blob: 49817555449e2956fcfe81555e646b5f9b53261a [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
David Ahern9ff74382016-06-13 13:44:19 -07001045struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table,
1046 int oif, 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}
David Ahern9ff74382016-06-13 13:44:19 -07001142EXPORT_SYMBOL_GPL(ip6_pol_route);
Thomas Grafc71099a2006-08-04 23:20:06 -07001143
Daniel Lezcano8ed67782008-03-04 13:48:30 -08001144static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table,
David S. Miller4c9483b2011-03-12 16:22:43 -05001145 struct flowi6 *fl6, int flags)
Pavel Emelyanov4acad722007-10-15 13:02:51 -07001146{
David S. Miller4c9483b2011-03-12 16:22:43 -05001147 return ip6_pol_route(net, table, fl6->flowi6_iif, fl6, flags);
Pavel Emelyanov4acad722007-10-15 13:02:51 -07001148}
1149
Shmulik Ladkani72331bc2012-04-01 04:03:45 +00001150static struct dst_entry *ip6_route_input_lookup(struct net *net,
1151 struct net_device *dev,
1152 struct flowi6 *fl6, int flags)
1153{
1154 if (rt6_need_strict(&fl6->daddr) && dev->type != ARPHRD_PIMREG)
1155 flags |= RT6_LOOKUP_F_IFACE;
1156
1157 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_input);
1158}
1159
Thomas Grafc71099a2006-08-04 23:20:06 -07001160void ip6_route_input(struct sk_buff *skb)
1161{
Eric Dumazetb71d1d42011-04-22 04:53:02 +00001162 const struct ipv6hdr *iph = ipv6_hdr(skb);
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09001163 struct net *net = dev_net(skb->dev);
Thomas Grafadaa70b2006-10-13 15:01:03 -07001164 int flags = RT6_LOOKUP_F_HAS_SADDR;
Jiri Benc904af042015-08-20 13:56:31 +02001165 struct ip_tunnel_info *tun_info;
David S. Miller4c9483b2011-03-12 16:22:43 -05001166 struct flowi6 fl6 = {
David Ahernca254492015-10-12 11:47:10 -07001167 .flowi6_iif = l3mdev_fib_oif(skb->dev),
David S. Miller4c9483b2011-03-12 16:22:43 -05001168 .daddr = iph->daddr,
1169 .saddr = iph->saddr,
YOSHIFUJI Hideaki / 吉藤英明6502ca52013-01-13 05:01:51 +00001170 .flowlabel = ip6_flowinfo(iph),
David S. Miller4c9483b2011-03-12 16:22:43 -05001171 .flowi6_mark = skb->mark,
1172 .flowi6_proto = iph->nexthdr,
Thomas Grafc71099a2006-08-04 23:20:06 -07001173 };
Thomas Grafadaa70b2006-10-13 15:01:03 -07001174
Jiri Benc904af042015-08-20 13:56:31 +02001175 tun_info = skb_tunnel_info(skb);
Jiri Benc46fa0622015-08-28 20:48:19 +02001176 if (tun_info && !(tun_info->mode & IP_TUNNEL_INFO_TX))
Jiri Benc904af042015-08-20 13:56:31 +02001177 fl6.flowi6_tun_key.tun_id = tun_info->key.tun_id;
Jiri Benc06e9d042015-08-20 13:56:26 +02001178 skb_dst_drop(skb);
Shmulik Ladkani72331bc2012-04-01 04:03:45 +00001179 skb_dst_set(skb, ip6_route_input_lookup(net, skb->dev, &fl6, flags));
Thomas Grafc71099a2006-08-04 23:20:06 -07001180}
1181
Daniel Lezcano8ed67782008-03-04 13:48:30 -08001182static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table,
David S. Miller4c9483b2011-03-12 16:22:43 -05001183 struct flowi6 *fl6, int flags)
Thomas Grafc71099a2006-08-04 23:20:06 -07001184{
David S. Miller4c9483b2011-03-12 16:22:43 -05001185 return ip6_pol_route(net, table, fl6->flowi6_oif, fl6, flags);
Thomas Grafc71099a2006-08-04 23:20:06 -07001186}
1187
Paolo Abeni6f21c962016-01-29 12:30:19 +01001188struct dst_entry *ip6_route_output_flags(struct net *net, const struct sock *sk,
1189 struct flowi6 *fl6, int flags)
Thomas Grafc71099a2006-08-04 23:20:06 -07001190{
David Ahernca254492015-10-12 11:47:10 -07001191 struct dst_entry *dst;
David Ahernd46a9d62015-10-21 08:42:22 -07001192 bool any_src;
Thomas Grafc71099a2006-08-04 23:20:06 -07001193
David Ahern4a658962016-05-07 16:48:59 -07001194 dst = l3mdev_get_rt6_dst(net, fl6);
David Ahernca254492015-10-12 11:47:10 -07001195 if (dst)
1196 return dst;
1197
Pavel Emelyanov1fb94892012-08-08 21:53:36 +00001198 fl6->flowi6_iif = LOOPBACK_IFINDEX;
David McCullough4dc27d1c2012-06-25 15:42:26 +00001199
David Ahernd46a9d62015-10-21 08:42:22 -07001200 any_src = ipv6_addr_any(&fl6->saddr);
David Ahern741a11d2015-09-28 10:12:13 -07001201 if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl6->daddr) ||
David Ahernd46a9d62015-10-21 08:42:22 -07001202 (fl6->flowi6_oif && any_src))
YOSHIFUJI Hideaki77d16f42006-08-23 17:25:05 -07001203 flags |= RT6_LOOKUP_F_IFACE;
Thomas Grafc71099a2006-08-04 23:20:06 -07001204
David Ahernd46a9d62015-10-21 08:42:22 -07001205 if (!any_src)
Thomas Grafadaa70b2006-10-13 15:01:03 -07001206 flags |= RT6_LOOKUP_F_HAS_SADDR;
YOSHIFUJI Hideaki / 吉藤英明0c9a2ac2010-03-07 00:14:44 +00001207 else if (sk)
1208 flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs);
Thomas Grafadaa70b2006-10-13 15:01:03 -07001209
David S. Miller4c9483b2011-03-12 16:22:43 -05001210 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_output);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001211}
Paolo Abeni6f21c962016-01-29 12:30:19 +01001212EXPORT_SYMBOL_GPL(ip6_route_output_flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001213
David S. Miller2774c132011-03-01 14:59:04 -08001214struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_orig)
David S. Miller14e50e52007-05-24 18:17:54 -07001215{
David S. Miller5c1e6aa2011-04-28 14:13:38 -07001216 struct rt6_info *rt, *ort = (struct rt6_info *) dst_orig;
David S. Miller14e50e52007-05-24 18:17:54 -07001217 struct dst_entry *new = NULL;
1218
David S. Millerf5b0a872012-07-19 12:31:33 -07001219 rt = dst_alloc(&ip6_dst_blackhole_ops, ort->dst.dev, 1, DST_OBSOLETE_NONE, 0);
David S. Miller14e50e52007-05-24 18:17:54 -07001220 if (rt) {
Martin KaFai Lau0a1f5962015-10-15 16:39:58 -07001221 rt6_info_init(rt);
1222
Changli Gaod8d1f302010-06-10 23:31:35 -07001223 new = &rt->dst;
David S. Miller14e50e52007-05-24 18:17:54 -07001224 new->__use = 1;
Herbert Xu352e5122007-11-13 21:34:06 -08001225 new->input = dst_discard;
Eric W. Biedermanede20592015-10-07 16:48:47 -05001226 new->output = dst_discard_out;
David S. Miller14e50e52007-05-24 18:17:54 -07001227
Martin KaFai Lau0a1f5962015-10-15 16:39:58 -07001228 dst_copy_metrics(new, &ort->dst);
David S. Miller14e50e52007-05-24 18:17:54 -07001229 rt->rt6i_idev = ort->rt6i_idev;
1230 if (rt->rt6i_idev)
1231 in6_dev_hold(rt->rt6i_idev);
David S. Miller14e50e52007-05-24 18:17:54 -07001232
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00001233 rt->rt6i_gateway = ort->rt6i_gateway;
Martin KaFai Lau0a1f5962015-10-15 16:39:58 -07001234 rt->rt6i_flags = ort->rt6i_flags & ~RTF_PCPU;
David S. Miller14e50e52007-05-24 18:17:54 -07001235 rt->rt6i_metric = 0;
1236
1237 memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key));
1238#ifdef CONFIG_IPV6_SUBTREES
1239 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1240#endif
1241
1242 dst_free(new);
1243 }
1244
David S. Miller69ead7a2011-03-01 14:45:33 -08001245 dst_release(dst_orig);
1246 return new ? new : ERR_PTR(-ENOMEM);
David S. Miller14e50e52007-05-24 18:17:54 -07001247}
David S. Miller14e50e52007-05-24 18:17:54 -07001248
Linus Torvalds1da177e2005-04-16 15:20:36 -07001249/*
1250 * Destination cache support functions
1251 */
1252
Martin KaFai Lau4b32b5a2015-04-28 13:03:06 -07001253static void rt6_dst_from_metrics_check(struct rt6_info *rt)
1254{
1255 if (rt->dst.from &&
1256 dst_metrics_ptr(&rt->dst) != dst_metrics_ptr(rt->dst.from))
1257 dst_init_metrics(&rt->dst, dst_metrics_ptr(rt->dst.from), true);
1258}
1259
Martin KaFai Lau3da59bd2015-05-22 20:56:03 -07001260static struct dst_entry *rt6_check(struct rt6_info *rt, u32 cookie)
1261{
1262 if (!rt->rt6i_node || (rt->rt6i_node->fn_sernum != cookie))
1263 return NULL;
1264
1265 if (rt6_check_expired(rt))
1266 return NULL;
1267
1268 return &rt->dst;
1269}
1270
1271static struct dst_entry *rt6_dst_from_check(struct rt6_info *rt, u32 cookie)
1272{
Martin KaFai Lau5973fb12015-11-11 11:51:07 -08001273 if (!__rt6_check_expired(rt) &&
1274 rt->dst.obsolete == DST_OBSOLETE_FORCE_CHK &&
Martin KaFai Lau3da59bd2015-05-22 20:56:03 -07001275 rt6_check((struct rt6_info *)(rt->dst.from), cookie))
1276 return &rt->dst;
1277 else
1278 return NULL;
1279}
1280
Linus Torvalds1da177e2005-04-16 15:20:36 -07001281static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie)
1282{
1283 struct rt6_info *rt;
1284
1285 rt = (struct rt6_info *) dst;
1286
Nicolas Dichtel6f3118b2012-09-10 22:09:46 +00001287 /* All IPV6 dsts are created with ->obsolete set to the value
1288 * DST_OBSOLETE_FORCE_CHK which forces validation calls down
1289 * into this function always.
1290 */
Hannes Frederic Sowae3bc10b2013-10-24 07:48:24 +02001291
Martin KaFai Lau4b32b5a2015-04-28 13:03:06 -07001292 rt6_dst_from_metrics_check(rt);
1293
Martin KaFai Lau02bcf4e2015-11-11 11:51:08 -08001294 if (rt->rt6i_flags & RTF_PCPU ||
1295 (unlikely(dst->flags & DST_NOCACHE) && rt->dst.from))
Martin KaFai Lau3da59bd2015-05-22 20:56:03 -07001296 return rt6_dst_from_check(rt, cookie);
1297 else
1298 return rt6_check(rt, cookie);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001299}
1300
1301static struct dst_entry *ip6_negative_advice(struct dst_entry *dst)
1302{
1303 struct rt6_info *rt = (struct rt6_info *) dst;
1304
1305 if (rt) {
YOSHIFUJI Hideaki / 吉藤英明54c1a852010-03-28 07:15:45 +00001306 if (rt->rt6i_flags & RTF_CACHE) {
1307 if (rt6_check_expired(rt)) {
1308 ip6_del_rt(rt);
1309 dst = NULL;
1310 }
1311 } else {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001312 dst_release(dst);
YOSHIFUJI Hideaki / 吉藤英明54c1a852010-03-28 07:15:45 +00001313 dst = NULL;
1314 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001315 }
YOSHIFUJI Hideaki / 吉藤英明54c1a852010-03-28 07:15:45 +00001316 return dst;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001317}
1318
1319static void ip6_link_failure(struct sk_buff *skb)
1320{
1321 struct rt6_info *rt;
1322
Alexey Dobriyan3ffe5332010-02-18 08:25:24 +00001323 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001324
Eric Dumazetadf30902009-06-02 05:19:30 +00001325 rt = (struct rt6_info *) skb_dst(skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001326 if (rt) {
Hannes Frederic Sowa1eb4f752013-07-10 23:00:57 +02001327 if (rt->rt6i_flags & RTF_CACHE) {
1328 dst_hold(&rt->dst);
Martin KaFai Lau8e3d5be2015-09-15 14:30:08 -07001329 ip6_del_rt(rt);
Hannes Frederic Sowa1eb4f752013-07-10 23:00:57 +02001330 } else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001331 rt->rt6i_node->fn_sernum = -1;
Hannes Frederic Sowa1eb4f752013-07-10 23:00:57 +02001332 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001333 }
1334}
1335
Martin KaFai Lau45e4fd22015-05-22 20:56:00 -07001336static void rt6_do_update_pmtu(struct rt6_info *rt, u32 mtu)
1337{
1338 struct net *net = dev_net(rt->dst.dev);
1339
1340 rt->rt6i_flags |= RTF_MODIFIED;
1341 rt->rt6i_pmtu = mtu;
1342 rt6_update_expires(rt, net->ipv6.sysctl.ip6_rt_mtu_expires);
1343}
1344
Martin KaFai Lau0d3f6d22015-11-11 11:51:06 -08001345static bool rt6_cache_allowed_for_pmtu(const struct rt6_info *rt)
1346{
1347 return !(rt->rt6i_flags & RTF_CACHE) &&
1348 (rt->rt6i_flags & RTF_PCPU || rt->rt6i_node);
1349}
1350
Martin KaFai Lau45e4fd22015-05-22 20:56:00 -07001351static void __ip6_rt_update_pmtu(struct dst_entry *dst, const struct sock *sk,
1352 const struct ipv6hdr *iph, u32 mtu)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001353{
Ian Morris67ba4152014-08-24 21:53:10 +01001354 struct rt6_info *rt6 = (struct rt6_info *)dst;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001355
Martin KaFai Lau45e4fd22015-05-22 20:56:00 -07001356 if (rt6->rt6i_flags & RTF_LOCAL)
1357 return;
1358
David S. Miller81aded22012-06-15 14:54:11 -07001359 dst_confirm(dst);
Martin KaFai Lau45e4fd22015-05-22 20:56:00 -07001360 mtu = max_t(u32, mtu, IPV6_MIN_MTU);
1361 if (mtu >= dst_mtu(dst))
1362 return;
David S. Miller81aded22012-06-15 14:54:11 -07001363
Martin KaFai Lau0d3f6d22015-11-11 11:51:06 -08001364 if (!rt6_cache_allowed_for_pmtu(rt6)) {
Martin KaFai Lau45e4fd22015-05-22 20:56:00 -07001365 rt6_do_update_pmtu(rt6, mtu);
1366 } else {
1367 const struct in6_addr *daddr, *saddr;
1368 struct rt6_info *nrt6;
Hagen Paul Pfeifer9d289712015-01-15 22:34:25 +01001369
Martin KaFai Lau45e4fd22015-05-22 20:56:00 -07001370 if (iph) {
1371 daddr = &iph->daddr;
1372 saddr = &iph->saddr;
1373 } else if (sk) {
1374 daddr = &sk->sk_v6_daddr;
1375 saddr = &inet6_sk(sk)->saddr;
1376 } else {
1377 return;
1378 }
1379 nrt6 = ip6_rt_cache_alloc(rt6, daddr, saddr);
1380 if (nrt6) {
1381 rt6_do_update_pmtu(nrt6, mtu);
1382
1383 /* ip6_ins_rt(nrt6) will bump the
1384 * rt6->rt6i_node->fn_sernum
1385 * which will fail the next rt6_check() and
1386 * invalidate the sk->sk_dst_cache.
1387 */
1388 ip6_ins_rt(nrt6);
1389 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001390 }
1391}
1392
Martin KaFai Lau45e4fd22015-05-22 20:56:00 -07001393static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
1394 struct sk_buff *skb, u32 mtu)
1395{
1396 __ip6_rt_update_pmtu(dst, sk, skb ? ipv6_hdr(skb) : NULL, mtu);
1397}
1398
David S. Miller42ae66c2012-06-15 20:01:57 -07001399void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu,
1400 int oif, u32 mark)
David S. Miller81aded22012-06-15 14:54:11 -07001401{
1402 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1403 struct dst_entry *dst;
1404 struct flowi6 fl6;
1405
1406 memset(&fl6, 0, sizeof(fl6));
1407 fl6.flowi6_oif = oif;
Lorenzo Colitti1b3c61d2014-05-13 10:17:34 -07001408 fl6.flowi6_mark = mark ? mark : IP6_REPLY_MARK(net, skb->mark);
David S. Miller81aded22012-06-15 14:54:11 -07001409 fl6.daddr = iph->daddr;
1410 fl6.saddr = iph->saddr;
YOSHIFUJI Hideaki / 吉藤英明6502ca52013-01-13 05:01:51 +00001411 fl6.flowlabel = ip6_flowinfo(iph);
David S. Miller81aded22012-06-15 14:54:11 -07001412
1413 dst = ip6_route_output(net, NULL, &fl6);
1414 if (!dst->error)
Martin KaFai Lau45e4fd22015-05-22 20:56:00 -07001415 __ip6_rt_update_pmtu(dst, NULL, iph, ntohl(mtu));
David S. Miller81aded22012-06-15 14:54:11 -07001416 dst_release(dst);
1417}
1418EXPORT_SYMBOL_GPL(ip6_update_pmtu);
1419
1420void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu)
1421{
Martin KaFai Lau33c162a2016-04-11 15:29:36 -07001422 struct dst_entry *dst;
1423
David S. Miller81aded22012-06-15 14:54:11 -07001424 ip6_update_pmtu(skb, sock_net(sk), mtu,
1425 sk->sk_bound_dev_if, sk->sk_mark);
Martin KaFai Lau33c162a2016-04-11 15:29:36 -07001426
1427 dst = __sk_dst_get(sk);
1428 if (!dst || !dst->obsolete ||
1429 dst->ops->check(dst, inet6_sk(sk)->dst_cookie))
1430 return;
1431
1432 bh_lock_sock(sk);
1433 if (!sock_owned_by_user(sk) && !ipv6_addr_v4mapped(&sk->sk_v6_daddr))
1434 ip6_datagram_dst_update(sk, false);
1435 bh_unlock_sock(sk);
David S. Miller81aded22012-06-15 14:54:11 -07001436}
1437EXPORT_SYMBOL_GPL(ip6_sk_update_pmtu);
1438
Duan Jiongb55b76b2013-09-04 19:44:21 +08001439/* Handle redirects */
1440struct ip6rd_flowi {
1441 struct flowi6 fl6;
1442 struct in6_addr gateway;
1443};
1444
1445static struct rt6_info *__ip6_route_redirect(struct net *net,
1446 struct fib6_table *table,
1447 struct flowi6 *fl6,
1448 int flags)
1449{
1450 struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl6;
1451 struct rt6_info *rt;
1452 struct fib6_node *fn;
1453
1454 /* Get the "current" route for this destination and
1455 * check if the redirect has come from approriate router.
