Commit eb4890059155d5de636acba5b4421cb826d6fa4f

Authored by Hagen Paul Pfeifer
Committed by Greg Kroah-Hartman
1 parent 9e4cc1ed83

ipv6: stop sending PTB packets for MTU < 1280

[ Upstream commit 9d289715eb5c252ae15bd547cb252ca547a3c4f2 ]

Reduce the attack vector and stop generating IPv6 Fragment Header for
paths with an MTU smaller than the minimum required IPv6 MTU
size (1280 byte) - called atomic fragments.

See IETF I-D "Deprecating the Generation of IPv6 Atomic Fragments" [1]
for more information and how this "feature" can be misused.

[1] https://tools.ietf.org/html/draft-ietf-6man-deprecate-atomfrag-generation-00

Signed-off-by: Fernando Gont <fgont@si6networks.com>
Signed-off-by: Hagen Paul Pfeifer <hagen@jauu.net>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Showing 1 changed file with 2 additions and 5 deletions Inline Diff

1 /* 1 /*
2 * Linux INET6 implementation 2 * Linux INET6 implementation
3 * FIB front-end. 3 * FIB front-end.
4 * 4 *
5 * Authors: 5 * Authors:
6 * Pedro Roque <roque@di.fc.ul.pt> 6 * Pedro Roque <roque@di.fc.ul.pt>
7 * 7 *
8 * This program is free software; you can redistribute it and/or 8 * This program is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU General Public License 9 * modify it under the terms of the GNU General Public License
10 * as published by the Free Software Foundation; either version 10 * as published by the Free Software Foundation; either version
11 * 2 of the License, or (at your option) any later version. 11 * 2 of the License, or (at your option) any later version.
12 */ 12 */
13 13
14 /* Changes: 14 /* Changes:
15 * 15 *
16 * YOSHIFUJI Hideaki @USAGI 16 * YOSHIFUJI Hideaki @USAGI
17 * reworked default router selection. 17 * reworked default router selection.
18 * - respect outgoing interface 18 * - respect outgoing interface
19 * - select from (probably) reachable routers (i.e. 19 * - select from (probably) reachable routers (i.e.
20 * routers in REACHABLE, STALE, DELAY or PROBE states). 20 * routers in REACHABLE, STALE, DELAY or PROBE states).
21 * - always select the same router if it is (probably) 21 * - always select the same router if it is (probably)
22 * reachable. otherwise, round-robin the list. 22 * reachable. otherwise, round-robin the list.
23 * Ville Nuorvala 23 * Ville Nuorvala
24 * Fixed routing subtrees. 24 * Fixed routing subtrees.
25 */ 25 */
26 26
27 #define pr_fmt(fmt) "IPv6: " fmt 27 #define pr_fmt(fmt) "IPv6: " fmt
28 28
29 #include <linux/capability.h> 29 #include <linux/capability.h>
30 #include <linux/errno.h> 30 #include <linux/errno.h>
31 #include <linux/export.h> 31 #include <linux/export.h>
32 #include <linux/types.h> 32 #include <linux/types.h>
33 #include <linux/times.h> 33 #include <linux/times.h>
34 #include <linux/socket.h> 34 #include <linux/socket.h>
35 #include <linux/sockios.h> 35 #include <linux/sockios.h>
36 #include <linux/net.h> 36 #include <linux/net.h>
37 #include <linux/route.h> 37 #include <linux/route.h>
38 #include <linux/netdevice.h> 38 #include <linux/netdevice.h>
39 #include <linux/in6.h> 39 #include <linux/in6.h>
40 #include <linux/mroute6.h> 40 #include <linux/mroute6.h>
41 #include <linux/init.h> 41 #include <linux/init.h>
42 #include <linux/if_arp.h> 42 #include <linux/if_arp.h>
43 #include <linux/proc_fs.h> 43 #include <linux/proc_fs.h>
44 #include <linux/seq_file.h> 44 #include <linux/seq_file.h>
45 #include <linux/nsproxy.h> 45 #include <linux/nsproxy.h>
46 #include <linux/slab.h> 46 #include <linux/slab.h>
47 #include <net/net_namespace.h> 47 #include <net/net_namespace.h>
48 #include <net/snmp.h> 48 #include <net/snmp.h>
49 #include <net/ipv6.h> 49 #include <net/ipv6.h>
50 #include <net/ip6_fib.h> 50 #include <net/ip6_fib.h>
51 #include <net/ip6_route.h> 51 #include <net/ip6_route.h>
52 #include <net/ndisc.h> 52 #include <net/ndisc.h>
53 #include <net/addrconf.h> 53 #include <net/addrconf.h>
54 #include <net/tcp.h> 54 #include <net/tcp.h>
55 #include <linux/rtnetlink.h> 55 #include <linux/rtnetlink.h>
56 #include <net/dst.h> 56 #include <net/dst.h>
57 #include <net/xfrm.h> 57 #include <net/xfrm.h>
58 #include <net/netevent.h> 58 #include <net/netevent.h>
59 #include <net/netlink.h> 59 #include <net/netlink.h>
60 #include <net/nexthop.h> 60 #include <net/nexthop.h>
61 61
62 #include <asm/uaccess.h> 62 #include <asm/uaccess.h>
63 63
64 #ifdef CONFIG_SYSCTL 64 #ifdef CONFIG_SYSCTL
65 #include <linux/sysctl.h> 65 #include <linux/sysctl.h>
66 #endif 66 #endif
67 67
68 enum rt6_nud_state { 68 enum rt6_nud_state {
69 RT6_NUD_FAIL_HARD = -3, 69 RT6_NUD_FAIL_HARD = -3,
70 RT6_NUD_FAIL_PROBE = -2, 70 RT6_NUD_FAIL_PROBE = -2,
71 RT6_NUD_FAIL_DO_RR = -1, 71 RT6_NUD_FAIL_DO_RR = -1,
72 RT6_NUD_SUCCEED = 1 72 RT6_NUD_SUCCEED = 1
73 }; 73 };
74 74
75 static struct rt6_info *ip6_rt_copy(struct rt6_info *ort, 75 static struct rt6_info *ip6_rt_copy(struct rt6_info *ort,
76 const struct in6_addr *dest); 76 const struct in6_addr *dest);
77 static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie); 77 static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie);
78 static unsigned int ip6_default_advmss(const struct dst_entry *dst); 78 static unsigned int ip6_default_advmss(const struct dst_entry *dst);
79 static unsigned int ip6_mtu(const struct dst_entry *dst); 79 static unsigned int ip6_mtu(const struct dst_entry *dst);
80 static struct dst_entry *ip6_negative_advice(struct dst_entry *); 80 static struct dst_entry *ip6_negative_advice(struct dst_entry *);
81 static void ip6_dst_destroy(struct dst_entry *); 81 static void ip6_dst_destroy(struct dst_entry *);
82 static void ip6_dst_ifdown(struct dst_entry *, 82 static void ip6_dst_ifdown(struct dst_entry *,
83 struct net_device *dev, int how); 83 struct net_device *dev, int how);
84 static int ip6_dst_gc(struct dst_ops *ops); 84 static int ip6_dst_gc(struct dst_ops *ops);
85 85
86 static int ip6_pkt_discard(struct sk_buff *skb); 86 static int ip6_pkt_discard(struct sk_buff *skb);
87 static int ip6_pkt_discard_out(struct sock *sk, struct sk_buff *skb); 87 static int ip6_pkt_discard_out(struct sock *sk, struct sk_buff *skb);
88 static int ip6_pkt_prohibit(struct sk_buff *skb); 88 static int ip6_pkt_prohibit(struct sk_buff *skb);
89 static int ip6_pkt_prohibit_out(struct sock *sk, struct sk_buff *skb); 89 static int ip6_pkt_prohibit_out(struct sock *sk, struct sk_buff *skb);
90 static void ip6_link_failure(struct sk_buff *skb); 90 static void ip6_link_failure(struct sk_buff *skb);
91 static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk, 91 static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
92 struct sk_buff *skb, u32 mtu); 92 struct sk_buff *skb, u32 mtu);
93 static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, 93 static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk,
94 struct sk_buff *skb); 94 struct sk_buff *skb);
95 static int rt6_score_route(struct rt6_info *rt, int oif, int strict); 95 static int rt6_score_route(struct rt6_info *rt, int oif, int strict);
96 96
97 #ifdef CONFIG_IPV6_ROUTE_INFO 97 #ifdef CONFIG_IPV6_ROUTE_INFO
98 static struct rt6_info *rt6_add_route_info(struct net *net, 98 static struct rt6_info *rt6_add_route_info(struct net *net,
99 const struct in6_addr *prefix, int prefixlen, 99 const struct in6_addr *prefix, int prefixlen,
100 const struct in6_addr *gwaddr, int ifindex, 100 const struct in6_addr *gwaddr, int ifindex,
101 unsigned int pref); 101 unsigned int pref);
102 static struct rt6_info *rt6_get_route_info(struct net *net, 102 static struct rt6_info *rt6_get_route_info(struct net *net,
103 const struct in6_addr *prefix, int prefixlen, 103 const struct in6_addr *prefix, int prefixlen,
104 const struct in6_addr *gwaddr, int ifindex); 104 const struct in6_addr *gwaddr, int ifindex);
105 #endif 105 #endif
106 106
107 static void rt6_bind_peer(struct rt6_info *rt, int create) 107 static void rt6_bind_peer(struct rt6_info *rt, int create)
108 { 108 {
109 struct inet_peer_base *base; 109 struct inet_peer_base *base;
110 struct inet_peer *peer; 110 struct inet_peer *peer;
111 111
112 base = inetpeer_base_ptr(rt->_rt6i_peer); 112 base = inetpeer_base_ptr(rt->_rt6i_peer);
113 if (!base) 113 if (!base)
114 return; 114 return;
115 115
116 peer = inet_getpeer_v6(base, &rt->rt6i_dst.addr, create); 116 peer = inet_getpeer_v6(base, &rt->rt6i_dst.addr, create);
117 if (peer) { 117 if (peer) {
118 if (!rt6_set_peer(rt, peer)) 118 if (!rt6_set_peer(rt, peer))
119 inet_putpeer(peer); 119 inet_putpeer(peer);
120 } 120 }
121 } 121 }
122 122
123 static struct inet_peer *__rt6_get_peer(struct rt6_info *rt, int create) 123 static struct inet_peer *__rt6_get_peer(struct rt6_info *rt, int create)
124 { 124 {
125 if (rt6_has_peer(rt)) 125 if (rt6_has_peer(rt))
126 return rt6_peer_ptr(rt); 126 return rt6_peer_ptr(rt);
127 127
128 rt6_bind_peer(rt, create); 128 rt6_bind_peer(rt, create);
129 return (rt6_has_peer(rt) ? rt6_peer_ptr(rt) : NULL); 129 return (rt6_has_peer(rt) ? rt6_peer_ptr(rt) : NULL);
130 } 130 }
131 131
132 static struct inet_peer *rt6_get_peer_create(struct rt6_info *rt) 132 static struct inet_peer *rt6_get_peer_create(struct rt6_info *rt)
133 { 133 {
134 return __rt6_get_peer(rt, 1); 134 return __rt6_get_peer(rt, 1);
135 } 135 }
136 136
137 static u32 *ipv6_cow_metrics(struct dst_entry *dst, unsigned long old) 137 static u32 *ipv6_cow_metrics(struct dst_entry *dst, unsigned long old)
138 { 138 {
139 struct rt6_info *rt = (struct rt6_info *) dst; 139 struct rt6_info *rt = (struct rt6_info *) dst;
140 struct inet_peer *peer; 140 struct inet_peer *peer;
141 u32 *p = NULL; 141 u32 *p = NULL;
142 142
143 if (!(rt->dst.flags & DST_HOST)) 143 if (!(rt->dst.flags & DST_HOST))
144 return NULL; 144 return NULL;
145 145
146 peer = rt6_get_peer_create(rt); 146 peer = rt6_get_peer_create(rt);
147 if (peer) { 147 if (peer) {
148 u32 *old_p = __DST_METRICS_PTR(old); 148 u32 *old_p = __DST_METRICS_PTR(old);
149 unsigned long prev, new; 149 unsigned long prev, new;
150 150
151 p = peer->metrics; 151 p = peer->metrics;
152 if (inet_metrics_new(peer) || 152 if (inet_metrics_new(peer) ||
153 (old & DST_METRICS_FORCE_OVERWRITE)) 153 (old & DST_METRICS_FORCE_OVERWRITE))
154 memcpy(p, old_p, sizeof(u32) * RTAX_MAX); 154 memcpy(p, old_p, sizeof(u32) * RTAX_MAX);
155 155
156 new = (unsigned long) p; 156 new = (unsigned long) p;
157 prev = cmpxchg(&dst->_metrics, old, new); 157 prev = cmpxchg(&dst->_metrics, old, new);
158 158
159 if (prev != old) { 159 if (prev != old) {
160 p = __DST_METRICS_PTR(prev); 160 p = __DST_METRICS_PTR(prev);
161 if (prev & DST_METRICS_READ_ONLY) 161 if (prev & DST_METRICS_READ_ONLY)
162 p = NULL; 162 p = NULL;
163 } 163 }
164 } 164 }
165 return p; 165 return p;
166 } 166 }
167 167
168 static inline const void *choose_neigh_daddr(struct rt6_info *rt, 168 static inline const void *choose_neigh_daddr(struct rt6_info *rt,
169 struct sk_buff *skb, 169 struct sk_buff *skb,
170 const void *daddr) 170 const void *daddr)
171 { 171 {
172 struct in6_addr *p = &rt->rt6i_gateway; 172 struct in6_addr *p = &rt->rt6i_gateway;
173 173
174 if (!ipv6_addr_any(p)) 174 if (!ipv6_addr_any(p))
175 return (const void *) p; 175 return (const void *) p;
176 else if (skb) 176 else if (skb)
177 return &ipv6_hdr(skb)->daddr; 177 return &ipv6_hdr(skb)->daddr;
178 return daddr; 178 return daddr;
179 } 179 }
180 180
181 static struct neighbour *ip6_neigh_lookup(const struct dst_entry *dst, 181 static struct neighbour *ip6_neigh_lookup(const struct dst_entry *dst,
182 struct sk_buff *skb, 182 struct sk_buff *skb,
183 const void *daddr) 183 const void *daddr)
184 { 184 {
185 struct rt6_info *rt = (struct rt6_info *) dst; 185 struct rt6_info *rt = (struct rt6_info *) dst;
186 struct neighbour *n; 186 struct neighbour *n;
187 187
188 daddr = choose_neigh_daddr(rt, skb, daddr); 188 daddr = choose_neigh_daddr(rt, skb, daddr);
189 n = __ipv6_neigh_lookup(dst->dev, daddr); 189 n = __ipv6_neigh_lookup(dst->dev, daddr);
190 if (n) 190 if (n)
191 return n; 191 return n;
192 return neigh_create(&nd_tbl, daddr, dst->dev); 192 return neigh_create(&nd_tbl, daddr, dst->dev);
193 } 193 }
194 194
195 static struct dst_ops ip6_dst_ops_template = { 195 static struct dst_ops ip6_dst_ops_template = {
196 .family = AF_INET6, 196 .family = AF_INET6,
197 .protocol = cpu_to_be16(ETH_P_IPV6), 197 .protocol = cpu_to_be16(ETH_P_IPV6),
198 .gc = ip6_dst_gc, 198 .gc = ip6_dst_gc,
199 .gc_thresh = 1024, 199 .gc_thresh = 1024,
200 .check = ip6_dst_check, 200 .check = ip6_dst_check,
201 .default_advmss = ip6_default_advmss, 201 .default_advmss = ip6_default_advmss,
202 .mtu = ip6_mtu, 202 .mtu = ip6_mtu,
203 .cow_metrics = ipv6_cow_metrics, 203 .cow_metrics = ipv6_cow_metrics,
204 .destroy = ip6_dst_destroy, 204 .destroy = ip6_dst_destroy,
205 .ifdown = ip6_dst_ifdown, 205 .ifdown = ip6_dst_ifdown,
206 .negative_advice = ip6_negative_advice, 206 .negative_advice = ip6_negative_advice,
207 .link_failure = ip6_link_failure, 207 .link_failure = ip6_link_failure,
208 .update_pmtu = ip6_rt_update_pmtu, 208 .update_pmtu = ip6_rt_update_pmtu,
209 .redirect = rt6_do_redirect, 209 .redirect = rt6_do_redirect,
210 .local_out = __ip6_local_out, 210 .local_out = __ip6_local_out,
211 .neigh_lookup = ip6_neigh_lookup, 211 .neigh_lookup = ip6_neigh_lookup,
212 }; 212 };
213 213
214 static unsigned int ip6_blackhole_mtu(const struct dst_entry *dst) 214 static unsigned int ip6_blackhole_mtu(const struct dst_entry *dst)
215 { 215 {
216 unsigned int mtu = dst_metric_raw(dst, RTAX_MTU); 216 unsigned int mtu = dst_metric_raw(dst, RTAX_MTU);
217 217
218 return mtu ? : dst->dev->mtu; 218 return mtu ? : dst->dev->mtu;
219 } 219 }
220 220
221 static void ip6_rt_blackhole_update_pmtu(struct dst_entry *dst, struct sock *sk, 221 static void ip6_rt_blackhole_update_pmtu(struct dst_entry *dst, struct sock *sk,
222 struct sk_buff *skb, u32 mtu) 222 struct sk_buff *skb, u32 mtu)
223 { 223 {
224 } 224 }
225 225
226 static void ip6_rt_blackhole_redirect(struct dst_entry *dst, struct sock *sk, 226 static void ip6_rt_blackhole_redirect(struct dst_entry *dst, struct sock *sk,
227 struct sk_buff *skb) 227 struct sk_buff *skb)
228 { 228 {
229 } 229 }
230 230
231 static u32 *ip6_rt_blackhole_cow_metrics(struct dst_entry *dst, 231 static u32 *ip6_rt_blackhole_cow_metrics(struct dst_entry *dst,
232 unsigned long old) 232 unsigned long old)
233 { 233 {
234 return NULL; 234 return NULL;
235 } 235 }
236 236
237 static struct dst_ops ip6_dst_blackhole_ops = { 237 static struct dst_ops ip6_dst_blackhole_ops = {
238 .family = AF_INET6, 238 .family = AF_INET6,
239 .protocol = cpu_to_be16(ETH_P_IPV6), 239 .protocol = cpu_to_be16(ETH_P_IPV6),
240 .destroy = ip6_dst_destroy, 240 .destroy = ip6_dst_destroy,
241 .check = ip6_dst_check, 241 .check = ip6_dst_check,
242 .mtu = ip6_blackhole_mtu, 242 .mtu = ip6_blackhole_mtu,
243 .default_advmss = ip6_default_advmss, 243 .default_advmss = ip6_default_advmss,
244 .update_pmtu = ip6_rt_blackhole_update_pmtu, 244 .update_pmtu = ip6_rt_blackhole_update_pmtu,
245 .redirect = ip6_rt_blackhole_redirect, 245 .redirect = ip6_rt_blackhole_redirect,
246 .cow_metrics = ip6_rt_blackhole_cow_metrics, 246 .cow_metrics = ip6_rt_blackhole_cow_metrics,
247 .neigh_lookup = ip6_neigh_lookup, 247 .neigh_lookup = ip6_neigh_lookup,
248 }; 248 };
249 249
250 static const u32 ip6_template_metrics[RTAX_MAX] = { 250 static const u32 ip6_template_metrics[RTAX_MAX] = {
251 [RTAX_HOPLIMIT - 1] = 0, 251 [RTAX_HOPLIMIT - 1] = 0,
252 }; 252 };
253 253
254 static const struct rt6_info ip6_null_entry_template = { 254 static const struct rt6_info ip6_null_entry_template = {
255 .dst = { 255 .dst = {
256 .__refcnt = ATOMIC_INIT(1), 256 .__refcnt = ATOMIC_INIT(1),
257 .__use = 1, 257 .__use = 1,
258 .obsolete = DST_OBSOLETE_FORCE_CHK, 258 .obsolete = DST_OBSOLETE_FORCE_CHK,
259 .error = -ENETUNREACH, 259 .error = -ENETUNREACH,
260 .input = ip6_pkt_discard, 260 .input = ip6_pkt_discard,
261 .output = ip6_pkt_discard_out, 261 .output = ip6_pkt_discard_out,
262 }, 262 },
263 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP), 263 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
264 .rt6i_protocol = RTPROT_KERNEL, 264 .rt6i_protocol = RTPROT_KERNEL,
265 .rt6i_metric = ~(u32) 0, 265 .rt6i_metric = ~(u32) 0,
266 .rt6i_ref = ATOMIC_INIT(1), 266 .rt6i_ref = ATOMIC_INIT(1),
267 }; 267 };
268 268
269 #ifdef CONFIG_IPV6_MULTIPLE_TABLES 269 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
270 270
271 static const struct rt6_info ip6_prohibit_entry_template = { 271 static const struct rt6_info ip6_prohibit_entry_template = {
272 .dst = { 272 .dst = {
273 .__refcnt = ATOMIC_INIT(1), 273 .__refcnt = ATOMIC_INIT(1),
274 .__use = 1, 274 .__use = 1,
275 .obsolete = DST_OBSOLETE_FORCE_CHK, 275 .obsolete = DST_OBSOLETE_FORCE_CHK,
276 .error = -EACCES, 276 .error = -EACCES,
277 .input = ip6_pkt_prohibit, 277 .input = ip6_pkt_prohibit,
278 .output = ip6_pkt_prohibit_out, 278 .output = ip6_pkt_prohibit_out,
279 }, 279 },
280 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP), 280 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
281 .rt6i_protocol = RTPROT_KERNEL, 281 .rt6i_protocol = RTPROT_KERNEL,
282 .rt6i_metric = ~(u32) 0, 282 .rt6i_metric = ~(u32) 0,
283 .rt6i_ref = ATOMIC_INIT(1), 283 .rt6i_ref = ATOMIC_INIT(1),
284 }; 284 };
285 285
286 static const struct rt6_info ip6_blk_hole_entry_template = { 286 static const struct rt6_info ip6_blk_hole_entry_template = {
287 .dst = { 287 .dst = {
288 .__refcnt = ATOMIC_INIT(1), 288 .__refcnt = ATOMIC_INIT(1),
289 .__use = 1, 289 .__use = 1,
290 .obsolete = DST_OBSOLETE_FORCE_CHK, 290 .obsolete = DST_OBSOLETE_FORCE_CHK,
291 .error = -EINVAL, 291 .error = -EINVAL,
292 .input = dst_discard, 292 .input = dst_discard,
293 .output = dst_discard_sk, 293 .output = dst_discard_sk,
294 }, 294 },
295 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP), 295 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
296 .rt6i_protocol = RTPROT_KERNEL, 296 .rt6i_protocol = RTPROT_KERNEL,
297 .rt6i_metric = ~(u32) 0, 297 .rt6i_metric = ~(u32) 0,
298 .rt6i_ref = ATOMIC_INIT(1), 298 .rt6i_ref = ATOMIC_INIT(1),
299 }; 299 };
300 300
301 #endif 301 #endif
302 302
303 /* allocate dst with ip6_dst_ops */ 303 /* allocate dst with ip6_dst_ops */
304 static inline struct rt6_info *ip6_dst_alloc(struct net *net, 304 static inline struct rt6_info *ip6_dst_alloc(struct net *net,
305 struct net_device *dev, 305 struct net_device *dev,
306 int flags, 306 int flags,
307 struct fib6_table *table) 307 struct fib6_table *table)
308 { 308 {
309 struct rt6_info *rt = dst_alloc(&net->ipv6.ip6_dst_ops, dev, 309 struct rt6_info *rt = dst_alloc(&net->ipv6.ip6_dst_ops, dev,
310 0, DST_OBSOLETE_FORCE_CHK, flags); 310 0, DST_OBSOLETE_FORCE_CHK, flags);
311 311
312 if (rt) { 312 if (rt) {
313 struct dst_entry *dst = &rt->dst; 313 struct dst_entry *dst = &rt->dst;
314 314
315 memset(dst + 1, 0, sizeof(*rt) - sizeof(*dst)); 315 memset(dst + 1, 0, sizeof(*rt) - sizeof(*dst));
316 rt6_init_peer(rt, table ? &table->tb6_peers : net->ipv6.peers); 316 rt6_init_peer(rt, table ? &table->tb6_peers : net->ipv6.peers);
317 INIT_LIST_HEAD(&rt->rt6i_siblings); 317 INIT_LIST_HEAD(&rt->rt6i_siblings);
318 } 318 }
319 return rt; 319 return rt;
320 } 320 }
321 321
322 static void ip6_dst_destroy(struct dst_entry *dst) 322 static void ip6_dst_destroy(struct dst_entry *dst)
323 { 323 {
324 struct rt6_info *rt = (struct rt6_info *)dst; 324 struct rt6_info *rt = (struct rt6_info *)dst;
325 struct inet6_dev *idev = rt->rt6i_idev; 325 struct inet6_dev *idev = rt->rt6i_idev;
326 struct dst_entry *from = dst->from; 326 struct dst_entry *from = dst->from;
327 327
328 if (!(rt->dst.flags & DST_HOST)) 328 if (!(rt->dst.flags & DST_HOST))
329 dst_destroy_metrics_generic(dst); 329 dst_destroy_metrics_generic(dst);
330 330
331 if (idev) { 331 if (idev) {
332 rt->rt6i_idev = NULL; 332 rt->rt6i_idev = NULL;
333 in6_dev_put(idev); 333 in6_dev_put(idev);
334 } 334 }
335 335
336 dst->from = NULL; 336 dst->from = NULL;
337 dst_release(from); 337 dst_release(from);
338 338
339 if (rt6_has_peer(rt)) { 339 if (rt6_has_peer(rt)) {
340 struct inet_peer *peer = rt6_peer_ptr(rt); 340 struct inet_peer *peer = rt6_peer_ptr(rt);
341 inet_putpeer(peer); 341 inet_putpeer(peer);
342 } 342 }
343 } 343 }
344 344
345 static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev, 345 static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
346 int how) 346 int how)
347 { 347 {
348 struct rt6_info *rt = (struct rt6_info *)dst; 348 struct rt6_info *rt = (struct rt6_info *)dst;
349 struct inet6_dev *idev = rt->rt6i_idev; 349 struct inet6_dev *idev = rt->rt6i_idev;
350 struct net_device *loopback_dev = 350 struct net_device *loopback_dev =
351 dev_net(dev)->loopback_dev; 351 dev_net(dev)->loopback_dev;
352 352
353 if (dev != loopback_dev) { 353 if (dev != loopback_dev) {
354 if (idev && idev->dev == dev) { 354 if (idev && idev->dev == dev) {
355 struct inet6_dev *loopback_idev = 355 struct inet6_dev *loopback_idev =
356 in6_dev_get(loopback_dev); 356 in6_dev_get(loopback_dev);
357 if (loopback_idev) { 357 if (loopback_idev) {
358 rt->rt6i_idev = loopback_idev; 358 rt->rt6i_idev = loopback_idev;
359 in6_dev_put(idev); 359 in6_dev_put(idev);
360 } 360 }
361 } 361 }
362 } 362 }
363 } 363 }
364 364
365 static bool rt6_check_expired(const struct rt6_info *rt) 365 static bool rt6_check_expired(const struct rt6_info *rt)
366 { 366 {
367 if (rt->rt6i_flags & RTF_EXPIRES) { 367 if (rt->rt6i_flags & RTF_EXPIRES) {
368 if (time_after(jiffies, rt->dst.expires)) 368 if (time_after(jiffies, rt->dst.expires))
369 return true; 369 return true;
370 } else if (rt->dst.from) { 370 } else if (rt->dst.from) {
371 return rt6_check_expired((struct rt6_info *) rt->dst.from); 371 return rt6_check_expired((struct rt6_info *) rt->dst.from);
372 } 372 }
373 return false; 373 return false;
374 } 374 }
375 375
376 /* Multipath route selection: 376 /* Multipath route selection:
377 * Hash based function using packet header and flowlabel. 377 * Hash based function using packet header and flowlabel.
