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); | |
81 | static void ip6_rt_update_pmtu(struct dst_entry *dst, u32 mtu); | |
82 | ||
70ceb4f5 | 83 | #ifdef CONFIG_IPV6_ROUTE_INFO |
efa2cea0 | 84 | static struct rt6_info *rt6_add_route_info(struct net *net, |
b71d1d42 ED |
85 | const struct in6_addr *prefix, int prefixlen, |
86 | const struct in6_addr *gwaddr, int ifindex, | |
95c96174 | 87 | unsigned int pref); |
efa2cea0 | 88 | static struct rt6_info *rt6_get_route_info(struct net *net, |
b71d1d42 ED |
89 | const struct in6_addr *prefix, int prefixlen, |
90 | const struct in6_addr *gwaddr, int ifindex); | |
70ceb4f5 YH |
91 | #endif |
92 | ||
06582540 DM |
93 | static u32 *ipv6_cow_metrics(struct dst_entry *dst, unsigned long old) |
94 | { | |
95 | struct rt6_info *rt = (struct rt6_info *) dst; | |
96 | struct inet_peer *peer; | |
97 | u32 *p = NULL; | |
98 | ||
8e2ec639 YZ |
99 | if (!(rt->dst.flags & DST_HOST)) |
100 | return NULL; | |
101 | ||
fbfe95a4 | 102 | peer = rt6_get_peer_create(rt); |
06582540 DM |
103 | if (peer) { |
104 | u32 *old_p = __DST_METRICS_PTR(old); | |
105 | unsigned long prev, new; | |
106 | ||
107 | p = peer->metrics; | |
108 | if (inet_metrics_new(peer)) | |
109 | memcpy(p, old_p, sizeof(u32) * RTAX_MAX); | |
110 | ||
111 | new = (unsigned long) p; | |
112 | prev = cmpxchg(&dst->_metrics, old, new); | |
113 | ||
114 | if (prev != old) { | |
115 | p = __DST_METRICS_PTR(prev); | |
116 | if (prev & DST_METRICS_READ_ONLY) | |
117 | p = NULL; | |
118 | } | |
119 | } | |
120 | return p; | |
121 | } | |
122 | ||
f894cbf8 DM |
123 | static inline const void *choose_neigh_daddr(struct rt6_info *rt, |
124 | struct sk_buff *skb, | |
125 | const void *daddr) | |
39232973 DM |
126 | { |
127 | struct in6_addr *p = &rt->rt6i_gateway; | |
128 | ||
a7563f34 | 129 | if (!ipv6_addr_any(p)) |
39232973 | 130 | return (const void *) p; |
f894cbf8 DM |
131 | else if (skb) |
132 | return &ipv6_hdr(skb)->daddr; | |
39232973 DM |
133 | return daddr; |
134 | } | |
135 | ||
f894cbf8 DM |
136 | static struct neighbour *ip6_neigh_lookup(const struct dst_entry *dst, |
137 | struct sk_buff *skb, | |
138 | const void *daddr) | |
d3aaeb38 | 139 | { |
39232973 DM |
140 | struct rt6_info *rt = (struct rt6_info *) dst; |
141 | struct neighbour *n; | |
142 | ||
f894cbf8 | 143 | daddr = choose_neigh_daddr(rt, skb, daddr); |
39232973 | 144 | n = __ipv6_neigh_lookup(&nd_tbl, dst->dev, daddr); |
f83c7790 DM |
145 | if (n) |
146 | return n; | |
147 | return neigh_create(&nd_tbl, daddr, dst->dev); | |
148 | } | |
149 | ||
8ade06c6 | 150 | static int rt6_bind_neighbour(struct rt6_info *rt, struct net_device *dev) |
f83c7790 | 151 | { |
8ade06c6 DM |
152 | struct neighbour *n = __ipv6_neigh_lookup(&nd_tbl, dev, &rt->rt6i_gateway); |
153 | if (!n) { | |
154 | n = neigh_create(&nd_tbl, &rt->rt6i_gateway, dev); | |
155 | if (IS_ERR(n)) | |
156 | return PTR_ERR(n); | |
157 | } | |
97cac082 | 158 | rt->n = n; |
f83c7790 DM |
159 | |
160 | return 0; | |
d3aaeb38 DM |
161 | } |
162 | ||
9a7ec3a9 | 163 | static struct dst_ops ip6_dst_ops_template = { |
1da177e4 | 164 | .family = AF_INET6, |
09640e63 | 165 | .protocol = cpu_to_be16(ETH_P_IPV6), |
1da177e4 LT |
166 | .gc = ip6_dst_gc, |
167 | .gc_thresh = 1024, | |
168 | .check = ip6_dst_check, | |
0dbaee3b | 169 | .default_advmss = ip6_default_advmss, |
ebb762f2 | 170 | .mtu = ip6_mtu, |
06582540 | 171 | .cow_metrics = ipv6_cow_metrics, |
1da177e4 LT |
172 | .destroy = ip6_dst_destroy, |
173 | .ifdown = ip6_dst_ifdown, | |
174 | .negative_advice = ip6_negative_advice, | |
175 | .link_failure = ip6_link_failure, | |
176 | .update_pmtu = ip6_rt_update_pmtu, | |
1ac06e03 | 177 | .local_out = __ip6_local_out, |
d3aaeb38 | 178 | .neigh_lookup = ip6_neigh_lookup, |
1da177e4 LT |
179 | }; |
180 | ||
ebb762f2 | 181 | static unsigned int ip6_blackhole_mtu(const struct dst_entry *dst) |
ec831ea7 | 182 | { |
618f9bc7 SK |
183 | unsigned int mtu = dst_metric_raw(dst, RTAX_MTU); |
184 | ||
185 | return mtu ? : dst->dev->mtu; | |
ec831ea7 RD |
186 | } |
187 | ||
14e50e57 DM |
188 | static void ip6_rt_blackhole_update_pmtu(struct dst_entry *dst, u32 mtu) |
189 | { | |
190 | } | |
191 | ||
0972ddb2 HB |
192 | static u32 *ip6_rt_blackhole_cow_metrics(struct dst_entry *dst, |
193 | unsigned long old) | |
194 | { | |
195 | return NULL; | |
196 | } | |
197 | ||
14e50e57 DM |
198 | static struct dst_ops ip6_dst_blackhole_ops = { |
199 | .family = AF_INET6, | |
09640e63 | 200 | .protocol = cpu_to_be16(ETH_P_IPV6), |
14e50e57 DM |
201 | .destroy = ip6_dst_destroy, |
202 | .check = ip6_dst_check, | |
ebb762f2 | 203 | .mtu = ip6_blackhole_mtu, |
214f45c9 | 204 | .default_advmss = ip6_default_advmss, |
14e50e57 | 205 | .update_pmtu = ip6_rt_blackhole_update_pmtu, |
0972ddb2 | 206 | .cow_metrics = ip6_rt_blackhole_cow_metrics, |
d3aaeb38 | 207 | .neigh_lookup = ip6_neigh_lookup, |
14e50e57 DM |
208 | }; |
209 | ||
62fa8a84 DM |
210 | static const u32 ip6_template_metrics[RTAX_MAX] = { |
211 | [RTAX_HOPLIMIT - 1] = 255, | |
212 | }; | |
213 | ||
bdb3289f | 214 | static struct rt6_info ip6_null_entry_template = { |
d8d1f30b CG |
215 | .dst = { |
216 | .__refcnt = ATOMIC_INIT(1), | |
217 | .__use = 1, | |
218 | .obsolete = -1, | |
219 | .error = -ENETUNREACH, | |
d8d1f30b CG |
220 | .input = ip6_pkt_discard, |
221 | .output = ip6_pkt_discard_out, | |
1da177e4 LT |
222 | }, |
223 | .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP), | |
4f724279 | 224 | .rt6i_protocol = RTPROT_KERNEL, |
1da177e4 LT |
225 | .rt6i_metric = ~(u32) 0, |
226 | .rt6i_ref = ATOMIC_INIT(1), | |
227 | }; | |
228 | ||
101367c2 TG |
229 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES |
230 | ||
6723ab54 DM |
231 | static int ip6_pkt_prohibit(struct sk_buff *skb); |
232 | static int ip6_pkt_prohibit_out(struct sk_buff *skb); | |
6723ab54 | 233 | |
280a34c8 | 234 | static struct rt6_info ip6_prohibit_entry_template = { |
d8d1f30b CG |
235 | .dst = { |
236 | .__refcnt = ATOMIC_INIT(1), | |
237 | .__use = 1, | |
238 | .obsolete = -1, | |
239 | .error = -EACCES, | |
d8d1f30b CG |
240 | .input = ip6_pkt_prohibit, |
241 | .output = ip6_pkt_prohibit_out, | |
101367c2 TG |
242 | }, |
243 | .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP), | |
4f724279 | 244 | .rt6i_protocol = RTPROT_KERNEL, |
101367c2 TG |
245 | .rt6i_metric = ~(u32) 0, |
246 | .rt6i_ref = ATOMIC_INIT(1), | |
247 | }; | |
248 | ||
bdb3289f | 249 | static struct rt6_info ip6_blk_hole_entry_template = { |
d8d1f30b CG |
250 | .dst = { |
251 | .__refcnt = ATOMIC_INIT(1), | |
252 | .__use = 1, | |
253 | .obsolete = -1, | |
254 | .error = -EINVAL, | |
d8d1f30b CG |
255 | .input = dst_discard, |
256 | .output = dst_discard, | |
101367c2 TG |
257 | }, |
258 | .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP), | |
4f724279 | 259 | .rt6i_protocol = RTPROT_KERNEL, |
101367c2 TG |
260 | .rt6i_metric = ~(u32) 0, |
261 | .rt6i_ref = ATOMIC_INIT(1), | |
262 | }; | |
263 | ||
264 | #endif | |
265 | ||
1da177e4 | 266 | /* allocate dst with ip6_dst_ops */ |
97bab73f | 267 | static inline struct rt6_info *ip6_dst_alloc(struct net *net, |
957c665f | 268 | struct net_device *dev, |
8b96d22d DM |
269 | int flags, |
270 | struct fib6_table *table) | |
1da177e4 | 271 | { |
97bab73f DM |
272 | struct rt6_info *rt = dst_alloc(&net->ipv6.ip6_dst_ops, dev, |
273 | 0, 0, flags); | |
cf911662 | 274 | |
97bab73f | 275 | if (rt) { |
fbe58186 | 276 | memset(&rt->rt6i_table, 0, |
38308473 | 277 | sizeof(*rt) - sizeof(struct dst_entry)); |
8b96d22d | 278 | rt6_init_peer(rt, table ? &table->tb6_peers : net->ipv6.peers); |
97bab73f | 279 | } |
cf911662 | 280 | return rt; |
1da177e4 LT |
281 | } |
282 | ||
283 | static void ip6_dst_destroy(struct dst_entry *dst) | |
284 | { | |
285 | struct rt6_info *rt = (struct rt6_info *)dst; | |
286 | struct inet6_dev *idev = rt->rt6i_idev; | |
287 | ||
97cac082 DM |
288 | if (rt->n) |
289 | neigh_release(rt->n); | |
290 | ||
8e2ec639 YZ |
291 | if (!(rt->dst.flags & DST_HOST)) |
292 | dst_destroy_metrics_generic(dst); | |
293 | ||
38308473 | 294 | if (idev) { |
1da177e4 LT |
295 | rt->rt6i_idev = NULL; |
296 | in6_dev_put(idev); | |
1ab1457c | 297 | } |
1716a961 G |
298 | |
299 | if (!(rt->rt6i_flags & RTF_EXPIRES) && dst->from) | |
300 | dst_release(dst->from); | |
301 | ||
97bab73f DM |
302 | if (rt6_has_peer(rt)) { |
303 | struct inet_peer *peer = rt6_peer_ptr(rt); | |
b3419363 DM |
304 | inet_putpeer(peer); |
305 | } | |
306 | } | |
307 | ||
6431cbc2 DM |
308 | static atomic_t __rt6_peer_genid = ATOMIC_INIT(0); |
309 | ||
310 | static u32 rt6_peer_genid(void) | |
311 | { | |
312 | return atomic_read(&__rt6_peer_genid); | |
313 | } | |
314 | ||
b3419363 DM |
315 | void rt6_bind_peer(struct rt6_info *rt, int create) |
316 | { | |
97bab73f | 317 | struct inet_peer_base *base; |
b3419363 DM |
318 | struct inet_peer *peer; |
319 | ||
97bab73f DM |
320 | base = inetpeer_base_ptr(rt->_rt6i_peer); |
321 | if (!base) | |
322 | return; | |
323 | ||
324 | peer = inet_getpeer_v6(base, &rt->rt6i_dst.addr, create); | |
7b34ca2a DM |
325 | if (peer) { |
326 | if (!rt6_set_peer(rt, peer)) | |
327 | inet_putpeer(peer); | |
328 | else | |
329 | rt->rt6i_peer_genid = rt6_peer_genid(); | |
330 | } | |
1da177e4 LT |
331 | } |
332 | ||
333 | static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev, | |
334 | int how) | |
335 | { | |
336 | struct rt6_info *rt = (struct rt6_info *)dst; | |
337 | struct inet6_dev *idev = rt->rt6i_idev; | |
5a3e55d6 | 338 | struct net_device *loopback_dev = |
c346dca1 | 339 | dev_net(dev)->loopback_dev; |
1da177e4 | 340 | |
97cac082 DM |
341 | if (dev != loopback_dev) { |
342 | if (idev && idev->dev == dev) { | |
343 | struct inet6_dev *loopback_idev = | |
344 | in6_dev_get(loopback_dev); | |
345 | if (loopback_idev) { | |
346 | rt->rt6i_idev = loopback_idev; | |
347 | in6_dev_put(idev); | |
348 | } | |
349 | } | |
350 | if (rt->n && rt->n->dev == dev) { | |
351 | rt->n->dev = loopback_dev; | |
352 | dev_hold(loopback_dev); | |
353 | dev_put(dev); | |
1da177e4 LT |
354 | } |
355 | } | |
356 | } | |
357 | ||
a50feda5 | 358 | static bool rt6_check_expired(const struct rt6_info *rt) |
1da177e4 | 359 | { |
1716a961 G |
360 | struct rt6_info *ort = NULL; |
361 | ||
362 | if (rt->rt6i_flags & RTF_EXPIRES) { | |
363 | if (time_after(jiffies, rt->dst.expires)) | |
a50feda5 | 364 | return true; |
1716a961 G |
365 | } else if (rt->dst.from) { |
366 | ort = (struct rt6_info *) rt->dst.from; | |
367 | return (ort->rt6i_flags & RTF_EXPIRES) && | |
368 | time_after(jiffies, ort->dst.expires); | |
369 | } | |
a50feda5 | 370 | return false; |
1da177e4 LT |
371 | } |
372 | ||
a50feda5 | 373 | static bool rt6_need_strict(const struct in6_addr *daddr) |
c71099ac | 374 | { |
a02cec21 ED |
375 | return ipv6_addr_type(daddr) & |
376 | (IPV6_ADDR_MULTICAST | IPV6_ADDR_LINKLOCAL | IPV6_ADDR_LOOPBACK); | |
c71099ac TG |
377 | } |
378 | ||
1da177e4 | 379 | /* |
c71099ac | 380 | * Route lookup. Any table->tb6_lock is implied. |
1da177e4 LT |
381 | */ |
382 | ||
8ed67789 DL |
383 | static inline struct rt6_info *rt6_device_match(struct net *net, |
384 | struct rt6_info *rt, | |
b71d1d42 | 385 | const struct in6_addr *saddr, |
1da177e4 | 386 | int oif, |
