blob: a275c6e1e25c23884d7d1859e46a2ee82c00acef [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
2 * Linux INET6 implementation
3 * FIB front-end.
4 *
5 * Authors:
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09006 * Pedro Roque <roque@di.fc.ul.pt>
Linus Torvalds1da177e2005-04-16 15:20:36 -07007 *
Linus Torvalds1da177e2005-04-16 15:20:36 -07008 * 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.
YOSHIFUJI Hideakic0bece92006-08-23 17:23:25 -070023 * Ville Nuorvala
24 * Fixed routing subtrees.
Linus Torvalds1da177e2005-04-16 15:20:36 -070025 */
26
Randy Dunlap4fc268d2006-01-11 12:17:47 -080027#include <linux/capability.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070028#include <linux/errno.h>
29#include <linux/types.h>
30#include <linux/times.h>
31#include <linux/socket.h>
32#include <linux/sockios.h>
33#include <linux/net.h>
34#include <linux/route.h>
35#include <linux/netdevice.h>
36#include <linux/in6.h>
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +090037#include <linux/mroute6.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070038#include <linux/init.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070039#include <linux/if_arp.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070040#include <linux/proc_fs.h>
41#include <linux/seq_file.h>
Daniel Lezcano5b7c9312008-03-03 23:28:58 -080042#include <linux/nsproxy.h>
Tejun Heo5a0e3ad2010-03-24 17:04:11 +090043#include <linux/slab.h>
Eric W. Biederman457c4cb2007-09-12 12:01:34 +020044#include <net/net_namespace.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070045#include <net/snmp.h>
46#include <net/ipv6.h>
47#include <net/ip6_fib.h>
48#include <net/ip6_route.h>
49#include <net/ndisc.h>
50#include <net/addrconf.h>
51#include <net/tcp.h>
52#include <linux/rtnetlink.h>
53#include <net/dst.h>
54#include <net/xfrm.h>
Tom Tucker8d717402006-07-30 20:43:36 -070055#include <net/netevent.h>
Thomas Graf21713eb2006-08-15 00:35:24 -070056#include <net/netlink.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070057
58#include <asm/uaccess.h>
59
60#ifdef CONFIG_SYSCTL
61#include <linux/sysctl.h>
62#endif
63
64/* Set to 3 to get tracing. */
65#define RT6_DEBUG 2
66
67#if RT6_DEBUG >= 3
68#define RDBG(x) printk x
69#define RT6_TRACE(x...) printk(KERN_DEBUG x)
70#else
71#define RDBG(x)
72#define RT6_TRACE(x...) do { ; } while (0)
73#endif
74
YOSHIFUJI Hideaki519fbd82006-03-20 17:00:05 -080075#define CLONE_OFFLINK_ROUTE 0
Linus Torvalds1da177e2005-04-16 15:20:36 -070076
Linus Torvalds1da177e2005-04-16 15:20:36 -070077static struct rt6_info * ip6_rt_copy(struct rt6_info *ort);
78static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie);
79static struct dst_entry *ip6_negative_advice(struct dst_entry *);
80static void ip6_dst_destroy(struct dst_entry *);
81static void ip6_dst_ifdown(struct dst_entry *,
82 struct net_device *dev, int how);
Daniel Lezcano569d3642008-01-18 03:56:57 -080083static int ip6_dst_gc(struct dst_ops *ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -070084
85static int ip6_pkt_discard(struct sk_buff *skb);
86static int ip6_pkt_discard_out(struct sk_buff *skb);
87static void ip6_link_failure(struct sk_buff *skb);
88static void ip6_rt_update_pmtu(struct dst_entry *dst, u32 mtu);
89
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -080090#ifdef CONFIG_IPV6_ROUTE_INFO
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -080091static struct rt6_info *rt6_add_route_info(struct net *net,
92 struct in6_addr *prefix, int prefixlen,
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -080093 struct in6_addr *gwaddr, int ifindex,
94 unsigned pref);
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -080095static struct rt6_info *rt6_get_route_info(struct net *net,
96 struct in6_addr *prefix, int prefixlen,
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -080097 struct in6_addr *gwaddr, int ifindex);
98#endif
99
Daniel Lezcano9a7ec3a2008-03-04 13:48:53 -0800100static struct dst_ops ip6_dst_ops_template = {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700101 .family = AF_INET6,
Harvey Harrison09640e632009-02-01 00:45:17 -0800102 .protocol = cpu_to_be16(ETH_P_IPV6),
Linus Torvalds1da177e2005-04-16 15:20:36 -0700103 .gc = ip6_dst_gc,
104 .gc_thresh = 1024,
105 .check = ip6_dst_check,
106 .destroy = ip6_dst_destroy,
107 .ifdown = ip6_dst_ifdown,
108 .negative_advice = ip6_negative_advice,
109 .link_failure = ip6_link_failure,
110 .update_pmtu = ip6_rt_update_pmtu,
Herbert Xu1ac06e02008-05-20 14:32:14 -0700111 .local_out = __ip6_local_out,
Eric Dumazete2422972008-01-30 20:07:45 -0800112 .entries = ATOMIC_INIT(0),
Linus Torvalds1da177e2005-04-16 15:20:36 -0700113};
114
David S. Miller14e50e52007-05-24 18:17:54 -0700115static void ip6_rt_blackhole_update_pmtu(struct dst_entry *dst, u32 mtu)
116{
117}
118
119static struct dst_ops ip6_dst_blackhole_ops = {
120 .family = AF_INET6,
Harvey Harrison09640e632009-02-01 00:45:17 -0800121 .protocol = cpu_to_be16(ETH_P_IPV6),
David S. Miller14e50e52007-05-24 18:17:54 -0700122 .destroy = ip6_dst_destroy,
123 .check = ip6_dst_check,
124 .update_pmtu = ip6_rt_blackhole_update_pmtu,
Eric Dumazete2422972008-01-30 20:07:45 -0800125 .entries = ATOMIC_INIT(0),
David S. Miller14e50e52007-05-24 18:17:54 -0700126};
127
Daniel Lezcanobdb32892008-03-04 13:48:10 -0800128static struct rt6_info ip6_null_entry_template = {
Changli Gaod8d1f302010-06-10 23:31:35 -0700129 .dst = {
130 .__refcnt = ATOMIC_INIT(1),
131 .__use = 1,
132 .obsolete = -1,
133 .error = -ENETUNREACH,
134 .metrics = { [RTAX_HOPLIMIT - 1] = 255, },
135 .input = ip6_pkt_discard,
136 .output = ip6_pkt_discard_out,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700137 },
138 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
Jean-Mickael Guerin4f724272009-05-20 17:38:59 -0700139 .rt6i_protocol = RTPROT_KERNEL,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700140 .rt6i_metric = ~(u32) 0,
141 .rt6i_ref = ATOMIC_INIT(1),
142};
143
Thomas Graf101367c2006-08-04 03:39:02 -0700144#ifdef CONFIG_IPV6_MULTIPLE_TABLES
145
David S. Miller6723ab52006-10-18 21:20:57 -0700146static int ip6_pkt_prohibit(struct sk_buff *skb);
147static int ip6_pkt_prohibit_out(struct sk_buff *skb);
David S. Miller6723ab52006-10-18 21:20:57 -0700148
Adrian Bunk280a34c2008-04-21 02:29:32 -0700149static struct rt6_info ip6_prohibit_entry_template = {
Changli Gaod8d1f302010-06-10 23:31:35 -0700150 .dst = {
151 .__refcnt = ATOMIC_INIT(1),
152 .__use = 1,
153 .obsolete = -1,
154 .error = -EACCES,
155 .metrics = { [RTAX_HOPLIMIT - 1] = 255, },
156 .input = ip6_pkt_prohibit,
157 .output = ip6_pkt_prohibit_out,
Thomas Graf101367c2006-08-04 03:39:02 -0700158 },
159 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
Jean-Mickael Guerin4f724272009-05-20 17:38:59 -0700160 .rt6i_protocol = RTPROT_KERNEL,
Thomas Graf101367c2006-08-04 03:39:02 -0700161 .rt6i_metric = ~(u32) 0,
162 .rt6i_ref = ATOMIC_INIT(1),
163};
164
Daniel Lezcanobdb32892008-03-04 13:48:10 -0800165static struct rt6_info ip6_blk_hole_entry_template = {
Changli Gaod8d1f302010-06-10 23:31:35 -0700166 .dst = {
167 .__refcnt = ATOMIC_INIT(1),
168 .__use = 1,
169 .obsolete = -1,
170 .error = -EINVAL,
171 .metrics = { [RTAX_HOPLIMIT - 1] = 255, },
172 .input = dst_discard,
173 .output = dst_discard,
Thomas Graf101367c2006-08-04 03:39:02 -0700174 },
175 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
Jean-Mickael Guerin4f724272009-05-20 17:38:59 -0700176 .rt6i_protocol = RTPROT_KERNEL,
Thomas Graf101367c2006-08-04 03:39:02 -0700177 .rt6i_metric = ~(u32) 0,
178 .rt6i_ref = ATOMIC_INIT(1),
179};
180
181#endif
182
Linus Torvalds1da177e2005-04-16 15:20:36 -0700183/* allocate dst with ip6_dst_ops */
Benjamin Theryf2fc6a52008-03-04 13:49:23 -0800184static inline struct rt6_info *ip6_dst_alloc(struct dst_ops *ops)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700185{
Benjamin Theryf2fc6a52008-03-04 13:49:23 -0800186 return (struct rt6_info *)dst_alloc(ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700187}
188
189static void ip6_dst_destroy(struct dst_entry *dst)
190{
191 struct rt6_info *rt = (struct rt6_info *)dst;
192 struct inet6_dev *idev = rt->rt6i_idev;
193
194 if (idev != NULL) {
195 rt->rt6i_idev = NULL;
196 in6_dev_put(idev);
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900197 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700198}
199
200static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
201 int how)
202{
203 struct rt6_info *rt = (struct rt6_info *)dst;
204 struct inet6_dev *idev = rt->rt6i_idev;
Denis V. Lunev5a3e55d2007-12-07 00:38:10 -0800205 struct net_device *loopback_dev =
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900206 dev_net(dev)->loopback_dev;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700207
Denis V. Lunev5a3e55d2007-12-07 00:38:10 -0800208 if (dev != loopback_dev && idev != NULL && idev->dev == dev) {
209 struct inet6_dev *loopback_idev =
210 in6_dev_get(loopback_dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700211 if (loopback_idev != NULL) {
212 rt->rt6i_idev = loopback_idev;
213 in6_dev_put(idev);
214 }
215 }
216}
217
218static __inline__ int rt6_check_expired(const struct rt6_info *rt)
219{
220 return (rt->rt6i_flags & RTF_EXPIRES &&
221 time_after(jiffies, rt->rt6i_expires));
222}
223
Thomas Grafc71099a2006-08-04 23:20:06 -0700224static inline int rt6_need_strict(struct in6_addr *daddr)
225{
226 return (ipv6_addr_type(daddr) &
YOSHIFUJI Hideaki5ce83af2008-06-25 16:58:17 +0900227 (IPV6_ADDR_MULTICAST | IPV6_ADDR_LINKLOCAL | IPV6_ADDR_LOOPBACK));
Thomas Grafc71099a2006-08-04 23:20:06 -0700228}
229
Linus Torvalds1da177e2005-04-16 15:20:36 -0700230/*
Thomas Grafc71099a2006-08-04 23:20:06 -0700231 * Route lookup. Any table->tb6_lock is implied.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700232 */
233
Daniel Lezcano8ed67782008-03-04 13:48:30 -0800234static inline struct rt6_info *rt6_device_match(struct net *net,
235 struct rt6_info *rt,
YOSHIFUJI Hideakidd3abc42008-07-02 18:30:18 +0900236 struct in6_addr *saddr,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700237 int oif,
YOSHIFUJI Hideakid4208952008-06-27 20:14:54 -0700238 int flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700239{
240 struct rt6_info *local = NULL;
241 struct rt6_info *sprt;
242
YOSHIFUJI Hideakidd3abc42008-07-02 18:30:18 +0900243 if (!oif && ipv6_addr_any(saddr))
244 goto out;
245
Changli Gaod8d1f302010-06-10 23:31:35 -0700246 for (sprt = rt; sprt; sprt = sprt->dst.rt6_next) {
YOSHIFUJI Hideakidd3abc42008-07-02 18:30:18 +0900247 struct net_device *dev = sprt->rt6i_dev;
248
249 if (oif) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700250 if (dev->ifindex == oif)
251 return sprt;
252 if (dev->flags & IFF_LOOPBACK) {
253 if (sprt->rt6i_idev == NULL ||
254 sprt->rt6i_idev->dev->ifindex != oif) {
YOSHIFUJI Hideakid4208952008-06-27 20:14:54 -0700255 if (flags & RT6_LOOKUP_F_IFACE && oif)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700256 continue;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900257 if (local && (!oif ||
Linus Torvalds1da177e2005-04-16 15:20:36 -0700258 local->rt6i_idev->dev->ifindex == oif))
259 continue;
260 }
261 local = sprt;
262 }
YOSHIFUJI Hideakidd3abc42008-07-02 18:30:18 +0900263 } else {
264 if (ipv6_chk_addr(net, saddr, dev,
265 flags & RT6_LOOKUP_F_IFACE))
266 return sprt;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700267 }
YOSHIFUJI Hideakidd3abc42008-07-02 18:30:18 +0900268 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700269
YOSHIFUJI Hideakidd3abc42008-07-02 18:30:18 +0900270 if (oif) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700271 if (local)
272 return local;
273
YOSHIFUJI Hideakid4208952008-06-27 20:14:54 -0700274 if (flags & RT6_LOOKUP_F_IFACE)
Daniel Lezcano8ed67782008-03-04 13:48:30 -0800275 return net->ipv6.ip6_null_entry;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700276 }
YOSHIFUJI Hideakidd3abc42008-07-02 18:30:18 +0900277out:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700278 return rt;
279}
280
YOSHIFUJI Hideaki27097252006-03-20 17:05:13 -0800281#ifdef CONFIG_IPV6_ROUTER_PREF
282static void rt6_probe(struct rt6_info *rt)
283{
284 struct neighbour *neigh = rt ? rt->rt6i_nexthop : NULL;
285 /*
286 * Okay, this does not seem to be appropriate
287 * for now, however, we need to check if it
288 * is really so; aka Router Reachability Probing.
289 *
290 * Router Reachability Probe MUST be rate-limited
291 * to no more than one per minute.