1456 *
1457 * RFC 4861 specifies that redirects should only be
1458 * accepted if they come from the nexthop to the target.
1459 * Due to the way the routes are chosen, this notion
1460 * is a bit fuzzy and one might need to check all possible
1461 * routes.
1462 */
1463
1464 read_lock_bh(&table->tb6_lock);
1465 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
1466restart:
1467 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1468 if (rt6_check_expired(rt))
1469 continue;
1470 if (rt->dst.error)
1471 break;
1472 if (!(rt->rt6i_flags & RTF_GATEWAY))
1473 continue;
1474 if (fl6->flowi6_oif != rt->dst.dev->ifindex)
1475 continue;
1476 if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway))
1477 continue;
1478 break;
1479 }
1480
1481 if (!rt)
1482 rt = net->ipv6.ip6_null_entry;
1483 else if (rt->dst.error) {
1484 rt = net->ipv6.ip6_null_entry;
Martin KaFai Laub0a1ba52015-01-20 19:16:02 -08001485 goto out;
1486 }
1487
1488 if (rt == net->ipv6.ip6_null_entry) {
Martin KaFai Laua3c00e42014-10-20 13:42:43 -07001489 fn = fib6_backtrack(fn, &fl6->saddr);
1490 if (fn)
1491 goto restart;
Duan Jiongb55b76b2013-09-04 19:44:21 +08001492 }
Martin KaFai Laua3c00e42014-10-20 13:42:43 -07001493
Martin KaFai Laub0a1ba52015-01-20 19:16:02 -08001494out:
Duan Jiongb55b76b2013-09-04 19:44:21 +08001495 dst_hold(&rt->dst);
1496
1497 read_unlock_bh(&table->tb6_lock);
1498
David Ahernb8115802015-11-19 12:24:22 -08001499 trace_fib6_table_lookup(net, rt, table->tb6_id, fl6);
Duan Jiongb55b76b2013-09-04 19:44:21 +08001500 return rt;
1501};
1502
1503static struct dst_entry *ip6_route_redirect(struct net *net,
1504 const struct flowi6 *fl6,
1505 const struct in6_addr *gateway)
1506{
1507 int flags = RT6_LOOKUP_F_HAS_SADDR;
1508 struct ip6rd_flowi rdfl;
1509
1510 rdfl.fl6 = *fl6;
1511 rdfl.gateway = *gateway;
1512
1513 return fib6_rule_lookup(net, &rdfl.fl6,
1514 flags, __ip6_route_redirect);
1515}
1516
David S. Miller3a5ad2e2012-07-12 00:08:07 -07001517void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark)
1518{
1519 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1520 struct dst_entry *dst;
1521 struct flowi6 fl6;
1522
1523 memset(&fl6, 0, sizeof(fl6));
Julian Anastasove374c612014-04-28 10:51:56 +03001524 fl6.flowi6_iif = LOOPBACK_IFINDEX;
David S. Miller3a5ad2e2012-07-12 00:08:07 -07001525 fl6.flowi6_oif = oif;
1526 fl6.flowi6_mark = mark;
David S. Miller3a5ad2e2012-07-12 00:08:07 -07001527 fl6.daddr = iph->daddr;
1528 fl6.saddr = iph->saddr;
YOSHIFUJI Hideaki / 吉藤英明6502ca52013-01-13 05:01:51 +00001529 fl6.flowlabel = ip6_flowinfo(iph);
David S. Miller3a5ad2e2012-07-12 00:08:07 -07001530
Duan Jiongb55b76b2013-09-04 19:44:21 +08001531 dst = ip6_route_redirect(net, &fl6, &ipv6_hdr(skb)->saddr);
1532 rt6_do_redirect(dst, NULL, skb);
David S. Miller3a5ad2e2012-07-12 00:08:07 -07001533 dst_release(dst);
1534}
1535EXPORT_SYMBOL_GPL(ip6_redirect);
1536
Duan Jiongc92a59e2013-08-22 12:07:35 +08001537void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif,
1538 u32 mark)
1539{
1540 const struct ipv6hdr *iph = ipv6_hdr(skb);
1541 const struct rd_msg *msg = (struct rd_msg *)icmp6_hdr(skb);
1542 struct dst_entry *dst;
1543 struct flowi6 fl6;
1544
1545 memset(&fl6, 0, sizeof(fl6));
Julian Anastasove374c612014-04-28 10:51:56 +03001546 fl6.flowi6_iif = LOOPBACK_IFINDEX;
Duan Jiongc92a59e2013-08-22 12:07:35 +08001547 fl6.flowi6_oif = oif;
1548 fl6.flowi6_mark = mark;
Duan Jiongc92a59e2013-08-22 12:07:35 +08001549 fl6.daddr = msg->dest;
1550 fl6.saddr = iph->daddr;
1551
Duan Jiongb55b76b2013-09-04 19:44:21 +08001552 dst = ip6_route_redirect(net, &fl6, &iph->saddr);
1553 rt6_do_redirect(dst, NULL, skb);
Duan Jiongc92a59e2013-08-22 12:07:35 +08001554 dst_release(dst);
1555}
1556
David S. Miller3a5ad2e2012-07-12 00:08:07 -07001557void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk)
1558{
1559 ip6_redirect(skb, sock_net(sk), sk->sk_bound_dev_if, sk->sk_mark);
1560}
1561EXPORT_SYMBOL_GPL(ip6_sk_redirect);
1562
David S. Miller0dbaee32010-12-13 12:52:14 -08001563static unsigned int ip6_default_advmss(const struct dst_entry *dst)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001564{
David S. Miller0dbaee32010-12-13 12:52:14 -08001565 struct net_device *dev = dst->dev;
1566 unsigned int mtu = dst_mtu(dst);
1567 struct net *net = dev_net(dev);
1568
Linus Torvalds1da177e2005-04-16 15:20:36 -07001569 mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr);
1570
Daniel Lezcano55786892008-03-04 13:47:47 -08001571 if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss)
1572 mtu = net->ipv6.sysctl.ip6_rt_min_advmss;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001573
1574 /*
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09001575 * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and
1576 * corresponding MSS is IPV6_MAXPLEN - tcp_header_size.
1577 * IPV6_MAXPLEN is also valid and means: "any MSS,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001578 * rely only on pmtu discovery"
1579 */
1580 if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr))
1581 mtu = IPV6_MAXPLEN;
1582 return mtu;
1583}
1584
Steffen Klassertebb762f2011-11-23 02:12:51 +00001585static unsigned int ip6_mtu(const struct dst_entry *dst)
David S. Millerd33e4552010-12-14 13:01:14 -08001586{
Martin KaFai Lau4b32b5a2015-04-28 13:03:06 -07001587 const struct rt6_info *rt = (const struct rt6_info *)dst;
1588 unsigned int mtu = rt->rt6i_pmtu;
David S. Millerd33e4552010-12-14 13:01:14 -08001589 struct inet6_dev *idev;
Steffen Klassert618f9bc2011-11-23 02:13:31 +00001590
1591 if (mtu)
Eric Dumazet30f78d82014-04-10 21:23:36 -07001592 goto out;
Steffen Klassert618f9bc2011-11-23 02:13:31 +00001593
Martin KaFai Lau4b32b5a2015-04-28 13:03:06 -07001594 mtu = dst_metric_raw(dst, RTAX_MTU);
1595 if (mtu)
1596 goto out;
1597
Steffen Klassert618f9bc2011-11-23 02:13:31 +00001598 mtu = IPV6_MIN_MTU;
David S. Millerd33e4552010-12-14 13:01:14 -08001599
1600 rcu_read_lock();
1601 idev = __in6_dev_get(dst->dev);
1602 if (idev)
1603 mtu = idev->cnf.mtu6;
1604 rcu_read_unlock();
1605
Eric Dumazet30f78d82014-04-10 21:23:36 -07001606out:
1607 return min_t(unsigned int, mtu, IP6_MAX_MTU);
David S. Millerd33e4552010-12-14 13:01:14 -08001608}
1609
YOSHIFUJI Hideaki3b009442007-12-06 16:11:48 -08001610static struct dst_entry *icmp6_dst_gc_list;
1611static DEFINE_SPINLOCK(icmp6_dst_lock);
Thomas Graf5d0bbee2006-08-04 03:37:36 -07001612
YOSHIFUJI Hideaki3b009442007-12-06 16:11:48 -08001613struct dst_entry *icmp6_dst_alloc(struct net_device *dev,
David S. Miller87a11572011-12-06 17:04:13 -05001614 struct flowi6 *fl6)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001615{
David S. Miller87a11572011-12-06 17:04:13 -05001616 struct dst_entry *dst;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001617 struct rt6_info *rt;
1618 struct inet6_dev *idev = in6_dev_get(dev);
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09001619 struct net *net = dev_net(dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001620
David S. Miller38308472011-12-03 18:02:47 -05001621 if (unlikely(!idev))
Eric Dumazet122bdf62012-03-14 21:13:11 +00001622 return ERR_PTR(-ENODEV);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001623
Martin KaFai Lauad706862015-08-14 11:05:52 -07001624 rt = ip6_dst_alloc(net, dev, 0);
David S. Miller38308472011-12-03 18:02:47 -05001625 if (unlikely(!rt)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001626 in6_dev_put(idev);
David S. Miller87a11572011-12-06 17:04:13 -05001627 dst = ERR_PTR(-ENOMEM);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001628 goto out;
1629 }
1630
Yan, Zheng8e2ec632011-09-05 21:34:30 +00001631 rt->dst.flags |= DST_HOST;
1632 rt->dst.output = ip6_output;
Changli Gaod8d1f302010-06-10 23:31:35 -07001633 atomic_set(&rt->dst.__refcnt, 1);
Julian Anastasov550bab42013-10-20 15:43:04 +03001634 rt->rt6i_gateway = fl6->daddr;
David S. Miller87a11572011-12-06 17:04:13 -05001635 rt->rt6i_dst.addr = fl6->daddr;
Yan, Zheng8e2ec632011-09-05 21:34:30 +00001636 rt->rt6i_dst.plen = 128;
1637 rt->rt6i_idev = idev;
Li RongQing14edd872012-10-24 14:01:18 +08001638 dst_metric_set(&rt->dst, RTAX_HOPLIMIT, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001639
YOSHIFUJI Hideaki3b009442007-12-06 16:11:48 -08001640 spin_lock_bh(&icmp6_dst_lock);
Changli Gaod8d1f302010-06-10 23:31:35 -07001641 rt->dst.next = icmp6_dst_gc_list;
1642 icmp6_dst_gc_list = &rt->dst;
YOSHIFUJI Hideaki3b009442007-12-06 16:11:48 -08001643 spin_unlock_bh(&icmp6_dst_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001644
Daniel Lezcano55786892008-03-04 13:47:47 -08001645 fib6_force_start_gc(net);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001646
David S. Miller87a11572011-12-06 17:04:13 -05001647 dst = xfrm_lookup(net, &rt->dst, flowi6_to_flowi(fl6), NULL, 0);
1648
Linus Torvalds1da177e2005-04-16 15:20:36 -07001649out:
David S. Miller87a11572011-12-06 17:04:13 -05001650 return dst;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001651}
1652
Stephen Hemminger3d0f24a2008-07-22 14:35:50 -07001653int icmp6_dst_gc(void)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001654{
Hagen Paul Pfeifere9476e92011-02-25 05:45:19 +00001655 struct dst_entry *dst, **pprev;
Stephen Hemminger3d0f24a2008-07-22 14:35:50 -07001656 int more = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001657
YOSHIFUJI Hideaki3b009442007-12-06 16:11:48 -08001658 spin_lock_bh(&icmp6_dst_lock);
1659 pprev = &icmp6_dst_gc_list;
Thomas Graf5d0bbee2006-08-04 03:37:36 -07001660
Linus Torvalds1da177e2005-04-16 15:20:36 -07001661 while ((dst = *pprev) != NULL) {
1662 if (!atomic_read(&dst->__refcnt)) {
1663 *pprev = dst->next;
1664 dst_free(dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001665 } else {
1666 pprev = &dst->next;
Stephen Hemminger3d0f24a2008-07-22 14:35:50 -07001667 ++more;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001668 }
1669 }
1670
YOSHIFUJI Hideaki3b009442007-12-06 16:11:48 -08001671 spin_unlock_bh(&icmp6_dst_lock);
Thomas Graf5d0bbee2006-08-04 03:37:36 -07001672
Stephen Hemminger3d0f24a2008-07-22 14:35:50 -07001673 return more;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001674}
1675
David S. Miller1e493d12008-09-10 17:27:15 -07001676static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg),
1677 void *arg)
1678{
1679 struct dst_entry *dst, **pprev;
1680
1681 spin_lock_bh(&icmp6_dst_lock);
1682 pprev = &icmp6_dst_gc_list;
1683 while ((dst = *pprev) != NULL) {
1684 struct rt6_info *rt = (struct rt6_info *) dst;
1685 if (func(rt, arg)) {
1686 *pprev = dst->next;
1687 dst_free(dst);
1688 } else {
1689 pprev = &dst->next;
1690 }
1691 }
1692 spin_unlock_bh(&icmp6_dst_lock);
1693}
1694
Daniel Lezcano569d3642008-01-18 03:56:57 -08001695static int ip6_dst_gc(struct dst_ops *ops)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001696{
Alexey Dobriyan86393e52009-08-29 01:34:49 +00001697 struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops);
Daniel Lezcano7019b782008-03-04 13:50:14 -08001698 int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval;
1699 int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size;
1700 int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity;
1701 int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout;
1702 unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc;
Eric Dumazetfc66f952010-10-08 06:37:34 +00001703 int entries;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001704
Eric Dumazetfc66f952010-10-08 06:37:34 +00001705 entries = dst_entries_get_fast(ops);
Michal Kubeček49a18d82013-08-01 10:04:24 +02001706 if (time_after(rt_last_gc + rt_min_interval, jiffies) &&
Eric Dumazetfc66f952010-10-08 06:37:34 +00001707 entries <= rt_max_size)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001708 goto out;
1709
Benjamin Thery6891a342008-03-04 13:49:47 -08001710 net->ipv6.ip6_rt_gc_expire++;
Li RongQing14956642014-05-19 17:30:28 +08001711 fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net, true);
Eric Dumazetfc66f952010-10-08 06:37:34 +00001712 entries = dst_entries_get_slow(ops);
1713 if (entries < ops->gc_thresh)
Daniel Lezcano7019b782008-03-04 13:50:14 -08001714 net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001715out:
Daniel Lezcano7019b782008-03-04 13:50:14 -08001716 net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity;
Eric Dumazetfc66f952010-10-08 06:37:34 +00001717 return entries > rt_max_size;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001718}
1719
Florian Westphale715b6d2015-01-05 23:57:44 +01001720static int ip6_convert_metrics(struct mx6_config *mxc,
1721 const struct fib6_config *cfg)
1722{
Daniel Borkmannc3a8d942015-08-31 15:58:47 +02001723 bool ecn_ca = false;
Florian Westphale715b6d2015-01-05 23:57:44 +01001724 struct nlattr *nla;
1725 int remaining;
1726 u32 *mp;
1727
Ian Morris63159f22015-03-29 14:00:04 +01001728 if (!cfg->fc_mx)
Florian Westphale715b6d2015-01-05 23:57:44 +01001729 return 0;
1730
1731 mp = kzalloc(sizeof(u32) * RTAX_MAX, GFP_KERNEL);
1732 if (unlikely(!mp))
1733 return -ENOMEM;
1734
1735 nla_for_each_attr(nla, cfg->fc_mx, cfg->fc_mx_len, remaining) {
1736 int type = nla_type(nla);
Daniel Borkmann1bb14802015-08-31 15:58:45 +02001737 u32 val;
Florian Westphale715b6d2015-01-05 23:57:44 +01001738
Daniel Borkmann1bb14802015-08-31 15:58:45 +02001739 if (!type)
1740 continue;
1741 if (unlikely(type > RTAX_MAX))
1742 goto err;
Daniel Borkmannea697632015-01-05 23:57:47 +01001743
Daniel Borkmann1bb14802015-08-31 15:58:45 +02001744 if (type == RTAX_CC_ALGO) {
1745 char tmp[TCP_CA_NAME_MAX];
1746
1747 nla_strlcpy(tmp, nla, sizeof(tmp));
Daniel Borkmannc3a8d942015-08-31 15:58:47 +02001748 val = tcp_ca_get_key_by_name(tmp, &ecn_ca);
Daniel Borkmann1bb14802015-08-31 15:58:45 +02001749 if (val == TCP_CA_UNSPEC)
Florian Westphale715b6d2015-01-05 23:57:44 +01001750 goto err;
Daniel Borkmann1bb14802015-08-31 15:58:45 +02001751 } else {
1752 val = nla_get_u32(nla);
Florian Westphale715b6d2015-01-05 23:57:44 +01001753 }
Paolo Abeni626abd52016-05-13 18:33:41 +02001754 if (type == RTAX_HOPLIMIT && val > 255)
1755 val = 255;
Daniel Borkmannb8d3e412015-08-31 15:58:46 +02001756 if (type == RTAX_FEATURES && (val & ~RTAX_FEATURE_MASK))
1757 goto err;
Daniel Borkmann1bb14802015-08-31 15:58:45 +02001758
1759 mp[type - 1] = val;
1760 __set_bit(type - 1, mxc->mx_valid);
Florian Westphale715b6d2015-01-05 23:57:44 +01001761 }
1762
Daniel Borkmannc3a8d942015-08-31 15:58:47 +02001763 if (ecn_ca) {
1764 __set_bit(RTAX_FEATURES - 1, mxc->mx_valid);
1765 mp[RTAX_FEATURES - 1] |= DST_FEATURE_ECN_CA;
1766 }
Florian Westphale715b6d2015-01-05 23:57:44 +01001767
Daniel Borkmannc3a8d942015-08-31 15:58:47 +02001768 mxc->mx = mp;
Florian Westphale715b6d2015-01-05 23:57:44 +01001769 return 0;
1770 err:
1771 kfree(mp);
1772 return -EINVAL;
1773}
Linus Torvalds1da177e2005-04-16 15:20:36 -07001774
David Ahern8c145862016-04-24 21:26:04 -07001775static struct rt6_info *ip6_nh_lookup_table(struct net *net,
1776 struct fib6_config *cfg,
1777 const struct in6_addr *gw_addr)
1778{
1779 struct flowi6 fl6 = {
1780 .flowi6_oif = cfg->fc_ifindex,
1781 .daddr = *gw_addr,
1782 .saddr = cfg->fc_prefsrc,
1783 };
1784 struct fib6_table *table;
1785 struct rt6_info *rt;
Paolo Abeni48f1dcb2016-06-23 15:25:09 +02001786 int flags = RT6_LOOKUP_F_IFACE;
David Ahern8c145862016-04-24 21:26:04 -07001787
1788 table = fib6_get_table(net, cfg->fc_table);
1789 if (!table)
1790 return NULL;
1791
1792 if (!ipv6_addr_any(&cfg->fc_prefsrc))
1793 flags |= RT6_LOOKUP_F_HAS_SADDR;
1794
1795 rt = ip6_pol_route(net, table, cfg->fc_ifindex, &fl6, flags);
1796
1797 /* if table lookup failed, fall back to full lookup */
1798 if (rt == net->ipv6.ip6_null_entry) {
1799 ip6_rt_put(rt);
1800 rt = NULL;
1801 }
1802
1803 return rt;
1804}
1805
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07001806static struct rt6_info *ip6_route_info_create(struct fib6_config *cfg)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001807{
Daniel Lezcano55786892008-03-04 13:47:47 -08001808 struct net *net = cfg->fc_nlinfo.nl_net;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001809 struct rt6_info *rt = NULL;
1810 struct net_device *dev = NULL;
1811 struct inet6_dev *idev = NULL;
Thomas Grafc71099a2006-08-04 23:20:06 -07001812 struct fib6_table *table;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001813 int addr_type;
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07001814 int err = -EINVAL;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001815
Thomas Graf86872cb2006-08-22 00:01:08 -07001816 if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128)
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07001817 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001818#ifndef CONFIG_IPV6_SUBTREES
Thomas Graf86872cb2006-08-22 00:01:08 -07001819 if (cfg->fc_src_len)
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07001820 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001821#endif
Thomas Graf86872cb2006-08-22 00:01:08 -07001822 if (cfg->fc_ifindex) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001823 err = -ENODEV;
Daniel Lezcano55786892008-03-04 13:47:47 -08001824 dev = dev_get_by_index(net, cfg->fc_ifindex);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001825 if (!dev)
1826 goto out;
1827 idev = in6_dev_get(dev);
1828 if (!idev)
1829 goto out;
1830 }
1831
Thomas Graf86872cb2006-08-22 00:01:08 -07001832 if (cfg->fc_metric == 0)
1833 cfg->fc_metric = IP6_RT_PRIO_USER;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001834
Matti Vaittinend71314b2011-11-14 00:14:49 +00001835 err = -ENOBUFS;
David S. Miller38308472011-12-03 18:02:47 -05001836 if (cfg->fc_nlinfo.nlh &&
1837 !(cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_CREATE)) {
Matti Vaittinend71314b2011-11-14 00:14:49 +00001838 table = fib6_get_table(net, cfg->fc_table);
David S. Miller38308472011-12-03 18:02:47 -05001839 if (!table) {
Joe Perchesf3213832012-05-15 14:11:53 +00001840 pr_warn("NLM_F_CREATE should be specified when creating new route\n");