378 * Adapted from fib_info_hashfn() 378 * Adapted from fib_info_hashfn()
379 */ 379 */
380 static int rt6_info_hash_nhsfn(unsigned int candidate_count, 380 static int rt6_info_hash_nhsfn(unsigned int candidate_count,
381 const struct flowi6 *fl6) 381 const struct flowi6 *fl6)
382 { 382 {
383 unsigned int val = fl6->flowi6_proto; 383 unsigned int val = fl6->flowi6_proto;
384 384
385 val ^= ipv6_addr_hash(&fl6->daddr); 385 val ^= ipv6_addr_hash(&fl6->daddr);
386 val ^= ipv6_addr_hash(&fl6->saddr); 386 val ^= ipv6_addr_hash(&fl6->saddr);
387 387
388 /* Work only if this not encapsulated */ 388 /* Work only if this not encapsulated */
389 switch (fl6->flowi6_proto) { 389 switch (fl6->flowi6_proto) {
390 case IPPROTO_UDP: 390 case IPPROTO_UDP:
391 case IPPROTO_TCP: 391 case IPPROTO_TCP:
392 case IPPROTO_SCTP: 392 case IPPROTO_SCTP:
393 val ^= (__force u16)fl6->fl6_sport; 393 val ^= (__force u16)fl6->fl6_sport;
394 val ^= (__force u16)fl6->fl6_dport; 394 val ^= (__force u16)fl6->fl6_dport;
395 break; 395 break;
396 396
397 case IPPROTO_ICMPV6: 397 case IPPROTO_ICMPV6:
398 val ^= (__force u16)fl6->fl6_icmp_type; 398 val ^= (__force u16)fl6->fl6_icmp_type;
399 val ^= (__force u16)fl6->fl6_icmp_code; 399 val ^= (__force u16)fl6->fl6_icmp_code;
400 break; 400 break;
401 } 401 }
402 /* RFC6438 recommands to use flowlabel */ 402 /* RFC6438 recommands to use flowlabel */
403 val ^= (__force u32)fl6->flowlabel; 403 val ^= (__force u32)fl6->flowlabel;
404 404
405 /* Perhaps, we need to tune, this function? */ 405 /* Perhaps, we need to tune, this function? */
406 val = val ^ (val >> 7) ^ (val >> 12); 406 val = val ^ (val >> 7) ^ (val >> 12);
407 return val % candidate_count; 407 return val % candidate_count;
408 } 408 }
409 409
410 static struct rt6_info *rt6_multipath_select(struct rt6_info *match, 410 static struct rt6_info *rt6_multipath_select(struct rt6_info *match,
411 struct flowi6 *fl6, int oif, 411 struct flowi6 *fl6, int oif,
412 int strict) 412 int strict)
413 { 413 {
414 struct rt6_info *sibling, *next_sibling; 414 struct rt6_info *sibling, *next_sibling;
415 int route_choosen; 415 int route_choosen;
416 416
417 route_choosen = rt6_info_hash_nhsfn(match->rt6i_nsiblings + 1, fl6); 417 route_choosen = rt6_info_hash_nhsfn(match->rt6i_nsiblings + 1, fl6);
418 /* Don't change the route, if route_choosen == 0 418 /* Don't change the route, if route_choosen == 0
419 * (siblings does not include ourself) 419 * (siblings does not include ourself)
420 */ 420 */
421 if (route_choosen) 421 if (route_choosen)
422 list_for_each_entry_safe(sibling, next_sibling, 422 list_for_each_entry_safe(sibling, next_sibling,
423 &match->rt6i_siblings, rt6i_siblings) { 423 &match->rt6i_siblings, rt6i_siblings) {
424 route_choosen--; 424 route_choosen--;
425 if (route_choosen == 0) { 425 if (route_choosen == 0) {
426 if (rt6_score_route(sibling, oif, strict) < 0) 426 if (rt6_score_route(sibling, oif, strict) < 0)
427 break; 427 break;
428 match = sibling; 428 match = sibling;
429 break; 429 break;
430 } 430 }
431 } 431 }
432 return match; 432 return match;
433 } 433 }
434 434
435 /* 435 /*
436 * Route lookup. Any table->tb6_lock is implied. 436 * Route lookup. Any table->tb6_lock is implied.
437 */ 437 */
438 438
439 static inline struct rt6_info *rt6_device_match(struct net *net, 439 static inline struct rt6_info *rt6_device_match(struct net *net,
440 struct rt6_info *rt, 440 struct rt6_info *rt,
441 const struct in6_addr *saddr, 441 const struct in6_addr *saddr,
442 int oif, 442 int oif,
443 int flags) 443 int flags)
444 { 444 {
445 struct rt6_info *local = NULL; 445 struct rt6_info *local = NULL;
446 struct rt6_info *sprt; 446 struct rt6_info *sprt;
447 447
448 if (!oif && ipv6_addr_any(saddr)) 448 if (!oif && ipv6_addr_any(saddr))
449 goto out; 449 goto out;
450 450
451 for (sprt = rt; sprt; sprt = sprt->dst.rt6_next) { 451 for (sprt = rt; sprt; sprt = sprt->dst.rt6_next) {
452 struct net_device *dev = sprt->dst.dev; 452 struct net_device *dev = sprt->dst.dev;
453 453
454 if (oif) { 454 if (oif) {
455 if (dev->ifindex == oif) 455 if (dev->ifindex == oif)
456 return sprt; 456 return sprt;
457 if (dev->flags & IFF_LOOPBACK) { 457 if (dev->flags & IFF_LOOPBACK) {
458 if (!sprt->rt6i_idev || 458 if (!sprt->rt6i_idev ||
459 sprt->rt6i_idev->dev->ifindex != oif) { 459 sprt->rt6i_idev->dev->ifindex != oif) {
460 if (flags & RT6_LOOKUP_F_IFACE && oif) 460 if (flags & RT6_LOOKUP_F_IFACE && oif)
461 continue; 461 continue;
462 if (local && (!oif || 462 if (local && (!oif ||
463 local->rt6i_idev->dev->ifindex == oif)) 463 local->rt6i_idev->dev->ifindex == oif))
464 continue; 464 continue;
465 } 465 }
466 local = sprt; 466 local = sprt;
467 } 467 }
468 } else { 468 } else {
469 if (ipv6_chk_addr(net, saddr, dev, 469 if (ipv6_chk_addr(net, saddr, dev,
470 flags & RT6_LOOKUP_F_IFACE)) 470 flags & RT6_LOOKUP_F_IFACE))
471 return sprt; 471 return sprt;
472 } 472 }
473 } 473 }
474 474
475 if (oif) { 475 if (oif) {
476 if (local) 476 if (local)
477 return local; 477 return local;
478 478
479 if (flags & RT6_LOOKUP_F_IFACE) 479 if (flags & RT6_LOOKUP_F_IFACE)
480 return net->ipv6.ip6_null_entry; 480 return net->ipv6.ip6_null_entry;
481 } 481 }
482 out: 482 out:
483 return rt; 483 return rt;
484 } 484 }
485 485
486 #ifdef CONFIG_IPV6_ROUTER_PREF 486 #ifdef CONFIG_IPV6_ROUTER_PREF
487 struct __rt6_probe_work { 487 struct __rt6_probe_work {
488 struct work_struct work; 488 struct work_struct work;
489 struct in6_addr target; 489 struct in6_addr target;
490 struct net_device *dev; 490 struct net_device *dev;
491 }; 491 };
492 492
493 static void rt6_probe_deferred(struct work_struct *w) 493 static void rt6_probe_deferred(struct work_struct *w)
494 { 494 {
495 struct in6_addr mcaddr; 495 struct in6_addr mcaddr;
496 struct __rt6_probe_work *work = 496 struct __rt6_probe_work *work =
497 container_of(w, struct __rt6_probe_work, work); 497 container_of(w, struct __rt6_probe_work, work);
498 498
499 addrconf_addr_solict_mult(&work->target, &mcaddr); 499 addrconf_addr_solict_mult(&work->target, &mcaddr);
500 ndisc_send_ns(work->dev, NULL, &work->target, &mcaddr, NULL); 500 ndisc_send_ns(work->dev, NULL, &work->target, &mcaddr, NULL);
501 dev_put(work->dev); 501 dev_put(work->dev);
502 kfree(w); 502 kfree(w);
503 } 503 }
504 504
505 static void rt6_probe(struct rt6_info *rt) 505 static void rt6_probe(struct rt6_info *rt)
506 { 506 {
507 struct neighbour *neigh; 507 struct neighbour *neigh;
508 /* 508 /*
509 * Okay, this does not seem to be appropriate 509 * Okay, this does not seem to be appropriate
510 * for now, however, we need to check if it 510 * for now, however, we need to check if it
511 * is really so; aka Router Reachability Probing. 511 * is really so; aka Router Reachability Probing.
512 * 512 *
513 * Router Reachability Probe MUST be rate-limited 513 * Router Reachability Probe MUST be rate-limited
514 * to no more than one per minute. 514 * to no more than one per minute.
515 */ 515 */
516 if (!rt || !(rt->rt6i_flags & RTF_GATEWAY)) 516 if (!rt || !(rt->rt6i_flags & RTF_GATEWAY))
517 return; 517 return;
518 rcu_read_lock_bh(); 518 rcu_read_lock_bh();
519 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway); 519 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
520 if (neigh) { 520 if (neigh) {
521 write_lock(&neigh->lock); 521 write_lock(&neigh->lock);
522 if (neigh->nud_state & NUD_VALID) 522 if (neigh->nud_state & NUD_VALID)
523 goto out; 523 goto out;
524 } 524 }
525 525
526 if (!neigh || 526 if (!neigh ||
527 time_after(jiffies, neigh->updated + rt->rt6i_idev->cnf.rtr_probe_interval)) { 527 time_after(jiffies, neigh->updated + rt->rt6i_idev->cnf.rtr_probe_interval)) {
528 struct __rt6_probe_work *work; 528 struct __rt6_probe_work *work;
529 529
530 work = kmalloc(sizeof(*work), GFP_ATOMIC); 530 work = kmalloc(sizeof(*work), GFP_ATOMIC);
531 531
532 if (neigh && work) 532 if (neigh && work)
533 __neigh_set_probe_once(neigh); 533 __neigh_set_probe_once(neigh);
534 534
535 if (neigh) 535 if (neigh)
536 write_unlock(&neigh->lock); 536 write_unlock(&neigh->lock);
537 537
538 if (work) { 538 if (work) {
539 INIT_WORK(&work->work, rt6_probe_deferred); 539 INIT_WORK(&work->work, rt6_probe_deferred);
540 work->target = rt->rt6i_gateway; 540 work->target = rt->rt6i_gateway;
541 dev_hold(rt->dst.dev); 541 dev_hold(rt->dst.dev);
542 work->dev = rt->dst.dev; 542 work->dev = rt->dst.dev;
543 schedule_work(&work->work); 543 schedule_work(&work->work);
544 } 544 }
545 } else { 545 } else {
546 out: 546 out:
547 write_unlock(&neigh->lock); 547 write_unlock(&neigh->lock);
548 } 548 }
549 rcu_read_unlock_bh(); 549 rcu_read_unlock_bh();
550 } 550 }
551 #else 551 #else
552 static inline void rt6_probe(struct rt6_info *rt) 552 static inline void rt6_probe(struct rt6_info *rt)
553 { 553 {
554 } 554 }
555 #endif 555 #endif
556 556
557 /* 557 /*
558 * Default Router Selection (RFC 2461 6.3.6) 558 * Default Router Selection (RFC 2461 6.3.6)
559 */ 559 */
560 static inline int rt6_check_dev(struct rt6_info *rt, int oif) 560 static inline int rt6_check_dev(struct rt6_info *rt, int oif)
561 { 561 {
562 struct net_device *dev = rt->dst.dev; 562 struct net_device *dev = rt->dst.dev;
563 if (!oif || dev->ifindex == oif) 563 if (!oif || dev->ifindex == oif)
564 return 2; 564 return 2;
565 if ((dev->flags & IFF_LOOPBACK) && 565 if ((dev->flags & IFF_LOOPBACK) &&
566 rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif) 566 rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif)
567 return 1; 567 return 1;
568 return 0; 568 return 0;
569 } 569 }
570 570
571 static inline enum rt6_nud_state rt6_check_neigh(struct rt6_info *rt) 571 static inline enum rt6_nud_state rt6_check_neigh(struct rt6_info *rt)
572 { 572 {
573 struct neighbour *neigh; 573 struct neighbour *neigh;
574 enum rt6_nud_state ret = RT6_NUD_FAIL_HARD; 574 enum rt6_nud_state ret = RT6_NUD_FAIL_HARD;
575 575
576 if (rt->rt6i_flags & RTF_NONEXTHOP || 576 if (rt->rt6i_flags & RTF_NONEXTHOP ||
577 !(rt->rt6i_flags & RTF_GATEWAY)) 577 !(rt->rt6i_flags & RTF_GATEWAY))
578 return RT6_NUD_SUCCEED; 578 return RT6_NUD_SUCCEED;
579 579
580 rcu_read_lock_bh(); 580 rcu_read_lock_bh();
581 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway); 581 neigh = __ipv6_neigh_lookup_noref(rt->dst.dev, &rt->rt6i_gateway);
582 if (neigh) { 582 if (neigh) {
583 read_lock(&neigh->lock); 583 read_lock(&neigh->lock);
584 if (neigh->nud_state & NUD_VALID) 584 if (neigh->nud_state & NUD_VALID)
585 ret = RT6_NUD_SUCCEED; 585 ret = RT6_NUD_SUCCEED;
586 #ifdef CONFIG_IPV6_ROUTER_PREF 586 #ifdef CONFIG_IPV6_ROUTER_PREF
587 else if (!(neigh->nud_state & NUD_FAILED)) 587 else if (!(neigh->nud_state & NUD_FAILED))
588 ret = RT6_NUD_SUCCEED; 588 ret = RT6_NUD_SUCCEED;
589 else 589 else
590 ret = RT6_NUD_FAIL_PROBE; 590 ret = RT6_NUD_FAIL_PROBE;
591 #endif 591 #endif
592 read_unlock(&neigh->lock); 592 read_unlock(&neigh->lock);
593 } else { 593 } else {
594 ret = IS_ENABLED(CONFIG_IPV6_ROUTER_PREF) ? 594 ret = IS_ENABLED(CONFIG_IPV6_ROUTER_PREF) ?