292 */
293 if (!neigh || (neigh->nud_state & NUD_VALID))
294 return;
295 read_lock_bh(&neigh->lock);
296 if (!(neigh->nud_state & NUD_VALID) &&
YOSHIFUJI Hideaki52e163562006-03-20 17:05:47 -0800297 time_after(jiffies, neigh->updated + rt->rt6i_idev->cnf.rtr_probe_interval)) {
YOSHIFUJI Hideaki27097252006-03-20 17:05:13 -0800298 struct in6_addr mcaddr;
299 struct in6_addr *target;
300
301 neigh->updated = jiffies;
302 read_unlock_bh(&neigh->lock);
303
304 target = (struct in6_addr *)&neigh->primary_key;
305 addrconf_addr_solict_mult(target, &mcaddr);
306 ndisc_send_ns(rt->rt6i_dev, NULL, target, &mcaddr, NULL);
307 } else
308 read_unlock_bh(&neigh->lock);
309}
310#else
311static inline void rt6_probe(struct rt6_info *rt)
312{
YOSHIFUJI Hideaki27097252006-03-20 17:05:13 -0800313}
314#endif
315
Linus Torvalds1da177e2005-04-16 15:20:36 -0700316/*
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800317 * Default Router Selection (RFC 2461 6.3.6)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700318 */
Dave Jonesb6f99a22007-03-22 12:27:49 -0700319static inline int rt6_check_dev(struct rt6_info *rt, int oif)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700320{
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800321 struct net_device *dev = rt->rt6i_dev;
David S. Miller161980f2007-04-06 11:42:27 -0700322 if (!oif || dev->ifindex == oif)
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800323 return 2;
David S. Miller161980f2007-04-06 11:42:27 -0700324 if ((dev->flags & IFF_LOOPBACK) &&
325 rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif)
326 return 1;
327 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700328}
329
Dave Jonesb6f99a22007-03-22 12:27:49 -0700330static inline int rt6_check_neigh(struct rt6_info *rt)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700331{
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800332 struct neighbour *neigh = rt->rt6i_nexthop;
YOSHIFUJI Hideaki398bcbe2008-01-19 00:35:16 -0800333 int m;
YOSHIFUJI Hideaki4d0c5912006-05-26 13:23:41 -0700334 if (rt->rt6i_flags & RTF_NONEXTHOP ||
335 !(rt->rt6i_flags & RTF_GATEWAY))
336 m = 1;
337 else if (neigh) {
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800338 read_lock_bh(&neigh->lock);
339 if (neigh->nud_state & NUD_VALID)
YOSHIFUJI Hideaki4d0c5912006-05-26 13:23:41 -0700340 m = 2;
YOSHIFUJI Hideaki398bcbe2008-01-19 00:35:16 -0800341#ifdef CONFIG_IPV6_ROUTER_PREF
342 else if (neigh->nud_state & NUD_FAILED)
343 m = 0;
344#endif
345 else
YOSHIFUJI Hideakiea73ee22006-11-06 09:45:44 -0800346 m = 1;
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800347 read_unlock_bh(&neigh->lock);
YOSHIFUJI Hideaki398bcbe2008-01-19 00:35:16 -0800348 } else
349 m = 0;
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800350 return m;
351}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700352
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800353static int rt6_score_route(struct rt6_info *rt, int oif,
354 int strict)
355{
YOSHIFUJI Hideaki4d0c5912006-05-26 13:23:41 -0700356 int m, n;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900357
YOSHIFUJI Hideaki4d0c5912006-05-26 13:23:41 -0700358 m = rt6_check_dev(rt, oif);
YOSHIFUJI Hideaki77d16f42006-08-23 17:25:05 -0700359 if (!m && (strict & RT6_LOOKUP_F_IFACE))
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800360 return -1;
YOSHIFUJI Hideakiebacaaa2006-03-20 17:04:53 -0800361#ifdef CONFIG_IPV6_ROUTER_PREF
362 m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2;
363#endif
YOSHIFUJI Hideaki4d0c5912006-05-26 13:23:41 -0700364 n = rt6_check_neigh(rt);
YOSHIFUJI Hideaki557e92e2006-11-06 09:45:45 -0800365 if (!n && (strict & RT6_LOOKUP_F_REACHABLE))
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800366 return -1;
367 return m;
368}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700369
David S. Millerf11e6652007-03-24 20:36:25 -0700370static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict,
371 int *mpri, struct rt6_info *match)
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800372{
David S. Millerf11e6652007-03-24 20:36:25 -0700373 int m;
374
375 if (rt6_check_expired(rt))
376 goto out;
377
378 m = rt6_score_route(rt, oif, strict);
379 if (m < 0)
380 goto out;
381
382 if (m > *mpri) {
383 if (strict & RT6_LOOKUP_F_REACHABLE)
384 rt6_probe(match);
385 *mpri = m;
386 match = rt;
387 } else if (strict & RT6_LOOKUP_F_REACHABLE) {
388 rt6_probe(rt);
389 }
390
391out:
392 return match;
393}
394
395static struct rt6_info *find_rr_leaf(struct fib6_node *fn,
396 struct rt6_info *rr_head,
397 u32 metric, int oif, int strict)
398{
399 struct rt6_info *rt, *match;
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800400 int mpri = -1;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700401
David S. Millerf11e6652007-03-24 20:36:25 -0700402 match = NULL;
403 for (rt = rr_head; rt && rt->rt6i_metric == metric;
Changli Gaod8d1f302010-06-10 23:31:35 -0700404 rt = rt->dst.rt6_next)
David S. Millerf11e6652007-03-24 20:36:25 -0700405 match = find_match(rt, oif, strict, &mpri, match);
406 for (rt = fn->leaf; rt && rt != rr_head && rt->rt6i_metric == metric;
Changli Gaod8d1f302010-06-10 23:31:35 -0700407 rt = rt->dst.rt6_next)
David S. Millerf11e6652007-03-24 20:36:25 -0700408 match = find_match(rt, oif, strict, &mpri, match);
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800409
David S. Millerf11e6652007-03-24 20:36:25 -0700410 return match;
411}
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800412
David S. Millerf11e6652007-03-24 20:36:25 -0700413static struct rt6_info *rt6_select(struct fib6_node *fn, int oif, int strict)
414{
415 struct rt6_info *match, *rt0;
Daniel Lezcano8ed67782008-03-04 13:48:30 -0800416 struct net *net;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700417
David S. Millerf11e6652007-03-24 20:36:25 -0700418 RT6_TRACE("%s(fn->leaf=%p, oif=%d)\n",
Harvey Harrison0dc47872008-03-05 20:47:47 -0800419 __func__, fn->leaf, oif);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700420
David S. Millerf11e6652007-03-24 20:36:25 -0700421 rt0 = fn->rr_ptr;
422 if (!rt0)
423 fn->rr_ptr = rt0 = fn->leaf;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700424
David S. Millerf11e6652007-03-24 20:36:25 -0700425 match = find_rr_leaf(fn, rt0, rt0->rt6i_metric, oif, strict);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700426
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800427 if (!match &&
David S. Millerf11e6652007-03-24 20:36:25 -0700428 (strict & RT6_LOOKUP_F_REACHABLE)) {
Changli Gaod8d1f302010-06-10 23:31:35 -0700429 struct rt6_info *next = rt0->dst.rt6_next;
David S. Millerf11e6652007-03-24 20:36:25 -0700430
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800431 /* no entries matched; do round-robin */
David S. Millerf11e6652007-03-24 20:36:25 -0700432 if (!next || next->rt6i_metric != rt0->rt6i_metric)
433 next = fn->leaf;
434
435 if (next != rt0)
436 fn->rr_ptr = next;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700437 }
438
David S. Millerf11e6652007-03-24 20:36:25 -0700439 RT6_TRACE("%s() => %p\n",
Harvey Harrison0dc47872008-03-05 20:47:47 -0800440 __func__, match);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700441
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900442 net = dev_net(rt0->rt6i_dev);
Daniel Lezcano8ed67782008-03-04 13:48:30 -0800443 return (match ? match : net->ipv6.ip6_null_entry);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700444}
445
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800446#ifdef CONFIG_IPV6_ROUTE_INFO
447int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
448 struct in6_addr *gwaddr)
449{
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900450 struct net *net = dev_net(dev);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800451 struct route_info *rinfo = (struct route_info *) opt;
452 struct in6_addr prefix_buf, *prefix;
453 unsigned int pref;
YOSHIFUJI Hideaki4bed72e2008-05-27 17:37:49 +0900454 unsigned long lifetime;
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800455 struct rt6_info *rt;
456
457 if (len < sizeof(struct route_info)) {
458 return -EINVAL;
459 }
460
461 /* Sanity check for prefix_len and length */
462 if (rinfo->length > 3) {
463 return -EINVAL;
464 } else if (rinfo->prefix_len > 128) {
465 return -EINVAL;
466 } else if (rinfo->prefix_len > 64) {
467 if (rinfo->length < 2) {
468 return -EINVAL;
469 }
470 } else if (rinfo->prefix_len > 0) {
471 if (rinfo->length < 1) {
472 return -EINVAL;
473 }
474 }
475
476 pref = rinfo->route_pref;
477 if (pref == ICMPV6_ROUTER_PREF_INVALID)
Jens Rosenboom3933fc92009-09-10 06:25:11 +0000478 return -EINVAL;
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800479
YOSHIFUJI Hideaki4bed72e2008-05-27 17:37:49 +0900480 lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800481
482 if (rinfo->length == 3)
483 prefix = (struct in6_addr *)rinfo->prefix;
484 else {
485 /* this function is safe */
486 ipv6_addr_prefix(&prefix_buf,
487 (struct in6_addr *)rinfo->prefix,
488 rinfo->prefix_len);
489 prefix = &prefix_buf;
490 }
491
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -0800492 rt = rt6_get_route_info(net, prefix, rinfo->prefix_len, gwaddr,
493 dev->ifindex);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800494
495 if (rt && !lifetime) {
Thomas Grafe0a1ad732006-08-22 00:00:21 -0700496 ip6_del_rt(rt);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800497 rt = NULL;
498 }
499
500 if (!rt && lifetime)
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -0800501 rt = rt6_add_route_info(net, prefix, rinfo->prefix_len, gwaddr, dev->ifindex,
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800502 pref);
503 else if (rt)
504 rt->rt6i_flags = RTF_ROUTEINFO |
505 (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref);
506
507 if (rt) {
YOSHIFUJI Hideaki4bed72e2008-05-27 17:37:49 +0900508 if (!addrconf_finite_timeout(lifetime)) {
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800509 rt->rt6i_flags &= ~RTF_EXPIRES;
510 } else {
511 rt->rt6i_expires = jiffies + HZ * lifetime;
512 rt->rt6i_flags |= RTF_EXPIRES;
513 }
Changli Gaod8d1f302010-06-10 23:31:35 -0700514 dst_release(&rt->dst);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800515 }
516 return 0;
517}
518#endif
519
Daniel Lezcano8ed67782008-03-04 13:48:30 -0800520#define BACKTRACK(__net, saddr) \
YOSHIFUJI Hideaki982f56f2006-08-23 17:22:39 -0700521do { \
Daniel Lezcano8ed67782008-03-04 13:48:30 -0800522 if (rt == __net->ipv6.ip6_null_entry) { \
YOSHIFUJI Hideaki982f56f2006-08-23 17:22:39 -0700523 struct fib6_node *pn; \
Ville Nuorvalae0eda7b2006-10-16 22:11:11 -0700524 while (1) { \
YOSHIFUJI Hideaki982f56f2006-08-23 17:22:39 -0700525 if (fn->fn_flags & RTN_TL_ROOT) \
526 goto out; \
527 pn = fn->parent; \
528 if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn) \
Kim Nordlund8bce65b2006-12-13 16:38:29 -0800529 fn = fib6_lookup(FIB6_SUBTREE(pn), NULL, saddr); \
YOSHIFUJI Hideaki982f56f2006-08-23 17:22:39 -0700530 else \
531 fn = pn; \
532 if (fn->fn_flags & RTN_RTINFO) \
533 goto restart; \
Thomas Grafc71099a2006-08-04 23:20:06 -0700534 } \
Thomas Grafc71099a2006-08-04 23:20:06 -0700535 } \
YOSHIFUJI Hideaki982f56f2006-08-23 17:22:39 -0700536} while(0)
Thomas Grafc71099a2006-08-04 23:20:06 -0700537
Daniel Lezcano8ed67782008-03-04 13:48:30 -0800538static struct rt6_info *ip6_pol_route_lookup(struct net *net,
539 struct fib6_table *table,
Thomas Grafc71099a2006-08-04 23:20:06 -0700540 struct flowi *fl, int flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700541{
542 struct fib6_node *fn;
543 struct rt6_info *rt;
544
Thomas Grafc71099a2006-08-04 23:20:06 -0700545 read_lock_bh(&table->tb6_lock);
546 fn = fib6_lookup(&table->tb6_root, &fl->fl6_dst, &fl->fl6_src);
547restart:
548 rt = fn->leaf;
YOSHIFUJI Hideakidd3abc42008-07-02 18:30:18 +0900549 rt = rt6_device_match(net, rt, &fl->fl6_src, fl->oif, flags);
Daniel Lezcano8ed67782008-03-04 13:48:30 -0800550 BACKTRACK(net, &fl->fl6_src);
Thomas Grafc71099a2006-08-04 23:20:06 -0700551out:
Changli Gaod8d1f302010-06-10 23:31:35 -0700552 dst_use(&rt->dst, jiffies);
Thomas Grafc71099a2006-08-04 23:20:06 -0700553 read_unlock_bh(&table->tb6_lock);
Thomas Grafc71099a2006-08-04 23:20:06 -0700554 return rt;
555
556}
557
YOSHIFUJI Hideaki9acd9f32008-04-10 15:42:10 +0900558struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
559 const struct in6_addr *saddr, int oif, int strict)
Thomas Grafc71099a2006-08-04 23:20:06 -0700560{
561 struct flowi fl = {
562 .oif = oif,
563 .nl_u = {
564 .ip6_u = {
565 .daddr = *daddr,
Thomas Grafc71099a2006-08-04 23:20:06 -0700566 },
567 },
568 };
569 struct dst_entry *dst;
YOSHIFUJI Hideaki77d16f42006-08-23 17:25:05 -0700570 int flags = strict ? RT6_LOOKUP_F_IFACE : 0;
Thomas Grafc71099a2006-08-04 23:20:06 -0700571
Thomas Grafadaa70b2006-10-13 15:01:03 -0700572 if (saddr) {
573 memcpy(&fl.fl6_src, saddr, sizeof(*saddr));
574 flags |= RT6_LOOKUP_F_HAS_SADDR;
575 }
576
Daniel Lezcano606a2b42008-03-04 13:45:59 -0800577 dst = fib6_rule_lookup(net, &fl, flags, ip6_pol_route_lookup);
Thomas Grafc71099a2006-08-04 23:20:06 -0700578 if (dst->error == 0)
579 return (struct rt6_info *) dst;
580
581 dst_release(dst);
582
Linus Torvalds1da177e2005-04-16 15:20:36 -0700583 return NULL;
584}
585
YOSHIFUJI Hideaki71590392007-02-22 22:05:40 +0900586EXPORT_SYMBOL(rt6_lookup);
587
Thomas Grafc71099a2006-08-04 23:20:06 -0700588/* ip6_ins_rt is called with FREE table->tb6_lock.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700589 It takes new route entry, the addition fails by any reason the
590 route is freed. In any case, if caller does not hold it, it may
591 be destroyed.
592 */
593
Thomas Graf86872cb2006-08-22 00:01:08 -0700594static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700595{
596 int err;
Thomas Grafc71099a2006-08-04 23:20:06 -0700597 struct fib6_table *table;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700598
Thomas Grafc71099a2006-08-04 23:20:06 -0700599 table = rt->rt6i_table;
600 write_lock_bh(&table->tb6_lock);
Thomas Graf86872cb2006-08-22 00:01:08 -0700601 err = fib6_add(&table->tb6_root, rt, info);
Thomas Grafc71099a2006-08-04 23:20:06 -0700602 write_unlock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700603
604 return err;
605}
606
Thomas Graf40e22e82006-08-22 00:00:45 -0700607int ip6_ins_rt(struct rt6_info *rt)
608{
Denis V. Lunev4d1169c2008-01-10 03:26:13 -0800609 struct nl_info info = {
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900610 .nl_net = dev_net(rt->rt6i_dev),
Denis V. Lunev4d1169c2008-01-10 03:26:13 -0800611 };
Denis V. Lunev528c4ce2007-12-13 09:45:12 -0800612 return __ip6_ins_rt(rt, &info);
Thomas Graf40e22e82006-08-22 00:00:45 -0700613}
614
YOSHIFUJI Hideaki95a9a5b2006-03-20 16:55:51 -0800615static struct rt6_info *rt6_alloc_cow(struct rt6_info *ort, struct in6_addr *daddr,
616 struct in6_addr *saddr)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700617{
Linus Torvalds1da177e2005-04-16 15:20:36 -0700618 struct rt6_info *rt;
619
620 /*
621 * Clone the route.
622 */
623
624 rt = ip6_rt_copy(ort);
625
626 if (rt) {
David S. Miller14deae42009-01-04 16:04:39 -0800627 struct neighbour *neigh;
628 int attempts = !in_softirq();
629
YOSHIFUJI Hideaki58c4fb82005-12-21 22:56:42 +0900630 if (!(rt->rt6i_flags&RTF_GATEWAY)) {
631 if (rt->rt6i_dst.plen != 128 &&
632 ipv6_addr_equal(&rt->rt6i_dst.addr, daddr))
633 rt->rt6i_flags |= RTF_ANYCAST;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700634 ipv6_addr_copy(&rt->rt6i_gateway, daddr);
YOSHIFUJI Hideaki58c4fb82005-12-21 22:56:42 +0900635 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700636
YOSHIFUJI Hideaki58c4fb82005-12-21 22:56:42 +0900637 ipv6_addr_copy(&rt->rt6i_dst.addr, daddr);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700638 rt->rt6i_dst.plen = 128;
639 rt->rt6i_flags |= RTF_CACHE;
Changli Gaod8d1f302010-06-10 23:31:35 -0700640 rt->dst.flags |= DST_HOST;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700641
642#ifdef CONFIG_IPV6_SUBTREES
643 if (rt->rt6i_src.plen && saddr) {
644 ipv6_addr_copy(&rt->rt6i_src.addr, saddr);
645 rt->rt6i_src.plen = 128;
646 }
647#endif
648
David S. Miller14deae42009-01-04 16:04:39 -0800649 retry:
650 neigh = ndisc_get_neigh(rt->rt6i_dev, &rt->rt6i_gateway);
651 if (IS_ERR(neigh)) {
652 struct net *net = dev_net(rt->rt6i_dev);
653 int saved_rt_min_interval =
654 net->ipv6.sysctl.ip6_rt_gc_min_interval;
655 int saved_rt_elasticity =
656 net->ipv6.sysctl.ip6_rt_gc_elasticity;
657
658 if (attempts-- > 0) {
659 net->ipv6.sysctl.ip6_rt_gc_elasticity = 1;
660 net->ipv6.sysctl.ip6_rt_gc_min_interval = 0;
661
Alexey Dobriyan86393e52009-08-29 01:34:49 +0000662 ip6_dst_gc(&net->ipv6.ip6_dst_ops);
David S. Miller14deae42009-01-04 16:04:39 -0800663
664 net->ipv6.sysctl.ip6_rt_gc_elasticity =
665 saved_rt_elasticity;
666 net->ipv6.sysctl.ip6_rt_gc_min_interval =
667 saved_rt_min_interval;
668 goto retry;
669 }
670
671 if (net_ratelimit())
672 printk(KERN_WARNING
Ulrich Weber7e1b33e2010-09-27 15:02:18 -0700673 "ipv6: Neighbour table overflow.\n");
Changli Gaod8d1f302010-06-10 23:31:35 -0700674 dst_free(&rt->dst);
David S. Miller14deae42009-01-04 16:04:39 -0800675 return NULL;
676 }
677 rt->rt6i_nexthop = neigh;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700678
YOSHIFUJI Hideaki95a9a5b2006-03-20 16:55:51 -0800679 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700680
YOSHIFUJI Hideaki95a9a5b2006-03-20 16:55:51 -0800681 return rt;
682}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700683
YOSHIFUJI Hideaki299d9932006-03-20 16:58:32 -0800684static struct rt6_info *rt6_alloc_clone(struct rt6_info *ort, struct in6_addr *daddr)
685{
686 struct rt6_info *rt = ip6_rt_copy(ort);
687 if (rt) {
688 ipv6_addr_copy(&rt->rt6i_dst.addr, daddr);
689 rt->rt6i_dst.plen = 128;
690 rt->rt6i_flags |= RTF_CACHE;
Changli Gaod8d1f302010-06-10 23:31:35 -0700691 rt->dst.flags |= DST_HOST;
YOSHIFUJI Hideaki299d9932006-03-20 16:58:32 -0800692 rt->rt6i_nexthop = neigh_clone(ort->rt6i_nexthop);
693 }
694 return rt;
695}
696
Daniel Lezcano8ed67782008-03-04 13:48:30 -0800697static struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table, int oif,
698 struct flowi *fl, int flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700699{
700 struct fib6_node *fn;
YOSHIFUJI Hideaki519fbd82006-03-20 17:00:05 -0800701 struct rt6_info *rt, *nrt;
Thomas Grafc71099a2006-08-04 23:20:06 -0700702 int strict = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700703 int attempts = 3;
YOSHIFUJI Hideaki519fbd82006-03-20 17:00:05 -0800704 int err;
YOSHIFUJI Hideaki53b79972008-07-19 22:35:03 -0700705 int reachable = net->ipv6.devconf_all->forwarding ? 0 : RT6_LOOKUP_F_REACHABLE;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700706
YOSHIFUJI Hideaki77d16f42006-08-23 17:25:05 -0700707 strict |= flags & RT6_LOOKUP_F_IFACE;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700708
709relookup:
Thomas Grafc71099a2006-08-04 23:20:06 -0700710 read_lock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700711
YOSHIFUJI Hideaki8238dd02006-03-20 17:04:35 -0800712restart_2:
Thomas Grafc71099a2006-08-04 23:20:06 -0700713 fn = fib6_lookup(&table->tb6_root, &fl->fl6_dst, &fl->fl6_src);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700714
715restart:
Pavel Emelyanov4acad722007-10-15 13:02:51 -0700716 rt = rt6_select(fn, oif, strict | reachable);
Daniel Lezcano8ed67782008-03-04 13:48:30 -0800717
718 BACKTRACK(net, &fl->fl6_src);
719 if (rt == net->ipv6.ip6_null_entry ||
YOSHIFUJI Hideaki8238dd02006-03-20 17:04:35 -0800720 rt->rt6i_flags & RTF_CACHE)
YOSHIFUJI Hideaki1ddef0442006-03-20 17:01:24 -0800721 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700722
Changli Gaod8d1f302010-06-10 23:31:35 -0700723 dst_hold(&rt->dst);
Thomas Grafc71099a2006-08-04 23:20:06 -0700724 read_unlock_bh(&table->tb6_lock);
YOSHIFUJI Hideakifb9de912006-03-20 16:59:08 -0800725
YOSHIFUJI Hideaki519fbd82006-03-20 17:00:05 -0800726 if (!rt->rt6i_nexthop && !(rt->rt6i_flags & RTF_NONEXTHOP))
YOSHIFUJI Hideakie40cf352006-03-20 16:59:27 -0800727 nrt = rt6_alloc_cow(rt, &fl->fl6_dst, &fl->fl6_src);
YOSHIFUJI Hideaki519fbd82006-03-20 17:00:05 -0800728 else {
729#if CLONE_OFFLINK_ROUTE
730 nrt = rt6_alloc_clone(rt, &fl->fl6_dst);
731#else
732 goto out2;
733#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700734 }
YOSHIFUJI Hideakie40cf352006-03-20 16:59:27 -0800735
Changli Gaod8d1f302010-06-10 23:31:35 -0700736 dst_release(&rt->dst);
Daniel Lezcano8ed67782008-03-04 13:48:30 -0800737 rt = nrt ? : net->ipv6.ip6_null_entry;
YOSHIFUJI Hideaki519fbd82006-03-20 17:00:05 -0800738
Changli Gaod8d1f302010-06-10 23:31:35 -0700739 dst_hold(&rt->dst);
YOSHIFUJI Hideaki519fbd82006-03-20 17:00:05 -0800740 if (nrt) {
Thomas Graf40e22e82006-08-22 00:00:45 -0700741 err = ip6_ins_rt(nrt);
YOSHIFUJI Hideaki519fbd82006-03-20 17:00:05 -0800742 if (!err)
743 goto out2;
744 }
745
746 if (--attempts <= 0)
747 goto out2;
748
749 /*
Thomas Grafc71099a2006-08-04 23:20:06 -0700750 * Race condition! In the gap, when table->tb6_lock was
YOSHIFUJI Hideaki519fbd82006-03-20 17:00:05 -0800751 * released someone could insert this route. Relookup.
