Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
2 | * Linux INET6 implementation | |
3 | * FIB front-end. | |
4 | * | |
5 | * Authors: | |
1ab1457c | 6 | * Pedro Roque <roque@di.fc.ul.pt> |
1da177e4 | 7 | * |
1da177e4 LT |
8 | * This program is free software; you can redistribute it and/or |
9 | * modify it under the terms of the GNU General Public License | |
10 | * as published by the Free Software Foundation; either version | |
11 | * 2 of the License, or (at your option) any later version. | |
12 | */ | |
13 | ||
14 | /* Changes: | |
15 | * | |
16 | * YOSHIFUJI Hideaki @USAGI | |
17 | * reworked default router selection. | |
18 | * - respect outgoing interface | |
19 | * - select from (probably) reachable routers (i.e. | |
20 | * routers in REACHABLE, STALE, DELAY or PROBE states). | |
21 | * - always select the same router if it is (probably) | |
22 | * reachable. otherwise, round-robin the list. | |
c0bece9f YH |
23 | * Ville Nuorvala |
24 | * Fixed routing subtrees. | |
1da177e4 LT |
25 | */ |
26 | ||
f3213831 JP |
27 | #define pr_fmt(fmt) "IPv6: " fmt |
28 | ||
4fc268d2 | 29 | #include <linux/capability.h> |
1da177e4 | 30 | #include <linux/errno.h> |
bc3b2d7f | 31 | #include <linux/export.h> |
1da177e4 LT |
32 | #include <linux/types.h> |
33 | #include <linux/times.h> | |
34 | #include <linux/socket.h> | |
35 | #include <linux/sockios.h> | |
36 | #include <linux/net.h> | |
37 | #include <linux/route.h> | |
38 | #include <linux/netdevice.h> | |
39 | #include <linux/in6.h> | |
7bc570c8 | 40 | #include <linux/mroute6.h> |
1da177e4 | 41 | #include <linux/init.h> |
1da177e4 | 42 | #include <linux/if_arp.h> |
1da177e4 LT |
43 | #include <linux/proc_fs.h> |
44 | #include <linux/seq_file.h> | |
5b7c931d | 45 | #include <linux/nsproxy.h> |
5a0e3ad6 | 46 | #include <linux/slab.h> |
457c4cbc | 47 | #include <net/net_namespace.h> |
1da177e4 LT |
48 | #include <net/snmp.h> |
49 | #include <net/ipv6.h> | |
50 | #include <net/ip6_fib.h> | |
51 | #include <net/ip6_route.h> | |
52 | #include <net/ndisc.h> | |
53 | #include <net/addrconf.h> | |
54 | #include <net/tcp.h> | |
55 | #include <linux/rtnetlink.h> | |
56 | #include <net/dst.h> | |
57 | #include <net/xfrm.h> | |
8d71740c | 58 | #include <net/netevent.h> |
21713ebc | 59 | #include <net/netlink.h> |
1da177e4 LT |
60 | |
61 | #include <asm/uaccess.h> | |
62 | ||
63 | #ifdef CONFIG_SYSCTL | |
64 | #include <linux/sysctl.h> | |
65 | #endif | |
66 | ||
1716a961 | 67 | static struct rt6_info *ip6_rt_copy(struct rt6_info *ort, |
21efcfa0 | 68 | const struct in6_addr *dest); |
1da177e4 | 69 | static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie); |
0dbaee3b | 70 | static unsigned int ip6_default_advmss(const struct dst_entry *dst); |
ebb762f2 | 71 | static unsigned int ip6_mtu(const struct dst_entry *dst); |
1da177e4 LT |
72 | static struct dst_entry *ip6_negative_advice(struct dst_entry *); |
73 | static void ip6_dst_destroy(struct dst_entry *); | |
74 | static void ip6_dst_ifdown(struct dst_entry *, | |
75 | struct net_device *dev, int how); | |
569d3645 | 76 | static int ip6_dst_gc(struct dst_ops *ops); |
1da177e4 LT |
77 | |
78 | static int ip6_pkt_discard(struct sk_buff *skb); | |
79 | static int ip6_pkt_discard_out(struct sk_buff *skb); | |
80 | static void ip6_link_failure(struct sk_buff *skb); | |
6700c270 DM |
81 | static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk, |
82 | struct sk_buff *skb, u32 mtu); | |
83 | static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, | |
84 | struct sk_buff *skb); | |
1da177e4 | 85 | |
70ceb4f5 | 86 | #ifdef CONFIG_IPV6_ROUTE_INFO |
efa2cea0 | 87 | static struct rt6_info *rt6_add_route_info(struct net *net, |
b71d1d42 ED |
88 | const struct in6_addr *prefix, int prefixlen, |
89 | const struct in6_addr *gwaddr, int ifindex, | |
95c96174 | 90 | unsigned int pref); |
efa2cea0 | 91 | static struct rt6_info *rt6_get_route_info(struct net *net, |
b71d1d42 ED |
92 | const struct in6_addr *prefix, int prefixlen, |
93 | const struct in6_addr *gwaddr, int ifindex); | |
70ceb4f5 YH |
94 | #endif |
95 | ||
06582540 DM |
96 | static u32 *ipv6_cow_metrics(struct dst_entry *dst, unsigned long old) |
97 | { | |
98 | struct rt6_info *rt = (struct rt6_info *) dst; | |
99 | struct inet_peer *peer; | |
100 | u32 *p = NULL; | |
101 | ||
8e2ec639 YZ |
102 | if (!(rt->dst.flags & DST_HOST)) |
103 | return NULL; | |
104 | ||
fbfe95a4 | 105 | peer = rt6_get_peer_create(rt); |
06582540 DM |
106 | if (peer) { |
107 | u32 *old_p = __DST_METRICS_PTR(old); | |
108 | unsigned long prev, new; | |
109 | ||
110 | p = peer->metrics; | |
111 | if (inet_metrics_new(peer)) | |
112 | memcpy(p, old_p, sizeof(u32) * RTAX_MAX); | |
113 | ||
114 | new = (unsigned long) p; | |
115 | prev = cmpxchg(&dst->_metrics, old, new); | |
116 | ||
117 | if (prev != old) { | |
118 | p = __DST_METRICS_PTR(prev); | |
119 | if (prev & DST_METRICS_READ_ONLY) | |
120 | p = NULL; | |
121 | } | |
122 | } | |
123 | return p; | |
124 | } | |
125 | ||
f894cbf8 DM |
126 | static inline const void *choose_neigh_daddr(struct rt6_info *rt, |
127 | struct sk_buff *skb, | |
128 | const void *daddr) | |
39232973 DM |
129 | { |
130 | struct in6_addr *p = &rt->rt6i_gateway; | |
131 | ||
a7563f34 | 132 | if (!ipv6_addr_any(p)) |
39232973 | 133 | return (const void *) p; |
f894cbf8 DM |
134 | else if (skb) |
135 | return &ipv6_hdr(skb)->daddr; | |
39232973 DM |
136 | return daddr; |
137 | } | |
138 | ||
f894cbf8 DM |
139 | static struct neighbour *ip6_neigh_lookup(const struct dst_entry *dst, |
140 | struct sk_buff *skb, | |
141 | const void *daddr) | |
d3aaeb38 | 142 | { |
39232973 DM |
143 | struct rt6_info *rt = (struct rt6_info *) dst; |
144 | struct neighbour *n; | |
145 | ||
f894cbf8 | 146 | daddr = choose_neigh_daddr(rt, skb, daddr); |
39232973 | 147 | n = __ipv6_neigh_lookup(&nd_tbl, dst->dev, daddr); |
f83c7790 DM |
148 | if (n) |
149 | return n; | |
150 | return neigh_create(&nd_tbl, daddr, dst->dev); | |
151 | } | |
152 | ||
8ade06c6 | 153 | static int rt6_bind_neighbour(struct rt6_info *rt, struct net_device *dev) |
f83c7790 | 154 | { |
8ade06c6 DM |
155 | struct neighbour *n = __ipv6_neigh_lookup(&nd_tbl, dev, &rt->rt6i_gateway); |
156 | if (!n) { | |
157 | n = neigh_create(&nd_tbl, &rt->rt6i_gateway, dev); | |
158 | if (IS_ERR(n)) | |
159 | return PTR_ERR(n); | |
160 | } | |
97cac082 | 161 | rt->n = n; |
f83c7790 DM |
162 | |
163 | return 0; | |
d3aaeb38 DM |
164 | } |
165 | ||
9a7ec3a9 | 166 | static struct dst_ops ip6_dst_ops_template = { |
1da177e4 | 167 | .family = AF_INET6, |
09640e63 | 168 | .protocol = cpu_to_be16(ETH_P_IPV6), |
1da177e4 LT |
169 | .gc = ip6_dst_gc, |
170 | .gc_thresh = 1024, | |
171 | .check = ip6_dst_check, | |
0dbaee3b | 172 | .default_advmss = ip6_default_advmss, |
ebb762f2 | 173 | .mtu = ip6_mtu, |
06582540 | 174 | .cow_metrics = ipv6_cow_metrics, |
1da177e4 LT |
175 | .destroy = ip6_dst_destroy, |
176 | .ifdown = ip6_dst_ifdown, | |
177 | .negative_advice = ip6_negative_advice, | |
178 | .link_failure = ip6_link_failure, | |
179 | .update_pmtu = ip6_rt_update_pmtu, | |
6e157b6a | 180 | .redirect = rt6_do_redirect, |
1ac06e03 | 181 | .local_out = __ip6_local_out, |
d3aaeb38 | 182 | .neigh_lookup = ip6_neigh_lookup, |
1da177e4 LT |
183 | }; |
184 | ||
ebb762f2 | 185 | static unsigned int ip6_blackhole_mtu(const struct dst_entry *dst) |
ec831ea7 | 186 | { |
618f9bc7 SK |
187 | unsigned int mtu = dst_metric_raw(dst, RTAX_MTU); |
188 | ||
189 | return mtu ? : dst->dev->mtu; | |
ec831ea7 RD |
190 | } |
191 | ||
6700c270 DM |
192 | static void ip6_rt_blackhole_update_pmtu(struct dst_entry *dst, struct sock *sk, |
193 | struct sk_buff *skb, u32 mtu) | |
14e50e57 DM |
194 | { |
195 | } | |
196 | ||
6700c270 DM |
197 | static void ip6_rt_blackhole_redirect(struct dst_entry *dst, struct sock *sk, |
198 | struct sk_buff *skb) | |
b587ee3b DM |
199 | { |
200 | } | |
201 | ||
0972ddb2 HB |
202 | static u32 *ip6_rt_blackhole_cow_metrics(struct dst_entry *dst, |
203 | unsigned long old) | |
204 | { | |
205 | return NULL; | |
206 | } | |
207 | ||
14e50e57 DM |
208 | static struct dst_ops ip6_dst_blackhole_ops = { |
209 | .family = AF_INET6, | |
09640e63 | 210 | .protocol = cpu_to_be16(ETH_P_IPV6), |
14e50e57 DM |
211 | .destroy = ip6_dst_destroy, |
212 | .check = ip6_dst_check, | |
ebb762f2 | 213 | .mtu = ip6_blackhole_mtu, |
214f45c9 | 214 | .default_advmss = ip6_default_advmss, |
14e50e57 | 215 | .update_pmtu = ip6_rt_blackhole_update_pmtu, |
b587ee3b | 216 | .redirect = ip6_rt_blackhole_redirect, |
0972ddb2 | 217 | .cow_metrics = ip6_rt_blackhole_cow_metrics, |
d3aaeb38 | 218 | .neigh_lookup = ip6_neigh_lookup, |
14e50e57 DM |
219 | }; |
220 | ||
62fa8a84 DM |
221 | static const u32 ip6_template_metrics[RTAX_MAX] = { |
222 | [RTAX_HOPLIMIT - 1] = 255, | |
223 | }; | |
224 | ||
bdb3289f | 225 | static struct rt6_info ip6_null_entry_template = { |
d8d1f30b CG |
226 | .dst = { |
227 | .__refcnt = ATOMIC_INIT(1), | |
228 | .__use = 1, | |
229 | .obsolete = -1, | |
230 | .error = -ENETUNREACH, | |
d8d1f30b CG |
231 | .input = ip6_pkt_discard, |
232 | .output = ip6_pkt_discard_out, | |
1da177e4 LT |
233 | }, |
234 | .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP), | |
4f724279 | 235 | .rt6i_protocol = RTPROT_KERNEL, |
1da177e4 LT |
236 | .rt6i_metric = ~(u32) 0, |
237 | .rt6i_ref = ATOMIC_INIT(1), | |
238 | }; | |
239 | ||
101367c2 TG |
240 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES |
241 | ||
6723ab54 DM |
242 | static int ip6_pkt_prohibit(struct sk_buff *skb); |
243 | static int ip6_pkt_prohibit_out(struct sk_buff *skb); | |
6723ab54 | 244 | |
280a34c8 | 245 | static struct rt6_info ip6_prohibit_entry_template = { |
d8d1f30b CG |
246 | .dst = { |
247 | .__refcnt = ATOMIC_INIT(1), | |
248 | .__use = 1, | |
249 | .obsolete = -1, | |
250 | .error = -EACCES, | |
d8d1f30b CG |
251 | .input = ip6_pkt_prohibit, |
252 | .output = ip6_pkt_prohibit_out, | |
101367c2 TG |
253 | }, |
254 | .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP), | |
4f724279 | 255 | .rt6i_protocol = RTPROT_KERNEL, |
101367c2 TG |
256 | .rt6i_metric = ~(u32) 0, |
257 | .rt6i_ref = ATOMIC_INIT(1), | |
258 | }; | |
259 | ||
bdb3289f | 260 | static struct rt6_info ip6_blk_hole_entry_template = { |
d8d1f30b CG |
261 | .dst = { |
262 | .__refcnt = ATOMIC_INIT(1), | |
263 | .__use = 1, | |
264 | .obsolete = -1, | |
265 | .error = -EINVAL, | |
d8d1f30b CG |
266 | .input = dst_discard, |
267 | .output = dst_discard, | |
101367c2 TG |
268 | }, |
269 | .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP), | |
4f724279 | 270 | .rt6i_protocol = RTPROT_KERNEL, |
101367c2 TG |
271 | .rt6i_metric = ~(u32) 0, |
272 | .rt6i_ref = ATOMIC_INIT(1), | |
273 | }; | |
274 | ||
275 | #endif | |
276 | ||
1da177e4 | 277 | /* allocate dst with ip6_dst_ops */ |
97bab73f | 278 | static inline struct rt6_info *ip6_dst_alloc(struct net *net, |
957c665f | 279 | struct net_device *dev, |
8b96d22d DM |
280 | int flags, |
281 | struct fib6_table *table) | |
1da177e4 | 282 | { |
97bab73f | 283 | struct rt6_info *rt = dst_alloc(&net->ipv6.ip6_dst_ops, dev, |
f5b0a874 | 284 | 0, DST_OBSOLETE_NONE, flags); |
cf911662 | 285 | |
97bab73f | 286 | if (rt) { |
8104891b SK |
287 | struct dst_entry *dst = &rt->dst; |
288 | ||
289 | memset(dst + 1, 0, sizeof(*rt) - sizeof(*dst)); | |
8b96d22d | 290 | rt6_init_peer(rt, table ? &table->tb6_peers : net->ipv6.peers); |
97bab73f | 291 | } |
cf911662 | 292 | return rt; |
1da177e4 LT |
293 | } |
294 | ||
295 | static void ip6_dst_destroy(struct dst_entry *dst) | |
296 | { | |
297 | struct rt6_info *rt = (struct rt6_info *)dst; | |
298 | struct inet6_dev *idev = rt->rt6i_idev; | |
299 | ||
97cac082 DM |
300 | if (rt->n) |
301 | neigh_release(rt->n); | |
302 | ||
8e2ec639 YZ |
303 | if (!(rt->dst.flags & DST_HOST)) |
304 | dst_destroy_metrics_generic(dst); | |
305 | ||
38308473 | 306 | if (idev) { |
1da177e4 LT |
307 | rt->rt6i_idev = NULL; |
308 | in6_dev_put(idev); | |
1ab1457c | 309 | } |
1716a961 G |
310 | |
311 | if (!(rt->rt6i_flags & RTF_EXPIRES) && dst->from) | |
312 | dst_release(dst->from); | |
313 | ||
97bab73f DM |
314 | if (rt6_has_peer(rt)) { |
315 | struct inet_peer *peer = rt6_peer_ptr(rt); | |
b3419363 DM |
316 | inet_putpeer(peer); |
317 | } | |
318 | } | |
319 | ||
6431cbc2 DM |
320 | static atomic_t __rt6_peer_genid = ATOMIC_INIT(0); |
321 | ||
322 | static u32 rt6_peer_genid(void) | |
323 | { | |
324 | return atomic_read(&__rt6_peer_genid); | |
325 | } | |
326 | ||
b3419363 DM |
327 | void rt6_bind_peer(struct rt6_info *rt, int create) |
328 | { | |
97bab73f | 329 | struct inet_peer_base *base; |
b3419363 DM |
330 | struct inet_peer *peer; |
331 | ||
97bab73f DM |
332 | base = inetpeer_base_ptr(rt->_rt6i_peer); |
333 | if (!base) | |
334 | return; | |
335 | ||
336 | peer = inet_getpeer_v6(base, &rt->rt6i_dst.addr, create); | |
7b34ca2a DM |
337 | if (peer) { |
338 | if (!rt6_set_peer(rt, peer)) | |
339 | inet_putpeer(peer); | |
340 | else | |
341 | rt->rt6i_peer_genid = rt6_peer_genid(); | |
342 | } | |
1da177e4 LT |
343 | } |
344 | ||
345 | static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev, | |
346 | int how) | |
347 | { | |
348 | struct rt6_info *rt = (struct rt6_info *)dst; | |
349 | struct inet6_dev *idev = rt->rt6i_idev; | |
5a3e55d6 | 350 | struct net_device *loopback_dev = |
c346dca1 | 351 | dev_net(dev)->loopback_dev; |
1da177e4 | 352 | |
97cac082 DM |
353 | if (dev != loopback_dev) { |
354 | if (idev && idev->dev == dev) { | |
355 | struct inet6_dev *loopback_idev = | |
356 | in6_dev_get(loopback_dev); | |
357 | if (loopback_idev) { | |
358 | rt->rt6i_idev = loopback_idev; | |
359 | in6_dev_put(idev); | |
360 | } | |
361 | } | |
362 | if (rt->n && rt->n->dev == dev) { | |
363 | rt->n->dev = loopback_dev; | |
364 | dev_hold(loopback_dev); | |
365 | dev_put(dev); | |
1da177e4 LT |
366 | } |
367 | } | |
368 | } | |
369 | ||
a50feda5 | 370 | static bool rt6_check_expired(const struct rt6_info *rt) |
1da177e4 | 371 | { |
1716a961 G |
372 | struct rt6_info *ort = NULL; |
373 | ||
374 | if (rt->rt6i_flags & RTF_EXPIRES) { | |
375 | if (time_after(jiffies, rt->dst.expires)) | |
a50feda5 | 376 | return true; |
1716a961 G |
377 | } else if (rt->dst.from) { |
378 | ort = (struct rt6_info *) rt->dst.from; | |
379 | return (ort->rt6i_flags & RTF_EXPIRES) && | |
380 | time_after(jiffies, ort->dst.expires); | |
381 | } | |
a50feda5 | 382 | return false; |
1da177e4 LT |
383 | } |
384 | ||
a50feda5 | 385 | static bool rt6_need_strict(const struct in6_addr *daddr) |
c71099ac | 386 | { |
a02cec21 ED |
387 | return ipv6_addr_type(daddr) & |
388 | (IPV6_ADDR_MULTICAST | IPV6_ADDR_LINKLOCAL | IPV6_ADDR_LOOPBACK); | |
c71099ac TG |
389 | } |
390 | ||
1da177e4 | 391 | /* |