Matti Vaittinend71314b2011-11-14 00:14:49 +00001841 table = fib6_new_table(net, cfg->fc_table);
1842 }
1843 } else {
1844 table = fib6_new_table(net, cfg->fc_table);
1845 }
David S. Miller38308472011-12-03 18:02:47 -05001846
1847 if (!table)
Thomas Grafc71099a2006-08-04 23:20:06 -07001848 goto out;
Thomas Grafc71099a2006-08-04 23:20:06 -07001849
Martin KaFai Lauad706862015-08-14 11:05:52 -07001850 rt = ip6_dst_alloc(net, NULL,
1851 (cfg->fc_flags & RTF_ADDRCONF) ? 0 : DST_NOCOUNT);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001852
David S. Miller38308472011-12-03 18:02:47 -05001853 if (!rt) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001854 err = -ENOMEM;
1855 goto out;
1856 }
1857
Gao feng1716a962012-04-06 00:13:10 +00001858 if (cfg->fc_flags & RTF_EXPIRES)
1859 rt6_set_expires(rt, jiffies +
1860 clock_t_to_jiffies(cfg->fc_expires));
1861 else
1862 rt6_clean_expires(rt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001863
Thomas Graf86872cb2006-08-22 00:01:08 -07001864 if (cfg->fc_protocol == RTPROT_UNSPEC)
1865 cfg->fc_protocol = RTPROT_BOOT;
1866 rt->rt6i_protocol = cfg->fc_protocol;
1867
1868 addr_type = ipv6_addr_type(&cfg->fc_dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001869
1870 if (addr_type & IPV6_ADDR_MULTICAST)
Changli Gaod8d1f302010-06-10 23:31:35 -07001871 rt->dst.input = ip6_mc_input;
Maciej Żenczykowskiab79ad12010-09-27 00:07:02 +00001872 else if (cfg->fc_flags & RTF_LOCAL)
1873 rt->dst.input = ip6_input;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001874 else
Changli Gaod8d1f302010-06-10 23:31:35 -07001875 rt->dst.input = ip6_forward;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001876
Changli Gaod8d1f302010-06-10 23:31:35 -07001877 rt->dst.output = ip6_output;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001878
Roopa Prabhu19e42e42015-07-21 10:43:48 +02001879 if (cfg->fc_encap) {
1880 struct lwtunnel_state *lwtstate;
1881
1882 err = lwtunnel_build_state(dev, cfg->fc_encap_type,
Tom Herbert127eb7c2015-08-24 09:45:41 -07001883 cfg->fc_encap, AF_INET6, cfg,
1884 &lwtstate);
Roopa Prabhu19e42e42015-07-21 10:43:48 +02001885 if (err)
1886 goto out;
Jiri Benc61adedf2015-08-20 13:56:25 +02001887 rt->dst.lwtstate = lwtstate_get(lwtstate);
1888 if (lwtunnel_output_redirect(rt->dst.lwtstate)) {
1889 rt->dst.lwtstate->orig_output = rt->dst.output;
1890 rt->dst.output = lwtunnel_output;
Tom Herbert25368622015-08-17 13:42:24 -07001891 }
Jiri Benc61adedf2015-08-20 13:56:25 +02001892 if (lwtunnel_input_redirect(rt->dst.lwtstate)) {
1893 rt->dst.lwtstate->orig_input = rt->dst.input;
1894 rt->dst.input = lwtunnel_input;
Tom Herbert25368622015-08-17 13:42:24 -07001895 }
Roopa Prabhu19e42e42015-07-21 10:43:48 +02001896 }
1897
Thomas Graf86872cb2006-08-22 00:01:08 -07001898 ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len);
1899 rt->rt6i_dst.plen = cfg->fc_dst_len;
Martin KaFai Lauafc4eef2015-04-28 13:03:07 -07001900 if (rt->rt6i_dst.plen == 128)
Michal Kubečeke5fd3872014-03-27 13:04:08 +01001901 rt->dst.flags |= DST_HOST;
Michal Kubečeke5fd3872014-03-27 13:04:08 +01001902
Linus Torvalds1da177e2005-04-16 15:20:36 -07001903#ifdef CONFIG_IPV6_SUBTREES
Thomas Graf86872cb2006-08-22 00:01:08 -07001904 ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len);
1905 rt->rt6i_src.plen = cfg->fc_src_len;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001906#endif
1907
Thomas Graf86872cb2006-08-22 00:01:08 -07001908 rt->rt6i_metric = cfg->fc_metric;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001909
1910 /* We cannot add true routes via loopback here,
1911 they would result in kernel looping; promote them to reject routes
1912 */
Thomas Graf86872cb2006-08-22 00:01:08 -07001913 if ((cfg->fc_flags & RTF_REJECT) ||
David S. Miller38308472011-12-03 18:02:47 -05001914 (dev && (dev->flags & IFF_LOOPBACK) &&
1915 !(addr_type & IPV6_ADDR_LOOPBACK) &&
1916 !(cfg->fc_flags & RTF_LOCAL))) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001917 /* hold loopback dev/idev if we haven't done so. */
Daniel Lezcano55786892008-03-04 13:47:47 -08001918 if (dev != net->loopback_dev) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001919 if (dev) {
1920 dev_put(dev);
1921 in6_dev_put(idev);
1922 }
Daniel Lezcano55786892008-03-04 13:47:47 -08001923 dev = net->loopback_dev;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001924 dev_hold(dev);
1925 idev = in6_dev_get(dev);
1926 if (!idev) {
1927 err = -ENODEV;
1928 goto out;
1929 }
1930 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001931 rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP;
Nicolas Dichtelef2c7d72012-09-05 02:12:42 +00001932 switch (cfg->fc_type) {
1933 case RTN_BLACKHOLE:
1934 rt->dst.error = -EINVAL;
Eric W. Biedermanede20592015-10-07 16:48:47 -05001935 rt->dst.output = dst_discard_out;
Kamala R7150aed2013-12-02 19:55:21 +05301936 rt->dst.input = dst_discard;
Nicolas Dichtelef2c7d72012-09-05 02:12:42 +00001937 break;
1938 case RTN_PROHIBIT:
1939 rt->dst.error = -EACCES;
Kamala R7150aed2013-12-02 19:55:21 +05301940 rt->dst.output = ip6_pkt_prohibit_out;
1941 rt->dst.input = ip6_pkt_prohibit;
Nicolas Dichtelef2c7d72012-09-05 02:12:42 +00001942 break;
Nicolas Dichtelb4949ab2012-09-06 05:53:35 +00001943 case RTN_THROW:
Nikola Forró0315e382015-09-17 16:01:32 +02001944 case RTN_UNREACHABLE:
Nicolas Dichtelef2c7d72012-09-05 02:12:42 +00001945 default:
Kamala R7150aed2013-12-02 19:55:21 +05301946 rt->dst.error = (cfg->fc_type == RTN_THROW) ? -EAGAIN
Nikola Forró0315e382015-09-17 16:01:32 +02001947 : (cfg->fc_type == RTN_UNREACHABLE)
1948 ? -EHOSTUNREACH : -ENETUNREACH;
Kamala R7150aed2013-12-02 19:55:21 +05301949 rt->dst.output = ip6_pkt_discard_out;
1950 rt->dst.input = ip6_pkt_discard;
Nicolas Dichtelef2c7d72012-09-05 02:12:42 +00001951 break;
1952 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001953 goto install_route;
1954 }
1955
Thomas Graf86872cb2006-08-22 00:01:08 -07001956 if (cfg->fc_flags & RTF_GATEWAY) {
Eric Dumazetb71d1d42011-04-22 04:53:02 +00001957 const struct in6_addr *gw_addr;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001958 int gwa_type;
1959
Thomas Graf86872cb2006-08-22 00:01:08 -07001960 gw_addr = &cfg->fc_gateway;
Florian Westphal330567b2015-08-07 10:54:28 +02001961 gwa_type = ipv6_addr_type(gw_addr);
Florian Westphal48ed7b22015-05-21 00:25:41 +02001962
1963 /* if gw_addr is local we will fail to detect this in case
1964 * address is still TENTATIVE (DAD in progress). rt6_lookup()
1965 * will return already-added prefix route via interface that
1966 * prefix route was assigned to, which might be non-loopback.
1967 */
1968 err = -EINVAL;
Florian Westphal330567b2015-08-07 10:54:28 +02001969 if (ipv6_chk_addr_and_flags(net, gw_addr,
1970 gwa_type & IPV6_ADDR_LINKLOCAL ?
1971 dev : NULL, 0, 0))
Florian Westphal48ed7b22015-05-21 00:25:41 +02001972 goto out;
1973
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00001974 rt->rt6i_gateway = *gw_addr;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001975
1976 if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) {
David Ahern8c145862016-04-24 21:26:04 -07001977 struct rt6_info *grt = NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001978
1979 /* IPv6 strictly inhibits using not link-local
1980 addresses as nexthop address.
1981 Otherwise, router will not able to send redirects.
1982 It is very good, but in some (rare!) circumstances
1983 (SIT, PtP, NBMA NOARP links) it is handy to allow
1984 some exceptions. --ANK
1985 */
David S. Miller38308472011-12-03 18:02:47 -05001986 if (!(gwa_type & IPV6_ADDR_UNICAST))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001987 goto out;
1988
David Ahern8c145862016-04-24 21:26:04 -07001989 if (cfg->fc_table)
1990 grt = ip6_nh_lookup_table(net, cfg, gw_addr);
1991
1992 if (!grt)
1993 grt = rt6_lookup(net, gw_addr, NULL,
1994 cfg->fc_ifindex, 1);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001995
1996 err = -EHOSTUNREACH;
David S. Miller38308472011-12-03 18:02:47 -05001997 if (!grt)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001998 goto out;
1999 if (dev) {
David S. Millerd1918542011-12-28 20:19:20 -05002000 if (dev != grt->dst.dev) {
Amerigo Wang94e187c2012-10-29 00:13:19 +00002001 ip6_rt_put(grt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002002 goto out;
2003 }
2004 } else {
David S. Millerd1918542011-12-28 20:19:20 -05002005 dev = grt->dst.dev;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002006 idev = grt->rt6i_idev;
2007 dev_hold(dev);
2008 in6_dev_hold(grt->rt6i_idev);
2009 }
David S. Miller38308472011-12-03 18:02:47 -05002010 if (!(grt->rt6i_flags & RTF_GATEWAY))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002011 err = 0;
Amerigo Wang94e187c2012-10-29 00:13:19 +00002012 ip6_rt_put(grt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002013
2014 if (err)
2015 goto out;
2016 }
2017 err = -EINVAL;
David S. Miller38308472011-12-03 18:02:47 -05002018 if (!dev || (dev->flags & IFF_LOOPBACK))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002019 goto out;
2020 }
2021
2022 err = -ENODEV;
David S. Miller38308472011-12-03 18:02:47 -05002023 if (!dev)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002024 goto out;
2025
Daniel Walterc3968a82011-04-13 21:10:57 +00002026 if (!ipv6_addr_any(&cfg->fc_prefsrc)) {
2027 if (!ipv6_chk_addr(net, &cfg->fc_prefsrc, dev, 0)) {
2028 err = -EINVAL;
2029 goto out;
2030 }
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00002031 rt->rt6i_prefsrc.addr = cfg->fc_prefsrc;
Daniel Walterc3968a82011-04-13 21:10:57 +00002032 rt->rt6i_prefsrc.plen = 128;
2033 } else
2034 rt->rt6i_prefsrc.plen = 0;
2035
Thomas Graf86872cb2006-08-22 00:01:08 -07002036 rt->rt6i_flags = cfg->fc_flags;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002037
2038install_route:
Changli Gaod8d1f302010-06-10 23:31:35 -07002039 rt->dst.dev = dev;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002040 rt->rt6i_idev = idev;
Thomas Grafc71099a2006-08-04 23:20:06 -07002041 rt->rt6i_table = table;
Daniel Lezcano63152fc2008-03-03 23:31:11 -08002042
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09002043 cfg->fc_nlinfo.nl_net = dev_net(dev);
Daniel Lezcano63152fc2008-03-03 23:31:11 -08002044
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07002045 return rt;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002046out:
2047 if (dev)
2048 dev_put(dev);
2049 if (idev)
2050 in6_dev_put(idev);
2051 if (rt)
Changli Gaod8d1f302010-06-10 23:31:35 -07002052 dst_free(&rt->dst);
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002053
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07002054 return ERR_PTR(err);
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002055}
2056
2057int ip6_route_add(struct fib6_config *cfg)
2058{
2059 struct mx6_config mxc = { .mx = NULL, };
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07002060 struct rt6_info *rt;
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002061 int err;
2062
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07002063 rt = ip6_route_info_create(cfg);
2064 if (IS_ERR(rt)) {
2065 err = PTR_ERR(rt);
2066 rt = NULL;
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002067 goto out;
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07002068 }
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002069
2070 err = ip6_convert_metrics(&mxc, cfg);
2071 if (err)
2072 goto out;
2073
2074 err = __ip6_ins_rt(rt, &cfg->fc_nlinfo, &mxc);
2075
2076 kfree(mxc.mx);
2077
2078 return err;
2079out:
2080 if (rt)
2081 dst_free(&rt->dst);
2082
Linus Torvalds1da177e2005-04-16 15:20:36 -07002083 return err;
2084}
2085
Thomas Graf86872cb2006-08-22 00:01:08 -07002086static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002087{
2088 int err;
Thomas Grafc71099a2006-08-04 23:20:06 -07002089 struct fib6_table *table;
David S. Millerd1918542011-12-28 20:19:20 -05002090 struct net *net = dev_net(rt->dst.dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002091
Martin KaFai Lau8e3d5be2015-09-15 14:30:08 -07002092 if (rt == net->ipv6.ip6_null_entry ||
2093 rt->dst.flags & DST_NOCACHE) {
Gao feng6825a262012-09-19 19:25:34 +00002094 err = -ENOENT;
2095 goto out;
2096 }
Patrick McHardy6c813a72006-08-06 22:22:47 -07002097
Thomas Grafc71099a2006-08-04 23:20:06 -07002098 table = rt->rt6i_table;
2099 write_lock_bh(&table->tb6_lock);
Thomas Graf86872cb2006-08-22 00:01:08 -07002100 err = fib6_del(rt, info);
Thomas Grafc71099a2006-08-04 23:20:06 -07002101 write_unlock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002102
Gao feng6825a262012-09-19 19:25:34 +00002103out:
Amerigo Wang94e187c2012-10-29 00:13:19 +00002104 ip6_rt_put(rt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002105 return err;
2106}
2107
Thomas Grafe0a1ad732006-08-22 00:00:21 -07002108int ip6_del_rt(struct rt6_info *rt)
2109{
Denis V. Lunev4d1169c2008-01-10 03:26:13 -08002110 struct nl_info info = {
David S. Millerd1918542011-12-28 20:19:20 -05002111 .nl_net = dev_net(rt->dst.dev),
Denis V. Lunev4d1169c2008-01-10 03:26:13 -08002112 };
Denis V. Lunev528c4ce2007-12-13 09:45:12 -08002113 return __ip6_del_rt(rt, &info);
Thomas Grafe0a1ad732006-08-22 00:00:21 -07002114}
2115
Thomas Graf86872cb2006-08-22 00:01:08 -07002116static int ip6_route_del(struct fib6_config *cfg)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002117{
Thomas Grafc71099a2006-08-04 23:20:06 -07002118 struct fib6_table *table;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002119 struct fib6_node *fn;
2120 struct rt6_info *rt;
2121 int err = -ESRCH;
2122
Daniel Lezcano55786892008-03-04 13:47:47 -08002123 table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table);
David S. Miller38308472011-12-03 18:02:47 -05002124 if (!table)
Thomas Grafc71099a2006-08-04 23:20:06 -07002125 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002126
Thomas Grafc71099a2006-08-04 23:20:06 -07002127 read_lock_bh(&table->tb6_lock);
2128
2129 fn = fib6_locate(&table->tb6_root,
Thomas Graf86872cb2006-08-22 00:01:08 -07002130 &cfg->fc_dst, cfg->fc_dst_len,
2131 &cfg->fc_src, cfg->fc_src_len);
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09002132
Linus Torvalds1da177e2005-04-16 15:20:36 -07002133 if (fn) {
Changli Gaod8d1f302010-06-10 23:31:35 -07002134 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
Martin KaFai Lau1f56a012015-04-28 13:03:03 -07002135 if ((rt->rt6i_flags & RTF_CACHE) &&
2136 !(cfg->fc_flags & RTF_CACHE))
2137 continue;
Thomas Graf86872cb2006-08-22 00:01:08 -07002138 if (cfg->fc_ifindex &&
David S. Millerd1918542011-12-28 20:19:20 -05002139 (!rt->dst.dev ||
2140 rt->dst.dev->ifindex != cfg->fc_ifindex))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002141 continue;
Thomas Graf86872cb2006-08-22 00:01:08 -07002142 if (cfg->fc_flags & RTF_GATEWAY &&
2143 !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002144 continue;
Thomas Graf86872cb2006-08-22 00:01:08 -07002145 if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002146 continue;
Changli Gaod8d1f302010-06-10 23:31:35 -07002147 dst_hold(&rt->dst);
Thomas Grafc71099a2006-08-04 23:20:06 -07002148 read_unlock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002149
Thomas Graf86872cb2006-08-22 00:01:08 -07002150 return __ip6_del_rt(rt, &cfg->fc_nlinfo);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002151 }
2152 }
Thomas Grafc71099a2006-08-04 23:20:06 -07002153 read_unlock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002154
2155 return err;
2156}
2157
David S. Miller6700c272012-07-17 03:29:28 -07002158static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_buff *skb)
YOSHIFUJI Hideakia6279452006-08-23 17:18:26 -07002159{
YOSHIFUJI Hideakia6279452006-08-23 17:18:26 -07002160 struct netevent_redirect netevent;
David S. Millere8599ff2012-07-11 23:43:53 -07002161 struct rt6_info *rt, *nrt = NULL;
David S. Millere8599ff2012-07-11 23:43:53 -07002162 struct ndisc_options ndopts;
2163 struct inet6_dev *in6_dev;
2164 struct neighbour *neigh;
YOSHIFUJI Hideaki / 吉藤英明71bcdba2013-01-05 16:34:51 +00002165 struct rd_msg *msg;
David S. Miller6e157b62012-07-12 00:05:02 -07002166 int optlen, on_link;
2167 u8 *lladdr;
David S. Millere8599ff2012-07-11 23:43:53 -07002168
Simon Horman29a3cad2013-05-28 20:34:26 +00002169 optlen = skb_tail_pointer(skb) - skb_transport_header(skb);
YOSHIFUJI Hideaki / 吉藤英明71bcdba2013-01-05 16:34:51 +00002170 optlen -= sizeof(*msg);
David S. Millere8599ff2012-07-11 23:43:53 -07002171
2172 if (optlen < 0) {
David S. Miller6e157b62012-07-12 00:05:02 -07002173 net_dbg_ratelimited("rt6_do_redirect: packet too short\n");
David S. Millere8599ff2012-07-11 23:43:53 -07002174 return;
2175 }
2176
YOSHIFUJI Hideaki / 吉藤英明71bcdba2013-01-05 16:34:51 +00002177 msg = (struct rd_msg *)icmp6_hdr(skb);
David S. Millere8599ff2012-07-11 23:43:53 -07002178
YOSHIFUJI Hideaki / 吉藤英明71bcdba2013-01-05 16:34:51 +00002179 if (ipv6_addr_is_multicast(&msg->dest)) {
David S. Miller6e157b62012-07-12 00:05:02 -07002180 net_dbg_ratelimited("rt6_do_redirect: destination address is multicast\n");
David S. Millere8599ff2012-07-11 23:43:53 -07002181 return;
2182 }
2183
David S. Miller6e157b62012-07-12 00:05:02 -07002184 on_link = 0;
YOSHIFUJI Hideaki / 吉藤英明71bcdba2013-01-05 16:34:51 +00002185 if (ipv6_addr_equal(&msg->dest, &msg->target)) {
David S. Millere8599ff2012-07-11 23:43:53 -07002186 on_link = 1;
YOSHIFUJI Hideaki / 吉藤英明71bcdba2013-01-05 16:34:51 +00002187 } else if (ipv6_addr_type(&msg->target) !=
David S. Millere8599ff2012-07-11 23:43:53 -07002188 (IPV6_ADDR_UNICAST|IPV6_ADDR_LINKLOCAL)) {
David S. Miller6e157b62012-07-12 00:05:02 -07002189 net_dbg_ratelimited("rt6_do_redirect: target address is not link-local unicast\n");
David S. Millere8599ff2012-07-11 23:43:53 -07002190 return;
2191 }
2192
2193 in6_dev = __in6_dev_get(skb->dev);
2194 if (!in6_dev)
2195 return;
2196 if (in6_dev->cnf.forwarding || !in6_dev->cnf.accept_redirects)
2197 return;
2198
2199 /* RFC2461 8.1:
2200 * The IP source address of the Redirect MUST be the same as the current
2201 * first-hop router for the specified ICMP Destination Address.