752 */
Changli Gaod8d1f302010-06-10 23:31:35 -0700753 dst_release(&rt->dst);
YOSHIFUJI Hideaki519fbd82006-03-20 17:00:05 -0800754 goto relookup;
755
756out:
YOSHIFUJI Hideaki8238dd02006-03-20 17:04:35 -0800757 if (reachable) {
758 reachable = 0;
759 goto restart_2;
760 }
Changli Gaod8d1f302010-06-10 23:31:35 -0700761 dst_hold(&rt->dst);
Thomas Grafc71099a2006-08-04 23:20:06 -0700762 read_unlock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700763out2:
Changli Gaod8d1f302010-06-10 23:31:35 -0700764 rt->dst.lastuse = jiffies;
765 rt->dst.__use++;
Thomas Grafc71099a2006-08-04 23:20:06 -0700766
767 return rt;
768}
769
Daniel Lezcano8ed67782008-03-04 13:48:30 -0800770static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table,
Pavel Emelyanov4acad722007-10-15 13:02:51 -0700771 struct flowi *fl, int flags)
772{
Daniel Lezcano8ed67782008-03-04 13:48:30 -0800773 return ip6_pol_route(net, table, fl->iif, fl, flags);
Pavel Emelyanov4acad722007-10-15 13:02:51 -0700774}
775
Thomas Grafc71099a2006-08-04 23:20:06 -0700776void ip6_route_input(struct sk_buff *skb)
777{
Arnaldo Carvalho de Melo0660e032007-04-25 17:54:47 -0700778 struct ipv6hdr *iph = ipv6_hdr(skb);
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900779 struct net *net = dev_net(skb->dev);
Thomas Grafadaa70b2006-10-13 15:01:03 -0700780 int flags = RT6_LOOKUP_F_HAS_SADDR;
Thomas Grafc71099a2006-08-04 23:20:06 -0700781 struct flowi fl = {
782 .iif = skb->dev->ifindex,
783 .nl_u = {
784 .ip6_u = {
785 .daddr = iph->daddr,
786 .saddr = iph->saddr,
Al Viro90bcaf72006-11-08 00:25:17 -0800787 .flowlabel = (* (__be32 *) iph)&IPV6_FLOWINFO_MASK,
Thomas Grafc71099a2006-08-04 23:20:06 -0700788 },
789 },
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900790 .mark = skb->mark,
Thomas Grafc71099a2006-08-04 23:20:06 -0700791 .proto = iph->nexthdr,
792 };
Thomas Grafadaa70b2006-10-13 15:01:03 -0700793
Thomas Goff1d6e55f2009-01-27 22:39:59 -0800794 if (rt6_need_strict(&iph->daddr) && skb->dev->type != ARPHRD_PIMREG)
Thomas Grafadaa70b2006-10-13 15:01:03 -0700795 flags |= RT6_LOOKUP_F_IFACE;
Thomas Grafc71099a2006-08-04 23:20:06 -0700796
Eric Dumazetadf30902009-06-02 05:19:30 +0000797 skb_dst_set(skb, fib6_rule_lookup(net, &fl, flags, ip6_pol_route_input));
Thomas Grafc71099a2006-08-04 23:20:06 -0700798}
799
Daniel Lezcano8ed67782008-03-04 13:48:30 -0800800static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table,
Thomas Grafc71099a2006-08-04 23:20:06 -0700801 struct flowi *fl, int flags)
802{
Daniel Lezcano8ed67782008-03-04 13:48:30 -0800803 return ip6_pol_route(net, table, fl->oif, fl, flags);
Thomas Grafc71099a2006-08-04 23:20:06 -0700804}
805
Daniel Lezcano4591db42008-03-05 10:48:10 -0800806struct dst_entry * ip6_route_output(struct net *net, struct sock *sk,
807 struct flowi *fl)
Thomas Grafc71099a2006-08-04 23:20:06 -0700808{
809 int flags = 0;
810
Brian Haley6057fd72010-05-28 23:02:35 -0700811 if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl->fl6_dst))
YOSHIFUJI Hideaki77d16f42006-08-23 17:25:05 -0700812 flags |= RT6_LOOKUP_F_IFACE;
Thomas Grafc71099a2006-08-04 23:20:06 -0700813
Thomas Grafadaa70b2006-10-13 15:01:03 -0700814 if (!ipv6_addr_any(&fl->fl6_src))
815 flags |= RT6_LOOKUP_F_HAS_SADDR;
YOSHIFUJI Hideaki / 吉藤英明0c9a2ac2010-03-07 00:14:44 +0000816 else if (sk)
817 flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs);
Thomas Grafadaa70b2006-10-13 15:01:03 -0700818
Daniel Lezcano4591db42008-03-05 10:48:10 -0800819 return fib6_rule_lookup(net, fl, flags, ip6_pol_route_output);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700820}
821
YOSHIFUJI Hideaki71590392007-02-22 22:05:40 +0900822EXPORT_SYMBOL(ip6_route_output);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700823
David S. Miller14e50e52007-05-24 18:17:54 -0700824int ip6_dst_blackhole(struct sock *sk, struct dst_entry **dstp, struct flowi *fl)
825{
826 struct rt6_info *ort = (struct rt6_info *) *dstp;
827 struct rt6_info *rt = (struct rt6_info *)
828 dst_alloc(&ip6_dst_blackhole_ops);
829 struct dst_entry *new = NULL;
830
831 if (rt) {
Changli Gaod8d1f302010-06-10 23:31:35 -0700832 new = &rt->dst;
David S. Miller14e50e52007-05-24 18:17:54 -0700833
834 atomic_set(&new->__refcnt, 1);
835 new->__use = 1;
Herbert Xu352e5122007-11-13 21:34:06 -0800836 new->input = dst_discard;
837 new->output = dst_discard;
David S. Miller14e50e52007-05-24 18:17:54 -0700838
Changli Gaod8d1f302010-06-10 23:31:35 -0700839 memcpy(new->metrics, ort->dst.metrics, RTAX_MAX*sizeof(u32));
840 new->dev = ort->dst.dev;
David S. Miller14e50e52007-05-24 18:17:54 -0700841 if (new->dev)
842 dev_hold(new->dev);
843 rt->rt6i_idev = ort->rt6i_idev;
844 if (rt->rt6i_idev)
845 in6_dev_hold(rt->rt6i_idev);
846 rt->rt6i_expires = 0;
847
848 ipv6_addr_copy(&rt->rt6i_gateway, &ort->rt6i_gateway);
849 rt->rt6i_flags = ort->rt6i_flags & ~RTF_EXPIRES;
850 rt->rt6i_metric = 0;
851
852 memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key));
853#ifdef CONFIG_IPV6_SUBTREES
854 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
855#endif
856
857 dst_free(new);
858 }
859
860 dst_release(*dstp);
861 *dstp = new;
862 return (new ? 0 : -ENOMEM);
863}
864EXPORT_SYMBOL_GPL(ip6_dst_blackhole);
865
Linus Torvalds1da177e2005-04-16 15:20:36 -0700866/*
867 * Destination cache support functions
868 */
869
870static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie)
871{
872 struct rt6_info *rt;
873
874 rt = (struct rt6_info *) dst;
875
Herbert Xu10414442010-03-18 23:00:22 +0000876 if (rt->rt6i_node && (rt->rt6i_node->fn_sernum == cookie))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700877 return dst;
878
879 return NULL;
880}
881
882static struct dst_entry *ip6_negative_advice(struct dst_entry *dst)
883{
884 struct rt6_info *rt = (struct rt6_info *) dst;
885
886 if (rt) {
YOSHIFUJI Hideaki / 吉藤英明54c1a852010-03-28 07:15:45 +0000887 if (rt->rt6i_flags & RTF_CACHE) {
888 if (rt6_check_expired(rt)) {
889 ip6_del_rt(rt);
890 dst = NULL;
891 }
892 } else {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700893 dst_release(dst);
YOSHIFUJI Hideaki / 吉藤英明54c1a852010-03-28 07:15:45 +0000894 dst = NULL;
895 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700896 }
YOSHIFUJI Hideaki / 吉藤英明54c1a852010-03-28 07:15:45 +0000897 return dst;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700898}
899
900static void ip6_link_failure(struct sk_buff *skb)
901{
902 struct rt6_info *rt;
903
Alexey Dobriyan3ffe5332010-02-18 08:25:24 +0000904 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700905
Eric Dumazetadf30902009-06-02 05:19:30 +0000906 rt = (struct rt6_info *) skb_dst(skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700907 if (rt) {
908 if (rt->rt6i_flags&RTF_CACHE) {
Changli Gaod8d1f302010-06-10 23:31:35 -0700909 dst_set_expires(&rt->dst, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700910 rt->rt6i_flags |= RTF_EXPIRES;
911 } else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT))
912 rt->rt6i_node->fn_sernum = -1;
913 }
914}
915
916static void ip6_rt_update_pmtu(struct dst_entry *dst, u32 mtu)
917{
918 struct rt6_info *rt6 = (struct rt6_info*)dst;
919
920 if (mtu < dst_mtu(dst) && rt6->rt6i_dst.plen == 128) {
921 rt6->rt6i_flags |= RTF_MODIFIED;
922 if (mtu < IPV6_MIN_MTU) {
923 mtu = IPV6_MIN_MTU;
924 dst->metrics[RTAX_FEATURES-1] |= RTAX_FEATURE_ALLFRAG;
925 }
926 dst->metrics[RTAX_MTU-1] = mtu;
Tom Tucker8d717402006-07-30 20:43:36 -0700927 call_netevent_notifiers(NETEVENT_PMTU_UPDATE, dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700928 }
929}
930
Linus Torvalds1da177e2005-04-16 15:20:36 -0700931static int ipv6_get_mtu(struct net_device *dev);
932
Daniel Lezcano55786892008-03-04 13:47:47 -0800933static inline unsigned int ipv6_advmss(struct net *net, unsigned int mtu)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700934{
935 mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr);
936
Daniel Lezcano55786892008-03-04 13:47:47 -0800937 if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss)
938 mtu = net->ipv6.sysctl.ip6_rt_min_advmss;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700939
940 /*
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900941 * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and
942 * corresponding MSS is IPV6_MAXPLEN - tcp_header_size.
943 * IPV6_MAXPLEN is also valid and means: "any MSS,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700944 * rely only on pmtu discovery"
945 */
946 if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr))
947 mtu = IPV6_MAXPLEN;
948 return mtu;
949}
950
YOSHIFUJI Hideaki3b009442007-12-06 16:11:48 -0800951static struct dst_entry *icmp6_dst_gc_list;
952static DEFINE_SPINLOCK(icmp6_dst_lock);
Thomas Graf5d0bbee2006-08-04 03:37:36 -0700953
YOSHIFUJI Hideaki3b009442007-12-06 16:11:48 -0800954struct dst_entry *icmp6_dst_alloc(struct net_device *dev,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700955 struct neighbour *neigh,
YOSHIFUJI Hideaki9acd9f32008-04-10 15:42:10 +0900956 const struct in6_addr *addr)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700957{
958 struct rt6_info *rt;
959 struct inet6_dev *idev = in6_dev_get(dev);
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900960 struct net *net = dev_net(dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700961
962 if (unlikely(idev == NULL))
963 return NULL;
964
Alexey Dobriyan86393e52009-08-29 01:34:49 +0000965 rt = ip6_dst_alloc(&net->ipv6.ip6_dst_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700966 if (unlikely(rt == NULL)) {
967 in6_dev_put(idev);
968 goto out;
969 }
970
971 dev_hold(dev);
972 if (neigh)
973 neigh_hold(neigh);
David S. Miller14deae42009-01-04 16:04:39 -0800974 else {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700975 neigh = ndisc_get_neigh(dev, addr);
David S. Miller14deae42009-01-04 16:04:39 -0800976 if (IS_ERR(neigh))
977 neigh = NULL;
978 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700979
980 rt->rt6i_dev = dev;
981 rt->rt6i_idev = idev;
982 rt->rt6i_nexthop = neigh;
Changli Gaod8d1f302010-06-10 23:31:35 -0700983 atomic_set(&rt->dst.__refcnt, 1);
984 rt->dst.metrics[RTAX_HOPLIMIT-1] = 255;
985 rt->dst.metrics[RTAX_MTU-1] = ipv6_get_mtu(rt->rt6i_dev);
986 rt->dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(net, dst_mtu(&rt->dst));
987 rt->dst.output = ip6_output;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700988
989#if 0 /* there's no chance to use these for ndisc */
Changli Gaod8d1f302010-06-10 23:31:35 -0700990 rt->dst.flags = ipv6_addr_type(addr) & IPV6_ADDR_UNICAST
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900991 ? DST_HOST
Linus Torvalds1da177e2005-04-16 15:20:36 -0700992 : 0;
993 ipv6_addr_copy(&rt->rt6i_dst.addr, addr);
994 rt->rt6i_dst.plen = 128;
995#endif
996
YOSHIFUJI Hideaki3b009442007-12-06 16:11:48 -0800997 spin_lock_bh(&icmp6_dst_lock);
Changli Gaod8d1f302010-06-10 23:31:35 -0700998 rt->dst.next = icmp6_dst_gc_list;
999 icmp6_dst_gc_list = &rt->dst;
YOSHIFUJI Hideaki3b009442007-12-06 16:11:48 -08001000 spin_unlock_bh(&icmp6_dst_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001001
Daniel Lezcano55786892008-03-04 13:47:47 -08001002 fib6_force_start_gc(net);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001003
1004out:
Changli Gaod8d1f302010-06-10 23:31:35 -07001005 return &rt->dst;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001006}
1007
Stephen Hemminger3d0f24a2008-07-22 14:35:50 -07001008int icmp6_dst_gc(void)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001009{
1010 struct dst_entry *dst, *next, **pprev;
Stephen Hemminger3d0f24a2008-07-22 14:35:50 -07001011 int more = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001012
1013 next = NULL;
Thomas Graf5d0bbee2006-08-04 03:37:36 -07001014
YOSHIFUJI Hideaki3b009442007-12-06 16:11:48 -08001015 spin_lock_bh(&icmp6_dst_lock);
1016 pprev = &icmp6_dst_gc_list;
Thomas Graf5d0bbee2006-08-04 03:37:36 -07001017
Linus Torvalds1da177e2005-04-16 15:20:36 -07001018 while ((dst = *pprev) != NULL) {
1019 if (!atomic_read(&dst->__refcnt)) {
1020 *pprev = dst->next;
1021 dst_free(dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001022 } else {
1023 pprev = &dst->next;
Stephen Hemminger3d0f24a2008-07-22 14:35:50 -07001024 ++more;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001025 }
1026 }
1027
YOSHIFUJI Hideaki3b009442007-12-06 16:11:48 -08001028 spin_unlock_bh(&icmp6_dst_lock);
Thomas Graf5d0bbee2006-08-04 03:37:36 -07001029
Stephen Hemminger3d0f24a2008-07-22 14:35:50 -07001030 return more;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001031}
1032
David S. Miller1e493d12008-09-10 17:27:15 -07001033static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg),
1034 void *arg)
1035{
1036 struct dst_entry *dst, **pprev;
1037
1038 spin_lock_bh(&icmp6_dst_lock);
1039 pprev = &icmp6_dst_gc_list;
1040 while ((dst = *pprev) != NULL) {
1041 struct rt6_info *rt = (struct rt6_info *) dst;
1042 if (func(rt, arg)) {
1043 *pprev = dst->next;
1044 dst_free(dst);
1045 } else {
1046 pprev = &dst->next;
1047 }
1048 }
1049 spin_unlock_bh(&icmp6_dst_lock);
1050}
1051
Daniel Lezcano569d3642008-01-18 03:56:57 -08001052static int ip6_dst_gc(struct dst_ops *ops)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001053{
Linus Torvalds1da177e2005-04-16 15:20:36 -07001054 unsigned long now = jiffies;
Alexey Dobriyan86393e52009-08-29 01:34:49 +00001055 struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops);
Daniel Lezcano7019b782008-03-04 13:50:14 -08001056 int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval;
1057 int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size;
1058 int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity;
1059 int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout;
1060 unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001061
Daniel Lezcano7019b782008-03-04 13:50:14 -08001062 if (time_after(rt_last_gc + rt_min_interval, now) &&
1063 atomic_read(&ops->entries) <= rt_max_size)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001064 goto out;
1065
Benjamin Thery6891a342008-03-04 13:49:47 -08001066 net->ipv6.ip6_rt_gc_expire++;
1067 fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net);
1068 net->ipv6.ip6_rt_last_gc = now;
Daniel Lezcano7019b782008-03-04 13:50:14 -08001069 if (atomic_read(&ops->entries) < ops->gc_thresh)
1070 net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001071out:
Daniel Lezcano7019b782008-03-04 13:50:14 -08001072 net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity;
1073 return (atomic_read(&ops->entries) > rt_max_size);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001074}
1075
1076/* Clean host part of a prefix. Not necessary in radix tree,
1077 but results in cleaner routing tables.