c71099ac | 392 | * Route lookup. Any table->tb6_lock is implied. |
1da177e4 LT |
393 | */ |
394 | ||
8ed67789 DL |
395 | static inline struct rt6_info *rt6_device_match(struct net *net, |
396 | struct rt6_info *rt, | |
b71d1d42 | 397 | const struct in6_addr *saddr, |
1da177e4 | 398 | int oif, |
d420895e | 399 | int flags) |
1da177e4 LT |
400 | { |
401 | struct rt6_info *local = NULL; | |
402 | struct rt6_info *sprt; | |
403 | ||
dd3abc4e YH |
404 | if (!oif && ipv6_addr_any(saddr)) |
405 | goto out; | |
406 | ||
d8d1f30b | 407 | for (sprt = rt; sprt; sprt = sprt->dst.rt6_next) { |
d1918542 | 408 | struct net_device *dev = sprt->dst.dev; |
dd3abc4e YH |
409 | |
410 | if (oif) { | |
1da177e4 LT |
411 | if (dev->ifindex == oif) |
412 | return sprt; | |
413 | if (dev->flags & IFF_LOOPBACK) { | |
38308473 | 414 | if (!sprt->rt6i_idev || |
1da177e4 | 415 | sprt->rt6i_idev->dev->ifindex != oif) { |
d420895e | 416 | if (flags & RT6_LOOKUP_F_IFACE && oif) |
1da177e4 | 417 | continue; |
1ab1457c | 418 | if (local && (!oif || |
1da177e4 LT |
419 | local->rt6i_idev->dev->ifindex == oif)) |
420 | continue; | |
421 | } | |
422 | local = sprt; | |
423 | } | |
dd3abc4e YH |
424 | } else { |
425 | if (ipv6_chk_addr(net, saddr, dev, | |
426 | flags & RT6_LOOKUP_F_IFACE)) | |
427 | return sprt; | |
1da177e4 | 428 | } |
dd3abc4e | 429 | } |
1da177e4 | 430 | |
dd3abc4e | 431 | if (oif) { |
1da177e4 LT |
432 | if (local) |
433 | return local; | |
434 | ||
d420895e | 435 | if (flags & RT6_LOOKUP_F_IFACE) |
8ed67789 | 436 | return net->ipv6.ip6_null_entry; |
1da177e4 | 437 | } |
dd3abc4e | 438 | out: |
1da177e4 LT |
439 | return rt; |
440 | } | |
441 | ||
27097255 YH |
442 | #ifdef CONFIG_IPV6_ROUTER_PREF |
443 | static void rt6_probe(struct rt6_info *rt) | |
444 | { | |
f2c31e32 | 445 | struct neighbour *neigh; |
27097255 YH |
446 | /* |
447 | * Okay, this does not seem to be appropriate | |
448 | * for now, however, we need to check if it | |
449 | * is really so; aka Router Reachability Probing. | |
450 | * | |
451 | * Router Reachability Probe MUST be rate-limited | |
452 | * to no more than one per minute. | |
453 | */ | |
f2c31e32 | 454 | rcu_read_lock(); |
97cac082 | 455 | neigh = rt ? rt->n : NULL; |
27097255 | 456 | if (!neigh || (neigh->nud_state & NUD_VALID)) |
f2c31e32 | 457 | goto out; |
27097255 YH |
458 | read_lock_bh(&neigh->lock); |
459 | if (!(neigh->nud_state & NUD_VALID) && | |
52e16356 | 460 | time_after(jiffies, neigh->updated + rt->rt6i_idev->cnf.rtr_probe_interval)) { |
27097255 YH |
461 | struct in6_addr mcaddr; |
462 | struct in6_addr *target; | |
463 | ||
464 | neigh->updated = jiffies; | |
465 | read_unlock_bh(&neigh->lock); | |
466 | ||
467 | target = (struct in6_addr *)&neigh->primary_key; | |
468 | addrconf_addr_solict_mult(target, &mcaddr); | |
d1918542 | 469 | ndisc_send_ns(rt->dst.dev, NULL, target, &mcaddr, NULL); |
f2c31e32 | 470 | } else { |
27097255 | 471 | read_unlock_bh(&neigh->lock); |
f2c31e32 ED |
472 | } |
473 | out: | |
474 | rcu_read_unlock(); | |
27097255 YH |
475 | } |
476 | #else | |
477 | static inline void rt6_probe(struct rt6_info *rt) | |
478 | { | |
27097255 YH |
479 | } |
480 | #endif | |
481 | ||
1da177e4 | 482 | /* |
554cfb7e | 483 | * Default Router Selection (RFC 2461 6.3.6) |
1da177e4 | 484 | */ |
b6f99a21 | 485 | static inline int rt6_check_dev(struct rt6_info *rt, int oif) |
554cfb7e | 486 | { |
d1918542 | 487 | struct net_device *dev = rt->dst.dev; |
161980f4 | 488 | if (!oif || dev->ifindex == oif) |
554cfb7e | 489 | return 2; |
161980f4 DM |
490 | if ((dev->flags & IFF_LOOPBACK) && |
491 | rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif) | |
492 | return 1; | |
493 | return 0; | |
554cfb7e | 494 | } |
1da177e4 | 495 | |
b6f99a21 | 496 | static inline int rt6_check_neigh(struct rt6_info *rt) |
1da177e4 | 497 | { |
f2c31e32 | 498 | struct neighbour *neigh; |
398bcbeb | 499 | int m; |
f2c31e32 ED |
500 | |
501 | rcu_read_lock(); | |
97cac082 | 502 | neigh = rt->n; |
4d0c5911 YH |
503 | if (rt->rt6i_flags & RTF_NONEXTHOP || |
504 | !(rt->rt6i_flags & RTF_GATEWAY)) | |
505 | m = 1; | |
506 | else if (neigh) { | |
554cfb7e YH |
507 | read_lock_bh(&neigh->lock); |
508 | if (neigh->nud_state & NUD_VALID) | |
4d0c5911 | 509 | m = 2; |
398bcbeb YH |
510 | #ifdef CONFIG_IPV6_ROUTER_PREF |
511 | else if (neigh->nud_state & NUD_FAILED) | |
512 | m = 0; | |
513 | #endif | |
514 | else | |
ea73ee23 | 515 | m = 1; |
554cfb7e | 516 | read_unlock_bh(&neigh->lock); |
398bcbeb YH |
517 | } else |
518 | m = 0; | |
f2c31e32 | 519 | rcu_read_unlock(); |
554cfb7e | 520 | return m; |
1da177e4 LT |
521 | } |
522 | ||
554cfb7e YH |
523 | static int rt6_score_route(struct rt6_info *rt, int oif, |
524 | int strict) | |
1da177e4 | 525 | { |
4d0c5911 | 526 | int m, n; |
1ab1457c | 527 | |
4d0c5911 | 528 | m = rt6_check_dev(rt, oif); |
77d16f45 | 529 | if (!m && (strict & RT6_LOOKUP_F_IFACE)) |
554cfb7e | 530 | return -1; |
ebacaaa0 YH |
531 | #ifdef CONFIG_IPV6_ROUTER_PREF |
532 | m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2; | |
533 | #endif | |
4d0c5911 | 534 | n = rt6_check_neigh(rt); |
557e92ef | 535 | if (!n && (strict & RT6_LOOKUP_F_REACHABLE)) |
554cfb7e YH |
536 | return -1; |
537 | return m; | |
538 | } | |
539 | ||
f11e6659 DM |
540 | static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict, |
541 | int *mpri, struct rt6_info *match) | |
554cfb7e | 542 | { |
f11e6659 DM |
543 | int m; |
544 | ||
545 | if (rt6_check_expired(rt)) | |
546 | goto out; | |
547 | ||
548 | m = rt6_score_route(rt, oif, strict); | |
549 | if (m < 0) | |
550 | goto out; | |
551 | ||
552 | if (m > *mpri) { | |
553 | if (strict & RT6_LOOKUP_F_REACHABLE) | |
554 | rt6_probe(match); | |
555 | *mpri = m; | |
556 | match = rt; | |
557 | } else if (strict & RT6_LOOKUP_F_REACHABLE) { | |
558 | rt6_probe(rt); | |
559 | } | |
560 | ||
561 | out: | |
562 | return match; | |
563 | } | |
564 | ||
565 | static struct rt6_info *find_rr_leaf(struct fib6_node *fn, | |
566 | struct rt6_info *rr_head, | |
567 | u32 metric, int oif, int strict) | |
568 | { | |
569 | struct rt6_info *rt, *match; | |
554cfb7e | 570 | int mpri = -1; |
1da177e4 | 571 | |
f11e6659 DM |
572 | match = NULL; |
573 | for (rt = rr_head; rt && rt->rt6i_metric == metric; | |
d8d1f30b | 574 | rt = rt->dst.rt6_next) |
f11e6659 DM |
575 | match = find_match(rt, oif, strict, &mpri, match); |
576 | for (rt = fn->leaf; rt && rt != rr_head && rt->rt6i_metric == metric; | |
d8d1f30b | 577 | rt = rt->dst.rt6_next) |
f11e6659 | 578 | match = find_match(rt, oif, strict, &mpri, match); |
1da177e4 | 579 | |
f11e6659 DM |
580 | return match; |
581 | } | |
1da177e4 | 582 | |
f11e6659 DM |
583 | static struct rt6_info *rt6_select(struct fib6_node *fn, int oif, int strict) |
584 | { | |
585 | struct rt6_info *match, *rt0; | |
8ed67789 | 586 | struct net *net; |
1da177e4 | 587 | |
f11e6659 DM |
588 | rt0 = fn->rr_ptr; |
589 | if (!rt0) | |
590 | fn->rr_ptr = rt0 = fn->leaf; | |
1da177e4 | 591 | |
f11e6659 | 592 | match = find_rr_leaf(fn, rt0, rt0->rt6i_metric, oif, strict); |
1da177e4 | 593 | |
554cfb7e | 594 | if (!match && |
f11e6659 | 595 | (strict & RT6_LOOKUP_F_REACHABLE)) { |
d8d1f30b | 596 | struct rt6_info *next = rt0->dst.rt6_next; |
f11e6659 | 597 | |
554cfb7e | 598 | /* no entries matched; do round-robin */ |
f11e6659 DM |
599 | if (!next || next->rt6i_metric != rt0->rt6i_metric) |
600 | next = fn->leaf; | |
601 | ||
602 | if (next != rt0) | |
603 | fn->rr_ptr = next; | |
1da177e4 | 604 | } |
1da177e4 | 605 | |
d1918542 | 606 | net = dev_net(rt0->dst.dev); |
a02cec21 | 607 | return match ? match : net->ipv6.ip6_null_entry; |
1da177e4 LT |
608 | } |
609 | ||
70ceb4f5 YH |
610 | #ifdef CONFIG_IPV6_ROUTE_INFO |
611 | int rt6_route_rcv(struct net_device *dev, u8 *opt, int len, | |
b71d1d42 | 612 | const struct in6_addr *gwaddr) |
70ceb4f5 | 613 | { |
c346dca1 | 614 | struct net *net = dev_net(dev); |
70ceb4f5 YH |
615 | struct route_info *rinfo = (struct route_info *) opt; |
616 | struct in6_addr prefix_buf, *prefix; | |
617 | unsigned int pref; | |
4bed72e4 | 618 | unsigned long lifetime; |
70ceb4f5 YH |
619 | struct rt6_info *rt; |
620 | ||
621 | if (len < sizeof(struct route_info)) { | |
622 | return -EINVAL; | |
623 | } | |
624 | ||
625 | /* Sanity check for prefix_len and length */ | |
626 | if (rinfo->length > 3) { | |
627 | return -EINVAL; | |
628 | } else if (rinfo->prefix_len > 128) { | |
629 | return -EINVAL; | |
630 | } else if (rinfo->prefix_len > 64) { | |
631 | if (rinfo->length < 2) { | |
632 | return -EINVAL; | |
633 | } | |
634 | } else if (rinfo->prefix_len > 0) { | |
635 | if (rinfo->length < 1) { | |
636 | return -EINVAL; | |
637 | } | |
638 | } | |
639 | ||
640 | pref = rinfo->route_pref; | |
641 | if (pref == ICMPV6_ROUTER_PREF_INVALID) | |
3933fc95 | 642 | return -EINVAL; |
70ceb4f5 | 643 | |
4bed72e4 | 644 | lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ); |
70ceb4f5 YH |
645 | |
646 | if (rinfo->length == 3) | |
647 | prefix = (struct in6_addr *)rinfo->prefix; | |
648 | else { | |
649 | /* this function is safe */ | |
650 | ipv6_addr_prefix(&prefix_buf, | |
651 | (struct in6_addr *)rinfo->prefix, | |
652 | rinfo->prefix_len); | |
653 | prefix = &prefix_buf; | |
654 | } | |
655 | ||
efa2cea0 DL |
656 | rt = rt6_get_route_info(net, prefix, rinfo->prefix_len, gwaddr, |
657 | dev->ifindex); | |
70ceb4f5 YH |
658 | |
659 | if (rt && !lifetime) { | |
e0a1ad73 | 660 | ip6_del_rt(rt); |
70ceb4f5 YH |
661 | rt = NULL; |
662 | } | |
663 | ||
664 | if (!rt && lifetime) | |
efa2cea0 | 665 | rt = rt6_add_route_info(net, prefix, rinfo->prefix_len, gwaddr, dev->ifindex, |
70ceb4f5 YH |
666 | pref); |
667 | else if (rt) | |
668 | rt->rt6i_flags = RTF_ROUTEINFO | | |
669 | (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref); | |
670 | ||
671 | if (rt) { | |
1716a961 G |
672 | if (!addrconf_finite_timeout(lifetime)) |
673 | rt6_clean_expires(rt); | |
674 | else | |
675 | rt6_set_expires(rt, jiffies + HZ * lifetime); | |
676 | ||
d8d1f30b | 677 | dst_release(&rt->dst); |
70ceb4f5 YH |
678 | } |
679 | return 0; | |
680 | } | |
681 | #endif | |
682 | ||
8ed67789 | 683 | #define BACKTRACK(__net, saddr) \ |
982f56f3 | 684 | do { \ |
8ed67789 | 685 | if (rt == __net->ipv6.ip6_null_entry) { \ |
982f56f3 | 686 | struct fib6_node *pn; \ |
e0eda7bb | 687 | while (1) { \ |
982f56f3 YH |
688 | if (fn->fn_flags & RTN_TL_ROOT) \ |
689 | goto out; \ | |
690 | pn = fn->parent; \ | |
691 | if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn) \ | |
8bce65b9 | 692 | fn = fib6_lookup(FIB6_SUBTREE(pn), NULL, saddr); \ |
982f56f3 YH |
693 | else \ |
694 | fn = pn; \ | |
695 | if (fn->fn_flags & RTN_RTINFO) \ | |
696 | goto restart; \ | |
c71099ac | 697 | } \ |
c71099ac | 698 | } \ |
38308473 | 699 | } while (0) |
c71099ac | 700 | |
8ed67789 DL |
701 | static struct rt6_info *ip6_pol_route_lookup(struct net *net, |
702 | struct fib6_table *table, | |
4c9483b2 | 703 | struct flowi6 *fl6, int flags) |
1da177e4 LT |
704 | { |
705 | struct fib6_node *fn; | |
706 | struct rt6_info *rt; | |
707 | ||
c71099ac | 708 | read_lock_bh(&table->tb6_lock); |
4c9483b2 | 709 | fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr); |
c71099ac TG |
710 | restart: |
711 | rt = fn->leaf; | |
4c9483b2 DM |
712 | rt = rt6_device_match(net, rt, &fl6->saddr, fl6->flowi6_oif, flags); |
713 | BACKTRACK(net, &fl6->saddr); | |
c71099ac | 714 | out: |
d8d1f30b | 715 | dst_use(&rt->dst, jiffies); |
c71099ac | 716 | read_unlock_bh(&table->tb6_lock); |
c71099ac TG |
717 | return rt; |
718 | ||
719 | } | |
720 | ||
ea6e574e FW |
721 | struct dst_entry * ip6_route_lookup(struct net *net, struct flowi6 *fl6, |
722 | int flags) | |
723 | { | |
724 | return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_lookup); | |
725 | } | |
726 | EXPORT_SYMBOL_GPL(ip6_route_lookup); | |
727 | ||
9acd9f3a YH |
728 | struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr, |
729 | const struct in6_addr *saddr, int oif, int strict) | |
c71099ac | 730 | { |
4c9483b2 DM |
731 | struct flowi6 fl6 = { |
732 | .flowi6_oif = oif, | |
733 | .daddr = *daddr, | |
c71099ac TG |
734 | }; |
735 | struct dst_entry *dst; | |
77d16f45 | 736 | int flags = strict ? RT6_LOOKUP_F_IFACE : 0; |
c71099ac | 737 | |
adaa70bb | 738 | if (saddr) { |
4c9483b2 | 739 | memcpy(&fl6.saddr, saddr, sizeof(*saddr)); |
adaa70bb TG |
740 | flags |= RT6_LOOKUP_F_HAS_SADDR; |
741 | } | |
742 | ||
4c9483b2 | 743 | dst = fib6_rule_lookup(net, &fl6, flags, ip6_pol_route_lookup); |
c71099ac TG |
744 | if (dst->error == 0) |
745 | return (struct rt6_info *) dst; | |
746 | ||
747 | dst_release(dst); | |
748 | ||
1da177e4 LT |
749 | return NULL; |
750 | } | |
751 | ||
7159039a YH |
752 | EXPORT_SYMBOL(rt6_lookup); |
753 | ||
c71099ac | 754 | /* ip6_ins_rt is called with FREE table->tb6_lock. |
1da177e4 LT |
755 | It takes new route entry, the addition fails by any reason the |
756 | route is freed. In any case, if caller does not hold it, it may | |
757 | be destroyed. | |
758 | */ | |
759 | ||
86872cb5 | 760 | static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info) |
1da177e4 LT |
761 | { |
762 | int err; | |
c71099ac | 763 | struct fib6_table *table; |
1da177e4 | 764 | |
c71099ac TG |
765 | table = rt->rt6i_table; |
766 | write_lock_bh(&table->tb6_lock); | |
86872cb5 | 767 | err = fib6_add(&table->tb6_root, rt, info); |
c71099ac | 768 | write_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
769 | |
770 | return err; | |
771 | } | |
772 | ||
40e22e8f TG |
773 | int ip6_ins_rt(struct rt6_info *rt) |
774 | { | |
4d1169c1 | 775 | struct nl_info info = { |
d1918542 | 776 | .nl_net = dev_net(rt->dst.dev), |
4d1169c1 | 777 | }; |
528c4ceb | 778 | return __ip6_ins_rt(rt, &info); |
40e22e8f TG |
779 | } |
780 | ||
1716a961 | 781 | static struct rt6_info *rt6_alloc_cow(struct rt6_info *ort, |
21efcfa0 | 782 | const struct in6_addr *daddr, |
b71d1d42 | 783 | const struct in6_addr *saddr) |
1da177e4 | 784 | { |
1da177e4 LT |
785 | struct rt6_info *rt; |
786 | ||
787 | /* | |
788 | * Clone the route. | |
789 | */ | |
790 | ||
21efcfa0 | 791 | rt = ip6_rt_copy(ort, daddr); |
1da177e4 LT |
792 | |
793 | if (rt) { | |
14deae41 DM |
794 | int attempts = !in_softirq(); |
795 | ||
38308473 | 796 | if (!(rt->rt6i_flags & RTF_GATEWAY)) { |
bb3c3686 | 797 | if (ort->rt6i_dst.plen != 128 && |
21efcfa0 | 798 | ipv6_addr_equal(&ort->rt6i_dst.addr, daddr)) |
58c4fb86 | 799 | rt->rt6i_flags |= RTF_ANYCAST; |
4e3fd7a0 | 800 | rt->rt6i_gateway = *daddr; |
58c4fb86 | 801 | } |
1da177e4 | 802 | |
1da177e4 | 803 | rt->rt6i_flags |= RTF_CACHE; |
1da177e4 LT |
804 | |
805 | #ifdef CONFIG_IPV6_SUBTREES | |
806 | if (rt->rt6i_src.plen && saddr) { | |
4e3fd7a0 | 807 | rt->rt6i_src.addr = *saddr; |
1da177e4 LT |
