_proc_arg |
2479 | { | |
2480 | char *buffer; | |
2481 | int offset; | |
2482 | int length; | |
2483 | int skip; | |
2484 | int len; | |
2485 | }; | |
2486 | ||
2487 | static int rt6_info_route(struct rt6_info *rt, void *p_arg) | |
2488 | { | |
33120b30 | 2489 | struct seq_file *m = p_arg; |
1da177e4 | 2490 | |
4b7a4274 | 2491 | seq_printf(m, "%pi6 %02x ", &rt->rt6i_dst.addr, rt->rt6i_dst.plen); |
1da177e4 LT |
2492 | |
2493 | #ifdef CONFIG_IPV6_SUBTREES | |
4b7a4274 | 2494 | seq_printf(m, "%pi6 %02x ", &rt->rt6i_src.addr, rt->rt6i_src.plen); |
1da177e4 | 2495 | #else |
33120b30 | 2496 | seq_puts(m, "00000000000000000000000000000000 00 "); |
1da177e4 LT |
2497 | #endif |
2498 | ||
2499 | if (rt->rt6i_nexthop) { | |
4b7a4274 | 2500 | seq_printf(m, "%pi6", rt->rt6i_nexthop->primary_key); |
1da177e4 | 2501 | } else { |
33120b30 | 2502 | seq_puts(m, "00000000000000000000000000000000"); |
1da177e4 | 2503 | } |
33120b30 | 2504 | seq_printf(m, " %08x %08x %08x %08x %8s\n", |
d8d1f30b CG |
2505 | rt->rt6i_metric, atomic_read(&rt->dst.__refcnt), |
2506 | rt->dst.__use, rt->rt6i_flags, | |
33120b30 | 2507 | rt->rt6i_dev ? rt->rt6i_dev->name : ""); |
1da177e4 LT |
2508 | return 0; |
2509 | } | |
2510 | ||
33120b30 | 2511 | static int ipv6_route_show(struct seq_file *m, void *v) |
1da177e4 | 2512 | { |
f3db4851 DL |
2513 | struct net *net = (struct net *)m->private; |
2514 | fib6_clean_all(net, rt6_info_route, 0, m); | |
33120b30 AD |
2515 | return 0; |
2516 | } | |
1da177e4 | 2517 | |
33120b30 AD |
2518 | static int ipv6_route_open(struct inode *inode, struct file *file) |
2519 | { | |
de05c557 | 2520 | return single_open_net(inode, file, ipv6_route_show); |
f3db4851 DL |
2521 | } |
2522 | ||
33120b30 AD |
2523 | static const struct file_operations ipv6_route_proc_fops = { |
2524 | .owner = THIS_MODULE, | |
2525 | .open = ipv6_route_open, | |
2526 | .read = seq_read, | |
2527 | .llseek = seq_lseek, | |
b6fcbdb4 | 2528 | .release = single_release_net, |
33120b30 AD |
2529 | }; |
2530 | ||
1da177e4 LT |
2531 | static int rt6_stats_seq_show(struct seq_file *seq, void *v) |
2532 | { | |
69ddb805 | 2533 | struct net *net = (struct net *)seq->private; |
1da177e4 | 2534 | seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n", |
69ddb805 DL |
2535 | net->ipv6.rt6_stats->fib_nodes, |
2536 | net->ipv6.rt6_stats->fib_route_nodes, | |
2537 | net->ipv6.rt6_stats->fib_rt_alloc, | |
2538 | net->ipv6.rt6_stats->fib_rt_entries, | |
2539 | net->ipv6.rt6_stats->fib_rt_cache, | |
fc66f95c | 2540 | dst_entries_get_slow(&net->ipv6.ip6_dst_ops), |
69ddb805 | 2541 | net->ipv6.rt6_stats->fib_discarded_routes); |
1da177e4 LT |
2542 | |
2543 | return 0; | |
2544 | } | |
2545 | ||
2546 | static int rt6_stats_seq_open(struct inode *inode, struct file *file) | |
2547 | { | |
de05c557 | 2548 | return single_open_net(inode, file, rt6_stats_seq_show); |
69ddb805 DL |
2549 | } |
2550 | ||
9a32144e | 2551 | static const struct file_operations rt6_stats_seq_fops = { |
1da177e4 LT |
2552 | .owner = THIS_MODULE, |
2553 | .open = rt6_stats_seq_open, | |
2554 | .read = seq_read, | |
2555 | .llseek = seq_lseek, | |
b6fcbdb4 | 2556 | .release = single_release_net, |
1da177e4 LT |
2557 | }; |