2202 */
2203
Alexander Aringf997c552016-06-15 21:20:23 +02002204 if (!ndisc_parse_options(skb->dev, msg->opt, optlen, &ndopts)) {
David S. Millere8599ff2012-07-11 23:43:53 -07002205 net_dbg_ratelimited("rt6_redirect: invalid ND options\n");
2206 return;
2207 }
David S. Miller6e157b62012-07-12 00:05:02 -07002208
2209 lladdr = NULL;
David S. Millere8599ff2012-07-11 23:43:53 -07002210 if (ndopts.nd_opts_tgt_lladdr) {
2211 lladdr = ndisc_opt_addr_data(ndopts.nd_opts_tgt_lladdr,
2212 skb->dev);
2213 if (!lladdr) {
2214 net_dbg_ratelimited("rt6_redirect: invalid link-layer address length\n");
2215 return;
2216 }
2217 }
2218
David S. Miller6e157b62012-07-12 00:05:02 -07002219 rt = (struct rt6_info *) dst;
Matthias Schifferec13ad12015-11-02 01:24:38 +01002220 if (rt->rt6i_flags & RTF_REJECT) {
David S. Miller6e157b62012-07-12 00:05:02 -07002221 net_dbg_ratelimited("rt6_redirect: source isn't a valid nexthop for redirect target\n");
2222 return;
2223 }
2224
2225 /* Redirect received -> path was valid.
2226 * Look, redirects are sent only in response to data packets,
2227 * so that this nexthop apparently is reachable. --ANK
2228 */
2229 dst_confirm(&rt->dst);
2230
YOSHIFUJI Hideaki / 吉藤英明71bcdba2013-01-05 16:34:51 +00002231 neigh = __neigh_lookup(&nd_tbl, &msg->target, skb->dev, 1);
David S. Millere8599ff2012-07-11 23:43:53 -07002232 if (!neigh)
2233 return;
2234
Linus Torvalds1da177e2005-04-16 15:20:36 -07002235 /*
2236 * We have finally decided to accept it.
2237 */
2238
Alexander Aringf997c552016-06-15 21:20:23 +02002239 ndisc_update(skb->dev, neigh, lladdr, NUD_STALE,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002240 NEIGH_UPDATE_F_WEAK_OVERRIDE|
2241 NEIGH_UPDATE_F_OVERRIDE|
2242 (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER|
Alexander Aringf997c552016-06-15 21:20:23 +02002243 NEIGH_UPDATE_F_ISROUTER)),
2244 NDISC_REDIRECT, &ndopts);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002245
Martin KaFai Lau83a09ab2015-05-22 20:56:05 -07002246 nrt = ip6_rt_cache_alloc(rt, &msg->dest, NULL);
David S. Miller38308472011-12-03 18:02:47 -05002247 if (!nrt)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002248 goto out;
2249
2250 nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE;
2251 if (on_link)
2252 nrt->rt6i_flags &= ~RTF_GATEWAY;
2253
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00002254 nrt->rt6i_gateway = *(struct in6_addr *)neigh->primary_key;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002255
Thomas Graf40e22e82006-08-22 00:00:45 -07002256 if (ip6_ins_rt(nrt))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002257 goto out;
2258
Changli Gaod8d1f302010-06-10 23:31:35 -07002259 netevent.old = &rt->dst;
2260 netevent.new = &nrt->dst;
YOSHIFUJI Hideaki / 吉藤英明71bcdba2013-01-05 16:34:51 +00002261 netevent.daddr = &msg->dest;
YOSHIFUJI Hideaki / 吉藤英明60592832013-01-14 09:28:27 +00002262 netevent.neigh = neigh;
Tom Tucker8d717402006-07-30 20:43:36 -07002263 call_netevent_notifiers(NETEVENT_REDIRECT, &netevent);
2264
David S. Miller38308472011-12-03 18:02:47 -05002265 if (rt->rt6i_flags & RTF_CACHE) {
David S. Miller6e157b62012-07-12 00:05:02 -07002266 rt = (struct rt6_info *) dst_clone(&rt->dst);
Thomas Grafe0a1ad732006-08-22 00:00:21 -07002267 ip6_del_rt(rt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002268 }
2269
2270out:
David S. Millere8599ff2012-07-11 23:43:53 -07002271 neigh_release(neigh);
David S. Miller6e157b62012-07-12 00:05:02 -07002272}
2273
Linus Torvalds1da177e2005-04-16 15:20:36 -07002274/*
Linus Torvalds1da177e2005-04-16 15:20:36 -07002275 * Misc support functions
2276 */
2277
Martin KaFai Lau4b32b5a2015-04-28 13:03:06 -07002278static void rt6_set_from(struct rt6_info *rt, struct rt6_info *from)
2279{
2280 BUG_ON(from->dst.from);
2281
2282 rt->rt6i_flags &= ~RTF_EXPIRES;
2283 dst_hold(&from->dst);
2284 rt->dst.from = &from->dst;
2285 dst_init_metrics(&rt->dst, dst_metrics_ptr(&from->dst), true);
2286}
2287
Martin KaFai Lau83a09ab2015-05-22 20:56:05 -07002288static void ip6_rt_copy_init(struct rt6_info *rt, struct rt6_info *ort)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002289{
Martin KaFai Lau83a09ab2015-05-22 20:56:05 -07002290 rt->dst.input = ort->dst.input;
2291 rt->dst.output = ort->dst.output;
2292 rt->rt6i_dst = ort->rt6i_dst;
2293 rt->dst.error = ort->dst.error;
2294 rt->rt6i_idev = ort->rt6i_idev;
2295 if (rt->rt6i_idev)
2296 in6_dev_hold(rt->rt6i_idev);
2297 rt->dst.lastuse = jiffies;
2298 rt->rt6i_gateway = ort->rt6i_gateway;
2299 rt->rt6i_flags = ort->rt6i_flags;
2300 rt6_set_from(rt, ort);
2301 rt->rt6i_metric = ort->rt6i_metric;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002302#ifdef CONFIG_IPV6_SUBTREES
Martin KaFai Lau83a09ab2015-05-22 20:56:05 -07002303 rt->rt6i_src = ort->rt6i_src;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002304#endif
Martin KaFai Lau83a09ab2015-05-22 20:56:05 -07002305 rt->rt6i_prefsrc = ort->rt6i_prefsrc;
2306 rt->rt6i_table = ort->rt6i_table;
Jiri Benc61adedf2015-08-20 13:56:25 +02002307 rt->dst.lwtstate = lwtstate_get(ort->dst.lwtstate);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002308}
2309
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002310#ifdef CONFIG_IPV6_ROUTE_INFO
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -08002311static struct rt6_info *rt6_get_route_info(struct net *net,
Eric Dumazetb71d1d42011-04-22 04:53:02 +00002312 const struct in6_addr *prefix, int prefixlen,
2313 const struct in6_addr *gwaddr, int ifindex)
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002314{
2315 struct fib6_node *fn;
2316 struct rt6_info *rt = NULL;
Thomas Grafc71099a2006-08-04 23:20:06 -07002317 struct fib6_table *table;
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002318
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -08002319 table = fib6_get_table(net, RT6_TABLE_INFO);
David S. Miller38308472011-12-03 18:02:47 -05002320 if (!table)
Thomas Grafc71099a2006-08-04 23:20:06 -07002321 return NULL;
2322
Li RongQing5744dd92012-09-11 21:59:01 +00002323 read_lock_bh(&table->tb6_lock);
Ian Morris67ba4152014-08-24 21:53:10 +01002324 fn = fib6_locate(&table->tb6_root, prefix, prefixlen, NULL, 0);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002325 if (!fn)
2326 goto out;
2327
Changli Gaod8d1f302010-06-10 23:31:35 -07002328 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
David S. Millerd1918542011-12-28 20:19:20 -05002329 if (rt->dst.dev->ifindex != ifindex)
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002330 continue;
2331 if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY))
2332 continue;
2333 if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr))
2334 continue;
Changli Gaod8d1f302010-06-10 23:31:35 -07002335 dst_hold(&rt->dst);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002336 break;
2337 }
2338out:
Li RongQing5744dd92012-09-11 21:59:01 +00002339 read_unlock_bh(&table->tb6_lock);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002340 return rt;
2341}
2342
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -08002343static struct rt6_info *rt6_add_route_info(struct net *net,
Eric Dumazetb71d1d42011-04-22 04:53:02 +00002344 const struct in6_addr *prefix, int prefixlen,
2345 const struct in6_addr *gwaddr, int ifindex,
Eric Dumazet95c96172012-04-15 05:58:06 +00002346 unsigned int pref)
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002347{
Thomas Graf86872cb2006-08-22 00:01:08 -07002348 struct fib6_config cfg = {
Rami Rosen238fc7e2008-02-09 23:43:11 -08002349 .fc_metric = IP6_RT_PRIO_USER,
Thomas Graf86872cb2006-08-22 00:01:08 -07002350 .fc_ifindex = ifindex,
2351 .fc_dst_len = prefixlen,
2352 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO |
2353 RTF_UP | RTF_PREF(pref),
Eric W. Biederman15e47302012-09-07 20:12:54 +00002354 .fc_nlinfo.portid = 0,
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -08002355 .fc_nlinfo.nlh = NULL,
2356 .fc_nlinfo.nl_net = net,
Thomas Graf86872cb2006-08-22 00:01:08 -07002357 };
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002358
David Ahernca254492015-10-12 11:47:10 -07002359 cfg.fc_table = l3mdev_fib_table_by_index(net, ifindex) ? : RT6_TABLE_INFO;
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00002360 cfg.fc_dst = *prefix;
2361 cfg.fc_gateway = *gwaddr;
Thomas Graf86872cb2006-08-22 00:01:08 -07002362
YOSHIFUJI Hideakie317da92006-03-20 17:06:42 -08002363 /* We should treat it as a default route if prefix length is 0. */
2364 if (!prefixlen)
Thomas Graf86872cb2006-08-22 00:01:08 -07002365 cfg.fc_flags |= RTF_DEFAULT;
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002366
Thomas Graf86872cb2006-08-22 00:01:08 -07002367 ip6_route_add(&cfg);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002368
Daniel Lezcanoefa2cea2008-03-04 13:46:48 -08002369 return rt6_get_route_info(net, prefix, prefixlen, gwaddr, ifindex);
YOSHIFUJI Hideaki70ceb4f2006-03-20 17:06:24 -08002370}
2371#endif
2372
Eric Dumazetb71d1d42011-04-22 04:53:02 +00002373struct rt6_info *rt6_get_dflt_router(const struct in6_addr *addr, struct net_device *dev)
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09002374{
Linus Torvalds1da177e2005-04-16 15:20:36 -07002375 struct rt6_info *rt;
Thomas Grafc71099a2006-08-04 23:20:06 -07002376 struct fib6_table *table;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002377
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09002378 table = fib6_get_table(dev_net(dev), RT6_TABLE_DFLT);
David S. Miller38308472011-12-03 18:02:47 -05002379 if (!table)
Thomas Grafc71099a2006-08-04 23:20:06 -07002380 return NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002381
Li RongQing5744dd92012-09-11 21:59:01 +00002382 read_lock_bh(&table->tb6_lock);
Ian Morris67ba4152014-08-24 21:53:10 +01002383 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
David S. Millerd1918542011-12-28 20:19:20 -05002384 if (dev == rt->dst.dev &&
YOSHIFUJI Hideaki045927f2006-03-20 17:00:48 -08002385 ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) &&
Linus Torvalds1da177e2005-04-16 15:20:36 -07002386 ipv6_addr_equal(&rt->rt6i_gateway, addr))
2387 break;
2388 }
2389 if (rt)
Changli Gaod8d1f302010-06-10 23:31:35 -07002390 dst_hold(&rt->dst);
Li RongQing5744dd92012-09-11 21:59:01 +00002391 read_unlock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002392 return rt;
2393}
2394
Eric Dumazetb71d1d42011-04-22 04:53:02 +00002395struct rt6_info *rt6_add_dflt_router(const struct in6_addr *gwaddr,
YOSHIFUJI Hideakiebacaaa2006-03-20 17:04:53 -08002396 struct net_device *dev,
2397 unsigned int pref)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002398{
Thomas Graf86872cb2006-08-22 00:01:08 -07002399 struct fib6_config cfg = {
David Ahernca254492015-10-12 11:47:10 -07002400 .fc_table = l3mdev_fib_table(dev) ? : RT6_TABLE_DFLT,
Rami Rosen238fc7e2008-02-09 23:43:11 -08002401 .fc_metric = IP6_RT_PRIO_USER,
Thomas Graf86872cb2006-08-22 00:01:08 -07002402 .fc_ifindex = dev->ifindex,
2403 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT |
2404 RTF_UP | RTF_EXPIRES | RTF_PREF(pref),
Eric W. Biederman15e47302012-09-07 20:12:54 +00002405 .fc_nlinfo.portid = 0,
Daniel Lezcano55786892008-03-04 13:47:47 -08002406 .fc_nlinfo.nlh = NULL,
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09002407 .fc_nlinfo.nl_net = dev_net(dev),
Thomas Graf86872cb2006-08-22 00:01:08 -07002408 };
Linus Torvalds1da177e2005-04-16 15:20:36 -07002409
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00002410 cfg.fc_gateway = *gwaddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002411
Thomas Graf86872cb2006-08-22 00:01:08 -07002412 ip6_route_add(&cfg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002413
Linus Torvalds1da177e2005-04-16 15:20:36 -07002414 return rt6_get_dflt_router(gwaddr, dev);
2415}
2416
Daniel Lezcano7b4da532008-03-04 13:47:14 -08002417void rt6_purge_dflt_routers(struct net *net)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002418{
2419 struct rt6_info *rt;
Thomas Grafc71099a2006-08-04 23:20:06 -07002420 struct fib6_table *table;
2421
2422 /* NOTE: Keep consistent with rt6_get_dflt_router */
Daniel Lezcano7b4da532008-03-04 13:47:14 -08002423 table = fib6_get_table(net, RT6_TABLE_DFLT);
David S. Miller38308472011-12-03 18:02:47 -05002424 if (!table)
Thomas Grafc71099a2006-08-04 23:20:06 -07002425 return;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002426
2427restart:
Thomas Grafc71099a2006-08-04 23:20:06 -07002428 read_lock_bh(&table->tb6_lock);
Changli Gaod8d1f302010-06-10 23:31:35 -07002429 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
Lorenzo Colitti3e8b0ac2013-03-03 20:46:46 +00002430 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF) &&
2431 (!rt->rt6i_idev || rt->rt6i_idev->cnf.accept_ra != 2)) {
Changli Gaod8d1f302010-06-10 23:31:35 -07002432 dst_hold(&rt->dst);
Thomas Grafc71099a2006-08-04 23:20:06 -07002433 read_unlock_bh(&table->tb6_lock);
Thomas Grafe0a1ad732006-08-22 00:00:21 -07002434 ip6_del_rt(rt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002435 goto restart;
2436 }
2437 }
Thomas Grafc71099a2006-08-04 23:20:06 -07002438 read_unlock_bh(&table->tb6_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002439}
2440
Daniel Lezcano55786892008-03-04 13:47:47 -08002441static void rtmsg_to_fib6_config(struct net *net,
2442 struct in6_rtmsg *rtmsg,
Thomas Graf86872cb2006-08-22 00:01:08 -07002443 struct fib6_config *cfg)
2444{
2445 memset(cfg, 0, sizeof(*cfg));
2446
David Ahernca254492015-10-12 11:47:10 -07002447 cfg->fc_table = l3mdev_fib_table_by_index(net, rtmsg->rtmsg_ifindex) ?