595 RT6_NUD_SUCCEED : RT6_NUD_FAIL_DO_RR; 595 RT6_NUD_SUCCEED : RT6_NUD_FAIL_DO_RR;
596 } 596 }
597 rcu_read_unlock_bh(); 597 rcu_read_unlock_bh();
598 598
599 return ret; 599 return ret;
600 } 600 }
601 601
602 static int rt6_score_route(struct rt6_info *rt, int oif, 602 static int rt6_score_route(struct rt6_info *rt, int oif,
603 int strict) 603 int strict)
604 { 604 {
605 int m; 605 int m;
606 606
607 m = rt6_check_dev(rt, oif); 607 m = rt6_check_dev(rt, oif);
608 if (!m && (strict & RT6_LOOKUP_F_IFACE)) 608 if (!m && (strict & RT6_LOOKUP_F_IFACE))
609 return RT6_NUD_FAIL_HARD; 609 return RT6_NUD_FAIL_HARD;
610 #ifdef CONFIG_IPV6_ROUTER_PREF 610 #ifdef CONFIG_IPV6_ROUTER_PREF
611 m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2; 611 m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2;
612 #endif 612 #endif
613 if (strict & RT6_LOOKUP_F_REACHABLE) { 613 if (strict & RT6_LOOKUP_F_REACHABLE) {
614 int n = rt6_check_neigh(rt); 614 int n = rt6_check_neigh(rt);
615 if (n < 0) 615 if (n < 0)
616 return n; 616 return n;
617 } 617 }
618 return m; 618 return m;
619 } 619 }
620 620
621 static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict, 621 static struct rt6_info *find_match(struct rt6_info *rt, int oif, int strict,
622 int *mpri, struct rt6_info *match, 622 int *mpri, struct rt6_info *match,
623 bool *do_rr) 623 bool *do_rr)
624 { 624 {
625 int m; 625 int m;
626 bool match_do_rr = false; 626 bool match_do_rr = false;
627 627
628 if (rt6_check_expired(rt)) 628 if (rt6_check_expired(rt))
629 goto out; 629 goto out;
630 630
631 m = rt6_score_route(rt, oif, strict); 631 m = rt6_score_route(rt, oif, strict);
632 if (m == RT6_NUD_FAIL_DO_RR) { 632 if (m == RT6_NUD_FAIL_DO_RR) {
633 match_do_rr = true; 633 match_do_rr = true;
634 m = 0; /* lowest valid score */ 634 m = 0; /* lowest valid score */
635 } else if (m == RT6_NUD_FAIL_HARD) { 635 } else if (m == RT6_NUD_FAIL_HARD) {
636 goto out; 636 goto out;
637 } 637 }
638 638
639 if (strict & RT6_LOOKUP_F_REACHABLE) 639 if (strict & RT6_LOOKUP_F_REACHABLE)
640 rt6_probe(rt); 640 rt6_probe(rt);
641 641
642 /* note that m can be RT6_NUD_FAIL_PROBE at this point */ 642 /* note that m can be RT6_NUD_FAIL_PROBE at this point */
643 if (m > *mpri) { 643 if (m > *mpri) {
644 *do_rr = match_do_rr; 644 *do_rr = match_do_rr;
645 *mpri = m; 645 *mpri = m;
646 match = rt; 646 match = rt;
647 } 647 }
648 out: 648 out:
649 return match; 649 return match;
650 } 650 }
651 651
652 static struct rt6_info *find_rr_leaf(struct fib6_node *fn, 652 static struct rt6_info *find_rr_leaf(struct fib6_node *fn,
653 struct rt6_info *rr_head, 653 struct rt6_info *rr_head,
654 u32 metric, int oif, int strict, 654 u32 metric, int oif, int strict,
655 bool *do_rr) 655 bool *do_rr)
656 { 656 {
657 struct rt6_info *rt, *match; 657 struct rt6_info *rt, *match;
658 int mpri = -1; 658 int mpri = -1;
659 659
660 match = NULL; 660 match = NULL;
661 for (rt = rr_head; rt && rt->rt6i_metric == metric; 661 for (rt = rr_head; rt && rt->rt6i_metric == metric;
662 rt = rt->dst.rt6_next) 662 rt = rt->dst.rt6_next)
663 match = find_match(rt, oif, strict, &mpri, match, do_rr); 663 match = find_match(rt, oif, strict, &mpri, match, do_rr);
664 for (rt = fn->leaf; rt && rt != rr_head && rt->rt6i_metric == metric; 664 for (rt = fn->leaf; rt && rt != rr_head && rt->rt6i_metric == metric;
665 rt = rt->dst.rt6_next) 665 rt = rt->dst.rt6_next)
666 match = find_match(rt, oif, strict, &mpri, match, do_rr); 666 match = find_match(rt, oif, strict, &mpri, match, do_rr);
667 667
668 return match; 668 return match;
669 } 669 }
670 670
671 static struct rt6_info *rt6_select(struct fib6_node *fn, int oif, int strict) 671 static struct rt6_info *rt6_select(struct fib6_node *fn, int oif, int strict)
672 { 672 {
673 struct rt6_info *match, *rt0; 673 struct rt6_info *match, *rt0;
674 struct net *net; 674 struct net *net;
675 bool do_rr = false; 675 bool do_rr = false;
676 676
677 rt0 = fn->rr_ptr; 677 rt0 = fn->rr_ptr;
678 if (!rt0) 678 if (!rt0)
679 fn->rr_ptr = rt0 = fn->leaf; 679 fn->rr_ptr = rt0 = fn->leaf;
680 680
681 match = find_rr_leaf(fn, rt0, rt0->rt6i_metric, oif, strict, 681 match = find_rr_leaf(fn, rt0, rt0->rt6i_metric, oif, strict,
682 &do_rr); 682 &do_rr);
683 683
684 if (do_rr) { 684 if (do_rr) {
685 struct rt6_info *next = rt0->dst.rt6_next; 685 struct rt6_info *next = rt0->dst.rt6_next;
686 686
687 /* no entries matched; do round-robin */ 687 /* no entries matched; do round-robin */
688 if (!next || next->rt6i_metric != rt0->rt6i_metric) 688 if (!next || next->rt6i_metric != rt0->rt6i_metric)
689 next = fn->leaf; 689 next = fn->leaf;
690 690
691 if (next != rt0) 691 if (next != rt0)
692 fn->rr_ptr = next; 692 fn->rr_ptr = next;
693 } 693 }
694 694
695 net = dev_net(rt0->dst.dev); 695 net = dev_net(rt0->dst.dev);
696 return match ? match : net->ipv6.ip6_null_entry; 696 return match ? match : net->ipv6.ip6_null_entry;
697 } 697 }
698 698
699 #ifdef CONFIG_IPV6_ROUTE_INFO 699 #ifdef CONFIG_IPV6_ROUTE_INFO
700 int rt6_route_rcv(struct net_device *dev, u8 *opt, int len, 700 int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
701 const struct in6_addr *gwaddr) 701 const struct in6_addr *gwaddr)
702 { 702 {
703 struct net *net = dev_net(dev); 703 struct net *net = dev_net(dev);
704 struct route_info *rinfo = (struct route_info *) opt; 704 struct route_info *rinfo = (struct route_info *) opt;
705 struct in6_addr prefix_buf, *prefix; 705 struct in6_addr prefix_buf, *prefix;
706 unsigned int pref; 706 unsigned int pref;
707 unsigned long lifetime; 707 unsigned long lifetime;
708 struct rt6_info *rt; 708 struct rt6_info *rt;
709 709
710 if (len < sizeof(struct route_info)) { 710 if (len < sizeof(struct route_info)) {
711 return -EINVAL; 711 return -EINVAL;
712 } 712 }
713 713
714 /* Sanity check for prefix_len and length */ 714 /* Sanity check for prefix_len and length */
715 if (rinfo->length > 3) { 715 if (rinfo->length > 3) {
716 return -EINVAL; 716 return -EINVAL;
717 } else if (rinfo->prefix_len > 128) { 717 } else if (rinfo->prefix_len > 128) {
718 return -EINVAL; 718 return -EINVAL;
719 } else if (rinfo->prefix_len > 64) { 719 } else if (rinfo->prefix_len > 64) {
720 if (rinfo->length < 2) { 720 if (rinfo->length < 2) {
721 return -EINVAL; 721 return -EINVAL;
722 } 722 }
723 } else if (rinfo->prefix_len > 0) { 723 } else if (rinfo->prefix_len > 0) {
724 if (rinfo->length < 1) { 724 if (rinfo->length < 1) {
725 return -EINVAL; 725 return -EINVAL;
726 } 726 }
727 } 727 }
728 728
729 pref = rinfo->route_pref; 729 pref = rinfo->route_pref;
730 if (pref == ICMPV6_ROUTER_PREF_INVALID) 730 if (pref == ICMPV6_ROUTER_PREF_INVALID)
731 return -EINVAL; 731 return -EINVAL;
732 732
733 lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ); 733 lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ);
734 734
735 if (rinfo->length == 3) 735 if (rinfo->length == 3)
736 prefix = (struct in6_addr *)rinfo->prefix; 736 prefix = (struct in6_addr *)rinfo->prefix;
737 else { 737 else {
738 /* this function is safe */ 738 /* this function is safe */
739 ipv6_addr_prefix(&prefix_buf, 739 ipv6_addr_prefix(&prefix_buf,
740 (struct in6_addr *)rinfo->prefix, 740 (struct in6_addr *)rinfo->prefix,
741 rinfo->prefix_len); 741 rinfo->prefix_len);
742 prefix = &prefix_buf; 742 prefix = &prefix_buf;
743 } 743 }
744 744
745 if (rinfo->prefix_len == 0) 745 if (rinfo->prefix_len == 0)
746 rt = rt6_get_dflt_router(gwaddr, dev); 746 rt = rt6_get_dflt_router(gwaddr, dev);
747 else 747 else
748 rt = rt6_get_route_info(net, prefix, rinfo->prefix_len, 748 rt = rt6_get_route_info(net, prefix, rinfo->prefix_len,
749 gwaddr, dev->ifindex); 749 gwaddr, dev->ifindex);
750 750
751 if (rt && !lifetime) { 751 if (rt && !lifetime) {
752 ip6_del_rt(rt); 752 ip6_del_rt(rt);
753 rt = NULL; 753 rt = NULL;
754 } 754 }
755 755
756 if (!rt && lifetime) 756 if (!rt && lifetime)
757 rt = rt6_add_route_info(net, prefix, rinfo->prefix_len, gwaddr, dev->ifindex, 757 rt = rt6_add_route_info(net, prefix, rinfo->prefix_len, gwaddr, dev->ifindex,
758 pref); 758 pref);
759 else if (rt) 759 else if (rt)
760 rt->rt6i_flags = RTF_ROUTEINFO | 760 rt->rt6i_flags = RTF_ROUTEINFO |
761 (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref); 761 (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref);
762 762
763 if (rt) { 763 if (rt) {
764 if (!addrconf_finite_timeout(lifetime)) 764 if (!addrconf_finite_timeout(lifetime))
765 rt6_clean_expires(rt); 765 rt6_clean_expires(rt);
766 else 766 else
767 rt6_set_expires(rt, jiffies + HZ * lifetime); 767 rt6_set_expires(rt, jiffies + HZ * lifetime);
768 768
769 ip6_rt_put(rt); 769 ip6_rt_put(rt);
770 } 770 }
771 return 0; 771 return 0;
772 } 772 }
773 #endif 773 #endif
774 774
775 #define BACKTRACK(__net, saddr) \ 775 #define BACKTRACK(__net, saddr) \
776 do { \ 776 do { \
777 if (rt == __net->ipv6.ip6_null_entry) { \ 777 if (rt == __net->ipv6.ip6_null_entry) { \
778 struct fib6_node *pn; \ 778 struct fib6_node *pn; \
779 while (1) { \ 779 while (1) { \
780 if (fn->fn_flags & RTN_TL_ROOT) \ 780 if (fn->fn_flags & RTN_TL_ROOT) \
781 goto out; \ 781 goto out; \
782 pn = fn->parent; \ 782 pn = fn->parent; \
783 if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn) \ 783 if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn) \
784 fn = fib6_lookup(FIB6_SUBTREE(pn), NULL, saddr); \ 784 fn = fib6_lookup(FIB6_SUBTREE(pn), NULL, saddr); \
785 else \ 785 else \
786 fn = pn; \ 786 fn = pn; \
787 if (fn->fn_flags & RTN_RTINFO) \ 787 if (fn->fn_flags & RTN_RTINFO) \
788 goto restart; \ 788 goto restart; \
789 } \ 789 } \
790 } \ 790 } \
791 } while (0) 791 } while (0)
792 792
793 static struct rt6_info *ip6_pol_route_lookup(struct net *net, 793 static struct rt6_info *ip6_pol_route_lookup(struct net *net,
794 struct fib6_table *table, 794 struct fib6_table *table,
795 struct flowi6 *fl6, int flags) 795 struct flowi6 *fl6, int flags)
796 { 796 {
797 struct fib6_node *fn; 797 struct fib6_node *fn;
798 struct rt6_info *rt; 798 struct rt6_info *rt;
799 799
800 read_lock_bh(&table->tb6_lock); 800 read_lock_bh(&table->tb6_lock);
801 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr); 801 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
802 restart: 802 restart:
803 rt = fn->leaf; 803 rt = fn->leaf;
804 rt = rt6_device_match(net, rt, &fl6->saddr, fl6->flowi6_oif, flags); 804 rt = rt6_device_match(net, rt, &fl6->saddr, fl6->flowi6_oif, flags);
805 if (rt->rt6i_nsiblings && fl6->flowi6_oif == 0) 805 if (rt->rt6i_nsiblings && fl6->flowi6_oif == 0)
806 rt = rt6_multipath_select(rt, fl6, fl6->flowi6_oif, flags); 806 rt = rt6_multipath_select(rt, fl6, fl6->flowi6_oif, flags);
807 BACKTRACK(net, &fl6->saddr); 807 BACKTRACK(net, &fl6->saddr);
808 out: 808 out:
809 dst_use(&rt->dst, jiffies); 809 dst_use(&rt->dst, jiffies);
810 read_unlock_bh(&table->tb6_lock); 810 read_unlock_bh(&table->tb6_lock);
811 return rt; 811 return rt;
812 812
813 } 813 }
814 814
815 struct dst_entry *ip6_route_lookup(struct net *net, struct flowi6 *fl6, 815 struct dst_entry *ip6_route_lookup(struct net *net, struct flowi6 *fl6,
816 int flags) 816 int flags)
817 { 817 {
818 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_lookup); 818 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_lookup);
819 } 819 }
820 EXPORT_SYMBOL_GPL(ip6_route_lookup); 820 EXPORT_SYMBOL_GPL(ip6_route_lookup);
821 821
822 struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr, 822 struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
823 const struct in6_addr *saddr, int oif, int strict) 823 const struct in6_addr *saddr, int oif, int strict)
824 { 824 {
825 struct flowi6 fl6 = { 825 struct flowi6 fl6 = {
826 .flowi6_oif = oif, 826 .flowi6_oif = oif,
827 .daddr = *daddr, 827 .daddr = *daddr,
828 }; 828 };
829 struct dst_entry *dst; 829 struct dst_entry *dst;
830 int flags = strict ? RT6_LOOKUP_F_IFACE : 0; 830 int flags = strict ? RT6_LOOKUP_F_IFACE : 0;
831 831
832 if (saddr) { 832 if (saddr) {
833 memcpy(&fl6.saddr, saddr, sizeof(*saddr)); 833 memcpy(&fl6.saddr, saddr, sizeof(*saddr));
834 flags |= RT6_LOOKUP_F_HAS_SADDR; 834 flags |= RT6_LOOKUP_F_HAS_SADDR;
835 } 835 }
836 836
837 dst = fib6_rule_lookup(net, &fl6, flags, ip6_pol_route_lookup); 837 dst = fib6_rule_lookup(net, &fl6, flags, ip6_pol_route_lookup);
838 if (dst->error == 0) 838 if (dst->error == 0)
839 return (struct rt6_info *) dst; 839 return (struct rt6_info *) dst;
840 840
841 dst_release(dst); 841 dst_release(dst);
842 842
843 return NULL; 843 return NULL;
844 } 844 }
845 EXPORT_SYMBOL(rt6_lookup); 845 EXPORT_SYMBOL(rt6_lookup);
846 846
847 /* ip6_ins_rt is called with FREE table->tb6_lock. 847 /* ip6_ins_rt is called with FREE table->tb6_lock.
848 It takes new route entry, the addition fails by any reason the 848 It takes new route entry, the addition fails by any reason the
849 route is freed. In any case, if caller does not hold it, it may 849 route is freed. In any case, if caller does not hold it, it may
850 be destroyed. 850 be destroyed.
851 */ 851 */
852 852
853 static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info, 853 static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info,
854 struct nlattr *mx, int mx_len) 854 struct nlattr *mx, int mx_len)
855 { 855 {
856 int err; 856 int err;
857 struct fib6_table *table; 857 struct fib6_table *table;
858 858
859 table = rt->rt6i_table; 859 table = rt->rt6i_table;
860 write_lock_bh(&table->tb6_lock); 860 write_lock_bh(&table->tb6_lock);
861 err = fib6_add(&table->tb6_root, rt, info, mx, mx_len); 861 err = fib6_add(&table->tb6_root, rt, info, mx, mx_len);
862 write_unlock_bh(&table->tb6_lock); 862 write_unlock_bh(&table->tb6_lock);
863 863
864 return err; 864 return err;
865 } 865 }
866 866
867 int ip6_ins_rt(struct rt6_info *rt) 867 int ip6_ins_rt(struct rt6_info *rt)
868 { 868 {
869 struct nl_info info = { 869 struct nl_info info = {
870 .nl_net = dev_net(rt->dst.dev), 870 .nl_net = dev_net(rt->dst.dev),
871 }; 871 };
872 return __ip6_ins_rt(rt, &info, NULL, 0); 872 return __ip6_ins_rt(rt, &info, NULL, 0);
873 } 873 }
874 874
875 static struct rt6_info *rt6_alloc_cow(struct rt6_info *ort, 875 static struct rt6_info *rt6_alloc_cow(struct rt6_info *ort,
876 const struct in6_addr *daddr, 876 const struct in6_addr *daddr,
877 const struct in6_addr *saddr) 877 const struct in6_addr *saddr)
878 { 878 {
879 struct rt6_info *rt; 879 struct rt6_info *rt;
880 880
881 /* 881 /*
882 * Clone the route. 882 * Clone the route.
883 */ 883 */
884 884
885 rt = ip6_rt_copy(ort, daddr); 885 rt = ip6_rt_copy(ort, daddr);
886 886
887 if (rt) { 887 if (rt) {
888 if (ort->rt6i_dst.plen != 128 && 888 if (ort->rt6i_dst.plen != 128 &&
889 ipv6_addr_equal(&ort->rt6i_dst.addr, daddr)) 889 ipv6_addr_equal(&ort->rt6i_dst.addr, daddr))
890 rt->rt6i_flags |= RTF_ANYCAST; 890 rt->rt6i_flags |= RTF_ANYCAST;
891 891
892 rt->rt6i_flags |= RTF_CACHE; 892 rt->rt6i_flags |= RTF_CACHE;
893 893
894 #ifdef CONFIG_IPV6_SUBTREES 894 #ifdef CONFIG_IPV6_SUBTREES
895 if (rt->rt6i_src.plen && saddr) { 895 if (rt->rt6i_src.plen && saddr) {
896 rt->rt6i_src.addr = *saddr; 896 rt->rt6i_src.addr = *saddr;
897 rt->rt6i_src.plen = 128; 897 rt->rt6i_src.plen = 128;
898 } 898 }
899 #endif 899 #endif
900 } 900 }
901 901
902 return rt; 902 return rt;
903 } 903 }
904 904
905 static struct rt6_info *rt6_alloc_clone(struct rt6_info *ort, 905 static struct rt6_info *rt6_alloc_clone(struct rt6_info *ort,
906 const struct in6_addr *daddr) 906 const struct in6_addr *daddr)
907 { 907 {
908 struct rt6_info *rt = ip6_rt_copy(ort, daddr); 908 struct rt6_info *rt = ip6_rt_copy(ort, daddr);
909 909
910 if (rt) 910 if (rt)
911 rt->rt6i_flags |= RTF_CACHE; 911 rt->rt6i_flags |= RTF_CACHE;
912 return rt; 912 return rt;
913 } 913 }
914 914
915 static struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table, int oif, 915 static struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table, int oif,
916 struct flowi6 *fl6, int flags) 916 struct flowi6 *fl6, int flags)
917 { 917 {
918 struct fib6_node *fn; 918 struct fib6_node *fn;
919 struct rt6_info *rt, *nrt; 919 struct rt6_info *rt, *nrt;
920 int strict = 0; 920 int strict = 0;
921 int attempts = 3; 921 int attempts = 3;
922 int err; 922 int err;
923 int reachable = net->ipv6.devconf_all->forwarding ? 0 : RT6_LOOKUP_F_REACHABLE; 923 int reachable = net->ipv6.devconf_all->forwarding ? 0 : RT6_LOOKUP_F_REACHABLE;
924 924
925 strict |= flags & RT6_LOOKUP_F_IFACE; 925 strict |= flags & RT6_LOOKUP_F_IFACE;
926 926
927 relookup: 927 relookup:
928 read_lock_bh(&table->tb6_lock); 928 read_lock_bh(&table->tb6_lock);
929 929
930 restart_2: 930 restart_2:
931 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr); 931 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
932 932
933 restart: 933 restart:
934 rt = rt6_select(fn, oif, strict | reachable); 934 rt = rt6_select(fn, oif, strict | reachable);
935 if (rt->rt6i_nsiblings) 935 if (rt->rt6i_nsiblings)
936 rt = rt6_multipath_select(rt, fl6, oif, strict | reachable); 936 rt = rt6_multipath_select(rt, fl6, oif, strict | reachable);
937 BACKTRACK(net, &fl6->saddr); 937 BACKTRACK(net, &fl6->saddr);
938 if (rt == net->ipv6.ip6_null_entry || 938 if (rt == net->ipv6.ip6_null_entry ||
939 rt->rt6i_flags & RTF_CACHE) 939 rt->rt6i_flags & RTF_CACHE)
940 goto out; 940 goto out;
941 941
942 dst_hold(&rt->dst); 942 dst_hold(&rt->dst);
943 read_unlock_bh(&table->tb6_lock); 943 read_unlock_bh(&table->tb6_lock);
944 944
945 if (!(rt->rt6i_flags & (RTF_NONEXTHOP | RTF_GATEWAY))) 945 if (!(rt->rt6i_flags & (RTF_NONEXTHOP | RTF_GATEWAY)))
946 nrt = rt6_alloc_cow(rt, &fl6->daddr, &fl6->saddr); 946 nrt = rt6_alloc_cow(rt, &fl6->daddr, &fl6->saddr);
947 else if (!(rt->dst.flags & DST_HOST)) 947 else if (!(rt->dst.flags & DST_HOST))
948 nrt = rt6_alloc_clone(rt, &fl6->daddr); 948 nrt = rt6_alloc_clone(rt, &fl6->daddr);
949 else 949 else
950 goto out2; 950 goto out2;
951 951
952 ip6_rt_put(rt); 952 ip6_rt_put(rt);
953 rt = nrt ? : net->ipv6.ip6_null_entry; 953 rt = nrt ? : net->ipv6.ip6_null_entry;
954 954
955 dst_hold(&rt->dst); 955 dst_hold(&rt->dst);
956 if (nrt) { 956 if (nrt) {
957 err = ip6_ins_rt(nrt); 957 err = ip6_ins_rt(nrt);
958 if (!err) 958 if (!err)
959 goto out2; 959 goto out2;
960 } 960 }
961 961
962 if (--attempts <= 0) 962 if (--attempts <= 0)
963 goto out2; 963 goto out2;
964 964
965 /* 965 /*
966 * Race condition! In the gap, when table->tb6_lock was 966 * Race condition! In the gap, when table->tb6_lock was
967 * released someone could insert this route. Relookup. 967 * released someone could insert this route. Relookup.
968 */ 968 */
969 ip6_rt_put(rt); 969 ip6_rt_put(rt);
970 goto relookup; 970 goto relookup;
971 971
972 out: 972 out:
973 if (reachable) { 973 if (reachable) {
974 reachable = 0; 974 reachable = 0;
975 goto restart_2; 975 goto restart_2;
976 } 976 }
977 dst_hold(&rt->dst); 977 dst_hold(&rt->dst);
978 read_unlock_bh(&table->tb6_lock); 978 read_unlock_bh(&table->tb6_lock);
979 out2: 979 out2:
980 rt->dst.lastuse = jiffies; 980 rt->dst.lastuse = jiffies;
981 rt->dst.__use++; 981 rt->dst.__use++;
982 982
983 return rt; 983 return rt;
984 } 984 }
985 985
986 static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table, 986 static struct rt6_info *ip6_pol_route_input(struct net *net, struct fib6_table *table,
987 struct flowi6 *fl6, int flags) 987 struct flowi6 *fl6, int flags)
988 { 988 {
989 return ip6_pol_route(net, table, fl6->flowi6_iif, fl6, flags); 989 return ip6_pol_route(net, table, fl6->flowi6_iif, fl6, flags);
990 } 990 }
991 991
992 static struct dst_entry *ip6_route_input_lookup(struct net *net, 992 static struct dst_entry *ip6_route_input_lookup(struct net *net,
993 struct net_device *dev, 993 struct net_device *dev,
994 struct flowi6 *fl6, int flags) 994 struct flowi6 *fl6, int flags)
995 { 995 {
996 if (rt6_need_strict(&fl6->daddr) && dev->type != ARPHRD_PIMREG) 996 if (rt6_need_strict(&fl6->daddr) && dev->type != ARPHRD_PIMREG)
997 flags |= RT6_LOOKUP_F_IFACE; 997 flags |= RT6_LOOKUP_F_IFACE;
998 998
999 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_input); 999 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_input);
1000 } 1000 }
1001 1001
1002 void ip6_route_input(struct sk_buff *skb) 1002 void ip6_route_input(struct sk_buff *skb)
1003 { 1003 {
1004 const struct ipv6hdr *iph = ipv6_hdr(skb); 1004 const struct ipv6hdr *iph = ipv6_hdr(skb);
1005 struct net *net = dev_net(skb->dev); 1005 struct net *net = dev_net(skb->dev);
1006 int flags = RT6_LOOKUP_F_HAS_SADDR; 1006 int flags = RT6_LOOKUP_F_HAS_SADDR;
1007 struct flowi6 fl6 = { 1007 struct flowi6 fl6 = {
1008 .flowi6_iif = skb->dev->ifindex, 1008 .flowi6_iif = skb->dev->ifindex,
1009 .daddr = iph->daddr, 1009 .daddr = iph->daddr,
1010 .saddr = iph->saddr, 1010 .saddr = iph->saddr,
1011 .flowlabel = ip6_flowinfo(iph), 1011 .flowlabel = ip6_flowinfo(iph),
1012 .flowi6_mark = skb->mark, 1012 .flowi6_mark = skb->mark,
1013 .flowi6_proto = iph->nexthdr, 1013 .flowi6_proto = iph->nexthdr,
1014 }; 1014 };
1015 1015
1016 skb_dst_set(skb, ip6_route_input_lookup(net, skb->dev, &fl6, flags)); 1016 skb_dst_set(skb, ip6_route_input_lookup(net, skb->dev, &fl6, flags));
1017 } 1017 }
1018 1018
1019 static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table, 1019 static struct rt6_info *ip6_pol_route_output(struct net *net, struct fib6_table *table,
1020 struct flowi6 *fl6, int flags) 1020 struct flowi6 *fl6, int flags)
1021 { 1021 {
1022 return ip6_pol_route(net, table, fl6->flowi6_oif, fl6, flags); 1022 return ip6_pol_route(net, table, fl6->flowi6_oif, fl6, flags);
1023 } 1023 }
1024 1024
1025 struct dst_entry *ip6_route_output(struct net *net, const struct sock *sk, 1025 struct dst_entry *ip6_route_output(struct net *net, const struct sock *sk,
1026 struct flowi6 *fl6) 1026 struct flowi6 *fl6)
1027 { 1027 {
1028 int flags = 0; 1028 int flags = 0;
1029 1029
1030 fl6->flowi6_iif = LOOPBACK_IFINDEX; 1030 fl6->flowi6_iif = LOOPBACK_IFINDEX;
1031 1031
1032 if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl6->daddr)) 1032 if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl6->daddr))
1033 flags |= RT6_LOOKUP_F_IFACE; 1033 flags |= RT6_LOOKUP_F_IFACE;
1034 1034
1035 if (!ipv6_addr_any(&fl6->saddr)) 1035 if (!ipv6_addr_any(&fl6->saddr))
1036 flags |= RT6_LOOKUP_F_HAS_SADDR; 1036 flags |= RT6_LOOKUP_F_HAS_SADDR;
1037 else if (sk) 1037 else if (sk)
1038 flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs); 1038 flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs);
1039 1039
1040 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_output); 1040 return fib6_rule_lookup(net, fl6, flags, ip6_pol_route_output);
1041 } 1041 }
1042 EXPORT_SYMBOL(ip6_route_output); 1042 EXPORT_SYMBOL(ip6_route_output);
1043 1043
1044 struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_orig) 1044 struct dst_entry *ip6_blackhole_route(struct net *net, struct dst_entry *dst_orig)
1045 { 1045 {
1046 struct rt6_info *rt, *ort = (struct rt6_info *) dst_orig; 1046 struct rt6_info *rt, *ort = (struct rt6_info *) dst_orig;
1047 struct dst_entry *new = NULL; 1047 struct dst_entry *new = NULL;
1048 1048
1049 rt = dst_alloc(&ip6_dst_blackhole_ops, ort->dst.dev, 1, DST_OBSOLETE_NONE, 0); 1049 rt = dst_alloc(&ip6_dst_blackhole_ops, ort->dst.dev, 1, DST_OBSOLETE_NONE, 0);
1050 if (rt) { 1050 if (rt) {
1051 new = &rt->dst; 1051 new = &rt->dst;
1052 1052
1053 memset(new + 1, 0, sizeof(*rt) - sizeof(*new)); 1053 memset(new + 1, 0, sizeof(*rt) - sizeof(*new));
1054 rt6_init_peer(rt, net->ipv6.peers); 1054 rt6_init_peer(rt, net->ipv6.peers);
1055 1055
1056 new->__use = 1; 1056 new->__use = 1;
1057 new->input = dst_discard; 1057 new->input = dst_discard;
1058 new->output = dst_discard_sk; 1058 new->output = dst_discard_sk;
1059 1059
1060 if (dst_metrics_read_only(&ort->dst)) 1060 if (dst_metrics_read_only(&ort->dst))
1061 new->_metrics = ort->dst._metrics; 1061 new->_metrics = ort->dst._metrics;
1062 else 1062 else
1063 dst_copy_metrics(new, &ort->dst); 1063 dst_copy_metrics(new, &ort->dst);
1064 rt->rt6i_idev = ort->rt6i_idev; 1064 rt->rt6i_idev = ort->rt6i_idev;
1065 if (rt->rt6i_idev) 1065 if (rt->rt6i_idev)
1066 in6_dev_hold(rt->rt6i_idev); 1066 in6_dev_hold(rt->rt6i_idev);
1067 1067
1068 rt->rt6i_gateway = ort->rt6i_gateway; 1068 rt->rt6i_gateway = ort->rt6i_gateway;
1069 rt->rt6i_flags = ort->rt6i_flags; 1069 rt->rt6i_flags = ort->rt6i_flags;
1070 rt->rt6i_metric = 0; 1070 rt->rt6i_metric = 0;
1071 1071
1072 memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key)); 1072 memcpy(&rt->rt6i_dst, &ort->rt6i_dst, sizeof(struct rt6key));
1073 #ifdef CONFIG_IPV6_SUBTREES 1073 #ifdef CONFIG_IPV6_SUBTREES
1074 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key)); 1074 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1075 #endif 1075 #endif
1076 1076
1077 dst_free(new); 1077 dst_free(new);
1078 } 1078 }
1079 1079
1080 dst_release(dst_orig); 1080 dst_release(dst_orig);
1081 return new ? new : ERR_PTR(-ENOMEM); 1081 return new ? new : ERR_PTR(-ENOMEM);
1082 } 1082 }
1083 1083
1084 /* 1084 /*
1085 * Destination cache support functions 1085 * Destination cache support functions
1086 */ 1086 */
1087 1087
1088 static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie) 1088 static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie)
1089 { 1089 {
1090 struct rt6_info *rt; 1090 struct rt6_info *rt;
1091 1091
1092 rt = (struct rt6_info *) dst; 1092 rt = (struct rt6_info *) dst;
1093 1093
1094 /* All IPV6 dsts are created with ->obsolete set to the value 1094 /* All IPV6 dsts are created with ->obsolete set to the value
1095 * DST_OBSOLETE_FORCE_CHK which forces validation calls down 1095 * DST_OBSOLETE_FORCE_CHK which forces validation calls down
1096 * into this function always. 1096 * into this function always.