2558 | #endif /* CONFIG_PROC_FS */ | |
2559 | ||
2560 | #ifdef CONFIG_SYSCTL | |
2561 | ||
1da177e4 | 2562 | static |
8d65af78 | 2563 | int ipv6_sysctl_rtcache_flush(ctl_table *ctl, int write, |
1da177e4 LT |
2564 | void __user *buffer, size_t *lenp, loff_t *ppos) |
2565 | { | |
5b7c931d DL |
2566 | struct net *net = current->nsproxy->net_ns; |
2567 | int delay = net->ipv6.sysctl.flush_delay; | |
1da177e4 | 2568 | if (write) { |
8d65af78 | 2569 | proc_dointvec(ctl, write, buffer, lenp, ppos); |
5b7c931d | 2570 | fib6_run_gc(delay <= 0 ? ~0UL : (unsigned long)delay, net); |
1da177e4 LT |
2571 | return 0; |
2572 | } else | |
2573 | return -EINVAL; | |
2574 | } | |
2575 | ||
760f2d01 | 2576 | ctl_table ipv6_route_table_template[] = { |
1ab1457c | 2577 | { |
1da177e4 | 2578 | .procname = "flush", |
4990509f | 2579 | .data = &init_net.ipv6.sysctl.flush_delay, |
1da177e4 | 2580 | .maxlen = sizeof(int), |
89c8b3a1 | 2581 | .mode = 0200, |
6d9f239a | 2582 | .proc_handler = ipv6_sysctl_rtcache_flush |
1da177e4 LT |
2583 | }, |
2584 | { | |
1da177e4 | 2585 | .procname = "gc_thresh", |
9a7ec3a9 | 2586 | .data = &ip6_dst_ops_template.gc_thresh, |
1da177e4 LT |
2587 | .maxlen = sizeof(int), |
2588 | .mode = 0644, | |
6d9f239a | 2589 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2590 | }, |
2591 | { | |
1da177e4 | 2592 | .procname = "max_size", |
4990509f | 2593 | .data = &init_net.ipv6.sysctl.ip6_rt_max_size, |
1da177e4 LT |
2594 | .maxlen = sizeof(int), |
2595 | .mode = 0644, | |
6d9f239a | 2596 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2597 | }, |
2598 | { | |
1da177e4 | 2599 | .procname = "gc_min_interval", |
4990509f | 2600 | .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval, |
1da177e4 LT |
2601 | .maxlen = sizeof(int), |
2602 | .mode = 0644, | |
6d9f239a | 2603 | .proc_handler = proc_dointvec_jiffies, |
1da177e4 LT |
2604 | }, |
2605 | { | |
1da177e4 | 2606 | .procname = "gc_timeout", |
4990509f | 2607 | .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout, |
1da177e4 LT |
2608 | .maxlen = sizeof(int), |
2609 | .mode = 0644, | |
6d9f239a | 2610 | .proc_handler = proc_dointvec_jiffies, |
1da177e4 LT |
2611 | }, |
2612 | { | |
1da177e4 | 2613 | .procname = "gc_interval", |
4990509f | 2614 | .data = &init_net.ipv6.sysctl.ip6_rt_gc_interval, |
1da177e4 LT |
2615 | .maxlen = sizeof(int), |
2616 | .mode = 0644, | |
6d9f239a | 2617 | .proc_handler = proc_dointvec_jiffies, |
1da177e4 LT |
2618 | }, |
2619 | { | |
1da177e4 | 2620 | .procname = "gc_elasticity", |
4990509f | 2621 | .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity, |
1da177e4 LT |
2622 | .maxlen = sizeof(int), |
2623 | .mode = 0644, | |
f3d3f616 | 2624 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2625 | }, |
2626 | { | |
1da177e4 | 2627 | .procname = "mtu_expires", |
4990509f | 2628 | .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires, |
1da177e4 LT |
2629 | .maxlen = sizeof(int), |
2630 | .mode = 0644, | |
6d9f239a | 2631 | .proc_handler = proc_dointvec_jiffies, |
1da177e4 LT |
2632 | }, |
2633 | { | |
1da177e4 | 2634 | .procname = "min_adv_mss", |
4990509f | 2635 | .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss, |
1da177e4 LT |
2636 | .maxlen = sizeof(int), |