1078
1079 Remove it only when all the things will work!
1080 */
1081
1082static int ipv6_get_mtu(struct net_device *dev)
1083{
1084 int mtu = IPV6_MIN_MTU;
1085 struct inet6_dev *idev;
1086
Eric Dumazetc68f24c2010-06-14 04:46:20 +00001087 rcu_read_lock();
1088 idev = __in6_dev_get(dev);
1089 if (idev)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001090 mtu = idev->cnf.mtu6;
Eric Dumazetc68f24c2010-06-14 04:46:20 +00001091 rcu_read_unlock();
Linus Torvalds1da177e2005-04-16 15:20:36 -07001092 return mtu;
1093}
1094
YOSHIFUJI Hideaki6b75d092008-03-10 06:00:30 -04001095int ip6_dst_hoplimit(struct dst_entry *dst)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001096{
YOSHIFUJI Hideaki6b75d092008-03-10 06:00:30 -04001097 int hoplimit = dst_metric(dst, RTAX_HOPLIMIT);
1098 if (hoplimit < 0) {
1099 struct net_device *dev = dst->dev;
Eric Dumazetc68f24c2010-06-14 04:46:20 +00001100 struct inet6_dev *idev;
1101
1102 rcu_read_lock();
1103 idev = __in6_dev_get(dev);
1104 if (idev)
YOSHIFUJI Hideaki6b75d092008-03-10 06:00:30 -04001105 hoplimit = idev->cnf.hop_limit;
Eric Dumazetc68f24c2010-06-14 04:46:20 +00001106 else
YOSHIFUJI Hideaki53b79972008-07-19 22:35:03 -07001107 hoplimit = dev_net(dev)->ipv6.devconf_all->hop_limit;
Eric Dumazetc68f24c2010-06-14 04:46:20 +00001108 rcu_read_unlock();
Linus Torvalds1da177e2005-04-16 15:20:36 -07001109 }
1110 return hoplimit;
1111}
1112
1113/*
1114 *
1115 */
1116
Thomas Graf86872cb2006-08-22 00:01:08 -07001117int ip6_route_add(struct fib6_config *cfg)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001118{
1119 int err;
Daniel Lezcano55786892008-03-04 13:47:47 -08001120 struct net *net = cfg->fc_nlinfo.nl_net;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001121 struct rt6_info *rt = NULL;
1122 struct net_device *dev = NULL;
1123 struct inet6_dev *idev = NULL;
Thomas Grafc71099a2006-08-04 23:20:06 -07001124 struct fib6_table *table;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001125 int addr_type;
1126
Thomas Graf86872cb2006-08-22 00:01:08 -07001127 if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001128 return -EINVAL;
1129#ifndef CONFIG_IPV6_SUBTREES
Thomas Graf86872cb2006-08-22 00:01:08 -07001130 if (cfg->fc_src_len)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001131 return -EINVAL;
1132#endif
Thomas Graf86872cb2006-08-22 00:01:08 -07001133 if (cfg->fc_ifindex) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001134 err = -ENODEV;
Daniel Lezcano55786892008-03-04 13:47:47 -08001135 dev = dev_get_by_index(net, cfg->fc_ifindex);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001136 if (!dev)
1137 goto out;
1138 idev = in6_dev_get(dev);
1139 if (!idev)
1140 goto out;
1141 }
1142
Thomas Graf86872cb2006-08-22 00:01:08 -07001143 if (cfg->fc_metric == 0)
1144 cfg->fc_metric = IP6_RT_PRIO_USER;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001145
Daniel Lezcano55786892008-03-04 13:47:47 -08001146 table = fib6_new_table(net, cfg->fc_table);
Thomas Grafc71099a2006-08-04 23:20:06 -07001147 if (table == NULL) {
1148 err = -ENOBUFS;
1149 goto out;
1150 }
1151
Alexey Dobriyan86393e52009-08-29 01:34:49 +00001152 rt = ip6_dst_alloc(&net->ipv6.ip6_dst_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001153
1154 if (rt == NULL) {
1155 err = -ENOMEM;
1156 goto out;
1157 }
1158
Changli Gaod8d1f302010-06-10 23:31:35 -07001159 rt->dst.obsolete = -1;
YOSHIFUJI Hideaki6f704992008-05-19 16:56:11 -07001160 rt->rt6i_expires = (cfg->fc_flags & RTF_EXPIRES) ?
1161 jiffies + clock_t_to_jiffies(cfg->fc_expires) :
1162 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001163
Thomas Graf86872cb2006-08-22 00:01:08 -07001164 if (cfg->fc_protocol == RTPROT_UNSPEC)
1165 cfg->fc_protocol = RTPROT_BOOT;
1166 rt->rt6i_protocol = cfg->fc_protocol;
1167
1168 addr_type = ipv6_addr_type(&cfg->fc_dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001169
1170 if (addr_type & IPV6_ADDR_MULTICAST)
Changli Gaod8d1f302010-06-10 23:31:35 -07001171 rt->dst.input = ip6_mc_input;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001172 else
Changli Gaod8d1f302010-06-10 23:31:35 -07001173 rt->dst.input = ip6_forward;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001174
Changli Gaod8d1f302010-06-10 23:31:35 -07001175 rt->dst.output = ip6_output;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001176
Thomas Graf86872cb2006-08-22 00:01:08 -07001177 ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len);
1178 rt->rt6i_dst.plen = cfg->fc_dst_len;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001179 if (rt->rt6i_dst.plen == 128)
Changli Gaod8d1f302010-06-10 23:31:35 -07001180 rt->dst.flags = DST_HOST;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001181
1182#ifdef CONFIG_IPV6_SUBTREES
Thomas Graf86872cb2006-08-22 00:01:08 -07001183 ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len);
1184 rt->rt6i_src.plen = cfg->fc_src_len;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001185#endif
1186
Thomas Graf86872cb2006-08-22 00:01:08 -07001187 rt->rt6i_metric = cfg->fc_metric;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001188
1189 /* We cannot add true routes via loopback here,
1190 they would result in kernel looping; promote them to reject routes
1191 */
Thomas Graf86872cb2006-08-22 00:01:08 -07001192 if ((cfg->fc_flags & RTF_REJECT) ||
Linus Torvalds1da177e2005-04-16 15:20:36 -07001193 (dev && (dev->flags&IFF_LOOPBACK) && !(addr_type&IPV6_ADDR_LOOPBACK))) {
1194 /* hold loopback dev/idev if we haven't done so. */
Daniel Lezcano55786892008-03-04 13:47:47 -08001195 if (dev != net->loopback_dev) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001196 if (dev) {
1197 dev_put(dev);
1198 in6_dev_put(idev);
1199 }
Daniel Lezcano55786892008-03-04 13:47:47 -08001200 dev = net->loopback_dev;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001201 dev_hold(dev);
1202 idev = in6_dev_get(dev);
1203 if (!idev) {
1204 err = -ENODEV;
1205 goto out;
1206 }
1207 }
Changli Gaod8d1f302010-06-10 23:31:35 -07001208 rt->dst.output = ip6_pkt_discard_out;
1209 rt->dst.input = ip6_pkt_discard;
1210 rt->dst.error = -ENETUNREACH;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001211 rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP;
1212 goto install_route;
1213 }
1214
Thomas Graf86872cb2006-08-22 00:01:08 -07001215 if (cfg->fc_flags & RTF_GATEWAY) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001216 struct in6_addr *gw_addr;
1217 int gwa_type;
1218
Thomas Graf86872cb2006-08-22 00:01:08 -07001219 gw_addr = &cfg->fc_gateway;
1220 ipv6_addr_copy(&rt->rt6i_gateway, gw_addr);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001221 gwa_type = ipv6_addr_type(gw_addr);
1222
1223 if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) {
1224 struct rt6_info *grt;
1225
1226 /* IPv6 strictly inhibits using not link-local
1227 addresses as nexthop address.
1228 Otherwise, router will not able to send redirects.
1229 It is very good, but in some (rare!) circumstances
1230 (SIT, PtP, NBMA NOARP links) it is handy to allow
1231 some exceptions. --ANK
1232 */
1233 err = -EINVAL;
1234 if (!(gwa_type&IPV6_ADDR_UNICAST))
1235 goto out;
1236
Daniel Lezcano55786892008-03-04 13:47:47 -08001237 grt = rt6_lookup(net, gw_addr, NULL, cfg->fc_ifindex, 1);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001238
1239 err = -EHOSTUNREACH;
1240 if (grt == NULL)
1241 goto out;
1242 if (dev) {
1243 if (dev != grt->rt6i_dev) {
Changli Gaod8d1f302010-06-10 23:31:35 -07001244 dst_release(&grt->dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001245 goto out;
1246 }
1247 } else {
1248 dev = grt->rt6i_dev;
1249 idev = grt->rt6i_idev;
1250 dev_hold(dev);
1251 in6_dev_hold(grt->rt6i_idev);
1252 }
1253 if (!(grt->rt6i_flags&RTF_GATEWAY))
1254 err = 0;
Changli Gaod8d1f302010-06-10 23:31:35 -07001255 dst_release(&grt->dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001256
1257 if (err)
1258 goto out;
1259 }
1260 err = -EINVAL;
1261 if (dev == NULL || (dev->flags&IFF_LOOPBACK))
1262 goto out;
1263 }
1264
1265 err = -ENODEV;
1266 if (dev == NULL)
1267 goto out;
1268
Thomas Graf86872cb2006-08-22 00:01:08 -07001269 if (cfg->fc_flags & (RTF_GATEWAY | RTF_NONEXTHOP)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001270 rt->rt6i_nexthop = __neigh_lookup_errno(&nd_tbl, &rt->rt6i_gateway, dev);
1271 if (IS_ERR(rt->rt6i_nexthop)) {
1272 err = PTR_ERR(rt->rt6i_nexthop);
1273 rt->rt6i_nexthop = NULL;
1274 goto out;
1275 }
1276 }
1277
Thomas Graf86872cb2006-08-22 00:01:08 -07001278 rt->rt6i_flags = cfg->fc_flags;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001279
1280install_route:
Thomas Graf86872cb2006-08-22 00:01:08 -07001281 if (cfg->fc_mx) {
1282 struct nlattr *nla;
1283 int remaining;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001284
Thomas Graf86872cb2006-08-22 00:01:08 -07001285 nla_for_each_attr(nla, cfg->fc_mx, cfg->fc_mx_len, remaining) {
Thomas Graf8f4c1f92007-09-12 14:44:36 +02001286 int type = nla_type(nla);
Thomas Graf86872cb2006-08-22 00:01:08 -07001287
1288 if (type) {
1289 if (type > RTAX_MAX) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001290 err = -EINVAL;
1291 goto out;
1292 }
Thomas Graf86872cb2006-08-22 00:01:08 -07001293
Changli Gaod8d1f302010-06-10 23:31:35 -07001294 rt->dst.metrics[type - 1] = nla_get_u32(nla);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001295 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001296 }
1297 }
1298
Changli Gaod8d1f302010-06-10 23:31:35 -07001299 if (dst_metric(&rt->dst, RTAX_HOPLIMIT) == 0)
1300 rt->dst.metrics[RTAX_HOPLIMIT-1] = -1;
1301 if (!dst_mtu(&rt->dst))
1302 rt->dst.metrics[RTAX_MTU-1] = ipv6_get_mtu(dev);
1303 if (!dst_metric(&rt->dst, RTAX_ADVMSS))
1304 rt->dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(net, dst_mtu(&rt->dst));
1305 rt->dst.dev = dev;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001306 rt->rt6i_idev = idev;
Thomas Grafc71099a2006-08-04 23:20:06 -07001307 rt->rt6i_table = table;
Daniel Lezcano63152fc2008-03-03 23:31:11 -08001308
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09001309 cfg->fc_nlinfo.nl_net = dev_net(dev);
Daniel Lezcano63152fc2008-03-03 23:31:11 -08001310
Thomas Graf86872cb2006-08-22 00:01:08 -07001311 return __ip6_ins_rt(rt, &cfg->fc_nlinfo);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001312
1313out:
1314 if (dev)
1315 dev_put(dev);
1316 if (idev)
1317 in6_dev_put(idev);
1318 if (rt)
Changli Gaod8d1f302010-06-10 23:31:35 -07001319 dst_free(&rt->dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001320 return err;
1321}
1322
Thomas Graf86872cb2006-08-22 00:01:08 -07001323static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001324{
1325 int err;
Thomas Grafc71099a2006-08-04 23:20:06 -07001326 struct fib6_table *table;
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09001327 struct net *net = dev_net(rt->rt6i_dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001328
Daniel Lezcano8ed67782008-03-04 13:48:30 -08001329 if (rt == net->ipv6.ip6_null_entry)
Patrick McHardy6c813a72006-08-06 22:22:47 -07001330 return -ENOENT;
1331
Thomas Grafc71099a2006-08-04 23:20:06 -07001332 table = rt->rt6i_table;
1333 write_lock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001334
Thomas Graf86872cb2006-08-22 00:01:08 -07001335 err = fib6_del(rt, info);
Changli Gaod8d1f302010-06-10 23:31:35 -07001336 dst_release(&rt->dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001337
Thomas Grafc71099a2006-08-04 23:20:06 -07001338 write_unlock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001339
1340 return err;
1341}
1342
Thomas Grafe0a1ad732006-08-22 00:00:21 -07001343int ip6_del_rt(struct rt6_info *rt)
1344{
Denis V. Lunev4d1169c2008-01-10 03:26:13 -08001345 struct nl_info info = {
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09001346 .nl_net = dev_net(rt->rt6i_dev),
Denis V. Lunev4d1169c2008-01-10 03:26:13 -08001347 };
Denis V. Lunev528c4ce2007-12-13 09:45:12 -08001348 return __ip6_del_rt(rt, &info);
Thomas Grafe0a1ad732006-08-22 00:00:21 -07001349}
1350
Thomas Graf86872cb2006-08-22 00:01:08 -07001351static int ip6_route_del(struct fib6_config *cfg)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001352{
Thomas Grafc71099a2006-08-04 23:20:06 -07001353 struct fib6_table *table;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001354 struct fib6_node *fn;
1355 struct rt6_info *rt;
1356 int err = -ESRCH;
1357
Daniel Lezcano55786892008-03-04 13:47:47 -08001358 table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table);
Thomas Grafc71099a2006-08-04 23:20:06 -07001359 if (table == NULL)
1360 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001361
Thomas Grafc71099a2006-08-04 23:20:06 -07001362 read_lock_bh(&table->tb6_lock);
1363
1364 fn = fib6_locate(&table->tb6_root,
Thomas Graf86872cb2006-08-22 00:01:08 -07001365 &cfg->fc_dst, cfg->fc_dst_len,
1366 &cfg->fc_src, cfg->fc_src_len);
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09001367
Linus Torvalds1da177e2005-04-16 15:20:36 -07001368 if (fn) {
Changli Gaod8d1f302010-06-10 23:31:35 -07001369 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
Thomas Graf86872cb2006-08-22 00:01:08 -07001370 if (cfg->fc_ifindex &&
Linus Torvalds1da177e2005-04-16 15:20:36 -07001371 (rt->rt6i_dev == NULL ||
Thomas Graf86872cb2006-08-22 00:01:08 -07001372 rt->rt6i_dev->ifindex != cfg->fc_ifindex))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001373 continue;
Thomas Graf86872cb2006-08-22 00:01:08 -07001374 if (cfg->fc_flags & RTF_GATEWAY &&
1375 !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001376 continue;
Thomas Graf86872cb2006-08-22 00:01:08 -07001377 if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001378 continue;
Changli Gaod8d1f302010-06-10 23:31:35 -07001379 dst_hold(&rt->dst);
Thomas Grafc71099a2006-08-04 23:20:06 -07001380 read_unlock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001381
Thomas Graf86872cb2006-08-22 00:01:08 -07001382 return __ip6_del_rt(rt, &cfg->fc_nlinfo);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001383 }
1384 }
Thomas Grafc71099a2006-08-04 23:20:06 -07001385 read_unlock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001386
1387 return err;
1388}
1389
1390/*
1391 * Handle redirects
1392 */
YOSHIFUJI Hideakia6279452006-08-23 17:18:26 -07001393struct ip6rd_flowi {
1394 struct flowi fl;
1395 struct in6_addr gateway;
1396};
Linus Torvalds1da177e2005-04-16 15:20:36 -07001397
Daniel Lezcano8ed67782008-03-04 13:48:30 -08001398static struct rt6_info *__ip6_route_redirect(struct net *net,
1399 struct fib6_table *table,
YOSHIFUJI Hideakia6279452006-08-23 17:18:26 -07001400 struct flowi *fl,
1401 int flags)
1402{
1403 struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl;
1404 struct rt6_info *rt;
1405 struct fib6_node *fn;
Thomas Grafc71099a2006-08-04 23:20:06 -07001406
Linus Torvalds1da177e2005-04-16 15:20:36 -07001407 /*
YOSHIFUJI Hideakie843b9e2006-03-20 17:07:49 -08001408 * Get the "current" route for this destination and
1409 * check if the redirect has come from approriate router.
1410 *
1411 * RFC 2461 specifies that redirects should only be
1412 * accepted if they come from the nexthop to the target.
1413 * Due to the way the routes are chosen, this notion
1414 * is a bit fuzzy and one might need to check all possible
1415 * routes.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001416 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001417
Thomas Grafc71099a2006-08-04 23:20:06 -07001418 read_lock_bh(&table->tb6_lock);
YOSHIFUJI Hideakia6279452006-08-23 17:18:26 -07001419 fn = fib6_lookup(&table->tb6_root, &fl->fl6_dst, &fl->fl6_src);
YOSHIFUJI Hideakie843b9e2006-03-20 17:07:49 -08001420restart:
Changli Gaod8d1f302010-06-10 23:31:35 -07001421 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
YOSHIFUJI Hideakie843b9e2006-03-20 17:07:49 -08001422 /*
1423 * Current route is on-link; redirect is always invalid.
1424 *
1425 * Seems, previous statement is not true. It could
1426 * be node, which looks for us as on-link (f.e. proxy ndisc)
1427 * But then router serving it might decide, that we should
1428 * know truth 8)8) --ANK (980726).
