blob: 969913da494fdf1d80ce674c3b6c421fdab18d3d [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
Joe Perchesf3213832012-05-15 14:11:53 +000027#define pr_fmt(fmt) "IPv6: " fmt
28
Randy Dunlap4fc268d2006-01-11 12:17:47 -080029#include <linux/capability.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070030#include <linux/errno.h>
Paul Gortmakerbc3b2d72011-07-15 11:47:34 -040031#include <linux/export.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070032#include <linux/types.h>
33#include <linux/times.h>
34#include <linux/socket.h>
35#include <linux/sockios.h>
36#include <linux/net.h>
37#include <linux/route.h>
38#include <linux/netdevice.h>
39#include <linux/in6.h>
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +090040#include <linux/mroute6.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070041#include <linux/init.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070042#include <linux/if_arp.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070043#include <linux/proc_fs.h>
44#include <linux/seq_file.h>
Daniel Lezcano5b7c9312008-03-03 23:28:58 -080045#include <linux/nsproxy.h>
Tejun Heo5a0e3ad2010-03-24 17:04:11 +090046#include <linux/slab.h>
Eric W. Biederman457c4cb2007-09-12 12:01:34 +020047#include <net/net_namespace.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070048#include <net/snmp.h>
49#include <net/ipv6.h>
50#include <net/ip6_fib.h>
51#include <net/ip6_route.h>
52#include <net/ndisc.h>
53#include <net/addrconf.h>
54#include <net/tcp.h>
55#include <linux/rtnetlink.h>
56#include <net/dst.h>
Jiri Benc904af042015-08-20 13:56:31 +020057#include <net/dst_metadata.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070058#include <net/xfrm.h>
Tom Tucker8d717402006-07-30 20:43:36 -070059#include <net/netevent.h>
Thomas Graf21713eb2006-08-15 00:35:24 -070060#include <net/netlink.h>
Nicolas Dichtel51ebd312012-10-22 03:42:09 +000061#include <net/nexthop.h>
Roopa Prabhu19e42e42015-07-21 10:43:48 +020062#include <net/lwtunnel.h>
Jiri Benc904af042015-08-20 13:56:31 +020063#include <net/ip_tunnels.h>
David Ahernca254492015-10-12 11:47:10 -070064#include <net/l3mdev.h>
David Ahernb8115802015-11-19 12:24:22 -080065#include <trace/events/fib6.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070066
67#include <asm/uaccess.h>
68
69#ifdef CONFIG_SYSCTL
70#include <linux/sysctl.h>
71#endif
72
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +020073enum rt6_nud_state {
Jiri Benc7e980562013-12-11 13:48:20 +010074 RT6_NUD_FAIL_HARD = -3,
75 RT6_NUD_FAIL_PROBE = -2,
76 RT6_NUD_FAIL_DO_RR = -1,
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +020077 RT6_NUD_SUCCEED = 1
78};
79
Martin KaFai Lau83a09ab2015-05-22 20:56:05 -070080static void ip6_rt_copy_init(struct rt6_info *rt, struct rt6_info *ort);
Linus Torvalds1da177e2005-04-16 15:20:36 -070081static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie);
David S. Miller0dbaee32010-12-13 12:52:14 -080082static unsigned int ip6_default_advmss(const struct dst_entry *dst);
Steffen Klassertebb762f2011-11-23 02:12:51 +000083static unsigned int ip6_mtu(const struct dst_entry *dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -070084static struct dst_entry *ip6_negative_advice(struct dst_entry *);
85static void ip6_dst_destroy(struct dst_entry *);
86static void ip6_dst_ifdown(struct dst_entry *,
87 struct net_device *dev, int how);
Daniel Lezcano569d3642008-01-18 03:56:57 -080088static int ip6_dst_gc(struct dst_ops *ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -070089
90static int ip6_pkt_discard(struct sk_buff *skb);
Eric W. Biedermanede20592015-10-07 16:48:47 -050091static int ip6_pkt_discard_out(struct net *net, struct sock *sk, struct sk_buff *skb);
Kamala R7150aed2013-12-02 19:55:21 +053092static int ip6_pkt_prohibit(struct sk_buff *skb);
Eric W. Biedermanede20592015-10-07 16:48:47 -050093static int ip6_pkt_prohibit_out(struct net *net, struct sock *sk, struct sk_buff *skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -070094static void ip6_link_failure(struct sk_buff *skb);
David S. Miller6700c272012-07-17 03:29:28 -070095static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
96 struct sk_buff *skb, u32 mtu);
97static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk,
98 struct sk_buff *skb);
Martin KaFai Lau4b32b5a2015-04-28 13:03:06 -070099static void rt6_dst_from_metrics_check(struct rt6_info *rt);
Nicolas Dichtel52bd4c02013-06-28 17:35:48 +0200100static int rt6_score_route(struct rt6_info *rt, int oif, int strict);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700101
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800102#ifdef CONFIG_IPV6_ROUTE_INFO
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -0800103static struct rt6_info *rt6_add_route_info(struct net *net,
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000104 const struct in6_addr *prefix, int prefixlen,
105 const struct in6_addr *gwaddr, int ifindex,
Eric Dumazet95c96172012-04-15 05:58:06 +0000106 unsigned int pref);
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -0800107static struct rt6_info *rt6_get_route_info(struct net *net,
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000108 const struct in6_addr *prefix, int prefixlen,
109 const struct in6_addr *gwaddr, int ifindex);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800110#endif
111
Martin KaFai Lau8d0b94a2015-05-22 20:56:04 -0700112struct uncached_list {
113 spinlock_t lock;
114 struct list_head head;
115};
116
117static DEFINE_PER_CPU_ALIGNED(struct uncached_list, rt6_uncached_list);
118
119static void rt6_uncached_list_add(struct rt6_info *rt)
120{
121 struct uncached_list *ul = raw_cpu_ptr(&rt6_uncached_list);
122
123 rt->dst.flags |= DST_NOCACHE;
124 rt->rt6i_uncached_list = ul;
125
126 spin_lock_bh(&ul->lock);
127 list_add_tail(&rt->rt6i_uncached, &ul->head);
128 spin_unlock_bh(&ul->lock);
129}
130
131static void rt6_uncached_list_del(struct rt6_info *rt)
132{
133 if (!list_empty(&rt->rt6i_uncached)) {
134 struct uncached_list *ul = rt->rt6i_uncached_list;
135
136 spin_lock_bh(&ul->lock);
137 list_del(&rt->rt6i_uncached);
138 spin_unlock_bh(&ul->lock);
139 }
140}
141
142static void rt6_uncached_list_flush_dev(struct net *net, struct net_device *dev)
143{
144 struct net_device *loopback_dev = net->loopback_dev;
145 int cpu;
146
Eric W. Biedermane332bc62015-10-12 11:02:08 -0500147 if (dev == loopback_dev)
148 return;
149
Martin KaFai Lau8d0b94a2015-05-22 20:56:04 -0700150 for_each_possible_cpu(cpu) {
151 struct uncached_list *ul = per_cpu_ptr(&rt6_uncached_list, cpu);
152 struct rt6_info *rt;
153
154 spin_lock_bh(&ul->lock);
155 list_for_each_entry(rt, &ul->head, rt6i_uncached) {
156 struct inet6_dev *rt_idev = rt->rt6i_idev;
157 struct net_device *rt_dev = rt->dst.dev;
158
Eric W. Biedermane332bc62015-10-12 11:02:08 -0500159 if (rt_idev->dev == dev) {
Martin KaFai Lau8d0b94a2015-05-22 20:56:04 -0700160 rt->rt6i_idev = in6_dev_get(loopback_dev);
161 in6_dev_put(rt_idev);
162 }
163
Eric W. Biedermane332bc62015-10-12 11:02:08 -0500164 if (rt_dev == dev) {
Martin KaFai Lau8d0b94a2015-05-22 20:56:04 -0700165 rt->dst.dev = loopback_dev;
166 dev_hold(rt->dst.dev);
167 dev_put(rt_dev);
168 }
169 }
170 spin_unlock_bh(&ul->lock);
171 }
172}
173
Martin KaFai Laud52d3992015-05-22 20:56:06 -0700174static u32 *rt6_pcpu_cow_metrics(struct rt6_info *rt)
175{
176 return dst_metrics_write_ptr(rt->dst.from);
177}
178
David S. Miller06582542011-01-27 14:58:42 -0800179static u32 *ipv6_cow_metrics(struct dst_entry *dst, unsigned long old)
180{
Martin KaFai Lau4b32b5a2015-04-28 13:03:06 -0700181 struct rt6_info *rt = (struct rt6_info *)dst;
David S. Miller06582542011-01-27 14:58:42 -0800182
Martin KaFai Laud52d3992015-05-22 20:56:06 -0700183 if (rt->rt6i_flags & RTF_PCPU)
184 return rt6_pcpu_cow_metrics(rt);
185 else if (rt->rt6i_flags & RTF_CACHE)
Martin KaFai Lau4b32b5a2015-04-28 13:03:06 -0700186 return NULL;
187 else
Martin KaFai Lau3b471172015-02-12 16:14:08 -0800188 return dst_cow_metrics_generic(dst, old);
David S. Miller06582542011-01-27 14:58:42 -0800189}
190
David S. Millerf894cbf2012-07-02 21:52:24 -0700191static inline const void *choose_neigh_daddr(struct rt6_info *rt,
192 struct sk_buff *skb,
193 const void *daddr)
David S. Miller39232972012-01-26 15:22:32 -0500194{
195 struct in6_addr *p = &rt->rt6i_gateway;
196
David S. Millera7563f32012-01-26 16:29:16 -0500197 if (!ipv6_addr_any(p))
David S. Miller39232972012-01-26 15:22:32 -0500198 return (const void *) p;
David S. Millerf894cbf2012-07-02 21:52:24 -0700199 else if (skb)
200 return &ipv6_hdr(skb)->daddr;
David S. Miller39232972012-01-26 15:22:32 -0500201 return daddr;
202}
203
David S. Millerf894cbf2012-07-02 21:52:24 -0700204static struct neighbour *ip6_neigh_lookup(const struct dst_entry *dst,
205 struct sk_buff *skb,
206 const void *daddr)
David S. Millerd3aaeb32011-07-18 00:40:17 -0700207{
David S. Miller39232972012-01-26 15:22:32 -0500208 struct rt6_info *rt = (struct rt6_info *) dst;
209 struct neighbour *n;
210
David S. Millerf894cbf2012-07-02 21:52:24 -0700211 daddr = choose_neigh_daddr(rt, skb, daddr);
YOSHIFUJI Hideaki / 吉藤英明8e022ee2013-01-17 12:53:09 +0000212 n = __ipv6_neigh_lookup(dst->dev, daddr);
David S. Millerf83c7792011-12-28 15:41:23 -0500213 if (n)
214 return n;
215 return neigh_create(&nd_tbl, daddr, dst->dev);
216}
217
Daniel Lezcano9a7ec3a2008-03-04 13:48:53 -0800218static struct dst_ops ip6_dst_ops_template = {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700219 .family = AF_INET6,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700220 .gc = ip6_dst_gc,
221 .gc_thresh = 1024,
222 .check = ip6_dst_check,
David S. Miller0dbaee32010-12-13 12:52:14 -0800223 .default_advmss = ip6_default_advmss,
Steffen Klassertebb762f2011-11-23 02:12:51 +0000224 .mtu = ip6_mtu,
David S. Miller06582542011-01-27 14:58:42 -0800225 .cow_metrics = ipv6_cow_metrics,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700226 .destroy = ip6_dst_destroy,
227 .ifdown = ip6_dst_ifdown,
228 .negative_advice = ip6_negative_advice,
229 .link_failure = ip6_link_failure,
230 .update_pmtu = ip6_rt_update_pmtu,
David S. Miller6e157b62012-07-12 00:05:02 -0700231 .redirect = rt6_do_redirect,
Eric W. Biederman9f8955c2015-10-07 16:48:39 -0500232 .local_out = __ip6_local_out,
David S. Millerd3aaeb32011-07-18 00:40:17 -0700233 .neigh_lookup = ip6_neigh_lookup,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700234};
235
Steffen Klassertebb762f2011-11-23 02:12:51 +0000236static unsigned int ip6_blackhole_mtu(const struct dst_entry *dst)
Roland Dreierec831ea2011-01-31 13:16:00 -0800237{
Steffen Klassert618f9bc2011-11-23 02:13:31 +0000238 unsigned int mtu = dst_metric_raw(dst, RTAX_MTU);
239
240 return mtu ? : dst->dev->mtu;
Roland Dreierec831ea2011-01-31 13:16:00 -0800241}
242
David S. Miller6700c272012-07-17 03:29:28 -0700243static void ip6_rt_blackhole_update_pmtu(struct dst_entry *dst, struct sock *sk,
244 struct sk_buff *skb, u32 mtu)
David S. Miller14e50e52007-05-24 18:17:54 -0700245{
246}
247
David S. Miller6700c272012-07-17 03:29:28 -0700248static void ip6_rt_blackhole_redirect(struct dst_entry *dst, struct sock *sk,
249 struct sk_buff *skb)
David S. Millerb587ee32012-07-12 00:39:24 -0700250{
251}
252
David S. Miller14e50e52007-05-24 18:17:54 -0700253static struct dst_ops ip6_dst_blackhole_ops = {
254 .family = AF_INET6,
David S. Miller14e50e52007-05-24 18:17:54 -0700255 .destroy = ip6_dst_destroy,
256 .check = ip6_dst_check,
Steffen Klassertebb762f2011-11-23 02:12:51 +0000257 .mtu = ip6_blackhole_mtu,
Eric Dumazet214f45c2011-02-18 11:39:01 -0800258 .default_advmss = ip6_default_advmss,
David S. Miller14e50e52007-05-24 18:17:54 -0700259 .update_pmtu = ip6_rt_blackhole_update_pmtu,
David S. Millerb587ee32012-07-12 00:39:24 -0700260 .redirect = ip6_rt_blackhole_redirect,
Martin KaFai Lau0a1f5962015-10-15 16:39:58 -0700261 .cow_metrics = dst_cow_metrics_generic,
David S. Millerd3aaeb32011-07-18 00:40:17 -0700262 .neigh_lookup = ip6_neigh_lookup,
David S. Miller14e50e52007-05-24 18:17:54 -0700263};
264
David S. Miller62fa8a82011-01-26 20:51:05 -0800265static const u32 ip6_template_metrics[RTAX_MAX] = {
Li RongQing14edd872012-10-24 14:01:18 +0800266 [RTAX_HOPLIMIT - 1] = 0,
David S. Miller62fa8a82011-01-26 20:51:05 -0800267};
268
Eric Dumazetfb0af4c2012-09-11 21:47:51 +0000269static const struct rt6_info ip6_null_entry_template = {
Changli Gaod8d1f302010-06-10 23:31:35 -0700270 .dst = {
271 .__refcnt = ATOMIC_INIT(1),
272 .__use = 1,
Nicolas Dichtel2c20cbd2012-09-10 22:09:47 +0000273 .obsolete = DST_OBSOLETE_FORCE_CHK,
Changli Gaod8d1f302010-06-10 23:31:35 -0700274 .error = -ENETUNREACH,
Changli Gaod8d1f302010-06-10 23:31:35 -0700275 .input = ip6_pkt_discard,
276 .output = ip6_pkt_discard_out,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700277 },
278 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
Jean-Mickael Guerin4f724272009-05-20 17:38:59 -0700279 .rt6i_protocol = RTPROT_KERNEL,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700280 .rt6i_metric = ~(u32) 0,
281 .rt6i_ref = ATOMIC_INIT(1),
282};
283
Thomas Graf101367c2006-08-04 03:39:02 -0700284#ifdef CONFIG_IPV6_MULTIPLE_TABLES
285
Eric Dumazetfb0af4c2012-09-11 21:47:51 +0000286static const struct rt6_info ip6_prohibit_entry_template = {
Changli Gaod8d1f302010-06-10 23:31:35 -0700287 .dst = {
288 .__refcnt = ATOMIC_INIT(1),
289 .__use = 1,
Nicolas Dichtel2c20cbd2012-09-10 22:09:47 +0000290 .obsolete = DST_OBSOLETE_FORCE_CHK,
Changli Gaod8d1f302010-06-10 23:31:35 -0700291 .error = -EACCES,
Changli Gaod8d1f302010-06-10 23:31:35 -0700292 .input = ip6_pkt_prohibit,
293 .output = ip6_pkt_prohibit_out,
Thomas Graf101367c2006-08-04 03:39:02 -0700294 },
295 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
Jean-Mickael Guerin4f724272009-05-20 17:38:59 -0700296 .rt6i_protocol = RTPROT_KERNEL,
Thomas Graf101367c2006-08-04 03:39:02 -0700297 .rt6i_metric = ~(u32) 0,
298 .rt6i_ref = ATOMIC_INIT(1),
299};
300
Eric Dumazetfb0af4c2012-09-11 21:47:51 +0000301static const struct rt6_info ip6_blk_hole_entry_template = {
Changli Gaod8d1f302010-06-10 23:31:35 -0700302 .dst = {
303 .__refcnt = ATOMIC_INIT(1),
304 .__use = 1,
Nicolas Dichtel2c20cbd2012-09-10 22:09:47 +0000305 .obsolete = DST_OBSOLETE_FORCE_CHK,
Changli Gaod8d1f302010-06-10 23:31:35 -0700306 .error = -EINVAL,
Changli Gaod8d1f302010-06-10 23:31:35 -0700307 .input = dst_discard,
Eric W. Biedermanede20592015-10-07 16:48:47 -0500308 .output = dst_discard_out,
Thomas Graf101367c2006-08-04 03:39:02 -0700309 },
310 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
Jean-Mickael Guerin4f724272009-05-20 17:38:59 -0700311 .rt6i_protocol = RTPROT_KERNEL,
Thomas Graf101367c2006-08-04 03:39:02 -0700312 .rt6i_metric = ~(u32) 0,
313 .rt6i_ref = ATOMIC_INIT(1),
314};
315
316#endif
317
Martin KaFai Lauebfa45f2015-10-15 16:39:57 -0700318static void rt6_info_init(struct rt6_info *rt)
319{
320 struct dst_entry *dst = &rt->dst;
321
322 memset(dst + 1, 0, sizeof(*rt) - sizeof(*dst));
323 INIT_LIST_HEAD(&rt->rt6i_siblings);
324 INIT_LIST_HEAD(&rt->rt6i_uncached);
325}
326
Linus Torvalds1da177e2005-04-16 15:20:36 -0700327/* allocate dst with ip6_dst_ops */
Martin KaFai Laud52d3992015-05-22 20:56:06 -0700328static struct rt6_info *__ip6_dst_alloc(struct net *net,
329 struct net_device *dev,
Martin KaFai Lauad706862015-08-14 11:05:52 -0700330 int flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700331{
David S. Miller97bab732012-06-09 22:36:36 -0700332 struct rt6_info *rt = dst_alloc(&net->ipv6.ip6_dst_ops, dev,
Nicolas Dichtel6f3118b2012-09-10 22:09:46 +0000333 0, DST_OBSOLETE_FORCE_CHK, flags);
David S. Millercf911662011-04-28 14:31:47 -0700334
Martin KaFai Lauebfa45f2015-10-15 16:39:57 -0700335 if (rt)
336 rt6_info_init(rt);
Steffen Klassert81048912012-07-05 23:37:09 +0000337
David S. Millercf911662011-04-28 14:31:47 -0700338 return rt;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700339}
340
David Ahern9ab179d2016-04-07 11:10:06 -0700341struct rt6_info *ip6_dst_alloc(struct net *net,
342 struct net_device *dev,
343 int flags)
Martin KaFai Laud52d3992015-05-22 20:56:06 -0700344{
Martin KaFai Lauad706862015-08-14 11:05:52 -0700345 struct rt6_info *rt = __ip6_dst_alloc(net, dev, flags);
Martin KaFai Laud52d3992015-05-22 20:56:06 -0700346
347 if (rt) {
348 rt->rt6i_pcpu = alloc_percpu_gfp(struct rt6_info *, GFP_ATOMIC);
349 if (rt->rt6i_pcpu) {
350 int cpu;
351
352 for_each_possible_cpu(cpu) {
353 struct rt6_info **p;
354
355 p = per_cpu_ptr(rt->rt6i_pcpu, cpu);
356 /* no one shares rt */
357 *p = NULL;
358 }
359 } else {
360 dst_destroy((struct dst_entry *)rt);
361 return NULL;
362 }
363 }
364
365 return rt;
366}
David Ahern9ab179d2016-04-07 11:10:06 -0700367EXPORT_SYMBOL(ip6_dst_alloc);
Martin KaFai Laud52d3992015-05-22 20:56:06 -0700368
Linus Torvalds1da177e2005-04-16 15:20:36 -0700369static void ip6_dst_destroy(struct dst_entry *dst)
370{
371 struct rt6_info *rt = (struct rt6_info *)dst;
YOSHIFUJI Hideaki / 吉藤英明ecd98832013-02-20 00:29:08 +0000372 struct dst_entry *from = dst->from;
Martin KaFai Lau8d0b94a2015-05-22 20:56:04 -0700373 struct inet6_dev *idev;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700374
Martin KaFai Lau4b32b5a2015-04-28 13:03:06 -0700375 dst_destroy_metrics_generic(dst);
Markus Elfring87775312015-07-02 16:30:24 +0200376 free_percpu(rt->rt6i_pcpu);
Martin KaFai Lau8d0b94a2015-05-22 20:56:04 -0700377 rt6_uncached_list_del(rt);
378
379 idev = rt->rt6i_idev;
David S. Miller38308472011-12-03 18:02:47 -0500380 if (idev) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700381 rt->rt6i_idev = NULL;
382 in6_dev_put(idev);
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900383 }
Gao feng1716a962012-04-06 00:13:10 +0000384
YOSHIFUJI Hideaki / 吉藤英明ecd98832013-02-20 00:29:08 +0000385 dst->from = NULL;
386 dst_release(from);
David S. Millerb3419362010-11-30 12:27:11 -0800387}
388
Linus Torvalds1da177e2005-04-16 15:20:36 -0700389static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
390 int how)
391{
392 struct rt6_info *rt = (struct rt6_info *)dst;
393 struct inet6_dev *idev = rt->rt6i_idev;
Denis V. Lunev5a3e55d2007-12-07 00:38:10 -0800394 struct net_device *loopback_dev =
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900395 dev_net(dev)->loopback_dev;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700396
David S. Miller97cac082012-07-02 22:43:47 -0700397 if (dev != loopback_dev) {
398 if (idev && idev->dev == dev) {
399 struct inet6_dev *loopback_idev =
400 in6_dev_get(loopback_dev);
401 if (loopback_idev) {
402 rt->rt6i_idev = loopback_idev;
403 in6_dev_put(idev);
404 }
405 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700406 }
407}
408
Martin KaFai Lau5973fb12015-11-11 11:51:07 -0800409static bool __rt6_check_expired(const struct rt6_info *rt)
410{
411 if (rt->rt6i_flags & RTF_EXPIRES)
412 return time_after(jiffies, rt->dst.expires);
413 else
414 return false;
415}
416
Eric Dumazeta50feda2012-05-18 18:57:34 +0000417static bool rt6_check_expired(const struct rt6_info *rt)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700418{
Gao feng1716a962012-04-06 00:13:10 +0000419 if (rt->rt6i_flags & RTF_EXPIRES) {
420 if (time_after(jiffies, rt->dst.expires))
Eric Dumazeta50feda2012-05-18 18:57:34 +0000421 return true;
Gao feng1716a962012-04-06 00:13:10 +0000422 } else if (rt->dst.from) {
Li RongQing3fd91fb2012-09-13 19:54:57 +0000423 return rt6_check_expired((struct rt6_info *) rt->dst.from);
Gao feng1716a962012-04-06 00:13:10 +0000424 }
Eric Dumazeta50feda2012-05-18 18:57:34 +0000425 return false;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700426}
427
Nicolas Dichtel51ebd312012-10-22 03:42:09 +0000428/* Multipath route selection:
429 * Hash based function using packet header and flowlabel.
430 * Adapted from fib_info_hashfn()
431 */
432static int rt6_info_hash_nhsfn(unsigned int candidate_count,
433 const struct flowi6 *fl6)
434{
Tom Herbert644d0e62015-09-23 14:13:35 -0700435 return get_hash_from_flowi6(fl6) % candidate_count;
Nicolas Dichtel51ebd312012-10-22 03:42:09 +0000436}
437
438static struct rt6_info *rt6_multipath_select(struct rt6_info *match,
Nicolas Dichtel52bd4c02013-06-28 17:35:48 +0200439 struct flowi6 *fl6, int oif,
440 int strict)
Nicolas Dichtel51ebd312012-10-22 03:42:09 +0000441{
442 struct rt6_info *sibling, *next_sibling;
443 int route_choosen;
444
445 route_choosen = rt6_info_hash_nhsfn(match->rt6i_nsiblings + 1, fl6);
446 /* Don't change the route, if route_choosen == 0
447 * (siblings does not include ourself)
448 */
449 if (route_choosen)
450 list_for_each_entry_safe(sibling, next_sibling,
451 &match->rt6i_siblings, rt6i_siblings) {
452 route_choosen--;
453 if (route_choosen == 0) {
Nicolas Dichtel52bd4c02013-06-28 17:35:48 +0200454 if (rt6_score_route(sibling, oif, strict) < 0)
455 break;
Nicolas Dichtel51ebd312012-10-22 03:42:09 +0000456 match = sibling;
457 break;
458 }
459 }
460 return match;
461}
462
Linus Torvalds1da177e2005-04-16 15:20:36 -0700463/*
Thomas Grafc71099a2006-08-04 23:20:06 -0700464 * Route lookup. Any table->tb6_lock is implied.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700465 */
466
Daniel Lezcano8ed67782008-03-04 13:48:30 -0800467static inline struct rt6_info *rt6_device_match(struct net *net,
468 struct rt6_info *rt,
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000469 const struct in6_addr *saddr,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700470 int oif,
YOSHIFUJI Hideakid4208952008-06-27 20:14:54 -0700471 int flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700472{
473 struct rt6_info *local = NULL;
474 struct rt6_info *sprt;
475
YOSHIFUJI Hideakidd3abc42008-07-02 18:30:18 +0900476 if (!oif && ipv6_addr_any(saddr))
477 goto out;
478
Changli Gaod8d1f302010-06-10 23:31:35 -0700479 for (sprt = rt; sprt; sprt = sprt->dst.rt6_next) {
David S. Millerd1918542011-12-28 20:19:20 -0500480 struct net_device *dev = sprt->dst.dev;
YOSHIFUJI Hideakidd3abc42008-07-02 18:30:18 +0900481
482 if (oif) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700483 if (dev->ifindex == oif)
484 return sprt;
485 if (dev->flags & IFF_LOOPBACK) {
David S. Miller38308472011-12-03 18:02:47 -0500486 if (!sprt->rt6i_idev ||
Linus Torvalds1da177e2005-04-16 15:20:36 -0700487 sprt->rt6i_idev->dev->ifindex != oif) {
David Ahern17fb0b22015-09-25 15:22:54 -0600488 if (flags & RT6_LOOKUP_F_IFACE)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700489 continue;
David Ahern17fb0b22015-09-25 15:22:54 -0600490 if (local &&
491 local->rt6i_idev->dev->ifindex == oif)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700492 continue;
493 }
494 local = sprt;
495 }
YOSHIFUJI Hideakidd3abc42008-07-02 18:30:18 +0900496 } else {
497 if (ipv6_chk_addr(net, saddr, dev,
498 flags & RT6_LOOKUP_F_IFACE))
499 return sprt;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700500 }
YOSHIFUJI Hideakidd3abc42008-07-02 18:30:18 +0900501 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700502
YOSHIFUJI Hideakidd3abc42008-07-02 18:30:18 +0900503 if (oif) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700504 if (local)
505 return local;
506
YOSHIFUJI Hideakid4208952008-06-27 20:14:54 -0700507 if (flags & RT6_LOOKUP_F_IFACE)
Daniel Lezcano8ed67782008-03-04 13:48:30 -0800508 return net->ipv6.ip6_null_entry;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700509 }
YOSHIFUJI Hideakidd3abc42008-07-02 18:30:18 +0900510out:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700511 return rt;
512}
513
YOSHIFUJI Hideaki27097252006-03-20 17:05:13 -0800514#ifdef CONFIG_IPV6_ROUTER_PREF
Hannes Frederic Sowac2f17e82013-10-21 06:17:15 +0200515struct __rt6_probe_work {
516 struct work_struct work;
517 struct in6_addr target;
518 struct net_device *dev;
519};
520
521static void rt6_probe_deferred(struct work_struct *w)
522{
523 struct in6_addr mcaddr;
524 struct __rt6_probe_work *work =
525 container_of(w, struct __rt6_probe_work, work);
526
527 addrconf_addr_solict_mult(&work->target, &mcaddr);
Nicolas Dichtel304d8882015-11-27 18:17:05 +0100528 ndisc_send_ns(work->dev, &work->target, &mcaddr, NULL);
Hannes Frederic Sowac2f17e82013-10-21 06:17:15 +0200529 dev_put(work->dev);
Michael Büsch662f5532015-02-08 10:14:07 +0100530 kfree(work);
Hannes Frederic Sowac2f17e82013-10-21 06:17:15 +0200531}
532
YOSHIFUJI Hideaki27097252006-03-20 17:05:13 -0800533static void rt6_probe(struct rt6_info *rt)
534{
Martin KaFai Lau990edb42015-07-24 09:57:42 -0700535 struct __rt6_probe_work *work;
Eric Dumazetf2c31e32011-07-29 19:00:53 +0000536 struct neighbour *neigh;
YOSHIFUJI Hideaki27097252006-03-20 17:05:13 -0800537 /*
538 * Okay, this does not seem to be appropriate
539 * for now, however, we need to check if it
540 * is really so; aka Router Reachability Probing.