d420895e | 387 | int flags) |
1da177e4 LT |
388 | { |
389 | struct rt6_info *local = NULL; | |
390 | struct rt6_info *sprt; | |
391 | ||
dd3abc4e YH |
392 | if (!oif && ipv6_addr_any(saddr)) |
393 | goto out; | |
394 | ||
d8d1f30b | 395 | for (sprt = rt; sprt; sprt = sprt->dst.rt6_next) { |
d1918542 | 396 | struct net_device *dev = sprt->dst.dev; |
dd3abc4e YH |
397 | |
398 | if (oif) { | |
1da177e4 LT |
399 | if (dev->ifindex == oif) |
400 | return sprt; | |
401 | if (dev->flags & IFF_LOOPBACK) { | |
38308473 | 402 | if (!sprt->rt6i_idev || |
1da177e4 | 403 | sprt->rt6i_idev->dev->ifindex != oif) { |
d420895e | 404 | if (flags & RT6_LOOKUP_F_IFACE && oif) |
1da177e4 | 405 | continue; |
1ab1457c | 406 | if (local && (!oif || |
1da177e4 LT |
407 | local->rt6i_idev->dev->ifindex == oif)) |
408 | continue; | |
409 | } | |
410 | local = sprt; | |
411 | } | |
dd3abc4e YH |
412 | } else { |
413 | if (ipv6_chk_addr(net, saddr, dev, | |
414 | flags & RT6_LOOKUP_F_IFACE)) | |
415 | return sprt; | |
1da177e4 | 416 | } |
dd3abc4e | 417 | } |
1da177e4 | 418 | |
dd3abc4e | 419 | if (oif) { |
1da177e4 LT |
420 | if (local) |
421 | return local; | |
422 | ||
d420895e | 423 | if (flags & RT6_LOOKUP_F_IFACE) |
8ed67789 | 424 | return net->ipv6.ip6_null_entry; |
1da177e4 | 425 | } |
dd3abc4e | 426 | out: |
1da177e4 LT |
427 | return rt; |
428 | } | |
429 | ||
27097255 YH |
430 | #ifdef CONFIG_IPV6_ROUTER_PREF |
431 | static void rt6_probe(struct rt6_info *rt) | |
432 | { | |
f2c31e32 | 433 | struct neighbour *neigh; |
27097255 YH |
434 | /* |
435 | * Okay, this does not seem to be appropriate | |
436 | * for now, however, we need to check if it | |
437 | * is really so; aka Router Reachability Probing. | |
438 | * | |
439 | * Router Reachability Probe MUST be rate-limited | |
440 | * to no more than one per minute. | |
441 | */ | |
f2c31e32 | 442 | rcu_read_lock(); |
97cac082 | 443 | neigh = rt ? rt->n : NULL; |
27097255 | 444 | if (!neigh || (neigh->nud_state & NUD_VALID)) |
f2c31e32 | 445 | goto out; |
27097255 YH |
446 | read_lock_bh(&neigh->lock); |
447 | if (!(neigh->nud_state & NUD_VALID) && | |
52e16356 | 448 | time_after(jiffies, neigh->updated + rt->rt6i_idev->cnf.rtr_probe_interval)) { |
27097255 YH |
449 | struct in6_addr mcaddr; |
450 | struct in6_addr *target; | |
451 | ||
452 | neigh->updated = jiffies; | |
453 | read_unlock_bh(&neigh->lock); | |
454 | ||
455 | target = (struct in6_addr *)&neigh->primary_key; | |
456 | addrconf_addr_solict_mult(target, &mcaddr); | |
d1918542 | 457 | ndisc_send_ns(rt->dst.dev, NULL, target, &mcaddr, NULL); |
f2c31e32 | 458 | } else { |
27097255 | 459 | read_unlock_bh(&neigh->lock); |
f2c31e32 ED |
460 | } |
461 | out: | |
462 | rcu_read_unlock(); | |
27097255 YH |
463 | } |
464 | #else | |
465 | static inline void rt6_probe(struct rt6_info *rt) | |
466 | { | |
27097255 YH |
467 | } |
468 | #endif | |
469 | ||
1da177e4 | 470 | /* |
554cfb7e | 471 | * Default Router Selection (RFC 2461 6.3.6) |
1da177e4 | 472 | */ |
b6f99a21 | 473 | static inline int rt6_check_dev(struct rt6_info *rt, int oif) |
554cfb7e | 474 | { |
d1918542 | 475 | struct net_device *dev = rt->dst.dev; |
161980f4 | 476 | if (!oif || dev->ifindex == oif) |
554cfb7e | 477 | return 2; |
161980f4 DM |
478 | if ((dev->flags & IFF_LOOPBACK) && |
479 | rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif) | |
480 | return 1; | |
481 | return 0; | |
554cfb7e | 482 | } |
1da177e4 | 483 | |
b6f99a21 | 484 | static inline int rt6_check_neigh(struct rt6_info *rt) |
1da177e4 | 485 | { |
f2c31e32 | 486 | struct neighbour *neigh; |
398bcbeb | 487 | int m; |
f2c31e32 ED |
488 | |
489 | rcu_read_lock(); | |
97cac082 | 490 | neigh = rt->n; |
4d0c5911 YH |
491 | if (rt->rt6i_flags & RTF_NONEXTHOP || |
492 | !(rt->rt6i_flags & RTF_GATEWAY)) | |
493 | m = 1; | |
494 | else if (neigh) { | |
554cfb7e YH |
495 | read_lock_bh(&neigh->lock); |
496 | if (neigh->nud_state & NUD_VALID) | |
4d0c5911 | 497 | m = 2; |
398bcbeb YH |
498 | #ifdef CONFIG_IPV6_ROUTER_PREF |
499 | else if (neigh->nud_state & NUD_FAILED) | |
500 | m = 0; | |
501 | #endif | |
502 | else | |
ea73ee23 | 503 | m = 1; |
554cfb7e | 504 | read_unlock_bh(&neigh->lock); |
398bcbeb YH |
505 | } else |
506 | m = 0; | |
f2c31e32 | 507 | rcu_read_unlock(); |
554cfb7e | 508 | return m; |
1da177e4 LT |
509 | } |
510 | ||
554cfb7e YH |
511 | static int rt6_score_route(struct rt6_info *rt, int oif, |
512 | int strict) | |
1da177e4 | 513 | { |
4d0c5911 | 514 | int m, n; |
1ab1457c | 515 | |
4d0c5911 | 516 | m = rt6_check_dev(rt, oif); |
77d16f45 | 517 | if (!m && (strict & RT6_LOOKUP_F_IFACE)) |
554cfb7e | 518 | return -1; |
ebacaaa0 YH |
519 | #ifdef CONFIG_IPV6_ROUTER_PREF |
520 | m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2; | |
521 | #endif | |
4d0c5911 | 522 | n = rt6_check_neigh(rt); |
557e92ef | 523 | if (!n && (strict & RT6_LOOKUP_F_REACHABLE)) |
554cfb7e YH |
524 | return -1; |
525 | return m; | |
526 | } | |
527 | ||
f11e6659 DM |
528 | static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict, |
529 | int *mpri, struct rt6_info *match) | |
554cfb7e | 530 | { |
f11e6659 DM |
531 | int m; |
532 | ||
533 | if (rt6_check_expired(rt)) | |
534 | goto out; | |
535 | ||
536 | m = rt6_score_route(rt, oif, strict); | |
537 | if (m < 0) | |
538 | goto out; | |
539 | ||
540 | if (m > *mpri) { | |
541 | if (strict & RT6_LOOKUP_F_REACHABLE) | |
542 | rt6_probe(match); | |
543 | *mpri = m; | |
544 | match = rt; | |
545 | } else if (strict & RT6_LOOKUP_F_REACHABLE) { | |
546 | rt6_probe(rt); | |
547 | } | |
548 | ||
549 | out: | |
550 | return match; | |
551 | } | |
552 | ||
553 | static struct rt6_info *find_rr_leaf(struct fib6_node *fn, | |
554 | struct rt6_info *rr_head, | |
555 | u32 metric, int oif, int strict) | |
556 | { | |
557 | struct rt6_info *rt, *match; | |
554cfb7e | 558 | int mpri = -1; |
1da177e4 | 559 | |
f11e6659 DM |
560 | match = NULL; |
561 | for (rt = rr_head; rt && rt->rt6i_metric == metric; | |
d8d1f30b | 562 | rt = rt->dst.rt6_next) |
f11e6659 DM |
563 | match = find_match(rt, oif, strict, &mpri, match); |
564 | for (rt = fn->leaf; rt && rt != rr_head && rt->rt6i_metric == metric; | |
d8d1f30b | 565 | rt = rt->dst.rt6_next) |
f11e6659 | 566 | match = find_match(rt, oif, strict, &mpri, match); |
1da177e4 | 567 | |
f11e6659 DM |
568 | return match; |
569 | } | |
1da177e4 | 570 | |
f11e6659 DM |
571 | static struct rt6_info *rt6_select(struct fib6_node *fn, int oif, int strict) |
572 | { | |
573 | struct rt6_info *match, *rt0; | |
8ed67789 | 574 | struct net *net; |
1da177e4 | 575 | |
f11e6659 DM |
576 | rt0 = fn->rr_ptr; |
577 | if (!rt0) | |
578 | fn->rr_ptr = rt0 = fn->leaf; | |
1da177e4 | 579 | |
f11e6659 | 580 | match = find_rr_leaf(fn, rt0, rt0->rt6i_metric, oif, strict); |
1da177e4 | 581 | |
554cfb7e | 582 | if (!match && |
f11e6659 | 583 | (strict & RT6_LOOKUP_F_REACHABLE)) { |
d8d1f30b | 584 | struct rt6_info *next = rt0->dst.rt6_next; |
f11e6659 | 585 | |
554cfb7e | 586 | /* no entries matched; do round-robin */ |
f11e6659 DM |
587 | if (!next || next->rt6i_metric != rt0->rt6i_metric) |
588 | next = fn->leaf; | |
589 | ||
590 | if (next != rt0) | |
591 | fn->rr_ptr = next; | |
1da177e4 | 592 | } |
1da177e4 | 593 | |
d1918542 | 594 | net = dev_net(rt0->dst.dev); |
a02cec21 | 595 | return match ? match : net->ipv6.ip6_null_entry; |
1da177e4 LT |
596 | } |
597 | ||
70ceb4f5 YH |
598 | #ifdef CONFIG_IPV6_ROUTE_INFO |
599 | int rt6_route_rcv(struct net_device *dev, u8 *opt, int len, | |
b71d1d42 | 600 | const struct in6_addr *gwaddr) |
70ceb4f5 | 601 | { |
c346dca1 | 602 | struct net *net = dev_net(dev); |
70ceb4f5 YH |
603 | struct route_info *rinfo = (struct route_info *) opt; |
604 | struct in6_addr prefix_buf, *prefix; | |
605 | unsigned int pref; | |
4bed72e4 | 606 | unsigned long lifetime; |
70ceb4f5 YH |
607 | struct rt6_info *rt; |
608 | ||
609 | if (len < sizeof(struct route_info)) { | |
610 | return -EINVAL; | |
611 | } | |
612 | ||
613 | /* Sanity check for prefix_len and length */ | |
614 | if (rinfo->length > 3) { | |
615 | return -EINVAL; | |
616 | } else if (rinfo->prefix_len > 128) { | |
617 | return -EINVAL; | |
618 | } else if (rinfo->prefix_len > 64) { | |
619 | if (rinfo->length < 2) { | |
620 | return -EINVAL; | |
621 | } | |
622 | } else if (rinfo->prefix_len > 0) { | |
623 | if (rinfo->length < 1) { | |
624 | return -EINVAL; | |
625 | } | |
626 | } | |
627 | ||
628 | pref = rinfo->route_pref; | |
629 | if (pref == ICMPV6_ROUTER_PREF_INVALID) | |
3933fc95 | 630 | return -EINVAL; |
70ceb4f5 | 631 | |
4bed72e4 | 632 | lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ); |
70ceb4f5 YH |
633 | |
634 | if (rinfo->length == 3) | |
635 | prefix = (struct in6_addr *)rinfo->prefix; | |
636 | else { | |
637 | /* this function is safe */ | |
638 | ipv6_addr_prefix(&prefix_buf, | |
639 | (struct in6_addr *)rinfo->prefix, | |
640 | rinfo->prefix_len); | |
641 | prefix = &prefix_buf; | |
642 | } | |
643 | ||
efa2cea0 DL |
644 | rt = rt6_get_route_info(net, prefix, rinfo->prefix_len, gwaddr, |
645 | dev->ifindex); | |
70ceb4f5 YH |
646 | |
647 | if (rt && !lifetime) { | |
e0a1ad73 | 648 | ip6_del_rt(rt); |
70ceb4f5 YH |
649 | rt = NULL; |
650 | } | |
651 | ||
652 | if (!rt && lifetime) | |
efa2cea0 | 653 | rt = rt6_add_route_info(net, prefix, rinfo->prefix_len, gwaddr, dev->ifindex, |
70ceb4f5 YH |
654 | pref); |
655 | else if (rt) | |
656 | rt->rt6i_flags = RTF_ROUTEINFO | | |
657 | (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref); | |
658 | ||
659 | if (rt) { | |
1716a961 G |
660 | if (!addrconf_finite_timeout(lifetime)) |
661 | rt6_clean_expires(rt); | |
662 | else | |
663 | rt6_set_expires(rt, jiffies + HZ * lifetime); | |
664 | ||
d8d1f30b | 665 | dst_release(&rt->dst); |
70ceb4f5 YH |
666 | } |
667 | return 0; | |
668 | } | |
669 | #endif | |
670 | ||
8ed67789 | 671 | #define BACKTRACK(__net, saddr) \ |
982f56f3 | 672 | do { \ |
8ed67789 | 673 | if (rt == __net->ipv6.ip6_null_entry) { \ |
982f56f3 | 674 | struct fib6_node *pn; \ |
e0eda7bb | 675 | while (1) { \ |
982f56f3 YH |
676 | if (fn->fn_flags & RTN_TL_ROOT) \ |
677 | goto out; \ | |
678 | pn = fn->parent; \ | |
679 | if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn) \ | |
8bce65b9 | 680 | fn = fib6_lookup(FIB6_SUBTREE(pn), NULL, saddr); \ |
982f56f3 YH |
681 | else \ |
682 | fn = pn; \ | |
683 | if (fn->fn_flags & RTN_RTINFO) \ | |
684 | goto restart; \ | |
c71099ac | 685 | } \ |
c71099ac | 686 | } \ |
38308473 | 687 | } while (0) |
c71099ac | 688 | |
8ed67789 DL |
689 | static struct rt6_info *ip6_pol_route_lookup(struct net *net, |
690 | struct fib6_table *table, | |
4c9483b2 | 691 | struct flowi6 *fl6, int flags) |
1da177e4 LT |
692 | { |
693 | struct fib6_node *fn; | |
694 | struct rt6_info *rt; | |
695 | ||
c71099ac | 696 | read_lock_bh(&table->tb6_lock); |
4c9483b2 | 697 | fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr); |
c71099ac TG |
698 | restart: |
699 | rt = fn->leaf; | |
4c9483b2 DM |
700 | rt = rt6_device_match(net, rt, &fl6->saddr, fl6->flowi6_oif, flags); |
701 | BACKTRACK(net, &fl6->saddr); | |
c71099ac | 702 | out: |
d8d1f30b | 703 | dst_use(&rt->dst, jiffies); |
c71099ac | 704 | read_unlock_bh(&table->tb6_lock); |
c71099ac TG |