808 | rt->rt6i_src.plen = 128; |
809 | } | |
810 | #endif | |
811 | ||
14deae41 | 812 | retry: |
8ade06c6 | 813 | if (rt6_bind_neighbour(rt, rt->dst.dev)) { |
d1918542 | 814 | struct net *net = dev_net(rt->dst.dev); |
14deae41 DM |
815 | int saved_rt_min_interval = |
816 | net->ipv6.sysctl.ip6_rt_gc_min_interval; | |
817 | int saved_rt_elasticity = | |
818 | net->ipv6.sysctl.ip6_rt_gc_elasticity; | |
819 | ||
820 | if (attempts-- > 0) { | |
821 | net->ipv6.sysctl.ip6_rt_gc_elasticity = 1; | |
822 | net->ipv6.sysctl.ip6_rt_gc_min_interval = 0; | |
823 | ||
86393e52 | 824 | ip6_dst_gc(&net->ipv6.ip6_dst_ops); |
14deae41 DM |
825 | |
826 | net->ipv6.sysctl.ip6_rt_gc_elasticity = | |
827 | saved_rt_elasticity; | |
828 | net->ipv6.sysctl.ip6_rt_gc_min_interval = | |
829 | saved_rt_min_interval; | |
830 | goto retry; | |
831 | } | |
832 | ||
f3213831 | 833 | net_warn_ratelimited("Neighbour table overflow\n"); |
d8d1f30b | 834 | dst_free(&rt->dst); |
14deae41 DM |
835 | return NULL; |
836 | } | |
95a9a5ba | 837 | } |
1da177e4 | 838 | |
95a9a5ba YH |
839 | return rt; |
840 | } | |
1da177e4 | 841 | |
21efcfa0 ED |
842 | static struct rt6_info *rt6_alloc_clone(struct rt6_info *ort, |
843 | const struct in6_addr *daddr) | |
299d9939 | 844 | { |
21efcfa0 ED |
845 | struct rt6_info *rt = ip6_rt_copy(ort, daddr); |
846 | ||
299d9939 | 847 | if (rt) { |
299d9939 | 848 | rt->rt6i_flags |= RTF_CACHE; |
97cac082 | 849 | rt->n = neigh_clone(ort->n); |
299d9939 YH |
850 | } |
851 | return rt; | |
852 | } | |
853 | ||
8ed67789 | 854 | static struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table, int oif, |
4c9483b2 | 855 | struct flowi6 *fl6, int flags) |
1da177e4 LT |
856 | { |
857 | struct fib6_node *fn; | |
519fbd87 | 858 | struct rt6_info *rt, *nrt; |
c71099ac | 859 | int strict = 0; |
1da177e4 | 860 | int attempts = 3; |
519fbd87 | 861 | int err; |
53b7997f | 862 | int reachable = net->ipv6.devconf_all->forwarding ? 0 : RT6_LOOKUP_F_REACHABLE; |
1da177e4 | 863 | |
77d16f45 | 864 | strict |= flags & RT6_LOOKUP_F_IFACE; |
1da177e4 LT |
865 | |
866 | relookup: | |
c71099ac | 867 | read_lock_bh(&table->tb6_lock); |
1da177e4 | 868 | |
8238dd06 | 869 | restart_2: |
4c9483b2 | 870 | fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr); |
1da177e4 LT |
871 | |
872 | restart: | |
4acad72d | 873 | rt = rt6_select(fn, oif, strict | reachable); |
8ed67789 | 874 | |
4c9483b2 | 875 | BACKTRACK(net, &fl6->saddr); |
8ed67789 | 876 | if (rt == net->ipv6.ip6_null_entry || |
8238dd06 | 877 | rt->rt6i_flags & RTF_CACHE) |
1ddef044 | 878 | goto out; |
1da177e4 | 879 | |
d8d1f30b | 880 | dst_hold(&rt->dst); |
c71099ac | 881 | read_unlock_bh(&table->tb6_lock); |
fb9de91e | 882 | |
97cac082 | 883 | if (!rt->n && !(rt->rt6i_flags & RTF_NONEXTHOP)) |
4c9483b2 | 884 | nrt = rt6_alloc_cow(rt, &fl6->daddr, &fl6->saddr); |
7343ff31 | 885 | else if (!(rt->dst.flags & DST_HOST)) |
4c9483b2 | 886 | nrt = rt6_alloc_clone(rt, &fl6->daddr); |
7343ff31 DM |
887 | else |
888 | goto out2; | |
e40cf353 | 889 | |
d8d1f30b | 890 | dst_release(&rt->dst); |
8ed67789 | 891 | rt = nrt ? : net->ipv6.ip6_null_entry; |
1da177e4 | 892 | |
d8d1f30b | 893 | dst_hold(&rt->dst); |
519fbd87 | 894 | if (nrt) { |
40e22e8f | 895 | err = ip6_ins_rt(nrt); |
519fbd87 | 896 | if (!err) |
1da177e4 | 897 | goto out2; |
1da177e4 | 898 | } |
1da177e4 | 899 | |
519fbd87 YH |
900 | if (--attempts <= 0) |
901 | goto out2; | |
902 | ||
903 | /* | |
c71099ac | 904 | * Race condition! In the gap, when table->tb6_lock was |
519fbd87 YH |
905 | * released someone could insert this route. Relookup. |
906 | */ | |
d8d1f30b | 907 | dst_release(&rt->dst); |
519fbd87 YH |
908 | goto relookup; |
909 | ||
910 | out: | |
8238dd06 YH |
911 | if (reachable) { |
912 | reachable = 0; | |
913 | goto restart_2; | |
914 | } | |
d8d1f30b | 915 | dst_hold(&rt->dst); |
c71099ac | 916 | read_unlock_bh(&table->tb6_lock); |
1da177e4 | 917 | out2: |
d8d1f30b CG |
918 | rt->dst.lastuse = jiffies; |
919 | rt->dst.__use++; | |
c71099ac TG |
920 | |
921 | return rt; | |
1da177e4 LT |
922 | } |
923 | ||
8ed67789 | 924 | static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table, |
4c9483b2 | 925 | struct flowi6 *fl6, int flags) |
4acad72d | 926 | { |
4c9483b2 | 927 | return ip6_pol_route(net, table, fl6->flowi6_iif, fl6, flags); |
4acad72d PE |
928 | } |
929 | ||
72331bc0 SL |
930 | static struct dst_entry *ip6_route_input_lookup(struct net *net, |
931 | struct net_device *dev, | |
932 | struct flowi6 *fl6, int flags) | |
933 | { | |
934 | if (rt6_need_strict(&fl6->daddr) && dev->type != ARPHRD_PIMREG) | |
935 | flags |= RT6_LOOKUP_F_IFACE; | |
936 | ||
937 | return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_input); | |
938 | } | |
939 | ||
c71099ac TG |
940 | void ip6_route_input(struct sk_buff *skb) |
941 | { | |
b71d1d42 | 942 | const struct ipv6hdr *iph = ipv6_hdr(skb); |
c346dca1 | 943 | struct net *net = dev_net(skb->dev); |
adaa70bb | 944 | int flags = RT6_LOOKUP_F_HAS_SADDR; |
4c9483b2 DM |
945 | struct flowi6 fl6 = { |
946 | .flowi6_iif = skb->dev->ifindex, | |
947 | .daddr = iph->daddr, | |
948 | .saddr = iph->saddr, | |
38308473 | 949 | .flowlabel = (* (__be32 *) iph) & IPV6_FLOWINFO_MASK, |
4c9483b2 DM |
950 | .flowi6_mark = skb->mark, |
951 | .flowi6_proto = iph->nexthdr, | |
c71099ac | 952 | }; |
adaa70bb | 953 | |
72331bc0 | 954 | skb_dst_set(skb, ip6_route_input_lookup(net, skb->dev, &fl6, flags)); |
c71099ac TG |
955 | } |
956 | ||
8ed67789 | 957 | static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table, |
4c9483b2 | 958 | struct flowi6 *fl6, int flags) |
1da177e4 | 959 | { |
4c9483b2 | 960 | return ip6_pol_route(net, table, fl6->flowi6_oif, fl6, flags); |
c71099ac TG |
961 | } |
962 | ||
9c7a4f9c | 963 | struct dst_entry * ip6_route_output(struct net *net, const struct sock *sk, |
4c9483b2 | 964 | struct flowi6 *fl6) |
c71099ac TG |
965 | { |
966 | int flags = 0; | |
967 | ||
4dc27d1c DM |
968 | fl6->flowi6_iif = net->loopback_dev->ifindex; |
969 | ||
4c9483b2 | 970 | if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl6->daddr)) |
77d16f45 | 971 | flags |= RT6_LOOKUP_F_IFACE; |
c71099ac | 972 | |
4c9483b2 | 973 | if (!ipv6_addr_any(&fl6->saddr)) |
adaa70bb | 974 | flags |= RT6_LOOKUP_F_HAS_SADDR; |
0c9a2ac1 YH |
975 | else if (sk) |
976 | flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs); | |
adaa70bb | 977 | |
4c9483b2 | 978 | return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_output); |
1da177e4 LT |
979 | } |
980 | ||
7159039a | 981 | EXPORT_SYMBOL(ip6_route_output); |
1da177e4 | 982 | |
2774c131 | 983 | struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_orig) |
14e50e57 | 984 | { |
5c1e6aa3 | 985 | struct rt6_info *rt, *ort = (struct rt6_info *) dst_orig; |
14e50e57 DM |
986 | struct dst_entry *new = NULL; |
987 | ||
f5b0a874 | 988 | rt = dst_alloc(&ip6_dst_blackhole_ops, ort->dst.dev, 1, DST_OBSOLETE_NONE, 0); |
14e50e57 | 989 | if (rt) { |
d8d1f30b | 990 | new = &rt->dst; |
14e50e57 | 991 | |
8104891b SK |
992 | memset(new + 1, 0, sizeof(*rt) - sizeof(*new)); |
993 | rt6_init_peer(rt, net->ipv6.peers); | |
994 | ||
14e50e57 | 995 | new->__use = 1; |
352e512c HX |
996 | new->input = dst_discard; |
997 | new->output = dst_discard; | |
14e50e57 | 998 | |
21efcfa0 ED |
999 | if (dst_metrics_read_only(&ort->dst)) |
1000 | new->_metrics = ort->dst._metrics; | |
1001 | else | |
1002 | dst_copy_metrics(new, &ort->dst); | |
14e50e57 DM |
1003 | rt->rt6i_idev = ort->rt6i_idev; |
1004 | if (rt->rt6i_idev) | |
1005 | in6_dev_hold(rt->rt6i_idev); | |
14e50e57 | 1006 | |
4e3fd7a0 | 1007 | rt->rt6i_gateway = ort->rt6i_gateway; |
1716a961 G |
1008 | rt->rt6i_flags = ort->rt6i_flags; |
1009 | rt6_clean_expires(rt); | |
14e50e57 DM |
1010 | rt->rt6i_metric = 0; |
1011 | ||
1012 | memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key)); | |
1013 | #ifdef CONFIG_IPV6_SUBTREES | |
1014 | memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key)); | |
1015 | #endif | |
1016 | ||
1017 | dst_free(new); | |
1018 | } | |
1019 | ||
69ead7af DM |
1020 | dst_release(dst_orig); |
1021 | return new ? new : ERR_PTR(-ENOMEM); | |
14e50e57 | 1022 | } |
14e50e57 | 1023 | |
1da177e4 LT |
1024 | /* |
1025 | * Destination cache support functions | |
1026 | */ | |
1027 | ||
1028 | static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie) | |
1029 | { | |
1030 | struct rt6_info *rt; | |
1031 | ||
1032 | rt = (struct rt6_info *) dst; | |
1033 | ||
6431cbc2 DM |
1034 | if (rt->rt6i_node && (rt->rt6i_node->fn_sernum == cookie)) { |
1035 | if (rt->rt6i_peer_genid != rt6_peer_genid()) { | |
97bab73f | 1036 | if (!rt6_has_peer(rt)) |
6431cbc2 DM |
1037 | rt6_bind_peer(rt, 0); |
1038 | rt->rt6i_peer_genid = rt6_peer_genid(); | |
1039 | } | |
1da177e4 | 1040 | return dst; |
6431cbc2 | 1041 | } |
1da177e4 LT |
1042 | return NULL; |
1043 | } | |
1044 | ||
1045 | static struct dst_entry *ip6_negative_advice(struct dst_entry *dst) | |
1046 | { | |
1047 | struct rt6_info *rt = (struct rt6_info *) dst; | |
1048 | ||
1049 | if (rt) { | |
54c1a859 YH |
1050 | if (rt->rt6i_flags & RTF_CACHE) { |
1051 | if (rt6_check_expired(rt)) { | |
1052 | ip6_del_rt(rt); | |
1053 | dst = NULL; | |
1054 | } | |
1055 | } else { | |
1da177e4 | 1056 | dst_release(dst); |
54c1a859 YH |
1057 | dst = NULL; |
1058 | } | |
1da177e4 | 1059 | } |
54c1a859 | 1060 | return dst; |
1da177e4 LT |
1061 | } |
1062 | ||
1063 | static void ip6_link_failure(struct sk_buff *skb) | |
1064 | { | |
1065 | struct rt6_info *rt; | |
1066 | ||
3ffe533c | 1067 | icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0); |
1da177e4 | 1068 | |
adf30907 | 1069 | rt = (struct rt6_info *) skb_dst(skb); |
1da177e4 | 1070 | if (rt) { |
1716a961 G |
1071 | if (rt->rt6i_flags & RTF_CACHE) |
1072 | rt6_update_expires(rt, 0); | |
1073 | else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT)) | |
1da177e4 LT |
1074 | rt->rt6i_node->fn_sernum = -1; |
1075 | } | |
1076 | } | |
1077 | ||
6700c270 DM |
1078 | static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk, |
1079 | struct sk_buff *skb, u32 mtu) | |
1da177e4 LT |
1080 | { |
1081 | struct rt6_info *rt6 = (struct rt6_info*)dst; | |
1082 | ||
81aded24 | 1083 | dst_confirm(dst); |
1da177e4 | 1084 | if (mtu < dst_mtu(dst) && rt6->rt6i_dst.plen == 128) { |
81aded24 DM |
1085 | struct net *net = dev_net(dst->dev); |
1086 | ||
1da177e4 LT |
1087 | rt6->rt6i_flags |= RTF_MODIFIED; |
1088 | if (mtu < IPV6_MIN_MTU) { | |
defb3519 | 1089 | u32 features = dst_metric(dst, RTAX_FEATURES); |
1da177e4 | 1090 | mtu = IPV6_MIN_MTU; |
defb3519 DM |
1091 | features |= RTAX_FEATURE_ALLFRAG; |
1092 | dst_metric_set(dst, RTAX_FEATURES, features); | |
1da177e4 | 1093 | } |
defb3519 | 1094 | dst_metric_set(dst, RTAX_MTU, mtu); |
81aded24 | 1095 | rt6_update_expires(rt6, net->ipv6.sysctl.ip6_rt_mtu_expires); |
1da177e4 LT |
1096 | } |
1097 | } | |
1098 | ||
42ae66c8 DM |
1099 | void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu, |
1100 | int oif, u32 mark) | |
81aded24 DM |
1101 | { |
1102 | const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data; | |
1103 | struct dst_entry *dst; | |
1104 | struct flowi6 fl6; | |
1105 | ||
1106 | memset(&fl6, 0, sizeof(fl6)); | |
1107 | fl6.flowi6_oif = oif; | |
1108 | fl6.flowi6_mark = mark; | |
3e12939a | 1109 | fl6.flowi6_flags = 0; |
81aded24 DM |
1110 | fl6.daddr = iph->daddr; |
1111 | fl6.saddr = iph->saddr; | |
1112 | fl6.flowlabel = (*(__be32 *) iph) & IPV6_FLOWINFO_MASK; | |
1113 | ||
1114 | dst = ip6_route_output(net, NULL, &fl6); | |
1115 | if (!dst->error) | |
6700c270 | 1116 | ip6_rt_update_pmtu(dst, NULL, skb, ntohl(mtu)); |
81aded24 DM |
1117 | dst_release(dst); |
1118 | } | |
1119 | EXPORT_SYMBOL_GPL(ip6_update_pmtu); | |
1120 | ||
1121 | void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu) | |
1122 | { | |
1123 | ip6_update_pmtu(skb, sock_net(sk), mtu, | |
1124 | sk->sk_bound_dev_if, sk->sk_mark); | |
1125 | } | |
1126 | EXPORT_SYMBOL_GPL(ip6_sk_update_pmtu); | |
1127 | ||
3a5ad2ee DM |
1128 | void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark) |
1129 | { | |
1130 | const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data; | |
1131 | struct dst_entry *dst; | |
1132 | struct flowi6 fl6; | |
1133 | ||
1134 | memset(&fl6, 0, sizeof(fl6)); | |
1135 | fl6.flowi6_oif = oif; | |
1136 | fl6.flowi6_mark = mark; | |
1137 | fl6.flowi6_flags = 0; | |
1138 | fl6.daddr = iph->daddr; | |
1139 | fl6.saddr = iph->saddr; | |
1140 | fl6.flowlabel = (*(__be32 *) iph) & IPV6_FLOWINFO_MASK; | |
1141 | ||
1142 | dst = ip6_route_output(net, NULL, &fl6); | |
1143 | if (!dst->error) | |
6700c270 | 1144 | rt6_do_redirect(dst, NULL, skb); |
3a5ad2ee DM |
1145 | dst_release(dst); |
1146 | } | |
1147 | EXPORT_SYMBOL_GPL(ip6_redirect); | |
1148 | ||
1149 | void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk) | |
1150 | { | |
1151 | ip6_redirect(skb, sock_net(sk), sk->sk_bound_dev_if, sk->sk_mark); | |
1152 | } | |
1153 | EXPORT_SYMBOL_GPL(ip6_sk_redirect); | |
1154 | ||
0dbaee3b | 1155 | static unsigned int ip6_default_advmss(const struct dst_entry *dst) |
1da177e4 | 1156 | { |
0dbaee3b DM |
1157 | struct net_device *dev = dst->dev; |
1158 | unsigned int mtu = dst_mtu(dst); | |
1159 | struct net *net = dev_net(dev); | |
1160 | ||
1da177e4 LT |
1161 | mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr); |
1162 | ||
5578689a DL |
1163 | if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss) |
1164 | mtu = net->ipv6.sysctl.ip6_rt_min_advmss; | |
1da177e4 LT |
1165 | |
1166 | /* | |
1ab1457c YH |
1167 | * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and |
1168 | * corresponding MSS is IPV6_MAXPLEN - tcp_header_size. | |
1169 | * IPV6_MAXPLEN is also valid and means: "any MSS, | |
1da177e4 LT |
1170 | * rely only on pmtu discovery" |
1171 | */ | |
1172 | if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr)) | |
1173 | mtu = IPV6_MAXPLEN; | |
1174 | return mtu; | |
1175 | } | |
1176 | ||
ebb762f2 | 1177 | static unsigned int ip6_mtu(const struct dst_entry *dst) |
d33e4553 | 1178 | { |
d33e4553 | 1179 | struct inet6_dev *idev; |
618f9bc7 SK |
1180 | unsigned int mtu = dst_metric_raw(dst, RTAX_MTU); |
1181 | ||
1182 | if (mtu) | |
1183 | return mtu; | |
1184 | ||
1185 | mtu = IPV6_MIN_MTU; | |
d33e4553 DM |
1186 | |
1187 | rcu_read_lock(); | |
1188 | idev = __in6_dev_get(dst->dev); | |
1189 | if (idev) | |
1190 | mtu = idev->cnf.mtu6; | |
1191 | rcu_read_unlock(); | |
1192 | ||
1193 | return mtu; | |
1194 | } | |
1195 | ||
3b00944c YH |
1196 | static struct dst_entry *icmp6_dst_gc_list; |
1197 | static DEFINE_SPINLOCK(icmp6_dst_lock); | |
5d0bbeeb | 1198 | |
3b00944c | 1199 | struct dst_entry *icmp6_dst_alloc(struct net_device *dev, |
1da177e4 | 1200 | struct neighbour *neigh, |