541 *
542 * Router Reachability Probe MUST be rate-limited
543 * to no more than one per minute.
544 */
YOSHIFUJI Hideaki / 吉藤英明2152cae2013-01-17 12:53:43 +0000545 if (!rt || !(rt->rt6i_flags & RTF_GATEWAY))
Amerigo Wangfdd66812012-09-10 02:48:44 +0000546 return;
YOSHIFUJI Hideaki / 吉藤英明2152cae2013-01-17 12:53:43 +0000547 rcu_read_lock_bh();
548 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
549 if (neigh) {
Martin KaFai Lau8d6c31b2015-07-24 09:57:43 -0700550 if (neigh->nud_state & NUD_VALID)
551 goto out;
552
Martin KaFai Lau990edb42015-07-24 09:57:42 -0700553 work = NULL;
YOSHIFUJI Hideaki / 吉藤英明2152cae2013-01-17 12:53:43 +0000554 write_lock(&neigh->lock);
Martin KaFai Lau990edb42015-07-24 09:57:42 -0700555 if (!(neigh->nud_state & NUD_VALID) &&
556 time_after(jiffies,
557 neigh->updated +
558 rt->rt6i_idev->cnf.rtr_probe_interval)) {
559 work = kmalloc(sizeof(*work), GFP_ATOMIC);
560 if (work)
561 __neigh_set_probe_once(neigh);
Hannes Frederic Sowac2f17e82013-10-21 06:17:15 +0200562 }
YOSHIFUJI Hideaki / 吉藤英明2152cae2013-01-17 12:53:43 +0000563 write_unlock(&neigh->lock);
Martin KaFai Lau990edb42015-07-24 09:57:42 -0700564 } else {
565 work = kmalloc(sizeof(*work), GFP_ATOMIC);
Eric Dumazetf2c31e32011-07-29 19:00:53 +0000566 }
Martin KaFai Lau990edb42015-07-24 09:57:42 -0700567
568 if (work) {
569 INIT_WORK(&work->work, rt6_probe_deferred);
570 work->target = rt->rt6i_gateway;
571 dev_hold(rt->dst.dev);
572 work->dev = rt->dst.dev;
573 schedule_work(&work->work);
574 }
575
Martin KaFai Lau8d6c31b2015-07-24 09:57:43 -0700576out:
YOSHIFUJI Hideaki / 吉藤英明2152cae2013-01-17 12:53:43 +0000577 rcu_read_unlock_bh();
YOSHIFUJI Hideaki27097252006-03-20 17:05:13 -0800578}
579#else
580static inline void rt6_probe(struct rt6_info *rt)
581{
YOSHIFUJI Hideaki27097252006-03-20 17:05:13 -0800582}
583#endif
584
Linus Torvalds1da177e2005-04-16 15:20:36 -0700585/*
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800586 * Default Router Selection (RFC 2461 6.3.6)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700587 */
Dave Jonesb6f99a22007-03-22 12:27:49 -0700588static inline int rt6_check_dev(struct rt6_info *rt, int oif)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700589{
David S. Millerd1918542011-12-28 20:19:20 -0500590 struct net_device *dev = rt->dst.dev;
David S. Miller161980f2007-04-06 11:42:27 -0700591 if (!oif || dev->ifindex == oif)
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800592 return 2;
David S. Miller161980f2007-04-06 11:42:27 -0700593 if ((dev->flags & IFF_LOOPBACK) &&
594 rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif)
595 return 1;
596 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700597}
598
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +0200599static inline enum rt6_nud_state rt6_check_neigh(struct rt6_info *rt)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700600{
Eric Dumazetf2c31e32011-07-29 19:00:53 +0000601 struct neighbour *neigh;
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +0200602 enum rt6_nud_state ret = RT6_NUD_FAIL_HARD;
Eric Dumazetf2c31e32011-07-29 19:00:53 +0000603
YOSHIFUJI Hideaki4d0c5912006-05-26 13:23:41 -0700604 if (rt->rt6i_flags & RTF_NONEXTHOP ||
605 !(rt->rt6i_flags & RTF_GATEWAY))
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +0200606 return RT6_NUD_SUCCEED;
YOSHIFUJI Hideaki / 吉藤英明145a3622013-01-17 12:53:38 +0000607
608 rcu_read_lock_bh();
609 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
610 if (neigh) {
611 read_lock(&neigh->lock);
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800612 if (neigh->nud_state & NUD_VALID)
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +0200613 ret = RT6_NUD_SUCCEED;
YOSHIFUJI Hideaki398bcbe2008-01-19 00:35:16 -0800614#ifdef CONFIG_IPV6_ROUTER_PREF
Paul Marksa5a81f02012-12-03 10:26:54 +0000615 else if (!(neigh->nud_state & NUD_FAILED))
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +0200616 ret = RT6_NUD_SUCCEED;
Jiri Benc7e980562013-12-11 13:48:20 +0100617 else
618 ret = RT6_NUD_FAIL_PROBE;
YOSHIFUJI Hideaki398bcbe2008-01-19 00:35:16 -0800619#endif
YOSHIFUJI Hideaki / 吉藤英明145a3622013-01-17 12:53:38 +0000620 read_unlock(&neigh->lock);
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +0200621 } else {
622 ret = IS_ENABLED(CONFIG_IPV6_ROUTER_PREF) ?
Jiri Benc7e980562013-12-11 13:48:20 +0100623 RT6_NUD_SUCCEED : RT6_NUD_FAIL_DO_RR;
Paul Marksa5a81f02012-12-03 10:26:54 +0000624 }
YOSHIFUJI Hideaki / 吉藤英明145a3622013-01-17 12:53:38 +0000625 rcu_read_unlock_bh();
626
Paul Marksa5a81f02012-12-03 10:26:54 +0000627 return ret;
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800628}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700629
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800630static int rt6_score_route(struct rt6_info *rt, int oif,
631 int strict)
632{
Paul Marksa5a81f02012-12-03 10:26:54 +0000633 int m;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900634
YOSHIFUJI Hideaki4d0c5912006-05-26 13:23:41 -0700635 m = rt6_check_dev(rt, oif);
YOSHIFUJI Hideaki77d16f42006-08-23 17:25:05 -0700636 if (!m && (strict & RT6_LOOKUP_F_IFACE))
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +0200637 return RT6_NUD_FAIL_HARD;
YOSHIFUJI Hideakiebacaaa2006-03-20 17:04:53 -0800638#ifdef CONFIG_IPV6_ROUTER_PREF
639 m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2;
640#endif
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +0200641 if (strict & RT6_LOOKUP_F_REACHABLE) {
642 int n = rt6_check_neigh(rt);
643 if (n < 0)
644 return n;
645 }
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800646 return m;
647}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700648
David S. Millerf11e6652007-03-24 20:36:25 -0700649static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict,
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +0200650 int *mpri, struct rt6_info *match,
651 bool *do_rr)
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800652{
David S. Millerf11e6652007-03-24 20:36:25 -0700653 int m;
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +0200654 bool match_do_rr = false;
Andy Gospodarek35103d12015-08-13 10:39:01 -0400655 struct inet6_dev *idev = rt->rt6i_idev;
656 struct net_device *dev = rt->dst.dev;
657
658 if (dev && !netif_carrier_ok(dev) &&
659 idev->cnf.ignore_routes_with_linkdown)
660 goto out;
David S. Millerf11e6652007-03-24 20:36:25 -0700661
662 if (rt6_check_expired(rt))
663 goto out;
664
665 m = rt6_score_route(rt, oif, strict);
Jiri Benc7e980562013-12-11 13:48:20 +0100666 if (m == RT6_NUD_FAIL_DO_RR) {
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +0200667 match_do_rr = true;
668 m = 0; /* lowest valid score */
Jiri Benc7e980562013-12-11 13:48:20 +0100669 } else if (m == RT6_NUD_FAIL_HARD) {
David S. Millerf11e6652007-03-24 20:36:25 -0700670 goto out;
David S. Millerf11e6652007-03-24 20:36:25 -0700671 }
672
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +0200673 if (strict & RT6_LOOKUP_F_REACHABLE)
674 rt6_probe(rt);
675
Jiri Benc7e980562013-12-11 13:48:20 +0100676 /* note that m can be RT6_NUD_FAIL_PROBE at this point */
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +0200677 if (m > *mpri) {
678 *do_rr = match_do_rr;
679 *mpri = m;
680 match = rt;
681 }
David S. Millerf11e6652007-03-24 20:36:25 -0700682out:
683 return match;
684}
685
686static struct rt6_info *find_rr_leaf(struct fib6_node *fn,
687 struct rt6_info *rr_head,
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +0200688 u32 metric, int oif, int strict,
689 bool *do_rr)
David S. Millerf11e6652007-03-24 20:36:25 -0700690{
Steffen Klassert9fbdcfa2015-04-28 13:03:04 -0700691 struct rt6_info *rt, *match, *cont;
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800692 int mpri = -1;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700693
David S. Millerf11e6652007-03-24 20:36:25 -0700694 match = NULL;
Steffen Klassert9fbdcfa2015-04-28 13:03:04 -0700695 cont = NULL;
696 for (rt = rr_head; rt; rt = rt->dst.rt6_next) {
697 if (rt->rt6i_metric != metric) {
698 cont = rt;
699 break;
700 }
701
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +0200702 match = find_match(rt, oif, strict, &mpri, match, do_rr);
Steffen Klassert9fbdcfa2015-04-28 13:03:04 -0700703 }
704
705 for (rt = fn->leaf; rt && rt != rr_head; rt = rt->dst.rt6_next) {
706 if (rt->rt6i_metric != metric) {
707 cont = rt;
708 break;
709 }
710
711 match = find_match(rt, oif, strict, &mpri, match, do_rr);
712 }
713
714 if (match || !cont)
715 return match;
716
717 for (rt = cont; rt; rt = rt->dst.rt6_next)
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +0200718 match = find_match(rt, oif, strict, &mpri, match, do_rr);
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800719
David S. Millerf11e6652007-03-24 20:36:25 -0700720 return match;
721}
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800722
David S. Millerf11e6652007-03-24 20:36:25 -0700723static struct rt6_info *rt6_select(struct fib6_node *fn, int oif, int strict)
724{
725 struct rt6_info *match, *rt0;
Daniel Lezcano8ed67782008-03-04 13:48:30 -0800726 struct net *net;
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +0200727 bool do_rr = false;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700728
David S. Millerf11e6652007-03-24 20:36:25 -0700729 rt0 = fn->rr_ptr;
730 if (!rt0)
731 fn->rr_ptr = rt0 = fn->leaf;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700732
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +0200733 match = find_rr_leaf(fn, rt0, rt0->rt6i_metric, oif, strict,
734 &do_rr);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700735
Hannes Frederic Sowaafc154e2013-07-11 12:43:42 +0200736 if (do_rr) {
Changli Gaod8d1f302010-06-10 23:31:35 -0700737 struct rt6_info *next = rt0->dst.rt6_next;
David S. Millerf11e6652007-03-24 20:36:25 -0700738
YOSHIFUJI Hideaki554cfb72006-03-20 17:00:26 -0800739 /* no entries matched; do round-robin */
David S. Millerf11e6652007-03-24 20:36:25 -0700740 if (!next || next->rt6i_metric != rt0->rt6i_metric)
741 next = fn->leaf;
742
743 if (next != rt0)
744 fn->rr_ptr = next;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700745 }
746
David S. Millerd1918542011-12-28 20:19:20 -0500747 net = dev_net(rt0->dst.dev);
Eric Dumazeta02cec22010-09-22 20:43:57 +0000748 return match ? match : net->ipv6.ip6_null_entry;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700749}
750
Martin KaFai Lau8b9df262015-05-22 20:55:59 -0700751static bool rt6_is_gw_or_nonexthop(const struct rt6_info *rt)
752{
753 return (rt->rt6i_flags & (RTF_NONEXTHOP | RTF_GATEWAY));
754}
755
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800756#ifdef CONFIG_IPV6_ROUTE_INFO
757int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000758 const struct in6_addr *gwaddr)
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800759{
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900760 struct net *net = dev_net(dev);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800761 struct route_info *rinfo = (struct route_info *) opt;
762 struct in6_addr prefix_buf, *prefix;
763 unsigned int pref;
YOSHIFUJI Hideaki4bed72e2008-05-27 17:37:49 +0900764 unsigned long lifetime;
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800765 struct rt6_info *rt;
766
767 if (len < sizeof(struct route_info)) {
768 return -EINVAL;
769 }
770
771 /* Sanity check for prefix_len and length */
772 if (rinfo->length > 3) {
773 return -EINVAL;
774 } else if (rinfo->prefix_len > 128) {
775 return -EINVAL;
776 } else if (rinfo->prefix_len > 64) {
777 if (rinfo->length < 2) {
778 return -EINVAL;
779 }
780 } else if (rinfo->prefix_len > 0) {
781 if (rinfo->length < 1) {
782 return -EINVAL;
783 }
784 }
785
786 pref = rinfo->route_pref;
787 if (pref == ICMPV6_ROUTER_PREF_INVALID)
Jens Rosenboom3933fc92009-09-10 06:25:11 +0000788 return -EINVAL;
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800789
YOSHIFUJI Hideaki4bed72e2008-05-27 17:37:49 +0900790 lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800791
792 if (rinfo->length == 3)
793 prefix = (struct in6_addr *)rinfo->prefix;
794 else {
795 /* this function is safe */
796 ipv6_addr_prefix(&prefix_buf,
797 (struct in6_addr *)rinfo->prefix,
798 rinfo->prefix_len);
799 prefix = &prefix_buf;
800 }
801
Duan Jiongf104a562013-11-08 09:56:53 +0800802 if (rinfo->prefix_len == 0)
803 rt = rt6_get_dflt_router(gwaddr, dev);
804 else
805 rt = rt6_get_route_info(net, prefix, rinfo->prefix_len,
806 gwaddr, dev->ifindex);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800807
808 if (rt && !lifetime) {
Thomas Grafe0a1ad732006-08-22 00:00:21 -0700809 ip6_del_rt(rt);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800810 rt = NULL;
811 }
812
813 if (!rt && lifetime)
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -0800814 rt = rt6_add_route_info(net, prefix, rinfo->prefix_len, gwaddr, dev->ifindex,
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800815 pref);
816 else if (rt)
817 rt->rt6i_flags = RTF_ROUTEINFO |
818 (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref);
819
820 if (rt) {
Gao feng1716a962012-04-06 00:13:10 +0000821 if (!addrconf_finite_timeout(lifetime))
822 rt6_clean_expires(rt);
823 else
824 rt6_set_expires(rt, jiffies + HZ * lifetime);
825
Amerigo Wang94e187c2012-10-29 00:13:19 +0000826 ip6_rt_put(rt);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -0800827 }
828 return 0;
829}
830#endif
831
Martin KaFai Laua3c00e42014-10-20 13:42:43 -0700832static struct fib6_node* fib6_backtrack(struct fib6_node *fn,
833 struct in6_addr *saddr)
834{
835 struct fib6_node *pn;
836 while (1) {
837 if (fn->fn_flags & RTN_TL_ROOT)
838 return NULL;
839 pn = fn->parent;
840 if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn)
841 fn = fib6_lookup(FIB6_SUBTREE(pn), NULL, saddr);
842 else
843 fn = pn;
844 if (fn->fn_flags & RTN_RTINFO)
845 return fn;
846 }
847}
Thomas Grafc71099a2006-08-04 23:20:06 -0700848
Daniel Lezcano8ed67782008-03-04 13:48:30 -0800849static struct rt6_info *ip6_pol_route_lookup(struct net *net,
850 struct fib6_table *table,
David S. Miller4c9483b2011-03-12 16:22:43 -0500851 struct flowi6 *fl6, int flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700852{
853 struct fib6_node *fn;
854 struct rt6_info *rt;
855
Thomas Grafc71099a2006-08-04 23:20:06 -0700856 read_lock_bh(&table->tb6_lock);
David S. Miller4c9483b2011-03-12 16:22:43 -0500857 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
Thomas Grafc71099a2006-08-04 23:20:06 -0700858restart:
859 rt = fn->leaf;
David S. Miller4c9483b2011-03-12 16:22:43 -0500860 rt = rt6_device_match(net, rt, &fl6->saddr, fl6->flowi6_oif, flags);
Nicolas Dichtel51ebd312012-10-22 03:42:09 +0000861 if (rt->rt6i_nsiblings && fl6->flowi6_oif == 0)
Nicolas Dichtel52bd4c02013-06-28 17:35:48 +0200862 rt = rt6_multipath_select(rt, fl6, fl6->flowi6_oif, flags);
Martin KaFai Laua3c00e42014-10-20 13:42:43 -0700863 if (rt == net->ipv6.ip6_null_entry) {
864 fn = fib6_backtrack(fn, &fl6->saddr);
865 if (fn)
866 goto restart;
867 }
Changli Gaod8d1f302010-06-10 23:31:35 -0700868 dst_use(&rt->dst, jiffies);
Thomas Grafc71099a2006-08-04 23:20:06 -0700869 read_unlock_bh(&table->tb6_lock);
David Ahernb8115802015-11-19 12:24:22 -0800870
871 trace_fib6_table_lookup(net, rt, table->tb6_id, fl6);
872
Thomas Grafc71099a2006-08-04 23:20:06 -0700873 return rt;
874
875}
876
Ian Morris67ba4152014-08-24 21:53:10 +0100877struct dst_entry *ip6_route_lookup(struct net *net, struct flowi6 *fl6,
Florian Westphalea6e5742011-09-05 16:05:44 +0200878 int flags)
879{
880 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_lookup);
881}
882EXPORT_SYMBOL_GPL(ip6_route_lookup);
883
YOSHIFUJI Hideaki9acd9f32008-04-10 15:42:10 +0900884struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
885 const struct in6_addr *saddr, int oif, int strict)
Thomas Grafc71099a2006-08-04 23:20:06 -0700886{
David S. Miller4c9483b2011-03-12 16:22:43 -0500887 struct flowi6 fl6 = {
888 .flowi6_oif = oif,
889 .daddr = *daddr,
Thomas Grafc71099a2006-08-04 23:20:06 -0700890 };
891 struct dst_entry *dst;
YOSHIFUJI Hideaki77d16f42006-08-23 17:25:05 -0700892 int flags = strict ? RT6_LOOKUP_F_IFACE : 0;
Thomas Grafc71099a2006-08-04 23:20:06 -0700893
Thomas Grafadaa70b2006-10-13 15:01:03 -0700894 if (saddr) {
David S. Miller4c9483b2011-03-12 16:22:43 -0500895 memcpy(&fl6.saddr, saddr, sizeof(*saddr));
Thomas Grafadaa70b2006-10-13 15:01:03 -0700896 flags |= RT6_LOOKUP_F_HAS_SADDR;
897 }
898
David S. Miller4c9483b2011-03-12 16:22:43 -0500899 dst = fib6_rule_lookup(net, &fl6, flags, ip6_pol_route_lookup);
Thomas Grafc71099a2006-08-04 23:20:06 -0700900 if (dst->error == 0)
901 return (struct rt6_info *) dst;
902
903 dst_release(dst);
904
Linus Torvalds1da177e2005-04-16 15:20:36 -0700905 return NULL;
906}
YOSHIFUJI Hideaki71590392007-02-22 22:05:40 +0900907EXPORT_SYMBOL(rt6_lookup);
908
Thomas Grafc71099a2006-08-04 23:20:06 -0700909/* ip6_ins_rt is called with FREE table->tb6_lock.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700910 It takes new route entry, the addition fails by any reason the
911 route is freed. In any case, if caller does not hold it, it may
912 be destroyed.
913 */
914
Michal Kubečeke5fd3872014-03-27 13:04:08 +0100915static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info,
Florian Westphale715b6d2015-01-05 23:57:44 +0100916 struct mx6_config *mxc)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700917{
918 int err;
Thomas Grafc71099a2006-08-04 23:20:06 -0700919 struct fib6_table *table;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700920
Thomas Grafc71099a2006-08-04 23:20:06 -0700921 table = rt->rt6i_table;
922 write_lock_bh(&table->tb6_lock);
Florian Westphale715b6d2015-01-05 23:57:44 +0100923 err = fib6_add(&table->tb6_root, rt, info, mxc);
Thomas Grafc71099a2006-08-04 23:20:06 -0700924 write_unlock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700925
926 return err;
927}
928
Thomas Graf40e22e82006-08-22 00:00:45 -0700929int ip6_ins_rt(struct rt6_info *rt)
930{
Florian Westphale715b6d2015-01-05 23:57:44 +0100931 struct nl_info info = { .nl_net = dev_net(rt->dst.dev), };
932 struct mx6_config mxc = { .mx = NULL, };
933
934 return __ip6_ins_rt(rt, &info, &mxc);
Thomas Graf40e22e82006-08-22 00:00:45 -0700935}
936
Martin KaFai Lau8b9df262015-05-22 20:55:59 -0700937static struct rt6_info *ip6_rt_cache_alloc(struct rt6_info *ort,
938 const struct in6_addr *daddr,
939 const struct in6_addr *saddr)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700940{
Linus Torvalds1da177e2005-04-16 15:20:36 -0700941 struct rt6_info *rt;
942
943 /*
944 * Clone the route.
945 */
946
Martin KaFai Laud52d3992015-05-22 20:56:06 -0700947 if (ort->rt6i_flags & (RTF_CACHE | RTF_PCPU))
Martin KaFai Lau83a09ab2015-05-22 20:56:05 -0700948 ort = (struct rt6_info *)ort->dst.from;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700949
Martin KaFai Lauad706862015-08-14 11:05:52 -0700950 rt = __ip6_dst_alloc(dev_net(ort->dst.dev), ort->dst.dev, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700951
Martin KaFai Lau83a09ab2015-05-22 20:56:05 -0700952 if (!rt)
953 return NULL;
954
955 ip6_rt_copy_init(rt, ort);
956 rt->rt6i_flags |= RTF_CACHE;
957 rt->rt6i_metric = 0;
958 rt->dst.flags |= DST_HOST;
959 rt->rt6i_dst.addr = *daddr;
960 rt->rt6i_dst.plen = 128;
961
962 if (!rt6_is_gw_or_nonexthop(ort)) {
963 if (ort->rt6i_dst.plen != 128 &&
964 ipv6_addr_equal(&ort->rt6i_dst.addr, daddr))
965 rt->rt6i_flags |= RTF_ANYCAST;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700966#ifdef CONFIG_IPV6_SUBTREES
Martin KaFai Lau83a09ab2015-05-22 20:56:05 -0700967 if (rt->rt6i_src.plen && saddr) {
968 rt->rt6i_src.addr = *saddr;
969 rt->rt6i_src.plen = 128;
Martin KaFai Lau8b9df262015-05-22 20:55:59 -0700970 }
Martin KaFai Lau83a09ab2015-05-22 20:56:05 -0700971#endif
YOSHIFUJI Hideaki95a9a5b2006-03-20 16:55:51 -0800972 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700973
YOSHIFUJI Hideaki95a9a5b2006-03-20 16:55:51 -0800974 return rt;
975}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700976
Martin KaFai Laud52d3992015-05-22 20:56:06 -0700977static struct rt6_info *ip6_rt_pcpu_alloc(struct rt6_info *rt)
978{
979 struct rt6_info *pcpu_rt;
980
981 pcpu_rt = __ip6_dst_alloc(dev_net(rt->dst.dev),
Martin KaFai Lauad706862015-08-14 11:05:52 -0700982 rt->dst.dev, rt->dst.flags);
Martin KaFai Laud52d3992015-05-22 20:56:06 -0700983
984 if (!pcpu_rt)
985 return NULL;
986 ip6_rt_copy_init(pcpu_rt, rt);
987 pcpu_rt->rt6i_protocol = rt->rt6i_protocol;
988 pcpu_rt->rt6i_flags |= RTF_PCPU;
989 return pcpu_rt;
990}
991
992/* It should be called with read_lock_bh(&tb6_lock) acquired */
993static struct rt6_info *rt6_get_pcpu_route(struct rt6_info *rt)
994{
Martin KaFai Laua73e4192015-08-14 11:05:53 -0700995 struct rt6_info *pcpu_rt, **p;
Martin KaFai Laud52d3992015-05-22 20:56:06 -0700996
997 p = this_cpu_ptr(rt->rt6i_pcpu);
998 pcpu_rt = *p;
999
Martin KaFai Laua73e4192015-08-14 11:05:53 -07001000 if (pcpu_rt) {
1001 dst_hold(&pcpu_rt->dst);
1002 rt6_dst_from_metrics_check(pcpu_rt);
1003 }
1004 return pcpu_rt;
1005}
1006
1007static struct rt6_info *rt6_make_pcpu_route(struct rt6_info *rt)
1008{
Martin KaFai Lau9c7370a2015-08-14 11:05:54 -07001009 struct fib6_table *table = rt->rt6i_table;
Martin KaFai Laua73e4192015-08-14 11:05:53 -07001010 struct rt6_info *pcpu_rt, *prev, **p;
Martin KaFai Laud52d3992015-05-22 20:56:06 -07001011
1012 pcpu_rt = ip6_rt_pcpu_alloc(rt);
1013 if (!pcpu_rt) {
1014 struct net *net = dev_net(rt->dst.dev);
1015
Martin KaFai Lau9c7370a2015-08-14 11:05:54 -07001016 dst_hold(&net->ipv6.ip6_null_entry->dst);
1017 return net->ipv6.ip6_null_entry;
Martin KaFai Laud52d3992015-05-22 20:56:06 -07001018 }
1019
Martin KaFai Lau9c7370a2015-08-14 11:05:54 -07001020 read_lock_bh(&table->tb6_lock);
1021 if (rt->rt6i_pcpu) {
1022 p = this_cpu_ptr(rt->rt6i_pcpu);
1023 prev = cmpxchg(p, NULL, pcpu_rt);
1024 if (prev) {
1025 /* If someone did it before us, return prev instead */
1026 dst_destroy(&pcpu_rt->dst);
1027 pcpu_rt = prev;
1028 }
1029 } else {
1030 /* rt has been removed from the fib6 tree
1031 * before we have a chance to acquire the read_lock.