705 | return rt; |
706 | ||
707 | } | |
708 | ||
ea6e574e FW |
709 | struct dst_entry * ip6_route_lookup(struct net *net, struct flowi6 *fl6, |
710 | int flags) | |
711 | { | |
712 | return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_lookup); | |
713 | } | |
714 | EXPORT_SYMBOL_GPL(ip6_route_lookup); | |
715 | ||
9acd9f3a YH |
716 | struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr, |
717 | const struct in6_addr *saddr, int oif, int strict) | |
c71099ac | 718 | { |
4c9483b2 DM |
719 | struct flowi6 fl6 = { |
720 | .flowi6_oif = oif, | |
721 | .daddr = *daddr, | |
c71099ac TG |
722 | }; |
723 | struct dst_entry *dst; | |
77d16f45 | 724 | int flags = strict ? RT6_LOOKUP_F_IFACE : 0; |
c71099ac | 725 | |
adaa70bb | 726 | if (saddr) { |
4c9483b2 | 727 | memcpy(&fl6.saddr, saddr, sizeof(*saddr)); |
adaa70bb TG |
728 | flags |= RT6_LOOKUP_F_HAS_SADDR; |
729 | } | |
730 | ||
4c9483b2 | 731 | dst = fib6_rule_lookup(net, &fl6, flags, ip6_pol_route_lookup); |
c71099ac TG |
732 | if (dst->error == 0) |
733 | return (struct rt6_info *) dst; | |
734 | ||
735 | dst_release(dst); | |
736 | ||
1da177e4 LT |
737 | return NULL; |
738 | } | |
739 | ||
7159039a YH |
740 | EXPORT_SYMBOL(rt6_lookup); |
741 | ||
c71099ac | 742 | /* ip6_ins_rt is called with FREE table->tb6_lock. |
1da177e4 LT |
743 | It takes new route entry, the addition fails by any reason the |
744 | route is freed. In any case, if caller does not hold it, it may | |
745 | be destroyed. | |
746 | */ | |
747 | ||
86872cb5 | 748 | static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info) |
1da177e4 LT |
749 | { |
750 | int err; | |
c71099ac | 751 | struct fib6_table *table; |
1da177e4 | 752 | |
c71099ac TG |
753 | table = rt->rt6i_table; |
754 | write_lock_bh(&table->tb6_lock); | |
86872cb5 | 755 | err = fib6_add(&table->tb6_root, rt, info); |
c71099ac | 756 | write_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
757 | |
758 | return err; | |
759 | } | |
760 | ||
40e22e8f TG |
761 | int ip6_ins_rt(struct rt6_info *rt) |
762 | { | |
4d1169c1 | 763 | struct nl_info info = { |
d1918542 | 764 | .nl_net = dev_net(rt->dst.dev), |
4d1169c1 | 765 | }; |
528c4ceb | 766 | return __ip6_ins_rt(rt, &info); |
40e22e8f TG |
767 | } |
768 | ||
1716a961 | 769 | static struct rt6_info *rt6_alloc_cow(struct rt6_info *ort, |
21efcfa0 | 770 | const struct in6_addr *daddr, |
b71d1d42 | 771 | const struct in6_addr *saddr) |
1da177e4 | 772 | { |
1da177e4 LT |
773 | struct rt6_info *rt; |
774 | ||
775 | /* | |
776 | * Clone the route. | |
777 | */ | |
778 | ||
21efcfa0 | 779 | rt = ip6_rt_copy(ort, daddr); |
1da177e4 LT |
780 | |
781 | if (rt) { | |
14deae41 DM |
782 | int attempts = !in_softirq(); |
783 | ||
38308473 | 784 | if (!(rt->rt6i_flags & RTF_GATEWAY)) { |
bb3c3686 | 785 | if (ort->rt6i_dst.plen != 128 && |
21efcfa0 | 786 | ipv6_addr_equal(&ort->rt6i_dst.addr, daddr)) |
58c4fb86 | 787 | rt->rt6i_flags |= RTF_ANYCAST; |
4e3fd7a0 | 788 | rt->rt6i_gateway = *daddr; |
58c4fb86 | 789 | } |
1da177e4 | 790 | |
1da177e4 | 791 | rt->rt6i_flags |= RTF_CACHE; |
1da177e4 LT |
792 | |
793 | #ifdef CONFIG_IPV6_SUBTREES | |
794 | if (rt->rt6i_src.plen && saddr) { | |
4e3fd7a0 | 795 | rt->rt6i_src.addr = *saddr; |
1da177e4 LT |
796 | rt->rt6i_src.plen = 128; |
797 | } | |
798 | #endif | |
799 | ||
14deae41 | 800 | retry: |
8ade06c6 | 801 | if (rt6_bind_neighbour(rt, rt->dst.dev)) { |
d1918542 | 802 | struct net *net = dev_net(rt->dst.dev); |
14deae41 DM |
803 | int saved_rt_min_interval = |
804 | net->ipv6.sysctl.ip6_rt_gc_min_interval; | |
805 | int saved_rt_elasticity = | |
806 | net->ipv6.sysctl.ip6_rt_gc_elasticity; | |
807 | ||
808 | if (attempts-- > 0) { | |
809 | net->ipv6.sysctl.ip6_rt_gc_elasticity = 1; | |
810 | net->ipv6.sysctl.ip6_rt_gc_min_interval = 0; | |
811 | ||
86393e52 | 812 | ip6_dst_gc(&net->ipv6.ip6_dst_ops); |
14deae41 DM |
813 | |
814 | net->ipv6.sysctl.ip6_rt_gc_elasticity = | |
815 | saved_rt_elasticity; | |
816 | net->ipv6.sysctl.ip6_rt_gc_min_interval = | |
817 | saved_rt_min_interval; | |
818 | goto retry; | |
819 | } | |
820 | ||
f3213831 | 821 | net_warn_ratelimited("Neighbour table overflow\n"); |
d8d1f30b | 822 | dst_free(&rt->dst); |
14deae41 DM |
823 | return NULL; |
824 | } | |
95a9a5ba | 825 | } |
1da177e4 | 826 | |
95a9a5ba YH |
827 | return rt; |
828 | } | |
1da177e4 | 829 | |
21efcfa0 ED |
830 | static struct rt6_info *rt6_alloc_clone(struct rt6_info *ort, |
831 | const struct in6_addr *daddr) | |
299d9939 | 832 | { |
21efcfa0 ED |
833 | struct rt6_info *rt = ip6_rt_copy(ort, daddr); |
834 | ||
299d9939 | 835 | if (rt) { |
299d9939 | 836 | rt->rt6i_flags |= RTF_CACHE; |
97cac082 | 837 | rt->n = neigh_clone(ort->n); |
299d9939 YH |
838 | } |
839 | return rt; | |
840 | } | |
841 | ||
8ed67789 | 842 | static struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table, int oif, |
4c9483b2 | 843 | struct flowi6 *fl6, int flags) |
1da177e4 LT |
844 | { |
845 | struct fib6_node *fn; | |
519fbd87 | 846 | struct rt6_info *rt, *nrt; |
c71099ac | 847 | int strict = 0; |
1da177e4 | 848 | int attempts = 3; |
519fbd87 | 849 | int err; |
53b7997f | 850 | int reachable = net->ipv6.devconf_all->forwarding ? 0 : RT6_LOOKUP_F_REACHABLE; |
1da177e4 | 851 | |
77d16f45 | 852 | strict |= flags & RT6_LOOKUP_F_IFACE; |
1da177e4 LT |
853 | |
854 | relookup: | |
c71099ac | 855 | read_lock_bh(&table->tb6_lock); |
1da177e4 | 856 | |
8238dd06 | 857 | restart_2: |
4c9483b2 | 858 | fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr); |
1da177e4 LT |
859 | |
860 | restart: | |
4acad72d | 861 | rt = rt6_select(fn, oif, strict | reachable); |
8ed67789 | 862 | |
4c9483b2 | 863 | BACKTRACK(net, &fl6->saddr); |
8ed67789 | 864 | if (rt == net->ipv6.ip6_null_entry || |
8238dd06 | 865 | rt->rt6i_flags & RTF_CACHE) |
1ddef044 | 866 | goto out; |
1da177e4 | 867 | |
d8d1f30b | 868 | dst_hold(&rt->dst); |
c71099ac | 869 | read_unlock_bh(&table->tb6_lock); |
fb9de91e | 870 | |
97cac082 | 871 | if (!rt->n && !(rt->rt6i_flags & RTF_NONEXTHOP)) |
4c9483b2 | 872 | nrt = rt6_alloc_cow(rt, &fl6->daddr, &fl6->saddr); |
7343ff31 | 873 | else if (!(rt->dst.flags & DST_HOST)) |
4c9483b2 | 874 | nrt = rt6_alloc_clone(rt, &fl6->daddr); |
7343ff31 DM |
875 | else |
876 | goto out2; | |
e40cf353 | 877 | |
d8d1f30b | 878 | dst_release(&rt->dst); |
8ed67789 | 879 | rt = nrt ? : net->ipv6.ip6_null_entry; |
1da177e4 | 880 | |
d8d1f30b | 881 | dst_hold(&rt->dst); |
519fbd87 | 882 | if (nrt) { |
40e22e8f | 883 | err = ip6_ins_rt(nrt); |
519fbd87 | 884 | if (!err) |
1da177e4 | 885 | goto out2; |
1da177e4 | 886 | } |
1da177e4 | 887 | |
519fbd87 YH |
888 | if (--attempts <= 0) |
889 | goto out2; | |
890 | ||
891 | /* | |
c71099ac | 892 | * Race condition! In the gap, when table->tb6_lock was |
519fbd87 YH |
893 | * released someone could insert this route. Relookup. |
894 | */ | |
d8d1f30b | 895 | dst_release(&rt->dst); |
519fbd87 YH |
896 | goto relookup; |
897 | ||
898 | out: | |
8238dd06 YH |
899 | if (reachable) { |
900 | reachable = 0; | |
901 | goto restart_2; | |
902 | } | |
d8d1f30b | 903 | dst_hold(&rt->dst); |
c71099ac | 904 | read_unlock_bh(&table->tb6_lock); |
1da177e4 | 905 | out2: |
d8d1f30b CG |
906 | rt->dst.lastuse = jiffies; |
907 | rt->dst.__use++; | |
c71099ac TG |
908 | |
909 | return rt; | |
1da177e4 LT |
910 | } |
911 | ||
8ed67789 | 912 | static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table, |
4c9483b2 | 913 | struct flowi6 *fl6, int flags) |
4acad72d | 914 | { |
4c9483b2 | 915 | return ip6_pol_route(net, table, fl6->flowi6_iif, fl6, flags); |
4acad72d PE |
916 | } |
917 | ||
72331bc0 SL |
918 | static struct dst_entry *ip6_route_input_lookup(struct net *net, |
919 | struct net_device *dev, | |
920 | struct flowi6 *fl6, int flags) | |
921 | { | |
922 | if (rt6_need_strict(&fl6->daddr) && dev->type != ARPHRD_PIMREG) | |
923 | flags |= RT6_LOOKUP_F_IFACE; | |
924 | ||
925 | return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_input); | |
926 | } | |
927 | ||
c71099ac TG |
928 | void ip6_route_input(struct sk_buff *skb) |
929 | { | |
b71d1d42 | 930 | const struct ipv6hdr *iph = ipv6_hdr(skb); |
c346dca1 | 931 | struct net *net = dev_net(skb->dev); |
adaa70bb | 932 | int flags = RT6_LOOKUP_F_HAS_SADDR; |
4c9483b2 DM |
933 | struct flowi6 fl6 = { |
934 | .flowi6_iif = skb->dev->ifindex, | |
935 | .daddr = iph->daddr, | |
936 | .saddr = iph->saddr, | |
38308473 | 937 | .flowlabel = (* (__be32 *) iph) & IPV6_FLOWINFO_MASK, |
4c9483b2 DM |
938 | .flowi6_mark = skb->mark, |
939 | .flowi6_proto = iph->nexthdr, | |
c71099ac | 940 | }; |
adaa70bb | 941 | |
72331bc0 | 942 | skb_dst_set(skb, ip6_route_input_lookup(net, skb->dev, &fl6, flags)); |
c71099ac TG |
943 | } |
944 | ||
8ed67789 | 945 | static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table, |
4c9483b2 | 946 | struct flowi6 *fl6, int flags) |
1da177e4 | 947 | { |
4c9483b2 | 948 | return ip6_pol_route(net, table, fl6->flowi6_oif, fl6, flags); |
c71099ac TG |
949 | } |
950 | ||
9c7a4f9c | 951 | struct dst_entry * ip6_route_output(struct net *net, const struct sock *sk, |
4c9483b2 | 952 | struct flowi6 *fl6) |
c71099ac TG |
953 | { |
954 | int flags = 0; | |
955 | ||
4dc27d1c DM |
956 | fl6->flowi6_iif = net->loopback_dev->ifindex; |
957 | ||
4c9483b2 | 958 | if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl6->daddr)) |
77d16f45 | 959 | flags |= RT6_LOOKUP_F_IFACE; |
c71099ac | 960 | |
4c9483b2 | 961 | if (!ipv6_addr_any(&fl6->saddr)) |
adaa70bb | 962 | flags |= RT6_LOOKUP_F_HAS_SADDR; |
0c9a2ac1 YH |
963 | else if (sk) |
964 | flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs); | |
adaa70bb | 965 | |
4c9483b2 | 966 | return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_output); |
1da177e4 LT |
967 | } |
968 | ||
7159039a | 969 | EXPORT_SYMBOL(ip6_route_output); |
1da177e4 | 970 | |
2774c131 | 971 | struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_orig) |
14e50e57 | 972 | { |
5c1e6aa3 | 973 | struct rt6_info *rt, *ort = (struct rt6_info *) dst_orig; |
14e50e57 DM |
974 | struct dst_entry *new = NULL; |
975 | ||
5c1e6aa3 | 976 | rt = dst_alloc(&ip6_dst_blackhole_ops, ort->dst.dev, 1, 0, 0); |
14e50e57 | 977 | if (rt) { |
cf911662 | 978 | memset(&rt->rt6i_table, 0, sizeof(*rt) - sizeof(struct dst_entry)); |
97bab73f | 979 | rt6_init_peer(rt, net->ipv6.peers); |
cf911662 | 980 | |
d8d1f30b | 981 | new = &rt->dst; |
14e50e57 | 982 | |
14e50e57 | 983 | new->__use = 1; |
352e512c HX |
984 | new->input = dst_discard; |
985 | new->output = dst_discard; | |
14e50e57 | 986 | |
21efcfa0 ED |
987 | if (dst_metrics_read_only(&ort->dst)) |
988 | new->_metrics = ort->dst._metrics; | |
989 | else | |
990 | dst_copy_metrics(new, &ort->dst); | |
14e50e57 DM |
991 | rt->rt6i_idev = ort->rt6i_idev; |
992 | if (rt->rt6i_idev) | |
993 | in6_dev_hold(rt->rt6i_idev); | |
14e50e57 | 994 | |
4e3fd7a0 | 995 | rt->rt6i_gateway = ort->rt6i_gateway; |
1716a961 G |
996 | rt->rt6i_flags = ort->rt6i_flags; |
997 | rt6_clean_expires(rt); | |
14e50e57 DM |
998 | rt->rt6i_metric = 0; |
999 | ||
1000 | memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key)); | |
1001 | #ifdef CONFIG_IPV6_SUBTREES | |
1002 | memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key)); | |
1003 | #endif | |
1004 | ||