1429 */
1430 if (rt6_check_expired(rt))
1431 continue;
1432 if (!(rt->rt6i_flags & RTF_GATEWAY))
1433 continue;
YOSHIFUJI Hideakia6279452006-08-23 17:18:26 -07001434 if (fl->oif != rt->rt6i_dev->ifindex)
YOSHIFUJI Hideakie843b9e2006-03-20 17:07:49 -08001435 continue;
YOSHIFUJI Hideakia6279452006-08-23 17:18:26 -07001436 if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway))
YOSHIFUJI Hideakie843b9e2006-03-20 17:07:49 -08001437 continue;
1438 break;
1439 }
YOSHIFUJI Hideakie843b9e2006-03-20 17:07:49 -08001440
YOSHIFUJI Hideakicb15d9c2006-08-23 17:23:11 -07001441 if (!rt)
Daniel Lezcano8ed67782008-03-04 13:48:30 -08001442 rt = net->ipv6.ip6_null_entry;
1443 BACKTRACK(net, &fl->fl6_src);
YOSHIFUJI Hideakicb15d9c2006-08-23 17:23:11 -07001444out:
Changli Gaod8d1f302010-06-10 23:31:35 -07001445 dst_hold(&rt->dst);
YOSHIFUJI Hideakia6279452006-08-23 17:18:26 -07001446
1447 read_unlock_bh(&table->tb6_lock);
1448
1449 return rt;
1450};
1451
1452static struct rt6_info *ip6_route_redirect(struct in6_addr *dest,
1453 struct in6_addr *src,
1454 struct in6_addr *gateway,
1455 struct net_device *dev)
1456{
Thomas Grafadaa70b2006-10-13 15:01:03 -07001457 int flags = RT6_LOOKUP_F_HAS_SADDR;
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09001458 struct net *net = dev_net(dev);
YOSHIFUJI Hideakia6279452006-08-23 17:18:26 -07001459 struct ip6rd_flowi rdfl = {
1460 .fl = {
1461 .oif = dev->ifindex,
1462 .nl_u = {
1463 .ip6_u = {
1464 .daddr = *dest,
1465 .saddr = *src,
1466 },
1467 },
1468 },
YOSHIFUJI Hideakia6279452006-08-23 17:18:26 -07001469 };
Thomas Grafadaa70b2006-10-13 15:01:03 -07001470
Brian Haley86c36ce2009-10-07 13:58:01 -07001471 ipv6_addr_copy(&rdfl.gateway, gateway);
1472
Thomas Grafadaa70b2006-10-13 15:01:03 -07001473 if (rt6_need_strict(dest))
1474 flags |= RT6_LOOKUP_F_IFACE;
YOSHIFUJI Hideakia6279452006-08-23 17:18:26 -07001475
Daniel Lezcano55786892008-03-04 13:47:47 -08001476 return (struct rt6_info *)fib6_rule_lookup(net, (struct flowi *)&rdfl,
Daniel Lezcano58f09b72008-03-03 23:25:27 -08001477 flags, __ip6_route_redirect);
YOSHIFUJI Hideakia6279452006-08-23 17:18:26 -07001478}
1479
1480void rt6_redirect(struct in6_addr *dest, struct in6_addr *src,
1481 struct in6_addr *saddr,
1482 struct neighbour *neigh, u8 *lladdr, int on_link)
1483{
1484 struct rt6_info *rt, *nrt = NULL;
1485 struct netevent_redirect netevent;
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09001486 struct net *net = dev_net(neigh->dev);
YOSHIFUJI Hideakia6279452006-08-23 17:18:26 -07001487
1488 rt = ip6_route_redirect(dest, src, saddr, neigh->dev);
1489
Daniel Lezcano8ed67782008-03-04 13:48:30 -08001490 if (rt == net->ipv6.ip6_null_entry) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001491 if (net_ratelimit())
1492 printk(KERN_DEBUG "rt6_redirect: source isn't a valid nexthop "
1493 "for redirect target\n");
YOSHIFUJI Hideakia6279452006-08-23 17:18:26 -07001494 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001495 }
1496
Linus Torvalds1da177e2005-04-16 15:20:36 -07001497 /*
1498 * We have finally decided to accept it.
1499 */
1500
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09001501 neigh_update(neigh, lladdr, NUD_STALE,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001502 NEIGH_UPDATE_F_WEAK_OVERRIDE|
1503 NEIGH_UPDATE_F_OVERRIDE|
1504 (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER|
1505 NEIGH_UPDATE_F_ISROUTER))
1506 );
1507
1508 /*
1509 * Redirect received -> path was valid.
1510 * Look, redirects are sent only in response to data packets,
1511 * so that this nexthop apparently is reachable. --ANK
1512 */
Changli Gaod8d1f302010-06-10 23:31:35 -07001513 dst_confirm(&rt->dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001514
1515 /* Duplicate redirect: silently ignore. */
Changli Gaod8d1f302010-06-10 23:31:35 -07001516 if (neigh == rt->dst.neighbour)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001517 goto out;
1518
1519 nrt = ip6_rt_copy(rt);
1520 if (nrt == NULL)
1521 goto out;
1522
1523 nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE;
1524 if (on_link)
1525 nrt->rt6i_flags &= ~RTF_GATEWAY;
1526
1527 ipv6_addr_copy(&nrt->rt6i_dst.addr, dest);
1528 nrt->rt6i_dst.plen = 128;
Changli Gaod8d1f302010-06-10 23:31:35 -07001529 nrt->dst.flags |= DST_HOST;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001530
1531 ipv6_addr_copy(&nrt->rt6i_gateway, (struct in6_addr*)neigh->primary_key);
1532 nrt->rt6i_nexthop = neigh_clone(neigh);
1533 /* Reset pmtu, it may be better */
Changli Gaod8d1f302010-06-10 23:31:35 -07001534 nrt->dst.metrics[RTAX_MTU-1] = ipv6_get_mtu(neigh->dev);
1535 nrt->dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(dev_net(neigh->dev),
1536 dst_mtu(&nrt->dst));
Linus Torvalds1da177e2005-04-16 15:20:36 -07001537
Thomas Graf40e22e82006-08-22 00:00:45 -07001538 if (ip6_ins_rt(nrt))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001539 goto out;
1540
Changli Gaod8d1f302010-06-10 23:31:35 -07001541 netevent.old = &rt->dst;
1542 netevent.new = &nrt->dst;
Tom Tucker8d717402006-07-30 20:43:36 -07001543 call_netevent_notifiers(NETEVENT_REDIRECT, &netevent);
1544
Linus Torvalds1da177e2005-04-16 15:20:36 -07001545 if (rt->rt6i_flags&RTF_CACHE) {
Thomas Grafe0a1ad732006-08-22 00:00:21 -07001546 ip6_del_rt(rt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001547 return;
1548 }
1549
1550out:
Changli Gaod8d1f302010-06-10 23:31:35 -07001551 dst_release(&rt->dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001552}
1553
1554/*
1555 * Handle ICMP "packet too big" messages
1556 * i.e. Path MTU discovery
1557 */
1558
Maciej Żenczykowskiae878ae2010-10-03 14:49:00 -07001559static void rt6_do_pmtu_disc(struct in6_addr *daddr, struct in6_addr *saddr,
1560 struct net *net, u32 pmtu, int ifindex)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001561{
1562 struct rt6_info *rt, *nrt;
1563 int allfrag = 0;
1564
Maciej Żenczykowskiae878ae2010-10-03 14:49:00 -07001565 rt = rt6_lookup(net, daddr, saddr, ifindex, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001566 if (rt == NULL)
1567 return;
1568
Changli Gaod8d1f302010-06-10 23:31:35 -07001569 if (pmtu >= dst_mtu(&rt->dst))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001570 goto out;
1571
1572 if (pmtu < IPV6_MIN_MTU) {
1573 /*
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09001574 * According to RFC2460, PMTU is set to the IPv6 Minimum Link
Linus Torvalds1da177e2005-04-16 15:20:36 -07001575 * MTU (1280) and a fragment header should always be included
1576 * after a node receiving Too Big message reporting PMTU is
1577 * less than the IPv6 Minimum Link MTU.
1578 */
1579 pmtu = IPV6_MIN_MTU;
1580 allfrag = 1;
1581 }
1582
1583 /* New mtu received -> path was valid.
1584 They are sent only in response to data packets,
1585 so that this nexthop apparently is reachable. --ANK
1586 */
Changli Gaod8d1f302010-06-10 23:31:35 -07001587 dst_confirm(&rt->dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001588
1589 /* Host route. If it is static, it would be better
1590 not to override it, but add new one, so that
1591 when cache entry will expire old pmtu
1592 would return automatically.
1593 */
1594 if (rt->rt6i_flags & RTF_CACHE) {
Changli Gaod8d1f302010-06-10 23:31:35 -07001595 rt->dst.metrics[RTAX_MTU-1] = pmtu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001596 if (allfrag)
Changli Gaod8d1f302010-06-10 23:31:35 -07001597 rt->dst.metrics[RTAX_FEATURES-1] |= RTAX_FEATURE_ALLFRAG;
1598 dst_set_expires(&rt->dst, net->ipv6.sysctl.ip6_rt_mtu_expires);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001599 rt->rt6i_flags |= RTF_MODIFIED|RTF_EXPIRES;
1600 goto out;
1601 }
1602
1603 /* Network route.
1604 Two cases are possible:
1605 1. It is connected route. Action: COW
1606 2. It is gatewayed route or NONEXTHOP route. Action: clone it.
1607 */
YOSHIFUJI Hideakid5315b502006-03-20 16:58:48 -08001608 if (!rt->rt6i_nexthop && !(rt->rt6i_flags & RTF_NONEXTHOP))
YOSHIFUJI Hideakia1e78362006-03-20 16:56:32 -08001609 nrt = rt6_alloc_cow(rt, daddr, saddr);
YOSHIFUJI Hideakid5315b502006-03-20 16:58:48 -08001610 else
1611 nrt = rt6_alloc_clone(rt, daddr);
YOSHIFUJI Hideakia1e78362006-03-20 16:56:32 -08001612
YOSHIFUJI Hideakid5315b502006-03-20 16:58:48 -08001613 if (nrt) {
Changli Gaod8d1f302010-06-10 23:31:35 -07001614 nrt->dst.metrics[RTAX_MTU-1] = pmtu;
YOSHIFUJI Hideakia1e78362006-03-20 16:56:32 -08001615 if (allfrag)
Changli Gaod8d1f302010-06-10 23:31:35 -07001616 nrt->dst.metrics[RTAX_FEATURES-1] |= RTAX_FEATURE_ALLFRAG;
YOSHIFUJI Hideakia1e78362006-03-20 16:56:32 -08001617
1618 /* According to RFC 1981, detecting PMTU increase shouldn't be
1619 * happened within 5 mins, the recommended timer is 10 mins.
1620 * Here this route expiration time is set to ip6_rt_mtu_expires
1621 * which is 10 mins. After 10 mins the decreased pmtu is expired
1622 * and detecting PMTU increase will be automatically happened.
1623 */
Changli Gaod8d1f302010-06-10 23:31:35 -07001624 dst_set_expires(&nrt->dst, net->ipv6.sysctl.ip6_rt_mtu_expires);
YOSHIFUJI Hideakia1e78362006-03-20 16:56:32 -08001625 nrt->rt6i_flags |= RTF_DYNAMIC|RTF_EXPIRES;
1626
Thomas Graf40e22e82006-08-22 00:00:45 -07001627 ip6_ins_rt(nrt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001628 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001629out:
Changli Gaod8d1f302010-06-10 23:31:35 -07001630 dst_release(&rt->dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001631}
1632
Maciej Żenczykowskiae878ae2010-10-03 14:49:00 -07001633void rt6_pmtu_discovery(struct in6_addr *daddr, struct in6_addr *saddr,
1634 struct net_device *dev, u32 pmtu)
1635{
1636 struct net *net = dev_net(dev);
1637
1638 /*
1639 * RFC 1981 states that a node "MUST reduce the size of the packets it
1640 * is sending along the path" that caused the Packet Too Big message.
1641 * Since it's not possible in the general case to determine which
1642 * interface was used to send the original packet, we update the MTU
1643 * on the interface that will be used to send future packets. We also
1644 * update the MTU on the interface that received the Packet Too Big in
1645 * case the original packet was forced out that interface with
1646 * SO_BINDTODEVICE or similar. This is the next best thing to the
1647 * correct behaviour, which would be to update the MTU on all
1648 * interfaces.
1649 */
1650 rt6_do_pmtu_disc(daddr, saddr, net, pmtu, 0);
1651 rt6_do_pmtu_disc(daddr, saddr, net, pmtu, dev->ifindex);
1652}
1653
Linus Torvalds1da177e2005-04-16 15:20:36 -07001654/*
1655 * Misc support functions
1656 */
1657
1658static struct rt6_info * ip6_rt_copy(struct rt6_info *ort)
1659{
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09001660 struct net *net = dev_net(ort->rt6i_dev);
Alexey Dobriyan86393e52009-08-29 01:34:49 +00001661 struct rt6_info *rt = ip6_dst_alloc(&net->ipv6.ip6_dst_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001662
1663 if (rt) {
Changli Gaod8d1f302010-06-10 23:31:35 -07001664 rt->dst.input = ort->dst.input;
1665 rt->dst.output = ort->dst.output;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001666
Changli Gaod8d1f302010-06-10 23:31:35 -07001667 memcpy(rt->dst.metrics, ort->dst.metrics, RTAX_MAX*sizeof(u32));
1668 rt->dst.error = ort->dst.error;
1669 rt->dst.dev = ort->dst.dev;
1670 if (rt->dst.dev)
1671 dev_hold(rt->dst.dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001672 rt->rt6i_idev = ort->rt6i_idev;
1673 if (rt->rt6i_idev)
1674 in6_dev_hold(rt->rt6i_idev);
Changli Gaod8d1f302010-06-10 23:31:35 -07001675 rt->dst.lastuse = jiffies;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001676 rt->rt6i_expires = 0;
1677
1678 ipv6_addr_copy(&rt->rt6i_gateway, &ort->rt6i_gateway);
1679 rt->rt6i_flags = ort->rt6i_flags & ~RTF_EXPIRES;
1680 rt->rt6i_metric = 0;
1681
1682 memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key));
1683#ifdef CONFIG_IPV6_SUBTREES
1684 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1685#endif
Thomas Grafc71099a2006-08-04 23:20:06 -07001686 rt->rt6i_table = ort->rt6i_table;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001687 }
1688 return rt;
1689}
1690
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08001691#ifdef CONFIG_IPV6_ROUTE_INFO
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -08001692static struct rt6_info *rt6_get_route_info(struct net *net,
1693 struct in6_addr *prefix, int prefixlen,
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08001694 struct in6_addr *gwaddr, int ifindex)
1695{
1696 struct fib6_node *fn;
1697 struct rt6_info *rt = NULL;
Thomas Grafc71099a2006-08-04 23:20:06 -07001698 struct fib6_table *table;
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08001699
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -08001700 table = fib6_get_table(net, RT6_TABLE_INFO);
Thomas Grafc71099a2006-08-04 23:20:06 -07001701 if (table == NULL)
1702 return NULL;
1703
1704 write_lock_bh(&table->tb6_lock);
1705 fn = fib6_locate(&table->tb6_root, prefix ,prefixlen, NULL, 0);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08001706 if (!fn)
1707 goto out;
1708
Changli Gaod8d1f302010-06-10 23:31:35 -07001709 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08001710 if (rt->rt6i_dev->ifindex != ifindex)
1711 continue;
1712 if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY))
1713 continue;
1714 if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr))
1715 continue;
Changli Gaod8d1f302010-06-10 23:31:35 -07001716 dst_hold(&rt->dst);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08001717 break;
1718 }
1719out:
Thomas Grafc71099a2006-08-04 23:20:06 -07001720 write_unlock_bh(&table->tb6_lock);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08001721 return rt;
1722}
1723
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -08001724static struct rt6_info *rt6_add_route_info(struct net *net,
1725 struct in6_addr *prefix, int prefixlen,
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08001726 struct in6_addr *gwaddr, int ifindex,
1727 unsigned pref)
1728{
Thomas Graf86872cb2006-08-22 00:01:08 -07001729 struct fib6_config cfg = {
1730 .fc_table = RT6_TABLE_INFO,
Rami Rosen238fc7e2008-02-09 23:43:11 -08001731 .fc_metric = IP6_RT_PRIO_USER,
Thomas Graf86872cb2006-08-22 00:01:08 -07001732 .fc_ifindex = ifindex,
1733 .fc_dst_len = prefixlen,
1734 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO |
1735 RTF_UP | RTF_PREF(pref),
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -08001736 .fc_nlinfo.pid = 0,
1737 .fc_nlinfo.nlh = NULL,
1738 .fc_nlinfo.nl_net = net,
Thomas Graf86872cb2006-08-22 00:01:08 -07001739 };