1032 * In this case, don't brother to create a pcpu rt
1033 * since rt is going away anyway. The next
1034 * dst_check() will trigger a re-lookup.
1035 */
Martin KaFai Laud52d3992015-05-22 20:56:06 -07001036 dst_destroy(&pcpu_rt->dst);
Martin KaFai Lau9c7370a2015-08-14 11:05:54 -07001037 pcpu_rt = rt;
Martin KaFai Laud52d3992015-05-22 20:56:06 -07001038 }
Martin KaFai Laud52d3992015-05-22 20:56:06 -07001039 dst_hold(&pcpu_rt->dst);
1040 rt6_dst_from_metrics_check(pcpu_rt);
Martin KaFai Lau9c7370a2015-08-14 11:05:54 -07001041 read_unlock_bh(&table->tb6_lock);
Martin KaFai Laud52d3992015-05-22 20:56:06 -07001042 return pcpu_rt;
1043}
1044
Daniel Lezcano8ed67782008-03-04 13:48:30 -08001045static struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table, int oif,
David S. Miller4c9483b2011-03-12 16:22:43 -05001046 struct flowi6 *fl6, int flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001047{
Martin KaFai Lau367efcb2014-10-20 13:42:45 -07001048 struct fib6_node *fn, *saved_fn;
Martin KaFai Lau45e4fd22015-05-22 20:56:00 -07001049 struct rt6_info *rt;
Thomas Grafc71099a2006-08-04 23:20:06 -07001050 int strict = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001051
YOSHIFUJI Hideaki77d16f42006-08-23 17:25:05 -07001052 strict |= flags & RT6_LOOKUP_F_IFACE;
Martin KaFai Lau367efcb2014-10-20 13:42:45 -07001053 if (net->ipv6.devconf_all->forwarding == 0)
1054 strict |= RT6_LOOKUP_F_REACHABLE;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001055
Thomas Grafc71099a2006-08-04 23:20:06 -07001056 read_lock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001057
David S. Miller4c9483b2011-03-12 16:22:43 -05001058 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
Martin KaFai Lau367efcb2014-10-20 13:42:45 -07001059 saved_fn = fn;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001060
David Ahernca254492015-10-12 11:47:10 -07001061 if (fl6->flowi6_flags & FLOWI_FLAG_SKIP_NH_OIF)
1062 oif = 0;
1063
Martin KaFai Laua3c00e42014-10-20 13:42:43 -07001064redo_rt6_select:
Martin KaFai Lau367efcb2014-10-20 13:42:45 -07001065 rt = rt6_select(fn, oif, strict);
Nicolas Dichtel52bd4c02013-06-28 17:35:48 +02001066 if (rt->rt6i_nsiblings)
Martin KaFai Lau367efcb2014-10-20 13:42:45 -07001067 rt = rt6_multipath_select(rt, fl6, oif, strict);
Martin KaFai Laua3c00e42014-10-20 13:42:43 -07001068 if (rt == net->ipv6.ip6_null_entry) {
1069 fn = fib6_backtrack(fn, &fl6->saddr);
1070 if (fn)
1071 goto redo_rt6_select;
Martin KaFai Lau367efcb2014-10-20 13:42:45 -07001072 else if (strict & RT6_LOOKUP_F_REACHABLE) {
1073 /* also consider unreachable route */
1074 strict &= ~RT6_LOOKUP_F_REACHABLE;
1075 fn = saved_fn;
1076 goto redo_rt6_select;
Martin KaFai Lau367efcb2014-10-20 13:42:45 -07001077 }
Martin KaFai Laua3c00e42014-10-20 13:42:43 -07001078 }
1079
YOSHIFUJI Hideakifb9de912006-03-20 16:59:08 -08001080
Martin KaFai Lau3da59bd2015-05-22 20:56:03 -07001081 if (rt == net->ipv6.ip6_null_entry || (rt->rt6i_flags & RTF_CACHE)) {
Martin KaFai Laud52d3992015-05-22 20:56:06 -07001082 dst_use(&rt->dst, jiffies);
1083 read_unlock_bh(&table->tb6_lock);
1084
1085 rt6_dst_from_metrics_check(rt);
David Ahernb8115802015-11-19 12:24:22 -08001086
1087 trace_fib6_table_lookup(net, rt, table->tb6_id, fl6);
Martin KaFai Laud52d3992015-05-22 20:56:06 -07001088 return rt;
Martin KaFai Lau3da59bd2015-05-22 20:56:03 -07001089 } else if (unlikely((fl6->flowi6_flags & FLOWI_FLAG_KNOWN_NH) &&
1090 !(rt->rt6i_flags & RTF_GATEWAY))) {
1091 /* Create a RTF_CACHE clone which will not be
1092 * owned by the fib6 tree. It is for the special case where
1093 * the daddr in the skb during the neighbor look-up is different
1094 * from the fl6->daddr used to look-up route here.
1095 */
Thomas Grafc71099a2006-08-04 23:20:06 -07001096
Martin KaFai Lau3da59bd2015-05-22 20:56:03 -07001097 struct rt6_info *uncached_rt;
1098
Martin KaFai Laud52d3992015-05-22 20:56:06 -07001099 dst_use(&rt->dst, jiffies);
1100 read_unlock_bh(&table->tb6_lock);
1101
Martin KaFai Lau3da59bd2015-05-22 20:56:03 -07001102 uncached_rt = ip6_rt_cache_alloc(rt, &fl6->daddr, NULL);
1103 dst_release(&rt->dst);
1104
1105 if (uncached_rt)
Martin KaFai Lau8d0b94a2015-05-22 20:56:04 -07001106 rt6_uncached_list_add(uncached_rt);
Martin KaFai Lau3da59bd2015-05-22 20:56:03 -07001107 else
1108 uncached_rt = net->ipv6.ip6_null_entry;
Martin KaFai Laud52d3992015-05-22 20:56:06 -07001109
Martin KaFai Lau3da59bd2015-05-22 20:56:03 -07001110 dst_hold(&uncached_rt->dst);
David Ahernb8115802015-11-19 12:24:22 -08001111
1112 trace_fib6_table_lookup(net, uncached_rt, table->tb6_id, fl6);
Martin KaFai Lau3da59bd2015-05-22 20:56:03 -07001113 return uncached_rt;
Martin KaFai Lau3da59bd2015-05-22 20:56:03 -07001114
Martin KaFai Laud52d3992015-05-22 20:56:06 -07001115 } else {
1116 /* Get a percpu copy */
1117
1118 struct rt6_info *pcpu_rt;
1119
1120 rt->dst.lastuse = jiffies;
1121 rt->dst.__use++;
1122 pcpu_rt = rt6_get_pcpu_route(rt);
Martin KaFai Laud52d3992015-05-22 20:56:06 -07001123
Martin KaFai Lau9c7370a2015-08-14 11:05:54 -07001124 if (pcpu_rt) {
1125 read_unlock_bh(&table->tb6_lock);
1126 } else {
1127 /* We have to do the read_unlock first
1128 * because rt6_make_pcpu_route() may trigger
1129 * ip6_dst_gc() which will take the write_lock.
1130 */
1131 dst_hold(&rt->dst);
1132 read_unlock_bh(&table->tb6_lock);
Martin KaFai Laua73e4192015-08-14 11:05:53 -07001133 pcpu_rt = rt6_make_pcpu_route(rt);
Martin KaFai Lau9c7370a2015-08-14 11:05:54 -07001134 dst_release(&rt->dst);
1135 }
Martin KaFai Laud52d3992015-05-22 20:56:06 -07001136
David Ahernb8115802015-11-19 12:24:22 -08001137 trace_fib6_table_lookup(net, pcpu_rt, table->tb6_id, fl6);
Martin KaFai Laud52d3992015-05-22 20:56:06 -07001138 return pcpu_rt;
Martin KaFai Lau9c7370a2015-08-14 11:05:54 -07001139
Martin KaFai Laud52d3992015-05-22 20:56:06 -07001140 }
Thomas Grafc71099a2006-08-04 23:20:06 -07001141}
1142
Daniel Lezcano8ed67782008-03-04 13:48:30 -08001143static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table,
David S. Miller4c9483b2011-03-12 16:22:43 -05001144 struct flowi6 *fl6, int flags)
Pavel Emelyanov4acad722007-10-15 13:02:51 -07001145{
David S. Miller4c9483b2011-03-12 16:22:43 -05001146 return ip6_pol_route(net, table, fl6->flowi6_iif, fl6, flags);
Pavel Emelyanov4acad722007-10-15 13:02:51 -07001147}
1148
Shmulik Ladkani72331bc2012-04-01 04:03:45 +00001149static struct dst_entry *ip6_route_input_lookup(struct net *net,
1150 struct net_device *dev,
1151 struct flowi6 *fl6, int flags)
1152{
1153 if (rt6_need_strict(&fl6->daddr) && dev->type != ARPHRD_PIMREG)
1154 flags |= RT6_LOOKUP_F_IFACE;
1155
1156 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_input);
1157}
1158
Thomas Grafc71099a2006-08-04 23:20:06 -07001159void ip6_route_input(struct sk_buff *skb)
1160{
Eric Dumazetb71d1d42011-04-22 04:53:02 +00001161 const struct ipv6hdr *iph = ipv6_hdr(skb);
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09001162 struct net *net = dev_net(skb->dev);
Thomas Grafadaa70b2006-10-13 15:01:03 -07001163 int flags = RT6_LOOKUP_F_HAS_SADDR;
Jiri Benc904af042015-08-20 13:56:31 +02001164 struct ip_tunnel_info *tun_info;
David S. Miller4c9483b2011-03-12 16:22:43 -05001165 struct flowi6 fl6 = {
David Ahernca254492015-10-12 11:47:10 -07001166 .flowi6_iif = l3mdev_fib_oif(skb->dev),
David S. Miller4c9483b2011-03-12 16:22:43 -05001167 .daddr = iph->daddr,
1168 .saddr = iph->saddr,
YOSHIFUJI Hideaki / 吉藤英明6502ca52013-01-13 05:01:51 +00001169 .flowlabel = ip6_flowinfo(iph),
David S. Miller4c9483b2011-03-12 16:22:43 -05001170 .flowi6_mark = skb->mark,
1171 .flowi6_proto = iph->nexthdr,
Thomas Grafc71099a2006-08-04 23:20:06 -07001172 };
Thomas Grafadaa70b2006-10-13 15:01:03 -07001173
Jiri Benc904af042015-08-20 13:56:31 +02001174 tun_info = skb_tunnel_info(skb);
Jiri Benc46fa0622015-08-28 20:48:19 +02001175 if (tun_info && !(tun_info->mode & IP_TUNNEL_INFO_TX))
Jiri Benc904af042015-08-20 13:56:31 +02001176 fl6.flowi6_tun_key.tun_id = tun_info->key.tun_id;
Jiri Benc06e9d042015-08-20 13:56:26 +02001177 skb_dst_drop(skb);
Shmulik Ladkani72331bc2012-04-01 04:03:45 +00001178 skb_dst_set(skb, ip6_route_input_lookup(net, skb->dev, &fl6, flags));
Thomas Grafc71099a2006-08-04 23:20:06 -07001179}
1180
Daniel Lezcano8ed67782008-03-04 13:48:30 -08001181static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table,
David S. Miller4c9483b2011-03-12 16:22:43 -05001182 struct flowi6 *fl6, int flags)
Thomas Grafc71099a2006-08-04 23:20:06 -07001183{
David S. Miller4c9483b2011-03-12 16:22:43 -05001184 return ip6_pol_route(net, table, fl6->flowi6_oif, fl6, flags);
Thomas Grafc71099a2006-08-04 23:20:06 -07001185}
1186
Paolo Abeni6f21c962016-01-29 12:30:19 +01001187struct dst_entry *ip6_route_output_flags(struct net *net, const struct sock *sk,
1188 struct flowi6 *fl6, int flags)
Thomas Grafc71099a2006-08-04 23:20:06 -07001189{
David Ahernca254492015-10-12 11:47:10 -07001190 struct dst_entry *dst;
David Ahernd46a9d62015-10-21 08:42:22 -07001191 bool any_src;
Thomas Grafc71099a2006-08-04 23:20:06 -07001192
David Ahern4a658962016-05-07 16:48:59 -07001193 dst = l3mdev_get_rt6_dst(net, fl6);
David Ahernca254492015-10-12 11:47:10 -07001194 if (dst)
1195 return dst;
1196
Pavel Emelyanov1fb94892012-08-08 21:53:36 +00001197 fl6->flowi6_iif = LOOPBACK_IFINDEX;
David McCullough4dc27d1c2012-06-25 15:42:26 +00001198
David Ahernd46a9d62015-10-21 08:42:22 -07001199 any_src = ipv6_addr_any(&fl6->saddr);
David Ahern741a11d2015-09-28 10:12:13 -07001200 if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl6->daddr) ||
David Ahernd46a9d62015-10-21 08:42:22 -07001201 (fl6->flowi6_oif && any_src))
YOSHIFUJI Hideaki77d16f42006-08-23 17:25:05 -07001202 flags |= RT6_LOOKUP_F_IFACE;
Thomas Grafc71099a2006-08-04 23:20:06 -07001203
David Ahernd46a9d62015-10-21 08:42:22 -07001204 if (!any_src)
Thomas Grafadaa70b2006-10-13 15:01:03 -07001205 flags |= RT6_LOOKUP_F_HAS_SADDR;
YOSHIFUJI Hideaki / 吉藤英明0c9a2ac2010-03-07 00:14:44 +00001206 else if (sk)
1207 flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs);
Thomas Grafadaa70b2006-10-13 15:01:03 -07001208
David S. Miller4c9483b2011-03-12 16:22:43 -05001209 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_output);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001210}
Paolo Abeni6f21c962016-01-29 12:30:19 +01001211EXPORT_SYMBOL_GPL(ip6_route_output_flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001212
David S. Miller2774c132011-03-01 14:59:04 -08001213struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_orig)
David S. Miller14e50e52007-05-24 18:17:54 -07001214{
David S. Miller5c1e6aa2011-04-28 14:13:38 -07001215 struct rt6_info *rt, *ort = (struct rt6_info *) dst_orig;
David S. Miller14e50e52007-05-24 18:17:54 -07001216 struct dst_entry *new = NULL;
1217
David S. Millerf5b0a872012-07-19 12:31:33 -07001218 rt = dst_alloc(&ip6_dst_blackhole_ops, ort->dst.dev, 1, DST_OBSOLETE_NONE, 0);
David S. Miller14e50e52007-05-24 18:17:54 -07001219 if (rt) {
Martin KaFai Lau0a1f5962015-10-15 16:39:58 -07001220 rt6_info_init(rt);
1221
Changli Gaod8d1f302010-06-10 23:31:35 -07001222 new = &rt->dst;
David S. Miller14e50e52007-05-24 18:17:54 -07001223 new->__use = 1;
Herbert Xu352e5122007-11-13 21:34:06 -08001224 new->input = dst_discard;
Eric W. Biedermanede20592015-10-07 16:48:47 -05001225 new->output = dst_discard_out;
David S. Miller14e50e52007-05-24 18:17:54 -07001226
Martin KaFai Lau0a1f5962015-10-15 16:39:58 -07001227 dst_copy_metrics(new, &ort->dst);
David S. Miller14e50e52007-05-24 18:17:54 -07001228 rt->rt6i_idev = ort->rt6i_idev;
1229 if (rt->rt6i_idev)
1230 in6_dev_hold(rt->rt6i_idev);
David S. Miller14e50e52007-05-24 18:17:54 -07001231
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00001232 rt->rt6i_gateway = ort->rt6i_gateway;
Martin KaFai Lau0a1f5962015-10-15 16:39:58 -07001233 rt->rt6i_flags = ort->rt6i_flags & ~RTF_PCPU;
David S. Miller14e50e52007-05-24 18:17:54 -07001234 rt->rt6i_metric = 0;
1235
1236 memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key));
1237#ifdef CONFIG_IPV6_SUBTREES
1238 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1239#endif
1240
1241 dst_free(new);
1242 }
1243
David S. Miller69ead7a2011-03-01 14:45:33 -08001244 dst_release(dst_orig);
1245 return new ? new : ERR_PTR(-ENOMEM);
David S. Miller14e50e52007-05-24 18:17:54 -07001246}
David S. Miller14e50e52007-05-24 18:17:54 -07001247
Linus Torvalds1da177e2005-04-16 15:20:36 -07001248/*
1249 * Destination cache support functions
1250 */
1251
Martin KaFai Lau4b32b5a2015-04-28 13:03:06 -07001252static void rt6_dst_from_metrics_check(struct rt6_info *rt)
1253{
1254 if (rt->dst.from &&
1255 dst_metrics_ptr(&rt->dst) != dst_metrics_ptr(rt->dst.from))
1256 dst_init_metrics(&rt->dst, dst_metrics_ptr(rt->dst.from), true);
1257}
1258
Martin KaFai Lau3da59bd2015-05-22 20:56:03 -07001259static struct dst_entry *rt6_check(struct rt6_info *rt, u32 cookie)
1260{
1261 if (!rt->rt6i_node || (rt->rt6i_node->fn_sernum != cookie))
1262 return NULL;
1263
1264 if (rt6_check_expired(rt))
1265 return NULL;
1266
1267 return &rt->dst;
1268}
1269
1270static struct dst_entry *rt6_dst_from_check(struct rt6_info *rt, u32 cookie)
1271{
Martin KaFai Lau5973fb12015-11-11 11:51:07 -08001272 if (!__rt6_check_expired(rt) &&
1273 rt->dst.obsolete == DST_OBSOLETE_FORCE_CHK &&
Martin KaFai Lau3da59bd2015-05-22 20:56:03 -07001274 rt6_check((struct rt6_info *)(rt->dst.from), cookie))
1275 return &rt->dst;
1276 else
1277 return NULL;
1278}
1279
Linus Torvalds1da177e2005-04-16 15:20:36 -07001280static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie)
1281{
1282 struct rt6_info *rt;
1283
1284 rt = (struct rt6_info *) dst;
1285
Nicolas Dichtel6f3118b2012-09-10 22:09:46 +00001286 /* All IPV6 dsts are created with ->obsolete set to the value
1287 * DST_OBSOLETE_FORCE_CHK which forces validation calls down
1288 * into this function always.
1289 */
Hannes Frederic Sowae3bc10b2013-10-24 07:48:24 +02001290
Martin KaFai Lau4b32b5a2015-04-28 13:03:06 -07001291 rt6_dst_from_metrics_check(rt);
1292
Martin KaFai Lau02bcf4e2015-11-11 11:51:08 -08001293 if (rt->rt6i_flags & RTF_PCPU ||
1294 (unlikely(dst->flags & DST_NOCACHE) && rt->dst.from))
Martin KaFai Lau3da59bd2015-05-22 20:56:03 -07001295 return rt6_dst_from_check(rt, cookie);
1296 else
1297 return rt6_check(rt, cookie);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001298}
1299
1300static struct dst_entry *ip6_negative_advice(struct dst_entry *dst)
1301{
1302 struct rt6_info *rt = (struct rt6_info *) dst;
1303
1304 if (rt) {
YOSHIFUJI Hideaki / 吉藤英明54c1a852010-03-28 07:15:45 +00001305 if (rt->rt6i_flags & RTF_CACHE) {
1306 if (rt6_check_expired(rt)) {
1307 ip6_del_rt(rt);
1308 dst = NULL;
1309 }
1310 } else {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001311 dst_release(dst);
YOSHIFUJI Hideaki / 吉藤英明54c1a852010-03-28 07:15:45 +00001312 dst = NULL;
1313 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001314 }
YOSHIFUJI Hideaki / 吉藤英明54c1a852010-03-28 07:15:45 +00001315 return dst;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001316}
1317
1318static void ip6_link_failure(struct sk_buff *skb)
1319{
1320 struct rt6_info *rt;
1321
Alexey Dobriyan3ffe5332010-02-18 08:25:24 +00001322 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001323
Eric Dumazetadf30902009-06-02 05:19:30 +00001324 rt = (struct rt6_info *) skb_dst(skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001325 if (rt) {
Hannes Frederic Sowa1eb4f752013-07-10 23:00:57 +02001326 if (rt->rt6i_flags & RTF_CACHE) {
1327 dst_hold(&rt->dst);
Martin KaFai Lau8e3d5be2015-09-15 14:30:08 -07001328 ip6_del_rt(rt);
Hannes Frederic Sowa1eb4f752013-07-10 23:00:57 +02001329 } else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001330 rt->rt6i_node->fn_sernum = -1;
Hannes Frederic Sowa1eb4f752013-07-10 23:00:57 +02001331 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001332 }
1333}
1334
Martin KaFai Lau45e4fd22015-05-22 20:56:00 -07001335static void rt6_do_update_pmtu(struct rt6_info *rt, u32 mtu)
1336{
1337 struct net *net = dev_net(rt->dst.dev);
1338
1339 rt->rt6i_flags |= RTF_MODIFIED;
1340 rt->rt6i_pmtu = mtu;
1341 rt6_update_expires(rt, net->ipv6.sysctl.ip6_rt_mtu_expires);
1342}
1343
Martin KaFai Lau0d3f6d22015-11-11 11:51:06 -08001344static bool rt6_cache_allowed_for_pmtu(const struct rt6_info *rt)
1345{
1346 return !(rt->rt6i_flags & RTF_CACHE) &&
1347 (rt->rt6i_flags & RTF_PCPU || rt->rt6i_node);
1348}
1349
Martin KaFai Lau45e4fd22015-05-22 20:56:00 -07001350static void __ip6_rt_update_pmtu(struct dst_entry *dst, const struct sock *sk,
1351 const struct ipv6hdr *iph, u32 mtu)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001352{
Ian Morris67ba4152014-08-24 21:53:10 +01001353 struct rt6_info *rt6 = (struct rt6_info *)dst;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001354
Martin KaFai Lau45e4fd22015-05-22 20:56:00 -07001355 if (rt6->rt6i_flags & RTF_LOCAL)
1356 return;
1357
David S. Miller81aded22012-06-15 14:54:11 -07001358 dst_confirm(dst);
Martin KaFai Lau45e4fd22015-05-22 20:56:00 -07001359 mtu = max_t(u32, mtu, IPV6_MIN_MTU);
1360 if (mtu >= dst_mtu(dst))
1361 return;
David S. Miller81aded22012-06-15 14:54:11 -07001362
Martin KaFai Lau0d3f6d22015-11-11 11:51:06 -08001363 if (!rt6_cache_allowed_for_pmtu(rt6)) {
Martin KaFai Lau45e4fd22015-05-22 20:56:00 -07001364 rt6_do_update_pmtu(rt6, mtu);
1365 } else {
1366 const struct in6_addr *daddr, *saddr;
1367 struct rt6_info *nrt6;
Hagen Paul Pfeifer9d289712015-01-15 22:34:25 +01001368
Martin KaFai Lau45e4fd22015-05-22 20:56:00 -07001369 if (iph) {
1370 daddr = &iph->daddr;
1371 saddr = &iph->saddr;
1372 } else if (sk) {
1373 daddr = &sk->sk_v6_daddr;
1374 saddr = &inet6_sk(sk)->saddr;
1375 } else {
1376 return;
1377 }
1378 nrt6 = ip6_rt_cache_alloc(rt6, daddr, saddr);
1379 if (nrt6) {
1380 rt6_do_update_pmtu(nrt6, mtu);
1381
1382 /* ip6_ins_rt(nrt6) will bump the
1383 * rt6->rt6i_node->fn_sernum
1384 * which will fail the next rt6_check() and
1385 * invalidate the sk->sk_dst_cache.
1386 */
1387 ip6_ins_rt(nrt6);
1388 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001389 }
1390}
1391
Martin KaFai Lau45e4fd22015-05-22 20:56:00 -07001392static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
1393 struct sk_buff *skb, u32 mtu)
1394{
1395 __ip6_rt_update_pmtu(dst, sk, skb ? ipv6_hdr(skb) : NULL, mtu);
1396}
1397
David S. Miller42ae66c2012-06-15 20:01:57 -07001398void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu,
1399 int oif, u32 mark)
David S. Miller81aded22012-06-15 14:54:11 -07001400{
1401 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1402 struct dst_entry *dst;
1403 struct flowi6 fl6;
1404
1405 memset(&fl6, 0, sizeof(fl6));
1406 fl6.flowi6_oif = oif;
Lorenzo Colitti1b3c61d2014-05-13 10:17:34 -07001407 fl6.flowi6_mark = mark ? mark : IP6_REPLY_MARK(net, skb->mark);
David S. Miller81aded22012-06-15 14:54:11 -07001408 fl6.daddr = iph->daddr;
1409 fl6.saddr = iph->saddr;
YOSHIFUJI Hideaki / 吉藤英明6502ca52013-01-13 05:01:51 +00001410 fl6.flowlabel = ip6_flowinfo(iph);
David S. Miller81aded22012-06-15 14:54:11 -07001411
1412 dst = ip6_route_output(net, NULL, &fl6);
1413 if (!dst->error)
Martin KaFai Lau45e4fd22015-05-22 20:56:00 -07001414 __ip6_rt_update_pmtu(dst, NULL, iph, ntohl(mtu));
David S. Miller81aded22012-06-15 14:54:11 -07001415 dst_release(dst);
1416}
1417EXPORT_SYMBOL_GPL(ip6_update_pmtu);
1418
1419void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu)
1420{
Martin KaFai Lau33c162a2016-04-11 15:29:36 -07001421 struct dst_entry *dst;
1422
David S. Miller81aded22012-06-15 14:54:11 -07001423 ip6_update_pmtu(skb, sock_net(sk), mtu,
1424 sk->sk_bound_dev_if, sk->sk_mark);
Martin KaFai Lau33c162a2016-04-11 15:29:36 -07001425
1426 dst = __sk_dst_get(sk);
1427 if (!dst || !dst->obsolete ||
1428 dst->ops->check(dst, inet6_sk(sk)->dst_cookie))
1429 return;
1430
1431 bh_lock_sock(sk);
1432 if (!sock_owned_by_user(sk) && !ipv6_addr_v4mapped(&sk->sk_v6_daddr))
1433 ip6_datagram_dst_update(sk, false);
1434 bh_unlock_sock(sk);
David S. Miller81aded22012-06-15 14:54:11 -07001435}
1436EXPORT_SYMBOL_GPL(ip6_sk_update_pmtu);
1437
Duan Jiongb55b76b2013-09-04 19:44:21 +08001438/* Handle redirects */
1439struct ip6rd_flowi {
1440 struct flowi6 fl6;
1441 struct in6_addr gateway;
1442};
1443
1444static struct rt6_info *__ip6_route_redirect(struct net *net,
1445 struct fib6_table *table,
1446 struct flowi6 *fl6,
1447 int flags)
1448{
1449 struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl6;
1450 struct rt6_info *rt;
1451 struct fib6_node *fn;
1452
1453 /* Get the "current" route for this destination and
1454 * check if the redirect has come from approriate router.