1005 | dst_free(new); | |
1006 | } | |
1007 | ||
69ead7af DM |
1008 | dst_release(dst_orig); |
1009 | return new ? new : ERR_PTR(-ENOMEM); | |
14e50e57 | 1010 | } |
14e50e57 | 1011 | |
1da177e4 LT |
1012 | /* |
1013 | * Destination cache support functions | |
1014 | */ | |
1015 | ||
1016 | static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie) | |
1017 | { | |
1018 | struct rt6_info *rt; | |
1019 | ||
1020 | rt = (struct rt6_info *) dst; | |
1021 | ||
6431cbc2 DM |
1022 | if (rt->rt6i_node && (rt->rt6i_node->fn_sernum == cookie)) { |
1023 | if (rt->rt6i_peer_genid != rt6_peer_genid()) { | |
97bab73f | 1024 | if (!rt6_has_peer(rt)) |
6431cbc2 DM |
1025 | rt6_bind_peer(rt, 0); |
1026 | rt->rt6i_peer_genid = rt6_peer_genid(); | |
1027 | } | |
1da177e4 | 1028 | return dst; |
6431cbc2 | 1029 | } |
1da177e4 LT |
1030 | return NULL; |
1031 | } | |
1032 | ||
1033 | static struct dst_entry *ip6_negative_advice(struct dst_entry *dst) | |
1034 | { | |
1035 | struct rt6_info *rt = (struct rt6_info *) dst; | |
1036 | ||
1037 | if (rt) { | |
54c1a859 YH |
1038 | if (rt->rt6i_flags & RTF_CACHE) { |
1039 | if (rt6_check_expired(rt)) { | |
1040 | ip6_del_rt(rt); | |
1041 | dst = NULL; | |
1042 | } | |
1043 | } else { | |
1da177e4 | 1044 | dst_release(dst); |
54c1a859 YH |
1045 | dst = NULL; |
1046 | } | |
1da177e4 | 1047 | } |
54c1a859 | 1048 | return dst; |
1da177e4 LT |
1049 | } |
1050 | ||
1051 | static void ip6_link_failure(struct sk_buff *skb) | |
1052 | { | |
1053 | struct rt6_info *rt; | |
1054 | ||
3ffe533c | 1055 | icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0); |
1da177e4 | 1056 | |
adf30907 | 1057 | rt = (struct rt6_info *) skb_dst(skb); |
1da177e4 | 1058 | if (rt) { |
1716a961 G |
1059 | if (rt->rt6i_flags & RTF_CACHE) |
1060 | rt6_update_expires(rt, 0); | |
1061 | else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT)) | |
1da177e4 LT |
1062 | rt->rt6i_node->fn_sernum = -1; |
1063 | } | |
1064 | } | |
1065 | ||
1066 | static void ip6_rt_update_pmtu(struct dst_entry *dst, u32 mtu) | |
1067 | { | |
1068 | struct rt6_info *rt6 = (struct rt6_info*)dst; | |
1069 | ||
81aded24 | 1070 | dst_confirm(dst); |
1da177e4 | 1071 | if (mtu < dst_mtu(dst) && rt6->rt6i_dst.plen == 128) { |
81aded24 DM |
1072 | struct net *net = dev_net(dst->dev); |
1073 | ||
1da177e4 LT |
1074 | rt6->rt6i_flags |= RTF_MODIFIED; |
1075 | if (mtu < IPV6_MIN_MTU) { | |
defb3519 | 1076 | u32 features = dst_metric(dst, RTAX_FEATURES); |
1da177e4 | 1077 | mtu = IPV6_MIN_MTU; |
defb3519 DM |
1078 | features |= RTAX_FEATURE_ALLFRAG; |
1079 | dst_metric_set(dst, RTAX_FEATURES, features); | |
1da177e4 | 1080 | } |
defb3519 | 1081 | dst_metric_set(dst, RTAX_MTU, mtu); |
81aded24 | 1082 | rt6_update_expires(rt6, net->ipv6.sysctl.ip6_rt_mtu_expires); |
1da177e4 LT |
1083 | } |
1084 | } | |
1085 | ||
42ae66c8 DM |
1086 | void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu, |
1087 | int oif, u32 mark) | |
81aded24 DM |
1088 | { |
1089 | const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data; | |
1090 | struct dst_entry *dst; | |
1091 | struct flowi6 fl6; | |
1092 | ||
1093 | memset(&fl6, 0, sizeof(fl6)); | |
1094 | fl6.flowi6_oif = oif; | |
1095 | fl6.flowi6_mark = mark; | |
1096 | fl6.flowi6_flags = FLOWI_FLAG_PRECOW_METRICS; | |
1097 | fl6.daddr = iph->daddr; | |
1098 | fl6.saddr = iph->saddr; | |
1099 | fl6.flowlabel = (*(__be32 *) iph) & IPV6_FLOWINFO_MASK; | |
1100 | ||
1101 | dst = ip6_route_output(net, NULL, &fl6); | |
1102 | if (!dst->error) | |
1103 | ip6_rt_update_pmtu(dst, ntohl(mtu)); | |
1104 | dst_release(dst); | |
1105 | } | |
1106 | EXPORT_SYMBOL_GPL(ip6_update_pmtu); | |
1107 | ||
1108 | void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu) | |
1109 | { | |
1110 | ip6_update_pmtu(skb, sock_net(sk), mtu, | |
1111 | sk->sk_bound_dev_if, sk->sk_mark); | |
1112 | } | |
1113 | EXPORT_SYMBOL_GPL(ip6_sk_update_pmtu); | |
1114 | ||
0dbaee3b | 1115 | static unsigned int ip6_default_advmss(const struct dst_entry *dst) |
1da177e4 | 1116 | { |
0dbaee3b DM |
1117 | struct net_device *dev = dst->dev; |
1118 | unsigned int mtu = dst_mtu(dst); | |
1119 | struct net *net = dev_net(dev); | |
1120 | ||
1da177e4 LT |
1121 | mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr); |
1122 | ||
5578689a DL |
1123 | if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss) |
1124 | mtu = net->ipv6.sysctl.ip6_rt_min_advmss; | |
1da177e4 LT |
1125 | |
1126 | /* | |
1ab1457c YH |
1127 | * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and |
1128 | * corresponding MSS is IPV6_MAXPLEN - tcp_header_size. | |
1129 | * IPV6_MAXPLEN is also valid and means: "any MSS, | |
1da177e4 LT |
1130 | * rely only on pmtu discovery" |
1131 | */ | |
1132 | if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr)) | |
1133 | mtu = IPV6_MAXPLEN; | |
1134 | return mtu; | |
1135 | } | |
1136 | ||
ebb762f2 | 1137 | static unsigned int ip6_mtu(const struct dst_entry *dst) |
d33e4553 | 1138 | { |
d33e4553 | 1139 | struct inet6_dev *idev; |
618f9bc7 SK |
1140 | unsigned int mtu = dst_metric_raw(dst, RTAX_MTU); |
1141 | ||
1142 | if (mtu) | |
1143 | return mtu; | |
1144 | ||
1145 | mtu = IPV6_MIN_MTU; | |
d33e4553 DM |
1146 | |
1147 | rcu_read_lock(); | |
1148 | idev = __in6_dev_get(dst->dev); | |
1149 | if (idev) | |
1150 | mtu = idev->cnf.mtu6; | |
1151 | rcu_read_unlock(); | |
1152 | ||
1153 | return mtu; | |
1154 | } | |
1155 | ||
3b00944c YH |
1156 | static struct dst_entry *icmp6_dst_gc_list; |
1157 | static DEFINE_SPINLOCK(icmp6_dst_lock); | |
5d0bbeeb | 1158 | |
3b00944c | 1159 | struct dst_entry *icmp6_dst_alloc(struct net_device *dev, |
1da177e4 | 1160 | struct neighbour *neigh, |
87a11578 | 1161 | struct flowi6 *fl6) |
1da177e4 | 1162 | { |
87a11578 | 1163 | struct dst_entry *dst; |
1da177e4 LT |
1164 | struct rt6_info *rt; |
1165 | struct inet6_dev *idev = in6_dev_get(dev); | |
c346dca1 | 1166 | struct net *net = dev_net(dev); |
1da177e4 | 1167 | |
38308473 | 1168 | if (unlikely(!idev)) |
122bdf67 | 1169 | return ERR_PTR(-ENODEV); |
1da177e4 | 1170 | |
8b96d22d | 1171 | rt = ip6_dst_alloc(net, dev, 0, NULL); |
38308473 | 1172 | if (unlikely(!rt)) { |
1da177e4 | 1173 | in6_dev_put(idev); |
87a11578 | 1174 | dst = ERR_PTR(-ENOMEM); |
1da177e4 LT |
1175 | goto out; |
1176 | } | |
1177 | ||
1da177e4 LT |
1178 | if (neigh) |
1179 | neigh_hold(neigh); | |
14deae41 | 1180 | else { |
f894cbf8 | 1181 | neigh = ip6_neigh_lookup(&rt->dst, NULL, &fl6->daddr); |
b43faac6 | 1182 | if (IS_ERR(neigh)) { |
252c3d84 | 1183 | in6_dev_put(idev); |
b43faac6 DM |
1184 | dst_free(&rt->dst); |
1185 | return ERR_CAST(neigh); | |
1186 | } | |
14deae41 | 1187 | } |
1da177e4 | 1188 | |
8e2ec639 YZ |
1189 | rt->dst.flags |= DST_HOST; |
1190 | rt->dst.output = ip6_output; | |
97cac082 | 1191 | rt->n = neigh; |
d8d1f30b | 1192 | atomic_set(&rt->dst.__refcnt, 1); |
87a11578 | 1193 | rt->rt6i_dst.addr = fl6->daddr; |
8e2ec639 YZ |
1194 | rt->rt6i_dst.plen = 128; |
1195 | rt->rt6i_idev = idev; | |
7011687f | 1196 | dst_metric_set(&rt->dst, RTAX_HOPLIMIT, 255); |
1da177e4 | 1197 | |
3b00944c | 1198 | spin_lock_bh(&icmp6_dst_lock); |
d8d1f30b CG |
1199 | rt->dst.next = icmp6_dst_gc_list; |
1200 | icmp6_dst_gc_list = &rt->dst; | |
3b00944c | 1201 | spin_unlock_bh(&icmp6_dst_lock); |
1da177e4 | 1202 | |
5578689a | 1203 | fib6_force_start_gc(net); |
1da177e4 | 1204 | |
87a11578 DM |
1205 | dst = xfrm_lookup(net, &rt->dst, flowi6_to_flowi(fl6), NULL, 0); |
1206 | ||
1da177e4 | 1207 | out: |
87a11578 | 1208 | return dst; |
1da177e4 LT |
1209 | } |
1210 | ||
3d0f24a7 | 1211 | int icmp6_dst_gc(void) |
1da177e4 | 1212 | { |
e9476e95 | 1213 | struct dst_entry *dst, **pprev; |
3d0f24a7 | 1214 | int more = 0; |
1da177e4 | 1215 | |
3b00944c YH |
1216 | spin_lock_bh(&icmp6_dst_lock); |
1217 | pprev = &icmp6_dst_gc_list; | |
5d0bbeeb | 1218 | |
1da177e4 LT |
1219 | while ((dst = *pprev) != NULL) { |
1220 | if (!atomic_read(&dst->__refcnt)) { | |
1221 | *pprev = dst->next; | |
1222 | dst_free(dst); | |
1da177e4 LT |
1223 | } else { |
1224 | pprev = &dst->next; | |
3d0f24a7 | 1225 | ++more; |
1da177e4 LT |
1226 | } |
1227 | } | |
1228 | ||
3b00944c | 1229 | spin_unlock_bh(&icmp6_dst_lock); |
5d0bbeeb | 1230 | |
3d0f24a7 | 1231 | return more; |
1da177e4 LT |
1232 | } |
1233 | ||
1e493d19 DM |
1234 | static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg), |
1235 | void *arg) | |
1236 | { | |
1237 | struct dst_entry *dst, **pprev; | |
1238 | ||
1239 | spin_lock_bh(&icmp6_dst_lock); | |
1240 | pprev = &icmp6_dst_gc_list; | |
1241 | while ((dst = *pprev) != NULL) { | |
1242 | struct rt6_info *rt = (struct rt6_info *) dst; | |
1243 | if (func(rt, arg)) { | |
1244 | *pprev = dst->next; | |
1245 | dst_free(dst); | |
1246 | } else { | |
1247 | pprev = &dst->next; | |
1248 | } | |
1249 | } | |
1250 | spin_unlock_bh(&icmp6_dst_lock); | |
1251 | } | |
1252 | ||
569d3645 | 1253 | static int ip6_dst_gc(struct dst_ops *ops) |
1da177e4 | 1254 | { |
1da177e4 | 1255 | unsigned long now = jiffies; |
86393e52 | 1256 | struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops); |
7019b78e DL |
1257 | int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval; |
1258 | int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size; | |
1259 | int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity; | |
1260 | int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout; | |
1261 | unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc; | |
fc66f95c | 1262 | int entries; |
7019b78e | 1263 | |
fc66f95c | 1264 | entries = dst_entries_get_fast(ops); |
7019b78e | 1265 | if (time_after(rt_last_gc + rt_min_interval, now) && |
fc66f95c | 1266 | entries <= rt_max_size) |
1da177e4 LT |
1267 | goto out; |
1268 | ||
6891a346 BT |
1269 | net->ipv6.ip6_rt_gc_expire++; |
1270 | fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net); | |
1271 | net->ipv6.ip6_rt_last_gc = now; | |
fc66f95c ED |
1272 | entries = dst_entries_get_slow(ops); |
1273 | if (entries < ops->gc_thresh) | |
7019b78e | 1274 | net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1; |
1da177e4 | 1275 | out: |
7019b78e | 1276 | net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity; |
fc66f95c | 1277 | return entries > rt_max_size; |
1da177e4 LT |
1278 | } |
1279 | ||
1280 | /* Clean host part of a prefix. Not necessary in radix tree, | |
1281 | but results in cleaner routing tables. | |
1282 | ||
1283 | Remove it only when all the things will work! | |
1284 | */ | |
1285 | ||
6b75d090 | 1286 | int ip6_dst_hoplimit(struct dst_entry *dst) |
1da177e4 | 1287 | { |
5170ae82 | 1288 | int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT); |
a02e4b7d | 1289 | if (hoplimit == 0) { |
6b75d090 | 1290 | struct net_device *dev = dst->dev; |
c68f24cc ED |
1291 | struct inet6_dev *idev; |
1292 | ||
1293 | rcu_read_lock(); | |
1294 | idev = __in6_dev_get(dev); | |
1295 | if (idev) | |
6b75d090 | 1296 | hoplimit = idev->cnf.hop_limit; |
c68f24cc | 1297 | else |
53b7997f | 1298 | hoplimit = dev_net(dev)->ipv6.devconf_all->hop_limit; |
c68f24cc | 1299 | rcu_read_unlock(); |
1da177e4 LT |
1300 | } |
1301 | return hoplimit; | |
1302 | } | |
abbf46ae | 1303 | EXPORT_SYMBOL(ip6_dst_hoplimit); |
1da177e4 LT |
1304 | |
1305 | /* | |
1306 | * | |