1097 */ 1097 */
1098 if (!rt->rt6i_node || (rt->rt6i_node->fn_sernum != cookie)) 1098 if (!rt->rt6i_node || (rt->rt6i_node->fn_sernum != cookie))
1099 return NULL; 1099 return NULL;
1100 1100
1101 if (rt6_check_expired(rt)) 1101 if (rt6_check_expired(rt))
1102 return NULL; 1102 return NULL;
1103 1103
1104 return dst; 1104 return dst;
1105 } 1105 }
1106 1106
1107 static struct dst_entry *ip6_negative_advice(struct dst_entry *dst) 1107 static struct dst_entry *ip6_negative_advice(struct dst_entry *dst)
1108 { 1108 {
1109 struct rt6_info *rt = (struct rt6_info *) dst; 1109 struct rt6_info *rt = (struct rt6_info *) dst;
1110 1110
1111 if (rt) { 1111 if (rt) {
1112 if (rt->rt6i_flags & RTF_CACHE) { 1112 if (rt->rt6i_flags & RTF_CACHE) {
1113 if (rt6_check_expired(rt)) { 1113 if (rt6_check_expired(rt)) {
1114 ip6_del_rt(rt); 1114 ip6_del_rt(rt);
1115 dst = NULL; 1115 dst = NULL;
1116 } 1116 }
1117 } else { 1117 } else {
1118 dst_release(dst); 1118 dst_release(dst);
1119 dst = NULL; 1119 dst = NULL;
1120 } 1120 }
1121 } 1121 }
1122 return dst; 1122 return dst;
1123 } 1123 }
1124 1124
1125 static void ip6_link_failure(struct sk_buff *skb) 1125 static void ip6_link_failure(struct sk_buff *skb)
1126 { 1126 {
1127 struct rt6_info *rt; 1127 struct rt6_info *rt;
1128 1128
1129 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0); 1129 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_ADDR_UNREACH, 0);
1130 1130
1131 rt = (struct rt6_info *) skb_dst(skb); 1131 rt = (struct rt6_info *) skb_dst(skb);
1132 if (rt) { 1132 if (rt) {
1133 if (rt->rt6i_flags & RTF_CACHE) { 1133 if (rt->rt6i_flags & RTF_CACHE) {
1134 dst_hold(&rt->dst); 1134 dst_hold(&rt->dst);
1135 if (ip6_del_rt(rt)) 1135 if (ip6_del_rt(rt))
1136 dst_free(&rt->dst); 1136 dst_free(&rt->dst);
1137 } else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT)) { 1137 } else if (rt->rt6i_node && (rt->rt6i_flags & RTF_DEFAULT)) {
1138 rt->rt6i_node->fn_sernum = -1; 1138 rt->rt6i_node->fn_sernum = -1;
1139 } 1139 }
1140 } 1140 }
1141 } 1141 }
1142 1142
1143 static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk, 1143 static void ip6_rt_update_pmtu(struct dst_entry *dst, struct sock *sk,
1144 struct sk_buff *skb, u32 mtu) 1144 struct sk_buff *skb, u32 mtu)
1145 { 1145 {
1146 struct rt6_info *rt6 = (struct rt6_info *)dst; 1146 struct rt6_info *rt6 = (struct rt6_info *)dst;
1147 1147
1148 dst_confirm(dst); 1148 dst_confirm(dst);
1149 if (mtu < dst_mtu(dst) && rt6->rt6i_dst.plen == 128) { 1149 if (mtu < dst_mtu(dst) && rt6->rt6i_dst.plen == 128) {
1150 struct net *net = dev_net(dst->dev); 1150 struct net *net = dev_net(dst->dev);
1151 1151
1152 rt6->rt6i_flags |= RTF_MODIFIED; 1152 rt6->rt6i_flags |= RTF_MODIFIED;
1153 if (mtu < IPV6_MIN_MTU) { 1153 if (mtu < IPV6_MIN_MTU)
1154 u32 features = dst_metric(dst, RTAX_FEATURES);
1155 mtu = IPV6_MIN_MTU; 1154 mtu = IPV6_MIN_MTU;
1156 features |= RTAX_FEATURE_ALLFRAG; 1155
1157 dst_metric_set(dst, RTAX_FEATURES, features);
1158 }
1159 dst_metric_set(dst, RTAX_MTU, mtu); 1156 dst_metric_set(dst, RTAX_MTU, mtu);
1160 rt6_update_expires(rt6, net->ipv6.sysctl.ip6_rt_mtu_expires); 1157 rt6_update_expires(rt6, net->ipv6.sysctl.ip6_rt_mtu_expires);
1161 } 1158 }
1162 } 1159 }
1163 1160
1164 void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu, 1161 void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu,
1165 int oif, u32 mark) 1162 int oif, u32 mark)
1166 { 1163 {
1167 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data; 1164 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1168 struct dst_entry *dst; 1165 struct dst_entry *dst;
1169 struct flowi6 fl6; 1166 struct flowi6 fl6;
1170 1167
1171 memset(&fl6, 0, sizeof(fl6)); 1168 memset(&fl6, 0, sizeof(fl6));
1172 fl6.flowi6_oif = oif; 1169 fl6.flowi6_oif = oif;
1173 fl6.flowi6_mark = mark ? mark : IP6_REPLY_MARK(net, skb->mark); 1170 fl6.flowi6_mark = mark ? mark : IP6_REPLY_MARK(net, skb->mark);
1174 fl6.daddr = iph->daddr; 1171 fl6.daddr = iph->daddr;
1175 fl6.saddr = iph->saddr; 1172 fl6.saddr = iph->saddr;
1176 fl6.flowlabel = ip6_flowinfo(iph); 1173 fl6.flowlabel = ip6_flowinfo(iph);
1177 1174
1178 dst = ip6_route_output(net, NULL, &fl6); 1175 dst = ip6_route_output(net, NULL, &fl6);
1179 if (!dst->error) 1176 if (!dst->error)
1180 ip6_rt_update_pmtu(dst, NULL, skb, ntohl(mtu)); 1177 ip6_rt_update_pmtu(dst, NULL, skb, ntohl(mtu));
1181 dst_release(dst); 1178 dst_release(dst);
1182 } 1179 }
1183 EXPORT_SYMBOL_GPL(ip6_update_pmtu); 1180 EXPORT_SYMBOL_GPL(ip6_update_pmtu);
1184 1181
1185 void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu) 1182 void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu)
1186 { 1183 {
1187 ip6_update_pmtu(skb, sock_net(sk), mtu, 1184 ip6_update_pmtu(skb, sock_net(sk), mtu,
1188 sk->sk_bound_dev_if, sk->sk_mark); 1185 sk->sk_bound_dev_if, sk->sk_mark);
1189 } 1186 }
1190 EXPORT_SYMBOL_GPL(ip6_sk_update_pmtu); 1187 EXPORT_SYMBOL_GPL(ip6_sk_update_pmtu);
1191 1188
1192 /* Handle redirects */ 1189 /* Handle redirects */
1193 struct ip6rd_flowi { 1190 struct ip6rd_flowi {
1194 struct flowi6 fl6; 1191 struct flowi6 fl6;
1195 struct in6_addr gateway; 1192 struct in6_addr gateway;
1196 }; 1193 };
1197 1194
1198 static struct rt6_info *__ip6_route_redirect(struct net *net, 1195 static struct rt6_info *__ip6_route_redirect(struct net *net,
1199 struct fib6_table *table, 1196 struct fib6_table *table,
1200 struct flowi6 *fl6, 1197 struct flowi6 *fl6,
1201 int flags) 1198 int flags)
1202 { 1199 {
1203 struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl6; 1200 struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl6;
1204 struct rt6_info *rt; 1201 struct rt6_info *rt;
1205 struct fib6_node *fn; 1202 struct fib6_node *fn;
1206 1203
1207 /* Get the "current" route for this destination and 1204 /* Get the "current" route for this destination and
1208 * check if the redirect has come from approriate router. 1205 * check if the redirect has come from approriate router.
1209 * 1206 *
1210 * RFC 4861 specifies that redirects should only be 1207 * RFC 4861 specifies that redirects should only be
1211 * accepted if they come from the nexthop to the target. 1208 * accepted if they come from the nexthop to the target.
1212 * Due to the way the routes are chosen, this notion 1209 * Due to the way the routes are chosen, this notion
1213 * is a bit fuzzy and one might need to check all possible 1210 * is a bit fuzzy and one might need to check all possible
1214 * routes. 1211 * routes.
1215 */ 1212 */
1216 1213
1217 read_lock_bh(&table->tb6_lock); 1214 read_lock_bh(&table->tb6_lock);
1218 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr); 1215 fn = fib6_lookup(&table->tb6_root, &fl6->daddr, &fl6->saddr);
1219 restart: 1216 restart:
1220 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) { 1217 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1221 if (rt6_check_expired(rt)) 1218 if (rt6_check_expired(rt))
1222 continue; 1219 continue;
1223 if (rt->dst.error) 1220 if (rt->dst.error)
1224 break; 1221 break;
1225 if (!(rt->rt6i_flags & RTF_GATEWAY)) 1222 if (!(rt->rt6i_flags & RTF_GATEWAY))
1226 continue; 1223 continue;
1227 if (fl6->flowi6_oif != rt->dst.dev->ifindex) 1224 if (fl6->flowi6_oif != rt->dst.dev->ifindex)
1228 continue; 1225 continue;
1229 if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway)) 1226 if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway))
1230 continue; 1227 continue;
1231 break; 1228 break;
1232 } 1229 }
1233 1230
1234 if (!rt) 1231 if (!rt)
1235 rt = net->ipv6.ip6_null_entry; 1232 rt = net->ipv6.ip6_null_entry;
1236 else if (rt->dst.error) { 1233 else if (rt->dst.error) {
1237 rt = net->ipv6.ip6_null_entry; 1234 rt = net->ipv6.ip6_null_entry;
1238 goto out; 1235 goto out;
1239 } 1236 }
1240 BACKTRACK(net, &fl6->saddr); 1237 BACKTRACK(net, &fl6->saddr);
1241 out: 1238 out:
1242 dst_hold(&rt->dst); 1239 dst_hold(&rt->dst);
1243 1240
1244 read_unlock_bh(&table->tb6_lock); 1241 read_unlock_bh(&table->tb6_lock);
1245 1242
1246 return rt; 1243 return rt;
1247 }; 1244 };
1248 1245
1249 static struct dst_entry *ip6_route_redirect(struct net *net, 1246 static struct dst_entry *ip6_route_redirect(struct net *net,
1250 const struct flowi6 *fl6, 1247 const struct flowi6 *fl6,
1251 const struct in6_addr *gateway) 1248 const struct in6_addr *gateway)
1252 { 1249 {
1253 int flags = RT6_LOOKUP_F_HAS_SADDR; 1250 int flags = RT6_LOOKUP_F_HAS_SADDR;
1254 struct ip6rd_flowi rdfl; 1251 struct ip6rd_flowi rdfl;
1255 1252
1256 rdfl.fl6 = *fl6; 1253 rdfl.fl6 = *fl6;
1257 rdfl.gateway = *gateway; 1254 rdfl.gateway = *gateway;
1258 1255
1259 return fib6_rule_lookup(net, &rdfl.fl6, 1256 return fib6_rule_lookup(net, &rdfl.fl6,
1260 flags, __ip6_route_redirect); 1257 flags, __ip6_route_redirect);
1261 } 1258 }
1262 1259
1263 void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark) 1260 void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark)
1264 { 1261 {
1265 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data; 1262 const struct ipv6hdr *iph = (struct ipv6hdr *) skb->data;
1266 struct dst_entry *dst; 1263 struct dst_entry *dst;
1267 struct flowi6 fl6; 1264 struct flowi6 fl6;
1268 1265
1269 memset(&fl6, 0, sizeof(fl6)); 1266 memset(&fl6, 0, sizeof(fl6));
1270 fl6.flowi6_iif = LOOPBACK_IFINDEX; 1267 fl6.flowi6_iif = LOOPBACK_IFINDEX;
1271 fl6.flowi6_oif = oif; 1268 fl6.flowi6_oif = oif;
1272 fl6.flowi6_mark = mark; 1269 fl6.flowi6_mark = mark;
1273 fl6.daddr = iph->daddr; 1270 fl6.daddr = iph->daddr;
1274 fl6.saddr = iph->saddr; 1271 fl6.saddr = iph->saddr;
1275 fl6.flowlabel = ip6_flowinfo(iph); 1272 fl6.flowlabel = ip6_flowinfo(iph);
1276 1273
1277 dst = ip6_route_redirect(net, &fl6, &ipv6_hdr(skb)->saddr); 1274 dst = ip6_route_redirect(net, &fl6, &ipv6_hdr(skb)->saddr);
1278 rt6_do_redirect(dst, NULL, skb); 1275 rt6_do_redirect(dst, NULL, skb);
1279 dst_release(dst); 1276 dst_release(dst);
1280 } 1277 }
1281 EXPORT_SYMBOL_GPL(ip6_redirect); 1278 EXPORT_SYMBOL_GPL(ip6_redirect);
1282 1279
1283 void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif, 1280 void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif,
1284 u32 mark) 1281 u32 mark)
1285 { 1282 {
1286 const struct ipv6hdr *iph = ipv6_hdr(skb); 1283 const struct ipv6hdr *iph = ipv6_hdr(skb);
1287 const struct rd_msg *msg = (struct rd_msg *)icmp6_hdr(skb); 1284 const struct rd_msg *msg = (struct rd_msg *)icmp6_hdr(skb);
1288 struct dst_entry *dst; 1285 struct dst_entry *dst;
1289 struct flowi6 fl6; 1286 struct flowi6 fl6;
1290 1287
1291 memset(&fl6, 0, sizeof(fl6)); 1288 memset(&fl6, 0, sizeof(fl6));
1292 fl6.flowi6_iif = LOOPBACK_IFINDEX; 1289 fl6.flowi6_iif = LOOPBACK_IFINDEX;
1293 fl6.flowi6_oif = oif; 1290 fl6.flowi6_oif = oif;
1294 fl6.flowi6_mark = mark; 1291 fl6.flowi6_mark = mark;
1295 fl6.daddr = msg->dest; 1292 fl6.daddr = msg->dest;
1296 fl6.saddr = iph->daddr; 1293 fl6.saddr = iph->daddr;
1297 1294
1298 dst = ip6_route_redirect(net, &fl6, &iph->saddr); 1295 dst = ip6_route_redirect(net, &fl6, &iph->saddr);
1299 rt6_do_redirect(dst, NULL, skb); 1296 rt6_do_redirect(dst, NULL, skb);
1300 dst_release(dst); 1297 dst_release(dst);
1301 } 1298 }
1302 1299
1303 void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk) 1300 void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk)
1304 { 1301 {
1305 ip6_redirect(skb, sock_net(sk), sk->sk_bound_dev_if, sk->sk_mark); 1302 ip6_redirect(skb, sock_net(sk), sk->sk_bound_dev_if, sk->sk_mark);
1306 } 1303 }
1307 EXPORT_SYMBOL_GPL(ip6_sk_redirect); 1304 EXPORT_SYMBOL_GPL(ip6_sk_redirect);
1308 1305
1309 static unsigned int ip6_default_advmss(const struct dst_entry *dst) 1306 static unsigned int ip6_default_advmss(const struct dst_entry *dst)
1310 { 1307 {
1311 struct net_device *dev = dst->dev; 1308 struct net_device *dev = dst->dev;
1312 unsigned int mtu = dst_mtu(dst); 1309 unsigned int mtu = dst_mtu(dst);
1313 struct net *net = dev_net(dev); 1310 struct net *net = dev_net(dev);
1314 1311
1315 mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr); 1312 mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr);
1316 1313
1317 if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss) 1314 if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss)
1318 mtu = net->ipv6.sysctl.ip6_rt_min_advmss; 1315 mtu = net->ipv6.sysctl.ip6_rt_min_advmss;
1319 1316
1320 /* 1317 /*
1321 * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and 1318 * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and
1322 * corresponding MSS is IPV6_MAXPLEN - tcp_header_size. 1319 * corresponding MSS is IPV6_MAXPLEN - tcp_header_size.