1455 *
1456 * RFC 4861 specifies that redirects should only be
1457 * accepted if they come from the nexthop to the target.
1458 * Due to the way the routes are chosen, this notion
1459 * is a bit fuzzy and one might need to check all possible
1460 * routes.
1461 */
1462
1463 read_lock_bh(&table->tb6_lock);
1464 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
1465restart:
1466 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1467 if (rt6_check_expired(rt))
1468 continue;
1469 if (rt->dst.error)
1470 break;
1471 if (!(rt->rt6i_flags & RTF_GATEWAY))
1472 continue;
1473 if (fl6->flowi6_oif != rt->dst.dev->ifindex)
1474 continue;
1475 if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway))
1476 continue;
1477 break;
1478 }
1479
1480 if (!rt)
1481 rt = net->ipv6.ip6_null_entry;
1482 else if (rt->dst.error) {
1483 rt = net->ipv6.ip6_null_entry;
Martin KaFai Laub0a1ba52015-01-20 19:16:02 -08001484 goto out;
1485 }
1486
1487 if (rt == net->ipv6.ip6_null_entry) {
Martin KaFai Laua3c00e42014-10-20 13:42:43 -07001488 fn = fib6_backtrack(fn, &fl6->saddr);
1489 if (fn)
1490 goto restart;
Duan Jiongb55b76b2013-09-04 19:44:21 +08001491 }
Martin KaFai Laua3c00e42014-10-20 13:42:43 -07001492
Martin KaFai Laub0a1ba52015-01-20 19:16:02 -08001493out:
Duan Jiongb55b76b2013-09-04 19:44:21 +08001494 dst_hold(&rt->dst);
1495
1496 read_unlock_bh(&table->tb6_lock);
1497
David Ahernb8115802015-11-19 12:24:22 -08001498 trace_fib6_table_lookup(net, rt, table->tb6_id, fl6);
Duan Jiongb55b76b2013-09-04 19:44:21 +08001499 return rt;
1500};
1501
1502static struct dst_entry *ip6_route_redirect(struct net *net,
1503 const struct flowi6 *fl6,
1504 const struct in6_addr *gateway)
1505{
1506 int flags = RT6_LOOKUP_F_HAS_SADDR;
1507 struct ip6rd_flowi rdfl;
1508
1509 rdfl.fl6 = *fl6;
1510 rdfl.gateway = *gateway;
1511
1512 return fib6_rule_lookup(net, &rdfl.fl6,
1513 flags, __ip6_route_redirect);
1514}
1515
David S. Miller3a5ad2e2012-07-12 00:08:07 -07001516void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark)
1517{
1518 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1519 struct dst_entry *dst;
1520 struct flowi6 fl6;
1521
1522 memset(&fl6, 0, sizeof(fl6));
Julian Anastasove374c612014-04-28 10:51:56 +03001523 fl6.flowi6_iif = LOOPBACK_IFINDEX;
David S. Miller3a5ad2e2012-07-12 00:08:07 -07001524 fl6.flowi6_oif = oif;
1525 fl6.flowi6_mark = mark;
David S. Miller3a5ad2e2012-07-12 00:08:07 -07001526 fl6.daddr = iph->daddr;
1527 fl6.saddr = iph->saddr;
YOSHIFUJI Hideaki / 吉藤英明6502ca52013-01-13 05:01:51 +00001528 fl6.flowlabel = ip6_flowinfo(iph);
David S. Miller3a5ad2e2012-07-12 00:08:07 -07001529
Duan Jiongb55b76b2013-09-04 19:44:21 +08001530 dst = ip6_route_redirect(net, &fl6, &ipv6_hdr(skb)->saddr);
1531 rt6_do_redirect(dst, NULL, skb);
David S. Miller3a5ad2e2012-07-12 00:08:07 -07001532 dst_release(dst);
1533}
1534EXPORT_SYMBOL_GPL(ip6_redirect);
1535
Duan Jiongc92a59e2013-08-22 12:07:35 +08001536void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif,
1537 u32 mark)
1538{
1539 const struct ipv6hdr *iph = ipv6_hdr(skb);
1540 const struct rd_msg *msg = (struct rd_msg *)icmp6_hdr(skb);
1541 struct dst_entry *dst;
1542 struct flowi6 fl6;
1543
1544 memset(&fl6, 0, sizeof(fl6));
Julian Anastasove374c612014-04-28 10:51:56 +03001545 fl6.flowi6_iif = LOOPBACK_IFINDEX;
Duan Jiongc92a59e2013-08-22 12:07:35 +08001546 fl6.flowi6_oif = oif;
1547 fl6.flowi6_mark = mark;
Duan Jiongc92a59e2013-08-22 12:07:35 +08001548 fl6.daddr = msg->dest;
1549 fl6.saddr = iph->daddr;
1550
Duan Jiongb55b76b2013-09-04 19:44:21 +08001551 dst = ip6_route_redirect(net, &fl6, &iph->saddr);
1552 rt6_do_redirect(dst, NULL, skb);
Duan Jiongc92a59e2013-08-22 12:07:35 +08001553 dst_release(dst);
1554}
1555
David S. Miller3a5ad2e2012-07-12 00:08:07 -07001556void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk)
1557{
1558 ip6_redirect(skb, sock_net(sk), sk->sk_bound_dev_if, sk->sk_mark);
1559}
1560EXPORT_SYMBOL_GPL(ip6_sk_redirect);
1561
David S. Miller0dbaee32010-12-13 12:52:14 -08001562static unsigned int ip6_default_advmss(const struct dst_entry *dst)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001563{
David S. Miller0dbaee32010-12-13 12:52:14 -08001564 struct net_device *dev = dst->dev;
1565 unsigned int mtu = dst_mtu(dst);
1566 struct net *net = dev_net(dev);
1567
Linus Torvalds1da177e2005-04-16 15:20:36 -07001568 mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr);
1569
Daniel Lezcano55786892008-03-04 13:47:47 -08001570 if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss)
1571 mtu = net->ipv6.sysctl.ip6_rt_min_advmss;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001572
1573 /*
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09001574 * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and
1575 * corresponding MSS is IPV6_MAXPLEN - tcp_header_size.
1576 * IPV6_MAXPLEN is also valid and means: "any MSS,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001577 * rely only on pmtu discovery"
1578 */
1579 if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr))
1580 mtu = IPV6_MAXPLEN;
1581 return mtu;
1582}
1583
Steffen Klassertebb762f2011-11-23 02:12:51 +00001584static unsigned int ip6_mtu(const struct dst_entry *dst)
David S. Millerd33e4552010-12-14 13:01:14 -08001585{
Martin KaFai Lau4b32b5a2015-04-28 13:03:06 -07001586 const struct rt6_info *rt = (const struct rt6_info *)dst;
1587 unsigned int mtu = rt->rt6i_pmtu;
David S. Millerd33e4552010-12-14 13:01:14 -08001588 struct inet6_dev *idev;
Steffen Klassert618f9bc2011-11-23 02:13:31 +00001589
1590 if (mtu)
Eric Dumazet30f78d82014-04-10 21:23:36 -07001591 goto out;
Steffen Klassert618f9bc2011-11-23 02:13:31 +00001592
Martin KaFai Lau4b32b5a2015-04-28 13:03:06 -07001593 mtu = dst_metric_raw(dst, RTAX_MTU);
1594 if (mtu)
1595 goto out;
1596
Steffen Klassert618f9bc2011-11-23 02:13:31 +00001597 mtu = IPV6_MIN_MTU;
David S. Millerd33e4552010-12-14 13:01:14 -08001598
1599 rcu_read_lock();
1600 idev = __in6_dev_get(dst->dev);
1601 if (idev)
1602 mtu = idev->cnf.mtu6;
1603 rcu_read_unlock();
1604
Eric Dumazet30f78d82014-04-10 21:23:36 -07001605out:
1606 return min_t(unsigned int, mtu, IP6_MAX_MTU);
David S. Millerd33e4552010-12-14 13:01:14 -08001607}
1608
YOSHIFUJI Hideaki3b009442007-12-06 16:11:48 -08001609static struct dst_entry *icmp6_dst_gc_list;
1610static DEFINE_SPINLOCK(icmp6_dst_lock);
Thomas Graf5d0bbee2006-08-04 03:37:36 -07001611
YOSHIFUJI Hideaki3b009442007-12-06 16:11:48 -08001612struct dst_entry *icmp6_dst_alloc(struct net_device *dev,
David S. Miller87a11572011-12-06 17:04:13 -05001613 struct flowi6 *fl6)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001614{
David S. Miller87a11572011-12-06 17:04:13 -05001615 struct dst_entry *dst;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001616 struct rt6_info *rt;
1617 struct inet6_dev *idev = in6_dev_get(dev);
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09001618 struct net *net = dev_net(dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001619
David S. Miller38308472011-12-03 18:02:47 -05001620 if (unlikely(!idev))
Eric Dumazet122bdf62012-03-14 21:13:11 +00001621 return ERR_PTR(-ENODEV);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001622
Martin KaFai Lauad706862015-08-14 11:05:52 -07001623 rt = ip6_dst_alloc(net, dev, 0);
David S. Miller38308472011-12-03 18:02:47 -05001624 if (unlikely(!rt)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001625 in6_dev_put(idev);
David S. Miller87a11572011-12-06 17:04:13 -05001626 dst = ERR_PTR(-ENOMEM);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001627 goto out;
1628 }
1629
Yan, Zheng8e2ec632011-09-05 21:34:30 +00001630 rt->dst.flags |= DST_HOST;
1631 rt->dst.output = ip6_output;
Changli Gaod8d1f302010-06-10 23:31:35 -07001632 atomic_set(&rt->dst.__refcnt, 1);
Julian Anastasov550bab42013-10-20 15:43:04 +03001633 rt->rt6i_gateway = fl6->daddr;
David S. Miller87a11572011-12-06 17:04:13 -05001634 rt->rt6i_dst.addr = fl6->daddr;
Yan, Zheng8e2ec632011-09-05 21:34:30 +00001635 rt->rt6i_dst.plen = 128;
1636 rt->rt6i_idev = idev;
Li RongQing14edd872012-10-24 14:01:18 +08001637 dst_metric_set(&rt->dst, RTAX_HOPLIMIT, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001638
YOSHIFUJI Hideaki3b009442007-12-06 16:11:48 -08001639 spin_lock_bh(&icmp6_dst_lock);
Changli Gaod8d1f302010-06-10 23:31:35 -07001640 rt->dst.next = icmp6_dst_gc_list;
1641 icmp6_dst_gc_list = &rt->dst;
YOSHIFUJI Hideaki3b009442007-12-06 16:11:48 -08001642 spin_unlock_bh(&icmp6_dst_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001643
Daniel Lezcano55786892008-03-04 13:47:47 -08001644 fib6_force_start_gc(net);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001645
David S. Miller87a11572011-12-06 17:04:13 -05001646 dst = xfrm_lookup(net, &rt->dst, flowi6_to_flowi(fl6), NULL, 0);
1647
Linus Torvalds1da177e2005-04-16 15:20:36 -07001648out:
David S. Miller87a11572011-12-06 17:04:13 -05001649 return dst;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001650}
1651
Stephen Hemminger3d0f24a2008-07-22 14:35:50 -07001652int icmp6_dst_gc(void)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001653{
Hagen Paul Pfeifere9476e92011-02-25 05:45:19 +00001654 struct dst_entry *dst, **pprev;
Stephen Hemminger3d0f24a2008-07-22 14:35:50 -07001655 int more = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001656
YOSHIFUJI Hideaki3b009442007-12-06 16:11:48 -08001657 spin_lock_bh(&icmp6_dst_lock);
1658 pprev = &icmp6_dst_gc_list;
Thomas Graf5d0bbee2006-08-04 03:37:36 -07001659
Linus Torvalds1da177e2005-04-16 15:20:36 -07001660 while ((dst = *pprev) != NULL) {
1661 if (!atomic_read(&dst->__refcnt)) {
1662 *pprev = dst->next;
1663 dst_free(dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001664 } else {
1665 pprev = &dst->next;
Stephen Hemminger3d0f24a2008-07-22 14:35:50 -07001666 ++more;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001667 }
1668 }
1669
YOSHIFUJI Hideaki3b009442007-12-06 16:11:48 -08001670 spin_unlock_bh(&icmp6_dst_lock);
Thomas Graf5d0bbee2006-08-04 03:37:36 -07001671
Stephen Hemminger3d0f24a2008-07-22 14:35:50 -07001672 return more;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001673}
1674
David S. Miller1e493d12008-09-10 17:27:15 -07001675static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg),
1676 void *arg)
1677{
1678 struct dst_entry *dst, **pprev;
1679
1680 spin_lock_bh(&icmp6_dst_lock);
1681 pprev = &icmp6_dst_gc_list;
1682 while ((dst = *pprev) != NULL) {
1683 struct rt6_info *rt = (struct rt6_info *) dst;
1684 if (func(rt, arg)) {
1685 *pprev = dst->next;
1686 dst_free(dst);
1687 } else {
1688 pprev = &dst->next;
1689 }
1690 }
1691 spin_unlock_bh(&icmp6_dst_lock);
1692}
1693
Daniel Lezcano569d3642008-01-18 03:56:57 -08001694static int ip6_dst_gc(struct dst_ops *ops)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001695{
Alexey Dobriyan86393e52009-08-29 01:34:49 +00001696 struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops);
Daniel Lezcano7019b782008-03-04 13:50:14 -08001697 int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval;
1698 int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size;
1699 int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity;
1700 int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout;
1701 unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc;
Eric Dumazetfc66f952010-10-08 06:37:34 +00001702 int entries;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001703
Eric Dumazetfc66f952010-10-08 06:37:34 +00001704 entries = dst_entries_get_fast(ops);
Michal Kubeček49a18d82013-08-01 10:04:24 +02001705 if (time_after(rt_last_gc + rt_min_interval, jiffies) &&
Eric Dumazetfc66f952010-10-08 06:37:34 +00001706 entries <= rt_max_size)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001707 goto out;
1708
Benjamin Thery6891a342008-03-04 13:49:47 -08001709 net->ipv6.ip6_rt_gc_expire++;
Li RongQing14956642014-05-19 17:30:28 +08001710 fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net, true);
Eric Dumazetfc66f952010-10-08 06:37:34 +00001711 entries = dst_entries_get_slow(ops);
1712 if (entries < ops->gc_thresh)
Daniel Lezcano7019b782008-03-04 13:50:14 -08001713 net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001714out:
Daniel Lezcano7019b782008-03-04 13:50:14 -08001715 net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity;
Eric Dumazetfc66f952010-10-08 06:37:34 +00001716 return entries > rt_max_size;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001717}
1718
Florian Westphale715b6d2015-01-05 23:57:44 +01001719static int ip6_convert_metrics(struct mx6_config *mxc,
1720 const struct fib6_config *cfg)
1721{
Daniel Borkmannc3a8d942015-08-31 15:58:47 +02001722 bool ecn_ca = false;
Florian Westphale715b6d2015-01-05 23:57:44 +01001723 struct nlattr *nla;
1724 int remaining;
1725 u32 *mp;
1726
Ian Morris63159f22015-03-29 14:00:04 +01001727 if (!cfg->fc_mx)
Florian Westphale715b6d2015-01-05 23:57:44 +01001728 return 0;
1729
1730 mp = kzalloc(sizeof(u32) * RTAX_MAX, GFP_KERNEL);
1731 if (unlikely(!mp))
1732 return -ENOMEM;
1733
1734 nla_for_each_attr(nla, cfg->fc_mx, cfg->fc_mx_len, remaining) {
1735 int type = nla_type(nla);
Daniel Borkmann1bb14802015-08-31 15:58:45 +02001736 u32 val;
Florian Westphale715b6d2015-01-05 23:57:44 +01001737
Daniel Borkmann1bb14802015-08-31 15:58:45 +02001738 if (!type)
1739 continue;
1740 if (unlikely(type > RTAX_MAX))
1741 goto err;
Daniel Borkmannea697632015-01-05 23:57:47 +01001742
Daniel Borkmann1bb14802015-08-31 15:58:45 +02001743 if (type == RTAX_CC_ALGO) {
1744 char tmp[TCP_CA_NAME_MAX];
1745
1746 nla_strlcpy(tmp, nla, sizeof(tmp));
Daniel Borkmannc3a8d942015-08-31 15:58:47 +02001747 val = tcp_ca_get_key_by_name(tmp, &ecn_ca);
Daniel Borkmann1bb14802015-08-31 15:58:45 +02001748 if (val == TCP_CA_UNSPEC)
Florian Westphale715b6d2015-01-05 23:57:44 +01001749 goto err;
Daniel Borkmann1bb14802015-08-31 15:58:45 +02001750 } else {
1751 val = nla_get_u32(nla);
Florian Westphale715b6d2015-01-05 23:57:44 +01001752 }
Paolo Abeni626abd52016-05-13 18:33:41 +02001753 if (type == RTAX_HOPLIMIT && val > 255)
1754 val = 255;
Daniel Borkmannb8d3e412015-08-31 15:58:46 +02001755 if (type == RTAX_FEATURES && (val & ~RTAX_FEATURE_MASK))
1756 goto err;
Daniel Borkmann1bb14802015-08-31 15:58:45 +02001757
1758 mp[type - 1] = val;
1759 __set_bit(type - 1, mxc->mx_valid);
Florian Westphale715b6d2015-01-05 23:57:44 +01001760 }
1761
Daniel Borkmannc3a8d942015-08-31 15:58:47 +02001762 if (ecn_ca) {
1763 __set_bit(RTAX_FEATURES - 1, mxc->mx_valid);
1764 mp[RTAX_FEATURES - 1] |= DST_FEATURE_ECN_CA;
1765 }
Florian Westphale715b6d2015-01-05 23:57:44 +01001766
Daniel Borkmannc3a8d942015-08-31 15:58:47 +02001767 mxc->mx = mp;
Florian Westphale715b6d2015-01-05 23:57:44 +01001768 return 0;
1769 err:
1770 kfree(mp);
1771 return -EINVAL;
1772}
Linus Torvalds1da177e2005-04-16 15:20:36 -07001773
David Ahern8c145862016-04-24 21:26:04 -07001774static struct rt6_info *ip6_nh_lookup_table(struct net *net,
1775 struct fib6_config *cfg,
1776 const struct in6_addr *gw_addr)
1777{
1778 struct flowi6 fl6 = {
1779 .flowi6_oif = cfg->fc_ifindex,
1780 .daddr = *gw_addr,
1781 .saddr = cfg->fc_prefsrc,
1782 };
1783 struct fib6_table *table;
1784 struct rt6_info *rt;
1785 int flags = 0;
1786
1787 table = fib6_get_table(net, cfg->fc_table);
1788 if (!table)
1789 return NULL;
1790
1791 if (!ipv6_addr_any(&cfg->fc_prefsrc))
1792 flags |= RT6_LOOKUP_F_HAS_SADDR;
1793
1794 rt = ip6_pol_route(net, table, cfg->fc_ifindex, &fl6, flags);
1795
1796 /* if table lookup failed, fall back to full lookup */
1797 if (rt == net->ipv6.ip6_null_entry) {
1798 ip6_rt_put(rt);
1799 rt = NULL;
1800 }
1801
1802 return rt;
1803}
1804
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07001805static struct rt6_info *ip6_route_info_create(struct fib6_config *cfg)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001806{
Daniel Lezcano55786892008-03-04 13:47:47 -08001807 struct net *net = cfg->fc_nlinfo.nl_net;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001808 struct rt6_info *rt = NULL;
1809 struct net_device *dev = NULL;
1810 struct inet6_dev *idev = NULL;
Thomas Grafc71099a2006-08-04 23:20:06 -07001811 struct fib6_table *table;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001812 int addr_type;
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07001813 int err = -EINVAL;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001814
Thomas Graf86872cb2006-08-22 00:01:08 -07001815 if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128)
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07001816 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001817#ifndef CONFIG_IPV6_SUBTREES
Thomas Graf86872cb2006-08-22 00:01:08 -07001818 if (cfg->fc_src_len)
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07001819 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001820#endif
Thomas Graf86872cb2006-08-22 00:01:08 -07001821 if (cfg->fc_ifindex) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001822 err = -ENODEV;
Daniel Lezcano55786892008-03-04 13:47:47 -08001823 dev = dev_get_by_index(net, cfg->fc_ifindex);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001824 if (!dev)
1825 goto out;
1826 idev = in6_dev_get(dev);
1827 if (!idev)
1828 goto out;
1829 }
1830
Thomas Graf86872cb2006-08-22 00:01:08 -07001831 if (cfg->fc_metric == 0)
1832 cfg->fc_metric = IP6_RT_PRIO_USER;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001833
Matti Vaittinend71314b2011-11-14 00:14:49 +00001834 err = -ENOBUFS;
David S. Miller38308472011-12-03 18:02:47 -05001835 if (cfg->fc_nlinfo.nlh &&
1836 !(cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_CREATE)) {
Matti Vaittinend71314b2011-11-14 00:14:49 +00001837 table = fib6_get_table(net, cfg->fc_table);
David S. Miller38308472011-12-03 18:02:47 -05001838 if (!table) {
Joe Perchesf3213832012-05-15 14:11:53 +00001839 pr_warn("NLM_F_CREATE should be specified when creating new route\n");
Matti Vaittinend71314b2011-11-14 00:14:49 +00001840 table = fib6_new_table(net, cfg->fc_table);
1841 }
1842 } else {
1843 table = fib6_new_table(net, cfg->fc_table);
1844 }
David S. Miller38308472011-12-03 18:02:47 -05001845
1846 if (!table)
Thomas Grafc71099a2006-08-04 23:20:06 -07001847 goto out;
Thomas Grafc71099a2006-08-04 23:20:06 -07001848
Martin KaFai Lauad706862015-08-14 11:05:52 -07001849 rt = ip6_dst_alloc(net, NULL,
1850 (cfg->fc_flags & RTF_ADDRCONF) ? 0 : DST_NOCOUNT);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001851
David S. Miller38308472011-12-03 18:02:47 -05001852 if (!rt) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001853 err = -ENOMEM;
1854 goto out;
1855 }
1856
Gao feng1716a962012-04-06 00:13:10 +00001857 if (cfg->fc_flags & RTF_EXPIRES)
1858 rt6_set_expires(rt, jiffies +
1859 clock_t_to_jiffies(cfg->fc_expires));
1860 else
1861 rt6_clean_expires(rt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001862
Thomas Graf86872cb2006-08-22 00:01:08 -07001863 if (cfg->fc_protocol == RTPROT_UNSPEC)
1864 cfg->fc_protocol = RTPROT_BOOT;
1865 rt->rt6i_protocol = cfg->fc_protocol;
1866
1867 addr_type = ipv6_addr_type(&cfg->fc_dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001868
1869 if (addr_type & IPV6_ADDR_MULTICAST)
Changli Gaod8d1f302010-06-10 23:31:35 -07001870 rt->dst.input = ip6_mc_input;
Maciej Żenczykowskiab79ad12010-09-27 00:07:02 +00001871 else if (cfg->fc_flags & RTF_LOCAL)
1872 rt->dst.input = ip6_input;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001873 else
Changli Gaod8d1f302010-06-10 23:31:35 -07001874 rt->dst.input = ip6_forward;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001875
Changli Gaod8d1f302010-06-10 23:31:35 -07001876 rt->dst.output = ip6_output;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001877
Roopa Prabhu19e42e42015-07-21 10:43:48 +02001878 if (cfg->fc_encap) {
1879 struct lwtunnel_state *lwtstate;
1880
1881 err = lwtunnel_build_state(dev, cfg->fc_encap_type,
Tom Herbert127eb7c2015-08-24 09:45:41 -07001882 cfg->fc_encap, AF_INET6, cfg,
1883 &lwtstate);
Roopa Prabhu19e42e42015-07-21 10:43:48 +02001884 if (err)
1885 goto out;
Jiri Benc61adedf2015-08-20 13:56:25 +02001886 rt->dst.lwtstate = lwtstate_get(lwtstate);
1887 if (lwtunnel_output_redirect(rt->dst.lwtstate)) {
1888 rt->dst.lwtstate->orig_output = rt->dst.output;
1889 rt->dst.output = lwtunnel_output;
Tom Herbert25368622015-08-17 13:42:24 -07001890 }
Jiri Benc61adedf2015-08-20 13:56:25 +02001891 if (lwtunnel_input_redirect(rt->dst.lwtstate)) {
1892 rt->dst.lwtstate->orig_input = rt->dst.input;
1893 rt->dst.input = lwtunnel_input;
Tom Herbert25368622015-08-17 13:42:24 -07001894 }
Roopa Prabhu19e42e42015-07-21 10:43:48 +02001895 }
1896
Thomas Graf86872cb2006-08-22 00:01:08 -07001897 ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len);
1898 rt->rt6i_dst.plen = cfg->fc_dst_len;
Martin KaFai Lauafc4eef2015-04-28 13:03:07 -07001899 if (rt->rt6i_dst.plen == 128)
Michal Kubečeke5fd3872014-03-27 13:04:08 +01001900 rt->dst.flags |= DST_HOST;
Michal Kubečeke5fd3872014-03-27 13:04:08 +01001901
Linus Torvalds1da177e2005-04-16 15:20:36 -07001902#ifdef CONFIG_IPV6_SUBTREES
Thomas Graf86872cb2006-08-22 00:01:08 -07001903 ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len);
1904 rt->rt6i_src.plen = cfg->fc_src_len;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001905#endif
1906
Thomas Graf86872cb2006-08-22 00:01:08 -07001907 rt->rt6i_metric = cfg->fc_metric;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001908
1909 /* We cannot add true routes via loopback here,
1910 they would result in kernel looping; promote them to reject routes
1911 */
Thomas Graf86872cb2006-08-22 00:01:08 -07001912 if ((cfg->fc_flags & RTF_REJECT) ||
David S. Miller38308472011-12-03 18:02:47 -05001913 (dev && (dev->flags & IFF_LOOPBACK) &&
1914 !(addr_type & IPV6_ADDR_LOOPBACK) &&
1915 !(cfg->fc_flags & RTF_LOCAL))) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001916 /* hold loopback dev/idev if we haven't done so. */
Daniel Lezcano55786892008-03-04 13:47:47 -08001917 if (dev != net->loopback_dev) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001918 if (dev) {
1919 dev_put(dev);
1920 in6_dev_put(idev);
1921 }
Daniel Lezcano55786892008-03-04 13:47:47 -08001922 dev = net->loopback_dev;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001923 dev_hold(dev);
1924 idev = in6_dev_get(dev);
1925 if (!idev) {
1926 err = -ENODEV;
1927 goto out;
1928 }
1929 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001930 rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP;
Nicolas Dichtelef2c7d72012-09-05 02:12:42 +00001931 switch (cfg->fc_type) {
1932 case RTN_BLACKHOLE:
1933 rt->dst.error = -EINVAL;
Eric W. Biedermanede20592015-10-07 16:48:47 -05001934 rt->dst.output = dst_discard_out;
Kamala R7150aed2013-12-02 19:55:21 +05301935 rt->dst.input = dst_discard;
Nicolas Dichtelef2c7d72012-09-05 02:12:42 +00001936 break;
1937 case RTN_PROHIBIT:
1938 rt->dst.error = -EACCES;
Kamala R7150aed2013-12-02 19:55:21 +05301939 rt->dst.output = ip6_pkt_prohibit_out;
1940 rt->dst.input = ip6_pkt_prohibit;
Nicolas Dichtelef2c7d72012-09-05 02:12:42 +00001941 break;
Nicolas Dichtelb4949ab2012-09-06 05:53:35 +00001942 case RTN_THROW:
Nikola Forró0315e382015-09-17 16:01:32 +02001943 case RTN_UNREACHABLE:
Nicolas Dichtelef2c7d72012-09-05 02:12:42 +00001944 default:
Kamala R7150aed2013-12-02 19:55:21 +05301945 rt->dst.error = (cfg->fc_type == RTN_THROW) ? -EAGAIN
Nikola Forró0315e382015-09-17 16:01:32 +02001946 : (cfg->fc_type == RTN_UNREACHABLE)
1947 ? -EHOSTUNREACH : -ENETUNREACH;
Kamala R7150aed2013-12-02 19:55:21 +05301948 rt->dst.output = ip6_pkt_discard_out;
1949 rt->dst.input = ip6_pkt_discard;
Nicolas Dichtelef2c7d72012-09-05 02:12:42 +00001950 break;
1951 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001952 goto install_route;
1953 }
1954
Thomas Graf86872cb2006-08-22 00:01:08 -07001955 if (cfg->fc_flags & RTF_GATEWAY) {
Eric Dumazetb71d1d42011-04-22 04:53:02 +00001956 const struct in6_addr *gw_addr;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001957 int gwa_type;
1958
Thomas Graf86872cb2006-08-22 00:01:08 -07001959 gw_addr = &cfg->fc_gateway;
Florian Westphal330567b2015-08-07 10:54:28 +02001960 gwa_type = ipv6_addr_type(gw_addr);
Florian Westphal48ed7b22015-05-21 00:25:41 +02001961
1962 /* if gw_addr is local we will fail to detect this in case
1963 * address is still TENTATIVE (DAD in progress). rt6_lookup()
1964 * will return already-added prefix route via interface that
1965 * prefix route was assigned to, which might be non-loopback.