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08001740
Thomas Graf86872cb2006-08-22 00:01:08 -07001741 ipv6_addr_copy(&cfg.fc_dst, prefix);
1742 ipv6_addr_copy(&cfg.fc_gateway, gwaddr);
1743
YOSHIFUJI Hideakie317da92006-03-20 17:06:42 -08001744 /* We should treat it as a default route if prefix length is 0. */
1745 if (!prefixlen)
Thomas Graf86872cb2006-08-22 00:01:08 -07001746 cfg.fc_flags |= RTF_DEFAULT;
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08001747
Thomas Graf86872cb2006-08-22 00:01:08 -07001748 ip6_route_add(&cfg);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08001749
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -08001750 return rt6_get_route_info(net, prefix, prefixlen, gwaddr, ifindex);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08001751}
1752#endif
1753
Linus Torvalds1da177e2005-04-16 15:20:36 -07001754struct rt6_info *rt6_get_dflt_router(struct in6_addr *addr, struct net_device *dev)
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09001755{
Linus Torvalds1da177e2005-04-16 15:20:36 -07001756 struct rt6_info *rt;
Thomas Grafc71099a2006-08-04 23:20:06 -07001757 struct fib6_table *table;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001758
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09001759 table = fib6_get_table(dev_net(dev), RT6_TABLE_DFLT);
Thomas Grafc71099a2006-08-04 23:20:06 -07001760 if (table == NULL)
1761 return NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001762
Thomas Grafc71099a2006-08-04 23:20:06 -07001763 write_lock_bh(&table->tb6_lock);
Changli Gaod8d1f302010-06-10 23:31:35 -07001764 for (rt = table->tb6_root.leaf; rt; rt=rt->dst.rt6_next) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001765 if (dev == rt->rt6i_dev &&
YOSHIFUJI Hideaki045927f2006-03-20 17:00:48 -08001766 ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) &&
Linus Torvalds1da177e2005-04-16 15:20:36 -07001767 ipv6_addr_equal(&rt->rt6i_gateway, addr))
1768 break;
1769 }
1770 if (rt)
Changli Gaod8d1f302010-06-10 23:31:35 -07001771 dst_hold(&rt->dst);
Thomas Grafc71099a2006-08-04 23:20:06 -07001772 write_unlock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001773 return rt;
1774}
1775
1776struct rt6_info *rt6_add_dflt_router(struct in6_addr *gwaddr,
YOSHIFUJI Hideakiebacaaa2006-03-20 17:04:53 -08001777 struct net_device *dev,
1778 unsigned int pref)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001779{
Thomas Graf86872cb2006-08-22 00:01:08 -07001780 struct fib6_config cfg = {
1781 .fc_table = RT6_TABLE_DFLT,
Rami Rosen238fc7e2008-02-09 23:43:11 -08001782 .fc_metric = IP6_RT_PRIO_USER,
Thomas Graf86872cb2006-08-22 00:01:08 -07001783 .fc_ifindex = dev->ifindex,
1784 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT |
1785 RTF_UP | RTF_EXPIRES | RTF_PREF(pref),
Daniel Lezcano55786892008-03-04 13:47:47 -08001786 .fc_nlinfo.pid = 0,
1787 .fc_nlinfo.nlh = NULL,
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09001788 .fc_nlinfo.nl_net = dev_net(dev),
Thomas Graf86872cb2006-08-22 00:01:08 -07001789 };
Linus Torvalds1da177e2005-04-16 15:20:36 -07001790
Thomas Graf86872cb2006-08-22 00:01:08 -07001791 ipv6_addr_copy(&cfg.fc_gateway, gwaddr);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001792
Thomas Graf86872cb2006-08-22 00:01:08 -07001793 ip6_route_add(&cfg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001794
Linus Torvalds1da177e2005-04-16 15:20:36 -07001795 return rt6_get_dflt_router(gwaddr, dev);
1796}
1797
Daniel Lezcano7b4da532008-03-04 13:47:14 -08001798void rt6_purge_dflt_routers(struct net *net)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001799{
1800 struct rt6_info *rt;
Thomas Grafc71099a2006-08-04 23:20:06 -07001801 struct fib6_table *table;
1802
1803 /* NOTE: Keep consistent with rt6_get_dflt_router */
Daniel Lezcano7b4da532008-03-04 13:47:14 -08001804 table = fib6_get_table(net, RT6_TABLE_DFLT);
Thomas Grafc71099a2006-08-04 23:20:06 -07001805 if (table == NULL)
1806 return;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001807
1808restart:
Thomas Grafc71099a2006-08-04 23:20:06 -07001809 read_lock_bh(&table->tb6_lock);
Changli Gaod8d1f302010-06-10 23:31:35 -07001810 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001811 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF)) {
Changli Gaod8d1f302010-06-10 23:31:35 -07001812 dst_hold(&rt->dst);
Thomas Grafc71099a2006-08-04 23:20:06 -07001813 read_unlock_bh(&table->tb6_lock);
Thomas Grafe0a1ad732006-08-22 00:00:21 -07001814 ip6_del_rt(rt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001815 goto restart;
1816 }
1817 }
Thomas Grafc71099a2006-08-04 23:20:06 -07001818 read_unlock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001819}
1820
Daniel Lezcano55786892008-03-04 13:47:47 -08001821static void rtmsg_to_fib6_config(struct net *net,
1822 struct in6_rtmsg *rtmsg,
Thomas Graf86872cb2006-08-22 00:01:08 -07001823 struct fib6_config *cfg)
1824{
1825 memset(cfg, 0, sizeof(*cfg));
1826
1827 cfg->fc_table = RT6_TABLE_MAIN;
1828 cfg->fc_ifindex = rtmsg->rtmsg_ifindex;
1829 cfg->fc_metric = rtmsg->rtmsg_metric;
1830 cfg->fc_expires = rtmsg->rtmsg_info;
1831 cfg->fc_dst_len = rtmsg->rtmsg_dst_len;
1832 cfg->fc_src_len = rtmsg->rtmsg_src_len;
1833 cfg->fc_flags = rtmsg->rtmsg_flags;
1834
Daniel Lezcano55786892008-03-04 13:47:47 -08001835 cfg->fc_nlinfo.nl_net = net;
Benjamin Theryf1243c22008-02-26 18:10:03 -08001836
Thomas Graf86872cb2006-08-22 00:01:08 -07001837 ipv6_addr_copy(&cfg->fc_dst, &rtmsg->rtmsg_dst);
1838 ipv6_addr_copy(&cfg->fc_src, &rtmsg->rtmsg_src);
1839 ipv6_addr_copy(&cfg->fc_gateway, &rtmsg->rtmsg_gateway);
1840}
1841
Daniel Lezcano55786892008-03-04 13:47:47 -08001842int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001843{
Thomas Graf86872cb2006-08-22 00:01:08 -07001844 struct fib6_config cfg;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001845 struct in6_rtmsg rtmsg;
1846 int err;
1847
1848 switch(cmd) {
1849 case SIOCADDRT: /* Add a route */
1850 case SIOCDELRT: /* Delete a route */
1851 if (!capable(CAP_NET_ADMIN))
1852 return -EPERM;
1853 err = copy_from_user(&rtmsg, arg,
1854 sizeof(struct in6_rtmsg));
1855 if (err)
1856 return -EFAULT;
Thomas Graf86872cb2006-08-22 00:01:08 -07001857
Daniel Lezcano55786892008-03-04 13:47:47 -08001858 rtmsg_to_fib6_config(net, &rtmsg, &cfg);
Thomas Graf86872cb2006-08-22 00:01:08 -07001859
Linus Torvalds1da177e2005-04-16 15:20:36 -07001860 rtnl_lock();
1861 switch (cmd) {
1862 case SIOCADDRT:
Thomas Graf86872cb2006-08-22 00:01:08 -07001863 err = ip6_route_add(&cfg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001864 break;
1865 case SIOCDELRT:
Thomas Graf86872cb2006-08-22 00:01:08 -07001866 err = ip6_route_del(&cfg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001867 break;
1868 default:
1869 err = -EINVAL;
1870 }
1871 rtnl_unlock();
1872
1873 return err;
Stephen Hemminger3ff50b72007-04-20 17:09:22 -07001874 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001875
1876 return -EINVAL;
1877}
1878
1879/*
1880 * Drop the packet on the floor
1881 */
1882
Brian Haleyd5fdd6b2009-06-23 04:31:07 -07001883static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001884{
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07001885 int type;
Eric Dumazetadf30902009-06-02 05:19:30 +00001886 struct dst_entry *dst = skb_dst(skb);
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07001887 switch (ipstats_mib_noroutes) {
1888 case IPSTATS_MIB_INNOROUTES:
Arnaldo Carvalho de Melo0660e032007-04-25 17:54:47 -07001889 type = ipv6_addr_type(&ipv6_hdr(skb)->daddr);
Ulrich Weber45bb0062010-02-25 23:28:58 +00001890 if (type == IPV6_ADDR_ANY) {
Denis V. Lunev3bd653c2008-10-08 10:54:51 -07001891 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
1892 IPSTATS_MIB_INADDRERRORS);
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07001893 break;
1894 }
1895 /* FALLTHROUGH */
1896 case IPSTATS_MIB_OUTNOROUTES:
Denis V. Lunev3bd653c2008-10-08 10:54:51 -07001897 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
1898 ipstats_mib_noroutes);
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07001899 break;
1900 }
Alexey Dobriyan3ffe5332010-02-18 08:25:24 +00001901 icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001902 kfree_skb(skb);
1903 return 0;
1904}
1905
Thomas Graf9ce8ade2006-10-18 20:46:54 -07001906static int ip6_pkt_discard(struct sk_buff *skb)
1907{
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07001908 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES);
Thomas Graf9ce8ade2006-10-18 20:46:54 -07001909}
1910
Arnaldo Carvalho de Melo20380732005-08-16 02:18:02 -03001911static int ip6_pkt_discard_out(struct sk_buff *skb)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001912{
Eric Dumazetadf30902009-06-02 05:19:30 +00001913 skb->dev = skb_dst(skb)->dev;
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07001914 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001915}
1916
David S. Miller6723ab52006-10-18 21:20:57 -07001917#ifdef CONFIG_IPV6_MULTIPLE_TABLES
1918
Thomas Graf9ce8ade2006-10-18 20:46:54 -07001919static int ip6_pkt_prohibit(struct sk_buff *skb)
1920{
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07001921 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES);
Thomas Graf9ce8ade2006-10-18 20:46:54 -07001922}
1923
1924static int ip6_pkt_prohibit_out(struct sk_buff *skb)
1925{
Eric Dumazetadf30902009-06-02 05:19:30 +00001926 skb->dev = skb_dst(skb)->dev;
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07001927 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES);
Thomas Graf9ce8ade2006-10-18 20:46:54 -07001928}
1929
David S. Miller6723ab52006-10-18 21:20:57 -07001930#endif
1931
Linus Torvalds1da177e2005-04-16 15:20:36 -07001932/*
1933 * Allocate a dst for local (unicast / anycast) address.
1934 */
1935
1936struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
1937 const struct in6_addr *addr,
1938 int anycast)
1939{
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09001940 struct net *net = dev_net(idev->dev);
Alexey Dobriyan86393e52009-08-29 01:34:49 +00001941 struct rt6_info *rt = ip6_dst_alloc(&net->ipv6.ip6_dst_ops);
David S. Miller14deae42009-01-04 16:04:39 -08001942 struct neighbour *neigh;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001943
1944 if (rt == NULL)
1945 return ERR_PTR(-ENOMEM);
1946
Daniel Lezcano55786892008-03-04 13:47:47 -08001947 dev_hold(net->loopback_dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001948 in6_dev_hold(idev);
1949
Changli Gaod8d1f302010-06-10 23:31:35 -07001950 rt->dst.flags = DST_HOST;
1951 rt->dst.input = ip6_input;
1952 rt->dst.output = ip6_output;
Daniel Lezcano55786892008-03-04 13:47:47 -08001953 rt->rt6i_dev = net->loopback_dev;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001954 rt->rt6i_idev = idev;
Changli Gaod8d1f302010-06-10 23:31:35 -07001955 rt->dst.metrics[RTAX_MTU-1] = ipv6_get_mtu(rt->rt6i_dev);
1956 rt->dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(net, dst_mtu(&rt->dst));
1957 rt->dst.metrics[RTAX_HOPLIMIT-1] = -1;
1958 rt->dst.obsolete = -1;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001959
1960 rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP;
YOSHIFUJI Hideaki58c4fb82005-12-21 22:56:42 +09001961 if (anycast)
1962 rt->rt6i_flags |= RTF_ANYCAST;
1963 else
Linus Torvalds1da177e2005-04-16 15:20:36 -07001964 rt->rt6i_flags |= RTF_LOCAL;
David S. Miller14deae42009-01-04 16:04:39 -08001965 neigh = ndisc_get_neigh(rt->rt6i_dev, &rt->rt6i_gateway);
1966 if (IS_ERR(neigh)) {
Changli Gaod8d1f302010-06-10 23:31:35 -07001967 dst_free(&rt->dst);
David S. Miller14deae42009-01-04 16:04:39 -08001968
1969 /* We are casting this because that is the return
1970 * value type. But an errno encoded pointer is the
1971 * same regardless of the underlying pointer type,
1972 * and that's what we are returning. So this is OK.
1973 */
1974 return (struct rt6_info *) neigh;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001975 }
David S. Miller14deae42009-01-04 16:04:39 -08001976 rt->rt6i_nexthop = neigh;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001977
1978 ipv6_addr_copy(&rt->rt6i_dst.addr, addr);
1979 rt->rt6i_dst.plen = 128;
Daniel Lezcano55786892008-03-04 13:47:47 -08001980 rt->rt6i_table = fib6_get_table(net, RT6_TABLE_LOCAL);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001981
Changli Gaod8d1f302010-06-10 23:31:35 -07001982 atomic_set(&rt->dst.__refcnt, 1);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001983
1984 return rt;
1985}
1986
Daniel Lezcano8ed67782008-03-04 13:48:30 -08001987struct arg_dev_net {
1988 struct net_device *dev;
1989 struct net *net;
1990};
1991
Linus Torvalds1da177e2005-04-16 15:20:36 -07001992static int fib6_ifdown(struct rt6_info *rt, void *arg)
1993{
Daniel Lezcano8ed67782008-03-04 13:48:30 -08001994 struct net_device *dev = ((struct arg_dev_net *)arg)->dev;
1995 struct net *net = ((struct arg_dev_net *)arg)->net;
1996
1997 if (((void *)rt->rt6i_dev == dev || dev == NULL) &&
1998 rt != net->ipv6.ip6_null_entry) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001999 RT6_TRACE("deleted by ifdown %p\n", rt);
2000 return -1;
2001 }
2002 return 0;
2003}
2004
Daniel Lezcanof3db4852008-03-03 23:27:06 -08002005void rt6_ifdown(struct net *net, struct net_device *dev)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002006{
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002007 struct arg_dev_net adn = {
2008 .dev = dev,
2009 .net = net,
2010 };
2011
2012 fib6_clean_all(net, fib6_ifdown, 0, &adn);
David S. Miller1e493d12008-09-10 17:27:15 -07002013 icmp6_clean_all(fib6_ifdown, &adn);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002014}
2015
2016struct rt6_mtu_change_arg
2017{
2018 struct net_device *dev;
2019 unsigned mtu;
2020};
2021
2022static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg)
2023{
2024 struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg;
2025 struct inet6_dev *idev;
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09002026 struct net *net = dev_net(arg->dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002027
2028 /* In IPv6 pmtu discovery is not optional,
2029 so that RTAX_MTU lock cannot disable it.
2030 We still use this lock to block changes
2031 caused by addrconf/ndisc.
2032 */
2033
2034 idev = __in6_dev_get(arg->dev);
2035 if (idev == NULL)
2036 return 0;
2037
2038 /* For administrative MTU increase, there is no way to discover
2039 IPv6 PMTU increase, so PMTU increase should be updated here.
2040 Since RFC 1981 doesn't include administrative MTU increase
2041 update PMTU increase is a MUST. (i.e. jumbo frame)