2448 : RT6_TABLE_MAIN;
Thomas Graf86872cb2006-08-22 00:01:08 -07002449 cfg->fc_ifindex = rtmsg->rtmsg_ifindex;
2450 cfg->fc_metric = rtmsg->rtmsg_metric;
2451 cfg->fc_expires = rtmsg->rtmsg_info;
2452 cfg->fc_dst_len = rtmsg->rtmsg_dst_len;
2453 cfg->fc_src_len = rtmsg->rtmsg_src_len;
2454 cfg->fc_flags = rtmsg->rtmsg_flags;
2455
Daniel Lezcano55786892008-03-04 13:47:47 -08002456 cfg->fc_nlinfo.nl_net = net;
Benjamin Theryf1243c22008-02-26 18:10:03 -08002457
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00002458 cfg->fc_dst = rtmsg->rtmsg_dst;
2459 cfg->fc_src = rtmsg->rtmsg_src;
2460 cfg->fc_gateway = rtmsg->rtmsg_gateway;
Thomas Graf86872cb2006-08-22 00:01:08 -07002461}
2462
Daniel Lezcano55786892008-03-04 13:47:47 -08002463int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002464{
Thomas Graf86872cb2006-08-22 00:01:08 -07002465 struct fib6_config cfg;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002466 struct in6_rtmsg rtmsg;
2467 int err;
2468
Ian Morris67ba4152014-08-24 21:53:10 +01002469 switch (cmd) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002470 case SIOCADDRT: /* Add a route */
2471 case SIOCDELRT: /* Delete a route */
Eric W. Biedermanaf31f412012-11-16 03:03:06 +00002472 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002473 return -EPERM;
2474 err = copy_from_user(&rtmsg, arg,
2475 sizeof(struct in6_rtmsg));
2476 if (err)
2477 return -EFAULT;
Thomas Graf86872cb2006-08-22 00:01:08 -07002478
Daniel Lezcano55786892008-03-04 13:47:47 -08002479 rtmsg_to_fib6_config(net, &rtmsg, &cfg);
Thomas Graf86872cb2006-08-22 00:01:08 -07002480
Linus Torvalds1da177e2005-04-16 15:20:36 -07002481 rtnl_lock();
2482 switch (cmd) {
2483 case SIOCADDRT:
Thomas Graf86872cb2006-08-22 00:01:08 -07002484 err = ip6_route_add(&cfg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002485 break;
2486 case SIOCDELRT:
Thomas Graf86872cb2006-08-22 00:01:08 -07002487 err = ip6_route_del(&cfg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002488 break;
2489 default:
2490 err = -EINVAL;
2491 }
2492 rtnl_unlock();
2493
2494 return err;
Stephen Hemminger3ff50b72007-04-20 17:09:22 -07002495 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07002496
2497 return -EINVAL;
2498}
2499
2500/*
2501 * Drop the packet on the floor
2502 */
2503
Brian Haleyd5fdd6b2009-06-23 04:31:07 -07002504static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002505{
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07002506 int type;
Eric Dumazetadf30902009-06-02 05:19:30 +00002507 struct dst_entry *dst = skb_dst(skb);
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07002508 switch (ipstats_mib_noroutes) {
2509 case IPSTATS_MIB_INNOROUTES:
Arnaldo Carvalho de Melo0660e032007-04-25 17:54:47 -07002510 type = ipv6_addr_type(&ipv6_hdr(skb)->daddr);
Ulrich Weber45bb0062010-02-25 23:28:58 +00002511 if (type == IPV6_ADDR_ANY) {
Denis V. Lunev3bd653c2008-10-08 10:54:51 -07002512 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2513 IPSTATS_MIB_INADDRERRORS);
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07002514 break;
2515 }
2516 /* FALLTHROUGH */
2517 case IPSTATS_MIB_OUTNOROUTES:
Denis V. Lunev3bd653c2008-10-08 10:54:51 -07002518 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2519 ipstats_mib_noroutes);
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07002520 break;
2521 }
Alexey Dobriyan3ffe5332010-02-18 08:25:24 +00002522 icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002523 kfree_skb(skb);
2524 return 0;
2525}
2526
Thomas Graf9ce8ade2006-10-18 20:46:54 -07002527static int ip6_pkt_discard(struct sk_buff *skb)
2528{
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07002529 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES);
Thomas Graf9ce8ade2006-10-18 20:46:54 -07002530}
2531
Eric W. Biedermanede20592015-10-07 16:48:47 -05002532static int ip6_pkt_discard_out(struct net *net, struct sock *sk, struct sk_buff *skb)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002533{
Eric Dumazetadf30902009-06-02 05:19:30 +00002534 skb->dev = skb_dst(skb)->dev;
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07002535 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002536}
2537
Thomas Graf9ce8ade2006-10-18 20:46:54 -07002538static int ip6_pkt_prohibit(struct sk_buff *skb)
2539{
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07002540 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES);
Thomas Graf9ce8ade2006-10-18 20:46:54 -07002541}
2542
Eric W. Biedermanede20592015-10-07 16:48:47 -05002543static int ip6_pkt_prohibit_out(struct net *net, struct sock *sk, struct sk_buff *skb)
Thomas Graf9ce8ade2006-10-18 20:46:54 -07002544{
Eric Dumazetadf30902009-06-02 05:19:30 +00002545 skb->dev = skb_dst(skb)->dev;
YOSHIFUJI Hideaki612f09e2007-04-13 16:18:02 -07002546 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES);
Thomas Graf9ce8ade2006-10-18 20:46:54 -07002547}
2548
Linus Torvalds1da177e2005-04-16 15:20:36 -07002549/*
2550 * Allocate a dst for local (unicast / anycast) address.
2551 */
2552
2553struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
2554 const struct in6_addr *addr,
David S. Miller8f031512011-12-06 16:48:14 -05002555 bool anycast)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002556{
David Ahernca254492015-10-12 11:47:10 -07002557 u32 tb_id;
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09002558 struct net *net = dev_net(idev->dev);
Hannes Frederic Sowaa3300ef2013-12-07 03:33:45 +01002559 struct rt6_info *rt = ip6_dst_alloc(net, net->loopback_dev,
Martin KaFai Lauad706862015-08-14 11:05:52 -07002560 DST_NOCOUNT);
Hannes Frederic Sowaa3300ef2013-12-07 03:33:45 +01002561 if (!rt)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002562 return ERR_PTR(-ENOMEM);
2563
Linus Torvalds1da177e2005-04-16 15:20:36 -07002564 in6_dev_hold(idev);
2565
David S. Miller11d53b42011-06-24 15:23:34 -07002566 rt->dst.flags |= DST_HOST;
Changli Gaod8d1f302010-06-10 23:31:35 -07002567 rt->dst.input = ip6_input;
2568 rt->dst.output = ip6_output;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002569 rt->rt6i_idev = idev;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002570
2571 rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP;
YOSHIFUJI Hideaki58c4fb82005-12-21 22:56:42 +09002572 if (anycast)
2573 rt->rt6i_flags |= RTF_ANYCAST;
2574 else
Linus Torvalds1da177e2005-04-16 15:20:36 -07002575 rt->rt6i_flags |= RTF_LOCAL;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002576
Julian Anastasov550bab42013-10-20 15:43:04 +03002577 rt->rt6i_gateway = *addr;
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00002578 rt->rt6i_dst.addr = *addr;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002579 rt->rt6i_dst.plen = 128;
David Ahernca254492015-10-12 11:47:10 -07002580 tb_id = l3mdev_fib_table(idev->dev) ? : RT6_TABLE_LOCAL;
2581 rt->rt6i_table = fib6_get_table(net, tb_id);
Martin KaFai Lau8e3d5be2015-09-15 14:30:08 -07002582 rt->dst.flags |= DST_NOCACHE;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002583
Changli Gaod8d1f302010-06-10 23:31:35 -07002584 atomic_set(&rt->dst.__refcnt, 1);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002585
2586 return rt;
2587}
2588
Daniel Walterc3968a82011-04-13 21:10:57 +00002589/* remove deleted ip from prefsrc entries */
2590struct arg_dev_net_ip {
2591 struct net_device *dev;
2592 struct net *net;
2593 struct in6_addr *addr;
2594};
2595
2596static int fib6_remove_prefsrc(struct rt6_info *rt, void *arg)
2597{
2598 struct net_device *dev = ((struct arg_dev_net_ip *)arg)->dev;
2599 struct net *net = ((struct arg_dev_net_ip *)arg)->net;
2600 struct in6_addr *addr = ((struct arg_dev_net_ip *)arg)->addr;
2601
David S. Millerd1918542011-12-28 20:19:20 -05002602 if (((void *)rt->dst.dev == dev || !dev) &&
Daniel Walterc3968a82011-04-13 21:10:57 +00002603 rt != net->ipv6.ip6_null_entry &&
2604 ipv6_addr_equal(addr, &rt->rt6i_prefsrc.addr)) {
2605 /* remove prefsrc entry */
2606 rt->rt6i_prefsrc.plen = 0;
2607 }
2608 return 0;
2609}
2610
2611void rt6_remove_prefsrc(struct inet6_ifaddr *ifp)
2612{
2613 struct net *net = dev_net(ifp->idev->dev);
2614 struct arg_dev_net_ip adni = {
2615 .dev = ifp->idev->dev,
2616 .net = net,
2617 .addr = &ifp->addr,
2618 };
Li RongQing0c3584d2013-12-27 16:32:38 +08002619 fib6_clean_all(net, fib6_remove_prefsrc, &adni);
Daniel Walterc3968a82011-04-13 21:10:57 +00002620}
2621
Duan Jiongbe7a0102014-05-15 15:56:14 +08002622#define RTF_RA_ROUTER (RTF_ADDRCONF | RTF_DEFAULT | RTF_GATEWAY)
2623#define RTF_CACHE_GATEWAY (RTF_GATEWAY | RTF_CACHE)
2624
2625/* Remove routers and update dst entries when gateway turn into host. */
2626static int fib6_clean_tohost(struct rt6_info *rt, void *arg)
2627{
2628 struct in6_addr *gateway = (struct in6_addr *)arg;
2629
2630 if ((((rt->rt6i_flags & RTF_RA_ROUTER) == RTF_RA_ROUTER) ||
2631 ((rt->rt6i_flags & RTF_CACHE_GATEWAY) == RTF_CACHE_GATEWAY)) &&
2632 ipv6_addr_equal(gateway, &rt->rt6i_gateway)) {
2633 return -1;
2634 }
2635 return 0;
2636}
2637
2638void rt6_clean_tohost(struct net *net, struct in6_addr *gateway)
2639{
2640 fib6_clean_all(net, fib6_clean_tohost, gateway);
2641}
2642
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002643struct arg_dev_net {
2644 struct net_device *dev;
2645 struct net *net;
2646};
2647
Linus Torvalds1da177e2005-04-16 15:20:36 -07002648static int fib6_ifdown(struct rt6_info *rt, void *arg)
2649{
stephen hemmingerbc3ef662010-12-16 17:42:40 +00002650 const struct arg_dev_net *adn = arg;
2651 const struct net_device *dev = adn->dev;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002652
David S. Millerd1918542011-12-28 20:19:20 -05002653 if ((rt->dst.dev == dev || !dev) &&
David S. Millerc159d302011-12-26 15:24:36 -05002654 rt != adn->net->ipv6.ip6_null_entry)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002655 return -1;
David S. Millerc159d302011-12-26 15:24:36 -05002656
Linus Torvalds1da177e2005-04-16 15:20:36 -07002657 return 0;
2658}
2659
Daniel Lezcanof3db4852008-03-03 23:27:06 -08002660void rt6_ifdown(struct net *net, struct net_device *dev)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002661{
Daniel Lezcano8ed67782008-03-04 13:48:30 -08002662 struct arg_dev_net adn = {
2663 .dev = dev,
2664 .net = net,
2665 };
2666
Li RongQing0c3584d2013-12-27 16:32:38 +08002667 fib6_clean_all(net, fib6_ifdown, &adn);
David S. Miller1e493d12008-09-10 17:27:15 -07002668 icmp6_clean_all(fib6_ifdown, &adn);
Eric W. Biedermane332bc62015-10-12 11:02:08 -05002669 if (dev)
2670 rt6_uncached_list_flush_dev(net, dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002671}
2672
Eric Dumazet95c96172012-04-15 05:58:06 +00002673struct rt6_mtu_change_arg {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002674 struct net_device *dev;
Eric Dumazet95c96172012-04-15 05:58:06 +00002675 unsigned int mtu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002676};
2677
2678static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg)
2679{
2680 struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg;
2681 struct inet6_dev *idev;
2682
2683 /* In IPv6 pmtu discovery is not optional,
2684 so that RTAX_MTU lock cannot disable it.
2685 We still use this lock to block changes
2686 caused by addrconf/ndisc.
2687 */
2688
2689 idev = __in6_dev_get(arg->dev);
David S. Miller38308472011-12-03 18:02:47 -05002690 if (!idev)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002691 return 0;
2692
2693 /* For administrative MTU increase, there is no way to discover
2694 IPv6 PMTU increase, so PMTU increase should be updated here.
2695 Since RFC 1981 doesn't include administrative MTU increase
2696 update PMTU increase is a MUST. (i.e. jumbo frame)
2697 */
2698 /*
2699 If new MTU is less than route PMTU, this new MTU will be the
2700 lowest MTU in the path, update the route PMTU to reflect PMTU
2701 decreases; if new MTU is greater than route PMTU, and the
2702 old MTU is the lowest MTU in the path, update the route PMTU
2703 to reflect the increase. In this case if the other nodes' MTU
2704 also have the lowest MTU, TOO BIG MESSAGE will be lead to
2705 PMTU discouvery.
2706 */
David S. Millerd1918542011-12-28 20:19:20 -05002707 if (rt->dst.dev == arg->dev &&
Martin KaFai Lau4b32b5a2015-04-28 13:03:06 -07002708 !dst_metric_locked(&rt->dst, RTAX_MTU)) {
2709 if (rt->rt6i_flags & RTF_CACHE) {
2710 /* For RTF_CACHE with rt6i_pmtu == 0
2711 * (i.e. a redirected route),
2712 * the metrics of its rt->dst.from has already
2713 * been updated.