87a11578 | 1201 | struct flowi6 *fl6) |
1da177e4 | 1202 | { |
87a11578 | 1203 | struct dst_entry *dst; |
1da177e4 LT |
1204 | struct rt6_info *rt; |
1205 | struct inet6_dev *idev = in6_dev_get(dev); | |
c346dca1 | 1206 | struct net *net = dev_net(dev); |
1da177e4 | 1207 | |
38308473 | 1208 | if (unlikely(!idev)) |
122bdf67 | 1209 | return ERR_PTR(-ENODEV); |
1da177e4 | 1210 | |
8b96d22d | 1211 | rt = ip6_dst_alloc(net, dev, 0, NULL); |
38308473 | 1212 | if (unlikely(!rt)) { |
1da177e4 | 1213 | in6_dev_put(idev); |
87a11578 | 1214 | dst = ERR_PTR(-ENOMEM); |
1da177e4 LT |
1215 | goto out; |
1216 | } | |
1217 | ||
1da177e4 LT |
1218 | if (neigh) |
1219 | neigh_hold(neigh); | |
14deae41 | 1220 | else { |
f894cbf8 | 1221 | neigh = ip6_neigh_lookup(&rt->dst, NULL, &fl6->daddr); |
b43faac6 | 1222 | if (IS_ERR(neigh)) { |
252c3d84 | 1223 | in6_dev_put(idev); |
b43faac6 DM |
1224 | dst_free(&rt->dst); |
1225 | return ERR_CAST(neigh); | |
1226 | } | |
14deae41 | 1227 | } |
1da177e4 | 1228 | |
8e2ec639 YZ |
1229 | rt->dst.flags |= DST_HOST; |
1230 | rt->dst.output = ip6_output; | |
97cac082 | 1231 | rt->n = neigh; |
d8d1f30b | 1232 | atomic_set(&rt->dst.__refcnt, 1); |
87a11578 | 1233 | rt->rt6i_dst.addr = fl6->daddr; |
8e2ec639 YZ |
1234 | rt->rt6i_dst.plen = 128; |
1235 | rt->rt6i_idev = idev; | |
7011687f | 1236 | dst_metric_set(&rt->dst, RTAX_HOPLIMIT, 255); |
1da177e4 | 1237 | |
3b00944c | 1238 | spin_lock_bh(&icmp6_dst_lock); |
d8d1f30b CG |
1239 | rt->dst.next = icmp6_dst_gc_list; |
1240 | icmp6_dst_gc_list = &rt->dst; | |
3b00944c | 1241 | spin_unlock_bh(&icmp6_dst_lock); |
1da177e4 | 1242 | |
5578689a | 1243 | fib6_force_start_gc(net); |
1da177e4 | 1244 | |
87a11578 DM |
1245 | dst = xfrm_lookup(net, &rt->dst, flowi6_to_flowi(fl6), NULL, 0); |
1246 | ||
1da177e4 | 1247 | out: |
87a11578 | 1248 | return dst; |
1da177e4 LT |
1249 | } |
1250 | ||
3d0f24a7 | 1251 | int icmp6_dst_gc(void) |
1da177e4 | 1252 | { |
e9476e95 | 1253 | struct dst_entry *dst, **pprev; |
3d0f24a7 | 1254 | int more = 0; |
1da177e4 | 1255 | |
3b00944c YH |
1256 | spin_lock_bh(&icmp6_dst_lock); |
1257 | pprev = &icmp6_dst_gc_list; | |
5d0bbeeb | 1258 | |
1da177e4 LT |
1259 | while ((dst = *pprev) != NULL) { |
1260 | if (!atomic_read(&dst->__refcnt)) { | |
1261 | *pprev = dst->next; | |
1262 | dst_free(dst); | |
1da177e4 LT |
1263 | } else { |
1264 | pprev = &dst->next; | |
3d0f24a7 | 1265 | ++more; |
1da177e4 LT |
1266 | } |
1267 | } | |
1268 | ||
3b00944c | 1269 | spin_unlock_bh(&icmp6_dst_lock); |
5d0bbeeb | 1270 | |
3d0f24a7 | 1271 | return more; |
1da177e4 LT |
1272 | } |
1273 | ||
1e493d19 DM |
1274 | static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg), |
1275 | void *arg) | |
1276 | { | |
1277 | struct dst_entry *dst, **pprev; | |
1278 | ||
1279 | spin_lock_bh(&icmp6_dst_lock); | |
1280 | pprev = &icmp6_dst_gc_list; | |
1281 | while ((dst = *pprev) != NULL) { | |
1282 | struct rt6_info *rt = (struct rt6_info *) dst; | |
1283 | if (func(rt, arg)) { | |
1284 | *pprev = dst->next; | |
1285 | dst_free(dst); | |
1286 | } else { | |
1287 | pprev = &dst->next; | |
1288 | } | |
1289 | } | |
1290 | spin_unlock_bh(&icmp6_dst_lock); | |
1291 | } | |
1292 | ||
569d3645 | 1293 | static int ip6_dst_gc(struct dst_ops *ops) |
1da177e4 | 1294 | { |
1da177e4 | 1295 | unsigned long now = jiffies; |
86393e52 | 1296 | struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops); |
7019b78e DL |
1297 | int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval; |
1298 | int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size; | |
1299 | int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity; | |
1300 | int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout; | |
1301 | unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc; | |
fc66f95c | 1302 | int entries; |
7019b78e | 1303 | |
fc66f95c | 1304 | entries = dst_entries_get_fast(ops); |
7019b78e | 1305 | if (time_after(rt_last_gc + rt_min_interval, now) && |
fc66f95c | 1306 | entries <= rt_max_size) |
1da177e4 LT |
1307 | goto out; |
1308 | ||
6891a346 BT |
1309 | net->ipv6.ip6_rt_gc_expire++; |
1310 | fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net); | |
1311 | net->ipv6.ip6_rt_last_gc = now; | |
fc66f95c ED |
1312 | entries = dst_entries_get_slow(ops); |
1313 | if (entries < ops->gc_thresh) | |
7019b78e | 1314 | net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1; |
1da177e4 | 1315 | out: |
7019b78e | 1316 | net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity; |
fc66f95c | 1317 | return entries > rt_max_size; |
1da177e4 LT |
1318 | } |
1319 | ||
1320 | /* Clean host part of a prefix. Not necessary in radix tree, | |
1321 | but results in cleaner routing tables. | |
1322 | ||
1323 | Remove it only when all the things will work! | |
1324 | */ | |
1325 | ||
6b75d090 | 1326 | int ip6_dst_hoplimit(struct dst_entry *dst) |
1da177e4 | 1327 | { |
5170ae82 | 1328 | int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT); |
a02e4b7d | 1329 | if (hoplimit == 0) { |
6b75d090 | 1330 | struct net_device *dev = dst->dev; |
c68f24cc ED |
1331 | struct inet6_dev *idev; |
1332 | ||
1333 | rcu_read_lock(); | |
1334 | idev = __in6_dev_get(dev); | |
1335 | if (idev) | |
6b75d090 | 1336 | hoplimit = idev->cnf.hop_limit; |
c68f24cc | 1337 | else |
53b7997f | 1338 | hoplimit = dev_net(dev)->ipv6.devconf_all->hop_limit; |
c68f24cc | 1339 | rcu_read_unlock(); |
1da177e4 LT |
1340 | } |
1341 | return hoplimit; | |
1342 | } | |
abbf46ae | 1343 | EXPORT_SYMBOL(ip6_dst_hoplimit); |
1da177e4 LT |
1344 | |
1345 | /* | |
1346 | * | |
1347 | */ | |
1348 | ||
86872cb5 | 1349 | int ip6_route_add(struct fib6_config *cfg) |
1da177e4 LT |
1350 | { |
1351 | int err; | |
5578689a | 1352 | struct net *net = cfg->fc_nlinfo.nl_net; |
1da177e4 LT |
1353 | struct rt6_info *rt = NULL; |
1354 | struct net_device *dev = NULL; | |
1355 | struct inet6_dev *idev = NULL; | |
c71099ac | 1356 | struct fib6_table *table; |
1da177e4 LT |
1357 | int addr_type; |
1358 | ||
86872cb5 | 1359 | if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128) |
1da177e4 LT |
1360 | return -EINVAL; |
1361 | #ifndef CONFIG_IPV6_SUBTREES | |
86872cb5 | 1362 | if (cfg->fc_src_len) |
1da177e4 LT |
1363 | return -EINVAL; |
1364 | #endif | |
86872cb5 | 1365 | if (cfg->fc_ifindex) { |
1da177e4 | 1366 | err = -ENODEV; |
5578689a | 1367 | dev = dev_get_by_index(net, cfg->fc_ifindex); |
1da177e4 LT |
1368 | if (!dev) |
1369 | goto out; | |
1370 | idev = in6_dev_get(dev); | |
1371 | if (!idev) | |
1372 | goto out; | |
1373 | } | |
1374 | ||
86872cb5 TG |
1375 | if (cfg->fc_metric == 0) |
1376 | cfg->fc_metric = IP6_RT_PRIO_USER; | |
1da177e4 | 1377 | |
d71314b4 | 1378 | err = -ENOBUFS; |
38308473 DM |
1379 | if (cfg->fc_nlinfo.nlh && |
1380 | !(cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_CREATE)) { | |
d71314b4 | 1381 | table = fib6_get_table(net, cfg->fc_table); |
38308473 | 1382 | if (!table) { |
f3213831 | 1383 | pr_warn("NLM_F_CREATE should be specified when creating new route\n"); |
d71314b4 MV |
1384 | table = fib6_new_table(net, cfg->fc_table); |
1385 | } | |
1386 | } else { | |
1387 | table = fib6_new_table(net, cfg->fc_table); | |
1388 | } | |
38308473 DM |
1389 | |
1390 | if (!table) | |
c71099ac | 1391 | goto out; |
c71099ac | 1392 | |
8b96d22d | 1393 | rt = ip6_dst_alloc(net, NULL, DST_NOCOUNT, table); |
1da177e4 | 1394 | |
38308473 | 1395 | if (!rt) { |
1da177e4 LT |
1396 | err = -ENOMEM; |
1397 | goto out; | |
1398 | } | |
1399 | ||
d8d1f30b | 1400 | rt->dst.obsolete = -1; |
1716a961 G |
1401 | |
1402 | if (cfg->fc_flags & RTF_EXPIRES) | |
1403 | rt6_set_expires(rt, jiffies + | |
1404 | clock_t_to_jiffies(cfg->fc_expires)); | |
1405 | else | |
1406 | rt6_clean_expires(rt); | |
1da177e4 | 1407 | |
86872cb5 TG |
1408 | if (cfg->fc_protocol == RTPROT_UNSPEC) |
1409 | cfg->fc_protocol = RTPROT_BOOT; | |
1410 | rt->rt6i_protocol = cfg->fc_protocol; | |
1411 | ||
1412 | addr_type = ipv6_addr_type(&cfg->fc_dst); | |
1da177e4 LT |
1413 | |
1414 | if (addr_type & IPV6_ADDR_MULTICAST) | |
d8d1f30b | 1415 | rt->dst.input = ip6_mc_input; |
ab79ad14 MŻ |
1416 | else if (cfg->fc_flags & RTF_LOCAL) |
1417 | rt->dst.input = ip6_input; | |
1da177e4 | 1418 | else |
d8d1f30b | 1419 | rt->dst.input = ip6_forward; |
1da177e4 | 1420 | |
d8d1f30b | 1421 | rt->dst.output = ip6_output; |
1da177e4 | 1422 | |
86872cb5 TG |
1423 | ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len); |
1424 | rt->rt6i_dst.plen = cfg->fc_dst_len; | |
1da177e4 | 1425 | if (rt->rt6i_dst.plen == 128) |
11d53b49 | 1426 | rt->dst.flags |= DST_HOST; |
1da177e4 | 1427 | |
8e2ec639 YZ |
1428 | if (!(rt->dst.flags & DST_HOST) && cfg->fc_mx) { |
1429 | u32 *metrics = kzalloc(sizeof(u32) * RTAX_MAX, GFP_KERNEL); | |
1430 | if (!metrics) { | |
1431 | err = -ENOMEM; | |
1432 | goto out; | |
1433 | } | |
1434 | dst_init_metrics(&rt->dst, metrics, 0); | |
1435 | } | |
1da177e4 | 1436 | #ifdef CONFIG_IPV6_SUBTREES |
86872cb5 TG |
1437 | ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len); |
1438 | rt->rt6i_src.plen = cfg->fc_src_len; | |
1da177e4 LT |
1439 | #endif |
1440 | ||
86872cb5 | 1441 | rt->rt6i_metric = cfg->fc_metric; |
1da177e4 LT |
1442 | |
1443 | /* We cannot add true routes via loopback here, | |
1444 | they would result in kernel looping; promote them to reject routes | |
1445 | */ | |
86872cb5 | 1446 | if ((cfg->fc_flags & RTF_REJECT) || |
38308473 DM |
1447 | (dev && (dev->flags & IFF_LOOPBACK) && |
1448 | !(addr_type & IPV6_ADDR_LOOPBACK) && | |
1449 | !(cfg->fc_flags & RTF_LOCAL))) { | |
1da177e4 | 1450 | /* hold loopback dev/idev if we haven't done so. */ |
5578689a | 1451 | if (dev != net->loopback_dev) { |
1da177e4 LT |
1452 | if (dev) { |
1453 | dev_put(dev); | |
1454 | in6_dev_put(idev); | |
1455 | } | |
5578689a | 1456 | dev = net->loopback_dev; |
1da177e4 LT |
1457 | dev_hold(dev); |
1458 | idev = in6_dev_get(dev); | |
1459 | if (!idev) { | |
1460 | err = -ENODEV; | |
1461 | goto out; | |
1462 | } | |
1463 | } | |
d8d1f30b CG |
1464 | rt->dst.output = ip6_pkt_discard_out; |
1465 | rt->dst.input = ip6_pkt_discard; | |
1466 | rt->dst.error = -ENETUNREACH; | |
1da177e4 LT |
1467 | rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP; |
1468 | goto install_route; | |
1469 | } | |
1470 | ||
86872cb5 | 1471 | if (cfg->fc_flags & RTF_GATEWAY) { |
b71d1d42 | 1472 | const struct in6_addr *gw_addr; |
1da177e4 LT |
1473 | int gwa_type; |
1474 | ||
86872cb5 | 1475 | gw_addr = &cfg->fc_gateway; |
4e3fd7a0 | 1476 | rt->rt6i_gateway = *gw_addr; |
1da177e4 LT |
1477 | gwa_type = ipv6_addr_type(gw_addr); |
1478 | ||
1479 | if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) { | |
1480 | struct rt6_info *grt; | |
1481 | ||
1482 | /* IPv6 strictly inhibits using not link-local | |
1483 | addresses as nexthop address. | |
1484 | Otherwise, router will not able to send redirects. | |
1485 | It is very good, but in some (rare!) circumstances | |
1486 | (SIT, PtP, NBMA NOARP links) it is handy to allow | |
1487 | some exceptions. --ANK | |
1488 | */ | |
1489 | err = -EINVAL; | |
38308473 | 1490 | if (!(gwa_type & IPV6_ADDR_UNICAST)) |
1da177e4 LT |
1491 | goto out; |
1492 | ||
5578689a | 1493 | grt = rt6_lookup(net, gw_addr, NULL, cfg->fc_ifindex, 1); |
1da177e4 LT |
1494 | |
1495 | err = -EHOSTUNREACH; | |
38308473 | 1496 | if (!grt) |
1da177e4 LT |
1497 | goto out; |
1498 | if (dev) { | |
d1918542 | 1499 | if (dev != grt->dst.dev) { |
d8d1f30b | 1500 | dst_release(&grt->dst); |
1da177e4 LT |
1501 | goto out; |
1502 | } | |
1503 | } else { | |
d1918542 | 1504 | dev = grt->dst.dev; |
1da177e4 LT |
1505 | idev = grt->rt6i_idev; |
1506 | dev_hold(dev); | |
1507 | in6_dev_hold(grt->rt6i_idev); | |
1508 | } | |
38308473 | 1509 | if (!(grt->rt6i_flags & RTF_GATEWAY)) |
1da177e4 | 1510 | err = 0; |
d8d1f30b | 1511 | dst_release(&grt->dst); |
1da177e4 LT |
1512 | |
1513 | if (err) | |
1514 | goto out; | |
1515 | } | |
1516 | err = -EINVAL; | |
38308473 | 1517 | if (!dev || (dev->flags & IFF_LOOPBACK)) |
1da177e4 LT |
1518 | goto out; |
1519 | } | |
1520 | ||
1521 | err = -ENODEV; | |
38308473 | 1522 | if (!dev) |
1da177e4 LT |
1523 | goto out; |
1524 | ||
c3968a85 DW |
1525 | if (!ipv6_addr_any(&cfg->fc_prefsrc)) { |
1526 | if (!ipv6_chk_addr(net, &cfg->fc_prefsrc, dev, 0)) { | |
1527 | err = -EINVAL; | |
1528 | goto out; | |
1529 | } | |
4e3fd7a0 | 1530 | rt->rt6i_prefsrc.addr = cfg->fc_prefsrc; |
c3968a85 DW |
1531 | rt->rt6i_prefsrc.plen = 128; |
1532 | } else | |
1533 | rt->rt6i_prefsrc.plen = 0; | |
1534 | ||
86872cb5 | 1535 | if (cfg->fc_flags & (RTF_GATEWAY | RTF_NONEXTHOP)) { |
8ade06c6 | 1536 | err = rt6_bind_neighbour(rt, dev); |
f83c7790 | 1537 | if (err) |
1da177e4 | 1538 | goto out; |
1da177e4 LT |
1539 | } |
1540 | ||
86872cb5 | 1541 | rt->rt6i_flags = cfg->fc_flags; |
1da177e4 LT |
1542 | |
1543 | install_route: | |
86872cb5 TG |
1544 | if (cfg->fc_mx) { |
1545 | struct nlattr *nla; | |
1546 | int remaining; | |
1547 | ||
1548 | nla_for_each_attr(nla, cfg->fc_mx, cfg->fc_mx_len, remaining) { | |
8f4c1f9b | 1549 | int type = nla_type(nla); |
86872cb5 TG |
1550 | |
1551 | if (type) { | |
1552 | if (type > RTAX_MAX) { | |
1da177e4 LT |
1553 | err = -EINVAL; |
1554 | goto out; | |
1555 | } | |
86872cb5 | 1556 | |
defb3519 | 1557 | dst_metric_set(&rt->dst, type, nla_get_u32(nla)); |
1da177e4 | 1558 | } |
1da177e4 LT |
1559 | } |
1560 | } | |
1561 | ||
d8d1f30b | 1562 | rt->dst.dev = dev; |
1da177e4 | 1563 | rt->rt6i_idev = idev; |
c71099ac | 1564 | rt->rt6i_table = table; |
63152fc0 | 1565 | |
c346dca1 | 1566 | cfg->fc_nlinfo.nl_net = dev_net(dev); |
63152fc0 | 1567 | |
86872cb5 | 1568 | return __ip6_ins_rt(rt, &cfg->fc_nlinfo); |
1da177e4 LT |
1569 | |
1570 | out: | |
1571 | if (dev) | |
1572 | dev_put(dev); | |
1573 | if (idev) | |
1574 | in6_dev_put(idev); | |
1575 | if (rt) | |
d8d1f30b | 1576 | dst_free(&rt->dst); |
1da177e4 LT |
1577 | return err; |
1578 | } | |
1579 | ||
86872cb5 | 1580 | static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info) |
1da177e4 LT |
1581 | { |
1582 | int err; | |
c71099ac | 1583 | struct fib6_table *table; |
d1918542 | 1584 | struct net *net = dev_net(rt->dst.dev); |
1da177e4 | 1585 | |
8ed67789 | 1586 | if (rt == net->ipv6.ip6_null_entry) |
6c813a72 PM |
1587 | return -ENOENT; |
1588 | ||
c71099ac TG |
1589 | table = rt->rt6i_table; |
1590 | write_lock_bh(&table->tb6_lock); | |
1da177e4 | 1591 | |
86872cb5 | 1592 | err = fib6_del(rt, info); |
d8d1f30b | 1593 | dst_release(&rt->dst); |
1da177e4 | 1594 | |
c71099ac | 1595 | write_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
1596 | |
1597 | return err; | |
1598 | } | |
1599 | ||
e0a1ad73 TG |
1600 | int ip6_del_rt(struct rt6_info *rt) |
1601 | { | |