2042 */
2043 /*
2044 If new MTU is less than route PMTU, this new MTU will be the
2045 lowest MTU in the path, update the route PMTU to reflect PMTU
2046 decreases; if new MTU is greater than route PMTU, and the
2047 old MTU is the lowest MTU in the path, update the route PMTU
2048 to reflect the increase. In this case if the other nodes' MTU
2049 also have the lowest MTU, TOO BIG MESSAGE will be lead to
2050 PMTU discouvery.
2051 */
2052 if (rt->rt6i_dev == arg->dev &&
Changli Gaod8d1f302010-06-10 23:31:35 -07002053 !dst_metric_locked(&rt->dst, RTAX_MTU) &&
2054 (dst_mtu(&rt->dst) >= arg->mtu ||
2055 (dst_mtu(&rt->dst) < arg->mtu &&
2056 dst_mtu(&rt->dst) == idev->cnf.mtu6))) {
2057 rt->dst.metrics[RTAX_MTU-1] = arg->mtu;
2058 rt->dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(net, arg->mtu);
Simon Arlott566cfd82007-07-26 00:09:55 -07002059 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07002060 return 0;
2061}
2062
2063void rt6_mtu_change(struct net_device *dev, unsigned mtu)
2064{
Thomas Grafc71099a2006-08-04 23:20:06 -07002065 struct rt6_mtu_change_arg arg = {
2066 .dev = dev,
2067 .mtu = mtu,
2068 };
Linus Torvalds1da177e2005-04-16 15:20:36 -07002069
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09002070 fib6_clean_all(dev_net(dev), rt6_mtu_change_route, 0, &arg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002071}
2072
Patrick McHardyef7c79e2007-06-05 12:38:30 -07002073static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = {
Thomas Graf5176f912006-08-26 20:13:18 -07002074 [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) },
Thomas Graf86872cb2006-08-22 00:01:08 -07002075 [RTA_OIF] = { .type = NLA_U32 },
Thomas Grafab364a62006-08-22 00:01:47 -07002076 [RTA_IIF] = { .type = NLA_U32 },
Thomas Graf86872cb2006-08-22 00:01:08 -07002077 [RTA_PRIORITY] = { .type = NLA_U32 },
2078 [RTA_METRICS] = { .type = NLA_NESTED },
2079};
2080
2081static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh,
2082 struct fib6_config *cfg)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002083{
Thomas Graf86872cb2006-08-22 00:01:08 -07002084 struct rtmsg *rtm;
2085 struct nlattr *tb[RTA_MAX+1];
2086 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002087
Thomas Graf86872cb2006-08-22 00:01:08 -07002088 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2089 if (err < 0)
2090 goto errout;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002091
Thomas Graf86872cb2006-08-22 00:01:08 -07002092 err = -EINVAL;
2093 rtm = nlmsg_data(nlh);
2094 memset(cfg, 0, sizeof(*cfg));
2095
2096 cfg->fc_table = rtm->rtm_table;
2097 cfg->fc_dst_len = rtm->rtm_dst_len;
2098 cfg->fc_src_len = rtm->rtm_src_len;
2099 cfg->fc_flags = RTF_UP;
2100 cfg->fc_protocol = rtm->rtm_protocol;
2101
2102 if (rtm->rtm_type == RTN_UNREACHABLE)
2103 cfg->fc_flags |= RTF_REJECT;
2104
2105 cfg->fc_nlinfo.pid = NETLINK_CB(skb).pid;
2106 cfg->fc_nlinfo.nlh = nlh;
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +09002107 cfg->fc_nlinfo.nl_net = sock_net(skb->sk);
Thomas Graf86872cb2006-08-22 00:01:08 -07002108
2109 if (tb[RTA_GATEWAY]) {
2110 nla_memcpy(&cfg->fc_gateway, tb[RTA_GATEWAY], 16);
2111 cfg->fc_flags |= RTF_GATEWAY;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002112 }
Thomas Graf86872cb2006-08-22 00:01:08 -07002113
2114 if (tb[RTA_DST]) {
2115 int plen = (rtm->rtm_dst_len + 7) >> 3;
2116
2117 if (nla_len(tb[RTA_DST]) < plen)
2118 goto errout;
2119
2120 nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002121 }
Thomas Graf86872cb2006-08-22 00:01:08 -07002122
2123 if (tb[RTA_SRC]) {
2124 int plen = (rtm->rtm_src_len + 7) >> 3;
2125
2126 if (nla_len(tb[RTA_SRC]) < plen)
2127 goto errout;
2128
2129 nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002130 }
Thomas Graf86872cb2006-08-22 00:01:08 -07002131
2132 if (tb[RTA_OIF])
2133 cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]);
2134
2135 if (tb[RTA_PRIORITY])
2136 cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]);
2137
2138 if (tb[RTA_METRICS]) {
2139 cfg->fc_mx = nla_data(tb[RTA_METRICS]);
2140 cfg->fc_mx_len = nla_len(tb[RTA_METRICS]);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002141 }
Thomas Graf86872cb2006-08-22 00:01:08 -07002142
2143 if (tb[RTA_TABLE])
2144 cfg->fc_table = nla_get_u32(tb[RTA_TABLE]);
2145
2146 err = 0;
2147errout:
2148 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002149}
2150
Thomas Grafc127ea22007-03-22 11:58:32 -07002151static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002152{
Thomas Graf86872cb2006-08-22 00:01:08 -07002153 struct fib6_config cfg;
2154 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002155
Thomas Graf86872cb2006-08-22 00:01:08 -07002156 err = rtm_to_fib6_config(skb, nlh, &cfg);
2157 if (err < 0)
2158 return err;
2159
2160 return ip6_route_del(&cfg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002161}
2162
Thomas Grafc127ea22007-03-22 11:58:32 -07002163static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002164{
Thomas Graf86872cb2006-08-22 00:01:08 -07002165 struct fib6_config cfg;
2166 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002167
Thomas Graf86872cb2006-08-22 00:01:08 -07002168 err = rtm_to_fib6_config(skb, nlh, &cfg);
2169 if (err < 0)
2170 return err;
2171
2172 return ip6_route_add(&cfg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002173}
2174
Thomas Graf339bf982006-11-10 14:10:15 -08002175static inline size_t rt6_nlmsg_size(void)
2176{
2177 return NLMSG_ALIGN(sizeof(struct rtmsg))
2178 + nla_total_size(16) /* RTA_SRC */
2179 + nla_total_size(16) /* RTA_DST */
2180 + nla_total_size(16) /* RTA_GATEWAY */
2181 + nla_total_size(16) /* RTA_PREFSRC */
2182 + nla_total_size(4) /* RTA_TABLE */
2183 + nla_total_size(4) /* RTA_IIF */
2184 + nla_total_size(4) /* RTA_OIF */
2185 + nla_total_size(4) /* RTA_PRIORITY */
Noriaki TAKAMIYA6a2b9ce2007-01-23 22:09:41 -08002186 + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */
Thomas Graf339bf982006-11-10 14:10:15 -08002187 + nla_total_size(sizeof(struct rta_cacheinfo));
2188}
2189
Brian Haley191cd582008-08-14 15:33:21 -07002190static int rt6_fill_node(struct net *net,
2191 struct sk_buff *skb, struct rt6_info *rt,
Jamal Hadi Salim0d51aa82005-06-21 13:51:04 -07002192 struct in6_addr *dst, struct in6_addr *src,
2193 int iif, int type, u32 pid, u32 seq,
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09002194 int prefix, int nowait, unsigned int flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002195{
2196 struct rtmsg *rtm;
Thomas Graf2d7202b2006-08-22 00:01:27 -07002197 struct nlmsghdr *nlh;
Thomas Grafe3703b32006-11-27 09:27:07 -08002198 long expires;
Patrick McHardy9e762a42006-08-10 23:09:48 -07002199 u32 table;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002200
2201 if (prefix) { /* user wants prefix routes only */
2202 if (!(rt->rt6i_flags & RTF_PREFIX_RT)) {
2203 /* success since this is not a prefix route */
2204 return 1;
2205 }
2206 }
2207
Thomas Graf2d7202b2006-08-22 00:01:27 -07002208 nlh = nlmsg_put(skb, pid, seq, type, sizeof(*rtm), flags);
2209 if (nlh == NULL)
Patrick McHardy26932562007-01-31 23:16:40 -08002210 return -EMSGSIZE;
Thomas Graf2d7202b2006-08-22 00:01:27 -07002211
2212 rtm = nlmsg_data(nlh);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002213 rtm->rtm_family = AF_INET6;
2214 rtm->rtm_dst_len = rt->rt6i_dst.plen;
2215 rtm->rtm_src_len = rt->rt6i_src.plen;
2216 rtm->rtm_tos = 0;
Thomas Grafc71099a2006-08-04 23:20:06 -07002217 if (rt->rt6i_table)
Patrick McHardy9e762a42006-08-10 23:09:48 -07002218 table = rt->rt6i_table->tb6_id;
Thomas Grafc71099a2006-08-04 23:20:06 -07002219 else
Patrick McHardy9e762a42006-08-10 23:09:48 -07002220 table = RT6_TABLE_UNSPEC;
2221 rtm->rtm_table = table;
Thomas Graf2d7202b2006-08-22 00:01:27 -07002222 NLA_PUT_U32(skb, RTA_TABLE, table);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002223 if (rt->rt6i_flags&RTF_REJECT)
2224 rtm->rtm_type = RTN_UNREACHABLE;
2225 else if (rt->rt6i_dev && (rt->rt6i_dev->flags&IFF_LOOPBACK))
2226 rtm->rtm_type = RTN_LOCAL;
2227 else
2228 rtm->rtm_type = RTN_UNICAST;
2229 rtm->rtm_flags = 0;
2230 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
2231 rtm->rtm_protocol = rt->rt6i_protocol;
2232 if (rt->rt6i_flags&RTF_DYNAMIC)
2233 rtm->rtm_protocol = RTPROT_REDIRECT;
2234 else if (rt->rt6i_flags & RTF_ADDRCONF)
2235 rtm->rtm_protocol = RTPROT_KERNEL;
2236 else if (rt->rt6i_flags&RTF_DEFAULT)
2237 rtm->rtm_protocol = RTPROT_RA;
2238
2239 if (rt->rt6i_flags&RTF_CACHE)
2240 rtm->rtm_flags |= RTM_F_CLONED;
2241
2242 if (dst) {
Thomas Graf2d7202b2006-08-22 00:01:27 -07002243 NLA_PUT(skb, RTA_DST, 16, dst);
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09002244 rtm->rtm_dst_len = 128;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002245 } else if (rtm->rtm_dst_len)
Thomas Graf2d7202b2006-08-22 00:01:27 -07002246 NLA_PUT(skb, RTA_DST, 16, &rt->rt6i_dst.addr);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002247#ifdef CONFIG_IPV6_SUBTREES
2248 if (src) {
Thomas Graf2d7202b2006-08-22 00:01:27 -07002249 NLA_PUT(skb, RTA_SRC, 16, src);
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09002250 rtm->rtm_src_len = 128;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002251 } else if (rtm->rtm_src_len)
Thomas Graf2d7202b2006-08-22 00:01:27 -07002252 NLA_PUT(skb, RTA_SRC, 16, &rt->rt6i_src.addr);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002253#endif
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09002254 if (iif) {
2255#ifdef CONFIG_IPV6_MROUTE
2256 if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) {
Benjamin Thery8229efd2008-12-10 16:30:15 -08002257 int err = ip6mr_get_route(net, skb, rtm, nowait);
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09002258 if (err <= 0) {
2259 if (!nowait) {
2260 if (err == 0)
2261 return 0;
2262 goto nla_put_failure;
2263 } else {
2264 if (err == -EMSGSIZE)
2265 goto nla_put_failure;
2266 }
2267 }
2268 } else
2269#endif
2270 NLA_PUT_U32(skb, RTA_IIF, iif);
2271 } else if (dst) {
Changli Gaod8d1f302010-06-10 23:31:35 -07002272 struct inet6_dev *idev = ip6_dst_idev(&rt->dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002273 struct in6_addr saddr_buf;
Brian Haley191cd582008-08-14 15:33:21 -07002274 if (ipv6_dev_get_saddr(net, idev ? idev->dev : NULL,
YOSHIFUJI Hideaki7cbca672008-03-25 09:37:42 +09002275 dst, 0, &saddr_buf) == 0)
Thomas Graf2d7202b2006-08-22 00:01:27 -07002276 NLA_PUT(skb, RTA_PREFSRC, 16, &saddr_buf);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002277 }
Thomas Graf2d7202b2006-08-22 00:01:27 -07002278
Changli Gaod8d1f302010-06-10 23:31:35 -07002279 if (rtnetlink_put_metrics(skb, rt->dst.metrics) < 0)
Thomas Graf2d7202b2006-08-22 00:01:27 -07002280 goto nla_put_failure;
2281
Changli Gaod8d1f302010-06-10 23:31:35 -07002282 if (rt->dst.neighbour)
2283 NLA_PUT(skb, RTA_GATEWAY, 16, &rt->dst.neighbour->primary_key);
Thomas Graf2d7202b2006-08-22 00:01:27 -07002284
Changli Gaod8d1f302010-06-10 23:31:35 -07002285 if (rt->dst.dev)
Thomas Graf2d7202b2006-08-22 00:01:27 -07002286 NLA_PUT_U32(skb, RTA_OIF, rt->rt6i_dev->ifindex);
2287
2288 NLA_PUT_U32(skb, RTA_PRIORITY, rt->rt6i_metric);
Thomas Grafe3703b32006-11-27 09:27:07 -08002289
YOSHIFUJI Hideaki36e3dea2008-05-13 02:52:55 +09002290 if (!(rt->rt6i_flags & RTF_EXPIRES))
2291 expires = 0;
2292 else if (rt->rt6i_expires - jiffies < INT_MAX)
2293 expires = rt->rt6i_expires - jiffies;
2294 else
2295 expires = INT_MAX;
YOSHIFUJI Hideaki69cdf8f2008-05-19 16:55:13 -07002296
Changli Gaod8d1f302010-06-10 23:31:35 -07002297 if (rtnl_put_cacheinfo(skb, &rt->dst, 0, 0, 0,
2298 expires, rt->dst.error) < 0)
Thomas Grafe3703b32006-11-27 09:27:07 -08002299 goto nla_put_failure;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002300
Thomas Graf2d7202b2006-08-22 00:01:27 -07002301 return nlmsg_end(skb, nlh);
2302
2303nla_put_failure:
Patrick McHardy26932562007-01-31 23:16:40 -08002304 nlmsg_cancel(skb, nlh);
2305 return -EMSGSIZE;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002306}
2307
Patrick McHardy1b43af52006-08-10 23:11:17 -07002308int rt6_dump_route(struct rt6_info *rt, void *p_arg)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002309{
2310 struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg;
2311 int prefix;
2312
Thomas Graf2d7202b2006-08-22 00:01:27 -07002313 if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) {
2314 struct rtmsg *rtm = nlmsg_data(arg->cb->nlh);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002315 prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0;
2316 } else
2317 prefix = 0;
2318
Brian Haley191cd582008-08-14 15:33:21 -07002319 return rt6_fill_node(arg->net,
2320 arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002321 NETLINK_CB(arg->cb->skb).pid, arg->cb->nlh->nlmsg_seq,
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09002322 prefix, 0, NLM_F_MULTI);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002323}
2324
Thomas Grafc127ea22007-03-22 11:58:32 -07002325static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr* nlh, void *arg)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002326{
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +09002327 struct net *net = sock_net(in_skb->sk);
Thomas Grafab364a62006-08-22 00:01:47 -07002328 struct nlattr *tb[RTA_MAX+1];
Linus Torvalds1da177e2005-04-16 15:20:36 -07002329 struct rt6_info *rt;
Thomas Grafab364a62006-08-22 00:01:47 -07002330 struct sk_buff *skb;
2331 struct rtmsg *rtm;
2332 struct flowi fl;
2333 int err, iif = 0;
2334
2335 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2336 if (err < 0)
2337 goto errout;
2338
2339 err = -EINVAL;
2340 memset(&fl, 0, sizeof(fl));
2341
2342 if (tb[RTA_SRC]) {
2343 if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr))
2344 goto errout;
2345
2346 ipv6_addr_copy(&fl.fl6_src, nla_data(tb[RTA_SRC]));
2347 }
2348
2349 if (tb[RTA_DST]) {
2350 if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr))
2351 goto errout;
2352
2353 ipv6_addr_copy(&fl.fl6_dst, nla_data(tb[RTA_DST]));
2354 }
2355
2356 if (tb[RTA_IIF])
2357 iif = nla_get_u32(tb[RTA_IIF]);
2358
2359 if (tb[RTA_OIF])
2360 fl.oif = nla_get_u32(tb[RTA_OIF]);
2361
2362 if (iif) {
2363 struct net_device *dev;
Daniel Lezcano55786892008-03-04 13:47:47 -08002364 dev = __dev_get_by_index(net, iif);
Thomas Grafab364a62006-08-22 00:01:47 -07002365 if (!dev) {
2366 err = -ENODEV;
2367 goto errout;
2368 }
2369 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07002370
2371 skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
Thomas Grafab364a62006-08-22 00:01:47 -07002372 if (skb == NULL) {
2373 err = -ENOBUFS;
2374 goto errout;
2375 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07002376
2377 /* Reserve room for dummy headers, this skb can pass
2378 through good chunk of routing engine.
2379 */
Arnaldo Carvalho de Melo459a98e2007-03-19 15:30:44 -07002380 skb_reset_mac_header(skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002381 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr));
2382
Daniel Lezcano8a3edd82008-03-07 11:14:16 -08002383 rt = (struct rt6_info*) ip6_route_output(net, NULL, &fl);
Changli Gaod8d1f302010-06-10 23:31:35 -07002384 skb_dst_set(skb, &rt->dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002385
Brian Haley191cd582008-08-14 15:33:21 -07002386 err = rt6_fill_node(net, skb, rt, &fl.fl6_dst, &fl.fl6_src, iif,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002387 RTM_NEWROUTE, NETLINK_CB(in_skb).pid,
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09002388 nlh->nlmsg_seq, 0, 0, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002389 if (err < 0) {
Thomas Grafab364a62006-08-22 00:01:47 -07002390 kfree_skb(skb);
2391 goto errout;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002392 }
2393
Daniel Lezcano55786892008-03-04 13:47:47 -08002394 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).pid);
Thomas Grafab364a62006-08-22 00:01:47 -07002395errout:
Linus Torvalds1da177e2005-04-16 15:20:36 -07002396 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002397}
2398
Thomas Graf86872cb2006-08-22 00:01:08 -07002399void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002400{
2401 struct sk_buff *skb;
Daniel Lezcano55786892008-03-04 13:47:47 -08002402 struct net *net = info->nl_net;
Denis V. Lunev528c4ce2007-12-13 09:45:12 -08002403 u32 seq;
2404 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002405
Denis V. Lunev528c4ce2007-12-13 09:45:12 -08002406 err = -ENOBUFS;
2407 seq = info->nlh != NULL ? info->nlh->nlmsg_seq : 0;
Thomas Graf86872cb2006-08-22 00:01:08 -07002408
Thomas Graf339bf982006-11-10 14:10:15 -08002409 skb = nlmsg_new(rt6_nlmsg_size(), gfp_any());
Thomas Graf21713eb2006-08-15 00:35:24 -07002410 if (skb == NULL)
2411 goto errout;
2412
Brian Haley191cd582008-08-14 15:33:21 -07002413 err = rt6_fill_node(net, skb, rt, NULL, NULL, 0,
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09002414 event, info->pid, seq, 0, 0, 0);
Patrick McHardy26932562007-01-31 23:16:40 -08002415 if (err < 0) {
2416 /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */
2417 WARN_ON(err == -EMSGSIZE);
2418 kfree_skb(skb);
2419 goto errout;
2420 }
Pablo Neira Ayuso1ce85fe2009-02-24 23:18:28 -08002421 rtnl_notify(skb, net, info->pid, RTNLGRP_IPV6_ROUTE,
2422 info->nlh, gfp_any());
2423 return;
Thomas Graf21713eb2006-08-15 00:35:24 -07002424errout:
2425 if (err < 0)
Daniel Lezcano55786892008-03-04 13:47:47 -08002426 rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002427}
2428
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002429static int ip6_route_dev_notify(struct notifier_block *this,
2430 unsigned long event, void *data)
2431{
2432 struct net_device *dev = (struct net_device *)data;
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09002433 struct net *net = dev_net(dev);
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002434
2435 if (event == NETDEV_REGISTER && (dev->flags & IFF_LOOPBACK)) {
Changli Gaod8d1f302010-06-10 23:31:35 -07002436 net->ipv6.ip6_null_entry->dst.dev = dev;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002437 net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev);
2438#ifdef CONFIG_IPV6_MULTIPLE_TABLES
Changli Gaod8d1f302010-06-10 23:31:35 -07002439 net->ipv6.ip6_prohibit_entry->dst.dev = dev;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002440 net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev);
Changli Gaod8d1f302010-06-10 23:31:35 -07002441 net->ipv6.ip6_blk_hole_entry->dst.dev = dev;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002442 net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev);
2443#endif
2444 }
2445
2446 return NOTIFY_OK;
2447}
2448
Linus Torvalds1da177e2005-04-16 15:20:36 -07002449/*
2450 * /proc
2451 */
2452
2453#ifdef CONFIG_PROC_FS
2454
2455#define RT6_INFO_LEN (32 + 4 + 32 + 4 + 32 + 40 + 5 + 1)
2456
2457struct rt6_proc_arg
2458{
2459 char *buffer;
2460 int offset;
2461 int length;
2462 int skip;
2463 int len;
2464};
2465
2466static int rt6_info_route(struct rt6_info *rt, void *p_arg)
2467{
Alexey Dobriyan33120b32007-11-06 05:27:11 -08002468 struct seq_file *m = p_arg;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002469
Harvey Harrison4b7a4272008-10-29 12:50:24 -07002470 seq_printf(m, "%pi6 %02x ", &rt->rt6i_dst.addr, rt->rt6i_dst.plen);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002471