2714 */
2715 if (rt->rt6i_pmtu && rt->rt6i_pmtu > arg->mtu)
2716 rt->rt6i_pmtu = arg->mtu;
2717 } else if (dst_mtu(&rt->dst) >= arg->mtu ||
2718 (dst_mtu(&rt->dst) < arg->mtu &&
2719 dst_mtu(&rt->dst) == idev->cnf.mtu6)) {
2720 dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu);
2721 }
Simon Arlott566cfd82007-07-26 00:09:55 -07002722 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07002723 return 0;
2724}
2725
Eric Dumazet95c96172012-04-15 05:58:06 +00002726void rt6_mtu_change(struct net_device *dev, unsigned int mtu)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002727{
Thomas Grafc71099a2006-08-04 23:20:06 -07002728 struct rt6_mtu_change_arg arg = {
2729 .dev = dev,
2730 .mtu = mtu,
2731 };
Linus Torvalds1da177e2005-04-16 15:20:36 -07002732
Li RongQing0c3584d2013-12-27 16:32:38 +08002733 fib6_clean_all(dev_net(dev), rt6_mtu_change_route, &arg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002734}
2735
Patrick McHardyef7c79e2007-06-05 12:38:30 -07002736static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = {
Thomas Graf5176f912006-08-26 20:13:18 -07002737 [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) },
Thomas Graf86872cb2006-08-22 00:01:08 -07002738 [RTA_OIF] = { .type = NLA_U32 },
Thomas Grafab364a62006-08-22 00:01:47 -07002739 [RTA_IIF] = { .type = NLA_U32 },
Thomas Graf86872cb2006-08-22 00:01:08 -07002740 [RTA_PRIORITY] = { .type = NLA_U32 },
2741 [RTA_METRICS] = { .type = NLA_NESTED },
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002742 [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) },
Lubomir Rintelc78ba6d2015-03-11 15:39:21 +01002743 [RTA_PREF] = { .type = NLA_U8 },
Roopa Prabhu19e42e42015-07-21 10:43:48 +02002744 [RTA_ENCAP_TYPE] = { .type = NLA_U16 },
2745 [RTA_ENCAP] = { .type = NLA_NESTED },
Xin Long32bc2012015-12-16 17:50:11 +08002746 [RTA_EXPIRES] = { .type = NLA_U32 },
Thomas Graf86872cb2006-08-22 00:01:08 -07002747};
2748
2749static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh,
2750 struct fib6_config *cfg)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002751{
Thomas Graf86872cb2006-08-22 00:01:08 -07002752 struct rtmsg *rtm;
2753 struct nlattr *tb[RTA_MAX+1];
Lubomir Rintelc78ba6d2015-03-11 15:39:21 +01002754 unsigned int pref;
Thomas Graf86872cb2006-08-22 00:01:08 -07002755 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002756
Thomas Graf86872cb2006-08-22 00:01:08 -07002757 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2758 if (err < 0)
2759 goto errout;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002760
Thomas Graf86872cb2006-08-22 00:01:08 -07002761 err = -EINVAL;
2762 rtm = nlmsg_data(nlh);
2763 memset(cfg, 0, sizeof(*cfg));
2764
2765 cfg->fc_table = rtm->rtm_table;
2766 cfg->fc_dst_len = rtm->rtm_dst_len;
2767 cfg->fc_src_len = rtm->rtm_src_len;
2768 cfg->fc_flags = RTF_UP;
2769 cfg->fc_protocol = rtm->rtm_protocol;
Nicolas Dichtelef2c7d72012-09-05 02:12:42 +00002770 cfg->fc_type = rtm->rtm_type;
Thomas Graf86872cb2006-08-22 00:01:08 -07002771
Nicolas Dichtelef2c7d72012-09-05 02:12:42 +00002772 if (rtm->rtm_type == RTN_UNREACHABLE ||
2773 rtm->rtm_type == RTN_BLACKHOLE ||
Nicolas Dichtelb4949ab2012-09-06 05:53:35 +00002774 rtm->rtm_type == RTN_PROHIBIT ||
2775 rtm->rtm_type == RTN_THROW)
Thomas Graf86872cb2006-08-22 00:01:08 -07002776 cfg->fc_flags |= RTF_REJECT;
2777
Maciej Żenczykowskiab79ad12010-09-27 00:07:02 +00002778 if (rtm->rtm_type == RTN_LOCAL)
2779 cfg->fc_flags |= RTF_LOCAL;
2780
Martin KaFai Lau1f56a012015-04-28 13:03:03 -07002781 if (rtm->rtm_flags & RTM_F_CLONED)
2782 cfg->fc_flags |= RTF_CACHE;
2783
Eric W. Biederman15e47302012-09-07 20:12:54 +00002784 cfg->fc_nlinfo.portid = NETLINK_CB(skb).portid;
Thomas Graf86872cb2006-08-22 00:01:08 -07002785 cfg->fc_nlinfo.nlh = nlh;
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +09002786 cfg->fc_nlinfo.nl_net = sock_net(skb->sk);
Thomas Graf86872cb2006-08-22 00:01:08 -07002787
2788 if (tb[RTA_GATEWAY]) {
Jiri Benc67b61f62015-03-29 16:59:26 +02002789 cfg->fc_gateway = nla_get_in6_addr(tb[RTA_GATEWAY]);
Thomas Graf86872cb2006-08-22 00:01:08 -07002790 cfg->fc_flags |= RTF_GATEWAY;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002791 }
Thomas Graf86872cb2006-08-22 00:01:08 -07002792
2793 if (tb[RTA_DST]) {
2794 int plen = (rtm->rtm_dst_len + 7) >> 3;
2795
2796 if (nla_len(tb[RTA_DST]) < plen)
2797 goto errout;
2798
2799 nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002800 }
Thomas Graf86872cb2006-08-22 00:01:08 -07002801
2802 if (tb[RTA_SRC]) {
2803 int plen = (rtm->rtm_src_len + 7) >> 3;
2804
2805 if (nla_len(tb[RTA_SRC]) < plen)
2806 goto errout;
2807
2808 nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002809 }
Thomas Graf86872cb2006-08-22 00:01:08 -07002810
Daniel Walterc3968a82011-04-13 21:10:57 +00002811 if (tb[RTA_PREFSRC])
Jiri Benc67b61f62015-03-29 16:59:26 +02002812 cfg->fc_prefsrc = nla_get_in6_addr(tb[RTA_PREFSRC]);
Daniel Walterc3968a82011-04-13 21:10:57 +00002813
Thomas Graf86872cb2006-08-22 00:01:08 -07002814 if (tb[RTA_OIF])
2815 cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]);
2816
2817 if (tb[RTA_PRIORITY])
2818 cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]);
2819
2820 if (tb[RTA_METRICS]) {
2821 cfg->fc_mx = nla_data(tb[RTA_METRICS]);
2822 cfg->fc_mx_len = nla_len(tb[RTA_METRICS]);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002823 }
Thomas Graf86872cb2006-08-22 00:01:08 -07002824
2825 if (tb[RTA_TABLE])
2826 cfg->fc_table = nla_get_u32(tb[RTA_TABLE]);
2827
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002828 if (tb[RTA_MULTIPATH]) {
2829 cfg->fc_mp = nla_data(tb[RTA_MULTIPATH]);
2830 cfg->fc_mp_len = nla_len(tb[RTA_MULTIPATH]);
2831 }
2832
Lubomir Rintelc78ba6d2015-03-11 15:39:21 +01002833 if (tb[RTA_PREF]) {
2834 pref = nla_get_u8(tb[RTA_PREF]);
2835 if (pref != ICMPV6_ROUTER_PREF_LOW &&
2836 pref != ICMPV6_ROUTER_PREF_HIGH)
2837 pref = ICMPV6_ROUTER_PREF_MEDIUM;
2838 cfg->fc_flags |= RTF_PREF(pref);
2839 }
2840
Roopa Prabhu19e42e42015-07-21 10:43:48 +02002841 if (tb[RTA_ENCAP])
2842 cfg->fc_encap = tb[RTA_ENCAP];
2843
2844 if (tb[RTA_ENCAP_TYPE])
2845 cfg->fc_encap_type = nla_get_u16(tb[RTA_ENCAP_TYPE]);
2846
Xin Long32bc2012015-12-16 17:50:11 +08002847 if (tb[RTA_EXPIRES]) {
2848 unsigned long timeout = addrconf_timeout_fixup(nla_get_u32(tb[RTA_EXPIRES]), HZ);
2849
2850 if (addrconf_finite_timeout(timeout)) {
2851 cfg->fc_expires = jiffies_to_clock_t(timeout * HZ);
2852 cfg->fc_flags |= RTF_EXPIRES;
2853 }
2854 }
2855
Thomas Graf86872cb2006-08-22 00:01:08 -07002856 err = 0;
2857errout:
2858 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002859}
2860
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002861struct rt6_nh {
2862 struct rt6_info *rt6_info;
2863 struct fib6_config r_cfg;
2864 struct mx6_config mxc;
2865 struct list_head next;
2866};
2867
2868static void ip6_print_replace_route_err(struct list_head *rt6_nh_list)
2869{
2870 struct rt6_nh *nh;
2871
2872 list_for_each_entry(nh, rt6_nh_list, next) {
2873 pr_warn("IPV6: multipath route replace failed (check consistency of installed routes): %pI6 nexthop %pI6 ifi %d\n",
2874 &nh->r_cfg.fc_dst, &nh->r_cfg.fc_gateway,
2875 nh->r_cfg.fc_ifindex);
2876 }
2877}
2878
2879static int ip6_route_info_append(struct list_head *rt6_nh_list,
2880 struct rt6_info *rt, struct fib6_config *r_cfg)
2881{
2882 struct rt6_nh *nh;
2883 struct rt6_info *rtnh;
2884 int err = -EEXIST;
2885
2886 list_for_each_entry(nh, rt6_nh_list, next) {
2887 /* check if rt6_info already exists */
2888 rtnh = nh->rt6_info;
2889
2890 if (rtnh->dst.dev == rt->dst.dev &&
2891 rtnh->rt6i_idev == rt->rt6i_idev &&
2892 ipv6_addr_equal(&rtnh->rt6i_gateway,
2893 &rt->rt6i_gateway))
2894 return err;
2895 }
2896
2897 nh = kzalloc(sizeof(*nh), GFP_KERNEL);
2898 if (!nh)
2899 return -ENOMEM;
2900 nh->rt6_info = rt;
2901 err = ip6_convert_metrics(&nh->mxc, r_cfg);
2902 if (err) {
2903 kfree(nh);
2904 return err;
2905 }
2906 memcpy(&nh->r_cfg, r_cfg, sizeof(*r_cfg));
2907 list_add_tail(&nh->next, rt6_nh_list);
2908
2909 return 0;
2910}
2911
2912static int ip6_route_multipath_add(struct fib6_config *cfg)
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002913{
2914 struct fib6_config r_cfg;
2915 struct rtnexthop *rtnh;
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002916 struct rt6_info *rt;
2917 struct rt6_nh *err_nh;
2918 struct rt6_nh *nh, *nh_safe;
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002919 int remaining;
2920 int attrlen;
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002921 int err = 1;
2922 int nhn = 0;
2923 int replace = (cfg->fc_nlinfo.nlh &&
2924 (cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_REPLACE));
2925 LIST_HEAD(rt6_nh_list);
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002926
Michal Kubeček35f1b4e2015-05-18 20:53:55 +02002927 remaining = cfg->fc_mp_len;
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002928 rtnh = (struct rtnexthop *)cfg->fc_mp;
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002929
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002930 /* Parse a Multipath Entry and build a list (rt6_nh_list) of
2931 * rt6_info structs per nexthop
2932 */
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002933 while (rtnh_ok(rtnh, remaining)) {
2934 memcpy(&r_cfg, cfg, sizeof(*cfg));
2935 if (rtnh->rtnh_ifindex)
2936 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
2937
2938 attrlen = rtnh_attrlen(rtnh);
2939 if (attrlen > 0) {
2940 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
2941
2942 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
2943 if (nla) {
Jiri Benc67b61f62015-03-29 16:59:26 +02002944 r_cfg.fc_gateway = nla_get_in6_addr(nla);
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002945 r_cfg.fc_flags |= RTF_GATEWAY;
2946 }
Roopa Prabhu19e42e42015-07-21 10:43:48 +02002947 r_cfg.fc_encap = nla_find(attrs, attrlen, RTA_ENCAP);
2948 nla = nla_find(attrs, attrlen, RTA_ENCAP_TYPE);
2949 if (nla)
2950 r_cfg.fc_encap_type = nla_get_u16(nla);
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002951 }
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002952
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07002953 rt = ip6_route_info_create(&r_cfg);
2954 if (IS_ERR(rt)) {
2955 err = PTR_ERR(rt);
2956 rt = NULL;
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002957 goto cleanup;
Roopa Prabhu8c5b83f2015-10-10 08:26:36 -07002958 }
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002959
2960 err = ip6_route_info_append(&rt6_nh_list, rt, &r_cfg);
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002961 if (err) {
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002962 dst_free(&rt->dst);
2963 goto cleanup;
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00002964 }
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002965
2966 rtnh = rtnh_next(rtnh, &remaining);
2967 }
2968
2969 err_nh = NULL;
2970 list_for_each_entry(nh, &rt6_nh_list, next) {
2971 err = __ip6_ins_rt(nh->rt6_info, &cfg->fc_nlinfo, &nh->mxc);
2972 /* nh->rt6_info is used or freed at this point, reset to NULL*/
2973 nh->rt6_info = NULL;
2974 if (err) {
2975 if (replace && nhn)
2976 ip6_print_replace_route_err(&rt6_nh_list);
2977 err_nh = nh;
2978 goto add_errout;
2979 }
2980
Nicolas Dichtel1a724182012-11-01 22:58:22 +00002981 /* Because each route is added like a single route we remove
Michal Kubeček27596472015-05-18 20:54:00 +02002982 * these flags after the first nexthop: if there is a collision,
2983 * we have already failed to add the first nexthop:
2984 * fib6_add_rt2node() has rejected it; when replacing, old
2985 * nexthops have been replaced by first new, the rest should
2986 * be added to it.
Nicolas Dichtel1a724182012-11-01 22:58:22 +00002987 */
Michal Kubeček27596472015-05-18 20:54:00 +02002988 cfg->fc_nlinfo.nlh->nlmsg_flags &= ~(NLM_F_EXCL |
2989 NLM_F_REPLACE);
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07002990 nhn++;
2991 }
2992
2993 goto cleanup;
2994
2995add_errout:
2996 /* Delete routes that were already added */
2997 list_for_each_entry(nh, &rt6_nh_list, next) {
2998 if (err_nh == nh)
2999 break;
3000 ip6_route_del(&nh->r_cfg);
3001 }
3002
3003cleanup:
3004 list_for_each_entry_safe(nh, nh_safe, &rt6_nh_list, next) {
3005 if (nh->rt6_info)
3006 dst_free(&nh->rt6_info->dst);
Wu Fengguang52fe51f2015-09-10 06:57:12 +08003007 kfree(nh->mxc.mx);
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07003008 list_del(&nh->next);
3009 kfree(nh);
3010 }
3011
3012 return err;
3013}
3014
3015static int ip6_route_multipath_del(struct fib6_config *cfg)
3016{
3017 struct fib6_config r_cfg;
3018 struct rtnexthop *rtnh;
3019 int remaining;
3020 int attrlen;
3021 int err = 1, last_err = 0;
3022
3023 remaining = cfg->fc_mp_len;
3024 rtnh = (struct rtnexthop *)cfg->fc_mp;
3025
3026 /* Parse a Multipath Entry */
3027 while (rtnh_ok(rtnh, remaining)) {
3028 memcpy(&r_cfg, cfg, sizeof(*cfg));
3029 if (rtnh->rtnh_ifindex)
3030 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
3031
3032 attrlen = rtnh_attrlen(rtnh);
3033 if (attrlen > 0) {
3034 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
3035
3036 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
3037 if (nla) {
3038 nla_memcpy(&r_cfg.fc_gateway, nla, 16);
3039 r_cfg.fc_flags |= RTF_GATEWAY;
3040 }
3041 }
3042 err = ip6_route_del(&r_cfg);
3043 if (err)
3044 last_err = err;
3045
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00003046 rtnh = rtnh_next(rtnh, &remaining);
3047 }
3048
3049 return last_err;
3050}
3051
Ian Morris67ba4152014-08-24 21:53:10 +01003052static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr *nlh)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003053{
Thomas Graf86872cb2006-08-22 00:01:08 -07003054 struct fib6_config cfg;
3055 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003056
Thomas Graf86872cb2006-08-22 00:01:08 -07003057 err = rtm_to_fib6_config(skb, nlh, &cfg);
3058 if (err < 0)
3059 return err;
3060
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00003061 if (cfg.fc_mp)
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07003062 return ip6_route_multipath_del(&cfg);
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00003063 else
3064 return ip6_route_del(&cfg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003065}
3066
Ian Morris67ba4152014-08-24 21:53:10 +01003067static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr *nlh)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003068{
Thomas Graf86872cb2006-08-22 00:01:08 -07003069 struct fib6_config cfg;
3070 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003071
Thomas Graf86872cb2006-08-22 00:01:08 -07003072 err = rtm_to_fib6_config(skb, nlh, &cfg);
3073 if (err < 0)
3074 return err;
3075
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00003076 if (cfg.fc_mp)
Roopa Prabhu6b9ea5a2015-09-08 10:53:04 -07003077 return ip6_route_multipath_add(&cfg);
Nicolas Dichtel51ebd312012-10-22 03:42:09 +00003078 else
3079 return ip6_route_add(&cfg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003080}
3081
Roopa Prabhu19e42e42015-07-21 10:43:48 +02003082static inline size_t rt6_nlmsg_size(struct rt6_info *rt)
Thomas Graf339bf982006-11-10 14:10:15 -08003083{
3084 return NLMSG_ALIGN(sizeof(struct rtmsg))
3085 + nla_total_size(16) /* RTA_SRC */
3086 + nla_total_size(16) /* RTA_DST */
3087 + nla_total_size(16) /* RTA_GATEWAY */
3088 + nla_total_size(16) /* RTA_PREFSRC */
3089 + nla_total_size(4) /* RTA_TABLE */
3090 + nla_total_size(4) /* RTA_IIF */
3091 + nla_total_size(4) /* RTA_OIF */
3092 + nla_total_size(4) /* RTA_PRIORITY */
Noriaki TAKAMIYA6a2b9ce2007-01-23 22:09:41 -08003093 + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */
Daniel Borkmannea697632015-01-05 23:57:47 +01003094 + nla_total_size(sizeof(struct rta_cacheinfo))
Lubomir Rintelc78ba6d2015-03-11 15:39:21 +01003095 + nla_total_size(TCP_CA_NAME_MAX) /* RTAX_CC_ALGO */
Roopa Prabhu19e42e42015-07-21 10:43:48 +02003096 + nla_total_size(1) /* RTA_PREF */
Jiri Benc61adedf2015-08-20 13:56:25 +02003097 + lwtunnel_get_encap_size(rt->dst.lwtstate);
Thomas Graf339bf982006-11-10 14:10:15 -08003098}
3099
Brian Haley191cd582008-08-14 15:33:21 -07003100static int rt6_fill_node(struct net *net,
3101 struct sk_buff *skb, struct rt6_info *rt,
Jamal Hadi Salim0d51aa82005-06-21 13:51:04 -07003102 struct in6_addr *dst, struct in6_addr *src,
Eric W. Biederman15e47302012-09-07 20:12:54 +00003103 int iif, int type, u32 portid, u32 seq,
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09003104 int prefix, int nowait, unsigned int flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003105{
Martin KaFai Lau4b32b5a2015-04-28 13:03:06 -07003106 u32 metrics[RTAX_MAX];
Linus Torvalds1da177e2005-04-16 15:20:36 -07003107 struct rtmsg *rtm;
Thomas Graf2d7202b2006-08-22 00:01:27 -07003108 struct nlmsghdr *nlh;
Thomas Grafe3703b32006-11-27 09:27:07 -08003109 long expires;
Patrick McHardy9e762a42006-08-10 23:09:48 -07003110 u32 table;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003111
3112 if (prefix) { /* user wants prefix routes only */
3113 if (!(rt->rt6i_flags & RTF_PREFIX_RT)) {
3114 /* success since this is not a prefix route */
3115 return 1;
3116 }
3117 }
3118
Eric W. Biederman15e47302012-09-07 20:12:54 +00003119 nlh = nlmsg_put(skb, portid, seq, type, sizeof(*rtm), flags);
David S. Miller38308472011-12-03 18:02:47 -05003120 if (!nlh)
Patrick McHardy26932562007-01-31 23:16:40 -08003121 return -EMSGSIZE;
Thomas Graf2d7202b2006-08-22 00:01:27 -07003122
3123 rtm = nlmsg_data(nlh);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003124 rtm->rtm_family = AF_INET6;
3125 rtm->rtm_dst_len = rt->rt6i_dst.plen;
3126 rtm->rtm_src_len = rt->rt6i_src.plen;
3127 rtm->rtm_tos = 0;
Thomas Grafc71099a2006-08-04 23:20:06 -07003128 if (rt->rt6i_table)
Patrick McHardy9e762a42006-08-10 23:09:48 -07003129 table = rt->rt6i_table->tb6_id;
Thomas Grafc71099a2006-08-04 23:20:06 -07003130 else
Patrick McHardy9e762a42006-08-10 23:09:48 -07003131 table = RT6_TABLE_UNSPEC;
3132 rtm->rtm_table = table;
David S. Millerc78679e2012-04-01 20:27:33 -04003133 if (nla_put_u32(skb, RTA_TABLE, table))
3134 goto nla_put_failure;
Nicolas Dichtelef2c7d72012-09-05 02:12:42 +00003135 if (rt->rt6i_flags & RTF_REJECT) {
3136 switch (rt->dst.error) {
3137 case -EINVAL:
3138 rtm->rtm_type = RTN_BLACKHOLE;
3139 break;
3140 case -EACCES:
3141 rtm->rtm_type = RTN_PROHIBIT;
3142 break;
Nicolas Dichtelb4949ab2012-09-06 05:53:35 +00003143 case -EAGAIN:
3144 rtm->rtm_type = RTN_THROW;
3145 break;
Nicolas Dichtelef2c7d72012-09-05 02:12:42 +00003146 default:
3147 rtm->rtm_type = RTN_UNREACHABLE;
3148 break;
3149 }
3150 }
David S. Miller38308472011-12-03 18:02:47 -05003151 else if (rt->rt6i_flags & RTF_LOCAL)
Maciej Żenczykowskiab79ad12010-09-27 00:07:02 +00003152 rtm->rtm_type = RTN_LOCAL;
David S. Millerd1918542011-12-28 20:19:20 -05003153 else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK))
Linus Torvalds1da177e2005-04-16 15:20:36 -07003154 rtm->rtm_type = RTN_LOCAL;
3155 else
3156 rtm->rtm_type = RTN_UNICAST;
3157 rtm->rtm_flags = 0;
Andy Gospodarek35103d12015-08-13 10:39:01 -04003158 if (!netif_carrier_ok(rt->dst.dev)) {
Andy Gospodarekcea45e22015-08-13 10:39:00 -04003159 rtm->rtm_flags |= RTNH_F_LINKDOWN;
Andy Gospodarek35103d12015-08-13 10:39:01 -04003160 if (rt->rt6i_idev->cnf.ignore_routes_with_linkdown)
3161 rtm->rtm_flags |= RTNH_F_DEAD;
3162 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07003163 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
3164 rtm->rtm_protocol = rt->rt6i_protocol;
David S. Miller38308472011-12-03 18:02:47 -05003165 if (rt->rt6i_flags & RTF_DYNAMIC)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003166 rtm->rtm_protocol = RTPROT_REDIRECT;
Denis Ovsienkof0396f602012-07-10 04:45:50 +00003167 else if (rt->rt6i_flags & RTF_ADDRCONF) {
3168 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ROUTEINFO))
3169 rtm->rtm_protocol = RTPROT_RA;
3170 else
3171 rtm->rtm_protocol = RTPROT_KERNEL;
3172 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07003173
David S. Miller38308472011-12-03 18:02:47 -05003174 if (rt->rt6i_flags & RTF_CACHE)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003175 rtm->rtm_flags |= RTM_F_CLONED;
3176
3177 if (dst) {
Jiri Benc930345e2015-03-29 16:59:25 +02003178 if (nla_put_in6_addr(skb, RTA_DST, dst))
David S. Millerc78679e2012-04-01 20:27:33 -04003179 goto nla_put_failure;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09003180 rtm->rtm_dst_len = 128;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003181 } else if (rtm->rtm_dst_len)
Jiri Benc930345e2015-03-29 16:59:25 +02003182 if (nla_put_in6_addr(skb, RTA_DST, &rt->rt6i_dst.addr))
David S. Millerc78679e2012-04-01 20:27:33 -04003183 goto nla_put_failure;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003184#ifdef CONFIG_IPV6_SUBTREES
3185 if (src) {
Jiri Benc930345e2015-03-29 16:59:25 +02003186 if (nla_put_in6_addr(skb, RTA_SRC, src))
David S. Millerc78679e2012-04-01 20:27:33 -04003187 goto nla_put_failure;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09003188 rtm->rtm_src_len = 128;
David S. Millerc78679e2012-04-01 20:27:33 -04003189 } else if (rtm->rtm_src_len &&
Jiri Benc930345e2015-03-29 16:59:25 +02003190 nla_put_in6_addr(skb, RTA_SRC, &rt->rt6i_src.addr))
David S. Millerc78679e2012-04-01 20:27:33 -04003191 goto nla_put_failure;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003192#endif
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09003193 if (iif) {
3194#ifdef CONFIG_IPV6_MROUTE
3195 if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) {
Benjamin Thery8229efd2008-12-10 16:30:15 -08003196 int err = ip6mr_get_route(net, skb, rtm, nowait);
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09003197 if (err <= 0) {
3198 if (!nowait) {
3199 if (err == 0)
3200 return 0;
3201 goto nla_put_failure;
3202 } else {
3203 if (err == -EMSGSIZE)
3204 goto nla_put_failure;
3205 }
3206 }
3207 } else
3208#endif
David S. Millerc78679e2012-04-01 20:27:33 -04003209 if (nla_put_u32(skb, RTA_IIF, iif))
3210 goto nla_put_failure;
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09003211 } else if (dst) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003212 struct in6_addr saddr_buf;
David S. Millerc78679e2012-04-01 20:27:33 -04003213 if (ip6_route_get_saddr(net, rt, dst, 0, &saddr_buf) == 0 &&
Jiri Benc930345e2015-03-29 16:59:25 +02003214 nla_put_in6_addr(skb, RTA_PREFSRC, &saddr_buf))
David S. Millerc78679e2012-04-01 20:27:33 -04003215 goto nla_put_failure;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003216 }
Thomas Graf2d7202b2006-08-22 00:01:27 -07003217
Daniel Walterc3968a82011-04-13 21:10:57 +00003218 if (rt->rt6i_prefsrc.plen) {
3219 struct in6_addr saddr_buf;
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00003220 saddr_buf = rt->rt6i_prefsrc.addr;
Jiri Benc930345e2015-03-29 16:59:25 +02003221 if (nla_put_in6_addr(skb, RTA_PREFSRC, &saddr_buf))
David S. Millerc78679e2012-04-01 20:27:33 -04003222 goto nla_put_failure;
Daniel Walterc3968a82011-04-13 21:10:57 +00003223 }
3224
Martin KaFai Lau4b32b5a2015-04-28 13:03:06 -07003225 memcpy(metrics, dst_metrics_ptr(&rt->dst), sizeof(metrics));
3226 if (rt->rt6i_pmtu)
3227 metrics[RTAX_MTU - 1] = rt->rt6i_pmtu;
3228 if (rtnetlink_put_metrics(skb, metrics) < 0)
Thomas Graf2d7202b2006-08-22 00:01:27 -07003229 goto nla_put_failure;
3230
YOSHIFUJI Hideaki / 吉藤英明dd0cbf22013-01-17 12:53:15 +00003231 if (rt->rt6i_flags & RTF_GATEWAY) {
Jiri Benc930345e2015-03-29 16:59:25 +02003232 if (nla_put_in6_addr(skb, RTA_GATEWAY, &rt->rt6i_gateway) < 0)
Eric Dumazet94f826b2012-03-27 09:53:52 +00003233 goto nla_put_failure;
Eric Dumazet94f826b2012-03-27 09:53:52 +00003234 }
Thomas Graf2d7202b2006-08-22 00:01:27 -07003235
David S. Millerc78679e2012-04-01 20:27:33 -04003236 if (rt->dst.dev &&
3237 nla_put_u32(skb, RTA_OIF, rt->dst.dev->ifindex))
3238 goto nla_put_failure;
3239 if (nla_put_u32(skb, RTA_PRIORITY, rt->rt6i_metric))
3240 goto nla_put_failure;
Li Wei82539472012-07-29 16:01:30 +00003241
3242 expires = (rt->rt6i_flags & RTF_EXPIRES) ? rt->dst.expires - jiffies : 0;
YOSHIFUJI Hideaki69cdf8f2008-05-19 16:55:13 -07003243
David S. Miller87a50692012-07-10 05:06:14 -07003244 if (rtnl_put_cacheinfo(skb, &rt->dst, 0, expires, rt->dst.error) < 0)
Thomas Grafe3703b32006-11-27 09:27:07 -08003245 goto nla_put_failure;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003246
Lubomir Rintelc78ba6d2015-03-11 15:39:21 +01003247 if (nla_put_u8(skb, RTA_PREF, IPV6_EXTRACT_PREF(rt->rt6i_flags)))
3248 goto nla_put_failure;
3249
Jiri Benc61adedf2015-08-20 13:56:25 +02003250 lwtunnel_fill_encap(skb, rt->dst.lwtstate);
Roopa Prabhu19e42e42015-07-21 10:43:48 +02003251
Johannes Berg053c0952015-01-16 22:09:00 +01003252 nlmsg_end(skb, nlh);
3253 return 0;
Thomas Graf2d7202b2006-08-22 00:01:27 -07003254
3255nla_put_failure:
Patrick McHardy26932562007-01-31 23:16:40 -08003256 nlmsg_cancel(skb, nlh);
3257 return -EMSGSIZE;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003258}
3259
Patrick McHardy1b43af52006-08-10 23:11:17 -07003260int rt6_dump_route(struct rt6_info *rt, void *p_arg)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003261{
3262 struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg;
3263 int prefix;
3264
Thomas Graf2d7202b2006-08-22 00:01:27 -07003265 if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) {
3266 struct rtmsg *rtm = nlmsg_data(arg->cb->nlh);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003267 prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0;
3268 } else
3269 prefix = 0;
3270
Brian Haley191cd582008-08-14 15:33:21 -07003271 return rt6_fill_node(arg->net,
3272 arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE,
Eric W. Biederman15e47302012-09-07 20:12:54 +00003273 NETLINK_CB(arg->cb->skb).portid, arg->cb->nlh->nlmsg_seq,
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09003274 prefix, 0, NLM_F_MULTI);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003275}
3276
Ian Morris67ba4152014-08-24 21:53:10 +01003277static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003278{
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +09003279 struct net *net = sock_net(in_skb->sk);
Thomas Grafab364a62006-08-22 00:01:47 -07003280 struct nlattr *tb[RTA_MAX+1];
Linus Torvalds1da177e2005-04-16 15:20:36 -07003281 struct rt6_info *rt;
Thomas Grafab364a62006-08-22 00:01:47 -07003282 struct sk_buff *skb;
3283 struct rtmsg *rtm;
David S. Miller4c9483b2011-03-12 16:22:43 -05003284 struct flowi6 fl6;
Shmulik Ladkani72331bc2012-04-01 04:03:45 +00003285 int err, iif = 0, oif = 0;
Thomas Grafab364a62006-08-22 00:01:47 -07003286
3287 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
3288 if (err < 0)
3289 goto errout;
3290
3291 err = -EINVAL;
David S. Miller4c9483b2011-03-12 16:22:43 -05003292 memset(&fl6, 0, sizeof(fl6));
Hannes Frederic Sowa38b70972016-06-11 20:08:19 +02003293 rtm = nlmsg_data(nlh);
3294 fl6.flowlabel = ip6_make_flowinfo(rtm->rtm_tos, 0);
Thomas Grafab364a62006-08-22 00:01:47 -07003295
3296 if (tb[RTA_SRC]) {
3297 if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr))
3298 goto errout;
3299
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00003300 fl6.saddr = *(struct in6_addr *)nla_data(tb[RTA_SRC]);
Thomas Grafab364a62006-08-22 00:01:47 -07003301 }
3302
3303 if (tb[RTA_DST]) {
3304 if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr))
3305 goto errout;
3306
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +00003307 fl6.daddr = *(struct in6_addr *)nla_data(tb[RTA_DST]);
Thomas Grafab364a62006-08-22 00:01:47 -07003308 }
3309
3310 if (tb[RTA_IIF])
3311 iif = nla_get_u32(tb[RTA_IIF]);
3312
3313 if (tb[RTA_OIF])
Shmulik Ladkani72331bc2012-04-01 04:03:45 +00003314 oif = nla_get_u32(tb[RTA_OIF]);
Thomas Grafab364a62006-08-22 00:01:47 -07003315
Lorenzo Colitti2e47b292014-05-15 16:38:41 -07003316 if (tb[RTA_MARK])
3317 fl6.flowi6_mark = nla_get_u32(tb[RTA_MARK]);
3318
Thomas Grafab364a62006-08-22 00:01:47 -07003319 if (iif) {
3320 struct net_device *dev;
Shmulik Ladkani72331bc2012-04-01 04:03:45 +00003321 int flags = 0;
3322
Daniel Lezcano55786892008-03-04 13:47:47 -08003323 dev = __dev_get_by_index(net, iif);
Thomas Grafab364a62006-08-22 00:01:47 -07003324 if (!dev) {
3325 err = -ENODEV;
3326 goto errout;
3327 }
Shmulik Ladkani72331bc2012-04-01 04:03:45 +00003328
3329 fl6.flowi6_iif = iif;
3330
3331 if (!ipv6_addr_any(&fl6.saddr))
3332 flags |= RT6_LOOKUP_F_HAS_SADDR;
3333
3334 rt = (struct rt6_info *)ip6_route_input_lookup(net, dev, &fl6,
3335 flags);
3336 } else {
3337 fl6.flowi6_oif = oif;
3338
David Ahernca254492015-10-12 11:47:10 -07003339 if (netif_index_is_l3_master(net, oif)) {
3340 fl6.flowi6_flags = FLOWI_FLAG_L3MDEV_SRC |
3341 FLOWI_FLAG_SKIP_NH_OIF;
3342 }
3343
Shmulik Ladkani72331bc2012-04-01 04:03:45 +00003344 rt = (struct rt6_info *)ip6_route_output(net, NULL, &fl6);
Thomas Grafab364a62006-08-22 00:01:47 -07003345 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07003346
3347 skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
David S. Miller38308472011-12-03 18:02:47 -05003348 if (!skb) {
Amerigo Wang94e187c2012-10-29 00:13:19 +00003349 ip6_rt_put(rt);
Thomas Grafab364a62006-08-22 00:01:47 -07003350 err = -ENOBUFS;
3351 goto errout;
3352 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07003353
3354 /* Reserve room for dummy headers, this skb can pass
3355 through good chunk of routing engine.
3356 */
Arnaldo Carvalho de Melo459a98e2007-03-19 15:30:44 -07003357 skb_reset_mac_header(skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003358 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr));
3359
Changli Gaod8d1f302010-06-10 23:31:35 -07003360 skb_dst_set(skb, &rt->dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003361
David S. Miller4c9483b2011-03-12 16:22:43 -05003362 err = rt6_fill_node(net, skb, rt, &fl6.daddr, &fl6.saddr, iif,
Eric W. Biederman15e47302012-09-07 20:12:54 +00003363 RTM_NEWROUTE, NETLINK_CB(in_skb).portid,
YOSHIFUJI Hideaki7bc570c2008-04-03 09:22:53 +09003364 nlh->nlmsg_seq, 0, 0, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003365 if (err < 0) {
Thomas Grafab364a62006-08-22 00:01:47 -07003366 kfree_skb(skb);
3367 goto errout;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003368 }
3369
Eric W. Biederman15e47302012-09-07 20:12:54 +00003370 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
Thomas Grafab364a62006-08-22 00:01:47 -07003371errout:
Linus Torvalds1da177e2005-04-16 15:20:36 -07003372 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003373}
3374
Roopa Prabhu37a1d362015-09-13 10:18:33 -07003375void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info,
3376 unsigned int nlm_flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003377{
3378 struct sk_buff *skb;
Daniel Lezcano55786892008-03-04 13:47:47 -08003379 struct net *net = info->nl_net;
Denis V. Lunev528c4ce2007-12-13 09:45:12 -08003380 u32 seq;
3381 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003382
Denis V. Lunev528c4ce2007-12-13 09:45:12 -08003383 err = -ENOBUFS;
David S. Miller38308472011-12-03 18:02:47 -05003384 seq = info->nlh ? info->nlh->nlmsg_seq : 0;
Thomas Graf86872cb2006-08-22 00:01:08 -07003385
Roopa Prabhu19e42e42015-07-21 10:43:48 +02003386 skb = nlmsg_new(rt6_nlmsg_size(rt), gfp_any());
David S. Miller38308472011-12-03 18:02:47 -05003387 if (!skb)
Thomas Graf21713eb2006-08-15 00:35:24 -07003388 goto errout;
3389
Brian Haley191cd582008-08-14 15:33:21 -07003390 err = rt6_fill_node(net, skb, rt, NULL, NULL, 0,
Roopa Prabhu37a1d362015-09-13 10:18:33 -07003391 event, info->portid, seq, 0, 0, nlm_flags);
Patrick McHardy26932562007-01-31 23:16:40 -08003392 if (err < 0) {
3393 /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */
3394 WARN_ON(err == -EMSGSIZE);
3395 kfree_skb(skb);
3396 goto errout;
3397 }
Eric W. Biederman15e47302012-09-07 20:12:54 +00003398 rtnl_notify(skb, net, info->portid, RTNLGRP_IPV6_ROUTE,
Pablo Neira Ayuso1ce85fe2009-02-24 23:18:28 -08003399 info->nlh, gfp_any());
3400 return;
Thomas Graf21713eb2006-08-15 00:35:24 -07003401errout:
3402 if (err < 0)
Daniel Lezcano55786892008-03-04 13:47:47 -08003403 rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003404}
3405
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003406static int ip6_route_dev_notify(struct notifier_block *this,
Jiri Pirko351638e2013-05-28 01:30:21 +00003407 unsigned long event, void *ptr)
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003408{
Jiri Pirko351638e2013-05-28 01:30:21 +00003409 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +09003410 struct net *net = dev_net(dev);
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003411
3412 if (event == NETDEV_REGISTER && (dev->flags & IFF_LOOPBACK)) {
Changli Gaod8d1f302010-06-10 23:31:35 -07003413 net->ipv6.ip6_null_entry->dst.dev = dev;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003414 net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev);
3415#ifdef CONFIG_IPV6_MULTIPLE_TABLES
Changli Gaod8d1f302010-06-10 23:31:35 -07003416 net->ipv6.ip6_prohibit_entry->dst.dev = dev;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003417 net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev);
Changli Gaod8d1f302010-06-10 23:31:35 -07003418 net->ipv6.ip6_blk_hole_entry->dst.dev = dev;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003419 net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev);
3420#endif
3421 }
3422
3423 return NOTIFY_OK;
3424}
3425
Linus Torvalds1da177e2005-04-16 15:20:36 -07003426/*
3427 * /proc
3428 */
3429
3430#ifdef CONFIG_PROC_FS
3431
Alexey Dobriyan33120b32007-11-06 05:27:11 -08003432static const struct file_operations ipv6_route_proc_fops = {
3433 .owner = THIS_MODULE,
3434 .open = ipv6_route_open,
3435 .read = seq_read,
3436 .llseek = seq_lseek,
Hannes Frederic Sowa8d2ca1d2013-09-21 16:55:59 +02003437 .release = seq_release_net,
Alexey Dobriyan33120b32007-11-06 05:27:11 -08003438};
3439
Linus Torvalds1da177e2005-04-16 15:20:36 -07003440static int rt6_stats_seq_show(struct seq_file *seq, void *v)
3441{
Daniel Lezcano69ddb802008-03-04 13:46:23 -08003442 struct net *net = (struct net *)seq->private;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003443 seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n",
Daniel Lezcano69ddb802008-03-04 13:46:23 -08003444 net->ipv6.rt6_stats->fib_nodes,
3445 net->ipv6.rt6_stats->fib_route_nodes,
3446 net->ipv6.rt6_stats->fib_rt_alloc,
3447 net->ipv6.rt6_stats->fib_rt_entries,
3448 net->ipv6.rt6_stats->fib_rt_cache,
Eric Dumazetfc66f952010-10-08 06:37:34 +00003449 dst_entries_get_slow(&net->ipv6.ip6_dst_ops),
Daniel Lezcano69ddb802008-03-04 13:46:23 -08003450 net->ipv6.rt6_stats->fib_discarded_routes);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003451
3452 return 0;
3453}
3454
3455static int rt6_stats_seq_open(struct inode *inode, struct file *file)
3456{
Pavel Emelyanovde05c552008-07-18 04:07:21 -07003457 return single_open_net(inode, file, rt6_stats_seq_show);
Daniel Lezcano69ddb802008-03-04 13:46:23 -08003458}
3459
Arjan van de Ven9a321442007-02-12 00:55:35 -08003460static const struct file_operations rt6_stats_seq_fops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003461 .owner = THIS_MODULE,
3462 .open = rt6_stats_seq_open,
3463 .read = seq_read,
3464 .llseek = seq_lseek,
Pavel Emelyanovb6fcbdb2008-07-18 04:07:44 -07003465 .release = single_release_net,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003466};
3467#endif /* CONFIG_PROC_FS */
3468
3469#ifdef CONFIG_SYSCTL
3470
Linus Torvalds1da177e2005-04-16 15:20:36 -07003471static
Joe Perchesfe2c6332013-06-11 23:04:25 -07003472int ipv6_sysctl_rtcache_flush(struct ctl_table *ctl, int write,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003473 void __user *buffer, size_t *lenp, loff_t *ppos)
3474{
Lucian Adrian Grijincuc486da32011-02-24 19:48:03 +00003475 struct net *net;
3476 int delay;
3477 if (!write)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003478 return -EINVAL;
Lucian Adrian Grijincuc486da32011-02-24 19:48:03 +00003479
3480 net = (struct net *)ctl->extra1;
3481 delay = net->ipv6.sysctl.flush_delay;
3482 proc_dointvec(ctl, write, buffer, lenp, ppos);
Michal Kubeček2ac3ac82013-08-01 10:04:14 +02003483 fib6_run_gc(delay <= 0 ? 0 : (unsigned long)delay, net, delay > 0);
Lucian Adrian Grijincuc486da32011-02-24 19:48:03 +00003484 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003485}
3486
Joe Perchesfe2c6332013-06-11 23:04:25 -07003487struct ctl_table ipv6_route_table_template[] = {