1966 */
1967 err = -EINVAL;
Florian Westphal330567b2015-08-07 10:54:28 +02001968 if (ipv6_chk_addr_and_flags(net, gw_addr,
1969 gwa_type & IPV6_ADDR_LINKLOCAL ?
1970 dev : NULL, 0, 0))
Florian Westphal48ed7b22015-05-21 00:25:41 +02001971 goto out;
1972
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00001973 rt->rt6i_gateway = *gw_addr;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001974
1975 if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) {
David Ahern8c145862016-04-24 21:26:04 -07001976 struct rt6_info *grt = NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001977
1978 /* IPv6 strictly inhibits using not link-local
1979 addresses as nexthop address.
1980 Otherwise, router will not able to send redirects.
1981 It is very good, but in some (rare!) circumstances
1982 (SIT, PtP, NBMA NOARP links) it is handy to allow
1983 some exceptions. --ANK
1984 */
David S. Miller38308472011-12-03 18:02:47 -05001985 if (!(gwa_type & IPV6_ADDR_UNICAST))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001986 goto out;
1987
David Ahern8c145862016-04-24 21:26:04 -07001988 if (cfg->fc_table)
1989 grt = ip6_nh_lookup_table(net, cfg, gw_addr);
1990
1991 if (!grt)
1992 grt = rt6_lookup(net, gw_addr, NULL,
1993 cfg->fc_ifindex, 1);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001994
1995 err = -EHOSTUNREACH;
David S. Miller38308472011-12-03 18:02:47 -05001996 if (!grt)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001997 goto out;
1998 if (dev) {
David S. Millerd1918542011-12-28 20:19:20 -05001999 if (dev != grt->dst.dev) {
Amerigo Wang94e187c2012-10-29 00:13:19 +00002000 ip6_rt_put(grt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002001 goto out;
2002 }
2003 } else {
David S. Millerd1918542011-12-28 20:19:20 -05002004 dev = grt->dst.dev;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002005 idev = grt->rt6i_idev;
2006 dev_hold(dev);
2007 in6_dev_hold(grt->rt6i_idev);
2008 }
David S. Miller38308472011-12-03 18:02:47 -05002009 if (!(grt->rt6i_flags & RTF_GATEWAY))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002010 err = 0;
Amerigo Wang94e187c2012-10-29 00:13:19 +00002011 ip6_rt_put(grt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002012
2013 if (err)
2014 goto out;
2015 }
2016 err = -EINVAL;
David S. Miller38308472011-12-03 18:02:47 -05002017 if (!dev || (dev->flags & IFF_LOOPBACK))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002018 goto out;
2019 }
2020
2021 err = -ENODEV;
David S. Miller38308472011-12-03 18:02:47 -05002022 if (!dev)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002023 goto out;
2024
Daniel Walterc3968a82011-04-13 21:10:57 +00002025 if (!ipv6_addr_any(&cfg->fc_prefsrc)) {
2026 if (!ipv6_chk_addr(net, &cfg->fc_prefsrc, dev, 0)) {
2027 err = -EINVAL;
2028 goto out;
2029 }
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00002030 rt->rt6i_prefsrc.addr = cfg->fc_prefsrc;
Daniel Walterc3968a82011-04-13 21:10:57 +00002031 rt->rt6i_prefsrc.plen = 128;
2032 } else
2033 rt->rt6i_prefsrc.plen = 0;
2034
Thomas Graf86872cb2006-08-22 00:01:08 -07002035 rt->rt6i_flags = cfg->fc_flags;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002036
2037install_route:
Changli Gaod8d1f302010-06-10 23:31:35 -07002038 rt->dst.dev = dev;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002039 rt->rt6i_idev = idev;
Thomas Grafc71099a2006-08-04 23:20:06 -07002040 rt->rt6i_table = table;
Daniel Lezcano63152fc2008-03-03 23:31:11 -08002041
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09002042 cfg->fc_nlinfo.nl_net = dev_net(dev);
Daniel Lezcano63152fc2008-03-03 23:31:11 -08002043
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07002044 return rt;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002045out:
2046 if (dev)
2047 dev_put(dev);
2048 if (idev)
2049 in6_dev_put(idev);
2050 if (rt)
Changli Gaod8d1f302010-06-10 23:31:35 -07002051 dst_free(&rt->dst);
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002052
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07002053 return ERR_PTR(err);
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002054}
2055
2056int ip6_route_add(struct fib6_config *cfg)
2057{
2058 struct mx6_config mxc = { .mx = NULL, };
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07002059 struct rt6_info *rt;
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002060 int err;
2061
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07002062 rt = ip6_route_info_create(cfg);
2063 if (IS_ERR(rt)) {
2064 err = PTR_ERR(rt);
2065 rt = NULL;
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002066 goto out;
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07002067 }
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002068
2069 err = ip6_convert_metrics(&mxc, cfg);
2070 if (err)
2071 goto out;
2072
2073 err = __ip6_ins_rt(rt, &cfg->fc_nlinfo, &mxc);
2074
2075 kfree(mxc.mx);
2076
2077 return err;
2078out:
2079 if (rt)
2080 dst_free(&rt->dst);
2081
Linus Torvalds1da177e2005-04-16 15:20:36 -07002082 return err;
2083}
2084
Thomas Graf86872cb2006-08-22 00:01:08 -07002085static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002086{
2087 int err;
Thomas Grafc71099a2006-08-04 23:20:06 -07002088 struct fib6_table *table;
David S. Millerd1918542011-12-28 20:19:20 -05002089 struct net *net = dev_net(rt->dst.dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002090
Martin KaFai Lau8e3d5be2015-09-15 14:30:08 -07002091 if (rt == net->ipv6.ip6_null_entry ||
2092 rt->dst.flags & DST_NOCACHE) {
Gao feng6825a262012-09-19 19:25:34 +00002093 err = -ENOENT;
2094 goto out;
2095 }
Patrick McHardy6c813a72006-08-06 22:22:47 -07002096
Thomas Grafc71099a2006-08-04 23:20:06 -07002097 table = rt->rt6i_table;
2098 write_lock_bh(&table->tb6_lock);
Thomas Graf86872cb2006-08-22 00:01:08 -07002099 err = fib6_del(rt, info);
Thomas Grafc71099a2006-08-04 23:20:06 -07002100 write_unlock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002101
Gao feng6825a262012-09-19 19:25:34 +00002102out:
Amerigo Wang94e187c2012-10-29 00:13:19 +00002103 ip6_rt_put(rt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002104 return err;
2105}
2106
Thomas Grafe0a1ad732006-08-22 00:00:21 -07002107int ip6_del_rt(struct rt6_info *rt)
2108{
Denis V. Lunev4d1169c2008-01-10 03:26:13 -08002109 struct nl_info info = {
David S. Millerd1918542011-12-28 20:19:20 -05002110 .nl_net = dev_net(rt->dst.dev),
Denis V. Lunev4d1169c2008-01-10 03:26:13 -08002111 };
Denis V. Lunev528c4ce2007-12-13 09:45:12 -08002112 return __ip6_del_rt(rt, &info);
Thomas Grafe0a1ad732006-08-22 00:00:21 -07002113}
2114
Thomas Graf86872cb2006-08-22 00:01:08 -07002115static int ip6_route_del(struct fib6_config *cfg)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002116{
Thomas Grafc71099a2006-08-04 23:20:06 -07002117 struct fib6_table *table;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002118 struct fib6_node *fn;
2119 struct rt6_info *rt;
2120 int err = -ESRCH;
2121
Daniel Lezcano55786892008-03-04 13:47:47 -08002122 table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table);
David S. Miller38308472011-12-03 18:02:47 -05002123 if (!table)
Thomas Grafc71099a2006-08-04 23:20:06 -07002124 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002125
Thomas Grafc71099a2006-08-04 23:20:06 -07002126 read_lock_bh(&table->tb6_lock);
2127
2128 fn = fib6_locate(&table->tb6_root,
Thomas Graf86872cb2006-08-22 00:01:08 -07002129 &cfg->fc_dst, cfg->fc_dst_len,
2130 &cfg->fc_src, cfg->fc_src_len);
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09002131
Linus Torvalds1da177e2005-04-16 15:20:36 -07002132 if (fn) {
Changli Gaod8d1f302010-06-10 23:31:35 -07002133 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
Martin KaFai Lau1f56a012015-04-28 13:03:03 -07002134 if ((rt->rt6i_flags & RTF_CACHE) &&
2135 !(cfg->fc_flags & RTF_CACHE))
2136 continue;
Thomas Graf86872cb2006-08-22 00:01:08 -07002137 if (cfg->fc_ifindex &&
David S. Millerd1918542011-12-28 20:19:20 -05002138 (!rt->dst.dev ||
2139 rt->dst.dev->ifindex != cfg->fc_ifindex))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002140 continue;
Thomas Graf86872cb2006-08-22 00:01:08 -07002141 if (cfg->fc_flags & RTF_GATEWAY &&
2142 !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002143 continue;
Thomas Graf86872cb2006-08-22 00:01:08 -07002144 if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002145 continue;
Changli Gaod8d1f302010-06-10 23:31:35 -07002146 dst_hold(&rt->dst);
Thomas Grafc71099a2006-08-04 23:20:06 -07002147 read_unlock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002148
Thomas Graf86872cb2006-08-22 00:01:08 -07002149 return __ip6_del_rt(rt, &cfg->fc_nlinfo);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002150 }
2151 }
Thomas Grafc71099a2006-08-04 23:20:06 -07002152 read_unlock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002153
2154 return err;
2155}
2156
David S. Miller6700c272012-07-17 03:29:28 -07002157static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_buff *skb)
YOSHIFUJI Hideakia6279452006-08-23 17:18:26 -07002158{
YOSHIFUJI Hideakia6279452006-08-23 17:18:26 -07002159 struct netevent_redirect netevent;
David S. Millere8599ff2012-07-11 23:43:53 -07002160 struct rt6_info *rt, *nrt = NULL;
David S. Millere8599ff2012-07-11 23:43:53 -07002161 struct ndisc_options ndopts;
2162 struct inet6_dev *in6_dev;
2163 struct neighbour *neigh;
YOSHIFUJI Hideaki / 吉藤英明71bcdba2013-01-05 16:34:51 +00002164 struct rd_msg *msg;
David S. Miller6e157b62012-07-12 00:05:02 -07002165 int optlen, on_link;
2166 u8 *lladdr;
David S. Millere8599ff2012-07-11 23:43:53 -07002167
Simon Horman29a3cad2013-05-28 20:34:26 +00002168 optlen = skb_tail_pointer(skb) - skb_transport_header(skb);
YOSHIFUJI Hideaki / 吉藤英明71bcdba2013-01-05 16:34:51 +00002169 optlen -= sizeof(*msg);
David S. Millere8599ff2012-07-11 23:43:53 -07002170
2171 if (optlen < 0) {
David S. Miller6e157b62012-07-12 00:05:02 -07002172 net_dbg_ratelimited("rt6_do_redirect: packet too short\n");
David S. Millere8599ff2012-07-11 23:43:53 -07002173 return;
2174 }
2175
YOSHIFUJI Hideaki / 吉藤英明71bcdba2013-01-05 16:34:51 +00002176 msg = (struct rd_msg *)icmp6_hdr(skb);
David S. Millere8599ff2012-07-11 23:43:53 -07002177
YOSHIFUJI Hideaki / 吉藤英明71bcdba2013-01-05 16:34:51 +00002178 if (ipv6_addr_is_multicast(&msg->dest)) {
David S. Miller6e157b62012-07-12 00:05:02 -07002179 net_dbg_ratelimited("rt6_do_redirect: destination address is multicast\n");
David S. Millere8599ff2012-07-11 23:43:53 -07002180 return;
2181 }
2182
David S. Miller6e157b62012-07-12 00:05:02 -07002183 on_link = 0;
YOSHIFUJI Hideaki / 吉藤英明71bcdba2013-01-05 16:34:51 +00002184 if (ipv6_addr_equal(&msg->dest, &msg->target)) {
David S. Millere8599ff2012-07-11 23:43:53 -07002185 on_link = 1;
YOSHIFUJI Hideaki / 吉藤英明71bcdba2013-01-05 16:34:51 +00002186 } else if (ipv6_addr_type(&msg->target) !=
David S. Millere8599ff2012-07-11 23:43:53 -07002187 (IPV6_ADDR_UNICAST|IPV6_ADDR_LINKLOCAL)) {
David S. Miller6e157b62012-07-12 00:05:02 -07002188 net_dbg_ratelimited("rt6_do_redirect: target address is not link-local unicast\n");
David S. Millere8599ff2012-07-11 23:43:53 -07002189 return;
2190 }
2191
2192 in6_dev = __in6_dev_get(skb->dev);
2193 if (!in6_dev)
2194 return;
2195 if (in6_dev->cnf.forwarding || !in6_dev->cnf.accept_redirects)
2196 return;
2197
2198 /* RFC2461 8.1:
2199 * The IP source address of the Redirect MUST be the same as the current
2200 * first-hop router for the specified ICMP Destination Address.
2201 */
2202
YOSHIFUJI Hideaki / 吉藤英明71bcdba2013-01-05 16:34:51 +00002203 if (!ndisc_parse_options(msg->opt, optlen, &ndopts)) {
David S. Millere8599ff2012-07-11 23:43:53 -07002204 net_dbg_ratelimited("rt6_redirect: invalid ND options\n");
2205 return;
2206 }
David S. Miller6e157b62012-07-12 00:05:02 -07002207
2208 lladdr = NULL;
David S. Millere8599ff2012-07-11 23:43:53 -07002209 if (ndopts.nd_opts_tgt_lladdr) {
2210 lladdr = ndisc_opt_addr_data(ndopts.nd_opts_tgt_lladdr,
2211 skb->dev);
2212 if (!lladdr) {
2213 net_dbg_ratelimited("rt6_redirect: invalid link-layer address length\n");
2214 return;
2215 }
2216 }
2217
David S. Miller6e157b62012-07-12 00:05:02 -07002218 rt = (struct rt6_info *) dst;
Matthias Schifferec13ad12015-11-02 01:24:38 +01002219 if (rt->rt6i_flags & RTF_REJECT) {
David S. Miller6e157b62012-07-12 00:05:02 -07002220 net_dbg_ratelimited("rt6_redirect: source isn't a valid nexthop for redirect target\n");
2221 return;
2222 }
2223
2224 /* Redirect received -> path was valid.
2225 * Look, redirects are sent only in response to data packets,
2226 * so that this nexthop apparently is reachable. --ANK
2227 */
2228 dst_confirm(&rt->dst);
2229
YOSHIFUJI Hideaki / 吉藤英明71bcdba2013-01-05 16:34:51 +00002230 neigh = __neigh_lookup(&nd_tbl, &msg->target, skb->dev, 1);
David S. Millere8599ff2012-07-11 23:43:53 -07002231 if (!neigh)
2232 return;
2233
Linus Torvalds1da177e2005-04-16 15:20:36 -07002234 /*
2235 * We have finally decided to accept it.
2236 */
2237
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09002238 neigh_update(neigh, lladdr, NUD_STALE,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002239 NEIGH_UPDATE_F_WEAK_OVERRIDE|
2240 NEIGH_UPDATE_F_OVERRIDE|
2241 (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER|
2242 NEIGH_UPDATE_F_ISROUTER))
2243 );
2244
Martin KaFai Lau83a09ab2015-05-22 20:56:05 -07002245 nrt = ip6_rt_cache_alloc(rt, &msg->dest, NULL);
David S. Miller38308472011-12-03 18:02:47 -05002246 if (!nrt)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002247 goto out;
2248
2249 nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE;
2250 if (on_link)
2251 nrt->rt6i_flags &= ~RTF_GATEWAY;
2252
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00002253 nrt->rt6i_gateway = *(struct in6_addr *)neigh->primary_key;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002254
Thomas Graf40e22e82006-08-22 00:00:45 -07002255 if (ip6_ins_rt(nrt))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002256 goto out;
2257
Changli Gaod8d1f302010-06-10 23:31:35 -07002258 netevent.old = &rt->dst;
2259 netevent.new = &nrt->dst;
YOSHIFUJI Hideaki / 吉藤英明71bcdba2013-01-05 16:34:51 +00002260 netevent.daddr = &msg->dest;
YOSHIFUJI Hideaki / 吉藤英明60592832013-01-14 09:28:27 +00002261 netevent.neigh = neigh;
Tom Tucker8d717402006-07-30 20:43:36 -07002262 call_netevent_notifiers(NETEVENT_REDIRECT, &netevent);
2263
David S. Miller38308472011-12-03 18:02:47 -05002264 if (rt->rt6i_flags & RTF_CACHE) {
David S. Miller6e157b62012-07-12 00:05:02 -07002265 rt = (struct rt6_info *) dst_clone(&rt->dst);
Thomas Grafe0a1ad732006-08-22 00:00:21 -07002266 ip6_del_rt(rt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002267 }
2268
2269out:
David S. Millere8599ff2012-07-11 23:43:53 -07002270 neigh_release(neigh);
David S. Miller6e157b62012-07-12 00:05:02 -07002271}
2272
Linus Torvalds1da177e2005-04-16 15:20:36 -07002273/*
Linus Torvalds1da177e2005-04-16 15:20:36 -07002274 * Misc support functions
2275 */
2276
Martin KaFai Lau4b32b5a2015-04-28 13:03:06 -07002277static void rt6_set_from(struct rt6_info *rt, struct rt6_info *from)
2278{
2279 BUG_ON(from->dst.from);
2280
2281 rt->rt6i_flags &= ~RTF_EXPIRES;
2282 dst_hold(&from->dst);
2283 rt->dst.from = &from->dst;
2284 dst_init_metrics(&rt->dst, dst_metrics_ptr(&from->dst), true);
2285}
2286
Martin KaFai Lau83a09ab2015-05-22 20:56:05 -07002287static void ip6_rt_copy_init(struct rt6_info *rt, struct rt6_info *ort)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002288{
Martin KaFai Lau83a09ab2015-05-22 20:56:05 -07002289 rt->dst.input = ort->dst.input;
2290 rt->dst.output = ort->dst.output;
2291 rt->rt6i_dst = ort->rt6i_dst;
2292 rt->dst.error = ort->dst.error;
2293 rt->rt6i_idev = ort->rt6i_idev;
2294 if (rt->rt6i_idev)
2295 in6_dev_hold(rt->rt6i_idev);
2296 rt->dst.lastuse = jiffies;
2297 rt->rt6i_gateway = ort->rt6i_gateway;
2298 rt->rt6i_flags = ort->rt6i_flags;
2299 rt6_set_from(rt, ort);
2300 rt->rt6i_metric = ort->rt6i_metric;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002301#ifdef CONFIG_IPV6_SUBTREES
Martin KaFai Lau83a09ab2015-05-22 20:56:05 -07002302 rt->rt6i_src = ort->rt6i_src;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002303#endif
Martin KaFai Lau83a09ab2015-05-22 20:56:05 -07002304 rt->rt6i_prefsrc = ort->rt6i_prefsrc;
2305 rt->rt6i_table = ort->rt6i_table;
Jiri Benc61adedf2015-08-20 13:56:25 +02002306 rt->dst.lwtstate = lwtstate_get(ort->dst.lwtstate);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002307}
2308
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002309#ifdef CONFIG_IPV6_ROUTE_INFO
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -08002310static struct rt6_info *rt6_get_route_info(struct net *net,
Eric Dumazetb71d1d42011-04-22 04:53:02 +00002311 const struct in6_addr *prefix, int prefixlen,
2312 const struct in6_addr *gwaddr, int ifindex)
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002313{
2314 struct fib6_node *fn;
2315 struct rt6_info *rt = NULL;
Thomas Grafc71099a2006-08-04 23:20:06 -07002316 struct fib6_table *table;
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002317
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -08002318 table = fib6_get_table(net, RT6_TABLE_INFO);
David S. Miller38308472011-12-03 18:02:47 -05002319 if (!table)
Thomas Grafc71099a2006-08-04 23:20:06 -07002320 return NULL;
2321
Li RongQing5744dd92012-09-11 21:59:01 +00002322 read_lock_bh(&table->tb6_lock);
Ian Morris67ba4152014-08-24 21:53:10 +01002323 fn = fib6_locate(&table->tb6_root, prefix, prefixlen, NULL, 0);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002324 if (!fn)
2325 goto out;
2326
Changli Gaod8d1f302010-06-10 23:31:35 -07002327 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
David S. Millerd1918542011-12-28 20:19:20 -05002328 if (rt->dst.dev->ifindex != ifindex)
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002329 continue;
2330 if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY))
2331 continue;
2332 if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr))
2333 continue;
Changli Gaod8d1f302010-06-10 23:31:35 -07002334 dst_hold(&rt->dst);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002335 break;
2336 }
2337out:
Li RongQing5744dd92012-09-11 21:59:01 +00002338 read_unlock_bh(&table->tb6_lock);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002339 return rt;
2340}
2341
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -08002342static struct rt6_info *rt6_add_route_info(struct net *net,
Eric Dumazetb71d1d42011-04-22 04:53:02 +00002343 const struct in6_addr *prefix, int prefixlen,
2344 const struct in6_addr *gwaddr, int ifindex,
Eric Dumazet95c96172012-04-15 05:58:06 +00002345 unsigned int pref)
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002346{
Thomas Graf86872cb2006-08-22 00:01:08 -07002347 struct fib6_config cfg = {
Rami Rosen238fc7e2008-02-09 23:43:11 -08002348 .fc_metric = IP6_RT_PRIO_USER,
Thomas Graf86872cb2006-08-22 00:01:08 -07002349 .fc_ifindex = ifindex,
2350 .fc_dst_len = prefixlen,
2351 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO |
2352 RTF_UP | RTF_PREF(pref),
Eric W. Biederman15e47302012-09-07 20:12:54 +00002353 .fc_nlinfo.portid = 0,
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -08002354 .fc_nlinfo.nlh = NULL,
2355 .fc_nlinfo.nl_net = net,
Thomas Graf86872cb2006-08-22 00:01:08 -07002356 };
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002357
David Ahernca254492015-10-12 11:47:10 -07002358 cfg.fc_table = l3mdev_fib_table_by_index(net, ifindex) ? : RT6_TABLE_INFO;
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00002359 cfg.fc_dst = *prefix;
2360 cfg.fc_gateway = *gwaddr;
Thomas Graf86872cb2006-08-22 00:01:08 -07002361
YOSHIFUJI Hideakie317da92006-03-20 17:06:42 -08002362 /* We should treat it as a default route if prefix length is 0. */
2363 if (!prefixlen)
Thomas Graf86872cb2006-08-22 00:01:08 -07002364 cfg.fc_flags |= RTF_DEFAULT;
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002365
Thomas Graf86872cb2006-08-22 00:01:08 -07002366 ip6_route_add(&cfg);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002367
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -08002368 return rt6_get_route_info(net, prefix, prefixlen, gwaddr, ifindex);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002369}
2370#endif
2371
Eric Dumazetb71d1d42011-04-22 04:53:02 +00002372struct rt6_info *rt6_get_dflt_router(const struct in6_addr *addr, struct net_device *dev)
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09002373{
Linus Torvalds1da177e2005-04-16 15:20:36 -07002374 struct rt6_info *rt;
Thomas Grafc71099a2006-08-04 23:20:06 -07002375 struct fib6_table *table;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002376
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09002377 table = fib6_get_table(dev_net(dev), RT6_TABLE_DFLT);
David S. Miller38308472011-12-03 18:02:47 -05002378 if (!table)
Thomas Grafc71099a2006-08-04 23:20:06 -07002379 return NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002380
Li RongQing5744dd92012-09-11 21:59:01 +00002381 read_lock_bh(&table->tb6_lock);
Ian Morris67ba4152014-08-24 21:53:10 +01002382 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
David S. Millerd1918542011-12-28 20:19:20 -05002383 if (dev == rt->dst.dev &&
YOSHIFUJI Hideaki045927f2006-03-20 17:00:48 -08002384 ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) &&
Linus Torvalds1da177e2005-04-16 15:20:36 -07002385 ipv6_addr_equal(&rt->rt6i_gateway, addr))
2386 break;
2387 }
2388 if (rt)
Changli Gaod8d1f302010-06-10 23:31:35 -07002389 dst_hold(&rt->dst);
Li RongQing5744dd92012-09-11 21:59:01 +00002390 read_unlock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002391 return rt;
2392}
2393
Eric Dumazetb71d1d42011-04-22 04:53:02 +00002394struct rt6_info *rt6_add_dflt_router(const struct in6_addr *gwaddr,
YOSHIFUJI Hideakiebacaaa2006-03-20 17:04:53 -08002395 struct net_device *dev,
2396 unsigned int pref)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002397{
Thomas Graf86872cb2006-08-22 00:01:08 -07002398 struct fib6_config cfg = {
David Ahernca254492015-10-12 11:47:10 -07002399 .fc_table = l3mdev_fib_table(dev) ? : RT6_TABLE_DFLT,
Rami Rosen238fc7e2008-02-09 23:43:11 -08002400 .fc_metric = IP6_RT_PRIO_USER,
Thomas Graf86872cb2006-08-22 00:01:08 -07002401 .fc_ifindex = dev->ifindex,
2402 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT |
2403 RTF_UP | RTF_EXPIRES | RTF_PREF(pref),
Eric W. Biederman15e47302012-09-07 20:12:54 +00002404 .fc_nlinfo.portid = 0,
Daniel Lezcano55786892008-03-04 13:47:47 -08002405 .fc_nlinfo.nlh = NULL,
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09002406 .fc_nlinfo.nl_net = dev_net(dev),
Thomas Graf86872cb2006-08-22 00:01:08 -07002407 };
Linus Torvalds1da177e2005-04-16 15:20:36 -07002408
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00002409 cfg.fc_gateway = *gwaddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002410
Thomas Graf86872cb2006-08-22 00:01:08 -07002411 ip6_route_add(&cfg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002412
Linus Torvalds1da177e2005-04-16 15:20:36 -07002413 return rt6_get_dflt_router(gwaddr, dev);
2414}
2415
Daniel Lezcano7b4da532008-03-04 13:47:14 -08002416void rt6_purge_dflt_routers(struct net *net)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002417{
2418 struct rt6_info *rt;
Thomas Grafc71099a2006-08-04 23:20:06 -07002419 struct fib6_table *table;
2420
2421 /* NOTE: Keep consistent with rt6_get_dflt_router */
Daniel Lezcano7b4da532008-03-04 13:47:14 -08002422 table = fib6_get_table(net, RT6_TABLE_DFLT);
David S. Miller38308472011-12-03 18:02:47 -05002423 if (!table)
Thomas Grafc71099a2006-08-04 23:20:06 -07002424 return;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002425
2426restart:
Thomas Grafc71099a2006-08-04 23:20:06 -07002427 read_lock_bh(&table->tb6_lock);
Changli Gaod8d1f302010-06-10 23:31:35 -07002428 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
Lorenzo Colitti3e8b0ac2013-03-03 20:46:46 +00002429 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF) &&
2430 (!rt->rt6i_idev || rt->rt6i_idev->cnf.accept_ra != 2)) {
Changli Gaod8d1f302010-06-10 23:31:35 -07002431 dst_hold(&rt->dst);
Thomas Grafc71099a2006-08-04 23:20:06 -07002432 read_unlock_bh(&table->tb6_lock);
Thomas Grafe0a1ad732006-08-22 00:00:21 -07002433 ip6_del_rt(rt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002434 goto restart;
2435 }
2436 }
Thomas Grafc71099a2006-08-04 23:20:06 -07002437 read_unlock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002438}
2439
Daniel Lezcano55786892008-03-04 13:47:47 -08002440static void rtmsg_to_fib6_config(struct net *net,
2441 struct in6_rtmsg *rtmsg,
Thomas Graf86872cb2006-08-22 00:01:08 -07002442 struct fib6_config *cfg)
2443{
2444 memset(cfg, 0, sizeof(*cfg));
2445
David Ahernca254492015-10-12 11:47:10 -07002446 cfg->fc_table = l3mdev_fib_table_by_index(net, rtmsg->rtmsg_ifindex) ?