1307 | */ | |
1308 | ||
86872cb5 | 1309 | int ip6_route_add(struct fib6_config *cfg) |
1da177e4 LT |
1310 | { |
1311 | int err; | |
5578689a | 1312 | struct net *net = cfg->fc_nlinfo.nl_net; |
1da177e4 LT |
1313 | struct rt6_info *rt = NULL; |
1314 | struct net_device *dev = NULL; | |
1315 | struct inet6_dev *idev = NULL; | |
c71099ac | 1316 | struct fib6_table *table; |
1da177e4 LT |
1317 | int addr_type; |
1318 | ||
86872cb5 | 1319 | if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128) |
1da177e4 LT |
1320 | return -EINVAL; |
1321 | #ifndef CONFIG_IPV6_SUBTREES | |
86872cb5 | 1322 | if (cfg->fc_src_len) |
1da177e4 LT |
1323 | return -EINVAL; |
1324 | #endif | |
86872cb5 | 1325 | if (cfg->fc_ifindex) { |
1da177e4 | 1326 | err = -ENODEV; |
5578689a | 1327 | dev = dev_get_by_index(net, cfg->fc_ifindex); |
1da177e4 LT |
1328 | if (!dev) |
1329 | goto out; | |
1330 | idev = in6_dev_get(dev); | |
1331 | if (!idev) | |
1332 | goto out; | |
1333 | } | |
1334 | ||
86872cb5 TG |
1335 | if (cfg->fc_metric == 0) |
1336 | cfg->fc_metric = IP6_RT_PRIO_USER; | |
1da177e4 | 1337 | |
d71314b4 | 1338 | err = -ENOBUFS; |
38308473 DM |
1339 | if (cfg->fc_nlinfo.nlh && |
1340 | !(cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_CREATE)) { | |
d71314b4 | 1341 | table = fib6_get_table(net, cfg->fc_table); |
38308473 | 1342 | if (!table) { |
f3213831 | 1343 | pr_warn("NLM_F_CREATE should be specified when creating new route\n"); |
d71314b4 MV |
1344 | table = fib6_new_table(net, cfg->fc_table); |
1345 | } | |
1346 | } else { | |
1347 | table = fib6_new_table(net, cfg->fc_table); | |
1348 | } | |
38308473 DM |
1349 | |
1350 | if (!table) | |
c71099ac | 1351 | goto out; |
c71099ac | 1352 | |
8b96d22d | 1353 | rt = ip6_dst_alloc(net, NULL, DST_NOCOUNT, table); |
1da177e4 | 1354 | |
38308473 | 1355 | if (!rt) { |
1da177e4 LT |
1356 | err = -ENOMEM; |
1357 | goto out; | |
1358 | } | |
1359 | ||
d8d1f30b | 1360 | rt->dst.obsolete = -1; |
1716a961 G |
1361 | |
1362 | if (cfg->fc_flags & RTF_EXPIRES) | |
1363 | rt6_set_expires(rt, jiffies + | |
1364 | clock_t_to_jiffies(cfg->fc_expires)); | |
1365 | else | |
1366 | rt6_clean_expires(rt); | |
1da177e4 | 1367 | |
86872cb5 TG |
1368 | if (cfg->fc_protocol == RTPROT_UNSPEC) |
1369 | cfg->fc_protocol = RTPROT_BOOT; | |
1370 | rt->rt6i_protocol = cfg->fc_protocol; | |
1371 | ||
1372 | addr_type = ipv6_addr_type(&cfg->fc_dst); | |
1da177e4 LT |
1373 | |
1374 | if (addr_type & IPV6_ADDR_MULTICAST) | |
d8d1f30b | 1375 | rt->dst.input = ip6_mc_input; |
ab79ad14 MŻ |
1376 | else if (cfg->fc_flags & RTF_LOCAL) |
1377 | rt->dst.input = ip6_input; | |
1da177e4 | 1378 | else |
d8d1f30b | 1379 | rt->dst.input = ip6_forward; |
1da177e4 | 1380 | |
d8d1f30b | 1381 | rt->dst.output = ip6_output; |
1da177e4 | 1382 | |
86872cb5 TG |
1383 | ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len); |
1384 | rt->rt6i_dst.plen = cfg->fc_dst_len; | |
1da177e4 | 1385 | if (rt->rt6i_dst.plen == 128) |
11d53b49 | 1386 | rt->dst.flags |= DST_HOST; |
1da177e4 | 1387 | |
8e2ec639 YZ |
1388 | if (!(rt->dst.flags & DST_HOST) && cfg->fc_mx) { |
1389 | u32 *metrics = kzalloc(sizeof(u32) * RTAX_MAX, GFP_KERNEL); | |
1390 | if (!metrics) { | |
1391 | err = -ENOMEM; | |
1392 | goto out; | |
1393 | } | |
1394 | dst_init_metrics(&rt->dst, metrics, 0); | |
1395 | } | |
1da177e4 | 1396 | #ifdef CONFIG_IPV6_SUBTREES |
86872cb5 TG |
1397 | ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len); |
1398 | rt->rt6i_src.plen = cfg->fc_src_len; | |
1da177e4 LT |
1399 | #endif |
1400 | ||
86872cb5 | 1401 | rt->rt6i_metric = cfg->fc_metric; |
1da177e4 LT |
1402 | |
1403 | /* We cannot add true routes via loopback here, | |
1404 | they would result in kernel looping; promote them to reject routes | |
1405 | */ | |
86872cb5 | 1406 | if ((cfg->fc_flags & RTF_REJECT) || |
38308473 DM |
1407 | (dev && (dev->flags & IFF_LOOPBACK) && |
1408 | !(addr_type & IPV6_ADDR_LOOPBACK) && | |
1409 | !(cfg->fc_flags & RTF_LOCAL))) { | |
1da177e4 | 1410 | /* hold loopback dev/idev if we haven't done so. */ |
5578689a | 1411 | if (dev != net->loopback_dev) { |
1da177e4 LT |
1412 | if (dev) { |
1413 | dev_put(dev); | |
1414 | in6_dev_put(idev); | |
1415 | } | |
5578689a | 1416 | dev = net->loopback_dev; |
1da177e4 LT |
1417 | dev_hold(dev); |
1418 | idev = in6_dev_get(dev); | |
1419 | if (!idev) { | |
1420 | err = -ENODEV; | |
1421 | goto out; | |
1422 | } | |
1423 | } | |
d8d1f30b CG |
1424 | rt->dst.output = ip6_pkt_discard_out; |
1425 | rt->dst.input = ip6_pkt_discard; | |
1426 | rt->dst.error = -ENETUNREACH; | |
1da177e4 LT |
1427 | rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP; |
1428 | goto install_route; | |
1429 | } | |
1430 | ||
86872cb5 | 1431 | if (cfg->fc_flags & RTF_GATEWAY) { |
b71d1d42 | 1432 | const struct in6_addr *gw_addr; |
1da177e4 LT |
1433 | int gwa_type; |
1434 | ||
86872cb5 | 1435 | gw_addr = &cfg->fc_gateway; |
4e3fd7a0 | 1436 | rt->rt6i_gateway = *gw_addr; |
1da177e4 LT |
1437 | gwa_type = ipv6_addr_type(gw_addr); |
1438 | ||
1439 | if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) { | |
1440 | struct rt6_info *grt; | |
1441 | ||
1442 | /* IPv6 strictly inhibits using not link-local | |
1443 | addresses as nexthop address. | |
1444 | Otherwise, router will not able to send redirects. | |
1445 | It is very good, but in some (rare!) circumstances | |
1446 | (SIT, PtP, NBMA NOARP links) it is handy to allow | |
1447 | some exceptions. --ANK | |
1448 | */ | |
1449 | err = -EINVAL; | |
38308473 | 1450 | if (!(gwa_type & IPV6_ADDR_UNICAST)) |
1da177e4 LT |
1451 | goto out; |
1452 | ||
5578689a | 1453 | grt = rt6_lookup(net, gw_addr, NULL, cfg->fc_ifindex, 1); |
1da177e4 LT |
1454 | |
1455 | err = -EHOSTUNREACH; | |
38308473 | 1456 | if (!grt) |
1da177e4 LT |
1457 | goto out; |
1458 | if (dev) { | |
d1918542 | 1459 | if (dev != grt->dst.dev) { |
d8d1f30b | 1460 | dst_release(&grt->dst); |
1da177e4 LT |
1461 | goto out; |
1462 | } | |
1463 | } else { | |
d1918542 | 1464 | dev = grt->dst.dev; |
1da177e4 LT |
1465 | idev = grt->rt6i_idev; |
1466 | dev_hold(dev); | |
1467 | in6_dev_hold(grt->rt6i_idev); | |
1468 | } | |
38308473 | 1469 | if (!(grt->rt6i_flags & RTF_GATEWAY)) |
1da177e4 | 1470 | err = 0; |
d8d1f30b | 1471 | dst_release(&grt->dst); |
1da177e4 LT |
1472 | |
1473 | if (err) | |
1474 | goto out; | |
1475 | } | |
1476 | err = -EINVAL; | |
38308473 | 1477 | if (!dev || (dev->flags & IFF_LOOPBACK)) |
1da177e4 LT |
1478 | goto out; |
1479 | } | |
1480 | ||
1481 | err = -ENODEV; | |
38308473 | 1482 | if (!dev) |
1da177e4 LT |
1483 | goto out; |
1484 | ||
c3968a85 DW |
1485 | if (!ipv6_addr_any(&cfg->fc_prefsrc)) { |
1486 | if (!ipv6_chk_addr(net, &cfg->fc_prefsrc, dev, 0)) { | |
1487 | err = -EINVAL; | |
1488 | goto out; | |
1489 | } | |
4e3fd7a0 | 1490 | rt->rt6i_prefsrc.addr = cfg->fc_prefsrc; |
c3968a85 DW |
1491 | rt->rt6i_prefsrc.plen = 128; |
1492 | } else | |
1493 | rt->rt6i_prefsrc.plen = 0; | |
1494 | ||
86872cb5 | 1495 | if (cfg->fc_flags & (RTF_GATEWAY | RTF_NONEXTHOP)) { |
8ade06c6 | 1496 | err = rt6_bind_neighbour(rt, dev); |
f83c7790 | 1497 | if (err) |
1da177e4 | 1498 | goto out; |
1da177e4 LT |
1499 | } |
1500 | ||
86872cb5 | 1501 | rt->rt6i_flags = cfg->fc_flags; |
1da177e4 LT |
1502 | |
1503 | install_route: | |
86872cb5 TG |
1504 | if (cfg->fc_mx) { |
1505 | struct nlattr *nla; | |
1506 | int remaining; | |
1507 | ||
1508 | nla_for_each_attr(nla, cfg->fc_mx, cfg->fc_mx_len, remaining) { | |
8f4c1f9b | 1509 | int type = nla_type(nla); |
86872cb5 TG |
1510 | |
1511 | if (type) { | |
1512 | if (type > RTAX_MAX) { | |
1da177e4 LT |
1513 | err = -EINVAL; |
1514 | goto out; | |
1515 | } | |
86872cb5 | 1516 | |
defb3519 | 1517 | dst_metric_set(&rt->dst, type, nla_get_u32(nla)); |
1da177e4 | 1518 | } |
1da177e4 LT |
1519 | } |
1520 | } | |
1521 | ||
d8d1f30b | 1522 | rt->dst.dev = dev; |
1da177e4 | 1523 | rt->rt6i_idev = idev; |
c71099ac | 1524 | rt->rt6i_table = table; |
63152fc0 | 1525 | |
c346dca1 | 1526 | cfg->fc_nlinfo.nl_net = dev_net(dev); |
63152fc0 | 1527 | |
86872cb5 | 1528 | return __ip6_ins_rt(rt, &cfg->fc_nlinfo); |
1da177e4 LT |
1529 | |
1530 | out: | |
1531 | if (dev) | |
1532 | dev_put(dev); | |
1533 | if (idev) | |
1534 | in6_dev_put(idev); | |
1535 | if (rt) | |
d8d1f30b | 1536 | dst_free(&rt->dst); |
1da177e4 LT |
1537 | return err; |
1538 | } | |
1539 | ||
86872cb5 | 1540 | static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info) |
1da177e4 LT |
1541 | { |
1542 | int err; | |
c71099ac | 1543 | struct fib6_table *table; |
d1918542 | 1544 | struct net *net = dev_net(rt->dst.dev); |
1da177e4 | 1545 | |
8ed67789 | 1546 | if (rt == net->ipv6.ip6_null_entry) |
6c813a72 PM |
1547 | return -ENOENT; |
1548 | ||
c71099ac TG |
1549 | table = rt->rt6i_table; |
1550 | write_lock_bh(&table->tb6_lock); | |
1da177e4 | 1551 | |
86872cb5 | 1552 | err = fib6_del(rt, info); |
d8d1f30b | 1553 | dst_release(&rt->dst); |
1da177e4 | 1554 | |
c71099ac | 1555 | write_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
1556 | |
1557 | return err; | |
1558 | } | |
1559 | ||
e0a1ad73 TG |
1560 | int ip6_del_rt(struct rt6_info *rt) |
1561 | { | |
4d1169c1 | 1562 | struct nl_info info = { |
d1918542 | 1563 | .nl_net = dev_net(rt->dst.dev), |
4d1169c1 | 1564 | }; |
528c4ceb | 1565 | return __ip6_del_rt(rt, &info); |
e0a1ad73 TG |
1566 | } |
1567 | ||
86872cb5 | 1568 | static int ip6_route_del(struct fib6_config *cfg) |
1da177e4 | 1569 | { |
c71099ac | 1570 | struct fib6_table *table; |
1da177e4 LT |
1571 | struct fib6_node *fn; |
1572 | struct rt6_info *rt; | |
1573 | int err = -ESRCH; | |
1574 | ||
5578689a | 1575 | table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table); |
38308473 | 1576 | if (!table) |
c71099ac TG |
1577 | return err; |
1578 | ||
1579 | read_lock_bh(&table->tb6_lock); | |
1da177e4 | 1580 | |
c71099ac | 1581 | fn = fib6_locate(&table->tb6_root, |
86872cb5 TG |
1582 | &cfg->fc_dst, cfg->fc_dst_len, |
1583 | &cfg->fc_src, cfg->fc_src_len); | |
1ab1457c | 1584 | |
1da177e4 | 1585 | if (fn) { |
d8d1f30b | 1586 | for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) { |
86872cb5 | 1587 | if (cfg->fc_ifindex && |
d1918542 DM |
1588 | (!rt->dst.dev || |
1589 | rt->dst.dev->ifindex != cfg->fc_ifindex)) | |
1da177e4 | 1590 | continue; |
86872cb5 TG |
1591 | if (cfg->fc_flags & RTF_GATEWAY && |
1592 | !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway)) | |
1da177e4 | 1593 | continue; |
86872cb5 | 1594 | if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric) |
1da177e4 | 1595 | continue; |
d8d1f30b | 1596 | dst_hold(&rt->dst); |
c71099ac | 1597 | read_unlock_bh(&table->tb6_lock); |
1da177e4 | 1598 | |
86872cb5 | 1599 | return __ip6_del_rt(rt, &cfg->fc_nlinfo); |
1da177e4 LT |
1600 | } |
1601 | } | |
c71099ac | 1602 | read_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
1603 | |
1604 | return err; | |
1605 | } | |
1606 | ||
1607 | /* | |
1608 | * Handle redirects | |
1609 | */ | |
a6279458 | 1610 | struct ip6rd_flowi { |
4c9483b2 | 1611 | struct flowi6 fl6; |
a6279458 YH |
1612 | struct in6_addr gateway; |
1613 | }; | |
1614 | ||
8ed67789 DL |
1615 | static struct rt6_info *__ip6_route_redirect(struct net *net, |
1616 | struct fib6_table *table, | |