1323 * IPV6_MAXPLEN is also valid and means: "any MSS, 1320 * IPV6_MAXPLEN is also valid and means: "any MSS,
1324 * rely only on pmtu discovery" 1321 * rely only on pmtu discovery"
1325 */ 1322 */
1326 if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr)) 1323 if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr))
1327 mtu = IPV6_MAXPLEN; 1324 mtu = IPV6_MAXPLEN;
1328 return mtu; 1325 return mtu;
1329 } 1326 }
1330 1327
1331 static unsigned int ip6_mtu(const struct dst_entry *dst) 1328 static unsigned int ip6_mtu(const struct dst_entry *dst)
1332 { 1329 {
1333 struct inet6_dev *idev; 1330 struct inet6_dev *idev;
1334 unsigned int mtu = dst_metric_raw(dst, RTAX_MTU); 1331 unsigned int mtu = dst_metric_raw(dst, RTAX_MTU);
1335 1332
1336 if (mtu) 1333 if (mtu)
1337 goto out; 1334 goto out;
1338 1335
1339 mtu = IPV6_MIN_MTU; 1336 mtu = IPV6_MIN_MTU;
1340 1337
1341 rcu_read_lock(); 1338 rcu_read_lock();
1342 idev = __in6_dev_get(dst->dev); 1339 idev = __in6_dev_get(dst->dev);
1343 if (idev) 1340 if (idev)
1344 mtu = idev->cnf.mtu6; 1341 mtu = idev->cnf.mtu6;
1345 rcu_read_unlock(); 1342 rcu_read_unlock();
1346 1343
1347 out: 1344 out:
1348 return min_t(unsigned int, mtu, IP6_MAX_MTU); 1345 return min_t(unsigned int, mtu, IP6_MAX_MTU);
1349 } 1346 }
1350 1347
1351 static struct dst_entry *icmp6_dst_gc_list; 1348 static struct dst_entry *icmp6_dst_gc_list;
1352 static DEFINE_SPINLOCK(icmp6_dst_lock); 1349 static DEFINE_SPINLOCK(icmp6_dst_lock);
1353 1350
1354 struct dst_entry *icmp6_dst_alloc(struct net_device *dev, 1351 struct dst_entry *icmp6_dst_alloc(struct net_device *dev,
1355 struct flowi6 *fl6) 1352 struct flowi6 *fl6)
1356 { 1353 {
1357 struct dst_entry *dst; 1354 struct dst_entry *dst;
1358 struct rt6_info *rt; 1355 struct rt6_info *rt;
1359 struct inet6_dev *idev = in6_dev_get(dev); 1356 struct inet6_dev *idev = in6_dev_get(dev);
1360 struct net *net = dev_net(dev); 1357 struct net *net = dev_net(dev);
1361 1358
1362 if (unlikely(!idev)) 1359 if (unlikely(!idev))
1363 return ERR_PTR(-ENODEV); 1360 return ERR_PTR(-ENODEV);
1364 1361
1365 rt = ip6_dst_alloc(net, dev, 0, NULL); 1362 rt = ip6_dst_alloc(net, dev, 0, NULL);
1366 if (unlikely(!rt)) { 1363 if (unlikely(!rt)) {
1367 in6_dev_put(idev); 1364 in6_dev_put(idev);
1368 dst = ERR_PTR(-ENOMEM); 1365 dst = ERR_PTR(-ENOMEM);
1369 goto out; 1366 goto out;
1370 } 1367 }
1371 1368
1372 rt->dst.flags |= DST_HOST; 1369 rt->dst.flags |= DST_HOST;
1373 rt->dst.output = ip6_output; 1370 rt->dst.output = ip6_output;
1374 atomic_set(&rt->dst.__refcnt, 1); 1371 atomic_set(&rt->dst.__refcnt, 1);
1375 rt->rt6i_gateway = fl6->daddr; 1372 rt->rt6i_gateway = fl6->daddr;
1376 rt->rt6i_dst.addr = fl6->daddr; 1373 rt->rt6i_dst.addr = fl6->daddr;
1377 rt->rt6i_dst.plen = 128; 1374 rt->rt6i_dst.plen = 128;
1378 rt->rt6i_idev = idev; 1375 rt->rt6i_idev = idev;
1379 dst_metric_set(&rt->dst, RTAX_HOPLIMIT, 0); 1376 dst_metric_set(&rt->dst, RTAX_HOPLIMIT, 0);
1380 1377
1381 spin_lock_bh(&icmp6_dst_lock); 1378 spin_lock_bh(&icmp6_dst_lock);
1382 rt->dst.next = icmp6_dst_gc_list; 1379 rt->dst.next = icmp6_dst_gc_list;
1383 icmp6_dst_gc_list = &rt->dst; 1380 icmp6_dst_gc_list = &rt->dst;
1384 spin_unlock_bh(&icmp6_dst_lock); 1381 spin_unlock_bh(&icmp6_dst_lock);
1385 1382
1386 fib6_force_start_gc(net); 1383 fib6_force_start_gc(net);
1387 1384
1388 dst = xfrm_lookup(net, &rt->dst, flowi6_to_flowi(fl6), NULL, 0); 1385 dst = xfrm_lookup(net, &rt->dst, flowi6_to_flowi(fl6), NULL, 0);
1389 1386
1390 out: 1387 out:
1391 return dst; 1388 return dst;
1392 } 1389 }
1393 1390
1394 int icmp6_dst_gc(void) 1391 int icmp6_dst_gc(void)
1395 { 1392 {
1396 struct dst_entry *dst, **pprev; 1393 struct dst_entry *dst, **pprev;
1397 int more = 0; 1394 int more = 0;
1398 1395
1399 spin_lock_bh(&icmp6_dst_lock); 1396 spin_lock_bh(&icmp6_dst_lock);
1400 pprev = &icmp6_dst_gc_list; 1397 pprev = &icmp6_dst_gc_list;
1401 1398
1402 while ((dst = *pprev) != NULL) { 1399 while ((dst = *pprev) != NULL) {
1403 if (!atomic_read(&dst->__refcnt)) { 1400 if (!atomic_read(&dst->__refcnt)) {
1404 *pprev = dst->next; 1401 *pprev = dst->next;
1405 dst_free(dst); 1402 dst_free(dst);
1406 } else { 1403 } else {
1407 pprev = &dst->next; 1404 pprev = &dst->next;
1408 ++more; 1405 ++more;
1409 } 1406 }
1410 } 1407 }
1411 1408
1412 spin_unlock_bh(&icmp6_dst_lock); 1409 spin_unlock_bh(&icmp6_dst_lock);
1413 1410
1414 return more; 1411 return more;
1415 } 1412 }
1416 1413
1417 static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg), 1414 static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg),
1418 void *arg) 1415 void *arg)
1419 { 1416 {
1420 struct dst_entry *dst, **pprev; 1417 struct dst_entry *dst, **pprev;
1421 1418
1422 spin_lock_bh(&icmp6_dst_lock); 1419 spin_lock_bh(&icmp6_dst_lock);
1423 pprev = &icmp6_dst_gc_list; 1420 pprev = &icmp6_dst_gc_list;
1424 while ((dst = *pprev) != NULL) { 1421 while ((dst = *pprev) != NULL) {
1425 struct rt6_info *rt = (struct rt6_info *) dst; 1422 struct rt6_info *rt = (struct rt6_info *) dst;
1426 if (func(rt, arg)) { 1423 if (func(rt, arg)) {
1427 *pprev = dst->next; 1424 *pprev = dst->next;
1428 dst_free(dst); 1425 dst_free(dst);
1429 } else { 1426 } else {
1430 pprev = &dst->next; 1427 pprev = &dst->next;
1431 } 1428 }
1432 } 1429 }
1433 spin_unlock_bh(&icmp6_dst_lock); 1430 spin_unlock_bh(&icmp6_dst_lock);
1434 } 1431 }
1435 1432
1436 static int ip6_dst_gc(struct dst_ops *ops) 1433 static int ip6_dst_gc(struct dst_ops *ops)
1437 { 1434 {
1438 struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops); 1435 struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops);
1439 int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval; 1436 int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval;
1440 int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size; 1437 int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size;
1441 int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity; 1438 int rt_elasticity = net->ipv6.sysctl.ip6_rt_gc_elasticity;
1442 int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout; 1439 int rt_gc_timeout = net->ipv6.sysctl.ip6_rt_gc_timeout;
1443 unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc; 1440 unsigned long rt_last_gc = net->ipv6.ip6_rt_last_gc;
1444 int entries; 1441 int entries;
1445 1442
1446 entries = dst_entries_get_fast(ops); 1443 entries = dst_entries_get_fast(ops);
1447 if (time_after(rt_last_gc + rt_min_interval, jiffies) && 1444 if (time_after(rt_last_gc + rt_min_interval, jiffies) &&
1448 entries <= rt_max_size) 1445 entries <= rt_max_size)
1449 goto out; 1446 goto out;
1450 1447
1451 net->ipv6.ip6_rt_gc_expire++; 1448 net->ipv6.ip6_rt_gc_expire++;
1452 fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net, true); 1449 fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net, true);
1453 entries = dst_entries_get_slow(ops); 1450 entries = dst_entries_get_slow(ops);
1454 if (entries < ops->gc_thresh) 1451 if (entries < ops->gc_thresh)
1455 net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1; 1452 net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1;
1456 out: 1453 out:
1457 net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity; 1454 net->ipv6.ip6_rt_gc_expire -= net->ipv6.ip6_rt_gc_expire>>rt_elasticity;
1458 return entries > rt_max_size; 1455 return entries > rt_max_size;
1459 } 1456 }
1460 1457
1461 /* 1458 /*
1462 * 1459 *
1463 */ 1460 */
1464 1461
1465 int ip6_route_add(struct fib6_config *cfg) 1462 int ip6_route_add(struct fib6_config *cfg)
1466 { 1463 {
1467 int err; 1464 int err;
1468 struct net *net = cfg->fc_nlinfo.nl_net; 1465 struct net *net = cfg->fc_nlinfo.nl_net;
1469 struct rt6_info *rt = NULL; 1466 struct rt6_info *rt = NULL;
1470 struct net_device *dev = NULL; 1467 struct net_device *dev = NULL;
1471 struct inet6_dev *idev = NULL; 1468 struct inet6_dev *idev = NULL;
1472 struct fib6_table *table; 1469 struct fib6_table *table;
1473 int addr_type; 1470 int addr_type;
1474 1471
1475 if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128) 1472 if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128)
1476 return -EINVAL; 1473 return -EINVAL;
1477 #ifndef CONFIG_IPV6_SUBTREES 1474 #ifndef CONFIG_IPV6_SUBTREES
1478 if (cfg->fc_src_len) 1475 if (cfg->fc_src_len)
1479 return -EINVAL; 1476 return -EINVAL;
1480 #endif 1477 #endif
1481 if (cfg->fc_ifindex) { 1478 if (cfg->fc_ifindex) {
1482 err = -ENODEV; 1479 err = -ENODEV;
1483 dev = dev_get_by_index(net, cfg->fc_ifindex); 1480 dev = dev_get_by_index(net, cfg->fc_ifindex);
1484 if (!dev) 1481 if (!dev)
1485 goto out; 1482 goto out;
1486 idev = in6_dev_get(dev); 1483 idev = in6_dev_get(dev);
1487 if (!idev) 1484 if (!idev)
1488 goto out; 1485 goto out;
1489 } 1486 }
1490 1487
1491 if (cfg->fc_metric == 0) 1488 if (cfg->fc_metric == 0)
1492 cfg->fc_metric = IP6_RT_PRIO_USER; 1489 cfg->fc_metric = IP6_RT_PRIO_USER;
1493 1490
1494 err = -ENOBUFS; 1491 err = -ENOBUFS;
1495 if (cfg->fc_nlinfo.nlh && 1492 if (cfg->fc_nlinfo.nlh &&
1496 !(cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_CREATE)) { 1493 !(cfg->fc_nlinfo.nlh->nlmsg_flags & NLM_F_CREATE)) {
1497 table = fib6_get_table(net, cfg->fc_table); 1494 table = fib6_get_table(net, cfg->fc_table);
1498 if (!table) { 1495 if (!table) {
1499 pr_warn("NLM_F_CREATE should be specified when creating new route\n"); 1496 pr_warn("NLM_F_CREATE should be specified when creating new route\n");
1500 table = fib6_new_table(net, cfg->fc_table); 1497 table = fib6_new_table(net, cfg->fc_table);
1501 } 1498 }
1502 } else { 1499 } else {
1503 table = fib6_new_table(net, cfg->fc_table); 1500 table = fib6_new_table(net, cfg->fc_table);
1504 } 1501 }
1505 1502
1506 if (!table) 1503 if (!table)
1507 goto out; 1504 goto out;
1508 1505
1509 rt = ip6_dst_alloc(net, NULL, (cfg->fc_flags & RTF_ADDRCONF) ? 0 : DST_NOCOUNT, table); 1506 rt = ip6_dst_alloc(net, NULL, (cfg->fc_flags & RTF_ADDRCONF) ? 0 : DST_NOCOUNT, table);
1510 1507
1511 if (!rt) { 1508 if (!rt) {
1512 err = -ENOMEM; 1509 err = -ENOMEM;
1513 goto out; 1510 goto out;
1514 } 1511 }
1515 1512
1516 if (cfg->fc_flags & RTF_EXPIRES) 1513 if (cfg->fc_flags & RTF_EXPIRES)
1517 rt6_set_expires(rt, jiffies + 1514 rt6_set_expires(rt, jiffies +
1518 clock_t_to_jiffies(cfg->fc_expires)); 1515 clock_t_to_jiffies(cfg->fc_expires));
1519 else 1516 else
1520 rt6_clean_expires(rt); 1517 rt6_clean_expires(rt);
1521 1518
1522 if (cfg->fc_protocol == RTPROT_UNSPEC) 1519 if (cfg->fc_protocol == RTPROT_UNSPEC)
1523 cfg->fc_protocol = RTPROT_BOOT; 1520 cfg->fc_protocol = RTPROT_BOOT;
1524 rt->rt6i_protocol = cfg->fc_protocol; 1521 rt->rt6i_protocol = cfg->fc_protocol;
1525 1522
1526 addr_type = ipv6_addr_type(&cfg->fc_dst); 1523 addr_type = ipv6_addr_type(&cfg->fc_dst);
1527 1524
1528 if (addr_type & IPV6_ADDR_MULTICAST) 1525 if (addr_type & IPV6_ADDR_MULTICAST)
1529 rt->dst.input = ip6_mc_input; 1526 rt->dst.input = ip6_mc_input;
1530 else if (cfg->fc_flags & RTF_LOCAL) 1527 else if (cfg->fc_flags & RTF_LOCAL)
1531 rt->dst.input = ip6_input; 1528 rt->dst.input = ip6_input;
1532 else 1529 else
1533 rt->dst.input = ip6_forward; 1530 rt->dst.input = ip6_forward;
1534 1531
1535 rt->dst.output = ip6_output; 1532 rt->dst.output = ip6_output;
1536 1533
1537 ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len); 1534 ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len);
1538 rt->rt6i_dst.plen = cfg->fc_dst_len; 1535 rt->rt6i_dst.plen = cfg->fc_dst_len;
1539 if (rt->rt6i_dst.plen == 128) { 1536 if (rt->rt6i_dst.plen == 128) {
1540 rt->dst.flags |= DST_HOST; 1537 rt->dst.flags |= DST_HOST;
1541 dst_metrics_set_force_overwrite(&rt->dst); 1538 dst_metrics_set_force_overwrite(&rt->dst);
1542 } 1539 }
1543 1540
1544 #ifdef CONFIG_IPV6_SUBTREES 1541 #ifdef CONFIG_IPV6_SUBTREES
1545 ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len); 1542 ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len);
1546 rt->rt6i_src.plen = cfg->fc_src_len; 1543 rt->rt6i_src.plen = cfg->fc_src_len;
1547 #endif 1544 #endif
1548 1545
1549 rt->rt6i_metric = cfg->fc_metric; 1546 rt->rt6i_metric = cfg->fc_metric;
1550 1547
1551 /* We cannot add true routes via loopback here, 1548 /* We cannot add true routes via loopback here,
1552 they would result in kernel looping; promote them to reject routes 1549 they would result in kernel looping; promote them to reject routes
1553 */ 1550 */
1554 if ((cfg->fc_flags & RTF_REJECT) || 1551 if ((cfg->fc_flags & RTF_REJECT) ||
1555 (dev && (dev->flags & IFF_LOOPBACK) && 1552 (dev && (dev->flags & IFF_LOOPBACK) &&
1556 !(addr_type & IPV6_ADDR_LOOPBACK) && 1553 !(addr_type & IPV6_ADDR_LOOPBACK) &&
1557 !(cfg->fc_flags & RTF_LOCAL))) { 1554 !(cfg->fc_flags & RTF_LOCAL))) {
1558 /* hold loopback dev/idev if we haven't done so. */ 1555 /* hold loopback dev/idev if we haven't done so. */
1559 if (dev != net->loopback_dev) { 1556 if (dev != net->loopback_dev) {
1560 if (dev) { 1557 if (dev) {
1561 dev_put(dev); 1558 dev_put(dev);
1562 in6_dev_put(idev); 1559 in6_dev_put(idev);
1563 } 1560 }
1564 dev = net->loopback_dev; 1561 dev = net->loopback_dev;
1565 dev_hold(dev); 1562 dev_hold(dev);
1566 idev = in6_dev_get(dev); 1563 idev = in6_dev_get(dev);
1567 if (!idev) { 1564 if (!idev) {
1568 err = -ENODEV; 1565 err = -ENODEV;
1569 goto out; 1566 goto out;
1570 } 1567 }
1571 } 1568 }
1572 rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP; 1569 rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP;
1573 switch (cfg->fc_type) { 1570 switch (cfg->fc_type) {
1574 case RTN_BLACKHOLE: 1571 case RTN_BLACKHOLE:
1575 rt->dst.error = -EINVAL; 1572 rt->dst.error = -EINVAL;
1576 rt->dst.output = dst_discard_sk; 1573 rt->dst.output = dst_discard_sk;
1577 rt->dst.input = dst_discard; 1574 rt->dst.input = dst_discard;
1578 break; 1575 break;
1579 case RTN_PROHIBIT: 1576 case RTN_PROHIBIT:
1580 rt->dst.error = -EACCES; 1577 rt->dst.error = -EACCES;
1581 rt->dst.output = ip6_pkt_prohibit_out; 1578 rt->dst.output = ip6_pkt_prohibit_out;
1582 rt->dst.input = ip6_pkt_prohibit; 1579 rt->dst.input = ip6_pkt_prohibit;
1583 break; 1580 break;
1584 case RTN_THROW: 1581 case RTN_THROW:
1585 default: 1582 default:
1586 rt->dst.error = (cfg->fc_type == RTN_THROW) ? -EAGAIN 1583 rt->dst.error = (cfg->fc_type == RTN_THROW) ? -EAGAIN
1587 : -ENETUNREACH; 1584 : -ENETUNREACH;
1588 rt->dst.output = ip6_pkt_discard_out; 1585 rt->dst.output = ip6_pkt_discard_out;
1589 rt->dst.input = ip6_pkt_discard; 1586 rt->dst.input = ip6_pkt_discard;
1590 break; 1587 break;
1591 } 1588 }
1592 goto install_route; 1589 goto install_route;
1593 } 1590 }
1594 1591
1595 if (cfg->fc_flags & RTF_GATEWAY) { 1592 if (cfg->fc_flags & RTF_GATEWAY) {
1596 const struct in6_addr *gw_addr; 1593 const struct in6_addr *gw_addr;
1597 int gwa_type; 1594 int gwa_type;
1598 1595
1599 gw_addr = &cfg->fc_gateway; 1596 gw_addr = &cfg->fc_gateway;
1600 rt->rt6i_gateway = *gw_addr; 1597 rt->rt6i_gateway = *gw_addr;
1601 gwa_type = ipv6_addr_type(gw_addr); 1598 gwa_type = ipv6_addr_type(gw_addr);
1602 1599
1603 if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) { 1600 if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) {
1604 struct rt6_info *grt; 1601 struct rt6_info *grt;
1605 1602
1606 /* IPv6 strictly inhibits using not link-local 1603 /* IPv6 strictly inhibits using not link-local
1607 addresses as nexthop address. 1604 addresses as nexthop address.
1608 Otherwise, router will not able to send redirects. 1605 Otherwise, router will not able to send redirects.
1609 It is very good, but in some (rare!) circumstances 1606 It is very good, but in some (rare!) circumstances
1610 (SIT, PtP, NBMA NOARP links) it is handy to allow 1607 (SIT, PtP, NBMA NOARP links) it is handy to allow
1611 some exceptions. --ANK 1608 some exceptions. --ANK
1612 */ 1609 */
1613 err = -EINVAL; 1610 err = -EINVAL;
1614 if (!(gwa_type & IPV6_ADDR_UNICAST)) 1611 if (!(gwa_type & IPV6_ADDR_UNICAST))
1615 goto out; 1612 goto out;
1616 1613
1617 grt = rt6_lookup(net, gw_addr, NULL, cfg->fc_ifindex, 1); 1614 grt = rt6_lookup(net, gw_addr, NULL, cfg->fc_ifindex, 1);
1618 1615
1619 err = -EHOSTUNREACH; 1616 err = -EHOSTUNREACH;
1620 if (!grt) 1617 if (!grt)
1621 goto out; 1618 goto out;
1622 if (dev) { 1619 if (dev) {
1623 if (dev != grt->dst.dev) { 1620 if (dev != grt->dst.dev) {
1624 ip6_rt_put(grt); 1621 ip6_rt_put(grt);
1625 goto out; 1622 goto out;
1626 } 1623 }
1627 } else { 1624 } else {
1628 dev = grt->dst.dev; 1625 dev = grt->dst.dev;
1629 idev = grt->rt6i_idev; 1626 idev = grt->rt6i_idev;
1630 dev_hold(dev); 1627 dev_hold(dev);
1631 in6_dev_hold(grt->rt6i_idev); 1628 in6_dev_hold(grt->rt6i_idev);
1632 } 1629 }
1633 if (!(grt->rt6i_flags & RTF_GATEWAY)) 1630 if (!(grt->rt6i_flags & RTF_GATEWAY))
1634 err = 0; 1631 err = 0;
1635 ip6_rt_put(grt); 1632 ip6_rt_put(grt);
1636 1633
1637 if (err) 1634 if (err)
1638 goto out; 1635 goto out;
1639 } 1636 }
1640 err = -EINVAL; 1637 err = -EINVAL;
1641 if (!dev || (dev->flags & IFF_LOOPBACK)) 1638 if (!dev || (dev->flags & IFF_LOOPBACK))
1642 goto out; 1639 goto out;
1643 } 1640 }
1644 1641
1645 err = -ENODEV; 1642 err = -ENODEV;
1646 if (!dev) 1643 if (!dev)
1647 goto out; 1644 goto out;
1648 1645
1649 if (!ipv6_addr_any(&cfg->fc_prefsrc)) { 1646 if (!ipv6_addr_any(&cfg->fc_prefsrc)) {
1650 if (!ipv6_chk_addr(net, &cfg->fc_prefsrc, dev, 0)) { 1647 if (!ipv6_chk_addr(net, &cfg->fc_prefsrc, dev, 0)) {
1651 err = -EINVAL; 1648 err = -EINVAL;
1652 goto out; 1649 goto out;
1653 } 1650 }
1654 rt->rt6i_prefsrc.addr = cfg->fc_prefsrc; 1651 rt->rt6i_prefsrc.addr = cfg->fc_prefsrc;
1655 rt->rt6i_prefsrc.plen = 128; 1652 rt->rt6i_prefsrc.plen = 128;
1656 } else 1653 } else
1657 rt->rt6i_prefsrc.plen = 0; 1654 rt->rt6i_prefsrc.plen = 0;
1658 1655
1659 rt->rt6i_flags = cfg->fc_flags; 1656 rt->rt6i_flags = cfg->fc_flags;
1660 1657
1661 install_route: 1658 install_route:
1662 rt->dst.dev = dev; 1659 rt->dst.dev = dev;
1663 rt->rt6i_idev = idev; 1660 rt->rt6i_idev = idev;
1664 rt->rt6i_table = table; 1661 rt->rt6i_table = table;
1665 1662
1666 cfg->fc_nlinfo.nl_net = dev_net(dev); 1663 cfg->fc_nlinfo.nl_net = dev_net(dev);
1667 1664
1668 return __ip6_ins_rt(rt, &cfg->fc_nlinfo, cfg->fc_mx, cfg->fc_mx_len); 1665 return __ip6_ins_rt(rt, &cfg->fc_nlinfo, cfg->fc_mx, cfg->fc_mx_len);
1669 1666
1670 out: 1667 out:
1671 if (dev) 1668 if (dev)
1672 dev_put(dev); 1669 dev_put(dev);
1673 if (idev) 1670 if (idev)
1674 in6_dev_put(idev); 1671 in6_dev_put(idev);
1675 if (rt) 1672 if (rt)
1676 dst_free(&rt->dst); 1673 dst_free(&rt->dst);
1677 return err; 1674 return err;
1678 } 1675 }
1679 1676
1680 static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info) 1677 static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info)
1681 { 1678 {
1682 int err; 1679 int err;
1683 struct fib6_table *table; 1680 struct fib6_table *table;
1684 struct net *net = dev_net(rt->dst.dev); 1681 struct net *net = dev_net(rt->dst.dev);
1685 1682
1686 if (rt == net->ipv6.ip6_null_entry) { 1683 if (rt == net->ipv6.ip6_null_entry) {
1687 err = -ENOENT; 1684 err = -ENOENT;
1688 goto out; 1685 goto out;
1689 } 1686 }
1690 1687
1691 table = rt->rt6i_table; 1688 table = rt->rt6i_table;
1692 write_lock_bh(&table->tb6_lock); 1689 write_lock_bh(&table->tb6_lock);
1693 err = fib6_del(rt, info); 1690 err = fib6_del(rt, info);
1694 write_unlock_bh(&table->tb6_lock); 1691 write_unlock_bh(&table->tb6_lock);
1695 1692
1696 out: 1693 out:
1697 ip6_rt_put(rt); 1694 ip6_rt_put(rt);
1698 return err; 1695 return err;
1699 } 1696 }
1700 1697
1701 int ip6_del_rt(struct rt6_info *rt) 1698 int ip6_del_rt(struct rt6_info *rt)
1702 { 1699 {
1703 struct nl_info info = { 1700 struct nl_info info = {
1704 .nl_net = dev_net(rt->dst.dev), 1701 .nl_net = dev_net(rt->dst.dev),
1705 }; 1702 };
1706 return __ip6_del_rt(rt, &info); 1703 return __ip6_del_rt(rt, &info);
1707 } 1704 }
1708 1705
1709 static int ip6_route_del(struct fib6_config *cfg) 1706 static int ip6_route_del(struct fib6_config *cfg)
1710 { 1707 {
1711 struct fib6_table *table; 1708 struct fib6_table *table;
1712 struct fib6_node *fn; 1709 struct fib6_node *fn;
1713 struct rt6_info *rt; 1710 struct rt6_info *rt;
1714 int err = -ESRCH; 1711 int err = -ESRCH;
1715 1712
1716 table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table); 1713 table = fib6_get_table(cfg->fc_nlinfo.nl_net, cfg->fc_table);
1717 if (!table) 1714 if (!table)
1718 return err; 1715 return err;
1719 1716
1720 read_lock_bh(&table->tb6_lock); 1717 read_lock_bh(&table->tb6_lock);
1721 1718
1722 fn = fib6_locate(&table->tb6_root, 1719 fn = fib6_locate(&table->tb6_root,
1723 &cfg->fc_dst, cfg->fc_dst_len, 1720 &cfg->fc_dst, cfg->fc_dst_len,
1724 &cfg->fc_src, cfg->fc_src_len); 1721 &cfg->fc_src, cfg->fc_src_len);
1725 1722
1726 if (fn) { 1723 if (fn) {
1727 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) { 1724 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1728 if (cfg->fc_ifindex && 1725 if (cfg->fc_ifindex &&
1729 (!rt->dst.dev || 1726 (!rt->dst.dev ||
1730 rt->dst.dev->ifindex != cfg->fc_ifindex)) 1727 rt->dst.dev->ifindex != cfg->fc_ifindex))
1731 continue; 1728 continue;
1732 if (cfg->fc_flags & RTF_GATEWAY && 1729 if (cfg->fc_flags & RTF_GATEWAY &&
1733 !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway)) 1730 !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway))
1734 continue; 1731 continue;
1735 if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric) 1732 if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric)
1736 continue; 1733 continue;
1737 dst_hold(&rt->dst); 1734 dst_hold(&rt->dst);
1738 read_unlock_bh(&table->tb6_lock); 1735 read_unlock_bh(&table->tb6_lock);
1739 1736
1740 return __ip6_del_rt(rt, &cfg->fc_nlinfo); 1737 return __ip6_del_rt(rt, &cfg->fc_nlinfo);
1741 } 1738 }
1742 } 1739 }
1743 read_unlock_bh(&table->tb6_lock); 1740 read_unlock_bh(&table->tb6_lock);
1744 1741
1745 return err; 1742 return err;
1746 } 1743 }
1747 1744
1748 static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_buff *skb) 1745 static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_buff *skb)
1749 { 1746 {
1750 struct net *net = dev_net(skb->dev); 1747 struct net *net = dev_net(skb->dev);
1751 struct netevent_redirect netevent; 1748 struct netevent_redirect netevent;
1752 struct rt6_info *rt, *nrt = NULL; 1749 struct rt6_info *rt, *nrt = NULL;
1753 struct ndisc_options ndopts; 1750 struct ndisc_options ndopts;
1754 struct inet6_dev *in6_dev; 1751 struct inet6_dev *in6_dev;
1755 struct neighbour *neigh; 1752 struct neighbour *neigh;
1756 struct rd_msg *msg; 1753 struct rd_msg *msg;
1757 int optlen, on_link; 1754 int optlen, on_link;
1758 u8 *lladdr; 1755 u8 *lladdr;
1759 1756
1760 optlen = skb_tail_pointer(skb) - skb_transport_header(skb); 1757 optlen = skb_tail_pointer(skb) - skb_transport_header(skb);
1761 optlen -= sizeof(*msg); 1758 optlen -= sizeof(*msg);
1762 1759
1763 if (optlen < 0) { 1760 if (optlen < 0) {
1764 net_dbg_ratelimited("rt6_do_redirect: packet too short\n"); 1761 net_dbg_ratelimited("rt6_do_redirect: packet too short\n");
1765 return; 1762 return;
1766 } 1763 }
1767 1764
1768 msg = (struct rd_msg *)icmp6_hdr(skb); 1765 msg = (struct rd_msg *)icmp6_hdr(skb);
1769 1766
1770 if (ipv6_addr_is_multicast(&msg->dest)) { 1767 if (ipv6_addr_is_multicast(&msg->dest)) {
1771 net_dbg_ratelimited("rt6_do_redirect: destination address is multicast\n"); 1768 net_dbg_ratelimited("rt6_do_redirect: destination address is multicast\n");
1772 return; 1769 return;
1773 } 1770 }
1774 1771
1775 on_link = 0; 1772 on_link = 0;
1776 if (ipv6_addr_equal(&msg->dest, &msg->target)) { 1773 if (ipv6_addr_equal(&msg->dest, &msg->target)) {
1777 on_link = 1; 1774 on_link = 1;
1778 } else if (ipv6_addr_type(&msg->target) != 1775 } else if (ipv6_addr_type(&msg->target) !=
1779 (IPV6_ADDR_UNICAST|IPV6_ADDR_LINKLOCAL)) { 1776 (IPV6_ADDR_UNICAST|IPV6_ADDR_LINKLOCAL)) {
1780 net_dbg_ratelimited("rt6_do_redirect: target address is not link-local unicast\n"); 1777 net_dbg_ratelimited("rt6_do_redirect: target address is not link-local unicast\n");
1781 return; 1778 return;
1782 } 1779 }
1783 1780
1784 in6_dev = __in6_dev_get(skb->dev); 1781 in6_dev = __in6_dev_get(skb->dev);
1785 if (!in6_dev) 1782 if (!in6_dev)
1786 return; 1783 return;
1787 if (in6_dev->cnf.forwarding || !in6_dev->cnf.accept_redirects) 1784 if (in6_dev->cnf.forwarding || !in6_dev->cnf.accept_redirects)
1788 return; 1785 return;
1789 1786
1790 /* RFC2461 8.1: 1787 /* RFC2461 8.1:
1791 * The IP source address of the Redirect MUST be the same as the current 1788 * The IP source address of the Redirect MUST be the same as the current
1792 * first-hop router for the specified ICMP Destination Address. 1789 * first-hop router for the specified ICMP Destination Address.