2447 : RT6_TABLE_MAIN;
Thomas Graf86872cb2006-08-22 00:01:08 -07002448 cfg->fc_ifindex = rtmsg->rtmsg_ifindex;
2449 cfg->fc_metric = rtmsg->rtmsg_metric;
2450 cfg->fc_expires = rtmsg->rtmsg_info;
2451 cfg->fc_dst_len = rtmsg->rtmsg_dst_len;
2452 cfg->fc_src_len = rtmsg->rtmsg_src_len;
2453 cfg->fc_flags = rtmsg->rtmsg_flags;
2454
Daniel Lezcano55786892008-03-04 13:47:47 -08002455 cfg->fc_nlinfo.nl_net = net;
Benjamin Theryf1243c22008-02-26 18:10:03 -08002456
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00002457 cfg->fc_dst = rtmsg->rtmsg_dst;
2458 cfg->fc_src = rtmsg->rtmsg_src;
2459 cfg->fc_gateway = rtmsg->rtmsg_gateway;
Thomas Graf86872cb2006-08-22 00:01:08 -07002460}
2461
Daniel Lezcano55786892008-03-04 13:47:47 -08002462int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002463{
Thomas Graf86872cb2006-08-22 00:01:08 -07002464 struct fib6_config cfg;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002465 struct in6_rtmsg rtmsg;
2466 int err;
2467
Ian Morris67ba4152014-08-24 21:53:10 +01002468 switch (cmd) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002469 case SIOCADDRT: /* Add a route */
2470 case SIOCDELRT: /* Delete a route */
Eric W. Biedermanaf31f412012-11-16 03:03:06 +00002471 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002472 return -EPERM;
2473 err = copy_from_user(&rtmsg, arg,
2474 sizeof(struct in6_rtmsg));
2475 if (err)
2476 return -EFAULT;
Thomas Graf86872cb2006-08-22 00:01:08 -07002477
Daniel Lezcano55786892008-03-04 13:47:47 -08002478 rtmsg_to_fib6_config(net, &rtmsg, &cfg);
Thomas Graf86872cb2006-08-22 00:01:08 -07002479
Linus Torvalds1da177e2005-04-16 15:20:36 -07002480 rtnl_lock();
2481 switch (cmd) {
2482 case SIOCADDRT:
Thomas Graf86872cb2006-08-22 00:01:08 -07002483 err = ip6_route_add(&cfg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002484 break;
2485 case SIOCDELRT:
Thomas Graf86872cb2006-08-22 00:01:08 -07002486 err = ip6_route_del(&cfg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002487 break;
2488 default:
2489 err = -EINVAL;
2490 }
2491 rtnl_unlock();
2492
2493 return err;
Stephen Hemminger3ff50b72007-04-20 17:09:22 -07002494 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07002495
2496 return -EINVAL;
2497}
2498
2499/*
2500 * Drop the packet on the floor
2501 */
2502
Brian Haleyd5fdd6b2009-06-23 04:31:07 -07002503static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002504{
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07002505 int type;
Eric Dumazetadf30902009-06-02 05:19:30 +00002506 struct dst_entry *dst = skb_dst(skb);
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07002507 switch (ipstats_mib_noroutes) {
2508 case IPSTATS_MIB_INNOROUTES:
Arnaldo Carvalho de Melo0660e032007-04-25 17:54:47 -07002509 type = ipv6_addr_type(&ipv6_hdr(skb)->daddr);
Ulrich Weber45bb0062010-02-25 23:28:58 +00002510 if (type == IPV6_ADDR_ANY) {
Denis V. Lunev3bd653c2008-10-08 10:54:51 -07002511 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2512 IPSTATS_MIB_INADDRERRORS);
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07002513 break;
2514 }
2515 /* FALLTHROUGH */
2516 case IPSTATS_MIB_OUTNOROUTES:
Denis V. Lunev3bd653c2008-10-08 10:54:51 -07002517 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2518 ipstats_mib_noroutes);
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07002519 break;
2520 }
Alexey Dobriyan3ffe5332010-02-18 08:25:24 +00002521 icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002522 kfree_skb(skb);
2523 return 0;
2524}
2525
Thomas Graf9ce8ade2006-10-18 20:46:54 -07002526static int ip6_pkt_discard(struct sk_buff *skb)
2527{
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07002528 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES);
Thomas Graf9ce8ade2006-10-18 20:46:54 -07002529}
2530
Eric W. Biedermanede20592015-10-07 16:48:47 -05002531static int ip6_pkt_discard_out(struct net *net, struct sock *sk, struct sk_buff *skb)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002532{
Eric Dumazetadf30902009-06-02 05:19:30 +00002533 skb->dev = skb_dst(skb)->dev;
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07002534 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002535}
2536
Thomas Graf9ce8ade2006-10-18 20:46:54 -07002537static int ip6_pkt_prohibit(struct sk_buff *skb)
2538{
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07002539 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES);
Thomas Graf9ce8ade2006-10-18 20:46:54 -07002540}
2541
Eric W. Biedermanede20592015-10-07 16:48:47 -05002542static int ip6_pkt_prohibit_out(struct net *net, struct sock *sk, struct sk_buff *skb)
Thomas Graf9ce8ade2006-10-18 20:46:54 -07002543{
Eric Dumazetadf30902009-06-02 05:19:30 +00002544 skb->dev = skb_dst(skb)->dev;
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07002545 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES);
Thomas Graf9ce8ade2006-10-18 20:46:54 -07002546}
2547
Linus Torvalds1da177e2005-04-16 15:20:36 -07002548/*
2549 * Allocate a dst for local (unicast / anycast) address.
2550 */
2551
2552struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
2553 const struct in6_addr *addr,
David S. Miller8f031512011-12-06 16:48:14 -05002554 bool anycast)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002555{
David Ahernca254492015-10-12 11:47:10 -07002556 u32 tb_id;
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09002557 struct net *net = dev_net(idev->dev);
Hannes Frederic Sowaa3300ef2013-12-07 03:33:45 +01002558 struct rt6_info *rt = ip6_dst_alloc(net, net->loopback_dev,
Martin KaFai Lauad706862015-08-14 11:05:52 -07002559 DST_NOCOUNT);
Hannes Frederic Sowaa3300ef2013-12-07 03:33:45 +01002560 if (!rt)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002561 return ERR_PTR(-ENOMEM);
2562
Linus Torvalds1da177e2005-04-16 15:20:36 -07002563 in6_dev_hold(idev);
2564
David S. Miller11d53b42011-06-24 15:23:34 -07002565 rt->dst.flags |= DST_HOST;
Changli Gaod8d1f302010-06-10 23:31:35 -07002566 rt->dst.input = ip6_input;
2567 rt->dst.output = ip6_output;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002568 rt->rt6i_idev = idev;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002569
2570 rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP;
YOSHIFUJI Hideaki58c4fb82005-12-21 22:56:42 +09002571 if (anycast)
2572 rt->rt6i_flags |= RTF_ANYCAST;
2573 else
Linus Torvalds1da177e2005-04-16 15:20:36 -07002574 rt->rt6i_flags |= RTF_LOCAL;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002575
Julian Anastasov550bab42013-10-20 15:43:04 +03002576 rt->rt6i_gateway = *addr;
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00002577 rt->rt6i_dst.addr = *addr;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002578 rt->rt6i_dst.plen = 128;
David Ahernca254492015-10-12 11:47:10 -07002579 tb_id = l3mdev_fib_table(idev->dev) ? : RT6_TABLE_LOCAL;
2580 rt->rt6i_table = fib6_get_table(net, tb_id);
Martin KaFai Lau8e3d5be2015-09-15 14:30:08 -07002581 rt->dst.flags |= DST_NOCACHE;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002582
Changli Gaod8d1f302010-06-10 23:31:35 -07002583 atomic_set(&rt->dst.__refcnt, 1);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002584
2585 return rt;
2586}
2587
Daniel Walterc3968a82011-04-13 21:10:57 +00002588int ip6_route_get_saddr(struct net *net,
2589 struct rt6_info *rt,
Eric Dumazetb71d1d42011-04-22 04:53:02 +00002590 const struct in6_addr *daddr,
Daniel Walterc3968a82011-04-13 21:10:57 +00002591 unsigned int prefs,
2592 struct in6_addr *saddr)
2593{
Markus Stenberge16e8882015-05-05 13:36:59 +03002594 struct inet6_dev *idev =
2595 rt ? ip6_dst_idev((struct dst_entry *)rt) : NULL;
Daniel Walterc3968a82011-04-13 21:10:57 +00002596 int err = 0;
Markus Stenberge16e8882015-05-05 13:36:59 +03002597 if (rt && rt->rt6i_prefsrc.plen)
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00002598 *saddr = rt->rt6i_prefsrc.addr;
Daniel Walterc3968a82011-04-13 21:10:57 +00002599 else
2600 err = ipv6_dev_get_saddr(net, idev ? idev->dev : NULL,
2601 daddr, prefs, saddr);
2602 return err;
2603}
2604
2605/* remove deleted ip from prefsrc entries */
2606struct arg_dev_net_ip {
2607 struct net_device *dev;
2608 struct net *net;
2609 struct in6_addr *addr;
2610};
2611
2612static int fib6_remove_prefsrc(struct rt6_info *rt, void *arg)
2613{
2614 struct net_device *dev = ((struct arg_dev_net_ip *)arg)->dev;
2615 struct net *net = ((struct arg_dev_net_ip *)arg)->net;
2616 struct in6_addr *addr = ((struct arg_dev_net_ip *)arg)->addr;
2617
David S. Millerd1918542011-12-28 20:19:20 -05002618 if (((void *)rt->dst.dev == dev || !dev) &&
Daniel Walterc3968a82011-04-13 21:10:57 +00002619 rt != net->ipv6.ip6_null_entry &&
2620 ipv6_addr_equal(addr, &rt->rt6i_prefsrc.addr)) {
2621 /* remove prefsrc entry */
2622 rt->rt6i_prefsrc.plen = 0;
2623 }
2624 return 0;
2625}
2626
2627void rt6_remove_prefsrc(struct inet6_ifaddr *ifp)
2628{
2629 struct net *net = dev_net(ifp->idev->dev);
2630 struct arg_dev_net_ip adni = {
2631 .dev = ifp->idev->dev,
2632 .net = net,
2633 .addr = &ifp->addr,
2634 };
Li RongQing0c3584d2013-12-27 16:32:38 +08002635 fib6_clean_all(net, fib6_remove_prefsrc, &adni);
Daniel Walterc3968a82011-04-13 21:10:57 +00002636}
2637
Duan Jiongbe7a0102014-05-15 15:56:14 +08002638#define RTF_RA_ROUTER (RTF_ADDRCONF | RTF_DEFAULT | RTF_GATEWAY)
2639#define RTF_CACHE_GATEWAY (RTF_GATEWAY | RTF_CACHE)
2640
2641/* Remove routers and update dst entries when gateway turn into host. */
2642static int fib6_clean_tohost(struct rt6_info *rt, void *arg)
2643{
2644 struct in6_addr *gateway = (struct in6_addr *)arg;
2645
2646 if ((((rt->rt6i_flags & RTF_RA_ROUTER) == RTF_RA_ROUTER) ||
2647 ((rt->rt6i_flags & RTF_CACHE_GATEWAY) == RTF_CACHE_GATEWAY)) &&
2648 ipv6_addr_equal(gateway, &rt->rt6i_gateway)) {
2649 return -1;
2650 }
2651 return 0;
2652}
2653
2654void rt6_clean_tohost(struct net *net, struct in6_addr *gateway)
2655{
2656 fib6_clean_all(net, fib6_clean_tohost, gateway);
2657}
2658
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002659struct arg_dev_net {
2660 struct net_device *dev;
2661 struct net *net;
2662};
2663
Linus Torvalds1da177e2005-04-16 15:20:36 -07002664static int fib6_ifdown(struct rt6_info *rt, void *arg)
2665{
stephen hemmingerbc3ef662010-12-16 17:42:40 +00002666 const struct arg_dev_net *adn = arg;
2667 const struct net_device *dev = adn->dev;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002668
David S. Millerd1918542011-12-28 20:19:20 -05002669 if ((rt->dst.dev == dev || !dev) &&
David S. Millerc159d302011-12-26 15:24:36 -05002670 rt != adn->net->ipv6.ip6_null_entry)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002671 return -1;
David S. Millerc159d302011-12-26 15:24:36 -05002672
Linus Torvalds1da177e2005-04-16 15:20:36 -07002673 return 0;
2674}
2675
Daniel Lezcanof3db4852008-03-03 23:27:06 -08002676void rt6_ifdown(struct net *net, struct net_device *dev)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002677{
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002678 struct arg_dev_net adn = {
2679 .dev = dev,
2680 .net = net,
2681 };
2682
Li RongQing0c3584d2013-12-27 16:32:38 +08002683 fib6_clean_all(net, fib6_ifdown, &adn);
David S. Miller1e493d12008-09-10 17:27:15 -07002684 icmp6_clean_all(fib6_ifdown, &adn);
Eric W. Biedermane332bc62015-10-12 11:02:08 -05002685 if (dev)
2686 rt6_uncached_list_flush_dev(net, dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002687}
2688
Eric Dumazet95c96172012-04-15 05:58:06 +00002689struct rt6_mtu_change_arg {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002690 struct net_device *dev;
Eric Dumazet95c96172012-04-15 05:58:06 +00002691 unsigned int mtu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002692};
2693
2694static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg)
2695{
2696 struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg;
2697 struct inet6_dev *idev;
2698
2699 /* In IPv6 pmtu discovery is not optional,
2700 so that RTAX_MTU lock cannot disable it.
2701 We still use this lock to block changes
2702 caused by addrconf/ndisc.
2703 */
2704
2705 idev = __in6_dev_get(arg->dev);
David S. Miller38308472011-12-03 18:02:47 -05002706 if (!idev)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002707 return 0;
2708
2709 /* For administrative MTU increase, there is no way to discover
2710 IPv6 PMTU increase, so PMTU increase should be updated here.
2711 Since RFC 1981 doesn't include administrative MTU increase
2712 update PMTU increase is a MUST. (i.e. jumbo frame)
2713 */
2714 /*
2715 If new MTU is less than route PMTU, this new MTU will be the
2716 lowest MTU in the path, update the route PMTU to reflect PMTU
2717 decreases; if new MTU is greater than route PMTU, and the
2718 old MTU is the lowest MTU in the path, update the route PMTU
2719 to reflect the increase. In this case if the other nodes' MTU
2720 also have the lowest MTU, TOO BIG MESSAGE will be lead to
2721 PMTU discouvery.
2722 */
David S. Millerd1918542011-12-28 20:19:20 -05002723 if (rt->dst.dev == arg->dev &&
Martin KaFai Lau4b32b5a2015-04-28 13:03:06 -07002724 !dst_metric_locked(&rt->dst, RTAX_MTU)) {
2725 if (rt->rt6i_flags & RTF_CACHE) {
2726 /* For RTF_CACHE with rt6i_pmtu == 0
2727 * (i.e. a redirected route),
2728 * the metrics of its rt->dst.from has already
2729 * been updated.
2730 */
2731 if (rt->rt6i_pmtu && rt->rt6i_pmtu > arg->mtu)
2732 rt->rt6i_pmtu = arg->mtu;
2733 } else if (dst_mtu(&rt->dst) >= arg->mtu ||
2734 (dst_mtu(&rt->dst) < arg->mtu &&
2735 dst_mtu(&rt->dst) == idev->cnf.mtu6)) {
2736 dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu);
2737 }
Simon Arlott566cfd82007-07-26 00:09:55 -07002738 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07002739 return 0;
2740}
2741
Eric Dumazet95c96172012-04-15 05:58:06 +00002742void rt6_mtu_change(struct net_device *dev, unsigned int mtu)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002743{
Thomas Grafc71099a2006-08-04 23:20:06 -07002744 struct rt6_mtu_change_arg arg = {
2745 .dev = dev,
2746 .mtu = mtu,
2747 };
Linus Torvalds1da177e2005-04-16 15:20:36 -07002748
Li RongQing0c3584d2013-12-27 16:32:38 +08002749 fib6_clean_all(dev_net(dev), rt6_mtu_change_route, &arg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002750}
2751
Patrick McHardyef7c79e2007-06-05 12:38:30 -07002752static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = {
Thomas Graf5176f912006-08-26 20:13:18 -07002753 [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) },
Thomas Graf86872cb2006-08-22 00:01:08 -07002754 [RTA_OIF] = { .type = NLA_U32 },
Thomas Grafab364a62006-08-22 00:01:47 -07002755 [RTA_IIF] = { .type = NLA_U32 },
Thomas Graf86872cb2006-08-22 00:01:08 -07002756 [RTA_PRIORITY] = { .type = NLA_U32 },
2757 [RTA_METRICS] = { .type = NLA_NESTED },
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002758 [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) },
Lubomir Rintelc78ba6d2015-03-11 15:39:21 +01002759 [RTA_PREF] = { .type = NLA_U8 },
Roopa Prabhu19e42e42015-07-21 10:43:48 +02002760 [RTA_ENCAP_TYPE] = { .type = NLA_U16 },
2761 [RTA_ENCAP] = { .type = NLA_NESTED },
Xin Long32bc2012015-12-16 17:50:11 +08002762 [RTA_EXPIRES] = { .type = NLA_U32 },
Thomas Graf86872cb2006-08-22 00:01:08 -07002763};
2764
2765static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh,
2766 struct fib6_config *cfg)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002767{
Thomas Graf86872cb2006-08-22 00:01:08 -07002768 struct rtmsg *rtm;
2769 struct nlattr *tb[RTA_MAX+1];
Lubomir Rintelc78ba6d2015-03-11 15:39:21 +01002770 unsigned int pref;
Thomas Graf86872cb2006-08-22 00:01:08 -07002771 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002772
Thomas Graf86872cb2006-08-22 00:01:08 -07002773 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2774 if (err < 0)
2775 goto errout;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002776
Thomas Graf86872cb2006-08-22 00:01:08 -07002777 err = -EINVAL;
2778 rtm = nlmsg_data(nlh);
2779 memset(cfg, 0, sizeof(*cfg));
2780
2781 cfg->fc_table = rtm->rtm_table;
2782 cfg->fc_dst_len = rtm->rtm_dst_len;
2783 cfg->fc_src_len = rtm->rtm_src_len;
2784 cfg->fc_flags = RTF_UP;
2785 cfg->fc_protocol = rtm->rtm_protocol;
Nicolas Dichtelef2c7d72012-09-05 02:12:42 +00002786 cfg->fc_type = rtm->rtm_type;
Thomas Graf86872cb2006-08-22 00:01:08 -07002787
Nicolas Dichtelef2c7d72012-09-05 02:12:42 +00002788 if (rtm->rtm_type == RTN_UNREACHABLE ||
2789 rtm->rtm_type == RTN_BLACKHOLE ||
Nicolas Dichtelb4949ab2012-09-06 05:53:35 +00002790 rtm->rtm_type == RTN_PROHIBIT ||
2791 rtm->rtm_type == RTN_THROW)
Thomas Graf86872cb2006-08-22 00:01:08 -07002792 cfg->fc_flags |= RTF_REJECT;
2793
Maciej Żenczykowskiab79ad12010-09-27 00:07:02 +00002794 if (rtm->rtm_type == RTN_LOCAL)
2795 cfg->fc_flags |= RTF_LOCAL;
2796
Martin KaFai Lau1f56a012015-04-28 13:03:03 -07002797 if (rtm->rtm_flags & RTM_F_CLONED)
2798 cfg->fc_flags |= RTF_CACHE;
2799
Eric W. Biederman15e47302012-09-07 20:12:54 +00002800 cfg->fc_nlinfo.portid = NETLINK_CB(skb).portid;
Thomas Graf86872cb2006-08-22 00:01:08 -07002801 cfg->fc_nlinfo.nlh = nlh;
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +09002802 cfg->fc_nlinfo.nl_net = sock_net(skb->sk);
Thomas Graf86872cb2006-08-22 00:01:08 -07002803
2804 if (tb[RTA_GATEWAY]) {
Jiri Benc67b61f62015-03-29 16:59:26 +02002805 cfg->fc_gateway = nla_get_in6_addr(tb[RTA_GATEWAY]);
Thomas Graf86872cb2006-08-22 00:01:08 -07002806 cfg->fc_flags |= RTF_GATEWAY;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002807 }
Thomas Graf86872cb2006-08-22 00:01:08 -07002808
2809 if (tb[RTA_DST]) {
2810 int plen = (rtm->rtm_dst_len + 7) >> 3;
2811
2812 if (nla_len(tb[RTA_DST]) < plen)
2813 goto errout;
2814
2815 nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002816 }
Thomas Graf86872cb2006-08-22 00:01:08 -07002817
2818 if (tb[RTA_SRC]) {
2819 int plen = (rtm->rtm_src_len + 7) >> 3;
2820
2821 if (nla_len(tb[RTA_SRC]) < plen)
2822 goto errout;
2823
2824 nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002825 }
Thomas Graf86872cb2006-08-22 00:01:08 -07002826
Daniel Walterc3968a82011-04-13 21:10:57 +00002827 if (tb[RTA_PREFSRC])
Jiri Benc67b61f62015-03-29 16:59:26 +02002828 cfg->fc_prefsrc = nla_get_in6_addr(tb[RTA_PREFSRC]);
Daniel Walterc3968a82011-04-13 21:10:57 +00002829
Thomas Graf86872cb2006-08-22 00:01:08 -07002830 if (tb[RTA_OIF])
2831 cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]);
2832
2833 if (tb[RTA_PRIORITY])
2834 cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]);
2835
2836 if (tb[RTA_METRICS]) {
2837 cfg->fc_mx = nla_data(tb[RTA_METRICS]);
2838 cfg->fc_mx_len = nla_len(tb[RTA_METRICS]);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002839 }
Thomas Graf86872cb2006-08-22 00:01:08 -07002840
2841 if (tb[RTA_TABLE])
2842 cfg->fc_table = nla_get_u32(tb[RTA_TABLE]);
2843
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002844 if (tb[RTA_MULTIPATH]) {
2845 cfg->fc_mp = nla_data(tb[RTA_MULTIPATH]);
2846 cfg->fc_mp_len = nla_len(tb[RTA_MULTIPATH]);
2847 }
2848
Lubomir Rintelc78ba6d2015-03-11 15:39:21 +01002849 if (tb[RTA_PREF]) {
2850 pref = nla_get_u8(tb[RTA_PREF]);
2851 if (pref != ICMPV6_ROUTER_PREF_LOW &&
2852 pref != ICMPV6_ROUTER_PREF_HIGH)
2853 pref = ICMPV6_ROUTER_PREF_MEDIUM;
2854 cfg->fc_flags |= RTF_PREF(pref);
2855 }
2856
Roopa Prabhu19e42e42015-07-21 10:43:48 +02002857 if (tb[RTA_ENCAP])
2858 cfg->fc_encap = tb[RTA_ENCAP];
2859
2860 if (tb[RTA_ENCAP_TYPE])
2861 cfg->fc_encap_type = nla_get_u16(tb[RTA_ENCAP_TYPE]);
2862
Xin Long32bc2012015-12-16 17:50:11 +08002863 if (tb[RTA_EXPIRES]) {
2864 unsigned long timeout = addrconf_timeout_fixup(nla_get_u32(tb[RTA_EXPIRES]), HZ);
2865
2866 if (addrconf_finite_timeout(timeout)) {
2867 cfg->fc_expires = jiffies_to_clock_t(timeout * HZ);
2868 cfg->fc_flags |= RTF_EXPIRES;
2869 }
2870 }
2871
Thomas Graf86872cb2006-08-22 00:01:08 -07002872 err = 0;
2873errout:
2874 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002875}
2876
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002877struct rt6_nh {
2878 struct rt6_info *rt6_info;
2879 struct fib6_config r_cfg;
2880 struct mx6_config mxc;
2881 struct list_head next;
2882};
2883
2884static void ip6_print_replace_route_err(struct list_head *rt6_nh_list)
2885{
2886 struct rt6_nh *nh;
2887
2888 list_for_each_entry(nh, rt6_nh_list, next) {
2889 pr_warn("IPV6: multipath route replace failed (check consistency of installed routes): %pI6 nexthop %pI6 ifi %d\n",
2890 &nh->r_cfg.fc_dst, &nh->r_cfg.fc_gateway,
2891 nh->r_cfg.fc_ifindex);
2892 }
2893}
2894
2895static int ip6_route_info_append(struct list_head *rt6_nh_list,
2896 struct rt6_info *rt, struct fib6_config *r_cfg)
2897{
2898 struct rt6_nh *nh;
2899 struct rt6_info *rtnh;
2900 int err = -EEXIST;
2901
2902 list_for_each_entry(nh, rt6_nh_list, next) {
2903 /* check if rt6_info already exists */
2904 rtnh = nh->rt6_info;
2905
2906 if (rtnh->dst.dev == rt->dst.dev &&
2907 rtnh->rt6i_idev == rt->rt6i_idev &&
2908 ipv6_addr_equal(&rtnh->rt6i_gateway,
2909 &rt->rt6i_gateway))
2910 return err;
2911 }
2912
2913 nh = kzalloc(sizeof(*nh), GFP_KERNEL);
2914 if (!nh)
2915 return -ENOMEM;
2916 nh->rt6_info = rt;
2917 err = ip6_convert_metrics(&nh->mxc, r_cfg);
2918 if (err) {
2919 kfree(nh);
2920 return err;
2921 }
2922 memcpy(&nh->r_cfg, r_cfg, sizeof(*r_cfg));
2923 list_add_tail(&nh->next, rt6_nh_list);
2924
2925 return 0;
2926}
2927
2928static int ip6_route_multipath_add(struct fib6_config *cfg)
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002929{
2930 struct fib6_config r_cfg;
2931 struct rtnexthop *rtnh;
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002932 struct rt6_info *rt;
2933 struct rt6_nh *err_nh;
2934 struct rt6_nh *nh, *nh_safe;
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002935 int remaining;
2936 int attrlen;
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002937 int err = 1;
2938 int nhn = 0;
2939 int replace = (cfg->fc_nlinfo.nlh &&
2940 (cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_REPLACE));
2941 LIST_HEAD(rt6_nh_list);
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002942
Michal Kubeček35f1b4e2015-05-18 20:53:55 +02002943 remaining = cfg->fc_mp_len;
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002944 rtnh = (struct rtnexthop *)cfg->fc_mp;
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002945
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002946 /* Parse a Multipath Entry and build a list (rt6_nh_list) of
2947 * rt6_info structs per nexthop
2948 */
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002949 while (rtnh_ok(rtnh, remaining)) {
2950 memcpy(&r_cfg, cfg, sizeof(*cfg));
2951 if (rtnh->rtnh_ifindex)
2952 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
2953
2954 attrlen = rtnh_attrlen(rtnh);
2955 if (attrlen > 0) {
2956 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
2957
2958 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
2959 if (nla) {
Jiri Benc67b61f62015-03-29 16:59:26 +02002960 r_cfg.fc_gateway = nla_get_in6_addr(nla);
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002961 r_cfg.fc_flags |= RTF_GATEWAY;
2962 }
Roopa Prabhu19e42e42015-07-21 10:43:48 +02002963 r_cfg.fc_encap = nla_find(attrs, attrlen, RTA_ENCAP);
2964 nla = nla_find(attrs, attrlen, RTA_ENCAP_TYPE);
2965 if (nla)
2966 r_cfg.fc_encap_type = nla_get_u16(nla);
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002967 }
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002968
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07002969 rt = ip6_route_info_create(&r_cfg);
2970 if (IS_ERR(rt)) {
2971 err = PTR_ERR(rt);
2972 rt = NULL;
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002973 goto cleanup;
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07002974 }
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002975
2976 err = ip6_route_info_append(&rt6_nh_list, rt, &r_cfg);
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002977 if (err) {
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002978 dst_free(&rt->dst);
2979 goto cleanup;
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002980 }
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002981
2982 rtnh = rtnh_next(rtnh, &remaining);
2983 }
2984
2985 err_nh = NULL;
2986 list_for_each_entry(nh, &rt6_nh_list, next) {
2987 err = __ip6_ins_rt(nh->rt6_info, &cfg->fc_nlinfo, &nh->mxc);
2988 /* nh->rt6_info is used or freed at this point, reset to NULL*/
2989 nh->rt6_info = NULL;
2990 if (err) {
2991 if (replace && nhn)
2992 ip6_print_replace_route_err(&rt6_nh_list);
2993 err_nh = nh;
2994 goto add_errout;
2995 }
2996
Nicolas Dichtel1a724182012-11-01 22:58:22 +00002997 /* Because each route is added like a single route we remove
Michal Kubeček27596472015-05-18 20:54:00 +02002998 * these flags after the first nexthop: if there is a collision,
2999 * we have already failed to add the first nexthop:
3000 * fib6_add_rt2node() has rejected it; when replacing, old
3001 * nexthops have been replaced by first new, the rest should
3002 * be added to it.