2637 | .mode = 0644, | |
f3d3f616 | 2638 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2639 | }, |
2640 | { | |
1da177e4 | 2641 | .procname = "gc_min_interval_ms", |
4990509f | 2642 | .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval, |
1da177e4 LT |
2643 | .maxlen = sizeof(int), |
2644 | .mode = 0644, | |
6d9f239a | 2645 | .proc_handler = proc_dointvec_ms_jiffies, |
1da177e4 | 2646 | }, |
f8572d8f | 2647 | { } |
1da177e4 LT |
2648 | }; |
2649 | ||
2c8c1e72 | 2650 | struct ctl_table * __net_init ipv6_route_sysctl_init(struct net *net) |
760f2d01 DL |
2651 | { |
2652 | struct ctl_table *table; | |
2653 | ||
2654 | table = kmemdup(ipv6_route_table_template, | |
2655 | sizeof(ipv6_route_table_template), | |
2656 | GFP_KERNEL); | |
5ee09105 YH |
2657 | |
2658 | if (table) { | |
2659 | table[0].data = &net->ipv6.sysctl.flush_delay; | |
86393e52 | 2660 | table[1].data = &net->ipv6.ip6_dst_ops.gc_thresh; |
5ee09105 YH |
2661 | table[2].data = &net->ipv6.sysctl.ip6_rt_max_size; |
2662 | table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval; | |
2663 | table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout; | |
2664 | table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval; | |
2665 | table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity; | |
2666 | table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires; | |
2667 | table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss; | |
9c69fabe | 2668 | table[9].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval; |
5ee09105 YH |
2669 | } |
2670 | ||
760f2d01 DL |
2671 | return table; |
2672 | } | |
1da177e4 LT |
2673 | #endif |
2674 | ||
2c8c1e72 | 2675 | static int __net_init ip6_route_net_init(struct net *net) |
cdb18761 | 2676 | { |
633d424b | 2677 | int ret = -ENOMEM; |
8ed67789 | 2678 | |
86393e52 AD |
2679 | memcpy(&net->ipv6.ip6_dst_ops, &ip6_dst_ops_template, |
2680 | sizeof(net->ipv6.ip6_dst_ops)); | |
f2fc6a54 | 2681 | |
fc66f95c ED |
2682 | if (dst_entries_init(&net->ipv6.ip6_dst_ops) < 0) |
2683 | goto out_ip6_dst_ops; | |
2684 | ||
8ed67789 DL |
2685 | net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template, |
2686 | sizeof(*net->ipv6.ip6_null_entry), | |
2687 | GFP_KERNEL); | |
2688 | if (!net->ipv6.ip6_null_entry) | |
fc66f95c | 2689 | goto out_ip6_dst_entries; |
d8d1f30b | 2690 | net->ipv6.ip6_null_entry->dst.path = |
8ed67789 | 2691 | (struct dst_entry *)net->ipv6.ip6_null_entry; |
d8d1f30b | 2692 | net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops; |
defb3519 | 2693 | dst_metric_set(&net->ipv6.ip6_null_entry->dst, RTAX_HOPLIMIT, 255); |
8ed67789 DL |
2694 | |
2695 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES | |
2696 | net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template, | |
2697 | sizeof(*net->ipv6.ip6_prohibit_entry), | |
2698 | GFP_KERNEL); | |
68fffc67 PZ |
2699 | if (!net->ipv6.ip6_prohibit_entry) |
2700 | goto out_ip6_null_entry; | |
d8d1f30b | 2701 | net->ipv6.ip6_prohibit_entry->dst.path = |
8ed67789 | 2702 | (struct dst_entry *)net->ipv6.ip6_prohibit_entry; |
d8d1f30b | 2703 | net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops; |
defb3519 | 2704 | dst_metric_set(&net->ipv6.ip6_prohibit_entry->dst, RTAX_HOPLIMIT, 255); |