2472#ifdef CONFIG_IPV6_SUBTREES
Harvey Harrison4b7a4272008-10-29 12:50:24 -07002473 seq_printf(m, "%pi6 %02x ", &rt->rt6i_src.addr, rt->rt6i_src.plen);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002474#else
Alexey Dobriyan33120b32007-11-06 05:27:11 -08002475 seq_puts(m, "00000000000000000000000000000000 00 ");
Linus Torvalds1da177e2005-04-16 15:20:36 -07002476#endif
2477
2478 if (rt->rt6i_nexthop) {
Harvey Harrison4b7a4272008-10-29 12:50:24 -07002479 seq_printf(m, "%pi6", rt->rt6i_nexthop->primary_key);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002480 } else {
Alexey Dobriyan33120b32007-11-06 05:27:11 -08002481 seq_puts(m, "00000000000000000000000000000000");
Linus Torvalds1da177e2005-04-16 15:20:36 -07002482 }
Alexey Dobriyan33120b32007-11-06 05:27:11 -08002483 seq_printf(m, " %08x %08x %08x %08x %8s\n",
Changli Gaod8d1f302010-06-10 23:31:35 -07002484 rt->rt6i_metric, atomic_read(&rt->dst.__refcnt),
2485 rt->dst.__use, rt->rt6i_flags,
Alexey Dobriyan33120b32007-11-06 05:27:11 -08002486 rt->rt6i_dev ? rt->rt6i_dev->name : "");
Linus Torvalds1da177e2005-04-16 15:20:36 -07002487 return 0;
2488}
2489
Alexey Dobriyan33120b32007-11-06 05:27:11 -08002490static int ipv6_route_show(struct seq_file *m, void *v)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002491{
Daniel Lezcanof3db4852008-03-03 23:27:06 -08002492 struct net *net = (struct net *)m->private;
2493 fib6_clean_all(net, rt6_info_route, 0, m);
Alexey Dobriyan33120b32007-11-06 05:27:11 -08002494 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002495}
2496
Alexey Dobriyan33120b32007-11-06 05:27:11 -08002497static int ipv6_route_open(struct inode *inode, struct file *file)
2498{
Pavel Emelyanovde05c552008-07-18 04:07:21 -07002499 return single_open_net(inode, file, ipv6_route_show);
Daniel Lezcanof3db4852008-03-03 23:27:06 -08002500}
2501
Alexey Dobriyan33120b32007-11-06 05:27:11 -08002502static const struct file_operations ipv6_route_proc_fops = {
2503 .owner = THIS_MODULE,
2504 .open = ipv6_route_open,
2505 .read = seq_read,
2506 .llseek = seq_lseek,
Pavel Emelyanovb6fcbdb2008-07-18 04:07:44 -07002507 .release = single_release_net,
Alexey Dobriyan33120b32007-11-06 05:27:11 -08002508};
2509
Linus Torvalds1da177e2005-04-16 15:20:36 -07002510static int rt6_stats_seq_show(struct seq_file *seq, void *v)
2511{
Daniel Lezcano69ddb802008-03-04 13:46:23 -08002512 struct net *net = (struct net *)seq->private;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002513 seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n",
Daniel Lezcano69ddb802008-03-04 13:46:23 -08002514 net->ipv6.rt6_stats->fib_nodes,
2515 net->ipv6.rt6_stats->fib_route_nodes,
2516 net->ipv6.rt6_stats->fib_rt_alloc,
2517 net->ipv6.rt6_stats->fib_rt_entries,
2518 net->ipv6.rt6_stats->fib_rt_cache,
Alexey Dobriyan86393e52009-08-29 01:34:49 +00002519 atomic_read(&net->ipv6.ip6_dst_ops.entries),
Daniel Lezcano69ddb802008-03-04 13:46:23 -08002520 net->ipv6.rt6_stats->fib_discarded_routes);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002521
2522 return 0;
2523}
2524
2525static int rt6_stats_seq_open(struct inode *inode, struct file *file)
2526{
Pavel Emelyanovde05c552008-07-18 04:07:21 -07002527 return single_open_net(inode, file, rt6_stats_seq_show);
Daniel Lezcano69ddb802008-03-04 13:46:23 -08002528}
2529
Arjan van de Ven9a321442007-02-12 00:55:35 -08002530static const struct file_operations rt6_stats_seq_fops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002531 .owner = THIS_MODULE,
2532 .open = rt6_stats_seq_open,
2533 .read = seq_read,
2534 .llseek = seq_lseek,
Pavel Emelyanovb6fcbdb2008-07-18 04:07:44 -07002535 .release = single_release_net,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002536};
2537#endif /* CONFIG_PROC_FS */
2538
2539#ifdef CONFIG_SYSCTL
2540
Linus Torvalds1da177e2005-04-16 15:20:36 -07002541static
Alexey Dobriyan8d65af72009-09-23 15:57:19 -07002542int ipv6_sysctl_rtcache_flush(ctl_table *ctl, int write,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002543 void __user *buffer, size_t *lenp, loff_t *ppos)
2544{
Daniel Lezcano5b7c9312008-03-03 23:28:58 -08002545 struct net *net = current->nsproxy->net_ns;
2546 int delay = net->ipv6.sysctl.flush_delay;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002547 if (write) {
Alexey Dobriyan8d65af72009-09-23 15:57:19 -07002548 proc_dointvec(ctl, write, buffer, lenp, ppos);
Daniel Lezcano5b7c9312008-03-03 23:28:58 -08002549 fib6_run_gc(delay <= 0 ? ~0UL : (unsigned long)delay, net);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002550 return 0;
2551 } else
2552 return -EINVAL;
2553}
2554
Daniel Lezcano760f2d02008-01-10 02:53:43 -08002555ctl_table ipv6_route_table_template[] = {
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09002556 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002557 .procname = "flush",
Daniel Lezcano49905092008-01-10 03:01:01 -08002558 .data = &init_net.ipv6.sysctl.flush_delay,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002559 .maxlen = sizeof(int),
Dave Jones89c8b3a12005-04-28 12:11:49 -07002560 .mode = 0200,
Alexey Dobriyan6d9f2392008-11-03 18:21:05 -08002561 .proc_handler = ipv6_sysctl_rtcache_flush
Linus Torvalds1da177e2005-04-16 15:20:36 -07002562 },
2563 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002564 .procname = "gc_thresh",
Daniel Lezcano9a7ec3a2008-03-04 13:48:53 -08002565 .data = &ip6_dst_ops_template.gc_thresh,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002566 .maxlen = sizeof(int),
2567 .mode = 0644,
Alexey Dobriyan6d9f2392008-11-03 18:21:05 -08002568 .proc_handler = proc_dointvec,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002569 },
2570 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002571 .procname = "max_size",
Daniel Lezcano49905092008-01-10 03:01:01 -08002572 .data = &init_net.ipv6.sysctl.ip6_rt_max_size,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002573 .maxlen = sizeof(int),
2574 .mode = 0644,
Alexey Dobriyan6d9f2392008-11-03 18:21:05 -08002575 .proc_handler = proc_dointvec,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002576 },
2577 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002578 .procname = "gc_min_interval",
Daniel Lezcano49905092008-01-10 03:01:01 -08002579 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002580 .maxlen = sizeof(int),
2581 .mode = 0644,
Alexey Dobriyan6d9f2392008-11-03 18:21:05 -08002582 .proc_handler = proc_dointvec_jiffies,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002583 },
2584 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002585 .procname = "gc_timeout",
Daniel Lezcano49905092008-01-10 03:01:01 -08002586 .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002587 .maxlen = sizeof(int),
2588 .mode = 0644,
Alexey Dobriyan6d9f2392008-11-03 18:21:05 -08002589 .proc_handler = proc_dointvec_jiffies,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002590 },
2591 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002592 .procname = "gc_interval",
Daniel Lezcano49905092008-01-10 03:01:01 -08002593 .data = &init_net.ipv6.sysctl.ip6_rt_gc_interval,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002594 .maxlen = sizeof(int),
2595 .mode = 0644,
Alexey Dobriyan6d9f2392008-11-03 18:21:05 -08002596 .proc_handler = proc_dointvec_jiffies,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002597 },
2598 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002599 .procname = "gc_elasticity",
Daniel Lezcano49905092008-01-10 03:01:01 -08002600 .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002601 .maxlen = sizeof(int),
2602 .mode = 0644,
Min Zhangf3d3f612010-08-14 22:42:51 -07002603 .proc_handler = proc_dointvec,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002604 },
2605 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002606 .procname = "mtu_expires",
Daniel Lezcano49905092008-01-10 03:01:01 -08002607 .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002608 .maxlen = sizeof(int),
2609 .mode = 0644,
Alexey Dobriyan6d9f2392008-11-03 18:21:05 -08002610 .proc_handler = proc_dointvec_jiffies,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002611 },
2612 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002613 .procname = "min_adv_mss",
Daniel Lezcano49905092008-01-10 03:01:01 -08002614 .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002615 .maxlen = sizeof(int),
2616 .mode = 0644,
Min Zhangf3d3f612010-08-14 22:42:51 -07002617 .proc_handler = proc_dointvec,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002618 },
2619 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002620 .procname = "gc_min_interval_ms",
Daniel Lezcano49905092008-01-10 03:01:01 -08002621 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002622 .maxlen = sizeof(int),
2623 .mode = 0644,
Alexey Dobriyan6d9f2392008-11-03 18:21:05 -08002624 .proc_handler = proc_dointvec_ms_jiffies,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002625 },
Eric W. Biedermanf8572d82009-11-05 13:32:03 -08002626 { }
Linus Torvalds1da177e2005-04-16 15:20:36 -07002627};
2628
Alexey Dobriyan2c8c1e72010-01-17 03:35:32 +00002629struct ctl_table * __net_init ipv6_route_sysctl_init(struct net *net)
Daniel Lezcano760f2d02008-01-10 02:53:43 -08002630{
2631 struct ctl_table *table;
2632
2633 table = kmemdup(ipv6_route_table_template,
2634 sizeof(ipv6_route_table_template),
2635 GFP_KERNEL);
YOSHIFUJI Hideaki5ee09102008-02-28 00:24:28 +09002636
2637 if (table) {
2638 table[0].data = &net->ipv6.sysctl.flush_delay;
Alexey Dobriyan86393e52009-08-29 01:34:49 +00002639 table[1].data = &net->ipv6.ip6_dst_ops.gc_thresh;
YOSHIFUJI Hideaki5ee09102008-02-28 00:24:28 +09002640 table[2].data = &net->ipv6.sysctl.ip6_rt_max_size;
2641 table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
2642 table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout;
2643 table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval;
2644 table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity;
2645 table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires;
2646 table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss;
Alexey Dobriyan9c69fab2009-12-18 20:11:03 -08002647 table[9].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
YOSHIFUJI Hideaki5ee09102008-02-28 00:24:28 +09002648 }
2649
Daniel Lezcano760f2d02008-01-10 02:53:43 -08002650 return table;
2651}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002652#endif
2653
Alexey Dobriyan2c8c1e72010-01-17 03:35:32 +00002654static int __net_init ip6_route_net_init(struct net *net)
Daniel Lezcanocdb18762008-03-04 13:45:33 -08002655{
Pavel Emelyanov633d424b2008-04-21 14:25:23 -07002656 int ret = -ENOMEM;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002657
Alexey Dobriyan86393e52009-08-29 01:34:49 +00002658 memcpy(&net->ipv6.ip6_dst_ops, &ip6_dst_ops_template,
2659 sizeof(net->ipv6.ip6_dst_ops));
Benjamin Theryf2fc6a52008-03-04 13:49:23 -08002660
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002661 net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template,
2662 sizeof(*net->ipv6.ip6_null_entry),
2663 GFP_KERNEL);
2664 if (!net->ipv6.ip6_null_entry)
Benjamin Theryf2fc6a52008-03-04 13:49:23 -08002665 goto out_ip6_dst_ops;
Changli Gaod8d1f302010-06-10 23:31:35 -07002666 net->ipv6.ip6_null_entry->dst.path =
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002667 (struct dst_entry *)net->ipv6.ip6_null_entry;
Changli Gaod8d1f302010-06-10 23:31:35 -07002668 net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002669
2670#ifdef CONFIG_IPV6_MULTIPLE_TABLES
2671 net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template,
2672 sizeof(*net->ipv6.ip6_prohibit_entry),
2673 GFP_KERNEL);
Peter Zijlstra68fffc62008-10-07 14:12:10 -07002674 if (!net->ipv6.ip6_prohibit_entry)
2675 goto out_ip6_null_entry;
Changli Gaod8d1f302010-06-10 23:31:35 -07002676 net->ipv6.ip6_prohibit_entry->dst.path =
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002677 (struct dst_entry *)net->ipv6.ip6_prohibit_entry;
Changli Gaod8d1f302010-06-10 23:31:35 -07002678 net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002679
2680 net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template,
2681 sizeof(*net->ipv6.ip6_blk_hole_entry),
2682 GFP_KERNEL);
Peter Zijlstra68fffc62008-10-07 14:12:10 -07002683 if (!net->ipv6.ip6_blk_hole_entry)
2684 goto out_ip6_prohibit_entry;
Changli Gaod8d1f302010-06-10 23:31:35 -07002685 net->ipv6.ip6_blk_hole_entry->dst.path =
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002686 (struct dst_entry *)net->ipv6.ip6_blk_hole_entry;
Changli Gaod8d1f302010-06-10 23:31:35 -07002687 net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002688#endif
2689
Peter Zijlstrab339a47c2008-10-07 14:15:00 -07002690 net->ipv6.sysctl.flush_delay = 0;
2691 net->ipv6.sysctl.ip6_rt_max_size = 4096;
2692 net->ipv6.sysctl.ip6_rt_gc_min_interval = HZ / 2;
2693 net->ipv6.sysctl.ip6_rt_gc_timeout = 60*HZ;
2694 net->ipv6.sysctl.ip6_rt_gc_interval = 30*HZ;
2695 net->ipv6.sysctl.ip6_rt_gc_elasticity = 9;
2696 net->ipv6.sysctl.ip6_rt_mtu_expires = 10*60*HZ;
2697 net->ipv6.sysctl.ip6_rt_min_advmss = IPV6_MIN_MTU - 20 - 40;
2698
Daniel Lezcanocdb18762008-03-04 13:45:33 -08002699#ifdef CONFIG_PROC_FS
2700 proc_net_fops_create(net, "ipv6_route", 0, &ipv6_route_proc_fops);
2701 proc_net_fops_create(net, "rt6_stats", S_IRUGO, &rt6_stats_seq_fops);
2702#endif
Benjamin Thery6891a342008-03-04 13:49:47 -08002703 net->ipv6.ip6_rt_gc_expire = 30*HZ;
2704
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002705 ret = 0;
2706out:
2707 return ret;
Benjamin Theryf2fc6a52008-03-04 13:49:23 -08002708
Peter Zijlstra68fffc62008-10-07 14:12:10 -07002709#ifdef CONFIG_IPV6_MULTIPLE_TABLES
2710out_ip6_prohibit_entry:
2711 kfree(net->ipv6.ip6_prohibit_entry);
2712out_ip6_null_entry:
2713 kfree(net->ipv6.ip6_null_entry);
2714#endif
Benjamin Theryf2fc6a52008-03-04 13:49:23 -08002715out_ip6_dst_ops:
Benjamin Theryf2fc6a52008-03-04 13:49:23 -08002716 goto out;
Daniel Lezcanocdb18762008-03-04 13:45:33 -08002717}
2718
Alexey Dobriyan2c8c1e72010-01-17 03:35:32 +00002719static void __net_exit ip6_route_net_exit(struct net *net)
Daniel Lezcanocdb18762008-03-04 13:45:33 -08002720{
2721#ifdef CONFIG_PROC_FS
2722 proc_net_remove(net, "ipv6_route");
2723 proc_net_remove(net, "rt6_stats");
2724#endif
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002725 kfree(net->ipv6.ip6_null_entry);
2726#ifdef CONFIG_IPV6_MULTIPLE_TABLES
2727 kfree(net->ipv6.ip6_prohibit_entry);
2728 kfree(net->ipv6.ip6_blk_hole_entry);
2729#endif
Daniel Lezcanocdb18762008-03-04 13:45:33 -08002730}
2731
2732static struct pernet_operations ip6_route_net_ops = {
2733 .init = ip6_route_net_init,
2734 .exit = ip6_route_net_exit,
2735};
2736
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002737static struct notifier_block ip6_route_dev_notifier = {
2738 .notifier_call = ip6_route_dev_notify,
2739 .priority = 0,
2740};
2741
Daniel Lezcano433d49c2007-12-07 00:43:48 -08002742int __init ip6_route_init(void)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002743{
Daniel Lezcano433d49c2007-12-07 00:43:48 -08002744 int ret;
2745
Daniel Lezcano9a7ec3a2008-03-04 13:48:53 -08002746 ret = -ENOMEM;
2747 ip6_dst_ops_template.kmem_cachep =
2748 kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0,
2749 SLAB_HWCACHE_ALIGN, NULL);
2750 if (!ip6_dst_ops_template.kmem_cachep)
Fernando Carrijoc19a28e2009-01-07 18:09:08 -08002751 goto out;
David S. Miller14e50e52007-05-24 18:17:54 -07002752
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002753 ret = register_pernet_subsys(&ip6_route_net_ops);
2754 if (ret)
Daniel Lezcanobdb32892008-03-04 13:48:10 -08002755 goto out_kmem_cache;
Daniel Lezcanobdb32892008-03-04 13:48:10 -08002756
Arnaud Ebalard5dc121e2008-10-01 02:37:56 -07002757 ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep;
2758
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002759 /* Registering of the loopback is done before this portion of code,
2760 * the loopback reference in rt6_info will not be taken, do it
2761 * manually for init_net */
Changli Gaod8d1f302010-06-10 23:31:35 -07002762 init_net.ipv6.ip6_null_entry->dst.dev = init_net.loopback_dev;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002763 init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
2764 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
Changli Gaod8d1f302010-06-10 23:31:35 -07002765 init_net.ipv6.ip6_prohibit_entry->dst.dev = init_net.loopback_dev;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002766 init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
Changli Gaod8d1f302010-06-10 23:31:35 -07002767 init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002768 init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
2769 #endif
Daniel Lezcano433d49c2007-12-07 00:43:48 -08002770 ret = fib6_init();
2771 if (ret)
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002772 goto out_register_subsys;
Daniel Lezcano433d49c2007-12-07 00:43:48 -08002773
Daniel Lezcano433d49c2007-12-07 00:43:48 -08002774 ret = xfrm6_init();
2775 if (ret)
Daniel Lezcanocdb18762008-03-04 13:45:33 -08002776 goto out_fib6_init;
Daniel Lezcanoc35b7e72007-12-08 00:14:11 -08002777
Daniel Lezcano433d49c2007-12-07 00:43:48 -08002778 ret = fib6_rules_init();
2779 if (ret)
2780 goto xfrm6_init;
Daniel Lezcano7e5449c2007-12-08 00:14:54 -08002781
Daniel Lezcano433d49c2007-12-07 00:43:48 -08002782 ret = -ENOBUFS;
2783 if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL) ||
2784 __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL) ||
2785 __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL))
2786 goto fib6_rules_init;
2787
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002788 ret = register_netdevice_notifier(&ip6_route_dev_notifier);
Daniel Lezcanocdb18762008-03-04 13:45:33 -08002789 if (ret)
2790 goto fib6_rules_init;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002791
Daniel Lezcano433d49c2007-12-07 00:43:48 -08002792out:
2793 return ret;
2794
2795fib6_rules_init:
Daniel Lezcano433d49c2007-12-07 00:43:48 -08002796 fib6_rules_cleanup();
2797xfrm6_init:
Daniel Lezcano433d49c2007-12-07 00:43:48 -08002798 xfrm6_fini();
Daniel Lezcano433d49c2007-12-07 00:43:48 -08002799out_fib6_init:
Daniel Lezcano433d49c2007-12-07 00:43:48 -08002800 fib6_gc_cleanup();
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002801out_register_subsys:
2802 unregister_pernet_subsys(&ip6_route_net_ops);
Daniel Lezcano433d49c2007-12-07 00:43:48 -08002803out_kmem_cache:
Benjamin Theryf2fc6a52008-03-04 13:49:23 -08002804 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
Daniel Lezcano433d49c2007-12-07 00:43:48 -08002805 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002806}
2807
2808void ip6_route_cleanup(void)
2809{
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002810 unregister_netdevice_notifier(&ip6_route_dev_notifier);
Thomas Graf101367c2006-08-04 03:39:02 -07002811 fib6_rules_cleanup();
Linus Torvalds1da177e2005-04-16 15:20:36 -07002812 xfrm6_fini();
Linus Torvalds1da177e2005-04-16 15:20:36 -07002813 fib6_gc_cleanup();
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002814 unregister_pernet_subsys(&ip6_route_net_ops);
Benjamin Theryf2fc6a52008-03-04 13:49:23 -08002815 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002816}