4d1169c1 | 1602 | struct nl_info info = { |
d1918542 | 1603 | .nl_net = dev_net(rt->dst.dev), |
4d1169c1 | 1604 | }; |
528c4ceb | 1605 | return __ip6_del_rt(rt, &info); |
e0a1ad73 TG |
1606 | } |
1607 | ||
86872cb5 | 1608 | static int ip6_route_del(struct fib6_config *cfg) |
1da177e4 | 1609 | { |
c71099ac | 1610 | struct fib6_table *table; |
1da177e4 LT |
1611 | struct fib6_node *fn; |
1612 | struct rt6_info *rt; | |
1613 | int err = -ESRCH; | |
1614 | ||
5578689a | 1615 | table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table); |
38308473 | 1616 | if (!table) |
c71099ac TG |
1617 | return err; |
1618 | ||
1619 | read_lock_bh(&table->tb6_lock); | |
1da177e4 | 1620 | |
c71099ac | 1621 | fn = fib6_locate(&table->tb6_root, |
86872cb5 TG |
1622 | &cfg->fc_dst, cfg->fc_dst_len, |
1623 | &cfg->fc_src, cfg->fc_src_len); | |
1ab1457c | 1624 | |
1da177e4 | 1625 | if (fn) { |
d8d1f30b | 1626 | for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) { |
86872cb5 | 1627 | if (cfg->fc_ifindex && |
d1918542 DM |
1628 | (!rt->dst.dev || |
1629 | rt->dst.dev->ifindex != cfg->fc_ifindex)) | |
1da177e4 | 1630 | continue; |
86872cb5 TG |
1631 | if (cfg->fc_flags & RTF_GATEWAY && |
1632 | !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway)) | |
1da177e4 | 1633 | continue; |
86872cb5 | 1634 | if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric) |
1da177e4 | 1635 | continue; |
d8d1f30b | 1636 | dst_hold(&rt->dst); |
c71099ac | 1637 | read_unlock_bh(&table->tb6_lock); |
1da177e4 | 1638 | |
86872cb5 | 1639 | return __ip6_del_rt(rt, &cfg->fc_nlinfo); |
1da177e4 LT |
1640 | } |
1641 | } | |
c71099ac | 1642 | read_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
1643 | |
1644 | return err; | |
1645 | } | |
1646 | ||
6700c270 | 1647 | static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_buff *skb) |
a6279458 | 1648 | { |
e8599ff4 | 1649 | struct net *net = dev_net(skb->dev); |
a6279458 | 1650 | struct netevent_redirect netevent; |
e8599ff4 DM |
1651 | struct rt6_info *rt, *nrt = NULL; |
1652 | const struct in6_addr *target; | |
e8599ff4 | 1653 | struct ndisc_options ndopts; |
6e157b6a DM |
1654 | const struct in6_addr *dest; |
1655 | struct neighbour *old_neigh; | |
e8599ff4 DM |
1656 | struct inet6_dev *in6_dev; |
1657 | struct neighbour *neigh; | |
1658 | struct icmp6hdr *icmph; | |
6e157b6a DM |
1659 | int optlen, on_link; |
1660 | u8 *lladdr; | |
e8599ff4 DM |
1661 | |
1662 | optlen = skb->tail - skb->transport_header; | |
1663 | optlen -= sizeof(struct icmp6hdr) + 2 * sizeof(struct in6_addr); | |
1664 | ||
1665 | if (optlen < 0) { | |
6e157b6a | 1666 | net_dbg_ratelimited("rt6_do_redirect: packet too short\n"); |
e8599ff4 DM |
1667 | return; |
1668 | } | |
1669 | ||
1670 | icmph = icmp6_hdr(skb); | |
1671 | target = (const struct in6_addr *) (icmph + 1); | |
1672 | dest = target + 1; | |
1673 | ||
1674 | if (ipv6_addr_is_multicast(dest)) { | |
6e157b6a | 1675 | net_dbg_ratelimited("rt6_do_redirect: destination address is multicast\n"); |
e8599ff4 DM |
1676 | return; |
1677 | } | |
1678 | ||
6e157b6a | 1679 | on_link = 0; |
e8599ff4 DM |
1680 | if (ipv6_addr_equal(dest, target)) { |
1681 | on_link = 1; | |
1682 | } else if (ipv6_addr_type(target) != | |
1683 | (IPV6_ADDR_UNICAST|IPV6_ADDR_LINKLOCAL)) { | |
6e157b6a | 1684 | net_dbg_ratelimited("rt6_do_redirect: target address is not link-local unicast\n"); |
e8599ff4 DM |
1685 | return; |
1686 | } | |
1687 | ||
1688 | in6_dev = __in6_dev_get(skb->dev); | |
1689 | if (!in6_dev) | |
1690 | return; | |
1691 | if (in6_dev->cnf.forwarding || !in6_dev->cnf.accept_redirects) | |
1692 | return; | |
1693 | ||
1694 | /* RFC2461 8.1: | |
1695 | * The IP source address of the Redirect MUST be the same as the current | |
1696 | * first-hop router for the specified ICMP Destination Address. | |
1697 | */ | |
1698 | ||
1699 | if (!ndisc_parse_options((u8*)(dest + 1), optlen, &ndopts)) { | |
1700 | net_dbg_ratelimited("rt6_redirect: invalid ND options\n"); | |
1701 | return; | |
1702 | } | |
6e157b6a DM |
1703 | |
1704 | lladdr = NULL; | |
e8599ff4 DM |
1705 | if (ndopts.nd_opts_tgt_lladdr) { |
1706 | lladdr = ndisc_opt_addr_data(ndopts.nd_opts_tgt_lladdr, | |
1707 | skb->dev); | |
1708 | if (!lladdr) { | |
1709 | net_dbg_ratelimited("rt6_redirect: invalid link-layer address length\n"); | |
1710 | return; | |
1711 | } | |
1712 | } | |
1713 | ||
6e157b6a DM |
1714 | rt = (struct rt6_info *) dst; |
1715 | if (rt == net->ipv6.ip6_null_entry) { | |
1716 | net_dbg_ratelimited("rt6_redirect: source isn't a valid nexthop for redirect target\n"); | |
e8599ff4 | 1717 | return; |
6e157b6a | 1718 | } |
e8599ff4 | 1719 | |
6e157b6a DM |
1720 | /* Redirect received -> path was valid. |
1721 | * Look, redirects are sent only in response to data packets, | |
1722 | * so that this nexthop apparently is reachable. --ANK | |
1723 | */ | |
1724 | dst_confirm(&rt->dst); | |
a6279458 | 1725 | |
6e157b6a DM |
1726 | neigh = __neigh_lookup(&nd_tbl, target, skb->dev, 1); |
1727 | if (!neigh) | |
1728 | return; | |
a6279458 | 1729 | |
6e157b6a DM |
1730 | /* Duplicate redirect: silently ignore. */ |
1731 | old_neigh = rt->n; | |
1732 | if (neigh == old_neigh) | |
a6279458 | 1733 | goto out; |
1da177e4 | 1734 | |
1da177e4 LT |
1735 | /* |
1736 | * We have finally decided to accept it. | |
1737 | */ | |
1738 | ||
1ab1457c | 1739 | neigh_update(neigh, lladdr, NUD_STALE, |
1da177e4 LT |
1740 | NEIGH_UPDATE_F_WEAK_OVERRIDE| |
1741 | NEIGH_UPDATE_F_OVERRIDE| | |
1742 | (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER| | |
1743 | NEIGH_UPDATE_F_ISROUTER)) | |
1744 | ); | |
1745 | ||
21efcfa0 | 1746 | nrt = ip6_rt_copy(rt, dest); |
38308473 | 1747 | if (!nrt) |
1da177e4 LT |
1748 | goto out; |
1749 | ||
1750 | nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE; | |
1751 | if (on_link) | |
1752 | nrt->rt6i_flags &= ~RTF_GATEWAY; | |
1753 | ||
4e3fd7a0 | 1754 | nrt->rt6i_gateway = *(struct in6_addr *)neigh->primary_key; |
97cac082 | 1755 | nrt->n = neigh_clone(neigh); |
1da177e4 | 1756 | |
40e22e8f | 1757 | if (ip6_ins_rt(nrt)) |
1da177e4 LT |
1758 | goto out; |
1759 | ||
d8d1f30b | 1760 | netevent.old = &rt->dst; |
1d248b1c | 1761 | netevent.old_neigh = old_neigh; |
d8d1f30b | 1762 | netevent.new = &nrt->dst; |
1d248b1c DM |
1763 | netevent.new_neigh = neigh; |
1764 | netevent.daddr = dest; | |
8d71740c TT |
1765 | call_netevent_notifiers(NETEVENT_REDIRECT, &netevent); |
1766 | ||
38308473 | 1767 | if (rt->rt6i_flags & RTF_CACHE) { |
6e157b6a | 1768 | rt = (struct rt6_info *) dst_clone(&rt->dst); |
e0a1ad73 | 1769 | ip6_del_rt(rt); |
1da177e4 LT |
1770 | } |
1771 | ||
1772 | out: | |
e8599ff4 | 1773 | neigh_release(neigh); |
6e157b6a DM |
1774 | } |
1775 | ||
1da177e4 LT |
1776 | /* |
1777 | * Misc support functions | |
1778 | */ | |
1779 | ||
1716a961 | 1780 | static struct rt6_info *ip6_rt_copy(struct rt6_info *ort, |
21efcfa0 | 1781 | const struct in6_addr *dest) |
1da177e4 | 1782 | { |
d1918542 | 1783 | struct net *net = dev_net(ort->dst.dev); |
8b96d22d DM |
1784 | struct rt6_info *rt = ip6_dst_alloc(net, ort->dst.dev, 0, |
1785 | ort->rt6i_table); | |
1da177e4 LT |
1786 | |
1787 | if (rt) { | |
d8d1f30b CG |
1788 | rt->dst.input = ort->dst.input; |
1789 | rt->dst.output = ort->dst.output; | |
8e2ec639 | 1790 | rt->dst.flags |= DST_HOST; |
d8d1f30b | 1791 | |
4e3fd7a0 | 1792 | rt->rt6i_dst.addr = *dest; |
8e2ec639 | 1793 | rt->rt6i_dst.plen = 128; |
defb3519 | 1794 | dst_copy_metrics(&rt->dst, &ort->dst); |
d8d1f30b | 1795 | rt->dst.error = ort->dst.error; |
1da177e4 LT |
1796 | rt->rt6i_idev = ort->rt6i_idev; |
1797 | if (rt->rt6i_idev) | |
1798 | in6_dev_hold(rt->rt6i_idev); | |
d8d1f30b | 1799 | rt->dst.lastuse = jiffies; |
1da177e4 | 1800 | |
4e3fd7a0 | 1801 | rt->rt6i_gateway = ort->rt6i_gateway; |
1716a961 G |
1802 | rt->rt6i_flags = ort->rt6i_flags; |
1803 | if ((ort->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF)) == | |
1804 | (RTF_DEFAULT | RTF_ADDRCONF)) | |
1805 | rt6_set_from(rt, ort); | |
1806 | else | |
1807 | rt6_clean_expires(rt); | |
1da177e4 LT |
1808 | rt->rt6i_metric = 0; |
1809 | ||
1da177e4 LT |
1810 | #ifdef CONFIG_IPV6_SUBTREES |
1811 | memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key)); | |
1812 | #endif | |
0f6c6392 | 1813 | memcpy(&rt->rt6i_prefsrc, &ort->rt6i_prefsrc, sizeof(struct rt6key)); |
c71099ac | 1814 | rt->rt6i_table = ort->rt6i_table; |
1da177e4 LT |
1815 | } |
1816 | return rt; | |
1817 | } | |
1818 | ||
70ceb4f5 | 1819 | #ifdef CONFIG_IPV6_ROUTE_INFO |
efa2cea0 | 1820 | static struct rt6_info *rt6_get_route_info(struct net *net, |
b71d1d42 ED |
1821 | const struct in6_addr *prefix, int prefixlen, |
1822 | const struct in6_addr *gwaddr, int ifindex) | |
70ceb4f5 YH |
1823 | { |
1824 | struct fib6_node *fn; | |
1825 | struct rt6_info *rt = NULL; | |
c71099ac TG |
1826 | struct fib6_table *table; |
1827 | ||
efa2cea0 | 1828 | table = fib6_get_table(net, RT6_TABLE_INFO); |
38308473 | 1829 | if (!table) |
c71099ac | 1830 | return NULL; |
70ceb4f5 | 1831 | |
c71099ac TG |
1832 | write_lock_bh(&table->tb6_lock); |
1833 | fn = fib6_locate(&table->tb6_root, prefix ,prefixlen, NULL, 0); | |
70ceb4f5 YH |
1834 | if (!fn) |
1835 | goto out; | |
1836 | ||
d8d1f30b | 1837 | for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) { |
d1918542 | 1838 | if (rt->dst.dev->ifindex != ifindex) |
70ceb4f5 YH |
1839 | continue; |
1840 | if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY)) | |
1841 | continue; | |
1842 | if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr)) | |
1843 | continue; | |
d8d1f30b | 1844 | dst_hold(&rt->dst); |
70ceb4f5 YH |
1845 | break; |
1846 | } | |
1847 | out: | |
c71099ac | 1848 | write_unlock_bh(&table->tb6_lock); |
70ceb4f5 YH |
1849 | return rt; |
1850 | } | |
1851 | ||
efa2cea0 | 1852 | static struct rt6_info *rt6_add_route_info(struct net *net, |
b71d1d42 ED |
1853 | const struct in6_addr *prefix, int prefixlen, |
1854 | const struct in6_addr *gwaddr, int ifindex, | |
95c96174 | 1855 | unsigned int pref) |
70ceb4f5 | 1856 | { |
86872cb5 TG |
1857 | struct fib6_config cfg = { |
1858 | .fc_table = RT6_TABLE_INFO, | |
238fc7ea | 1859 | .fc_metric = IP6_RT_PRIO_USER, |
86872cb5 TG |
1860 | .fc_ifindex = ifindex, |
1861 | .fc_dst_len = prefixlen, | |
1862 | .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO | | |
1863 | RTF_UP | RTF_PREF(pref), | |
efa2cea0 DL |
1864 | .fc_nlinfo.pid = 0, |
1865 | .fc_nlinfo.nlh = NULL, | |
1866 | .fc_nlinfo.nl_net = net, | |
86872cb5 TG |
1867 | }; |
1868 | ||
4e3fd7a0 AD |
1869 | cfg.fc_dst = *prefix; |
1870 | cfg.fc_gateway = *gwaddr; | |
70ceb4f5 | 1871 | |
e317da96 YH |
1872 | /* We should treat it as a default route if prefix length is 0. */ |
1873 | if (!prefixlen) | |
86872cb5 | 1874 | cfg.fc_flags |= RTF_DEFAULT; |
70ceb4f5 | 1875 | |
86872cb5 | 1876 | ip6_route_add(&cfg); |
70ceb4f5 | 1877 | |
efa2cea0 | 1878 | return rt6_get_route_info(net, prefix, prefixlen, gwaddr, ifindex); |
70ceb4f5 YH |
1879 | } |
1880 | #endif | |
1881 | ||
b71d1d42 | 1882 | struct rt6_info *rt6_get_dflt_router(const struct in6_addr *addr, struct net_device *dev) |
1ab1457c | 1883 | { |
1da177e4 | 1884 | struct rt6_info *rt; |
c71099ac | 1885 | struct fib6_table *table; |
1da177e4 | 1886 | |
c346dca1 | 1887 | table = fib6_get_table(dev_net(dev), RT6_TABLE_DFLT); |
38308473 | 1888 | if (!table) |
c71099ac | 1889 | return NULL; |
1da177e4 | 1890 | |
c71099ac | 1891 | write_lock_bh(&table->tb6_lock); |
d8d1f30b | 1892 | for (rt = table->tb6_root.leaf; rt; rt=rt->dst.rt6_next) { |
d1918542 | 1893 | if (dev == rt->dst.dev && |
045927ff | 1894 | ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) && |
1da177e4 LT |
1895 | ipv6_addr_equal(&rt->rt6i_gateway, addr)) |
1896 | break; | |
1897 | } | |
1898 | if (rt) | |
d8d1f30b | 1899 | dst_hold(&rt->dst); |
c71099ac | 1900 | write_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
1901 | return rt; |
1902 | } | |
1903 | ||
b71d1d42 | 1904 | struct rt6_info *rt6_add_dflt_router(const struct in6_addr *gwaddr, |
ebacaaa0 YH |
1905 | struct net_device *dev, |
1906 | unsigned int pref) | |
1da177e4 | 1907 | { |
86872cb5 TG |
1908 | struct fib6_config cfg = { |
1909 | .fc_table = RT6_TABLE_DFLT, | |
238fc7ea | 1910 | .fc_metric = IP6_RT_PRIO_USER, |
86872cb5 TG |
1911 | .fc_ifindex = dev->ifindex, |
1912 | .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT | | |
1913 | RTF_UP | RTF_EXPIRES | RTF_PREF(pref), | |
5578689a DL |
1914 | .fc_nlinfo.pid = 0, |
1915 | .fc_nlinfo.nlh = NULL, | |
c346dca1 | 1916 | .fc_nlinfo.nl_net = dev_net(dev), |
86872cb5 | 1917 | }; |
1da177e4 | 1918 | |
4e3fd7a0 | 1919 | cfg.fc_gateway = *gwaddr; |
1da177e4 | 1920 | |
86872cb5 | 1921 | ip6_route_add(&cfg); |
1da177e4 | 1922 | |
1da177e4 LT |
1923 | return rt6_get_dflt_router(gwaddr, dev); |
1924 | } | |
1925 | ||
7b4da532 | 1926 | void rt6_purge_dflt_routers(struct net *net) |
1da177e4 LT |
1927 | { |
1928 | struct rt6_info *rt; | |
c71099ac TG |
1929 | struct fib6_table *table; |
1930 | ||
1931 | /* NOTE: Keep consistent with rt6_get_dflt_router */ | |
7b4da532 | 1932 | table = fib6_get_table(net, RT6_TABLE_DFLT); |
38308473 | 1933 | if (!table) |
c71099ac | 1934 | return; |
1da177e4 LT |
1935 | |
1936 | restart: | |
c71099ac | 1937 | read_lock_bh(&table->tb6_lock); |
d8d1f30b | 1938 | for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) { |
1da177e4 | 1939 | if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF)) { |
d8d1f30b | 1940 | dst_hold(&rt->dst); |
c71099ac | 1941 | read_unlock_bh(&table->tb6_lock); |
e0a1ad73 | 1942 | ip6_del_rt(rt); |
1da177e4 LT |
1943 | goto restart; |
1944 | } | |
1945 | } | |
c71099ac | 1946 | read_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
1947 | } |
1948 | ||
5578689a DL |
1949 | static void rtmsg_to_fib6_config(struct net *net, |
1950 | struct in6_rtmsg *rtmsg, | |
86872cb5 TG |
1951 | struct fib6_config *cfg) |
1952 | { | |
1953 | memset(cfg, 0, sizeof(*cfg)); | |
1954 | ||
1955 | cfg->fc_table = RT6_TABLE_MAIN; | |
1956 | cfg->fc_ifindex = rtmsg->rtmsg_ifindex; | |
1957 | cfg->fc_metric = rtmsg->rtmsg_metric; | |
1958 | cfg->fc_expires = rtmsg->rtmsg_info; | |
1959 | cfg->fc_dst_len = rtmsg->rtmsg_dst_len; | |
1960 | cfg->fc_src_len = rtmsg->rtmsg_src_len; | |
1961 | cfg->fc_flags = rtmsg->rtmsg_flags; | |
1962 | ||
5578689a | 1963 | cfg->fc_nlinfo.nl_net = net; |
f1243c2d | 1964 | |
4e3fd7a0 AD |
1965 | cfg->fc_dst = rtmsg->rtmsg_dst; |
1966 | cfg->fc_src = rtmsg->rtmsg_src; | |
1967 | cfg->fc_gateway = rtmsg->rtmsg_gateway; | |
86872cb5 TG |
1968 | } |
1969 | ||
5578689a | 1970 | int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg) |
1da177e4 | 1971 | { |
86872cb5 | 1972 | struct fib6_config cfg; |
1da177e4 LT |
1973 | struct in6_rtmsg rtmsg; |