8ed67789 DL |
2705 | |
2706 | net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template, | |
2707 | sizeof(*net->ipv6.ip6_blk_hole_entry), | |
2708 | GFP_KERNEL); | |
68fffc67 PZ |
2709 | if (!net->ipv6.ip6_blk_hole_entry) |
2710 | goto out_ip6_prohibit_entry; | |
d8d1f30b | 2711 | net->ipv6.ip6_blk_hole_entry->dst.path = |
8ed67789 | 2712 | (struct dst_entry *)net->ipv6.ip6_blk_hole_entry; |
d8d1f30b | 2713 | net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops; |
defb3519 | 2714 | dst_metric_set(&net->ipv6.ip6_blk_hole_entry->dst, RTAX_HOPLIMIT, 255); |
8ed67789 DL |
2715 | #endif |
2716 | ||
b339a47c PZ |
2717 | net->ipv6.sysctl.flush_delay = 0; |
2718 | net->ipv6.sysctl.ip6_rt_max_size = 4096; | |
2719 | net->ipv6.sysctl.ip6_rt_gc_min_interval = HZ / 2; | |
2720 | net->ipv6.sysctl.ip6_rt_gc_timeout = 60*HZ; | |
2721 | net->ipv6.sysctl.ip6_rt_gc_interval = 30*HZ; | |
2722 | net->ipv6.sysctl.ip6_rt_gc_elasticity = 9; | |
2723 | net->ipv6.sysctl.ip6_rt_mtu_expires = 10*60*HZ; | |
2724 | net->ipv6.sysctl.ip6_rt_min_advmss = IPV6_MIN_MTU - 20 - 40; | |
2725 | ||
cdb18761 DL |
2726 | #ifdef CONFIG_PROC_FS |
2727 | proc_net_fops_create(net, "ipv6_route", 0, &ipv6_route_proc_fops); | |
2728 | proc_net_fops_create(net, "rt6_stats", S_IRUGO, &rt6_stats_seq_fops); | |
2729 | #endif | |
6891a346 BT |
2730 | net->ipv6.ip6_rt_gc_expire = 30*HZ; |
2731 | ||
8ed67789 DL |
2732 | ret = 0; |
2733 | out: | |
2734 | return ret; | |
f2fc6a54 | 2735 | |
68fffc67 PZ |
2736 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES |
2737 | out_ip6_prohibit_entry: | |
2738 | kfree(net->ipv6.ip6_prohibit_entry); | |
2739 | out_ip6_null_entry: | |
2740 | kfree(net->ipv6.ip6_null_entry); | |
2741 | #endif | |
fc66f95c ED |
2742 | out_ip6_dst_entries: |
2743 | dst_entries_destroy(&net->ipv6.ip6_dst_ops); | |
f2fc6a54 | 2744 | out_ip6_dst_ops: |
f2fc6a54 | 2745 | goto out; |
cdb18761 DL |
2746 | } |
2747 | ||
2c8c1e72 | 2748 | static void __net_exit ip6_route_net_exit(struct net *net) |
cdb18761 DL |
2749 | { |
2750 | #ifdef CONFIG_PROC_FS | |
2751 | proc_net_remove(net, "ipv6_route"); | |
2752 | proc_net_remove(net, "rt6_stats"); | |
2753 | #endif | |
8ed67789 DL |
2754 | kfree(net->ipv6.ip6_null_entry); |
2755 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES | |
2756 | kfree(net->ipv6.ip6_prohibit_entry); | |
2757 | kfree(net->ipv6.ip6_blk_hole_entry); | |
2758 | #endif | |
41bb78b4 | 2759 | dst_entries_destroy(&net->ipv6.ip6_dst_ops); |
cdb18761 DL |
2760 | } |
2761 | ||
2762 | static struct pernet_operations ip6_route_net_ops = { | |
2763 | .init = ip6_route_net_init, | |
2764 | .exit = ip6_route_net_exit, | |
2765 | }; | |
2766 | ||
8ed67789 DL |
2767 | static struct notifier_block ip6_route_dev_notifier = { |
2768 | .notifier_call = ip6_route_dev_notify, | |
2769 | .priority = 0, | |
2770 | }; | |
2771 | ||
433d49c3 | 2772 | int __init ip6_route_init(void) |
1da177e4 | 2773 | { |
433d49c3 DL |
2774 | int ret; |
2775 | ||
9a7ec3a9 DL |
2776 | ret = -ENOMEM; |
2777 | ip6_dst_ops_template.kmem_cachep = | |
e5d679f3 | 2778 | kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0, |
f845ab6b | 2779 | SLAB_HWCACHE_ALIGN, NULL); |