1793 */ 1790 */
1794 1791
1795 if (!ndisc_parse_options(msg->opt, optlen, &ndopts)) { 1792 if (!ndisc_parse_options(msg->opt, optlen, &ndopts)) {
1796 net_dbg_ratelimited("rt6_redirect: invalid ND options\n"); 1793 net_dbg_ratelimited("rt6_redirect: invalid ND options\n");
1797 return; 1794 return;
1798 } 1795 }
1799 1796
1800 lladdr = NULL; 1797 lladdr = NULL;
1801 if (ndopts.nd_opts_tgt_lladdr) { 1798 if (ndopts.nd_opts_tgt_lladdr) {
1802 lladdr = ndisc_opt_addr_data(ndopts.nd_opts_tgt_lladdr, 1799 lladdr = ndisc_opt_addr_data(ndopts.nd_opts_tgt_lladdr,
1803 skb->dev); 1800 skb->dev);
1804 if (!lladdr) { 1801 if (!lladdr) {
1805 net_dbg_ratelimited("rt6_redirect: invalid link-layer address length\n"); 1802 net_dbg_ratelimited("rt6_redirect: invalid link-layer address length\n");
1806 return; 1803 return;
1807 } 1804 }
1808 } 1805 }
1809 1806
1810 rt = (struct rt6_info *) dst; 1807 rt = (struct rt6_info *) dst;
1811 if (rt == net->ipv6.ip6_null_entry) { 1808 if (rt == net->ipv6.ip6_null_entry) {
1812 net_dbg_ratelimited("rt6_redirect: source isn't a valid nexthop for redirect target\n"); 1809 net_dbg_ratelimited("rt6_redirect: source isn't a valid nexthop for redirect target\n");
1813 return; 1810 return;
1814 } 1811 }
1815 1812
1816 /* Redirect received -> path was valid. 1813 /* Redirect received -> path was valid.
1817 * Look, redirects are sent only in response to data packets, 1814 * Look, redirects are sent only in response to data packets,
1818 * so that this nexthop apparently is reachable. --ANK 1815 * so that this nexthop apparently is reachable. --ANK
1819 */ 1816 */
1820 dst_confirm(&rt->dst); 1817 dst_confirm(&rt->dst);
1821 1818
1822 neigh = __neigh_lookup(&nd_tbl, &msg->target, skb->dev, 1); 1819 neigh = __neigh_lookup(&nd_tbl, &msg->target, skb->dev, 1);
1823 if (!neigh) 1820 if (!neigh)
1824 return; 1821 return;
1825 1822
1826 /* 1823 /*
1827 * We have finally decided to accept it. 1824 * We have finally decided to accept it.
1828 */ 1825 */
1829 1826
1830 neigh_update(neigh, lladdr, NUD_STALE, 1827 neigh_update(neigh, lladdr, NUD_STALE,
1831 NEIGH_UPDATE_F_WEAK_OVERRIDE| 1828 NEIGH_UPDATE_F_WEAK_OVERRIDE|
1832 NEIGH_UPDATE_F_OVERRIDE| 1829 NEIGH_UPDATE_F_OVERRIDE|
1833 (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER| 1830 (on_link ? 0 : (NEIGH_UPDATE_F_OVERRIDE_ISROUTER|
1834 NEIGH_UPDATE_F_ISROUTER)) 1831 NEIGH_UPDATE_F_ISROUTER))
1835 ); 1832 );
1836 1833
1837 nrt = ip6_rt_copy(rt, &msg->dest); 1834 nrt = ip6_rt_copy(rt, &msg->dest);
1838 if (!nrt) 1835 if (!nrt)
1839 goto out; 1836 goto out;
1840 1837
1841 nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE; 1838 nrt->rt6i_flags = RTF_GATEWAY|RTF_UP|RTF_DYNAMIC|RTF_CACHE;
1842 if (on_link) 1839 if (on_link)
1843 nrt->rt6i_flags &= ~RTF_GATEWAY; 1840 nrt->rt6i_flags &= ~RTF_GATEWAY;
1844 1841
1845 nrt->rt6i_gateway = *(struct in6_addr *)neigh->primary_key; 1842 nrt->rt6i_gateway = *(struct in6_addr *)neigh->primary_key;
1846 1843
1847 if (ip6_ins_rt(nrt)) 1844 if (ip6_ins_rt(nrt))
1848 goto out; 1845 goto out;
1849 1846
1850 netevent.old = &rt->dst; 1847 netevent.old = &rt->dst;
1851 netevent.new = &nrt->dst; 1848 netevent.new = &nrt->dst;
1852 netevent.daddr = &msg->dest; 1849 netevent.daddr = &msg->dest;
1853 netevent.neigh = neigh; 1850 netevent.neigh = neigh;
1854 call_netevent_notifiers(NETEVENT_REDIRECT, &netevent); 1851 call_netevent_notifiers(NETEVENT_REDIRECT, &netevent);
1855 1852
1856 if (rt->rt6i_flags & RTF_CACHE) { 1853 if (rt->rt6i_flags & RTF_CACHE) {
1857 rt = (struct rt6_info *) dst_clone(&rt->dst); 1854 rt = (struct rt6_info *) dst_clone(&rt->dst);
1858 ip6_del_rt(rt); 1855 ip6_del_rt(rt);
1859 } 1856 }
1860 1857
1861 out: 1858 out:
1862 neigh_release(neigh); 1859 neigh_release(neigh);
1863 } 1860 }
1864 1861
1865 /* 1862 /*
1866 * Misc support functions 1863 * Misc support functions
1867 */ 1864 */
1868 1865
1869 static struct rt6_info *ip6_rt_copy(struct rt6_info *ort, 1866 static struct rt6_info *ip6_rt_copy(struct rt6_info *ort,
1870 const struct in6_addr *dest) 1867 const struct in6_addr *dest)
1871 { 1868 {
1872 struct net *net = dev_net(ort->dst.dev); 1869 struct net *net = dev_net(ort->dst.dev);
1873 struct rt6_info *rt = ip6_dst_alloc(net, ort->dst.dev, 0, 1870 struct rt6_info *rt = ip6_dst_alloc(net, ort->dst.dev, 0,
1874 ort->rt6i_table); 1871 ort->rt6i_table);
1875 1872
1876 if (rt) { 1873 if (rt) {
1877 rt->dst.input = ort->dst.input; 1874 rt->dst.input = ort->dst.input;
1878 rt->dst.output = ort->dst.output; 1875 rt->dst.output = ort->dst.output;
1879 rt->dst.flags |= DST_HOST; 1876 rt->dst.flags |= DST_HOST;
1880 1877
1881 rt->rt6i_dst.addr = *dest; 1878 rt->rt6i_dst.addr = *dest;
1882 rt->rt6i_dst.plen = 128; 1879 rt->rt6i_dst.plen = 128;
1883 dst_copy_metrics(&rt->dst, &ort->dst); 1880 dst_copy_metrics(&rt->dst, &ort->dst);
1884 rt->dst.error = ort->dst.error; 1881 rt->dst.error = ort->dst.error;
1885 rt->rt6i_idev = ort->rt6i_idev; 1882 rt->rt6i_idev = ort->rt6i_idev;
1886 if (rt->rt6i_idev) 1883 if (rt->rt6i_idev)
1887 in6_dev_hold(rt->rt6i_idev); 1884 in6_dev_hold(rt->rt6i_idev);
1888 rt->dst.lastuse = jiffies; 1885 rt->dst.lastuse = jiffies;
1889 1886
1890 if (ort->rt6i_flags & RTF_GATEWAY) 1887 if (ort->rt6i_flags & RTF_GATEWAY)
1891 rt->rt6i_gateway = ort->rt6i_gateway; 1888 rt->rt6i_gateway = ort->rt6i_gateway;
1892 else 1889 else
1893 rt->rt6i_gateway = *dest; 1890 rt->rt6i_gateway = *dest;
1894 rt->rt6i_flags = ort->rt6i_flags; 1891 rt->rt6i_flags = ort->rt6i_flags;
1895 rt6_set_from(rt, ort); 1892 rt6_set_from(rt, ort);
1896 rt->rt6i_metric = 0; 1893 rt->rt6i_metric = 0;
1897 1894
1898 #ifdef CONFIG_IPV6_SUBTREES 1895 #ifdef CONFIG_IPV6_SUBTREES
1899 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key)); 1896 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1900 #endif 1897 #endif
1901 memcpy(&rt->rt6i_prefsrc, &ort->rt6i_prefsrc, sizeof(struct rt6key)); 1898 memcpy(&rt->rt6i_prefsrc, &ort->rt6i_prefsrc, sizeof(struct rt6key));
1902 rt->rt6i_table = ort->rt6i_table; 1899 rt->rt6i_table = ort->rt6i_table;
1903 } 1900 }
1904 return rt; 1901 return rt;
1905 } 1902 }
1906 1903
1907 #ifdef CONFIG_IPV6_ROUTE_INFO 1904 #ifdef CONFIG_IPV6_ROUTE_INFO
1908 static struct rt6_info *rt6_get_route_info(struct net *net, 1905 static struct rt6_info *rt6_get_route_info(struct net *net,
1909 const struct in6_addr *prefix, int prefixlen, 1906 const struct in6_addr *prefix, int prefixlen,
1910 const struct in6_addr *gwaddr, int ifindex) 1907 const struct in6_addr *gwaddr, int ifindex)
1911 { 1908 {
1912 struct fib6_node *fn; 1909 struct fib6_node *fn;
1913 struct rt6_info *rt = NULL; 1910 struct rt6_info *rt = NULL;
1914 struct fib6_table *table; 1911 struct fib6_table *table;
1915 1912
1916 table = fib6_get_table(net, RT6_TABLE_INFO); 1913 table = fib6_get_table(net, RT6_TABLE_INFO);
1917 if (!table) 1914 if (!table)
1918 return NULL; 1915 return NULL;
1919 1916
1920 read_lock_bh(&table->tb6_lock); 1917 read_lock_bh(&table->tb6_lock);
1921 fn = fib6_locate(&table->tb6_root, prefix, prefixlen, NULL, 0); 1918 fn = fib6_locate(&table->tb6_root, prefix, prefixlen, NULL, 0);
1922 if (!fn) 1919 if (!fn)
1923 goto out; 1920 goto out;
1924 1921
1925 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) { 1922 for (rt = fn->leaf; rt; rt = rt->dst.rt6_next) {
1926 if (rt->dst.dev->ifindex != ifindex) 1923 if (rt->dst.dev->ifindex != ifindex)
1927 continue; 1924 continue;
1928 if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY)) 1925 if ((rt->rt6i_flags & (RTF_ROUTEINFO|RTF_GATEWAY)) != (RTF_ROUTEINFO|RTF_GATEWAY))
1929 continue; 1926 continue;
1930 if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr)) 1927 if (!ipv6_addr_equal(&rt->rt6i_gateway, gwaddr))
1931 continue; 1928 continue;
1932 dst_hold(&rt->dst); 1929 dst_hold(&rt->dst);
1933 break; 1930 break;
1934 } 1931 }
1935 out: 1932 out:
1936 read_unlock_bh(&table->tb6_lock); 1933 read_unlock_bh(&table->tb6_lock);
1937 return rt; 1934 return rt;
1938 } 1935 }
1939 1936
1940 static struct rt6_info *rt6_add_route_info(struct net *net, 1937 static struct rt6_info *rt6_add_route_info(struct net *net,
1941 const struct in6_addr *prefix, int prefixlen, 1938 const struct in6_addr *prefix, int prefixlen,
1942 const struct in6_addr *gwaddr, int ifindex, 1939 const struct in6_addr *gwaddr, int ifindex,
1943 unsigned int pref) 1940 unsigned int pref)
1944 { 1941 {
1945 struct fib6_config cfg = { 1942 struct fib6_config cfg = {
1946 .fc_table = RT6_TABLE_INFO, 1943 .fc_table = RT6_TABLE_INFO,
1947 .fc_metric = IP6_RT_PRIO_USER, 1944 .fc_metric = IP6_RT_PRIO_USER,
1948 .fc_ifindex = ifindex, 1945 .fc_ifindex = ifindex,
1949 .fc_dst_len = prefixlen, 1946 .fc_dst_len = prefixlen,
1950 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO | 1947 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO |
1951 RTF_UP | RTF_PREF(pref), 1948 RTF_UP | RTF_PREF(pref),
1952 .fc_nlinfo.portid = 0, 1949 .fc_nlinfo.portid = 0,
1953 .fc_nlinfo.nlh = NULL, 1950 .fc_nlinfo.nlh = NULL,
1954 .fc_nlinfo.nl_net = net, 1951 .fc_nlinfo.nl_net = net,
1955 }; 1952 };
1956 1953
1957 cfg.fc_dst = *prefix; 1954 cfg.fc_dst = *prefix;
1958 cfg.fc_gateway = *gwaddr; 1955 cfg.fc_gateway = *gwaddr;
1959 1956
1960 /* We should treat it as a default route if prefix length is 0. */ 1957 /* We should treat it as a default route if prefix length is 0. */
1961 if (!prefixlen) 1958 if (!prefixlen)
1962 cfg.fc_flags |= RTF_DEFAULT; 1959 cfg.fc_flags |= RTF_DEFAULT;
1963 1960
1964 ip6_route_add(&cfg); 1961 ip6_route_add(&cfg);
1965 1962
1966 return rt6_get_route_info(net, prefix, prefixlen, gwaddr, ifindex); 1963 return rt6_get_route_info(net, prefix, prefixlen, gwaddr, ifindex);
1967 } 1964 }
1968 #endif 1965 #endif
1969 1966
1970 struct rt6_info *rt6_get_dflt_router(const struct in6_addr *addr, struct net_device *dev) 1967 struct rt6_info *rt6_get_dflt_router(const struct in6_addr *addr, struct net_device *dev)
1971 { 1968 {
1972 struct rt6_info *rt; 1969 struct rt6_info *rt;
1973 struct fib6_table *table; 1970 struct fib6_table *table;
1974 1971
1975 table = fib6_get_table(dev_net(dev), RT6_TABLE_DFLT); 1972 table = fib6_get_table(dev_net(dev), RT6_TABLE_DFLT);
1976 if (!table) 1973 if (!table)
1977 return NULL; 1974 return NULL;
1978 1975
1979 read_lock_bh(&table->tb6_lock); 1976 read_lock_bh(&table->tb6_lock);
1980 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) { 1977 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
1981 if (dev == rt->dst.dev && 1978 if (dev == rt->dst.dev &&
1982 ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) && 1979 ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) &&
1983 ipv6_addr_equal(&rt->rt6i_gateway, addr)) 1980 ipv6_addr_equal(&rt->rt6i_gateway, addr))
1984 break; 1981 break;
1985 } 1982 }
1986 if (rt) 1983 if (rt)
1987 dst_hold(&rt->dst); 1984 dst_hold(&rt->dst);
1988 read_unlock_bh(&table->tb6_lock); 1985 read_unlock_bh(&table->tb6_lock);
1989 return rt; 1986 return rt;
1990 } 1987 }
1991 1988
1992 struct rt6_info *rt6_add_dflt_router(const struct in6_addr *gwaddr, 1989 struct rt6_info *rt6_add_dflt_router(const struct in6_addr *gwaddr,
1993 struct net_device *dev, 1990 struct net_device *dev,
1994 unsigned int pref) 1991 unsigned int pref)
1995 { 1992 {
1996 struct fib6_config cfg = { 1993 struct fib6_config cfg = {
1997 .fc_table = RT6_TABLE_DFLT, 1994 .fc_table = RT6_TABLE_DFLT,
1998 .fc_metric = IP6_RT_PRIO_USER, 1995 .fc_metric = IP6_RT_PRIO_USER,
1999 .fc_ifindex = dev->ifindex, 1996 .fc_ifindex = dev->ifindex,
2000 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT | 1997 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT |
2001 RTF_UP | RTF_EXPIRES | RTF_PREF(pref), 1998 RTF_UP | RTF_EXPIRES | RTF_PREF(pref),
2002 .fc_nlinfo.portid = 0, 1999 .fc_nlinfo.portid = 0,
2003 .fc_nlinfo.nlh = NULL, 2000 .fc_nlinfo.nlh = NULL,
2004 .fc_nlinfo.nl_net = dev_net(dev), 2001 .fc_nlinfo.nl_net = dev_net(dev),
2005 }; 2002 };
2006 2003
2007 cfg.fc_gateway = *gwaddr; 2004 cfg.fc_gateway = *gwaddr;
2008 2005
2009 ip6_route_add(&cfg); 2006 ip6_route_add(&cfg);
2010 2007
2011 return rt6_get_dflt_router(gwaddr, dev); 2008 return rt6_get_dflt_router(gwaddr, dev);
2012 } 2009 }
2013 2010
2014 void rt6_purge_dflt_routers(struct net *net) 2011 void rt6_purge_dflt_routers(struct net *net)
2015 { 2012 {
2016 struct rt6_info *rt; 2013 struct rt6_info *rt;
2017 struct fib6_table *table; 2014 struct fib6_table *table;
2018 2015
2019 /* NOTE: Keep consistent with rt6_get_dflt_router */ 2016 /* NOTE: Keep consistent with rt6_get_dflt_router */
2020 table = fib6_get_table(net, RT6_TABLE_DFLT); 2017 table = fib6_get_table(net, RT6_TABLE_DFLT);
2021 if (!table) 2018 if (!table)
2022 return; 2019 return;
2023 2020
2024 restart: 2021 restart:
2025 read_lock_bh(&table->tb6_lock); 2022 read_lock_bh(&table->tb6_lock);
2026 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) { 2023 for (rt = table->tb6_root.leaf; rt; rt = rt->dst.rt6_next) {
2027 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF) && 2024 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF) &&
2028 (!rt->rt6i_idev || rt->rt6i_idev->cnf.accept_ra != 2)) { 2025 (!rt->rt6i_idev || rt->rt6i_idev->cnf.accept_ra != 2)) {
2029 dst_hold(&rt->dst); 2026 dst_hold(&rt->dst);
2030 read_unlock_bh(&table->tb6_lock); 2027 read_unlock_bh(&table->tb6_lock);
2031 ip6_del_rt(rt); 2028 ip6_del_rt(rt);
2032 goto restart; 2029 goto restart;
2033 } 2030 }
2034 } 2031 }
2035 read_unlock_bh(&table->tb6_lock); 2032 read_unlock_bh(&table->tb6_lock);
2036 } 2033 }
2037 2034
2038 static void rtmsg_to_fib6_config(struct net *net, 2035 static void rtmsg_to_fib6_config(struct net *net,
2039 struct in6_rtmsg *rtmsg, 2036 struct in6_rtmsg *rtmsg,
2040 struct fib6_config *cfg) 2037 struct fib6_config *cfg)
2041 { 2038 {
2042 memset(cfg, 0, sizeof(*cfg)); 2039 memset(cfg, 0, sizeof(*cfg));
2043 2040
2044 cfg->fc_table = RT6_TABLE_MAIN; 2041 cfg->fc_table = RT6_TABLE_MAIN;
2045 cfg->fc_ifindex = rtmsg->rtmsg_ifindex; 2042 cfg->fc_ifindex = rtmsg->rtmsg_ifindex;
2046 cfg->fc_metric = rtmsg->rtmsg_metric; 2043 cfg->fc_metric = rtmsg->rtmsg_metric;
2047 cfg->fc_expires = rtmsg->rtmsg_info; 2044 cfg->fc_expires = rtmsg->rtmsg_info;
2048 cfg->fc_dst_len = rtmsg->rtmsg_dst_len; 2045 cfg->fc_dst_len = rtmsg->rtmsg_dst_len;
2049 cfg->fc_src_len = rtmsg->rtmsg_src_len; 2046 cfg->fc_src_len = rtmsg->rtmsg_src_len;
2050 cfg->fc_flags = rtmsg->rtmsg_flags; 2047 cfg->fc_flags = rtmsg->rtmsg_flags;
2051 2048
2052 cfg->fc_nlinfo.nl_net = net; 2049 cfg->fc_nlinfo.nl_net = net;
2053 2050
2054 cfg->fc_dst = rtmsg->rtmsg_dst; 2051 cfg->fc_dst = rtmsg->rtmsg_dst;
2055 cfg->fc_src = rtmsg->rtmsg_src; 2052 cfg->fc_src = rtmsg->rtmsg_src;
2056 cfg->fc_gateway = rtmsg->rtmsg_gateway; 2053 cfg->fc_gateway = rtmsg->rtmsg_gateway;
2057 } 2054 }
2058 2055
2059 int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg) 2056 int ipv6_route_ioctl(struct net *net, unsigned int cmd, void __user *arg)
2060 { 2057 {
2061 struct fib6_config cfg; 2058 struct fib6_config cfg;
2062 struct in6_rtmsg rtmsg; 2059 struct in6_rtmsg rtmsg;
2063 int err; 2060 int err;
2064 2061
2065 switch (cmd) { 2062 switch (cmd) {
2066 case SIOCADDRT: /* Add a route */ 2063 case SIOCADDRT: /* Add a route */
2067 case SIOCDELRT: /* Delete a route */ 2064 case SIOCDELRT: /* Delete a route */
2068 if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) 2065 if (!ns_capable(net->user_ns, CAP_NET_ADMIN))
2069 return -EPERM; 2066 return -EPERM;
2070 err = copy_from_user(&rtmsg, arg, 2067 err = copy_from_user(&rtmsg, arg,
2071 sizeof(struct in6_rtmsg)); 2068 sizeof(struct in6_rtmsg));
2072 if (err) 2069 if (err)
2073 return -EFAULT; 2070 return -EFAULT;
2074 2071
2075 rtmsg_to_fib6_config(net, &rtmsg, &cfg); 2072 rtmsg_to_fib6_config(net, &rtmsg, &cfg);
2076 2073
2077 rtnl_lock(); 2074 rtnl_lock();
2078 switch (cmd) { 2075 switch (cmd) {
2079 case SIOCADDRT: 2076 case SIOCADDRT:
2080 err = ip6_route_add(&cfg); 2077 err = ip6_route_add(&cfg);
2081 break; 2078 break;
2082 case SIOCDELRT: 2079 case SIOCDELRT:
2083 err = ip6_route_del(&cfg); 2080 err = ip6_route_del(&cfg);
2084 break; 2081 break;
2085 default: 2082 default:
2086 err = -EINVAL; 2083 err = -EINVAL;
2087 } 2084 }
2088 rtnl_unlock(); 2085 rtnl_unlock();
2089 2086
2090 return err; 2087 return err;
2091 } 2088 }
2092 2089
2093 return -EINVAL; 2090 return -EINVAL;
2094 } 2091 }
2095 2092
2096 /* 2093 /*
2097 * Drop the packet on the floor 2094 * Drop the packet on the floor
2098 */ 2095 */
2099 2096
2100 static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes) 2097 static int ip6_pkt_drop(struct sk_buff *skb, u8 code, int ipstats_mib_noroutes)
2101 { 2098 {
2102 int type; 2099 int type;
2103 struct dst_entry *dst = skb_dst(skb); 2100 struct dst_entry *dst = skb_dst(skb);
2104 switch (ipstats_mib_noroutes) { 2101 switch (ipstats_mib_noroutes) {
2105 case IPSTATS_MIB_INNOROUTES: 2102 case IPSTATS_MIB_INNOROUTES:
2106 type = ipv6_addr_type(&ipv6_hdr(skb)->daddr); 2103 type = ipv6_addr_type(&ipv6_hdr(skb)->daddr);
2107 if (type == IPV6_ADDR_ANY) { 2104 if (type == IPV6_ADDR_ANY) {
2108 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst), 2105 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2109 IPSTATS_MIB_INADDRERRORS); 2106 IPSTATS_MIB_INADDRERRORS);
2110 break; 2107 break;
2111 } 2108 }
2112 /* FALLTHROUGH */ 2109 /* FALLTHROUGH */
2113 case IPSTATS_MIB_OUTNOROUTES: 2110 case IPSTATS_MIB_OUTNOROUTES:
2114 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst), 2111 IP6_INC_STATS(dev_net(dst->dev), ip6_dst_idev(dst),
2115 ipstats_mib_noroutes); 2112 ipstats_mib_noroutes);
2116 break; 2113 break;
2117 } 2114 }
2118 icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0); 2115 icmpv6_send(skb, ICMPV6_DEST_UNREACH, code, 0);
2119 kfree_skb(skb); 2116 kfree_skb(skb);
2120 return 0; 2117 return 0;
2121 } 2118 }
2122 2119
2123 static int ip6_pkt_discard(struct sk_buff *skb) 2120 static int ip6_pkt_discard(struct sk_buff *skb)
2124 { 2121 {
2125 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES); 2122 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_INNOROUTES);
2126 } 2123 }
2127 2124
2128 static int ip6_pkt_discard_out(struct sock *sk, struct sk_buff *skb) 2125 static int ip6_pkt_discard_out(struct sock *sk, struct sk_buff *skb)
2129 { 2126 {
2130 skb->dev = skb_dst(skb)->dev; 2127 skb->dev = skb_dst(skb)->dev;
2131 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES); 2128 return ip6_pkt_drop(skb, ICMPV6_NOROUTE, IPSTATS_MIB_OUTNOROUTES);
2132 } 2129 }
2133 2130
2134 static int ip6_pkt_prohibit(struct sk_buff *skb) 2131 static int ip6_pkt_prohibit(struct sk_buff *skb)
2135 { 2132 {
2136 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES); 2133 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_INNOROUTES);
2137 } 2134 }
2138 2135
2139 static int ip6_pkt_prohibit_out(struct sock *sk, struct sk_buff *skb) 2136 static int ip6_pkt_prohibit_out(struct sock *sk, struct sk_buff *skb)
2140 { 2137 {
2141 skb->dev = skb_dst(skb)->dev; 2138 skb->dev = skb_dst(skb)->dev;
2142 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES); 2139 return ip6_pkt_drop(skb, ICMPV6_ADM_PROHIBITED, IPSTATS_MIB_OUTNOROUTES);
2143 } 2140 }
2144 2141
2145 /* 2142 /*
2146 * Allocate a dst for local (unicast / anycast) address. 2143 * Allocate a dst for local (unicast / anycast) address.