1974 | int err; | |
1975 | ||
1976 | switch(cmd) { | |
1977 | case SIOCADDRT: /* Add a route */ | |
1978 | case SIOCDELRT: /* Delete a route */ | |
1979 | if (!capable(CAP_NET_ADMIN)) | |
1980 | return -EPERM; | |
1981 | err = copy_from_user(&rtmsg, arg, | |
1982 | sizeof(struct in6_rtmsg)); | |
1983 | if (err) | |
1984 | return -EFAULT; | |
86872cb5 | 1985 | |
5578689a | 1986 | rtmsg_to_fib6_config(net, &rtmsg, &cfg); |
86872cb5 | 1987 | |
1da177e4 LT |
1988 | rtnl_lock(); |
1989 | switch (cmd) { | |
1990 | case SIOCADDRT: | |
86872cb5 | 1991 | err = ip6_route_add(&cfg); |
1da177e4 LT |
1992 | break; |
1993 | case SIOCDELRT: | |
86872cb5 | 1994 | err = ip6_route_del(&cfg); |
1da177e4 LT |
1995 | break; |
1996 | default: | |
1997 | err = -EINVAL; | |
1998 | } | |
1999 | rtnl_unlock(); | |
2000 | ||
2001 | return err; | |
3ff50b79 | 2002 | } |
1da177e4 LT |
2003 | |
2004 | return -EINVAL; | |
2005 | } | |
2006 | ||
2007 | /* | |
2008 | * Drop the packet on the floor | |
2009 | */ | |
2010 | ||
d5fdd6ba | 2011 | static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes) |
1da177e4 | 2012 | { |
612f09e8 | 2013 | int type; |
adf30907 | 2014 | struct dst_entry *dst = skb_dst(skb); |
612f09e8 YH |
2015 | switch (ipstats_mib_noroutes) { |
2016 | case IPSTATS_MIB_INNOROUTES: | |
0660e03f | 2017 | type = ipv6_addr_type(&ipv6_hdr(skb)->daddr); |
45bb0060 | 2018 | if (type == IPV6_ADDR_ANY) { |
3bd653c8 DL |
2019 | IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst), |
2020 | IPSTATS_MIB_INADDRERRORS); | |
612f09e8 YH |
2021 | break; |
2022 | } | |
2023 | /* FALLTHROUGH */ | |
2024 | case IPSTATS_MIB_OUTNOROUTES: | |
3bd653c8 DL |
2025 | IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst), |
2026 | ipstats_mib_noroutes); | |
612f09e8 YH |
2027 | break; |
2028 | } | |
3ffe533c | 2029 | icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0); |
1da177e4 LT |
2030 | kfree_skb(skb); |
2031 | return 0; | |
2032 | } | |
2033 | ||
9ce8ade0 TG |
2034 | static int ip6_pkt_discard(struct sk_buff *skb) |
2035 | { | |
612f09e8 | 2036 | return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES); |
9ce8ade0 TG |
2037 | } |
2038 | ||
20380731 | 2039 | static int ip6_pkt_discard_out(struct sk_buff *skb) |
1da177e4 | 2040 | { |
adf30907 | 2041 | skb->dev = skb_dst(skb)->dev; |
612f09e8 | 2042 | return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES); |
1da177e4 LT |
2043 | } |
2044 | ||
6723ab54 DM |
2045 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES |
2046 | ||
9ce8ade0 TG |
2047 | static int ip6_pkt_prohibit(struct sk_buff *skb) |
2048 | { | |
612f09e8 | 2049 | return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES); |
9ce8ade0 TG |
2050 | } |
2051 | ||
2052 | static int ip6_pkt_prohibit_out(struct sk_buff *skb) | |
2053 | { | |
adf30907 | 2054 | skb->dev = skb_dst(skb)->dev; |
612f09e8 | 2055 | return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES); |
9ce8ade0 TG |
2056 | } |
2057 | ||
6723ab54 DM |
2058 | #endif |
2059 | ||
1da177e4 LT |
2060 | /* |
2061 | * Allocate a dst for local (unicast / anycast) address. | |
2062 | */ | |
2063 | ||
2064 | struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev, | |
2065 | const struct in6_addr *addr, | |
8f031519 | 2066 | bool anycast) |
1da177e4 | 2067 | { |
c346dca1 | 2068 | struct net *net = dev_net(idev->dev); |
8b96d22d | 2069 | struct rt6_info *rt = ip6_dst_alloc(net, net->loopback_dev, 0, NULL); |
f83c7790 | 2070 | int err; |
1da177e4 | 2071 | |
38308473 | 2072 | if (!rt) { |
f3213831 | 2073 | net_warn_ratelimited("Maximum number of routes reached, consider increasing route/max_size\n"); |
1da177e4 | 2074 | return ERR_PTR(-ENOMEM); |
40385653 | 2075 | } |
1da177e4 | 2076 | |
1da177e4 LT |
2077 | in6_dev_hold(idev); |
2078 | ||
11d53b49 | 2079 | rt->dst.flags |= DST_HOST; |
d8d1f30b CG |
2080 | rt->dst.input = ip6_input; |
2081 | rt->dst.output = ip6_output; | |
1da177e4 | 2082 | rt->rt6i_idev = idev; |
d8d1f30b | 2083 | rt->dst.obsolete = -1; |
1da177e4 LT |
2084 | |
2085 | rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP; | |
58c4fb86 YH |
2086 | if (anycast) |
2087 | rt->rt6i_flags |= RTF_ANYCAST; | |
2088 | else | |
1da177e4 | 2089 | rt->rt6i_flags |= RTF_LOCAL; |
8ade06c6 | 2090 | err = rt6_bind_neighbour(rt, rt->dst.dev); |
f83c7790 | 2091 | if (err) { |
d8d1f30b | 2092 | dst_free(&rt->dst); |
f83c7790 | 2093 | return ERR_PTR(err); |
1da177e4 LT |
2094 | } |
2095 | ||
4e3fd7a0 | 2096 | rt->rt6i_dst.addr = *addr; |
1da177e4 | 2097 | rt->rt6i_dst.plen = 128; |
5578689a | 2098 | rt->rt6i_table = fib6_get_table(net, RT6_TABLE_LOCAL); |
1da177e4 | 2099 | |
d8d1f30b | 2100 | atomic_set(&rt->dst.__refcnt, 1); |
1da177e4 LT |
2101 | |
2102 | return rt; | |
2103 | } | |
2104 | ||
c3968a85 DW |
2105 | int ip6_route_get_saddr(struct net *net, |
2106 | struct rt6_info *rt, | |
b71d1d42 | 2107 | const struct in6_addr *daddr, |
c3968a85 DW |
2108 | unsigned int prefs, |
2109 | struct in6_addr *saddr) | |
2110 | { | |
2111 | struct inet6_dev *idev = ip6_dst_idev((struct dst_entry*)rt); | |
2112 | int err = 0; | |
2113 | if (rt->rt6i_prefsrc.plen) | |
4e3fd7a0 | 2114 | *saddr = rt->rt6i_prefsrc.addr; |
c3968a85 DW |
2115 | else |
2116 | err = ipv6_dev_get_saddr(net, idev ? idev->dev : NULL, | |
2117 | daddr, prefs, saddr); | |
2118 | return err; | |
2119 | } | |
2120 | ||
2121 | /* remove deleted ip from prefsrc entries */ | |
2122 | struct arg_dev_net_ip { | |
2123 | struct net_device *dev; | |
2124 | struct net *net; | |
2125 | struct in6_addr *addr; | |
2126 | }; | |
2127 | ||
2128 | static int fib6_remove_prefsrc(struct rt6_info *rt, void *arg) | |
2129 | { | |
2130 | struct net_device *dev = ((struct arg_dev_net_ip *)arg)->dev; | |
2131 | struct net *net = ((struct arg_dev_net_ip *)arg)->net; | |
2132 | struct in6_addr *addr = ((struct arg_dev_net_ip *)arg)->addr; | |
2133 | ||
d1918542 | 2134 | if (((void *)rt->dst.dev == dev || !dev) && |
c3968a85 DW |
2135 | rt != net->ipv6.ip6_null_entry && |
2136 | ipv6_addr_equal(addr, &rt->rt6i_prefsrc.addr)) { | |
2137 | /* remove prefsrc entry */ | |
2138 | rt->rt6i_prefsrc.plen = 0; | |
2139 | } | |
2140 | return 0; | |
2141 | } | |
2142 | ||
2143 | void rt6_remove_prefsrc(struct inet6_ifaddr *ifp) | |
2144 | { | |
2145 | struct net *net = dev_net(ifp->idev->dev); | |
2146 | struct arg_dev_net_ip adni = { | |
2147 | .dev = ifp->idev->dev, | |
2148 | .net = net, | |
2149 | .addr = &ifp->addr, | |
2150 | }; | |
2151 | fib6_clean_all(net, fib6_remove_prefsrc, 0, &adni); | |
2152 | } | |
2153 | ||
8ed67789 DL |
2154 | struct arg_dev_net { |
2155 | struct net_device *dev; | |
2156 | struct net *net; | |
2157 | }; | |
2158 | ||
1da177e4 LT |
2159 | static int fib6_ifdown(struct rt6_info *rt, void *arg) |
2160 | { | |
bc3ef660 | 2161 | const struct arg_dev_net *adn = arg; |
2162 | const struct net_device *dev = adn->dev; | |
8ed67789 | 2163 | |
d1918542 | 2164 | if ((rt->dst.dev == dev || !dev) && |
c159d30c | 2165 | rt != adn->net->ipv6.ip6_null_entry) |
1da177e4 | 2166 | return -1; |
c159d30c | 2167 | |
1da177e4 LT |
2168 | return 0; |
2169 | } | |
2170 | ||
f3db4851 | 2171 | void rt6_ifdown(struct net *net, struct net_device *dev) |
1da177e4 | 2172 | { |
8ed67789 DL |
2173 | struct arg_dev_net adn = { |
2174 | .dev = dev, | |
2175 | .net = net, | |
2176 | }; | |
2177 | ||
2178 | fib6_clean_all(net, fib6_ifdown, 0, &adn); | |
1e493d19 | 2179 | icmp6_clean_all(fib6_ifdown, &adn); |
1da177e4 LT |
2180 | } |
2181 | ||
95c96174 | 2182 | struct rt6_mtu_change_arg { |
1da177e4 | 2183 | struct net_device *dev; |
95c96174 | 2184 | unsigned int mtu; |
1da177e4 LT |
2185 | }; |
2186 | ||
2187 | static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg) | |
2188 | { | |
2189 | struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg; | |
2190 | struct inet6_dev *idev; | |
2191 | ||
2192 | /* In IPv6 pmtu discovery is not optional, | |
2193 | so that RTAX_MTU lock cannot disable it. | |
2194 | We still use this lock to block changes | |
2195 | caused by addrconf/ndisc. | |
2196 | */ | |
2197 | ||
2198 | idev = __in6_dev_get(arg->dev); | |
38308473 | 2199 | if (!idev) |
1da177e4 LT |
2200 | return 0; |
2201 | ||
2202 | /* For administrative MTU increase, there is no way to discover | |
2203 | IPv6 PMTU increase, so PMTU increase should be updated here. | |
2204 | Since RFC 1981 doesn't include administrative MTU increase | |
2205 | update PMTU increase is a MUST. (i.e. jumbo frame) | |
2206 | */ | |
2207 | /* | |
2208 | If new MTU is less than route PMTU, this new MTU will be the | |
2209 | lowest MTU in the path, update the route PMTU to reflect PMTU | |
2210 | decreases; if new MTU is greater than route PMTU, and the | |
2211 | old MTU is the lowest MTU in the path, update the route PMTU | |
2212 | to reflect the increase. In this case if the other nodes' MTU | |
2213 | also have the lowest MTU, TOO BIG MESSAGE will be lead to | |
2214 | PMTU discouvery. | |
2215 | */ | |
d1918542 | 2216 | if (rt->dst.dev == arg->dev && |
d8d1f30b CG |
2217 | !dst_metric_locked(&rt->dst, RTAX_MTU) && |
2218 | (dst_mtu(&rt->dst) >= arg->mtu || | |
2219 | (dst_mtu(&rt->dst) < arg->mtu && | |
2220 | dst_mtu(&rt->dst) == idev->cnf.mtu6))) { | |
defb3519 | 2221 | dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu); |
566cfd8f | 2222 | } |
1da177e4 LT |
2223 | return 0; |
2224 | } | |
2225 | ||
95c96174 | 2226 | void rt6_mtu_change(struct net_device *dev, unsigned int mtu) |
1da177e4 | 2227 | { |
c71099ac TG |
2228 | struct rt6_mtu_change_arg arg = { |
2229 | .dev = dev, | |
2230 | .mtu = mtu, | |
2231 | }; | |
1da177e4 | 2232 | |
c346dca1 | 2233 | fib6_clean_all(dev_net(dev), rt6_mtu_change_route, 0, &arg); |
1da177e4 LT |
2234 | } |
2235 | ||
ef7c79ed | 2236 | static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = { |
5176f91e | 2237 | [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) }, |
86872cb5 | 2238 | [RTA_OIF] = { .type = NLA_U32 }, |
ab364a6f | 2239 | [RTA_IIF] = { .type = NLA_U32 }, |
86872cb5 TG |
2240 | [RTA_PRIORITY] = { .type = NLA_U32 }, |
2241 | [RTA_METRICS] = { .type = NLA_NESTED }, | |
2242 | }; | |
2243 | ||
2244 | static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh, | |
2245 | struct fib6_config *cfg) | |
1da177e4 | 2246 | { |
86872cb5 TG |
2247 | struct rtmsg *rtm; |
2248 | struct nlattr *tb[RTA_MAX+1]; | |
2249 | int err; | |
1da177e4 | 2250 | |
86872cb5 TG |
2251 | err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy); |
2252 | if (err < 0) | |
2253 | goto errout; | |
1da177e4 | 2254 | |
86872cb5 TG |
2255 | err = -EINVAL; |
2256 | rtm = nlmsg_data(nlh); | |
2257 | memset(cfg, 0, sizeof(*cfg)); | |
2258 | ||
2259 | cfg->fc_table = rtm->rtm_table; | |
2260 | cfg->fc_dst_len = rtm->rtm_dst_len; | |
2261 | cfg->fc_src_len = rtm->rtm_src_len; | |
2262 | cfg->fc_flags = RTF_UP; | |
2263 | cfg->fc_protocol = rtm->rtm_protocol; | |
2264 | ||
2265 | if (rtm->rtm_type == RTN_UNREACHABLE) | |
2266 | cfg->fc_flags |= RTF_REJECT; | |
2267 | ||
ab79ad14 MŻ |
2268 | if (rtm->rtm_type == RTN_LOCAL) |
2269 | cfg->fc_flags |= RTF_LOCAL; | |
2270 | ||
86872cb5 TG |
2271 | cfg->fc_nlinfo.pid = NETLINK_CB(skb).pid; |
2272 | cfg->fc_nlinfo.nlh = nlh; | |
3b1e0a65 | 2273 | cfg->fc_nlinfo.nl_net = sock_net(skb->sk); |
86872cb5 TG |
2274 | |
2275 | if (tb[RTA_GATEWAY]) { | |
2276 | nla_memcpy(&cfg->fc_gateway, tb[RTA_GATEWAY], 16); | |
2277 | cfg->fc_flags |= RTF_GATEWAY; | |
1da177e4 | 2278 | } |
86872cb5 TG |
2279 | |
2280 | if (tb[RTA_DST]) { | |
2281 | int plen = (rtm->rtm_dst_len + 7) >> 3; | |
2282 | ||
2283 | if (nla_len(tb[RTA_DST]) < plen) | |
2284 | goto errout; | |
2285 | ||
2286 | nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen); | |
1da177e4 | 2287 | } |
86872cb5 TG |
2288 | |
2289 | if (tb[RTA_SRC]) { | |
2290 | int plen = (rtm->rtm_src_len + 7) >> 3; | |
2291 | ||
2292 | if (nla_len(tb[RTA_SRC]) < plen) | |
2293 | goto errout; | |
2294 | ||
2295 | nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen); | |
1da177e4 | 2296 | } |
86872cb5 | 2297 | |
c3968a85 DW |
2298 | if (tb[RTA_PREFSRC]) |
2299 | nla_memcpy(&cfg->fc_prefsrc, tb[RTA_PREFSRC], 16); | |
2300 | ||
86872cb5 TG |
2301 | if (tb[RTA_OIF]) |
2302 | cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]); | |
2303 | ||
2304 | if (tb[RTA_PRIORITY]) | |
2305 | cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]); | |
2306 | ||
2307 | if (tb[RTA_METRICS]) { | |
2308 | cfg->fc_mx = nla_data(tb[RTA_METRICS]); | |
2309 | cfg->fc_mx_len = nla_len(tb[RTA_METRICS]); | |
1da177e4 | 2310 | } |
86872cb5 TG |
2311 | |
2312 | if (tb[RTA_TABLE]) | |
2313 | cfg->fc_table = nla_get_u32(tb[RTA_TABLE]); | |
2314 | ||
2315 | err = 0; | |
2316 | errout: | |
2317 | return err; | |
1da177e4 LT |
2318 | } |
2319 | ||
c127ea2c | 2320 | static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg) |
1da177e4 | 2321 | { |
86872cb5 TG |
2322 | struct fib6_config cfg; |
2323 | int err; | |
1da177e4 | 2324 | |
86872cb5 TG |
2325 | err = rtm_to_fib6_config(skb, nlh, &cfg); |
2326 | if (err < 0) | |
2327 | return err; | |
2328 | ||
2329 | return ip6_route_del(&cfg); | |
1da177e4 LT |
2330 | } |
2331 | ||
c127ea2c | 2332 | static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg) |
1da177e4 | 2333 | { |
86872cb5 TG |
2334 | struct fib6_config cfg; |
2335 | int err; | |
1da177e4 | 2336 | |
86872cb5 TG |
2337 | err = rtm_to_fib6_config(skb, nlh, &cfg); |
2338 | if (err < 0) | |
2339 | return err; | |
2340 | ||
2341 | return ip6_route_add(&cfg); | |
1da177e4 LT |
2342 | } |
2343 | ||
339bf98f TG |
2344 | static inline size_t rt6_nlmsg_size(void) |
2345 | { | |
2346 | return NLMSG_ALIGN(sizeof(struct rtmsg)) | |
2347 | + nla_total_size(16) /* RTA_SRC */ | |
2348 | + nla_total_size(16) /* RTA_DST */ | |
2349 | + nla_total_size(16) /* RTA_GATEWAY */ | |
2350 | + nla_total_size(16) /* RTA_PREFSRC */ | |
2351 | + nla_total_size(4) /* RTA_TABLE */ | |
2352 | + nla_total_size(4) /* RTA_IIF */ | |
2353 | + nla_total_size(4) /* RTA_OIF */ | |
2354 | + nla_total_size(4) /* RTA_PRIORITY */ | |
6a2b9ce0 | 2355 | + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */ |
339bf98f TG |
2356 | + nla_total_size(sizeof(struct rta_cacheinfo)); |
2357 | } | |
2358 | ||
191cd582 BH |
2359 | static int rt6_fill_node(struct net *net, |
2360 | struct sk_buff *skb, struct rt6_info *rt, | |
0d51aa80 JHS |
2361 | struct in6_addr *dst, struct in6_addr *src, |
2362 | int iif, int type, u32 pid, u32 seq, | |
7bc570c8 | 2363 | int prefix, int nowait, unsigned int flags) |
1da177e4 LT |
2364 | { |
2365 | struct rtmsg *rtm; | |
2d7202bf | 2366 | struct nlmsghdr *nlh; |
e3703b3d | 2367 | long expires; |
9e762a4a | 2368 | u32 table; |
f2c31e32 | 2369 | struct neighbour *n; |
1da177e4 LT |
2370 | |