4c9483b2 | 1617 | struct flowi6 *fl6, |
a6279458 | 1618 | int flags) |
1da177e4 | 1619 | { |
4c9483b2 | 1620 | struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl6; |
a6279458 | 1621 | struct rt6_info *rt; |
e843b9e1 | 1622 | struct fib6_node *fn; |
c71099ac | 1623 | |
1da177e4 | 1624 | /* |
e843b9e1 YH |
1625 | * Get the "current" route for this destination and |
1626 | * check if the redirect has come from approriate router. | |
1627 | * | |
1628 | * RFC 2461 specifies that redirects should only be | |
1629 | * accepted if they come from the nexthop to the target. | |
1630 | * Due to the way the routes are chosen, this notion | |
1631 | * is a bit fuzzy and one might need to check all possible | |
1632 | * routes. | |
1da177e4 | 1633 | */ |
1da177e4 | 1634 | |
c71099ac | 1635 | read_lock_bh(&table->tb6_lock); |
4c9483b2 | 1636 | fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr); |
e843b9e1 | 1637 | restart: |
d8d1f30b | 1638 | for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) { |
e843b9e1 YH |
1639 | /* |
1640 | * Current route is on-link; redirect is always invalid. | |
1641 | * | |
1642 | * Seems, previous statement is not true. It could | |
1643 | * be node, which looks for us as on-link (f.e. proxy ndisc) | |
1644 | * But then router serving it might decide, that we should | |
1645 | * know truth 8)8) --ANK (980726). | |
1646 | */ | |
1647 | if (rt6_check_expired(rt)) | |
1648 | continue; | |
1649 | if (!(rt->rt6i_flags & RTF_GATEWAY)) | |
1650 | continue; | |
d1918542 | 1651 | if (fl6->flowi6_oif != rt->dst.dev->ifindex) |
e843b9e1 | 1652 | continue; |
a6279458 | 1653 | if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway)) |
e843b9e1 YH |
1654 | continue; |
1655 | break; | |
1656 | } | |
a6279458 | 1657 | |
cb15d9c2 | 1658 | if (!rt) |
8ed67789 | 1659 | rt = net->ipv6.ip6_null_entry; |
4c9483b2 | 1660 | BACKTRACK(net, &fl6->saddr); |
cb15d9c2 | 1661 | out: |
d8d1f30b | 1662 | dst_hold(&rt->dst); |
a6279458 | 1663 | |
c71099ac | 1664 | read_unlock_bh(&table->tb6_lock); |
e843b9e1 | 1665 | |
a6279458 YH |
1666 | return rt; |
1667 | }; | |
1668 | ||
b71d1d42 ED |
1669 | static struct rt6_info *ip6_route_redirect(const struct in6_addr *dest, |
1670 | const struct in6_addr *src, | |
1671 | const struct in6_addr *gateway, | |
a6279458 YH |
1672 | struct net_device *dev) |
1673 | { | |
adaa70bb | 1674 | int flags = RT6_LOOKUP_F_HAS_SADDR; |
c346dca1 | 1675 | struct net *net = dev_net(dev); |
a6279458 | 1676 | struct ip6rd_flowi rdfl = { |
4c9483b2 DM |
1677 | .fl6 = { |
1678 | .flowi6_oif = dev->ifindex, | |
1679 | .daddr = *dest, | |
1680 | .saddr = *src, | |
a6279458 | 1681 | }, |
a6279458 | 1682 | }; |
adaa70bb | 1683 | |
4e3fd7a0 | 1684 | rdfl.gateway = *gateway; |
86c36ce4 | 1685 | |
adaa70bb TG |
1686 | if (rt6_need_strict(dest)) |
1687 | flags |= RT6_LOOKUP_F_IFACE; | |
a6279458 | 1688 | |
4c9483b2 | 1689 | return (struct rt6_info *)fib6_rule_lookup(net, &rdfl.fl6, |
58f09b78 | 1690 | flags, __ip6_route_redirect); |
a6279458 YH |
1691 | } |
1692 | ||
b71d1d42 ED |
1693 | void rt6_redirect(const struct in6_addr *dest, const struct in6_addr *src, |
1694 | const struct in6_addr *saddr, | |
a6279458 YH |
1695 | struct neighbour *neigh, u8 *lladdr, int on_link) |
1696 | { | |
1697 | struct rt6_info *rt, *nrt = NULL; | |
1698 | struct netevent_redirect netevent; | |
c346dca1 | 1699 | struct net *net = dev_net(neigh->dev); |
1d248b1c | 1700 | struct neighbour *old_neigh; |
a6279458 YH |
1701 | |
1702 | rt = ip6_route_redirect(dest, src, saddr, neigh->dev); | |
1703 | ||
8ed67789 | 1704 | if (rt == net->ipv6.ip6_null_entry) { |
e87cc472 | 1705 | net_dbg_ratelimited("rt6_redirect: source isn't a valid nexthop for redirect target\n"); |
a6279458 | 1706 | goto out; |
1da177e4 LT |
1707 | } |
1708 | ||
1da177e4 LT |
1709 | /* |
1710 | * We have finally decided to accept it. | |
1711 | */ | |
1712 | ||
1ab1457c | 1713 | neigh_update(neigh, lladdr, NUD_STALE, |
1da177e4 LT |
1714 | NEIGH_UPDATE_F_WEAK_OVERRIDE| |
1715 | NEIGH_UPDATE_F_OVERRIDE| | |
1716 | (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER| | |
1717 | NEIGH_UPDATE_F_ISROUTER)) | |
1718 | ); | |
1719 | ||
1720 | /* | |
1721 | * Redirect received -> path was valid. | |
1722 | * Look, redirects are sent only in response to data packets, | |
1723 | * so that this nexthop apparently is reachable. --ANK | |
1724 | */ | |
d8d1f30b | 1725 | dst_confirm(&rt->dst); |
1da177e4 LT |
1726 | |
1727 | /* Duplicate redirect: silently ignore. */ | |
97cac082 | 1728 | old_neigh = rt->n; |
1d248b1c | 1729 | if (neigh == old_neigh) |
1da177e4 LT |
1730 | goto out; |
1731 | ||
21efcfa0 | 1732 | nrt = ip6_rt_copy(rt, dest); |
38308473 | 1733 | if (!nrt) |
1da177e4 LT |
1734 | goto out; |
1735 | ||
1736 | nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE; | |
1737 | if (on_link) | |
1738 | nrt->rt6i_flags &= ~RTF_GATEWAY; | |
1739 | ||
4e3fd7a0 | 1740 | nrt->rt6i_gateway = *(struct in6_addr *)neigh->primary_key; |
97cac082 | 1741 | nrt->n = neigh_clone(neigh); |
1da177e4 | 1742 | |
40e22e8f | 1743 | if (ip6_ins_rt(nrt)) |
1da177e4 LT |
1744 | goto out; |
1745 | ||
d8d1f30b | 1746 | netevent.old = &rt->dst; |
1d248b1c | 1747 | netevent.old_neigh = old_neigh; |
d8d1f30b | 1748 | netevent.new = &nrt->dst; |
1d248b1c DM |
1749 | netevent.new_neigh = neigh; |
1750 | netevent.daddr = dest; | |
8d71740c TT |
1751 | call_netevent_notifiers(NETEVENT_REDIRECT, &netevent); |
1752 | ||
38308473 | 1753 | if (rt->rt6i_flags & RTF_CACHE) { |
e0a1ad73 | 1754 | ip6_del_rt(rt); |
1da177e4 LT |
1755 | return; |
1756 | } | |
1757 | ||
1758 | out: | |
d8d1f30b | 1759 | dst_release(&rt->dst); |
1da177e4 LT |
1760 | } |
1761 | ||
1da177e4 LT |
1762 | /* |
1763 | * Misc support functions | |
1764 | */ | |
1765 | ||
1716a961 | 1766 | static struct rt6_info *ip6_rt_copy(struct rt6_info *ort, |
21efcfa0 | 1767 | const struct in6_addr *dest) |
1da177e4 | 1768 | { |
d1918542 | 1769 | struct net *net = dev_net(ort->dst.dev); |
8b96d22d DM |
1770 | struct rt6_info *rt = ip6_dst_alloc(net, ort->dst.dev, 0, |
1771 | ort->rt6i_table); | |
1da177e4 LT |
1772 | |
1773 | if (rt) { | |
d8d1f30b CG |
1774 | rt->dst.input = ort->dst.input; |
1775 | rt->dst.output = ort->dst.output; | |
8e2ec639 | 1776 | rt->dst.flags |= DST_HOST; |
d8d1f30b | 1777 | |
4e3fd7a0 | 1778 | rt->rt6i_dst.addr = *dest; |
8e2ec639 | 1779 | rt->rt6i_dst.plen = 128; |
defb3519 | 1780 | dst_copy_metrics(&rt->dst, &ort->dst); |
d8d1f30b | 1781 | rt->dst.error = ort->dst.error; |
1da177e4 LT |
1782 | rt->rt6i_idev = ort->rt6i_idev; |
1783 | if (rt->rt6i_idev) | |
1784 | in6_dev_hold(rt->rt6i_idev); | |
d8d1f30b | 1785 | rt->dst.lastuse = jiffies; |
1da177e4 | 1786 | |
4e3fd7a0 | 1787 | rt->rt6i_gateway = ort->rt6i_gateway; |
1716a961 G |
1788 | rt->rt6i_flags = ort->rt6i_flags; |
1789 | if ((ort->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF)) == | |
1790 | (RTF_DEFAULT | RTF_ADDRCONF)) | |
1791 | rt6_set_from(rt, ort); | |
1792 | else | |
1793 | rt6_clean_expires(rt); | |
1da177e4 LT |
1794 | rt->rt6i_metric = 0; |
1795 | ||
1da177e4 LT |
1796 | #ifdef CONFIG_IPV6_SUBTREES |
1797 | memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key)); | |
1798 | #endif | |
0f6c6392 | 1799 | memcpy(&rt->rt6i_prefsrc, &ort->rt6i_prefsrc, sizeof(struct rt6key)); |
c71099ac | 1800 | rt->rt6i_table = ort->rt6i_table; |
1da177e4 LT |
1801 | } |
1802 | return rt; | |
1803 | } | |
1804 | ||
70ceb4f5 | 1805 | #ifdef CONFIG_IPV6_ROUTE_INFO |
efa2cea0 | 1806 | static struct rt6_info *rt6_get_route_info(struct net *net, |
b71d1d42 ED |
1807 | const struct in6_addr *prefix, int prefixlen, |
1808 | const struct in6_addr *gwaddr, int ifindex) | |
70ceb4f5 YH |
1809 | { |
1810 | struct fib6_node *fn; | |
1811 | struct rt6_info *rt = NULL; | |
c71099ac TG |
1812 | struct fib6_table *table; |
1813 | ||
efa2cea0 | 1814 | table = fib6_get_table(net, RT6_TABLE_INFO); |
38308473 | 1815 | if (!table) |
c71099ac | 1816 | return NULL; |
70ceb4f5 | 1817 | |
c71099ac TG |
1818 | write_lock_bh(&table->tb6_lock); |
1819 | fn = fib6_locate(&table->tb6_root, prefix ,prefixlen, NULL, 0); | |
70ceb4f5 YH |
1820 | if (!fn) |
1821 | goto out; | |
1822 | ||
d8d1f30b | 1823 | for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) { |
d1918542 | 1824 | if (rt->dst.dev->ifindex != ifindex) |
70ceb4f5 YH |
1825 | continue; |
1826 | if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY)) | |
1827 | continue; | |
1828 | if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr)) | |
1829 | continue; | |
d8d1f30b | 1830 | dst_hold(&rt->dst); |
70ceb4f5 YH |
1831 | break; |
1832 | } | |
1833 | out: | |
c71099ac | 1834 | write_unlock_bh(&table->tb6_lock); |
70ceb4f5 YH |
1835 | return rt; |
1836 | } | |
1837 | ||
efa2cea0 | 1838 | static struct rt6_info *rt6_add_route_info(struct net *net, |
b71d1d42 ED |
1839 | const struct in6_addr *prefix, int prefixlen, |
1840 | const struct in6_addr *gwaddr, int ifindex, | |
95c96174 | 1841 | unsigned int pref) |
70ceb4f5 | 1842 | { |
86872cb5 TG |
1843 | struct fib6_config cfg = { |
1844 | .fc_table = RT6_TABLE_INFO, | |
238fc7ea | 1845 | .fc_metric = IP6_RT_PRIO_USER, |
86872cb5 TG |
1846 | .fc_ifindex = ifindex, |
1847 | .fc_dst_len = prefixlen, | |
1848 | .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO | | |
1849 | RTF_UP | RTF_PREF(pref), | |
efa2cea0 DL |
1850 | .fc_nlinfo.pid = 0, |
1851 | .fc_nlinfo.nlh = NULL, | |
1852 | .fc_nlinfo.nl_net = net, | |
86872cb5 TG |
1853 | }; |
1854 | ||
4e3fd7a0 AD |
1855 | cfg.fc_dst = *prefix; |
1856 | cfg.fc_gateway = *gwaddr; | |
70ceb4f5 | 1857 | |
e317da96 YH |
1858 | /* We should treat it as a default route if prefix length is 0. */ |
1859 | if (!prefixlen) | |
86872cb5 | 1860 | cfg.fc_flags |= RTF_DEFAULT; |
70ceb4f5 | 1861 | |
86872cb5 | 1862 | ip6_route_add(&cfg); |
70ceb4f5 | 1863 | |
efa2cea0 | 1864 | return rt6_get_route_info(net, prefix, prefixlen, gwaddr, ifindex); |
70ceb4f5 YH |
1865 | } |
1866 | #endif | |
1867 | ||
b71d1d42 | 1868 | struct rt6_info *rt6_get_dflt_router(const struct in6_addr *addr, struct net_device *dev) |
1ab1457c | 1869 | { |
1da177e4 | 1870 | struct rt6_info *rt; |
c71099ac | 1871 | struct fib6_table *table; |
1da177e4 | 1872 | |
c346dca1 | 1873 | table = fib6_get_table(dev_net(dev), RT6_TABLE_DFLT); |
38308473 | 1874 | if (!table) |
c71099ac | 1875 | return NULL; |
1da177e4 | 1876 | |
c71099ac | 1877 | write_lock_bh(&table->tb6_lock); |
d8d1f30b | 1878 | for (rt = table->tb6_root.leaf; rt; rt=rt->dst.rt6_next) { |
d1918542 | 1879 | if (dev == rt->dst.dev && |
045927ff | 1880 | ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) && |
1da177e4 LT |
1881 | ipv6_addr_equal(&rt->rt6i_gateway, addr)) |
1882 | break; | |
1883 | } | |
1884 | if (rt) | |
d8d1f30b | 1885 | dst_hold(&rt->dst); |
c71099ac | 1886 | write_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
1887 | return rt; |
1888 | } | |
1889 | ||
b71d1d42 | 1890 | struct rt6_info *rt6_add_dflt_router(const struct in6_addr *gwaddr, |
ebacaaa0 YH |
1891 | struct net_device *dev, |
1892 | unsigned int pref) | |
1da177e4 | 1893 | { |
86872cb5 TG |
1894 | struct fib6_config cfg = { |
1895 | .fc_table = RT6_TABLE_DFLT, | |
238fc7ea | 1896 | .fc_metric = IP6_RT_PRIO_USER, |