Nicolas Dichtel1a724182012-11-01 22:58:22 +00003003 */
Michal Kubeček27596472015-05-18 20:54:00 +02003004 cfg->fc_nlinfo.nlh->nlmsg_flags &= ~(NLM_F_EXCL |
3005 NLM_F_REPLACE);
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07003006 nhn++;
3007 }
3008
3009 goto cleanup;
3010
3011add_errout:
3012 /* Delete routes that were already added */
3013 list_for_each_entry(nh, &rt6_nh_list, next) {
3014 if (err_nh == nh)
3015 break;
3016 ip6_route_del(&nh->r_cfg);
3017 }
3018
3019cleanup:
3020 list_for_each_entry_safe(nh, nh_safe, &rt6_nh_list, next) {
3021 if (nh->rt6_info)
3022 dst_free(&nh->rt6_info->dst);
Wu Fengguang52fe51f2015-09-10 06:57:12 +08003023 kfree(nh->mxc.mx);
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07003024 list_del(&nh->next);
3025 kfree(nh);
3026 }
3027
3028 return err;
3029}
3030
3031static int ip6_route_multipath_del(struct fib6_config *cfg)
3032{
3033 struct fib6_config r_cfg;
3034 struct rtnexthop *rtnh;
3035 int remaining;
3036 int attrlen;
3037 int err = 1, last_err = 0;
3038
3039 remaining = cfg->fc_mp_len;
3040 rtnh = (struct rtnexthop *)cfg->fc_mp;
3041
3042 /* Parse a Multipath Entry */
3043 while (rtnh_ok(rtnh, remaining)) {
3044 memcpy(&r_cfg, cfg, sizeof(*cfg));
3045 if (rtnh->rtnh_ifindex)
3046 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
3047
3048 attrlen = rtnh_attrlen(rtnh);
3049 if (attrlen > 0) {
3050 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
3051
3052 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
3053 if (nla) {
3054 nla_memcpy(&r_cfg.fc_gateway, nla, 16);
3055 r_cfg.fc_flags |= RTF_GATEWAY;
3056 }
3057 }
3058 err = ip6_route_del(&r_cfg);
3059 if (err)
3060 last_err = err;
3061
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00003062 rtnh = rtnh_next(rtnh, &remaining);
3063 }
3064
3065 return last_err;
3066}
3067
Ian Morris67ba4152014-08-24 21:53:10 +01003068static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr *nlh)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003069{
Thomas Graf86872cb2006-08-22 00:01:08 -07003070 struct fib6_config cfg;
3071 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003072
Thomas Graf86872cb2006-08-22 00:01:08 -07003073 err = rtm_to_fib6_config(skb, nlh, &cfg);
3074 if (err < 0)
3075 return err;
3076
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00003077 if (cfg.fc_mp)
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07003078 return ip6_route_multipath_del(&cfg);
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00003079 else
3080 return ip6_route_del(&cfg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003081}
3082
Ian Morris67ba4152014-08-24 21:53:10 +01003083static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr *nlh)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003084{
Thomas Graf86872cb2006-08-22 00:01:08 -07003085 struct fib6_config cfg;
3086 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003087
Thomas Graf86872cb2006-08-22 00:01:08 -07003088 err = rtm_to_fib6_config(skb, nlh, &cfg);
3089 if (err < 0)
3090 return err;
3091
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00003092 if (cfg.fc_mp)
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07003093 return ip6_route_multipath_add(&cfg);
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00003094 else
3095 return ip6_route_add(&cfg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003096}
3097
Roopa Prabhu19e42e42015-07-21 10:43:48 +02003098static inline size_t rt6_nlmsg_size(struct rt6_info *rt)
Thomas Graf339bf982006-11-10 14:10:15 -08003099{
3100 return NLMSG_ALIGN(sizeof(struct rtmsg))
3101 + nla_total_size(16) /* RTA_SRC */
3102 + nla_total_size(16) /* RTA_DST */
3103 + nla_total_size(16) /* RTA_GATEWAY */
3104 + nla_total_size(16) /* RTA_PREFSRC */
3105 + nla_total_size(4) /* RTA_TABLE */
3106 + nla_total_size(4) /* RTA_IIF */
3107 + nla_total_size(4) /* RTA_OIF */
3108 + nla_total_size(4) /* RTA_PRIORITY */
Noriaki TAKAMIYA6a2b9ce2007-01-23 22:09:41 -08003109 + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */
Daniel Borkmannea697632015-01-05 23:57:47 +01003110 + nla_total_size(sizeof(struct rta_cacheinfo))
Lubomir Rintelc78ba6d2015-03-11 15:39:21 +01003111 + nla_total_size(TCP_CA_NAME_MAX) /* RTAX_CC_ALGO */
Roopa Prabhu19e42e42015-07-21 10:43:48 +02003112 + nla_total_size(1) /* RTA_PREF */
Jiri Benc61adedf2015-08-20 13:56:25 +02003113 + lwtunnel_get_encap_size(rt->dst.lwtstate);
Thomas Graf339bf982006-11-10 14:10:15 -08003114}
3115
Brian Haley191cd582008-08-14 15:33:21 -07003116static int rt6_fill_node(struct net *net,
3117 struct sk_buff *skb, struct rt6_info *rt,
Jamal Hadi Salim0d51aa82005-06-21 13:51:04 -07003118 struct in6_addr *dst, struct in6_addr *src,
Eric W. Biederman15e47302012-09-07 20:12:54 +00003119 int iif, int type, u32 portid, u32 seq,
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09003120 int prefix, int nowait, unsigned int flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003121{
Martin KaFai Lau4b32b5a2015-04-28 13:03:06 -07003122 u32 metrics[RTAX_MAX];
Linus Torvalds1da177e2005-04-16 15:20:36 -07003123 struct rtmsg *rtm;
Thomas Graf2d7202b2006-08-22 00:01:27 -07003124 struct nlmsghdr *nlh;
Thomas Grafe3703b32006-11-27 09:27:07 -08003125 long expires;
Patrick McHardy9e762a42006-08-10 23:09:48 -07003126 u32 table;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003127
3128 if (prefix) { /* user wants prefix routes only */
3129 if (!(rt->rt6i_flags & RTF_PREFIX_RT)) {
3130 /* success since this is not a prefix route */
3131 return 1;
3132 }
3133 }
3134
Eric W. Biederman15e47302012-09-07 20:12:54 +00003135 nlh = nlmsg_put(skb, portid, seq, type, sizeof(*rtm), flags);
David S. Miller38308472011-12-03 18:02:47 -05003136 if (!nlh)
Patrick McHardy26932562007-01-31 23:16:40 -08003137 return -EMSGSIZE;
Thomas Graf2d7202b2006-08-22 00:01:27 -07003138
3139 rtm = nlmsg_data(nlh);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003140 rtm->rtm_family = AF_INET6;
3141 rtm->rtm_dst_len = rt->rt6i_dst.plen;
3142 rtm->rtm_src_len = rt->rt6i_src.plen;
3143 rtm->rtm_tos = 0;
Thomas Grafc71099a2006-08-04 23:20:06 -07003144 if (rt->rt6i_table)
Patrick McHardy9e762a42006-08-10 23:09:48 -07003145 table = rt->rt6i_table->tb6_id;
Thomas Grafc71099a2006-08-04 23:20:06 -07003146 else
Patrick McHardy9e762a42006-08-10 23:09:48 -07003147 table = RT6_TABLE_UNSPEC;
3148 rtm->rtm_table = table;
David S. Millerc78679e2012-04-01 20:27:33 -04003149 if (nla_put_u32(skb, RTA_TABLE, table))
3150 goto nla_put_failure;
Nicolas Dichtelef2c7d72012-09-05 02:12:42 +00003151 if (rt->rt6i_flags & RTF_REJECT) {
3152 switch (rt->dst.error) {
3153 case -EINVAL:
3154 rtm->rtm_type = RTN_BLACKHOLE;
3155 break;
3156 case -EACCES:
3157 rtm->rtm_type = RTN_PROHIBIT;
3158 break;
Nicolas Dichtelb4949ab2012-09-06 05:53:35 +00003159 case -EAGAIN:
3160 rtm->rtm_type = RTN_THROW;
3161 break;
Nicolas Dichtelef2c7d72012-09-05 02:12:42 +00003162 default:
3163 rtm->rtm_type = RTN_UNREACHABLE;
3164 break;
3165 }
3166 }
David S. Miller38308472011-12-03 18:02:47 -05003167 else if (rt->rt6i_flags & RTF_LOCAL)
Maciej Żenczykowskiab79ad12010-09-27 00:07:02 +00003168 rtm->rtm_type = RTN_LOCAL;
David S. Millerd1918542011-12-28 20:19:20 -05003169 else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK))
Linus Torvalds1da177e2005-04-16 15:20:36 -07003170 rtm->rtm_type = RTN_LOCAL;
3171 else
3172 rtm->rtm_type = RTN_UNICAST;
3173 rtm->rtm_flags = 0;
Andy Gospodarek35103d12015-08-13 10:39:01 -04003174 if (!netif_carrier_ok(rt->dst.dev)) {
Andy Gospodarekcea45e22015-08-13 10:39:00 -04003175 rtm->rtm_flags |= RTNH_F_LINKDOWN;
Andy Gospodarek35103d12015-08-13 10:39:01 -04003176 if (rt->rt6i_idev->cnf.ignore_routes_with_linkdown)
3177 rtm->rtm_flags |= RTNH_F_DEAD;
3178 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07003179 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
3180 rtm->rtm_protocol = rt->rt6i_protocol;
David S. Miller38308472011-12-03 18:02:47 -05003181 if (rt->rt6i_flags & RTF_DYNAMIC)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003182 rtm->rtm_protocol = RTPROT_REDIRECT;
Denis Ovsienkof0396f602012-07-10 04:45:50 +00003183 else if (rt->rt6i_flags & RTF_ADDRCONF) {
3184 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ROUTEINFO))
3185 rtm->rtm_protocol = RTPROT_RA;
3186 else
3187 rtm->rtm_protocol = RTPROT_KERNEL;
3188 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07003189
David S. Miller38308472011-12-03 18:02:47 -05003190 if (rt->rt6i_flags & RTF_CACHE)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003191 rtm->rtm_flags |= RTM_F_CLONED;
3192
3193 if (dst) {
Jiri Benc930345e2015-03-29 16:59:25 +02003194 if (nla_put_in6_addr(skb, RTA_DST, dst))
David S. Millerc78679e2012-04-01 20:27:33 -04003195 goto nla_put_failure;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09003196 rtm->rtm_dst_len = 128;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003197 } else if (rtm->rtm_dst_len)
Jiri Benc930345e2015-03-29 16:59:25 +02003198 if (nla_put_in6_addr(skb, RTA_DST, &rt->rt6i_dst.addr))
David S. Millerc78679e2012-04-01 20:27:33 -04003199 goto nla_put_failure;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003200#ifdef CONFIG_IPV6_SUBTREES
3201 if (src) {
Jiri Benc930345e2015-03-29 16:59:25 +02003202 if (nla_put_in6_addr(skb, RTA_SRC, src))
David S. Millerc78679e2012-04-01 20:27:33 -04003203 goto nla_put_failure;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09003204 rtm->rtm_src_len = 128;
David S. Millerc78679e2012-04-01 20:27:33 -04003205 } else if (rtm->rtm_src_len &&
Jiri Benc930345e2015-03-29 16:59:25 +02003206 nla_put_in6_addr(skb, RTA_SRC, &rt->rt6i_src.addr))
David S. Millerc78679e2012-04-01 20:27:33 -04003207 goto nla_put_failure;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003208#endif
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09003209 if (iif) {
3210#ifdef CONFIG_IPV6_MROUTE
3211 if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) {
Benjamin Thery8229efd2008-12-10 16:30:15 -08003212 int err = ip6mr_get_route(net, skb, rtm, nowait);
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09003213 if (err <= 0) {
3214 if (!nowait) {
3215 if (err == 0)
3216 return 0;
3217 goto nla_put_failure;
3218 } else {
3219 if (err == -EMSGSIZE)
3220 goto nla_put_failure;
3221 }
3222 }
3223 } else
3224#endif
David S. Millerc78679e2012-04-01 20:27:33 -04003225 if (nla_put_u32(skb, RTA_IIF, iif))
3226 goto nla_put_failure;
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09003227 } else if (dst) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003228 struct in6_addr saddr_buf;
David S. Millerc78679e2012-04-01 20:27:33 -04003229 if (ip6_route_get_saddr(net, rt, dst, 0, &saddr_buf) == 0 &&
Jiri Benc930345e2015-03-29 16:59:25 +02003230 nla_put_in6_addr(skb, RTA_PREFSRC, &saddr_buf))
David S. Millerc78679e2012-04-01 20:27:33 -04003231 goto nla_put_failure;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003232 }
Thomas Graf2d7202b2006-08-22 00:01:27 -07003233
Daniel Walterc3968a82011-04-13 21:10:57 +00003234 if (rt->rt6i_prefsrc.plen) {
3235 struct in6_addr saddr_buf;
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00003236 saddr_buf = rt->rt6i_prefsrc.addr;
Jiri Benc930345e2015-03-29 16:59:25 +02003237 if (nla_put_in6_addr(skb, RTA_PREFSRC, &saddr_buf))
David S. Millerc78679e2012-04-01 20:27:33 -04003238 goto nla_put_failure;
Daniel Walterc3968a82011-04-13 21:10:57 +00003239 }
3240
Martin KaFai Lau4b32b5a2015-04-28 13:03:06 -07003241 memcpy(metrics, dst_metrics_ptr(&rt->dst), sizeof(metrics));
3242 if (rt->rt6i_pmtu)
3243 metrics[RTAX_MTU - 1] = rt->rt6i_pmtu;
3244 if (rtnetlink_put_metrics(skb, metrics) < 0)
Thomas Graf2d7202b2006-08-22 00:01:27 -07003245 goto nla_put_failure;
3246
YOSHIFUJI Hideaki / 吉藤英明dd0cbf22013-01-17 12:53:15 +00003247 if (rt->rt6i_flags & RTF_GATEWAY) {
Jiri Benc930345e2015-03-29 16:59:25 +02003248 if (nla_put_in6_addr(skb, RTA_GATEWAY, &rt->rt6i_gateway) < 0)
Eric Dumazet94f826b2012-03-27 09:53:52 +00003249 goto nla_put_failure;
Eric Dumazet94f826b2012-03-27 09:53:52 +00003250 }
Thomas Graf2d7202b2006-08-22 00:01:27 -07003251
David S. Millerc78679e2012-04-01 20:27:33 -04003252 if (rt->dst.dev &&
3253 nla_put_u32(skb, RTA_OIF, rt->dst.dev->ifindex))
3254 goto nla_put_failure;
3255 if (nla_put_u32(skb, RTA_PRIORITY, rt->rt6i_metric))
3256 goto nla_put_failure;
Li Wei82539472012-07-29 16:01:30 +00003257
3258 expires = (rt->rt6i_flags & RTF_EXPIRES) ? rt->dst.expires - jiffies : 0;
YOSHIFUJI Hideaki69cdf8f2008-05-19 16:55:13 -07003259
David S. Miller87a50692012-07-10 05:06:14 -07003260 if (rtnl_put_cacheinfo(skb, &rt->dst, 0, expires, rt->dst.error) < 0)
Thomas Grafe3703b32006-11-27 09:27:07 -08003261 goto nla_put_failure;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003262
Lubomir Rintelc78ba6d2015-03-11 15:39:21 +01003263 if (nla_put_u8(skb, RTA_PREF, IPV6_EXTRACT_PREF(rt->rt6i_flags)))
3264 goto nla_put_failure;
3265
Jiri Benc61adedf2015-08-20 13:56:25 +02003266 lwtunnel_fill_encap(skb, rt->dst.lwtstate);
Roopa Prabhu19e42e42015-07-21 10:43:48 +02003267
Johannes Berg053c0952015-01-16 22:09:00 +01003268 nlmsg_end(skb, nlh);
3269 return 0;
Thomas Graf2d7202b2006-08-22 00:01:27 -07003270
3271nla_put_failure:
Patrick McHardy26932562007-01-31 23:16:40 -08003272 nlmsg_cancel(skb, nlh);
3273 return -EMSGSIZE;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003274}
3275
Patrick McHardy1b43af52006-08-10 23:11:17 -07003276int rt6_dump_route(struct rt6_info *rt, void *p_arg)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003277{
3278 struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg;
3279 int prefix;
3280
Thomas Graf2d7202b2006-08-22 00:01:27 -07003281 if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) {
3282 struct rtmsg *rtm = nlmsg_data(arg->cb->nlh);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003283 prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0;
3284 } else
3285 prefix = 0;
3286
Brian Haley191cd582008-08-14 15:33:21 -07003287 return rt6_fill_node(arg->net,
3288 arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE,
Eric W. Biederman15e47302012-09-07 20:12:54 +00003289 NETLINK_CB(arg->cb->skb).portid, arg->cb->nlh->nlmsg_seq,
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09003290 prefix, 0, NLM_F_MULTI);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003291}
3292
Ian Morris67ba4152014-08-24 21:53:10 +01003293static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003294{
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +09003295 struct net *net = sock_net(in_skb->sk);
Thomas Grafab364a62006-08-22 00:01:47 -07003296 struct nlattr *tb[RTA_MAX+1];
Linus Torvalds1da177e2005-04-16 15:20:36 -07003297 struct rt6_info *rt;
Thomas Grafab364a62006-08-22 00:01:47 -07003298 struct sk_buff *skb;
3299 struct rtmsg *rtm;
David S. Miller4c9483b2011-03-12 16:22:43 -05003300 struct flowi6 fl6;
Shmulik Ladkani72331bc2012-04-01 04:03:45 +00003301 int err, iif = 0, oif = 0;
Thomas Grafab364a62006-08-22 00:01:47 -07003302
3303 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
3304 if (err < 0)
3305 goto errout;
3306
3307 err = -EINVAL;
David S. Miller4c9483b2011-03-12 16:22:43 -05003308 memset(&fl6, 0, sizeof(fl6));
Thomas Grafab364a62006-08-22 00:01:47 -07003309
3310 if (tb[RTA_SRC]) {
3311 if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr))
3312 goto errout;
3313
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00003314 fl6.saddr = *(struct in6_addr *)nla_data(tb[RTA_SRC]);
Thomas Grafab364a62006-08-22 00:01:47 -07003315 }
3316
3317 if (tb[RTA_DST]) {
3318 if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr))
3319 goto errout;
3320
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00003321 fl6.daddr = *(struct in6_addr *)nla_data(tb[RTA_DST]);
Thomas Grafab364a62006-08-22 00:01:47 -07003322 }
3323
3324 if (tb[RTA_IIF])
3325 iif = nla_get_u32(tb[RTA_IIF]);
3326
3327 if (tb[RTA_OIF])
Shmulik Ladkani72331bc2012-04-01 04:03:45 +00003328 oif = nla_get_u32(tb[RTA_OIF]);
Thomas Grafab364a62006-08-22 00:01:47 -07003329
Lorenzo Colitti2e47b292014-05-15 16:38:41 -07003330 if (tb[RTA_MARK])
3331 fl6.flowi6_mark = nla_get_u32(tb[RTA_MARK]);
3332
Thomas Grafab364a62006-08-22 00:01:47 -07003333 if (iif) {
3334 struct net_device *dev;
Shmulik Ladkani72331bc2012-04-01 04:03:45 +00003335 int flags = 0;
3336
Daniel Lezcano55786892008-03-04 13:47:47 -08003337 dev = __dev_get_by_index(net, iif);
Thomas Grafab364a62006-08-22 00:01:47 -07003338 if (!dev) {
3339 err = -ENODEV;
3340 goto errout;
3341 }
Shmulik Ladkani72331bc2012-04-01 04:03:45 +00003342
3343 fl6.flowi6_iif = iif;
3344
3345 if (!ipv6_addr_any(&fl6.saddr))
3346 flags |= RT6_LOOKUP_F_HAS_SADDR;
3347
3348 rt = (struct rt6_info *)ip6_route_input_lookup(net, dev, &fl6,
3349 flags);
3350 } else {
3351 fl6.flowi6_oif = oif;
3352
David Ahernca254492015-10-12 11:47:10 -07003353 if (netif_index_is_l3_master(net, oif)) {
3354 fl6.flowi6_flags = FLOWI_FLAG_L3MDEV_SRC |
3355 FLOWI_FLAG_SKIP_NH_OIF;
3356 }
3357
Shmulik Ladkani72331bc2012-04-01 04:03:45 +00003358 rt = (struct rt6_info *)ip6_route_output(net, NULL, &fl6);
Thomas Grafab364a62006-08-22 00:01:47 -07003359 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07003360
3361 skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
David S. Miller38308472011-12-03 18:02:47 -05003362 if (!skb) {
Amerigo Wang94e187c2012-10-29 00:13:19 +00003363 ip6_rt_put(rt);
Thomas Grafab364a62006-08-22 00:01:47 -07003364 err = -ENOBUFS;
3365 goto errout;
3366 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07003367
3368 /* Reserve room for dummy headers, this skb can pass
3369 through good chunk of routing engine.