2147 */ 2144 */
2148 2145
2149 struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev, 2146 struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
2150 const struct in6_addr *addr, 2147 const struct in6_addr *addr,
2151 bool anycast) 2148 bool anycast)
2152 { 2149 {
2153 struct net *net = dev_net(idev->dev); 2150 struct net *net = dev_net(idev->dev);
2154 struct rt6_info *rt = ip6_dst_alloc(net, net->loopback_dev, 2151 struct rt6_info *rt = ip6_dst_alloc(net, net->loopback_dev,
2155 DST_NOCOUNT, NULL); 2152 DST_NOCOUNT, NULL);
2156 if (!rt) 2153 if (!rt)
2157 return ERR_PTR(-ENOMEM); 2154 return ERR_PTR(-ENOMEM);
2158 2155
2159 in6_dev_hold(idev); 2156 in6_dev_hold(idev);
2160 2157
2161 rt->dst.flags |= DST_HOST; 2158 rt->dst.flags |= DST_HOST;
2162 rt->dst.input = ip6_input; 2159 rt->dst.input = ip6_input;
2163 rt->dst.output = ip6_output; 2160 rt->dst.output = ip6_output;
2164 rt->rt6i_idev = idev; 2161 rt->rt6i_idev = idev;
2165 2162
2166 rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP; 2163 rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP;
2167 if (anycast) 2164 if (anycast)
2168 rt->rt6i_flags |= RTF_ANYCAST; 2165 rt->rt6i_flags |= RTF_ANYCAST;
2169 else 2166 else
2170 rt->rt6i_flags |= RTF_LOCAL; 2167 rt->rt6i_flags |= RTF_LOCAL;
2171 2168
2172 rt->rt6i_gateway = *addr; 2169 rt->rt6i_gateway = *addr;
2173 rt->rt6i_dst.addr = *addr; 2170 rt->rt6i_dst.addr = *addr;
2174 rt->rt6i_dst.plen = 128; 2171 rt->rt6i_dst.plen = 128;
2175 rt->rt6i_table = fib6_get_table(net, RT6_TABLE_LOCAL); 2172 rt->rt6i_table = fib6_get_table(net, RT6_TABLE_LOCAL);
2176 2173
2177 atomic_set(&rt->dst.__refcnt, 1); 2174 atomic_set(&rt->dst.__refcnt, 1);
2178 2175
2179 return rt; 2176 return rt;
2180 } 2177 }
2181 2178
2182 int ip6_route_get_saddr(struct net *net, 2179 int ip6_route_get_saddr(struct net *net,
2183 struct rt6_info *rt, 2180 struct rt6_info *rt,
2184 const struct in6_addr *daddr, 2181 const struct in6_addr *daddr,
2185 unsigned int prefs, 2182 unsigned int prefs,
2186 struct in6_addr *saddr) 2183 struct in6_addr *saddr)
2187 { 2184 {
2188 struct inet6_dev *idev = ip6_dst_idev((struct dst_entry *)rt); 2185 struct inet6_dev *idev = ip6_dst_idev((struct dst_entry *)rt);
2189 int err = 0; 2186 int err = 0;
2190 if (rt->rt6i_prefsrc.plen) 2187 if (rt->rt6i_prefsrc.plen)
2191 *saddr = rt->rt6i_prefsrc.addr; 2188 *saddr = rt->rt6i_prefsrc.addr;
2192 else 2189 else
2193 err = ipv6_dev_get_saddr(net, idev ? idev->dev : NULL, 2190 err = ipv6_dev_get_saddr(net, idev ? idev->dev : NULL,
2194 daddr, prefs, saddr); 2191 daddr, prefs, saddr);
2195 return err; 2192 return err;
2196 } 2193 }
2197 2194
2198 /* remove deleted ip from prefsrc entries */ 2195 /* remove deleted ip from prefsrc entries */
2199 struct arg_dev_net_ip { 2196 struct arg_dev_net_ip {
2200 struct net_device *dev; 2197 struct net_device *dev;
2201 struct net *net; 2198 struct net *net;
2202 struct in6_addr *addr; 2199 struct in6_addr *addr;
2203 }; 2200 };
2204 2201
2205 static int fib6_remove_prefsrc(struct rt6_info *rt, void *arg) 2202 static int fib6_remove_prefsrc(struct rt6_info *rt, void *arg)
2206 { 2203 {
2207 struct net_device *dev = ((struct arg_dev_net_ip *)arg)->dev; 2204 struct net_device *dev = ((struct arg_dev_net_ip *)arg)->dev;
2208 struct net *net = ((struct arg_dev_net_ip *)arg)->net; 2205 struct net *net = ((struct arg_dev_net_ip *)arg)->net;
2209 struct in6_addr *addr = ((struct arg_dev_net_ip *)arg)->addr; 2206 struct in6_addr *addr = ((struct arg_dev_net_ip *)arg)->addr;
2210 2207
2211 if (((void *)rt->dst.dev == dev || !dev) && 2208 if (((void *)rt->dst.dev == dev || !dev) &&
2212 rt != net->ipv6.ip6_null_entry && 2209 rt != net->ipv6.ip6_null_entry &&
2213 ipv6_addr_equal(addr, &rt->rt6i_prefsrc.addr)) { 2210 ipv6_addr_equal(addr, &rt->rt6i_prefsrc.addr)) {
2214 /* remove prefsrc entry */ 2211 /* remove prefsrc entry */
2215 rt->rt6i_prefsrc.plen = 0; 2212 rt->rt6i_prefsrc.plen = 0;
2216 } 2213 }
2217 return 0; 2214 return 0;
2218 } 2215 }
2219 2216
2220 void rt6_remove_prefsrc(struct inet6_ifaddr *ifp) 2217 void rt6_remove_prefsrc(struct inet6_ifaddr *ifp)
2221 { 2218 {
2222 struct net *net = dev_net(ifp->idev->dev); 2219 struct net *net = dev_net(ifp->idev->dev);
2223 struct arg_dev_net_ip adni = { 2220 struct arg_dev_net_ip adni = {
2224 .dev = ifp->idev->dev, 2221 .dev = ifp->idev->dev,
2225 .net = net, 2222 .net = net,
2226 .addr = &ifp->addr, 2223 .addr = &ifp->addr,
2227 }; 2224 };
2228 fib6_clean_all(net, fib6_remove_prefsrc, &adni); 2225 fib6_clean_all(net, fib6_remove_prefsrc, &adni);
2229 } 2226 }
2230 2227
2231 #define RTF_RA_ROUTER (RTF_ADDRCONF | RTF_DEFAULT | RTF_GATEWAY) 2228 #define RTF_RA_ROUTER (RTF_ADDRCONF | RTF_DEFAULT | RTF_GATEWAY)
2232 #define RTF_CACHE_GATEWAY (RTF_GATEWAY | RTF_CACHE) 2229 #define RTF_CACHE_GATEWAY (RTF_GATEWAY | RTF_CACHE)
2233 2230
2234 /* Remove routers and update dst entries when gateway turn into host. */ 2231 /* Remove routers and update dst entries when gateway turn into host. */
2235 static int fib6_clean_tohost(struct rt6_info *rt, void *arg) 2232 static int fib6_clean_tohost(struct rt6_info *rt, void *arg)
2236 { 2233 {
2237 struct in6_addr *gateway = (struct in6_addr *)arg; 2234 struct in6_addr *gateway = (struct in6_addr *)arg;
2238 2235
2239 if ((((rt->rt6i_flags & RTF_RA_ROUTER) == RTF_RA_ROUTER) || 2236 if ((((rt->rt6i_flags & RTF_RA_ROUTER) == RTF_RA_ROUTER) ||
2240 ((rt->rt6i_flags & RTF_CACHE_GATEWAY) == RTF_CACHE_GATEWAY)) && 2237 ((rt->rt6i_flags & RTF_CACHE_GATEWAY) == RTF_CACHE_GATEWAY)) &&
2241 ipv6_addr_equal(gateway, &rt->rt6i_gateway)) { 2238 ipv6_addr_equal(gateway, &rt->rt6i_gateway)) {
2242 return -1; 2239 return -1;
2243 } 2240 }
2244 return 0; 2241 return 0;
2245 } 2242 }
2246 2243
2247 void rt6_clean_tohost(struct net *net, struct in6_addr *gateway) 2244 void rt6_clean_tohost(struct net *net, struct in6_addr *gateway)
2248 { 2245 {
2249 fib6_clean_all(net, fib6_clean_tohost, gateway); 2246 fib6_clean_all(net, fib6_clean_tohost, gateway);
2250 } 2247 }
2251 2248
2252 struct arg_dev_net { 2249 struct arg_dev_net {
2253 struct net_device *dev; 2250 struct net_device *dev;
2254 struct net *net; 2251 struct net *net;
2255 }; 2252 };
2256 2253
2257 static int fib6_ifdown(struct rt6_info *rt, void *arg) 2254 static int fib6_ifdown(struct rt6_info *rt, void *arg)
2258 { 2255 {
2259 const struct arg_dev_net *adn = arg; 2256 const struct arg_dev_net *adn = arg;
2260 const struct net_device *dev = adn->dev; 2257 const struct net_device *dev = adn->dev;
2261 2258
2262 if ((rt->dst.dev == dev || !dev) && 2259 if ((rt->dst.dev == dev || !dev) &&
2263 rt != adn->net->ipv6.ip6_null_entry) 2260 rt != adn->net->ipv6.ip6_null_entry)
2264 return -1; 2261 return -1;
2265 2262
2266 return 0; 2263 return 0;
2267 } 2264 }
2268 2265
2269 void rt6_ifdown(struct net *net, struct net_device *dev) 2266 void rt6_ifdown(struct net *net, struct net_device *dev)
2270 { 2267 {
2271 struct arg_dev_net adn = { 2268 struct arg_dev_net adn = {
2272 .dev = dev, 2269 .dev = dev,
2273 .net = net, 2270 .net = net,
2274 }; 2271 };
2275 2272
2276 fib6_clean_all(net, fib6_ifdown, &adn); 2273 fib6_clean_all(net, fib6_ifdown, &adn);
2277 icmp6_clean_all(fib6_ifdown, &adn); 2274 icmp6_clean_all(fib6_ifdown, &adn);
2278 } 2275 }
2279 2276
2280 struct rt6_mtu_change_arg { 2277 struct rt6_mtu_change_arg {
2281 struct net_device *dev; 2278 struct net_device *dev;
2282 unsigned int mtu; 2279 unsigned int mtu;
2283 }; 2280 };
2284 2281
2285 static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg) 2282 static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg)
2286 { 2283 {
2287 struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg; 2284 struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg;
2288 struct inet6_dev *idev; 2285 struct inet6_dev *idev;
2289 2286
2290 /* In IPv6 pmtu discovery is not optional, 2287 /* In IPv6 pmtu discovery is not optional,
2291 so that RTAX_MTU lock cannot disable it. 2288 so that RTAX_MTU lock cannot disable it.
2292 We still use this lock to block changes 2289 We still use this lock to block changes
2293 caused by addrconf/ndisc. 2290 caused by addrconf/ndisc.
2294 */ 2291 */
2295 2292
2296 idev = __in6_dev_get(arg->dev); 2293 idev = __in6_dev_get(arg->dev);
2297 if (!idev) 2294 if (!idev)
2298 return 0; 2295 return 0;
2299 2296
2300 /* For administrative MTU increase, there is no way to discover 2297 /* For administrative MTU increase, there is no way to discover
2301 IPv6 PMTU increase, so PMTU increase should be updated here. 2298 IPv6 PMTU increase, so PMTU increase should be updated here.
2302 Since RFC 1981 doesn't include administrative MTU increase 2299 Since RFC 1981 doesn't include administrative MTU increase
2303 update PMTU increase is a MUST. (i.e. jumbo frame) 2300 update PMTU increase is a MUST. (i.e. jumbo frame)
2304 */ 2301 */
2305 /* 2302 /*
2306 If new MTU is less than route PMTU, this new MTU will be the 2303 If new MTU is less than route PMTU, this new MTU will be the
2307 lowest MTU in the path, update the route PMTU to reflect PMTU 2304 lowest MTU in the path, update the route PMTU to reflect PMTU
2308 decreases; if new MTU is greater than route PMTU, and the 2305 decreases; if new MTU is greater than route PMTU, and the
2309 old MTU is the lowest MTU in the path, update the route PMTU 2306 old MTU is the lowest MTU in the path, update the route PMTU
2310 to reflect the increase. In this case if the other nodes' MTU 2307 to reflect the increase. In this case if the other nodes' MTU
2311 also have the lowest MTU, TOO BIG MESSAGE will be lead to 2308 also have the lowest MTU, TOO BIG MESSAGE will be lead to