86872cb5 TG |
1897 | .fc_ifindex = dev->ifindex, |
1898 | .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT | | |
1899 | RTF_UP | RTF_EXPIRES | RTF_PREF(pref), | |
5578689a DL |
1900 | .fc_nlinfo.pid = 0, |
1901 | .fc_nlinfo.nlh = NULL, | |
c346dca1 | 1902 | .fc_nlinfo.nl_net = dev_net(dev), |
86872cb5 | 1903 | }; |
1da177e4 | 1904 | |
4e3fd7a0 | 1905 | cfg.fc_gateway = *gwaddr; |
1da177e4 | 1906 | |
86872cb5 | 1907 | ip6_route_add(&cfg); |
1da177e4 | 1908 | |
1da177e4 LT |
1909 | return rt6_get_dflt_router(gwaddr, dev); |
1910 | } | |
1911 | ||
7b4da532 | 1912 | void rt6_purge_dflt_routers(struct net *net) |
1da177e4 LT |
1913 | { |
1914 | struct rt6_info *rt; | |
c71099ac TG |
1915 | struct fib6_table *table; |
1916 | ||
1917 | /* NOTE: Keep consistent with rt6_get_dflt_router */ | |
7b4da532 | 1918 | table = fib6_get_table(net, RT6_TABLE_DFLT); |
38308473 | 1919 | if (!table) |
c71099ac | 1920 | return; |
1da177e4 LT |
1921 | |
1922 | restart: | |
c71099ac | 1923 | read_lock_bh(&table->tb6_lock); |
d8d1f30b | 1924 | for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) { |
1da177e4 | 1925 | if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF)) { |
d8d1f30b | 1926 | dst_hold(&rt->dst); |
c71099ac | 1927 | read_unlock_bh(&table->tb6_lock); |
e0a1ad73 | 1928 | ip6_del_rt(rt); |
1da177e4 LT |
1929 | goto restart; |
1930 | } | |
1931 | } | |
c71099ac | 1932 | read_unlock_bh(&table->tb6_lock); |
1da177e4 LT |
1933 | } |
1934 | ||
5578689a DL |
1935 | static void rtmsg_to_fib6_config(struct net *net, |
1936 | struct in6_rtmsg *rtmsg, | |
86872cb5 TG |
1937 | struct fib6_config *cfg) |
1938 | { | |
1939 | memset(cfg, 0, sizeof(*cfg)); | |
1940 | ||
1941 | cfg->fc_table = RT6_TABLE_MAIN; | |
1942 | cfg->fc_ifindex = rtmsg->rtmsg_ifindex; | |
1943 | cfg->fc_metric = rtmsg->rtmsg_metric; | |
1944 | cfg->fc_expires = rtmsg->rtmsg_info; | |
1945 | cfg->fc_dst_len = rtmsg->rtmsg_dst_len; | |
1946 | cfg->fc_src_len = rtmsg->rtmsg_src_len; | |
1947 | cfg->fc_flags = rtmsg->rtmsg_flags; | |
1948 | ||
5578689a | 1949 | cfg->fc_nlinfo.nl_net = net; |
f1243c2d | 1950 | |
4e3fd7a0 AD |
1951 | cfg->fc_dst = rtmsg->rtmsg_dst; |
1952 | cfg->fc_src = rtmsg->rtmsg_src; | |
1953 | cfg->fc_gateway = rtmsg->rtmsg_gateway; | |
86872cb5 TG |
1954 | } |
1955 | ||
5578689a | 1956 | int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg) |
1da177e4 | 1957 | { |
86872cb5 | 1958 | struct fib6_config cfg; |
1da177e4 LT |
1959 | struct in6_rtmsg rtmsg; |
1960 | int err; | |
1961 | ||
1962 | switch(cmd) { | |
1963 | case SIOCADDRT: /* Add a route */ | |
1964 | case SIOCDELRT: /* Delete a route */ | |
1965 | if (!capable(CAP_NET_ADMIN)) | |
1966 | return -EPERM; | |
1967 | err = copy_from_user(&rtmsg, arg, | |
1968 | sizeof(struct in6_rtmsg)); | |
1969 | if (err) | |
1970 | return -EFAULT; | |
86872cb5 | 1971 | |
5578689a | 1972 | rtmsg_to_fib6_config(net, &rtmsg, &cfg); |
86872cb5 | 1973 | |
1da177e4 LT |
1974 | rtnl_lock(); |
1975 | switch (cmd) { | |
1976 | case SIOCADDRT: | |
86872cb5 | 1977 | err = ip6_route_add(&cfg); |
1da177e4 LT |
1978 | break; |
1979 | case SIOCDELRT: | |
86872cb5 | 1980 | err = ip6_route_del(&cfg); |
1da177e4 LT |
1981 | break; |
1982 | default: | |
1983 | err = -EINVAL; | |
1984 | } | |
1985 | rtnl_unlock(); | |
1986 | ||
1987 | return err; | |
3ff50b79 | 1988 | } |
1da177e4 LT |
1989 | |
1990 | return -EINVAL; | |
1991 | } | |
1992 | ||
1993 | /* | |
1994 | * Drop the packet on the floor | |
1995 | */ | |
1996 | ||
d5fdd6ba | 1997 | static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes) |
1da177e4 | 1998 | { |
612f09e8 | 1999 | int type; |
adf30907 | 2000 | struct dst_entry *dst = skb_dst(skb); |
612f09e8 YH |
2001 | switch (ipstats_mib_noroutes) { |
2002 | case IPSTATS_MIB_INNOROUTES: | |
0660e03f | 2003 | type = ipv6_addr_type(&ipv6_hdr(skb)->daddr); |
45bb0060 | 2004 | if (type == IPV6_ADDR_ANY) { |
3bd653c8 DL |
2005 | IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst), |
2006 | IPSTATS_MIB_INADDRERRORS); | |
612f09e8 YH |
2007 | break; |
2008 | } | |
2009 | /* FALLTHROUGH */ | |
2010 | case IPSTATS_MIB_OUTNOROUTES: | |
3bd653c8 DL |
2011 | IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst), |
2012 | ipstats_mib_noroutes); | |
612f09e8 YH |
2013 | break; |
2014 | } | |
3ffe533c | 2015 | icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0); |
1da177e4 LT |
2016 | kfree_skb(skb); |
2017 | return 0; | |
2018 | } | |
2019 | ||
9ce8ade0 TG |
2020 | static int ip6_pkt_discard(struct sk_buff *skb) |
2021 | { | |
612f09e8 | 2022 | return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES); |
9ce8ade0 TG |
2023 | } |
2024 | ||
20380731 | 2025 | static int ip6_pkt_discard_out(struct sk_buff *skb) |
1da177e4 | 2026 | { |
adf30907 | 2027 | skb->dev = skb_dst(skb)->dev; |
612f09e8 | 2028 | return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES); |
1da177e4 LT |
2029 | } |
2030 | ||
6723ab54 DM |
2031 | #ifdef CONFIG_IPV6_MULTIPLE_TABLES |
2032 | ||
9ce8ade0 TG |
2033 | static int ip6_pkt_prohibit(struct sk_buff *skb) |
2034 | { | |
612f09e8 | 2035 | return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES); |
9ce8ade0 TG |
2036 | } |
2037 | ||
2038 | static int ip6_pkt_prohibit_out(struct sk_buff *skb) | |
2039 | { | |
adf30907 | 2040 | skb->dev = skb_dst(skb)->dev; |
612f09e8 | 2041 | return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES); |
9ce8ade0 TG |
2042 | } |
2043 | ||
6723ab54 DM |
2044 | #endif |
2045 | ||
1da177e4 LT |
2046 | /* |
2047 | * Allocate a dst for local (unicast / anycast) address. | |
2048 | */ | |
2049 | ||
2050 | struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev, | |
2051 | const struct in6_addr *addr, | |
8f031519 | 2052 | bool anycast) |
1da177e4 | 2053 | { |
c346dca1 | 2054 | struct net *net = dev_net(idev->dev); |
8b96d22d | 2055 | struct rt6_info *rt = ip6_dst_alloc(net, net->loopback_dev, 0, NULL); |
f83c7790 | 2056 | int err; |
1da177e4 | 2057 | |
38308473 | 2058 | if (!rt) { |
f3213831 | 2059 | net_warn_ratelimited("Maximum number of routes reached, consider increasing route/max_size\n"); |
1da177e4 | 2060 | return ERR_PTR(-ENOMEM); |
40385653 | 2061 | } |
1da177e4 | 2062 | |
1da177e4 LT |
2063 | in6_dev_hold(idev); |
2064 | ||
11d53b49 | 2065 | rt->dst.flags |= DST_HOST; |
d8d1f30b CG |
2066 | rt->dst.input = ip6_input; |
2067 | rt->dst.output = ip6_output; | |
1da177e4 | 2068 | rt->rt6i_idev = idev; |
d8d1f30b | 2069 | rt->dst.obsolete = -1; |
1da177e4 LT |
2070 | |
2071 | rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP; | |
58c4fb86 YH |
2072 | if (anycast) |
2073 | rt->rt6i_flags |= RTF_ANYCAST; | |
2074 | else | |
1da177e4 | 2075 | rt->rt6i_flags |= RTF_LOCAL; |
8ade06c6 | 2076 | err = rt6_bind_neighbour(rt, rt->dst.dev); |
f83c7790 | 2077 | if (err) { |
d8d1f30b | 2078 | dst_free(&rt->dst); |
f83c7790 | 2079 | return ERR_PTR(err); |
1da177e4 LT |
2080 | } |
2081 | ||
4e3fd7a0 | 2082 | rt->rt6i_dst.addr = *addr; |
1da177e4 | 2083 | rt->rt6i_dst.plen = 128; |
5578689a | 2084 | rt->rt6i_table = fib6_get_table(net, RT6_TABLE_LOCAL); |
1da177e4 | 2085 | |
d8d1f30b | 2086 | atomic_set(&rt->dst.__refcnt, 1); |
1da177e4 LT |
2087 | |
2088 | return rt; | |
2089 | } | |
2090 | ||
c3968a85 DW |
2091 | int ip6_route_get_saddr(struct net *net, |
2092 | struct rt6_info *rt, | |
b71d1d42 | 2093 | const struct in6_addr *daddr, |
c3968a85 DW |
2094 | unsigned int prefs, |
2095 | struct in6_addr *saddr) | |
2096 | { | |
2097 | struct inet6_dev *idev = ip6_dst_idev((struct dst_entry*)rt); | |
2098 | int err = 0; | |
2099 | if (rt->rt6i_prefsrc.plen) | |
4e3fd7a0 | 2100 | *saddr = rt->rt6i_prefsrc.addr; |
c3968a85 DW |
2101 | else |
2102 | err = ipv6_dev_get_saddr(net, idev ? idev->dev : NULL, | |
2103 | daddr, prefs, saddr); | |
2104 | return err; | |
2105 | } | |
2106 | ||
2107 | /* remove deleted ip from prefsrc entries */ | |
2108 | struct arg_dev_net_ip { | |
2109 | struct net_device *dev; | |
2110 | struct net *net; | |
2111 | struct in6_addr *addr; | |
2112 | }; | |
2113 | ||
2114 | static int fib6_remove_prefsrc(struct rt6_info *rt, void *arg) | |
2115 | { | |
2116 | struct net_device *dev = ((struct arg_dev_net_ip *)arg)->dev; | |
2117 | struct net *net = ((struct arg_dev_net_ip *)arg)->net; | |
2118 | struct in6_addr *addr = ((struct arg_dev_net_ip *)arg)->addr; | |
2119 | ||
d1918542 | 2120 | if (((void *)rt->dst.dev == dev || !dev) && |
c3968a85 DW |
2121 | rt != net->ipv6.ip6_null_entry && |
2122 | ipv6_addr_equal(addr, &rt->rt6i_prefsrc.addr)) { | |
2123 | /* remove prefsrc entry */ | |
2124 | rt->rt6i_prefsrc.plen = 0; | |
2125 | } | |
2126 | return 0; | |
2127 | } | |
2128 | ||
2129 | void rt6_remove_prefsrc(struct inet6_ifaddr *ifp) | |
2130 | { | |
2131 | struct net *net = dev_net(ifp->idev->dev); | |
2132 | struct arg_dev_net_ip adni = { | |
2133 | .dev = ifp->idev->dev, | |
2134 | .net = net, | |
2135 | .addr = &ifp->addr, | |
2136 | }; | |
2137 | fib6_clean_all(net, fib6_remove_prefsrc, 0, &adni); | |
2138 | } | |
2139 | ||
8ed67789 DL |
2140 | struct arg_dev_net { |
2141 | struct net_device *dev; | |
2142 | struct net *net; | |
2143 | }; | |
2144 | ||
1da177e4 LT |
2145 | static int fib6_ifdown(struct rt6_info *rt, void *arg) |
2146 | { | |
bc3ef660 | 2147 | const struct arg_dev_net *adn = arg; |
2148 | const struct net_device *dev = adn->dev; | |
8ed67789 | 2149 | |
d1918542 | 2150 | if ((rt->dst.dev == dev || !dev) && |
c159d30c | 2151 | rt != adn->net->ipv6.ip6_null_entry) |
1da177e4 | 2152 | return -1; |
c159d30c | 2153 | |
1da177e4 LT |
2154 | return 0; |
2155 | } | |
2156 | ||
f3db4851 | 2157 | void rt6_ifdown(struct net *net, struct net_device *dev) |
1da177e4 | 2158 | { |
8ed67789 DL |
2159 | struct arg_dev_net adn = { |
2160 | .dev = dev, | |
2161 | .net = net, | |
2162 | }; | |
2163 | ||
2164 | fib6_clean_all(net, fib6_ifdown, 0, &adn); | |
1e493d19 | 2165 | icmp6_clean_all(fib6_ifdown, &adn); |
1da177e4 LT |
2166 | } |
2167 | ||
95c96174 | 2168 | struct rt6_mtu_change_arg { |
1da177e4 | 2169 | struct net_device *dev; |
95c96174 | 2170 | unsigned int mtu; |
1da177e4 LT |
2171 | }; |
2172 | ||
2173 | static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg) | |
2174 | { | |
2175 | struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg; | |
2176 | struct inet6_dev *idev; | |
2177 | ||
2178 | /* In IPv6 pmtu discovery is not optional, | |
2179 | so that RTAX_MTU lock cannot disable it. | |
2180 | We still use this lock to block changes | |
2181 | caused by addrconf/ndisc. | |
2182 | */ | |
2183 | ||
2184 | idev = __in6_dev_get(arg->dev); | |
38308473 | 2185 | if (!idev) |
1da177e4 LT |
2186 | return 0; |
2187 | ||
2188 | /* For administrative MTU increase, there is no way to discover | |
2189 | IPv6 PMTU increase, so PMTU increase should be updated here. | |
2190 | Since RFC 1981 doesn't include administrative MTU increase | |
2191 | update PMTU increase is a MUST. (i.e. jumbo frame) | |
2192 | */ | |
2193 | /* | |
2194 | If new MTU is less than route PMTU, this new MTU will be the | |
2195 | lowest MTU in the path, update the route PMTU to reflect PMTU | |
2196 | decreases; if new MTU is greater than route PMTU, and the | |
2197 | old MTU is the lowest MTU in the path, update the route PMTU | |