3370 */
Arnaldo Carvalho de Melo459a98e2007-03-19 15:30:44 -07003371 skb_reset_mac_header(skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003372 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr));
3373
Changli Gaod8d1f302010-06-10 23:31:35 -07003374 skb_dst_set(skb, &rt->dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003375
David S. Miller4c9483b2011-03-12 16:22:43 -05003376 err = rt6_fill_node(net, skb, rt, &fl6.daddr, &fl6.saddr, iif,
Eric W. Biederman15e47302012-09-07 20:12:54 +00003377 RTM_NEWROUTE, NETLINK_CB(in_skb).portid,
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09003378 nlh->nlmsg_seq, 0, 0, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003379 if (err < 0) {
Thomas Grafab364a62006-08-22 00:01:47 -07003380 kfree_skb(skb);
3381 goto errout;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003382 }
3383
Eric W. Biederman15e47302012-09-07 20:12:54 +00003384 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
Thomas Grafab364a62006-08-22 00:01:47 -07003385errout:
Linus Torvalds1da177e2005-04-16 15:20:36 -07003386 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003387}
3388
Roopa Prabhu37a1d362015-09-13 10:18:33 -07003389void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info,
3390 unsigned int nlm_flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003391{
3392 struct sk_buff *skb;
Daniel Lezcano55786892008-03-04 13:47:47 -08003393 struct net *net = info->nl_net;
Denis V. Lunev528c4ce2007-12-13 09:45:12 -08003394 u32 seq;
3395 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003396
Denis V. Lunev528c4ce2007-12-13 09:45:12 -08003397 err = -ENOBUFS;
David S. Miller38308472011-12-03 18:02:47 -05003398 seq = info->nlh ? info->nlh->nlmsg_seq : 0;
Thomas Graf86872cb2006-08-22 00:01:08 -07003399
Roopa Prabhu19e42e42015-07-21 10:43:48 +02003400 skb = nlmsg_new(rt6_nlmsg_size(rt), gfp_any());
David S. Miller38308472011-12-03 18:02:47 -05003401 if (!skb)
Thomas Graf21713eb2006-08-15 00:35:24 -07003402 goto errout;
3403
Brian Haley191cd582008-08-14 15:33:21 -07003404 err = rt6_fill_node(net, skb, rt, NULL, NULL, 0,
Roopa Prabhu37a1d362015-09-13 10:18:33 -07003405 event, info->portid, seq, 0, 0, nlm_flags);
Patrick McHardy26932562007-01-31 23:16:40 -08003406 if (err < 0) {
3407 /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */
3408 WARN_ON(err == -EMSGSIZE);
3409 kfree_skb(skb);
3410 goto errout;
3411 }
Eric W. Biederman15e47302012-09-07 20:12:54 +00003412 rtnl_notify(skb, net, info->portid, RTNLGRP_IPV6_ROUTE,
Pablo Neira Ayuso1ce85fe2009-02-24 23:18:28 -08003413 info->nlh, gfp_any());
3414 return;
Thomas Graf21713eb2006-08-15 00:35:24 -07003415errout:
3416 if (err < 0)
Daniel Lezcano55786892008-03-04 13:47:47 -08003417 rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003418}
3419
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003420static int ip6_route_dev_notify(struct notifier_block *this,
Jiri Pirko351638e2013-05-28 01:30:21 +00003421 unsigned long event, void *ptr)
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003422{
Jiri Pirko351638e2013-05-28 01:30:21 +00003423 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09003424 struct net *net = dev_net(dev);
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003425
3426 if (event == NETDEV_REGISTER && (dev->flags & IFF_LOOPBACK)) {
Changli Gaod8d1f302010-06-10 23:31:35 -07003427 net->ipv6.ip6_null_entry->dst.dev = dev;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003428 net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev);
3429#ifdef CONFIG_IPV6_MULTIPLE_TABLES
Changli Gaod8d1f302010-06-10 23:31:35 -07003430 net->ipv6.ip6_prohibit_entry->dst.dev = dev;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003431 net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev);
Changli Gaod8d1f302010-06-10 23:31:35 -07003432 net->ipv6.ip6_blk_hole_entry->dst.dev = dev;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003433 net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev);
3434#endif
3435 }
3436
3437 return NOTIFY_OK;
3438}
3439
Linus Torvalds1da177e2005-04-16 15:20:36 -07003440/*
3441 * /proc
3442 */
3443
3444#ifdef CONFIG_PROC_FS
3445
Alexey Dobriyan33120b32007-11-06 05:27:11 -08003446static const struct file_operations ipv6_route_proc_fops = {
3447 .owner = THIS_MODULE,
3448 .open = ipv6_route_open,
3449 .read = seq_read,
3450 .llseek = seq_lseek,
Hannes Frederic Sowa8d2ca1d2013-09-21 16:55:59 +02003451 .release = seq_release_net,
Alexey Dobriyan33120b32007-11-06 05:27:11 -08003452};
3453
Linus Torvalds1da177e2005-04-16 15:20:36 -07003454static int rt6_stats_seq_show(struct seq_file *seq, void *v)
3455{
Daniel Lezcano69ddb802008-03-04 13:46:23 -08003456 struct net *net = (struct net *)seq->private;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003457 seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n",
Daniel Lezcano69ddb802008-03-04 13:46:23 -08003458 net->ipv6.rt6_stats->fib_nodes,
3459 net->ipv6.rt6_stats->fib_route_nodes,
3460 net->ipv6.rt6_stats->fib_rt_alloc,
3461 net->ipv6.rt6_stats->fib_rt_entries,
3462 net->ipv6.rt6_stats->fib_rt_cache,
Eric Dumazetfc66f952010-10-08 06:37:34 +00003463 dst_entries_get_slow(&net->ipv6.ip6_dst_ops),
Daniel Lezcano69ddb802008-03-04 13:46:23 -08003464 net->ipv6.rt6_stats->fib_discarded_routes);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003465
3466 return 0;
3467}
3468
3469static int rt6_stats_seq_open(struct inode *inode, struct file *file)
3470{
Pavel Emelyanovde05c552008-07-18 04:07:21 -07003471 return single_open_net(inode, file, rt6_stats_seq_show);
Daniel Lezcano69ddb802008-03-04 13:46:23 -08003472}
3473
Arjan van de Ven9a321442007-02-12 00:55:35 -08003474static const struct file_operations rt6_stats_seq_fops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003475 .owner = THIS_MODULE,
3476 .open = rt6_stats_seq_open,
3477 .read = seq_read,
3478 .llseek = seq_lseek,
Pavel Emelyanovb6fcbdb2008-07-18 04:07:44 -07003479 .release = single_release_net,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003480};
3481#endif /* CONFIG_PROC_FS */
3482
3483#ifdef CONFIG_SYSCTL
3484
Linus Torvalds1da177e2005-04-16 15:20:36 -07003485static
Joe Perchesfe2c6332013-06-11 23:04:25 -07003486int ipv6_sysctl_rtcache_flush(struct ctl_table *ctl, int write,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003487 void __user *buffer, size_t *lenp, loff_t *ppos)
3488{
Lucian Adrian Grijincuc486da32011-02-24 19:48:03 +00003489 struct net *net;
3490 int delay;
3491 if (!write)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003492 return -EINVAL;
Lucian Adrian Grijincuc486da32011-02-24 19:48:03 +00003493
3494 net = (struct net *)ctl->extra1;
3495 delay = net->ipv6.sysctl.flush_delay;
3496 proc_dointvec(ctl, write, buffer, lenp, ppos);
Michal Kubeček2ac3ac82013-08-01 10:04:14 +02003497 fib6_run_gc(delay <= 0 ? 0 : (unsigned long)delay, net, delay > 0);
Lucian Adrian Grijincuc486da32011-02-24 19:48:03 +00003498 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003499}
3500
Joe Perchesfe2c6332013-06-11 23:04:25 -07003501struct ctl_table ipv6_route_table_template[] = {
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09003502 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003503 .procname = "flush",
Daniel Lezcano49905092008-01-10 03:01:01 -08003504 .data = &init_net.ipv6.sysctl.flush_delay,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003505 .maxlen = sizeof(int),
Dave Jones89c8b3a12005-04-28 12:11:49 -07003506 .mode = 0200,
Alexey Dobriyan6d9f2392008-11-03 18:21:05 -08003507 .proc_handler = ipv6_sysctl_rtcache_flush
Linus Torvalds1da177e2005-04-16 15:20:36 -07003508 },
3509 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003510 .procname = "gc_thresh",
Daniel Lezcano9a7ec3a2008-03-04 13:48:53 -08003511 .data = &ip6_dst_ops_template.gc_thresh,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003512 .maxlen = sizeof(int),
3513 .mode = 0644,
Alexey Dobriyan6d9f2392008-11-03 18:21:05 -08003514 .proc_handler = proc_dointvec,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003515 },
3516 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003517 .procname = "max_size",
Daniel Lezcano49905092008-01-10 03:01:01 -08003518 .data = &init_net.ipv6.sysctl.ip6_rt_max_size,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003519 .maxlen = sizeof(int),
3520 .mode = 0644,
Alexey Dobriyan6d9f2392008-11-03 18:21:05 -08003521 .proc_handler = proc_dointvec,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003522 },
3523 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003524 .procname = "gc_min_interval",
Daniel Lezcano49905092008-01-10 03:01:01 -08003525 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003526 .maxlen = sizeof(int),
3527 .mode = 0644,
Alexey Dobriyan6d9f2392008-11-03 18:21:05 -08003528 .proc_handler = proc_dointvec_jiffies,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003529 },
3530 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003531 .procname = "gc_timeout",
Daniel Lezcano49905092008-01-10 03:01:01 -08003532 .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003533 .maxlen = sizeof(int),
3534 .mode = 0644,
Alexey Dobriyan6d9f2392008-11-03 18:21:05 -08003535 .proc_handler = proc_dointvec_jiffies,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003536 },
3537 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003538 .procname = "gc_interval",
Daniel Lezcano49905092008-01-10 03:01:01 -08003539 .data = &init_net.ipv6.sysctl.ip6_rt_gc_interval,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003540 .maxlen = sizeof(int),
3541 .mode = 0644,
Alexey Dobriyan6d9f2392008-11-03 18:21:05 -08003542 .proc_handler = proc_dointvec_jiffies,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003543 },
3544 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003545 .procname = "gc_elasticity",
Daniel Lezcano49905092008-01-10 03:01:01 -08003546 .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003547 .maxlen = sizeof(int),
3548 .mode = 0644,
Min Zhangf3d3f612010-08-14 22:42:51 -07003549 .proc_handler = proc_dointvec,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003550 },
3551 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003552 .procname = "mtu_expires",
Daniel Lezcano49905092008-01-10 03:01:01 -08003553 .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003554 .maxlen = sizeof(int),
3555 .mode = 0644,
Alexey Dobriyan6d9f2392008-11-03 18:21:05 -08003556 .proc_handler = proc_dointvec_jiffies,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003557 },
3558 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003559 .procname = "min_adv_mss",
Daniel Lezcano49905092008-01-10 03:01:01 -08003560 .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003561 .maxlen = sizeof(int),
3562 .mode = 0644,
Min Zhangf3d3f612010-08-14 22:42:51 -07003563 .proc_handler = proc_dointvec,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003564 },
3565 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003566 .procname = "gc_min_interval_ms",
Daniel Lezcano49905092008-01-10 03:01:01 -08003567 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003568 .maxlen = sizeof(int),
3569 .mode = 0644,
Alexey Dobriyan6d9f2392008-11-03 18:21:05 -08003570 .proc_handler = proc_dointvec_ms_jiffies,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003571 },
Eric W. Biedermanf8572d82009-11-05 13:32:03 -08003572 { }
Linus Torvalds1da177e2005-04-16 15:20:36 -07003573};
3574
Alexey Dobriyan2c8c1e72010-01-17 03:35:32 +00003575struct ctl_table * __net_init ipv6_route_sysctl_init(struct net *net)
Daniel Lezcano760f2d02008-01-10 02:53:43 -08003576{
3577 struct ctl_table *table;
3578
3579 table = kmemdup(ipv6_route_table_template,
3580 sizeof(ipv6_route_table_template),
3581 GFP_KERNEL);
YOSHIFUJI Hideaki5ee09102008-02-28 00:24:28 +09003582
3583 if (table) {
3584 table[0].data = &net->ipv6.sysctl.flush_delay;
Lucian Adrian Grijincuc486da32011-02-24 19:48:03 +00003585 table[0].extra1 = net;
Alexey Dobriyan86393e52009-08-29 01:34:49 +00003586 table[1].data = &net->ipv6.ip6_dst_ops.gc_thresh;
YOSHIFUJI Hideaki5ee09102008-02-28 00:24:28 +09003587 table[2].data = &net->ipv6.sysctl.ip6_rt_max_size;
3588 table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
3589 table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout;
3590 table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval;
3591 table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity;
3592 table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires;
3593 table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss;
Alexey Dobriyan9c69fab2009-12-18 20:11:03 -08003594 table[9].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
Eric W. Biederman464dc802012-11-16 03:02:59 +00003595
3596 /* Don't export sysctls to unprivileged users */
3597 if (net->user_ns != &init_user_ns)
3598 table[0].procname = NULL;
YOSHIFUJI Hideaki5ee09102008-02-28 00:24:28 +09003599 }
3600
Daniel Lezcano760f2d02008-01-10 02:53:43 -08003601 return table;
3602}
Linus Torvalds1da177e2005-04-16 15:20:36 -07003603#endif
3604
Alexey Dobriyan2c8c1e72010-01-17 03:35:32 +00003605static int __net_init ip6_route_net_init(struct net *net)
Daniel Lezcanocdb18762008-03-04 13:45:33 -08003606{
Pavel Emelyanov633d424b2008-04-21 14:25:23 -07003607 int ret = -ENOMEM;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003608
Alexey Dobriyan86393e52009-08-29 01:34:49 +00003609 memcpy(&net->ipv6.ip6_dst_ops, &ip6_dst_ops_template,
3610 sizeof(net->ipv6.ip6_dst_ops));
Benjamin Theryf2fc6a52008-03-04 13:49:23 -08003611
Eric Dumazetfc66f952010-10-08 06:37:34 +00003612 if (dst_entries_init(&net->ipv6.ip6_dst_ops) < 0)
3613 goto out_ip6_dst_ops;
3614
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003615 net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template,
3616 sizeof(*net->ipv6.ip6_null_entry),
3617 GFP_KERNEL);
3618 if (!net->ipv6.ip6_null_entry)
Eric Dumazetfc66f952010-10-08 06:37:34 +00003619 goto out_ip6_dst_entries;
Changli Gaod8d1f302010-06-10 23:31:35 -07003620 net->ipv6.ip6_null_entry->dst.path =
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003621 (struct dst_entry *)net->ipv6.ip6_null_entry;
Changli Gaod8d1f302010-06-10 23:31:35 -07003622 net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops;
David S. Miller62fa8a82011-01-26 20:51:05 -08003623 dst_init_metrics(&net->ipv6.ip6_null_entry->dst,
3624 ip6_template_metrics, true);
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003625
3626#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3627 net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template,
3628 sizeof(*net->ipv6.ip6_prohibit_entry),
3629 GFP_KERNEL);
Peter Zijlstra68fffc62008-10-07 14:12:10 -07003630 if (!net->ipv6.ip6_prohibit_entry)
3631 goto out_ip6_null_entry;
Changli Gaod8d1f302010-06-10 23:31:35 -07003632 net->ipv6.ip6_prohibit_entry->dst.path =
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003633 (struct dst_entry *)net->ipv6.ip6_prohibit_entry;
Changli Gaod8d1f302010-06-10 23:31:35 -07003634 net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops;
David S. Miller62fa8a82011-01-26 20:51:05 -08003635 dst_init_metrics(&net->ipv6.ip6_prohibit_entry->dst,
3636 ip6_template_metrics, true);
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003637
3638 net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template,
3639 sizeof(*net->ipv6.ip6_blk_hole_entry),
3640 GFP_KERNEL);
Peter Zijlstra68fffc62008-10-07 14:12:10 -07003641 if (!net->ipv6.ip6_blk_hole_entry)
3642 goto out_ip6_prohibit_entry;
Changli Gaod8d1f302010-06-10 23:31:35 -07003643 net->ipv6.ip6_blk_hole_entry->dst.path =
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003644 (struct dst_entry *)net->ipv6.ip6_blk_hole_entry;
Changli Gaod8d1f302010-06-10 23:31:35 -07003645 net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops;
David S. Miller62fa8a82011-01-26 20:51:05 -08003646 dst_init_metrics(&net->ipv6.ip6_blk_hole_entry->dst,
3647 ip6_template_metrics, true);
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003648#endif
3649
Peter Zijlstrab339a47c2008-10-07 14:15:00 -07003650 net->ipv6.sysctl.flush_delay = 0;
3651 net->ipv6.sysctl.ip6_rt_max_size = 4096;
3652 net->ipv6.sysctl.ip6_rt_gc_min_interval = HZ / 2;
3653 net->ipv6.sysctl.ip6_rt_gc_timeout = 60*HZ;
3654 net->ipv6.sysctl.ip6_rt_gc_interval = 30*HZ;
3655 net->ipv6.sysctl.ip6_rt_gc_elasticity = 9;
3656 net->ipv6.sysctl.ip6_rt_mtu_expires = 10*60*HZ;
3657 net->ipv6.sysctl.ip6_rt_min_advmss = IPV6_MIN_MTU - 20 - 40;
3658
Benjamin Thery6891a342008-03-04 13:49:47 -08003659 net->ipv6.ip6_rt_gc_expire = 30*HZ;
3660
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003661 ret = 0;
3662out:
3663 return ret;
Benjamin Theryf2fc6a52008-03-04 13:49:23 -08003664
Peter Zijlstra68fffc62008-10-07 14:12:10 -07003665#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3666out_ip6_prohibit_entry:
3667 kfree(net->ipv6.ip6_prohibit_entry);
3668out_ip6_null_entry:
3669 kfree(net->ipv6.ip6_null_entry);
3670#endif
Eric Dumazetfc66f952010-10-08 06:37:34 +00003671out_ip6_dst_entries:
3672 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
Benjamin Theryf2fc6a52008-03-04 13:49:23 -08003673out_ip6_dst_ops:
Benjamin Theryf2fc6a52008-03-04 13:49:23 -08003674 goto out;
Daniel Lezcanocdb18762008-03-04 13:45:33 -08003675}
3676
Alexey Dobriyan2c8c1e72010-01-17 03:35:32 +00003677static void __net_exit ip6_route_net_exit(struct net *net)
Daniel Lezcanocdb18762008-03-04 13:45:33 -08003678{
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003679 kfree(net->ipv6.ip6_null_entry);
3680#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3681 kfree(net->ipv6.ip6_prohibit_entry);
3682 kfree(net->ipv6.ip6_blk_hole_entry);
3683#endif
Xiaotian Feng41bb78b2010-11-02 16:11:05 +00003684 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
Daniel Lezcanocdb18762008-03-04 13:45:33 -08003685}
3686
Thomas Grafd1896342012-06-18 12:08:33 +00003687static int __net_init ip6_route_net_init_late(struct net *net)
3688{
3689#ifdef CONFIG_PROC_FS
Gao fengd4beaa62013-02-18 01:34:54 +00003690 proc_create("ipv6_route", 0, net->proc_net, &ipv6_route_proc_fops);
3691 proc_create("rt6_stats", S_IRUGO, net->proc_net, &rt6_stats_seq_fops);
Thomas Grafd1896342012-06-18 12:08:33 +00003692#endif
3693 return 0;
3694}
3695
3696static void __net_exit ip6_route_net_exit_late(struct net *net)
3697{
3698#ifdef CONFIG_PROC_FS
Gao fengece31ff2013-02-18 01:34:56 +00003699 remove_proc_entry("ipv6_route", net->proc_net);
3700 remove_proc_entry("rt6_stats", net->proc_net);
Thomas Grafd1896342012-06-18 12:08:33 +00003701#endif
3702}
3703
Daniel Lezcanocdb18762008-03-04 13:45:33 -08003704static struct pernet_operations ip6_route_net_ops = {
3705 .init = ip6_route_net_init,
3706 .exit = ip6_route_net_exit,
3707};
3708
David S. Millerc3426b42012-06-09 16:27:05 -07003709static int __net_init ipv6_inetpeer_init(struct net *net)
3710{
3711 struct inet_peer_base *bp = kmalloc(sizeof(*bp), GFP_KERNEL);
3712
3713 if (!bp)
3714 return -ENOMEM;
3715 inet_peer_base_init(bp);
3716 net->ipv6.peers = bp;
3717 return 0;
3718}
3719
3720static void __net_exit ipv6_inetpeer_exit(struct net *net)
3721{
3722 struct inet_peer_base *bp = net->ipv6.peers;
3723
3724 net->ipv6.peers = NULL;
David S. Miller56a6b242012-06-09 16:32:41 -07003725 inetpeer_invalidate_tree(bp);
David S. Millerc3426b42012-06-09 16:27:05 -07003726 kfree(bp);
3727}
3728
David S. Miller2b823f72012-06-09 19:00:16 -07003729static struct pernet_operations ipv6_inetpeer_ops = {
David S. Millerc3426b42012-06-09 16:27:05 -07003730 .init = ipv6_inetpeer_init,
3731 .exit = ipv6_inetpeer_exit,
3732};
3733
Thomas Grafd1896342012-06-18 12:08:33 +00003734static struct pernet_operations ip6_route_net_late_ops = {
3735 .init = ip6_route_net_init_late,
3736 .exit = ip6_route_net_exit_late,
3737};
3738
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003739static struct notifier_block ip6_route_dev_notifier = {
3740 .notifier_call = ip6_route_dev_notify,
3741 .priority = 0,
3742};
3743
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003744int __init ip6_route_init(void)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003745{
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003746 int ret;
Martin KaFai Lau8d0b94a2015-05-22 20:56:04 -07003747 int cpu;
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003748
Daniel Lezcano9a7ec3a2008-03-04 13:48:53 -08003749 ret = -ENOMEM;
3750 ip6_dst_ops_template.kmem_cachep =
3751 kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0,
3752 SLAB_HWCACHE_ALIGN, NULL);
3753 if (!ip6_dst_ops_template.kmem_cachep)
Fernando Carrijoc19a28e2009-01-07 18:09:08 -08003754 goto out;
David S. Miller14e50e52007-05-24 18:17:54 -07003755
Eric Dumazetfc66f952010-10-08 06:37:34 +00003756 ret = dst_entries_init(&ip6_dst_blackhole_ops);
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003757 if (ret)
Daniel Lezcanobdb32892008-03-04 13:48:10 -08003758 goto out_kmem_cache;
Daniel Lezcanobdb32892008-03-04 13:48:10 -08003759
David S. Millerc3426b42012-06-09 16:27:05 -07003760 ret = register_pernet_subsys(&ipv6_inetpeer_ops);
3761 if (ret)
David S. Millere8803b62012-06-16 01:12:19 -07003762 goto out_dst_entries;
Thomas Graf2a0c4512012-06-14 23:00:17 +00003763
David S. Miller7e52b332012-06-15 15:51:55 -07003764 ret = register_pernet_subsys(&ip6_route_net_ops);
3765 if (ret)
3766 goto out_register_inetpeer;
David S. Millerc3426b42012-06-09 16:27:05 -07003767
Arnaud Ebalard5dc121e2008-10-01 02:37:56 -07003768 ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep;
3769
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003770 /* Registering of the loopback is done before this portion of code,
3771 * the loopback reference in rt6_info will not be taken, do it
3772 * manually for init_net */
Changli Gaod8d1f302010-06-10 23:31:35 -07003773 init_net.ipv6.ip6_null_entry->dst.dev = init_net.loopback_dev;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003774 init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3775 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
Changli Gaod8d1f302010-06-10 23:31:35 -07003776 init_net.ipv6.ip6_prohibit_entry->dst.dev = init_net.loopback_dev;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003777 init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
Changli Gaod8d1f302010-06-10 23:31:35 -07003778 init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003779 init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3780 #endif
David S. Millere8803b62012-06-16 01:12:19 -07003781 ret = fib6_init();
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003782 if (ret)
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003783 goto out_register_subsys;
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003784
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003785 ret = xfrm6_init();
3786 if (ret)
David S. Millere8803b62012-06-16 01:12:19 -07003787 goto out_fib6_init;
Daniel Lezcanoc35b7e72007-12-08 00:14:11 -08003788
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003789 ret = fib6_rules_init();
3790 if (ret)
3791 goto xfrm6_init;
Daniel Lezcano7e5449c2007-12-08 00:14:54 -08003792
Thomas Grafd1896342012-06-18 12:08:33 +00003793 ret = register_pernet_subsys(&ip6_route_net_late_ops);
3794 if (ret)
3795 goto fib6_rules_init;
3796
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003797 ret = -ENOBUFS;
Greg Rosec7ac8672011-06-10 01:27:09 +00003798 if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL, NULL) ||
3799 __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL, NULL) ||
3800 __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL, NULL))
Thomas Grafd1896342012-06-18 12:08:33 +00003801 goto out_register_late_subsys;
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003802
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003803 ret = register_netdevice_notifier(&ip6_route_dev_notifier);
Daniel Lezcanocdb18762008-03-04 13:45:33 -08003804 if (ret)
Thomas Grafd1896342012-06-18 12:08:33 +00003805 goto out_register_late_subsys;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003806
Martin KaFai Lau8d0b94a2015-05-22 20:56:04 -07003807 for_each_possible_cpu(cpu) {
3808 struct uncached_list *ul = per_cpu_ptr(&rt6_uncached_list, cpu);
3809
3810 INIT_LIST_HEAD(&ul->head);
3811 spin_lock_init(&ul->lock);
3812 }
3813
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003814out:
3815 return ret;
3816
Thomas Grafd1896342012-06-18 12:08:33 +00003817out_register_late_subsys:
3818 unregister_pernet_subsys(&ip6_route_net_late_ops);
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003819fib6_rules_init:
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003820 fib6_rules_cleanup();
3821xfrm6_init:
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003822 xfrm6_fini();
Thomas Graf2a0c4512012-06-14 23:00:17 +00003823out_fib6_init:
3824 fib6_gc_cleanup();
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003825out_register_subsys:
3826 unregister_pernet_subsys(&ip6_route_net_ops);
David S. Miller7e52b332012-06-15 15:51:55 -07003827out_register_inetpeer:
3828 unregister_pernet_subsys(&ipv6_inetpeer_ops);
Eric Dumazetfc66f952010-10-08 06:37:34 +00003829out_dst_entries:
3830 dst_entries_destroy(&ip6_dst_blackhole_ops);
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003831out_kmem_cache:
Benjamin Theryf2fc6a52008-03-04 13:49:23 -08003832 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003833 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003834}
3835
3836void ip6_route_cleanup(void)
3837{
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003838 unregister_netdevice_notifier(&ip6_route_dev_notifier);
Thomas Grafd1896342012-06-18 12:08:33 +00003839 unregister_pernet_subsys(&ip6_route_net_late_ops);
Thomas Graf101367c2006-08-04 03:39:02 -07003840 fib6_rules_cleanup();
Linus Torvalds1da177e2005-04-16 15:20:36 -07003841 xfrm6_fini();
Linus Torvalds1da177e2005-04-16 15:20:36 -07003842 fib6_gc_cleanup();
David S. Millerc3426b42012-06-09 16:27:05 -07003843 unregister_pernet_subsys(&ipv6_inetpeer_ops);
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003844 unregister_pernet_subsys(&ip6_route_net_ops);
Xiaotian Feng41bb78b2010-11-02 16:11:05 +00003845 dst_entries_destroy(&ip6_dst_blackhole_ops);
Benjamin Theryf2fc6a52008-03-04 13:49:23 -08003846 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003847}