2371 | if (prefix) { /* user wants prefix routes only */ | |
2372 | if (!(rt->rt6i_flags & RTF_PREFIX_RT)) { | |
2373 | /* success since this is not a prefix route */ | |
2374 | return 1; | |
2375 | } | |
2376 | } | |
2377 | ||
2d7202bf | 2378 | nlh = nlmsg_put(skb, pid, seq, type, sizeof(*rtm), flags); |
38308473 | 2379 | if (!nlh) |
26932566 | 2380 | return -EMSGSIZE; |
2d7202bf TG |
2381 | |
2382 | rtm = nlmsg_data(nlh); | |
1da177e4 LT |
2383 | rtm->rtm_family = AF_INET6; |
2384 | rtm->rtm_dst_len = rt->rt6i_dst.plen; | |
2385 | rtm->rtm_src_len = rt->rt6i_src.plen; | |
2386 | rtm->rtm_tos = 0; | |
c71099ac | 2387 | if (rt->rt6i_table) |
9e762a4a | 2388 | table = rt->rt6i_table->tb6_id; |
c71099ac | 2389 | else |
9e762a4a PM |
2390 | table = RT6_TABLE_UNSPEC; |
2391 | rtm->rtm_table = table; | |
c78679e8 DM |
2392 | if (nla_put_u32(skb, RTA_TABLE, table)) |
2393 | goto nla_put_failure; | |
38308473 | 2394 | if (rt->rt6i_flags & RTF_REJECT) |
1da177e4 | 2395 | rtm->rtm_type = RTN_UNREACHABLE; |
38308473 | 2396 | else if (rt->rt6i_flags & RTF_LOCAL) |
ab79ad14 | 2397 | rtm->rtm_type = RTN_LOCAL; |
d1918542 | 2398 | else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK)) |
1da177e4 LT |
2399 | rtm->rtm_type = RTN_LOCAL; |
2400 | else | |
2401 | rtm->rtm_type = RTN_UNICAST; | |
2402 | rtm->rtm_flags = 0; | |
2403 | rtm->rtm_scope = RT_SCOPE_UNIVERSE; | |
2404 | rtm->rtm_protocol = rt->rt6i_protocol; | |
38308473 | 2405 | if (rt->rt6i_flags & RTF_DYNAMIC) |
1da177e4 | 2406 | rtm->rtm_protocol = RTPROT_REDIRECT; |
f0396f60 DO |
2407 | else if (rt->rt6i_flags & RTF_ADDRCONF) { |
2408 | if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ROUTEINFO)) | |
2409 | rtm->rtm_protocol = RTPROT_RA; | |
2410 | else | |
2411 | rtm->rtm_protocol = RTPROT_KERNEL; | |
2412 | } | |
1da177e4 | 2413 | |
38308473 | 2414 | if (rt->rt6i_flags & RTF_CACHE) |
1da177e4 LT |
2415 | rtm->rtm_flags |= RTM_F_CLONED; |
2416 | ||
2417 | if (dst) { | |
c78679e8 DM |
2418 | if (nla_put(skb, RTA_DST, 16, dst)) |
2419 | goto nla_put_failure; | |
1ab1457c | 2420 | rtm->rtm_dst_len = 128; |
1da177e4 | 2421 | } else if (rtm->rtm_dst_len) |
c78679e8 DM |
2422 | if (nla_put(skb, RTA_DST, 16, &rt->rt6i_dst.addr)) |
2423 | goto nla_put_failure; | |
1da177e4 LT |
2424 | #ifdef CONFIG_IPV6_SUBTREES |
2425 | if (src) { | |
c78679e8 DM |
2426 | if (nla_put(skb, RTA_SRC, 16, src)) |
2427 | goto nla_put_failure; | |
1ab1457c | 2428 | rtm->rtm_src_len = 128; |
c78679e8 DM |
2429 | } else if (rtm->rtm_src_len && |
2430 | nla_put(skb, RTA_SRC, 16, &rt->rt6i_src.addr)) | |
2431 | goto nla_put_failure; | |
1da177e4 | 2432 | #endif |
7bc570c8 YH |
2433 | if (iif) { |
2434 | #ifdef CONFIG_IPV6_MROUTE | |
2435 | if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) { | |
8229efda | 2436 | int err = ip6mr_get_route(net, skb, rtm, nowait); |
7bc570c8 YH |
2437 | if (err <= 0) { |
2438 | if (!nowait) { | |
2439 | if (err == 0) | |
2440 | return 0; | |
2441 | goto nla_put_failure; | |
2442 | } else { | |
2443 | if (err == -EMSGSIZE) | |
2444 | goto nla_put_failure; | |
2445 | } | |
2446 | } | |
2447 | } else | |
2448 | #endif | |
c78679e8 DM |
2449 | if (nla_put_u32(skb, RTA_IIF, iif)) |
2450 | goto nla_put_failure; | |
7bc570c8 | 2451 | } else if (dst) { |
1da177e4 | 2452 | struct in6_addr saddr_buf; |
c78679e8 DM |
2453 | if (ip6_route_get_saddr(net, rt, dst, 0, &saddr_buf) == 0 && |
2454 | nla_put(skb, RTA_PREFSRC, 16, &saddr_buf)) | |
2455 | goto nla_put_failure; | |
1da177e4 | 2456 | } |
2d7202bf | 2457 | |
c3968a85 DW |
2458 | if (rt->rt6i_prefsrc.plen) { |
2459 | struct in6_addr saddr_buf; | |
4e3fd7a0 | 2460 | saddr_buf = rt->rt6i_prefsrc.addr; |
c78679e8 DM |
2461 | if (nla_put(skb, RTA_PREFSRC, 16, &saddr_buf)) |
2462 | goto nla_put_failure; | |
c3968a85 DW |
2463 | } |
2464 | ||
defb3519 | 2465 | if (rtnetlink_put_metrics(skb, dst_metrics_ptr(&rt->dst)) < 0) |
2d7202bf TG |
2466 | goto nla_put_failure; |
2467 | ||
f2c31e32 | 2468 | rcu_read_lock(); |
97cac082 | 2469 | n = rt->n; |
94f826b8 ED |
2470 | if (n) { |
2471 | if (nla_put(skb, RTA_GATEWAY, 16, &n->primary_key) < 0) { | |
2472 | rcu_read_unlock(); | |
2473 | goto nla_put_failure; | |
2474 | } | |
2475 | } | |
f2c31e32 | 2476 | rcu_read_unlock(); |
2d7202bf | 2477 | |
c78679e8 DM |
2478 | if (rt->dst.dev && |
2479 | nla_put_u32(skb, RTA_OIF, rt->dst.dev->ifindex)) | |
2480 | goto nla_put_failure; | |
2481 | if (nla_put_u32(skb, RTA_PRIORITY, rt->rt6i_metric)) | |
2482 | goto nla_put_failure; | |
8253947e LW |
2483 | |
2484 | expires = (rt->rt6i_flags & RTF_EXPIRES) ? rt->dst.expires - jiffies : 0; | |
69cdf8f9 | 2485 | |
87a50699 | 2486 | if (rtnl_put_cacheinfo(skb, &rt->dst, 0, expires, rt->dst.error) < 0) |
e3703b3d | 2487 | goto nla_put_failure; |
2d7202bf TG |
2488 | |
2489 | return nlmsg_end(skb, nlh); | |
2490 | ||
2491 | nla_put_failure: | |
26932566 PM |
2492 | nlmsg_cancel(skb, nlh); |
2493 | return -EMSGSIZE; | |
1da177e4 LT |
2494 | } |
2495 | ||
1b43af54 | 2496 | int rt6_dump_route(struct rt6_info *rt, void *p_arg) |
1da177e4 LT |
2497 | { |
2498 | struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg; | |
2499 | int prefix; | |
2500 | ||
2d7202bf TG |
2501 | if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) { |
2502 | struct rtmsg *rtm = nlmsg_data(arg->cb->nlh); | |
1da177e4 LT |
2503 | prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0; |
2504 | } else | |
2505 | prefix = 0; | |
2506 | ||
191cd582 BH |
2507 | return rt6_fill_node(arg->net, |
2508 | arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE, | |
1da177e4 | 2509 | NETLINK_CB(arg->cb->skb).pid, arg->cb->nlh->nlmsg_seq, |
7bc570c8 | 2510 | prefix, 0, NLM_F_MULTI); |
1da177e4 LT |
2511 | } |
2512 | ||
c127ea2c | 2513 | static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr* nlh, void *arg) |
1da177e4 | 2514 | { |
3b1e0a65 | 2515 | struct net *net = sock_net(in_skb->sk); |
ab364a6f TG |
2516 | struct nlattr *tb[RTA_MAX+1]; |
2517 | struct rt6_info *rt; | |
1da177e4 | 2518 | struct sk_buff *skb; |
ab364a6f | 2519 | struct rtmsg *rtm; |
4c9483b2 | 2520 | struct flowi6 fl6; |
72331bc0 | 2521 | int err, iif = 0, oif = 0; |
1da177e4 | 2522 | |
ab364a6f TG |
2523 | err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy); |
2524 | if (err < 0) | |
2525 | goto errout; | |
1da177e4 | 2526 | |
ab364a6f | 2527 | err = -EINVAL; |
4c9483b2 | 2528 | memset(&fl6, 0, sizeof(fl6)); |
1da177e4 | 2529 | |
ab364a6f TG |
2530 | if (tb[RTA_SRC]) { |
2531 | if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr)) | |
2532 | goto errout; | |
2533 | ||
4e3fd7a0 | 2534 | fl6.saddr = *(struct in6_addr *)nla_data(tb[RTA_SRC]); |
ab364a6f TG |
2535 | } |
2536 | ||
2537 | if (tb[RTA_DST]) { | |
2538 | if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr)) | |
2539 | goto errout; | |
2540 | ||
4e3fd7a0 | 2541 | fl6.daddr = *(struct in6_addr *)nla_data(tb[RTA_DST]); |
ab364a6f TG |
2542 | } |
2543 | ||
2544 | if (tb[RTA_IIF]) | |
2545 | iif = nla_get_u32(tb[RTA_IIF]); | |
2546 | ||
2547 | if (tb[RTA_OIF]) | |
72331bc0 | 2548 | oif = nla_get_u32(tb[RTA_OIF]); |
1da177e4 LT |
2549 | |
2550 | if (iif) { | |
2551 | struct net_device *dev; | |
72331bc0 SL |
2552 | int flags = 0; |
2553 | ||
5578689a | 2554 | dev = __dev_get_by_index(net, iif); |
1da177e4 LT |
2555 | if (!dev) { |
2556 | err = -ENODEV; | |
ab364a6f | 2557 | goto errout; |
1da177e4 | 2558 | } |
72331bc0 SL |
2559 | |
2560 | fl6.flowi6_iif = iif; | |
2561 | ||
2562 | if (!ipv6_addr_any(&fl6.saddr)) | |
2563 | flags |= RT6_LOOKUP_F_HAS_SADDR; | |
2564 | ||
2565 | rt = (struct rt6_info *)ip6_route_input_lookup(net, dev, &fl6, | |
2566 | flags); | |
2567 | } else { | |
2568 | fl6.flowi6_oif = oif; | |
2569 | ||
2570 | rt = (struct rt6_info *)ip6_route_output(net, NULL, &fl6); | |
1da177e4 LT |
2571 | } |
2572 | ||
ab364a6f | 2573 | skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL); |
38308473 | 2574 | if (!skb) { |
2173bff5 | 2575 | dst_release(&rt->dst); |
ab364a6f TG |
2576 | err = -ENOBUFS; |
2577 | goto errout; | |
2578 | } | |
1da177e4 | 2579 | |
ab364a6f TG |
2580 | /* Reserve room for dummy headers, this skb can pass |
2581 | through good chunk of routing engine. | |
2582 | */ | |
459a98ed | 2583 | skb_reset_mac_header(skb); |
ab364a6f | 2584 | skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr)); |
1da177e4 | 2585 | |
d8d1f30b | 2586 | skb_dst_set(skb, &rt->dst); |
1da177e4 | 2587 | |
4c9483b2 | 2588 | err = rt6_fill_node(net, skb, rt, &fl6.daddr, &fl6.saddr, iif, |
1da177e4 | 2589 | RTM_NEWROUTE, NETLINK_CB(in_skb).pid, |
7bc570c8 | 2590 | nlh->nlmsg_seq, 0, 0, 0); |
1da177e4 | 2591 | if (err < 0) { |
ab364a6f TG |
2592 | kfree_skb(skb); |
2593 | goto errout; | |
1da177e4 LT |
2594 | } |
2595 | ||
5578689a | 2596 | err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).pid); |
ab364a6f | 2597 | errout: |
1da177e4 | 2598 | return err; |
1da177e4 LT |
2599 | } |
2600 | ||
86872cb5 | 2601 | void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info) |
1da177e4 LT |
2602 | { |
2603 | struct sk_buff *skb; | |
5578689a | 2604 | struct net *net = info->nl_net; |
528c4ceb DL |
2605 | u32 seq; |
2606 | int err; | |
2607 | ||
2608 | err = -ENOBUFS; | |
38308473 | 2609 | seq = info->nlh ? info->nlh->nlmsg_seq : 0; |
86872cb5 | 2610 | |
339bf98f | 2611 | skb = nlmsg_new(rt6_nlmsg_size(), gfp_any()); |
38308473 | 2612 | if (!skb) |
21713ebc TG |
2613 | goto errout; |
2614 | ||
191cd582 | 2615 | err = rt6_fill_node(net, skb, rt, NULL, NULL, 0, |
7bc570c8 | 2616 | event, info->pid, seq, 0, 0, 0); |
26932566 PM |
2617 | if (err < 0) { |
2618 | /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */ | |
2619 | WARN_ON(err == -EMSGSIZE); | |
2620 | kfree_skb(skb); | |
2621 | goto errout; | |
2622 | } | |
1ce85fe4 PNA |
2623 | rtnl_notify(skb, net, info->pid, RTNLGRP_IPV6_ROUTE, |
2624 | info->nlh, gfp_any()); | |
2625 | return; | |
21713ebc TG |
2626 | errout: |
2627 | if (err < 0) | |
5578689a | 2628 | rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err); |
1da177e4 LT |
2629 | } |
2630 | ||
8ed67789 DL |
2631 | static int ip6_route_dev_notify(struct notifier_block *this, |
2632 | unsigned long event, void *data) | |
2633 | { | |
2634 | struct net_device *dev = (struct net_device *)data; | |
c346dca1 | 2635 | struct net *net = dev_net(dev); |
8ed67789 DL |
2636 | |
2637 | if (event == NETDEV_REGISTER && (dev->flags & IFF_LOOPBACK)) { | |
d8d1f30b | 2638 | net->ipv6.ip6_null_entry->dst.dev = dev; |
8ed67789 DL |
2639 | net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev); |
2640 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES | |
d8d1f30b | 2641 | net->ipv6.ip6_prohibit_entry->dst.dev = dev; |
8ed67789 | 2642 | net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev); |
d8d1f30b | 2643 | net->ipv6.ip6_blk_hole_entry->dst.dev = dev; |
8ed67789 DL |
2644 | net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev); |
2645 | #endif | |
2646 | } | |
2647 | ||
2648 | return NOTIFY_OK; | |
2649 | } | |
2650 | ||
1da177e4 LT |
2651 | /* |
2652 | * /proc | |
2653 | */ | |
2654 | ||
2655 | #ifdef CONFIG_PROC_FS | |
2656 | ||
1da177e4 LT |
2657 | struct rt6_proc_arg |
2658 | { | |
2659 | char *buffer; | |
2660 | int offset; | |
2661 | int length; | |
2662 | int skip; | |
2663 | int len; | |
2664 | }; | |
2665 | ||
2666 | static int rt6_info_route(struct rt6_info *rt, void *p_arg) | |
2667 | { | |
33120b30 | 2668 | struct seq_file *m = p_arg; |
69cce1d1 | 2669 | struct neighbour *n; |
1da177e4 | 2670 | |
4b7a4274 | 2671 | seq_printf(m, "%pi6 %02x ", &rt->rt6i_dst.addr, rt->rt6i_dst.plen); |
1da177e4 LT |
2672 | |
2673 | #ifdef CONFIG_IPV6_SUBTREES | |
4b7a4274 | 2674 | seq_printf(m, "%pi6 %02x ", &rt->rt6i_src.addr, rt->rt6i_src.plen); |
1da177e4 | 2675 | #else |
33120b30 | 2676 | seq_puts(m, "00000000000000000000000000000000 00 "); |
1da177e4 | 2677 | #endif |
f2c31e32 | 2678 | rcu_read_lock(); |
97cac082 | 2679 | n = rt->n; |
69cce1d1 DM |
2680 | if (n) { |
2681 | seq_printf(m, "%pi6", n->primary_key); | |
1da177e4 | 2682 | } else { |
33120b30 | 2683 | seq_puts(m, "00000000000000000000000000000000"); |
1da177e4 | 2684 | } |
f2c31e32 | 2685 | rcu_read_unlock(); |
33120b30 | 2686 | seq_printf(m, " %08x %08x %08x %08x %8s\n", |
d8d1f30b CG |
2687 | rt->rt6i_metric, atomic_read(&rt->dst.__refcnt), |
2688 | rt->dst.__use, rt->rt6i_flags, | |
d1918542 | 2689 | rt->dst.dev ? rt->dst.dev->name : ""); |
1da177e4 LT |
2690 | return 0; |
2691 | } | |
2692 | ||
33120b30 | 2693 | static int ipv6_route_show(struct seq_file *m, void *v) |
1da177e4 | 2694 | { |
f3db4851 | 2695 | struct net *net = (struct net *)m->private; |
32b293a5 | 2696 | fib6_clean_all_ro(net, rt6_info_route, 0, m); |
33120b30 AD |
2697 | return 0; |
2698 | } | |
1da177e4 | 2699 | |
33120b30 AD |
2700 | static int ipv6_route_open(struct inode *inode, struct file *file) |
2701 | { | |
de05c557 | 2702 | return single_open_net(inode, file, ipv6_route_show); |
f3db4851 DL |
2703 | } |
2704 | ||
33120b30 AD |
2705 | static const struct file_operations ipv6_route_proc_fops = { |
2706 | .owner = THIS_MODULE, | |
2707 | .open = ipv6_route_open, | |
2708 | .read = seq_read, | |
2709 | .llseek = seq_lseek, | |
b6fcbdb4 | 2710 | .release = single_release_net, |
33120b30 AD |
2711 | }; |
2712 | ||
1da177e4 LT |
2713 | static int rt6_stats_seq_show(struct seq_file *seq, void *v) |
2714 | { | |
69ddb805 | 2715 | struct net *net = (struct net *)seq->private; |
1da177e4 | 2716 | seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n", |
69ddb805 DL |
2717 | net->ipv6.rt6_stats->fib_nodes, |
2718 | net->ipv6.rt6_stats->fib_route_nodes, | |
2719 | net->ipv6.rt6_stats->fib_rt_alloc, | |
2720 | net->ipv6.rt6_stats->fib_rt_entries, | |
2721 | net->ipv6.rt6_stats->fib_rt_cache, | |
fc66f95c | 2722 | dst_entries_get_slow(&net->ipv6.ip6_dst_ops), |
69ddb805 | 2723 | net->ipv6.rt6_stats->fib_discarded_routes); |
1da177e4 LT |
2724 | |
2725 | return 0; | |
2726 | } | |
2727 | ||
2728 | static int rt6_stats_seq_open(struct inode *inode, struct file *file) | |
2729 | { | |
de05c557 | 2730 | return single_open_net(inode, file, rt6_stats_seq_show); |
69ddb805 DL |
2731 | } |
2732 | ||
9a32144e | 2733 | static const struct file_operations rt6_stats_seq_fops = { |
1da177e4 LT |
2734 | .owner = THIS_MODULE, |
2735 | .open = rt6_stats_seq_open, | |
2736 | .read = seq_read, | |
2737 | .llseek = seq_lseek, | |
b6fcbdb4 | 2738 | .release = single_release_net, |
1da177e4 LT |
2739 | }; |
2740 | #endif /* CONFIG_PROC_FS */ | |
2741 | ||
2742 | #ifdef CONFIG_SYSCTL | |
2743 | ||
1da177e4 | 2744 | static |
8d65af78 | 2745 | int ipv6_sysctl_rtcache_flush(ctl_table *ctl, int write, |
1da177e4 LT |
2746 | void __user *buffer, size_t *lenp, loff_t *ppos) |
2747 | { | |
c486da34 LAG |