2198 | to reflect the increase. In this case if the other nodes' MTU | |
2199 | also have the lowest MTU, TOO BIG MESSAGE will be lead to | |
2200 | PMTU discouvery. | |
2201 | */ | |
d1918542 | 2202 | if (rt->dst.dev == arg->dev && |
d8d1f30b CG |
2203 | !dst_metric_locked(&rt->dst, RTAX_MTU) && |
2204 | (dst_mtu(&rt->dst) >= arg->mtu || | |
2205 | (dst_mtu(&rt->dst) < arg->mtu && | |
2206 | dst_mtu(&rt->dst) == idev->cnf.mtu6))) { | |
defb3519 | 2207 | dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu); |
566cfd8f | 2208 | } |
1da177e4 LT |
2209 | return 0; |
2210 | } | |
2211 | ||
95c96174 | 2212 | void rt6_mtu_change(struct net_device *dev, unsigned int mtu) |
1da177e4 | 2213 | { |
c71099ac TG |
2214 | struct rt6_mtu_change_arg arg = { |
2215 | .dev = dev, | |
2216 | .mtu = mtu, | |
2217 | }; | |
1da177e4 | 2218 | |
c346dca1 | 2219 | fib6_clean_all(dev_net(dev), rt6_mtu_change_route, 0, &arg); |
1da177e4 LT |
2220 | } |
2221 | ||
ef7c79ed | 2222 | static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = { |
5176f91e | 2223 | [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) }, |
86872cb5 | 2224 | [RTA_OIF] = { .type = NLA_U32 }, |
ab364a6f | 2225 | [RTA_IIF] = { .type = NLA_U32 }, |
86872cb5 TG |
2226 | [RTA_PRIORITY] = { .type = NLA_U32 }, |
2227 | [RTA_METRICS] = { .type = NLA_NESTED }, | |
2228 | }; | |
2229 | ||
2230 | static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh, | |
2231 | struct fib6_config *cfg) | |
1da177e4 | 2232 | { |
86872cb5 TG |
2233 | struct rtmsg *rtm; |
2234 | struct nlattr *tb[RTA_MAX+1]; | |
2235 | int err; | |
1da177e4 | 2236 | |
86872cb5 TG |
2237 | err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy); |
2238 | if (err < 0) | |
2239 | goto errout; | |
1da177e4 | 2240 | |
86872cb5 TG |
2241 | err = -EINVAL; |
2242 | rtm = nlmsg_data(nlh); | |
2243 | memset(cfg, 0, sizeof(*cfg)); | |
2244 | ||
2245 | cfg->fc_table = rtm->rtm_table; | |
2246 | cfg->fc_dst_len = rtm->rtm_dst_len; | |
2247 | cfg->fc_src_len = rtm->rtm_src_len; | |
2248 | cfg->fc_flags = RTF_UP; | |
2249 | cfg->fc_protocol = rtm->rtm_protocol; | |
2250 | ||
2251 | if (rtm->rtm_type == RTN_UNREACHABLE) | |
2252 | cfg->fc_flags |= RTF_REJECT; | |
2253 | ||
ab79ad14 MŻ |
2254 | if (rtm->rtm_type == RTN_LOCAL) |
2255 | cfg->fc_flags |= RTF_LOCAL; | |
2256 | ||
86872cb5 TG |
2257 | cfg->fc_nlinfo.pid = NETLINK_CB(skb).pid; |
2258 | cfg->fc_nlinfo.nlh = nlh; | |
3b1e0a65 | 2259 | cfg->fc_nlinfo.nl_net = sock_net(skb->sk); |
86872cb5 TG |
2260 | |
2261 | if (tb[RTA_GATEWAY]) { | |
2262 | nla_memcpy(&cfg->fc_gateway, tb[RTA_GATEWAY], 16); | |
2263 | cfg->fc_flags |= RTF_GATEWAY; | |
1da177e4 | 2264 | } |
86872cb5 TG |
2265 | |
2266 | if (tb[RTA_DST]) { | |
2267 | int plen = (rtm->rtm_dst_len + 7) >> 3; | |
2268 | ||
2269 | if (nla_len(tb[RTA_DST]) < plen) | |
2270 | goto errout; | |
2271 | ||
2272 | nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen); | |
1da177e4 | 2273 | } |
86872cb5 TG |
2274 | |
2275 | if (tb[RTA_SRC]) { | |
2276 | int plen = (rtm->rtm_src_len + 7) >> 3; | |
2277 | ||
2278 | if (nla_len(tb[RTA_SRC]) < plen) | |
2279 | goto errout; | |
2280 | ||
2281 | nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen); | |
1da177e4 | 2282 | } |
86872cb5 | 2283 | |
c3968a85 DW |
2284 | if (tb[RTA_PREFSRC]) |
2285 | nla_memcpy(&cfg->fc_prefsrc, tb[RTA_PREFSRC], 16); | |
2286 | ||
86872cb5 TG |
2287 | if (tb[RTA_OIF]) |
2288 | cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]); | |
2289 | ||
2290 | if (tb[RTA_PRIORITY]) | |
2291 | cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]); | |
2292 | ||
2293 | if (tb[RTA_METRICS]) { | |
2294 | cfg->fc_mx = nla_data(tb[RTA_METRICS]); | |
2295 | cfg->fc_mx_len = nla_len(tb[RTA_METRICS]); | |
1da177e4 | 2296 | } |
86872cb5 TG |
2297 | |
2298 | if (tb[RTA_TABLE]) | |
2299 | cfg->fc_table = nla_get_u32(tb[RTA_TABLE]); | |
2300 | ||
2301 | err = 0; | |
2302 | errout: | |
2303 | return err; | |
1da177e4 LT |
2304 | } |
2305 | ||
c127ea2c | 2306 | static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg) |
1da177e4 | 2307 | { |
86872cb5 TG |
2308 | struct fib6_config cfg; |
2309 | int err; | |
1da177e4 | 2310 | |
86872cb5 TG |
2311 | err = rtm_to_fib6_config(skb, nlh, &cfg); |
2312 | if (err < 0) | |
2313 | return err; | |
2314 | ||
2315 | return ip6_route_del(&cfg); | |
1da177e4 LT |
2316 | } |
2317 | ||
c127ea2c | 2318 | static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg) |
1da177e4 | 2319 | { |
86872cb5 TG |
2320 | struct fib6_config cfg; |
2321 | int err; | |
1da177e4 | 2322 | |
86872cb5 TG |
2323 | err = rtm_to_fib6_config(skb, nlh, &cfg); |
2324 | if (err < 0) | |
2325 | return err; | |
2326 | ||
2327 | return ip6_route_add(&cfg); | |
1da177e4 LT |
2328 | } |
2329 | ||
339bf98f TG |
2330 | static inline size_t rt6_nlmsg_size(void) |
2331 | { | |
2332 | return NLMSG_ALIGN(sizeof(struct rtmsg)) | |
2333 | + nla_total_size(16) /* RTA_SRC */ | |
2334 | + nla_total_size(16) /* RTA_DST */ | |
2335 | + nla_total_size(16) /* RTA_GATEWAY */ | |
2336 | + nla_total_size(16) /* RTA_PREFSRC */ | |
2337 | + nla_total_size(4) /* RTA_TABLE */ | |
2338 | + nla_total_size(4) /* RTA_IIF */ | |
2339 | + nla_total_size(4) /* RTA_OIF */ | |
2340 | + nla_total_size(4) /* RTA_PRIORITY */ | |
6a2b9ce0 | 2341 | + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */ |
339bf98f TG |
2342 | + nla_total_size(sizeof(struct rta_cacheinfo)); |
2343 | } | |
2344 | ||
191cd582 BH |
2345 | static int rt6_fill_node(struct net *net, |
2346 | struct sk_buff *skb, struct rt6_info *rt, | |
0d51aa80 JHS |
2347 | struct in6_addr *dst, struct in6_addr *src, |
2348 | int iif, int type, u32 pid, u32 seq, | |
7bc570c8 | 2349 | int prefix, int nowait, unsigned int flags) |
1da177e4 | 2350 | { |
346f870b | 2351 | const struct inet_peer *peer; |
1da177e4 | 2352 | struct rtmsg *rtm; |
2d7202bf | 2353 | struct nlmsghdr *nlh; |
e3703b3d | 2354 | long expires; |
9e762a4a | 2355 | u32 table; |
f2c31e32 | 2356 | struct neighbour *n; |
346f870b | 2357 | u32 ts, tsage; |
1da177e4 LT |
2358 | |
2359 | if (prefix) { /* user wants prefix routes only */ | |
2360 | if (!(rt->rt6i_flags & RTF_PREFIX_RT)) { | |
2361 | /* success since this is not a prefix route */ | |
2362 | return 1; | |
2363 | } | |
2364 | } | |
2365 | ||
2d7202bf | 2366 | nlh = nlmsg_put(skb, pid, seq, type, sizeof(*rtm), flags); |
38308473 | 2367 | if (!nlh) |
26932566 | 2368 | return -EMSGSIZE; |
2d7202bf TG |
2369 | |
2370 | rtm = nlmsg_data(nlh); | |
1da177e4 LT |
2371 | rtm->rtm_family = AF_INET6; |
2372 | rtm->rtm_dst_len = rt->rt6i_dst.plen; | |
2373 | rtm->rtm_src_len = rt->rt6i_src.plen; | |
2374 | rtm->rtm_tos = 0; | |
c71099ac | 2375 | if (rt->rt6i_table) |
9e762a4a | 2376 | table = rt->rt6i_table->tb6_id; |
c71099ac | 2377 | else |
9e762a4a PM |
2378 | table = RT6_TABLE_UNSPEC; |
2379 | rtm->rtm_table = table; | |
c78679e8 DM |
2380 | if (nla_put_u32(skb, RTA_TABLE, table)) |
2381 | goto nla_put_failure; | |
38308473 | 2382 | if (rt->rt6i_flags & RTF_REJECT) |
1da177e4 | 2383 | rtm->rtm_type = RTN_UNREACHABLE; |
38308473 | 2384 | else if (rt->rt6i_flags & RTF_LOCAL) |
ab79ad14 | 2385 | rtm->rtm_type = RTN_LOCAL; |
d1918542 | 2386 | else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK)) |
1da177e4 LT |
2387 | rtm->rtm_type = RTN_LOCAL; |
2388 | else | |
2389 | rtm->rtm_type = RTN_UNICAST; | |
2390 | rtm->rtm_flags = 0; | |
2391 | rtm->rtm_scope = RT_SCOPE_UNIVERSE; | |
2392 | rtm->rtm_protocol = rt->rt6i_protocol; | |
38308473 | 2393 | if (rt->rt6i_flags & RTF_DYNAMIC) |
1da177e4 LT |
2394 | rtm->rtm_protocol = RTPROT_REDIRECT; |
2395 | else if (rt->rt6i_flags & RTF_ADDRCONF) | |
2396 | rtm->rtm_protocol = RTPROT_KERNEL; | |
38308473 | 2397 | else if (rt->rt6i_flags & RTF_DEFAULT) |
1da177e4 LT |
2398 | rtm->rtm_protocol = RTPROT_RA; |
2399 | ||
38308473 | 2400 | if (rt->rt6i_flags & RTF_CACHE) |
1da177e4 LT |
2401 | rtm->rtm_flags |= RTM_F_CLONED; |
2402 | ||
2403 | if (dst) { | |
c78679e8 DM |
2404 | if (nla_put(skb, RTA_DST, 16, dst)) |
2405 | goto nla_put_failure; | |
1ab1457c | 2406 | rtm->rtm_dst_len = 128; |
1da177e4 | 2407 | } else if (rtm->rtm_dst_len) |
c78679e8 DM |
2408 | if (nla_put(skb, RTA_DST, 16, &rt->rt6i_dst.addr)) |
2409 | goto nla_put_failure; | |
1da177e4 LT |
2410 | #ifdef CONFIG_IPV6_SUBTREES |
2411 | if (src) { | |
c78679e8 DM |
2412 | if (nla_put(skb, RTA_SRC, 16, src)) |
2413 | goto nla_put_failure; | |
1ab1457c | 2414 | rtm->rtm_src_len = 128; |
c78679e8 DM |
2415 | } else if (rtm->rtm_src_len && |
2416 | nla_put(skb, RTA_SRC, 16, &rt->rt6i_src.addr)) | |
2417 | goto nla_put_failure; | |
1da177e4 | 2418 | #endif |
7bc570c8 YH |
2419 | if (iif) { |
2420 | #ifdef CONFIG_IPV6_MROUTE | |
2421 | if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) { | |
8229efda | 2422 | int err = ip6mr_get_route(net, skb, rtm, nowait); |
7bc570c8 YH |
2423 | if (err <= 0) { |
2424 | if (!nowait) { | |
2425 | if (err == 0) | |
2426 | return 0; | |
2427 | goto nla_put_failure; | |
2428 | } else { | |
2429 | if (err == -EMSGSIZE) | |
2430 | goto nla_put_failure; | |
2431 | } | |
2432 | } | |
2433 | } else | |
2434 | #endif | |
c78679e8 DM |
2435 | if (nla_put_u32(skb, RTA_IIF, iif)) |
2436 | goto nla_put_failure; | |
7bc570c8 | 2437 | } else if (dst) { |
1da177e4 | 2438 | struct in6_addr saddr_buf; |
c78679e8 DM |
2439 | if (ip6_route_get_saddr(net, rt, dst, 0, &saddr_buf) == 0 && |
2440 | nla_put(skb, RTA_PREFSRC, 16, &saddr_buf)) | |
2441 | goto nla_put_failure; | |
1da177e4 | 2442 | } |
2d7202bf | 2443 | |
c3968a85 DW |
2444 | if (rt->rt6i_prefsrc.plen) { |
2445 | struct in6_addr saddr_buf; | |
4e3fd7a0 | 2446 | saddr_buf = rt->rt6i_prefsrc.addr; |
c78679e8 DM |
2447 | if (nla_put(skb, RTA_PREFSRC, 16, &saddr_buf)) |
2448 | goto nla_put_failure; | |
c3968a85 DW |
2449 | } |
2450 | ||
defb3519 | 2451 | if (rtnetlink_put_metrics(skb, dst_metrics_ptr(&rt->dst)) < 0) |
2d7202bf TG |
2452 | goto nla_put_failure; |
2453 | ||
f2c31e32 | 2454 | rcu_read_lock(); |
97cac082 | 2455 | n = rt->n; |
94f826b8 ED |
2456 | if (n) { |
2457 | if (nla_put(skb, RTA_GATEWAY, 16, &n->primary_key) < 0) { | |
2458 | rcu_read_unlock(); | |
2459 | goto nla_put_failure; | |
2460 | } | |
2461 | } | |
f2c31e32 | 2462 | rcu_read_unlock(); |
2d7202bf | 2463 | |
c78679e8 DM |
2464 | if (rt->dst.dev && |
2465 | nla_put_u32(skb, RTA_OIF, rt->dst.dev->ifindex)) | |
2466 | goto nla_put_failure; | |
2467 | if (nla_put_u32(skb, RTA_PRIORITY, rt->rt6i_metric)) | |
2468 | goto nla_put_failure; | |
36e3deae YH |
2469 | if (!(rt->rt6i_flags & RTF_EXPIRES)) |
2470 | expires = 0; | |
d1918542 DM |
2471 | else if (rt->dst.expires - jiffies < INT_MAX) |
2472 | expires = rt->dst.expires - jiffies; | |
36e3deae YH |
2473 | else |
2474 | expires = INT_MAX; | |
69cdf8f9 | 2475 | |
97bab73f DM |
2476 | peer = NULL; |
2477 | if (rt6_has_peer(rt)) | |
2478 | peer = rt6_peer_ptr(rt); | |
346f870b DM |
2479 | ts = tsage = 0; |
2480 | if (peer && peer->tcp_ts_stamp) { | |
2481 | ts = peer->tcp_ts; | |
2482 | tsage = get_seconds() - peer->tcp_ts_stamp; | |
2483 | } | |
2484 | ||
2485 | if (rtnl_put_cacheinfo(skb, &rt->dst, 0, ts, tsage, | |
d8d1f30b | 2486 | 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 | } |