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +09003488 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003489 .procname = "flush",
Daniel Lezcano49905092008-01-10 03:01:01 -08003490 .data = &init_net.ipv6.sysctl.flush_delay,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003491 .maxlen = sizeof(int),
Dave Jones89c8b3a12005-04-28 12:11:49 -07003492 .mode = 0200,
Alexey Dobriyan6d9f2392008-11-03 18:21:05 -08003493 .proc_handler = ipv6_sysctl_rtcache_flush
Linus Torvalds1da177e2005-04-16 15:20:36 -07003494 },
3495 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003496 .procname = "gc_thresh",
Daniel Lezcano9a7ec3a2008-03-04 13:48:53 -08003497 .data = &ip6_dst_ops_template.gc_thresh,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003498 .maxlen = sizeof(int),
3499 .mode = 0644,
Alexey Dobriyan6d9f2392008-11-03 18:21:05 -08003500 .proc_handler = proc_dointvec,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003501 },
3502 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003503 .procname = "max_size",
Daniel Lezcano49905092008-01-10 03:01:01 -08003504 .data = &init_net.ipv6.sysctl.ip6_rt_max_size,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003505 .maxlen = sizeof(int),
3506 .mode = 0644,
Alexey Dobriyan6d9f2392008-11-03 18:21:05 -08003507 .proc_handler = proc_dointvec,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003508 },
3509 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003510 .procname = "gc_min_interval",
Daniel Lezcano49905092008-01-10 03:01:01 -08003511 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
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_jiffies,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003515 },
3516 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003517 .procname = "gc_timeout",
Daniel Lezcano49905092008-01-10 03:01:01 -08003518 .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout,
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_jiffies,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003522 },
3523 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003524 .procname = "gc_interval",
Daniel Lezcano49905092008-01-10 03:01:01 -08003525 .data = &init_net.ipv6.sysctl.ip6_rt_gc_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_elasticity",
Daniel Lezcano49905092008-01-10 03:01:01 -08003532 .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003533 .maxlen = sizeof(int),
3534 .mode = 0644,
Min Zhangf3d3f612010-08-14 22:42:51 -07003535 .proc_handler = proc_dointvec,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003536 },
3537 {
Linus Torvalds1da177e2005-04-16 15:20:36 -07003538 .procname = "mtu_expires",
Daniel Lezcano49905092008-01-10 03:01:01 -08003539 .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires,
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 = "min_adv_mss",
Daniel Lezcano49905092008-01-10 03:01:01 -08003546 .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss,
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 = "gc_min_interval_ms",
Daniel Lezcano49905092008-01-10 03:01:01 -08003553 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
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_ms_jiffies,
Linus Torvalds1da177e2005-04-16 15:20:36 -07003557 },
Eric W. Biedermanf8572d82009-11-05 13:32:03 -08003558 { }
Linus Torvalds1da177e2005-04-16 15:20:36 -07003559};
3560
Alexey Dobriyan2c8c1e72010-01-17 03:35:32 +00003561struct ctl_table * __net_init ipv6_route_sysctl_init(struct net *net)
Daniel Lezcano760f2d02008-01-10 02:53:43 -08003562{
3563 struct ctl_table *table;
3564
3565 table = kmemdup(ipv6_route_table_template,
3566 sizeof(ipv6_route_table_template),
3567 GFP_KERNEL);
YOSHIFUJI Hideaki5ee09102008-02-28 00:24:28 +09003568
3569 if (table) {
3570 table[0].data = &net->ipv6.sysctl.flush_delay;
Lucian Adrian Grijincuc486da32011-02-24 19:48:03 +00003571 table[0].extra1 = net;
Alexey Dobriyan86393e52009-08-29 01:34:49 +00003572 table[1].data = &net->ipv6.ip6_dst_ops.gc_thresh;
YOSHIFUJI Hideaki5ee09102008-02-28 00:24:28 +09003573 table[2].data = &net->ipv6.sysctl.ip6_rt_max_size;
3574 table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
3575 table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout;
3576 table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval;
3577 table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity;
3578 table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires;
3579 table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss;
Alexey Dobriyan9c69fab2009-12-18 20:11:03 -08003580 table[9].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
Eric W. Biederman464dc802012-11-16 03:02:59 +00003581
3582 /* Don't export sysctls to unprivileged users */
3583 if (net->user_ns != &init_user_ns)
3584 table[0].procname = NULL;
YOSHIFUJI Hideaki5ee09102008-02-28 00:24:28 +09003585 }
3586
Daniel Lezcano760f2d02008-01-10 02:53:43 -08003587 return table;
3588}
Linus Torvalds1da177e2005-04-16 15:20:36 -07003589#endif
3590
Alexey Dobriyan2c8c1e72010-01-17 03:35:32 +00003591static int __net_init ip6_route_net_init(struct net *net)
Daniel Lezcanocdb18762008-03-04 13:45:33 -08003592{
Pavel Emelyanov633d424b2008-04-21 14:25:23 -07003593 int ret = -ENOMEM;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003594
Alexey Dobriyan86393e52009-08-29 01:34:49 +00003595 memcpy(&net->ipv6.ip6_dst_ops, &ip6_dst_ops_template,
3596 sizeof(net->ipv6.ip6_dst_ops));
Benjamin Theryf2fc6a52008-03-04 13:49:23 -08003597
Eric Dumazetfc66f952010-10-08 06:37:34 +00003598 if (dst_entries_init(&net->ipv6.ip6_dst_ops) < 0)
3599 goto out_ip6_dst_ops;
3600
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003601 net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template,
3602 sizeof(*net->ipv6.ip6_null_entry),
3603 GFP_KERNEL);
3604 if (!net->ipv6.ip6_null_entry)
Eric Dumazetfc66f952010-10-08 06:37:34 +00003605 goto out_ip6_dst_entries;
Changli Gaod8d1f302010-06-10 23:31:35 -07003606 net->ipv6.ip6_null_entry->dst.path =
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003607 (struct dst_entry *)net->ipv6.ip6_null_entry;
Changli Gaod8d1f302010-06-10 23:31:35 -07003608 net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops;
David S. Miller62fa8a82011-01-26 20:51:05 -08003609 dst_init_metrics(&net->ipv6.ip6_null_entry->dst,
3610 ip6_template_metrics, true);
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003611
3612#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3613 net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template,
3614 sizeof(*net->ipv6.ip6_prohibit_entry),
3615 GFP_KERNEL);
Peter Zijlstra68fffc62008-10-07 14:12:10 -07003616 if (!net->ipv6.ip6_prohibit_entry)
3617 goto out_ip6_null_entry;
Changli Gaod8d1f302010-06-10 23:31:35 -07003618 net->ipv6.ip6_prohibit_entry->dst.path =
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003619 (struct dst_entry *)net->ipv6.ip6_prohibit_entry;
Changli Gaod8d1f302010-06-10 23:31:35 -07003620 net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops;
David S. Miller62fa8a82011-01-26 20:51:05 -08003621 dst_init_metrics(&net->ipv6.ip6_prohibit_entry->dst,
3622 ip6_template_metrics, true);
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003623
3624 net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template,
3625 sizeof(*net->ipv6.ip6_blk_hole_entry),
3626 GFP_KERNEL);
Peter Zijlstra68fffc62008-10-07 14:12:10 -07003627 if (!net->ipv6.ip6_blk_hole_entry)
3628 goto out_ip6_prohibit_entry;
Changli Gaod8d1f302010-06-10 23:31:35 -07003629 net->ipv6.ip6_blk_hole_entry->dst.path =
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003630 (struct dst_entry *)net->ipv6.ip6_blk_hole_entry;
Changli Gaod8d1f302010-06-10 23:31:35 -07003631 net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops;
David S. Miller62fa8a82011-01-26 20:51:05 -08003632 dst_init_metrics(&net->ipv6.ip6_blk_hole_entry->dst,
3633 ip6_template_metrics, true);
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003634#endif
3635
Peter Zijlstrab339a47c2008-10-07 14:15:00 -07003636 net->ipv6.sysctl.flush_delay = 0;
3637 net->ipv6.sysctl.ip6_rt_max_size = 4096;
3638 net->ipv6.sysctl.ip6_rt_gc_min_interval = HZ / 2;
3639 net->ipv6.sysctl.ip6_rt_gc_timeout = 60*HZ;
3640 net->ipv6.sysctl.ip6_rt_gc_interval = 30*HZ;
3641 net->ipv6.sysctl.ip6_rt_gc_elasticity = 9;
3642 net->ipv6.sysctl.ip6_rt_mtu_expires = 10*60*HZ;
3643 net->ipv6.sysctl.ip6_rt_min_advmss = IPV6_MIN_MTU - 20 - 40;
3644
Benjamin Thery6891a342008-03-04 13:49:47 -08003645 net->ipv6.ip6_rt_gc_expire = 30*HZ;
3646
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003647 ret = 0;
3648out:
3649 return ret;
Benjamin Theryf2fc6a52008-03-04 13:49:23 -08003650
Peter Zijlstra68fffc62008-10-07 14:12:10 -07003651#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3652out_ip6_prohibit_entry:
3653 kfree(net->ipv6.ip6_prohibit_entry);
3654out_ip6_null_entry:
3655 kfree(net->ipv6.ip6_null_entry);
3656#endif
Eric Dumazetfc66f952010-10-08 06:37:34 +00003657out_ip6_dst_entries:
3658 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
Benjamin Theryf2fc6a52008-03-04 13:49:23 -08003659out_ip6_dst_ops:
Benjamin Theryf2fc6a52008-03-04 13:49:23 -08003660 goto out;
Daniel Lezcanocdb18762008-03-04 13:45:33 -08003661}
3662
Alexey Dobriyan2c8c1e72010-01-17 03:35:32 +00003663static void __net_exit ip6_route_net_exit(struct net *net)
Daniel Lezcanocdb18762008-03-04 13:45:33 -08003664{
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003665 kfree(net->ipv6.ip6_null_entry);
3666#ifdef CONFIG_IPV6_MULTIPLE_TABLES
3667 kfree(net->ipv6.ip6_prohibit_entry);
3668 kfree(net->ipv6.ip6_blk_hole_entry);
3669#endif
Xiaotian Feng41bb78b2010-11-02 16:11:05 +00003670 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
Daniel Lezcanocdb18762008-03-04 13:45:33 -08003671}
3672
Thomas Grafd1896342012-06-18 12:08:33 +00003673static int __net_init ip6_route_net_init_late(struct net *net)
3674{
3675#ifdef CONFIG_PROC_FS
Gao fengd4beaa62013-02-18 01:34:54 +00003676 proc_create("ipv6_route", 0, net->proc_net, &ipv6_route_proc_fops);
3677 proc_create("rt6_stats", S_IRUGO, net->proc_net, &rt6_stats_seq_fops);
Thomas Grafd1896342012-06-18 12:08:33 +00003678#endif
3679 return 0;
3680}
3681
3682static void __net_exit ip6_route_net_exit_late(struct net *net)
3683{
3684#ifdef CONFIG_PROC_FS
Gao fengece31ff2013-02-18 01:34:56 +00003685 remove_proc_entry("ipv6_route", net->proc_net);
3686 remove_proc_entry("rt6_stats", net->proc_net);
Thomas Grafd1896342012-06-18 12:08:33 +00003687#endif
3688}
3689
Daniel Lezcanocdb18762008-03-04 13:45:33 -08003690static struct pernet_operations ip6_route_net_ops = {
3691 .init = ip6_route_net_init,
3692 .exit = ip6_route_net_exit,
3693};
3694
David S. Millerc3426b42012-06-09 16:27:05 -07003695static int __net_init ipv6_inetpeer_init(struct net *net)
3696{
3697 struct inet_peer_base *bp = kmalloc(sizeof(*bp), GFP_KERNEL);
3698
3699 if (!bp)
3700 return -ENOMEM;
3701 inet_peer_base_init(bp);
3702 net->ipv6.peers = bp;
3703 return 0;
3704}
3705
3706static void __net_exit ipv6_inetpeer_exit(struct net *net)
3707{
3708 struct inet_peer_base *bp = net->ipv6.peers;
3709
3710 net->ipv6.peers = NULL;
David S. Miller56a6b242012-06-09 16:32:41 -07003711 inetpeer_invalidate_tree(bp);
David S. Millerc3426b42012-06-09 16:27:05 -07003712 kfree(bp);
3713}
3714
David S. Miller2b823f72012-06-09 19:00:16 -07003715static struct pernet_operations ipv6_inetpeer_ops = {
David S. Millerc3426b42012-06-09 16:27:05 -07003716 .init = ipv6_inetpeer_init,
3717 .exit = ipv6_inetpeer_exit,
3718};
3719
Thomas Grafd1896342012-06-18 12:08:33 +00003720static struct pernet_operations ip6_route_net_late_ops = {
3721 .init = ip6_route_net_init_late,
3722 .exit = ip6_route_net_exit_late,
3723};
3724
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003725static struct notifier_block ip6_route_dev_notifier = {
3726 .notifier_call = ip6_route_dev_notify,
3727 .priority = 0,
3728};
3729
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003730int __init ip6_route_init(void)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003731{
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003732 int ret;
Martin KaFai Lau8d0b94a2015-05-22 20:56:04 -07003733 int cpu;
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003734
Daniel Lezcano9a7ec3a2008-03-04 13:48:53 -08003735 ret = -ENOMEM;
3736 ip6_dst_ops_template.kmem_cachep =
3737 kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0,
3738 SLAB_HWCACHE_ALIGN, NULL);
3739 if (!ip6_dst_ops_template.kmem_cachep)
Fernando Carrijoc19a28e2009-01-07 18:09:08 -08003740 goto out;
David S. Miller14e50e52007-05-24 18:17:54 -07003741
Eric Dumazetfc66f952010-10-08 06:37:34 +00003742 ret = dst_entries_init(&ip6_dst_blackhole_ops);
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003743 if (ret)
Daniel Lezcanobdb32892008-03-04 13:48:10 -08003744 goto out_kmem_cache;
Daniel Lezcanobdb32892008-03-04 13:48:10 -08003745
David S. Millerc3426b42012-06-09 16:27:05 -07003746 ret = register_pernet_subsys(&ipv6_inetpeer_ops);
3747 if (ret)
David S. Millere8803b62012-06-16 01:12:19 -07003748 goto out_dst_entries;
Thomas Graf2a0c4512012-06-14 23:00:17 +00003749
David S. Miller7e52b332012-06-15 15:51:55 -07003750 ret = register_pernet_subsys(&ip6_route_net_ops);
3751 if (ret)
3752 goto out_register_inetpeer;
David S. Millerc3426b42012-06-09 16:27:05 -07003753
Arnaud Ebalard5dc121e2008-10-01 02:37:56 -07003754 ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep;
3755
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003756 /* Registering of the loopback is done before this portion of code,
3757 * the loopback reference in rt6_info will not be taken, do it
3758 * manually for init_net */
Changli Gaod8d1f302010-06-10 23:31:35 -07003759 init_net.ipv6.ip6_null_entry->dst.dev = init_net.loopback_dev;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003760 init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3761 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
Changli Gaod8d1f302010-06-10 23:31:35 -07003762 init_net.ipv6.ip6_prohibit_entry->dst.dev = init_net.loopback_dev;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003763 init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
Changli Gaod8d1f302010-06-10 23:31:35 -07003764 init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003765 init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3766 #endif
David S. Millere8803b62012-06-16 01:12:19 -07003767 ret = fib6_init();
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003768 if (ret)
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003769 goto out_register_subsys;
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003770
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003771 ret = xfrm6_init();
3772 if (ret)
David S. Millere8803b62012-06-16 01:12:19 -07003773 goto out_fib6_init;
Daniel Lezcanoc35b7e72007-12-08 00:14:11 -08003774
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003775 ret = fib6_rules_init();
3776 if (ret)
3777 goto xfrm6_init;
Daniel Lezcano7e5449c2007-12-08 00:14:54 -08003778
Thomas Grafd1896342012-06-18 12:08:33 +00003779 ret = register_pernet_subsys(&ip6_route_net_late_ops);
3780 if (ret)
3781 goto fib6_rules_init;
3782
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003783 ret = -ENOBUFS;
Greg Rosec7ac8672011-06-10 01:27:09 +00003784 if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL, NULL) ||
3785 __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL, NULL) ||
3786 __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL, NULL))
Thomas Grafd1896342012-06-18 12:08:33 +00003787 goto out_register_late_subsys;
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003788
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003789 ret = register_netdevice_notifier(&ip6_route_dev_notifier);
Daniel Lezcanocdb18762008-03-04 13:45:33 -08003790 if (ret)
Thomas Grafd1896342012-06-18 12:08:33 +00003791 goto out_register_late_subsys;
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003792
Martin KaFai Lau8d0b94a2015-05-22 20:56:04 -07003793 for_each_possible_cpu(cpu) {
3794 struct uncached_list *ul = per_cpu_ptr(&rt6_uncached_list, cpu);
3795
3796 INIT_LIST_HEAD(&ul->head);
3797 spin_lock_init(&ul->lock);
3798 }
3799
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003800out:
3801 return ret;
3802
Thomas Grafd1896342012-06-18 12:08:33 +00003803out_register_late_subsys:
3804 unregister_pernet_subsys(&ip6_route_net_late_ops);
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003805fib6_rules_init:
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003806 fib6_rules_cleanup();
3807xfrm6_init:
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003808 xfrm6_fini();
Thomas Graf2a0c4512012-06-14 23:00:17 +00003809out_fib6_init:
3810 fib6_gc_cleanup();
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003811out_register_subsys:
3812 unregister_pernet_subsys(&ip6_route_net_ops);
David S. Miller7e52b332012-06-15 15:51:55 -07003813out_register_inetpeer:
3814 unregister_pernet_subsys(&ipv6_inetpeer_ops);
Eric Dumazetfc66f952010-10-08 06:37:34 +00003815out_dst_entries:
3816 dst_entries_destroy(&ip6_dst_blackhole_ops);
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003817out_kmem_cache:
Benjamin Theryf2fc6a52008-03-04 13:49:23 -08003818 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
Daniel Lezcano433d49c2007-12-07 00:43:48 -08003819 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003820}
3821
3822void ip6_route_cleanup(void)
3823{
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003824 unregister_netdevice_notifier(&ip6_route_dev_notifier);
Thomas Grafd1896342012-06-18 12:08:33 +00003825 unregister_pernet_subsys(&ip6_route_net_late_ops);
Thomas Graf101367c2006-08-04 03:39:02 -07003826 fib6_rules_cleanup();
Linus Torvalds1da177e2005-04-16 15:20:36 -07003827 xfrm6_fini();
Linus Torvalds1da177e2005-04-16 15:20:36 -07003828 fib6_gc_cleanup();
David S. Millerc3426b42012-06-09 16:27:05 -07003829 unregister_pernet_subsys(&ipv6_inetpeer_ops);
Daniel Lezcano8ed67782008-03-04 13:48:30 -08003830 unregister_pernet_subsys(&ip6_route_net_ops);
Xiaotian Feng41bb78b2010-11-02 16:11:05 +00003831 dst_entries_destroy(&ip6_dst_blackhole_ops);
Benjamin Theryf2fc6a52008-03-04 13:49:23 -08003832 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003833}