2748 | struct net *net; |
2749 | int delay; | |
2750 | if (!write) | |
1da177e4 | 2751 | return -EINVAL; |
c486da34 LAG |
2752 | |
2753 | net = (struct net *)ctl->extra1; | |
2754 | delay = net->ipv6.sysctl.flush_delay; | |
2755 | proc_dointvec(ctl, write, buffer, lenp, ppos); | |
2756 | fib6_run_gc(delay <= 0 ? ~0UL : (unsigned long)delay, net); | |
2757 | return 0; | |
1da177e4 LT |
2758 | } |
2759 | ||
760f2d01 | 2760 | ctl_table ipv6_route_table_template[] = { |
1ab1457c | 2761 | { |
1da177e4 | 2762 | .procname = "flush", |
4990509f | 2763 | .data = &init_net.ipv6.sysctl.flush_delay, |
1da177e4 | 2764 | .maxlen = sizeof(int), |
89c8b3a1 | 2765 | .mode = 0200, |
6d9f239a | 2766 | .proc_handler = ipv6_sysctl_rtcache_flush |
1da177e4 LT |
2767 | }, |
2768 | { | |
1da177e4 | 2769 | .procname = "gc_thresh", |
9a7ec3a9 | 2770 | .data = &ip6_dst_ops_template.gc_thresh, |
1da177e4 LT |
2771 | .maxlen = sizeof(int), |
2772 | .mode = 0644, | |
6d9f239a | 2773 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2774 | }, |
2775 | { | |
1da177e4 | 2776 | .procname = "max_size", |
4990509f | 2777 | .data = &init_net.ipv6.sysctl.ip6_rt_max_size, |
1da177e4 LT |
2778 | .maxlen = sizeof(int), |
2779 | .mode = 0644, | |
6d9f239a | 2780 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2781 | }, |
2782 | { | |
1da177e4 | 2783 | .procname = "gc_min_interval", |
4990509f | 2784 | .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval, |
1da177e4 LT |
2785 | .maxlen = sizeof(int), |
2786 | .mode = 0644, | |
6d9f239a | 2787 | .proc_handler = proc_dointvec_jiffies, |
1da177e4 LT |
2788 | }, |
2789 | { | |
1da177e4 | 2790 | .procname = "gc_timeout", |
4990509f | 2791 | .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout, |
1da177e4 LT |
2792 | .maxlen = sizeof(int), |
2793 | .mode = 0644, | |
6d9f239a | 2794 | .proc_handler = proc_dointvec_jiffies, |
1da177e4 LT |
2795 | }, |
2796 | { | |
1da177e4 | 2797 | .procname = "gc_interval", |
4990509f | 2798 | .data = &init_net.ipv6.sysctl.ip6_rt_gc_interval, |
1da177e4 LT |
2799 | .maxlen = sizeof(int), |
2800 | .mode = 0644, | |
6d9f239a | 2801 | .proc_handler = proc_dointvec_jiffies, |
1da177e4 LT |
2802 | }, |
2803 | { | |
1da177e4 | 2804 | .procname = "gc_elasticity", |
4990509f | 2805 | .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity, |
1da177e4 LT |
2806 | .maxlen = sizeof(int), |
2807 | .mode = 0644, | |
f3d3f616 | 2808 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2809 | }, |
2810 | { | |
1da177e4 | 2811 | .procname = "mtu_expires", |
4990509f | 2812 | .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires, |
1da177e4 LT |
2813 | .maxlen = sizeof(int), |
2814 | .mode = 0644, | |
6d9f239a | 2815 | .proc_handler = proc_dointvec_jiffies, |
1da177e4 LT |
2816 | }, |
2817 | { | |
1da177e4 | 2818 | .procname = "min_adv_mss", |
4990509f | 2819 | .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss, |
1da177e4 LT |
2820 | .maxlen = sizeof(int), |
2821 | .mode = 0644, | |
f3d3f616 | 2822 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2823 | }, |
2824 | { | |
1da177e4 | 2825 | .procname = "gc_min_interval_ms", |
4990509f | 2826 | .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval, |
1da177e4 LT |
2827 | .maxlen = sizeof(int), |
2828 | .mode = 0644, | |
6d9f239a | 2829 | .proc_handler = proc_dointvec_ms_jiffies, |
1da177e4 | 2830 | }, |
f8572d8f | 2831 | { } |
1da177e4 LT |
2832 | }; |
2833 | ||
2c8c1e72 | 2834 | struct ctl_table * __net_init ipv6_route_sysctl_init(struct net *net) |
760f2d01 DL |
2835 | { |
2836 | struct ctl_table *table; | |
2837 | ||
2838 | table = kmemdup(ipv6_route_table_template, | |
2839 | sizeof(ipv6_route_table_template), | |
2840 | GFP_KERNEL); | |
5ee09105 YH |
2841 | |
2842 | if (table) { | |
2843 | table[0].data = &net->ipv6.sysctl.flush_delay; | |
c486da34 | 2844 | table[0].extra1 = net; |
86393e52 | 2845 | table[1].data = &net->ipv6.ip6_dst_ops.gc_thresh; |
5ee09105 YH |
2846 | table[2].data = &net->ipv6.sysctl.ip6_rt_max_size; |
2847 | table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval; | |
2848 | table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout; | |
2849 | table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval; | |
2850 | table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity; | |
2851 | table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires; | |
2852 | table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss; | |
9c69fabe | 2853 | table[9].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval; |
5ee09105 YH |
2854 | } |
2855 | ||
760f2d01 DL |
2856 | return table; |
2857 | } | |
1da177e4 LT |
2858 | #endif |
2859 | ||
2c8c1e72 | 2860 | static int __net_init ip6_route_net_init(struct net *net) |
cdb18761 | 2861 | { |
633d424b | 2862 | int ret = -ENOMEM; |
8ed67789 | 2863 | |
86393e52 AD |
2864 | memcpy(&net->ipv6.ip6_dst_ops, &ip6_dst_ops_template, |
2865 | sizeof(net->ipv6.ip6_dst_ops)); | |
f2fc6a54 | 2866 | |
fc66f95c ED |
2867 | if (dst_entries_init(&net->ipv6.ip6_dst_ops) < 0) |
2868 | goto out_ip6_dst_ops; | |
2869 | ||
8ed67789 DL |
2870 | net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template, |
2871 | sizeof(*net->ipv6.ip6_null_entry), | |
2872 | GFP_KERNEL); | |
2873 | if (!net->ipv6.ip6_null_entry) | |
fc66f95c | 2874 | goto out_ip6_dst_entries; |
d8d1f30b | 2875 | net->ipv6.ip6_null_entry->dst.path = |
8ed67789 | 2876 | (struct dst_entry *)net->ipv6.ip6_null_entry; |
d8d1f30b | 2877 | net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops; |
62fa8a84 DM |
2878 | dst_init_metrics(&net->ipv6.ip6_null_entry->dst, |
2879 | ip6_template_metrics, true); | |
8ed67789 DL |
2880 | |
2881 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES | |
2882 | net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template, | |
2883 | sizeof(*net->ipv6.ip6_prohibit_entry), | |
2884 | GFP_KERNEL); | |
68fffc67 PZ |
2885 | if (!net->ipv6.ip6_prohibit_entry) |
2886 | goto out_ip6_null_entry; | |
d8d1f30b | 2887 | net->ipv6.ip6_prohibit_entry->dst.path = |
8ed67789 | 2888 | (struct dst_entry *)net->ipv6.ip6_prohibit_entry; |
d8d1f30b | 2889 | net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops; |
62fa8a84 DM |
2890 | dst_init_metrics(&net->ipv6.ip6_prohibit_entry->dst, |
2891 | ip6_template_metrics, true); | |
8ed67789 DL |
2892 | |
2893 | net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template, | |
2894 | sizeof(*net->ipv6.ip6_blk_hole_entry), | |
2895 | GFP_KERNEL); | |
68fffc67 PZ |
2896 | if (!net->ipv6.ip6_blk_hole_entry) |
2897 | goto out_ip6_prohibit_entry; | |
d8d1f30b | 2898 | net->ipv6.ip6_blk_hole_entry->dst.path = |
8ed67789 | 2899 | (struct dst_entry *)net->ipv6.ip6_blk_hole_entry; |
d8d1f30b | 2900 | net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops; |
62fa8a84 DM |
2901 | dst_init_metrics(&net->ipv6.ip6_blk_hole_entry->dst, |
2902 | ip6_template_metrics, true); | |
8ed67789 DL |
2903 | #endif |
2904 | ||
b339a47c PZ |
2905 | net->ipv6.sysctl.flush_delay = 0; |
2906 | net->ipv6.sysctl.ip6_rt_max_size = 4096; | |
2907 | net->ipv6.sysctl.ip6_rt_gc_min_interval = HZ / 2; | |
2908 | net->ipv6.sysctl.ip6_rt_gc_timeout = 60*HZ; | |
2909 | net->ipv6.sysctl.ip6_rt_gc_interval = 30*HZ; | |
2910 | net->ipv6.sysctl.ip6_rt_gc_elasticity = 9; | |
2911 | net->ipv6.sysctl.ip6_rt_mtu_expires = 10*60*HZ; | |
2912 | net->ipv6.sysctl.ip6_rt_min_advmss = IPV6_MIN_MTU - 20 - 40; | |
2913 | ||
6891a346 BT |
2914 | net->ipv6.ip6_rt_gc_expire = 30*HZ; |
2915 | ||
8ed67789 DL |
2916 | ret = 0; |
2917 | out: | |
2918 | return ret; | |
f2fc6a54 | 2919 | |
68fffc67 PZ |
2920 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES |
2921 | out_ip6_prohibit_entry: | |
2922 | kfree(net->ipv6.ip6_prohibit_entry); | |
2923 | out_ip6_null_entry: | |
2924 | kfree(net->ipv6.ip6_null_entry); | |
2925 | #endif | |
fc66f95c ED |
2926 | out_ip6_dst_entries: |
2927 | dst_entries_destroy(&net->ipv6.ip6_dst_ops); | |
f2fc6a54 | 2928 | out_ip6_dst_ops: |
f2fc6a54 | 2929 | goto out; |
cdb18761 DL |
2930 | } |
2931 | ||
2c8c1e72 | 2932 | static void __net_exit ip6_route_net_exit(struct net *net) |
cdb18761 | 2933 | { |
8ed67789 DL |
2934 | kfree(net->ipv6.ip6_null_entry); |
2935 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES | |
2936 | kfree(net->ipv6.ip6_prohibit_entry); | |
2937 | kfree(net->ipv6.ip6_blk_hole_entry); | |
2938 | #endif | |
41bb78b4 | 2939 | dst_entries_destroy(&net->ipv6.ip6_dst_ops); |
cdb18761 DL |
2940 | } |
2941 | ||
d189634e TG |
2942 | static int __net_init ip6_route_net_init_late(struct net *net) |
2943 | { | |
2944 | #ifdef CONFIG_PROC_FS | |
2945 | proc_net_fops_create(net, "ipv6_route", 0, &ipv6_route_proc_fops); | |
2946 | proc_net_fops_create(net, "rt6_stats", S_IRUGO, &rt6_stats_seq_fops); | |
2947 | #endif | |
2948 | return 0; | |
2949 | } | |
2950 | ||
2951 | static void __net_exit ip6_route_net_exit_late(struct net *net) | |
2952 | { | |
2953 | #ifdef CONFIG_PROC_FS | |
2954 | proc_net_remove(net, "ipv6_route"); | |
2955 | proc_net_remove(net, "rt6_stats"); | |
2956 | #endif | |
2957 | } | |
2958 | ||
cdb18761 DL |
2959 | static struct pernet_operations ip6_route_net_ops = { |
2960 | .init = ip6_route_net_init, | |
2961 | .exit = ip6_route_net_exit, | |
2962 | }; | |
2963 | ||
c3426b47 DM |
2964 | static int __net_init ipv6_inetpeer_init(struct net *net) |
2965 | { | |
2966 | struct inet_peer_base *bp = kmalloc(sizeof(*bp), GFP_KERNEL); | |
2967 | ||
2968 | if (!bp) | |
2969 | return -ENOMEM; | |
2970 | inet_peer_base_init(bp); | |
2971 | net->ipv6.peers = bp; | |
2972 | return 0; | |
2973 | } | |
2974 | ||
2975 | static void __net_exit ipv6_inetpeer_exit(struct net *net) | |
2976 | { | |
2977 | struct inet_peer_base *bp = net->ipv6.peers; | |
2978 | ||
2979 | net->ipv6.peers = NULL; | |
56a6b248 | 2980 | inetpeer_invalidate_tree(bp); |
c3426b47 DM |
2981 | kfree(bp); |
2982 | } | |
2983 | ||
2b823f72 | 2984 | static struct pernet_operations ipv6_inetpeer_ops = { |
c3426b47 DM |
2985 | .init = ipv6_inetpeer_init, |
2986 | .exit = ipv6_inetpeer_exit, | |
2987 | }; | |
2988 | ||
d189634e TG |
2989 | static struct pernet_operations ip6_route_net_late_ops = { |
2990 | .init = ip6_route_net_init_late, | |
2991 | .exit = ip6_route_net_exit_late, | |
2992 | }; | |
2993 | ||
8ed67789 DL |
2994 | static struct notifier_block ip6_route_dev_notifier = { |
2995 | .notifier_call = ip6_route_dev_notify, | |
2996 | .priority = 0, | |
2997 | }; | |
2998 | ||
433d49c3 | 2999 | int __init ip6_route_init(void) |
1da177e4 | 3000 | { |
433d49c3 DL |
3001 | int ret; |
3002 | ||
9a7ec3a9 DL |
3003 | ret = -ENOMEM; |
3004 | ip6_dst_ops_template.kmem_cachep = | |
e5d679f3 | 3005 | kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0, |
f845ab6b | 3006 | SLAB_HWCACHE_ALIGN, NULL); |
9a7ec3a9 | 3007 | if (!ip6_dst_ops_template.kmem_cachep) |
c19a28e1 | 3008 | goto out; |
14e50e57 | 3009 | |
fc66f95c | 3010 | ret = dst_entries_init(&ip6_dst_blackhole_ops); |
8ed67789 | 3011 | if (ret) |
bdb3289f | 3012 | goto out_kmem_cache; |
bdb3289f | 3013 | |
c3426b47 DM |
3014 | ret = register_pernet_subsys(&ipv6_inetpeer_ops); |
3015 | if (ret) | |
e8803b6c | 3016 | goto out_dst_entries; |
2a0c451a | 3017 | |
7e52b33b DM |
3018 | ret = register_pernet_subsys(&ip6_route_net_ops); |
3019 | if (ret) | |
3020 | goto out_register_inetpeer; | |
c3426b47 | 3021 | |
5dc121e9 AE |
3022 | ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep; |
3023 | ||
8ed67789 DL |
3024 | /* Registering of the loopback is done before this portion of code, |
3025 | * the loopback reference in rt6_info will not be taken, do it | |
3026 | * manually for init_net */ | |
d8d1f30b | 3027 | init_net.ipv6.ip6_null_entry->dst.dev = init_net.loopback_dev; |
8ed67789 DL |
3028 | init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev); |
3029 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES | |
d8d1f30b | 3030 | init_net.ipv6.ip6_prohibit_entry->dst.dev = init_net.loopback_dev; |
8ed67789 | 3031 | init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev); |
d8d1f30b | 3032 | init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev; |
8ed67789 DL |
3033 | init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev); |
3034 | #endif | |
e8803b6c | 3035 | ret = fib6_init(); |
433d49c3 | 3036 | if (ret) |
8ed67789 | 3037 | goto out_register_subsys; |
433d49c3 | 3038 | |
433d49c3 DL |
3039 | ret = xfrm6_init(); |
3040 | if (ret) | |
e8803b6c | 3041 | goto out_fib6_init; |
c35b7e72 | 3042 | |
433d49c3 DL |
3043 | ret = fib6_rules_init(); |
3044 | if (ret) | |
3045 | goto xfrm6_init; | |
7e5449c2 | 3046 | |
d189634e TG |
3047 | ret = register_pernet_subsys(&ip6_route_net_late_ops); |
3048 | if (ret) | |
3049 | goto fib6_rules_init; | |
3050 | ||
433d49c3 | 3051 | ret = -ENOBUFS; |
c7ac8679 GR |
3052 | if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL, NULL) || |
3053 | __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL, NULL) || | |
3054 | __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL, NULL)) | |
d189634e | 3055 | goto out_register_late_subsys; |
c127ea2c | 3056 | |
8ed67789 | 3057 | ret = register_netdevice_notifier(&ip6_route_dev_notifier); |
cdb18761 | 3058 | if (ret) |
d189634e | 3059 | goto out_register_late_subsys; |
8ed67789 | 3060 | |
433d49c3 DL |
3061 | out: |
3062 | return ret; | |
3063 | ||
d189634e TG |
3064 | out_register_late_subsys: |
3065 | unregister_pernet_subsys(&ip6_route_net_late_ops); | |
433d49c3 | 3066 | fib6_rules_init: |
433d49c3 DL |
3067 | fib6_rules_cleanup(); |
3068 | xfrm6_init: | |
433d49c3 | 3069 | xfrm6_fini(); |
2a0c451a TG |
3070 | out_fib6_init: |
3071 | fib6_gc_cleanup(); | |
8ed67789 DL |
3072 | out_register_subsys: |
3073 | unregister_pernet_subsys(&ip6_route_net_ops); | |
7e52b33b DM |
3074 | out_register_inetpeer: |
3075 | unregister_pernet_subsys(&ipv6_inetpeer_ops); | |
fc66f95c ED |
3076 | out_dst_entries: |
3077 | dst_entries_destroy(&ip6_dst_blackhole_ops); | |
433d49c3 | 3078 | out_kmem_cache: |
f2fc6a54 | 3079 | kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep); |
433d49c3 | 3080 | goto out; |
1da177e4 LT |
3081 | } |
3082 | ||
3083 | void ip6_route_cleanup(void) | |
3084 | { | |
8ed67789 | 3085 | unregister_netdevice_notifier(&ip6_route_dev_notifier); |
d189634e | 3086 | unregister_pernet_subsys(&ip6_route_net_late_ops); |
101367c2 | 3087 | fib6_rules_cleanup(); |
1da177e4 | 3088 | xfrm6_fini(); |
1da177e4 | 3089 | fib6_gc_cleanup(); |
c3426b47 | 3090 | unregister_pernet_subsys(&ipv6_inetpeer_ops); |
8ed67789 | 3091 | unregister_pernet_subsys(&ip6_route_net_ops); |
41bb78b4 | 3092 | dst_entries_destroy(&ip6_dst_blackhole_ops); |
f2fc6a54 | 3093 | kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep); |
1da177e4 | 3094 | } |