9a7ec3a9 | 2780 | if (!ip6_dst_ops_template.kmem_cachep) |
c19a28e1 | 2781 | goto out; |
14e50e57 | 2782 | |
fc66f95c | 2783 | ret = dst_entries_init(&ip6_dst_blackhole_ops); |
8ed67789 | 2784 | if (ret) |
bdb3289f | 2785 | goto out_kmem_cache; |
bdb3289f | 2786 | |
fc66f95c ED |
2787 | ret = register_pernet_subsys(&ip6_route_net_ops); |
2788 | if (ret) | |
2789 | goto out_dst_entries; | |
2790 | ||
5dc121e9 AE |
2791 | ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep; |
2792 | ||
8ed67789 DL |
2793 | /* Registering of the loopback is done before this portion of code, |
2794 | * the loopback reference in rt6_info will not be taken, do it | |
2795 | * manually for init_net */ | |
d8d1f30b | 2796 | init_net.ipv6.ip6_null_entry->dst.dev = init_net.loopback_dev; |
8ed67789 DL |
2797 | init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev); |
2798 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES | |
d8d1f30b | 2799 | init_net.ipv6.ip6_prohibit_entry->dst.dev = init_net.loopback_dev; |
8ed67789 | 2800 | init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev); |
d8d1f30b | 2801 | init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev; |
8ed67789 DL |
2802 | init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev); |
2803 | #endif | |
433d49c3 DL |
2804 | ret = fib6_init(); |
2805 | if (ret) | |
8ed67789 | 2806 | goto out_register_subsys; |
433d49c3 | 2807 | |
433d49c3 DL |
2808 | ret = xfrm6_init(); |
2809 | if (ret) | |
cdb18761 | 2810 | goto out_fib6_init; |
c35b7e72 | 2811 | |
433d49c3 DL |
2812 | ret = fib6_rules_init(); |
2813 | if (ret) | |
2814 | goto xfrm6_init; | |
7e5449c2 | 2815 | |
433d49c3 DL |
2816 | ret = -ENOBUFS; |
2817 | if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL) || | |
2818 | __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL) || | |
2819 | __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL)) | |
2820 | goto fib6_rules_init; | |
c127ea2c | 2821 | |
8ed67789 | 2822 | ret = register_netdevice_notifier(&ip6_route_dev_notifier); |
cdb18761 DL |
2823 | if (ret) |
2824 | goto fib6_rules_init; | |
8ed67789 | 2825 | |
433d49c3 DL |
2826 | out: |
2827 | return ret; | |
2828 | ||
2829 | fib6_rules_init: | |
433d49c3 DL |
2830 | fib6_rules_cleanup(); |
2831 | xfrm6_init: | |
433d49c3 | 2832 | xfrm6_fini(); |
433d49c3 | 2833 | out_fib6_init: |
433d49c3 | 2834 | fib6_gc_cleanup(); |
8ed67789 DL |
2835 | out_register_subsys: |
2836 | unregister_pernet_subsys(&ip6_route_net_ops); | |
fc66f95c ED |
2837 | out_dst_entries: |
2838 | dst_entries_destroy(&ip6_dst_blackhole_ops); | |
433d49c3 | 2839 | out_kmem_cache: |
f2fc6a54 | 2840 | kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep); |
433d49c3 | 2841 | goto out; |
1da177e4 LT |
2842 | } |
2843 | ||
2844 | void ip6_route_cleanup(void) | |
2845 | { | |
8ed67789 | 2846 | unregister_netdevice_notifier(&ip6_route_dev_notifier); |
101367c2 | 2847 | fib6_rules_cleanup(); |
1da177e4 | 2848 | xfrm6_fini(); |
1da177e4 | 2849 | fib6_gc_cleanup(); |
8ed67789 | 2850 | unregister_pernet_subsys(&ip6_route_net_ops); |
41bb78b4 | 2851 | dst_entries_destroy(&ip6_dst_blackhole_ops); |
f2fc6a54 | 2852 | kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep); |
1da177e4 | 2853 | } |