2312 PMTU discouvery. 2309 PMTU discouvery.
2313 */ 2310 */
2314 if (rt->dst.dev == arg->dev && 2311 if (rt->dst.dev == arg->dev &&
2315 !dst_metric_locked(&rt->dst, RTAX_MTU) && 2312 !dst_metric_locked(&rt->dst, RTAX_MTU) &&
2316 (dst_mtu(&rt->dst) >= arg->mtu || 2313 (dst_mtu(&rt->dst) >= arg->mtu ||
2317 (dst_mtu(&rt->dst) < arg->mtu && 2314 (dst_mtu(&rt->dst) < arg->mtu &&
2318 dst_mtu(&rt->dst) == idev->cnf.mtu6))) { 2315 dst_mtu(&rt->dst) == idev->cnf.mtu6))) {
2319 dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu); 2316 dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu);
2320 } 2317 }
2321 return 0; 2318 return 0;
2322 } 2319 }
2323 2320
2324 void rt6_mtu_change(struct net_device *dev, unsigned int mtu) 2321 void rt6_mtu_change(struct net_device *dev, unsigned int mtu)
2325 { 2322 {
2326 struct rt6_mtu_change_arg arg = { 2323 struct rt6_mtu_change_arg arg = {
2327 .dev = dev, 2324 .dev = dev,
2328 .mtu = mtu, 2325 .mtu = mtu,
2329 }; 2326 };
2330 2327
2331 fib6_clean_all(dev_net(dev), rt6_mtu_change_route, &arg); 2328 fib6_clean_all(dev_net(dev), rt6_mtu_change_route, &arg);
2332 } 2329 }
2333 2330
2334 static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = { 2331 static const struct nla_policy rtm_ipv6_policy[RTA_MAX+1] = {
2335 [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) }, 2332 [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) },
2336 [RTA_OIF] = { .type = NLA_U32 }, 2333 [RTA_OIF] = { .type = NLA_U32 },
2337 [RTA_IIF] = { .type = NLA_U32 }, 2334 [RTA_IIF] = { .type = NLA_U32 },
2338 [RTA_PRIORITY] = { .type = NLA_U32 }, 2335 [RTA_PRIORITY] = { .type = NLA_U32 },
2339 [RTA_METRICS] = { .type = NLA_NESTED }, 2336 [RTA_METRICS] = { .type = NLA_NESTED },
2340 [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) }, 2337 [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) },
2341 }; 2338 };
2342 2339
2343 static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh, 2340 static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh,
2344 struct fib6_config *cfg) 2341 struct fib6_config *cfg)
2345 { 2342 {
2346 struct rtmsg *rtm; 2343 struct rtmsg *rtm;
2347 struct nlattr *tb[RTA_MAX+1]; 2344 struct nlattr *tb[RTA_MAX+1];
2348 int err; 2345 int err;
2349 2346
2350 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy); 2347 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2351 if (err < 0) 2348 if (err < 0)
2352 goto errout; 2349 goto errout;
2353 2350
2354 err = -EINVAL; 2351 err = -EINVAL;
2355 rtm = nlmsg_data(nlh); 2352 rtm = nlmsg_data(nlh);
2356 memset(cfg, 0, sizeof(*cfg)); 2353 memset(cfg, 0, sizeof(*cfg));
2357 2354
2358 cfg->fc_table = rtm->rtm_table; 2355 cfg->fc_table = rtm->rtm_table;
2359 cfg->fc_dst_len = rtm->rtm_dst_len; 2356 cfg->fc_dst_len = rtm->rtm_dst_len;
2360 cfg->fc_src_len = rtm->rtm_src_len; 2357 cfg->fc_src_len = rtm->rtm_src_len;
2361 cfg->fc_flags = RTF_UP; 2358 cfg->fc_flags = RTF_UP;
2362 cfg->fc_protocol = rtm->rtm_protocol; 2359 cfg->fc_protocol = rtm->rtm_protocol;
2363 cfg->fc_type = rtm->rtm_type; 2360 cfg->fc_type = rtm->rtm_type;
2364 2361
2365 if (rtm->rtm_type == RTN_UNREACHABLE || 2362 if (rtm->rtm_type == RTN_UNREACHABLE ||
2366 rtm->rtm_type == RTN_BLACKHOLE || 2363 rtm->rtm_type == RTN_BLACKHOLE ||
2367 rtm->rtm_type == RTN_PROHIBIT || 2364 rtm->rtm_type == RTN_PROHIBIT ||
2368 rtm->rtm_type == RTN_THROW) 2365 rtm->rtm_type == RTN_THROW)
2369 cfg->fc_flags |= RTF_REJECT; 2366 cfg->fc_flags |= RTF_REJECT;
2370 2367
2371 if (rtm->rtm_type == RTN_LOCAL) 2368 if (rtm->rtm_type == RTN_LOCAL)
2372 cfg->fc_flags |= RTF_LOCAL; 2369 cfg->fc_flags |= RTF_LOCAL;
2373 2370
2374 cfg->fc_nlinfo.portid = NETLINK_CB(skb).portid; 2371 cfg->fc_nlinfo.portid = NETLINK_CB(skb).portid;
2375 cfg->fc_nlinfo.nlh = nlh; 2372 cfg->fc_nlinfo.nlh = nlh;
2376 cfg->fc_nlinfo.nl_net = sock_net(skb->sk); 2373 cfg->fc_nlinfo.nl_net = sock_net(skb->sk);
2377 2374
2378 if (tb[RTA_GATEWAY]) { 2375 if (tb[RTA_GATEWAY]) {
2379 nla_memcpy(&cfg->fc_gateway, tb[RTA_GATEWAY], 16); 2376 nla_memcpy(&cfg->fc_gateway, tb[RTA_GATEWAY], 16);
2380 cfg->fc_flags |= RTF_GATEWAY; 2377 cfg->fc_flags |= RTF_GATEWAY;
2381 } 2378 }
2382 2379
2383 if (tb[RTA_DST]) { 2380 if (tb[RTA_DST]) {
2384 int plen = (rtm->rtm_dst_len + 7) >> 3; 2381 int plen = (rtm->rtm_dst_len + 7) >> 3;
2385 2382
2386 if (nla_len(tb[RTA_DST]) < plen) 2383 if (nla_len(tb[RTA_DST]) < plen)
2387 goto errout; 2384 goto errout;
2388 2385
2389 nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen); 2386 nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen);
2390 } 2387 }
2391 2388
2392 if (tb[RTA_SRC]) { 2389 if (tb[RTA_SRC]) {
2393 int plen = (rtm->rtm_src_len + 7) >> 3; 2390 int plen = (rtm->rtm_src_len + 7) >> 3;
2394 2391
2395 if (nla_len(tb[RTA_SRC]) < plen) 2392 if (nla_len(tb[RTA_SRC]) < plen)
2396 goto errout; 2393 goto errout;
2397 2394
2398 nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen); 2395 nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen);
2399 } 2396 }
2400 2397
2401 if (tb[RTA_PREFSRC]) 2398 if (tb[RTA_PREFSRC])
2402 nla_memcpy(&cfg->fc_prefsrc, tb[RTA_PREFSRC], 16); 2399 nla_memcpy(&cfg->fc_prefsrc, tb[RTA_PREFSRC], 16);
2403 2400
2404 if (tb[RTA_OIF]) 2401 if (tb[RTA_OIF])
2405 cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]); 2402 cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]);
2406 2403
2407 if (tb[RTA_PRIORITY]) 2404 if (tb[RTA_PRIORITY])
2408 cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]); 2405 cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]);
2409 2406
2410 if (tb[RTA_METRICS]) { 2407 if (tb[RTA_METRICS]) {
2411 cfg->fc_mx = nla_data(tb[RTA_METRICS]); 2408 cfg->fc_mx = nla_data(tb[RTA_METRICS]);
2412 cfg->fc_mx_len = nla_len(tb[RTA_METRICS]); 2409 cfg->fc_mx_len = nla_len(tb[RTA_METRICS]);
2413 } 2410 }
2414 2411
2415 if (tb[RTA_TABLE]) 2412 if (tb[RTA_TABLE])
2416 cfg->fc_table = nla_get_u32(tb[RTA_TABLE]); 2413 cfg->fc_table = nla_get_u32(tb[RTA_TABLE]);
2417 2414
2418 if (tb[RTA_MULTIPATH]) { 2415 if (tb[RTA_MULTIPATH]) {
2419 cfg->fc_mp = nla_data(tb[RTA_MULTIPATH]); 2416 cfg->fc_mp = nla_data(tb[RTA_MULTIPATH]);
2420 cfg->fc_mp_len = nla_len(tb[RTA_MULTIPATH]); 2417 cfg->fc_mp_len = nla_len(tb[RTA_MULTIPATH]);
2421 } 2418 }
2422 2419
2423 err = 0; 2420 err = 0;
2424 errout: 2421 errout:
2425 return err; 2422 return err;
2426 } 2423 }
2427 2424
2428 static int ip6_route_multipath(struct fib6_config *cfg, int add) 2425 static int ip6_route_multipath(struct fib6_config *cfg, int add)
2429 { 2426 {
2430 struct fib6_config r_cfg; 2427 struct fib6_config r_cfg;
2431 struct rtnexthop *rtnh; 2428 struct rtnexthop *rtnh;
2432 int remaining; 2429 int remaining;
2433 int attrlen; 2430 int attrlen;
2434 int err = 0, last_err = 0; 2431 int err = 0, last_err = 0;
2435 2432
2436 beginning: 2433 beginning:
2437 rtnh = (struct rtnexthop *)cfg->fc_mp; 2434 rtnh = (struct rtnexthop *)cfg->fc_mp;
2438 remaining = cfg->fc_mp_len; 2435 remaining = cfg->fc_mp_len;
2439 2436
2440 /* Parse a Multipath Entry */ 2437 /* Parse a Multipath Entry */
2441 while (rtnh_ok(rtnh, remaining)) { 2438 while (rtnh_ok(rtnh, remaining)) {
2442 memcpy(&r_cfg, cfg, sizeof(*cfg)); 2439 memcpy(&r_cfg, cfg, sizeof(*cfg));
2443 if (rtnh->rtnh_ifindex) 2440 if (rtnh->rtnh_ifindex)
2444 r_cfg.fc_ifindex = rtnh->rtnh_ifindex; 2441 r_cfg.fc_ifindex = rtnh->rtnh_ifindex;
2445 2442
2446 attrlen = rtnh_attrlen(rtnh); 2443 attrlen = rtnh_attrlen(rtnh);
2447 if (attrlen > 0) { 2444 if (attrlen > 0) {
2448 struct nlattr *nla, *attrs = rtnh_attrs(rtnh); 2445 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
2449 2446
2450 nla = nla_find(attrs, attrlen, RTA_GATEWAY); 2447 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
2451 if (nla) { 2448 if (nla) {
2452 nla_memcpy(&r_cfg.fc_gateway, nla, 16); 2449 nla_memcpy(&r_cfg.fc_gateway, nla, 16);
2453 r_cfg.fc_flags |= RTF_GATEWAY; 2450 r_cfg.fc_flags |= RTF_GATEWAY;
2454 } 2451 }
2455 } 2452 }
2456 err = add ? ip6_route_add(&r_cfg) : ip6_route_del(&r_cfg); 2453 err = add ? ip6_route_add(&r_cfg) : ip6_route_del(&r_cfg);
2457 if (err) { 2454 if (err) {
2458 last_err = err; 2455 last_err = err;
2459 /* If we are trying to remove a route, do not stop the 2456 /* If we are trying to remove a route, do not stop the
2460 * loop when ip6_route_del() fails (because next hop is 2457 * loop when ip6_route_del() fails (because next hop is
2461 * already gone), we should try to remove all next hops. 2458 * already gone), we should try to remove all next hops.
2462 */ 2459 */
2463 if (add) { 2460 if (add) {
2464 /* If add fails, we should try to delete all 2461 /* If add fails, we should try to delete all
2465 * next hops that have been already added. 2462 * next hops that have been already added.
2466 */ 2463 */
2467 add = 0; 2464 add = 0;
2468 goto beginning; 2465 goto beginning;
2469 } 2466 }
2470 } 2467 }
2471 /* Because each route is added like a single route we remove 2468 /* Because each route is added like a single route we remove
2472 * this flag after the first nexthop (if there is a collision, 2469 * this flag after the first nexthop (if there is a collision,
2473 * we have already fail to add the first nexthop: 2470 * we have already fail to add the first nexthop:
2474 * fib6_add_rt2node() has reject it). 2471 * fib6_add_rt2node() has reject it).
2475 */ 2472 */
2476 cfg->fc_nlinfo.nlh->nlmsg_flags &= ~NLM_F_EXCL; 2473 cfg->fc_nlinfo.nlh->nlmsg_flags &= ~NLM_F_EXCL;
2477 rtnh = rtnh_next(rtnh, &remaining); 2474 rtnh = rtnh_next(rtnh, &remaining);
2478 } 2475 }
2479 2476
2480 return last_err; 2477 return last_err;
2481 } 2478 }
2482 2479
2483 static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr *nlh) 2480 static int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr *nlh)
2484 { 2481 {
2485 struct fib6_config cfg; 2482 struct fib6_config cfg;
2486 int err; 2483 int err;
2487 2484
2488 err = rtm_to_fib6_config(skb, nlh, &cfg); 2485 err = rtm_to_fib6_config(skb, nlh, &cfg);
2489 if (err < 0) 2486 if (err < 0)
2490 return err; 2487 return err;
2491 2488
2492 if (cfg.fc_mp) 2489 if (cfg.fc_mp)
2493 return ip6_route_multipath(&cfg, 0); 2490 return ip6_route_multipath(&cfg, 0);
2494 else 2491 else
2495 return ip6_route_del(&cfg); 2492 return ip6_route_del(&cfg);
2496 } 2493 }
2497 2494
2498 static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr *nlh) 2495 static int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr *nlh)
2499 { 2496 {
2500 struct fib6_config cfg; 2497 struct fib6_config cfg;
2501 int err; 2498 int err;
2502 2499
2503 err = rtm_to_fib6_config(skb, nlh, &cfg); 2500 err = rtm_to_fib6_config(skb, nlh, &cfg);
2504 if (err < 0) 2501 if (err < 0)
2505 return err; 2502 return err;
2506 2503
2507 if (cfg.fc_mp) 2504 if (cfg.fc_mp)
2508 return ip6_route_multipath(&cfg, 1); 2505 return ip6_route_multipath(&cfg, 1);
2509 else 2506 else
2510 return ip6_route_add(&cfg); 2507 return ip6_route_add(&cfg);
2511 } 2508 }
2512 2509
2513 static inline size_t rt6_nlmsg_size(void) 2510 static inline size_t rt6_nlmsg_size(void)
2514 { 2511 {
2515 return NLMSG_ALIGN(sizeof(struct rtmsg)) 2512 return NLMSG_ALIGN(sizeof(struct rtmsg))
2516 + nla_total_size(16) /* RTA_SRC */ 2513 + nla_total_size(16) /* RTA_SRC */
2517 + nla_total_size(16) /* RTA_DST */ 2514 + nla_total_size(16) /* RTA_DST */
2518 + nla_total_size(16) /* RTA_GATEWAY */ 2515 + nla_total_size(16) /* RTA_GATEWAY */
2519 + nla_total_size(16) /* RTA_PREFSRC */ 2516 + nla_total_size(16) /* RTA_PREFSRC */
2520 + nla_total_size(4) /* RTA_TABLE */ 2517 + nla_total_size(4) /* RTA_TABLE */
2521 + nla_total_size(4) /* RTA_IIF */ 2518 + nla_total_size(4) /* RTA_IIF */
2522 + nla_total_size(4) /* RTA_OIF */ 2519 + nla_total_size(4) /* RTA_OIF */
2523 + nla_total_size(4) /* RTA_PRIORITY */ 2520 + nla_total_size(4) /* RTA_PRIORITY */
2524 + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */ 2521 + RTAX_MAX * nla_total_size(4) /* RTA_METRICS */
2525 + nla_total_size(sizeof(struct rta_cacheinfo)); 2522 + nla_total_size(sizeof(struct rta_cacheinfo));
2526 } 2523 }
2527 2524
2528 static int rt6_fill_node(struct net *net, 2525 static int rt6_fill_node(struct net *net,
2529 struct sk_buff *skb, struct rt6_info *rt, 2526 struct sk_buff *skb, struct rt6_info *rt,
2530 struct in6_addr *dst, struct in6_addr *src, 2527 struct in6_addr *dst, struct in6_addr *src,
2531 int iif, int type, u32 portid, u32 seq, 2528 int iif, int type, u32 portid, u32 seq,
2532 int prefix, int nowait, unsigned int flags) 2529 int prefix, int nowait, unsigned int flags)
2533 { 2530 {
2534 struct rtmsg *rtm; 2531 struct rtmsg *rtm;
2535 struct nlmsghdr *nlh; 2532 struct nlmsghdr *nlh;
2536 long expires; 2533 long expires;
2537 u32 table; 2534 u32 table;
2538 2535
2539 if (prefix) { /* user wants prefix routes only */ 2536 if (prefix) { /* user wants prefix routes only */
2540 if (!(rt->rt6i_flags & RTF_PREFIX_RT)) { 2537 if (!(rt->rt6i_flags & RTF_PREFIX_RT)) {
2541 /* success since this is not a prefix route */ 2538 /* success since this is not a prefix route */
2542 return 1; 2539 return 1;
2543 } 2540 }
2544 } 2541 }
2545 2542
2546 nlh = nlmsg_put(skb, portid, seq, type, sizeof(*rtm), flags); 2543 nlh = nlmsg_put(skb, portid, seq, type, sizeof(*rtm), flags);
2547 if (!nlh) 2544 if (!nlh)
2548 return -EMSGSIZE; 2545 return -EMSGSIZE;
2549 2546
2550 rtm = nlmsg_data(nlh); 2547 rtm = nlmsg_data(nlh);
2551 rtm->rtm_family = AF_INET6; 2548 rtm->rtm_family = AF_INET6;
2552 rtm->rtm_dst_len = rt->rt6i_dst.plen; 2549 rtm->rtm_dst_len = rt->rt6i_dst.plen;
2553 rtm->rtm_src_len = rt->rt6i_src.plen; 2550 rtm->rtm_src_len = rt->rt6i_src.plen;
2554 rtm->rtm_tos = 0; 2551 rtm->rtm_tos = 0;
2555 if (rt->rt6i_table) 2552 if (rt->rt6i_table)
2556 table = rt->rt6i_table->tb6_id; 2553 table = rt->rt6i_table->tb6_id;
2557 else 2554 else
2558 table = RT6_TABLE_UNSPEC; 2555 table = RT6_TABLE_UNSPEC;
2559 rtm->rtm_table = table; 2556 rtm->rtm_table = table;
2560 if (nla_put_u32(skb, RTA_TABLE, table)) 2557 if (nla_put_u32(skb, RTA_TABLE, table))
2561 goto nla_put_failure; 2558 goto nla_put_failure;
2562 if (rt->rt6i_flags & RTF_REJECT) { 2559 if (rt->rt6i_flags & RTF_REJECT) {
2563 switch (rt->dst.error) { 2560 switch (rt->dst.error) {
2564 case -EINVAL: 2561 case -EINVAL:
2565 rtm->rtm_type = RTN_BLACKHOLE; 2562 rtm->rtm_type = RTN_BLACKHOLE;
2566 break; 2563 break;
2567 case -EACCES: 2564 case -EACCES:
2568 rtm->rtm_type = RTN_PROHIBIT; 2565 rtm->rtm_type = RTN_PROHIBIT;
2569 break; 2566 break;
2570 case -EAGAIN: 2567 case -EAGAIN:
2571 rtm->rtm_type = RTN_THROW; 2568 rtm->rtm_type = RTN_THROW;
2572 break; 2569 break;
2573 default: 2570 default:
2574 rtm->rtm_type = RTN_UNREACHABLE; 2571 rtm->rtm_type = RTN_UNREACHABLE;
2575 break; 2572 break;
2576 } 2573 }
2577 } 2574 }
2578 else if (rt->rt6i_flags & RTF_LOCAL) 2575 else if (rt->rt6i_flags & RTF_LOCAL)
2579 rtm->rtm_type = RTN_LOCAL; 2576 rtm->rtm_type = RTN_LOCAL;
2580 else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK)) 2577 else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK))
2581 rtm->rtm_type = RTN_LOCAL; 2578 rtm->rtm_type = RTN_LOCAL;
2582 else 2579 else
2583 rtm->rtm_type = RTN_UNICAST; 2580 rtm->rtm_type = RTN_UNICAST;
2584 rtm->rtm_flags = 0; 2581 rtm->rtm_flags = 0;
2585 rtm->rtm_scope = RT_SCOPE_UNIVERSE; 2582 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
2586 rtm->rtm_protocol = rt->rt6i_protocol; 2583 rtm->rtm_protocol = rt->rt6i_protocol;
2587 if (rt->rt6i_flags & RTF_DYNAMIC) 2584 if (rt->rt6i_flags & RTF_DYNAMIC)
2588 rtm->rtm_protocol = RTPROT_REDIRECT; 2585 rtm->rtm_protocol = RTPROT_REDIRECT;
2589 else if (rt->rt6i_flags & RTF_ADDRCONF) { 2586 else if (rt->rt6i_flags & RTF_ADDRCONF) {
2590 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ROUTEINFO)) 2587 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ROUTEINFO))
2591 rtm->rtm_protocol = RTPROT_RA; 2588 rtm->rtm_protocol = RTPROT_RA;
2592 else 2589 else
2593 rtm->rtm_protocol = RTPROT_KERNEL; 2590 rtm->rtm_protocol = RTPROT_KERNEL;
2594 } 2591 }
2595 2592
2596 if (rt->rt6i_flags & RTF_CACHE) 2593 if (rt->rt6i_flags & RTF_CACHE)
2597 rtm->rtm_flags |= RTM_F_CLONED; 2594 rtm->rtm_flags |= RTM_F_CLONED;
2598 2595
2599 if (dst) { 2596 if (dst) {
2600 if (nla_put(skb, RTA_DST, 16, dst)) 2597 if (nla_put(skb, RTA_DST, 16, dst))
2601 goto nla_put_failure; 2598 goto nla_put_failure;
2602 rtm->rtm_dst_len = 128; 2599 rtm->rtm_dst_len = 128;
2603 } else if (rtm->rtm_dst_len) 2600 } else if (rtm->rtm_dst_len)
2604 if (nla_put(skb, RTA_DST, 16, &rt->rt6i_dst.addr)) 2601 if (nla_put(skb, RTA_DST, 16, &rt->rt6i_dst.addr))
2605 goto nla_put_failure; 2602 goto nla_put_failure;
2606 #ifdef CONFIG_IPV6_SUBTREES 2603 #ifdef CONFIG_IPV6_SUBTREES
2607 if (src) { 2604 if (src) {
2608 if (nla_put(skb, RTA_SRC, 16, src)) 2605 if (nla_put(skb, RTA_SRC, 16, src))
2609 goto nla_put_failure; 2606 goto nla_put_failure;
2610 rtm->rtm_src_len = 128; 2607 rtm->rtm_src_len = 128;
2611 } else if (rtm->rtm_src_len && 2608 } else if (rtm->rtm_src_len &&
2612 nla_put(skb, RTA_SRC, 16, &rt->rt6i_src.addr)) 2609 nla_put(skb, RTA_SRC, 16, &rt->rt6i_src.addr))
2613 goto nla_put_failure; 2610 goto nla_put_failure;
2614 #endif 2611 #endif
2615 if (iif) { 2612 if (iif) {
2616 #ifdef CONFIG_IPV6_MROUTE 2613 #ifdef CONFIG_IPV6_MROUTE
2617 if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) { 2614 if (ipv6_addr_is_multicast(&rt->rt6i_dst.addr)) {
2618 int err = ip6mr_get_route(net, skb, rtm, nowait); 2615 int err = ip6mr_get_route(net, skb, rtm, nowait);
2619 if (err <= 0) { 2616 if (err <= 0) {
2620 if (!nowait) { 2617 if (!nowait) {
2621 if (err == 0) 2618 if (err == 0)
2622 return 0; 2619 return 0;
2623 goto nla_put_failure; 2620 goto nla_put_failure;
2624 } else { 2621 } else {
2625 if (err == -EMSGSIZE) 2622 if (err == -EMSGSIZE)
2626 goto nla_put_failure; 2623 goto nla_put_failure;
2627 } 2624 }
2628 } 2625 }
2629 } else 2626 } else
2630 #endif 2627 #endif
2631 if (nla_put_u32(skb, RTA_IIF, iif)) 2628 if (nla_put_u32(skb, RTA_IIF, iif))
2632 goto nla_put_failure; 2629 goto nla_put_failure;
2633 } else if (dst) { 2630 } else if (dst) {
2634 struct in6_addr saddr_buf; 2631 struct in6_addr saddr_buf;
2635 if (ip6_route_get_saddr(net, rt, dst, 0, &saddr_buf) == 0 && 2632 if (ip6_route_get_saddr(net, rt, dst, 0, &saddr_buf) == 0 &&
2636 nla_put(skb, RTA_PREFSRC, 16, &saddr_buf)) 2633 nla_put(skb, RTA_PREFSRC, 16, &saddr_buf))
2637 goto nla_put_failure; 2634 goto nla_put_failure;
2638 } 2635 }
2639 2636
2640 if (rt->rt6i_prefsrc.plen) { 2637 if (rt->rt6i_prefsrc.plen) {
2641 struct in6_addr saddr_buf; 2638 struct in6_addr saddr_buf;
2642 saddr_buf = rt->rt6i_prefsrc.addr; 2639 saddr_buf = rt->rt6i_prefsrc.addr;
2643 if (nla_put(skb, RTA_PREFSRC, 16, &saddr_buf)) 2640 if (nla_put(skb, RTA_PREFSRC, 16, &saddr_buf))
2644 goto nla_put_failure; 2641 goto nla_put_failure;
2645 } 2642 }
2646 2643
2647 if (rtnetlink_put_metrics(skb, dst_metrics_ptr(&rt->dst)) < 0) 2644 if (rtnetlink_put_metrics(skb, dst_metrics_ptr(&rt->dst)) < 0)
2648 goto nla_put_failure; 2645 goto nla_put_failure;
2649 2646
2650 if (rt->rt6i_flags & RTF_GATEWAY) { 2647 if (rt->rt6i_flags & RTF_GATEWAY) {
2651 if (nla_put(skb, RTA_GATEWAY, 16, &rt->rt6i_gateway) < 0) 2648 if (nla_put(skb, RTA_GATEWAY, 16, &rt->rt6i_gateway) < 0)
2652 goto nla_put_failure; 2649 goto nla_put_failure;
2653 } 2650 }
2654 2651
2655 if (rt->dst.dev && 2652 if (rt->dst.dev &&
2656 nla_put_u32(skb, RTA_OIF, rt->dst.dev->ifindex)) 2653 nla_put_u32(skb, RTA_OIF, rt->dst.dev->ifindex))
2657 goto nla_put_failure; 2654 goto nla_put_failure;
2658 if (nla_put_u32(skb, RTA_PRIORITY, rt->rt6i_metric)) 2655 if (nla_put_u32(skb, RTA_PRIORITY, rt->rt6i_metric))
2659 goto nla_put_failure; 2656 goto nla_put_failure;
2660 2657
2661 expires = (rt->rt6i_flags & RTF_EXPIRES) ? rt->dst.expires - jiffies : 0; 2658 expires = (rt->rt6i_flags & RTF_EXPIRES) ? rt->dst.expires - jiffies : 0;
2662 2659
2663 if (rtnl_put_cacheinfo(skb, &rt->dst, 0, expires, rt->dst.error) < 0) 2660 if (rtnl_put_cacheinfo(skb, &rt->dst, 0, expires, rt->dst.error) < 0)
2664 goto nla_put_failure; 2661 goto nla_put_failure;
2665 2662
2666 return nlmsg_end(skb, nlh); 2663 return nlmsg_end(skb, nlh);
2667 2664
2668 nla_put_failure: 2665 nla_put_failure:
2669 nlmsg_cancel(skb, nlh); 2666 nlmsg_cancel(skb, nlh);
2670 return -EMSGSIZE; 2667 return -EMSGSIZE;
2671 } 2668 }
2672 2669
2673 int rt6_dump_route(struct rt6_info *rt, void *p_arg) 2670 int rt6_dump_route(struct rt6_info *rt, void *p_arg)
2674 { 2671 {
2675 struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg; 2672 struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg;
2676 int prefix; 2673 int prefix;
2677 2674
2678 if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) { 2675 if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) {
2679 struct rtmsg *rtm = nlmsg_data(arg->cb->nlh); 2676 struct rtmsg *rtm = nlmsg_data(arg->cb->nlh);
2680 prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0; 2677 prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0;
2681 } else 2678 } else
2682 prefix = 0; 2679 prefix = 0;
2683 2680
2684 return rt6_fill_node(arg->net, 2681 return rt6_fill_node(arg->net,
2685 arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE, 2682 arg->skb, rt, NULL, NULL, 0, RTM_NEWROUTE,
2686 NETLINK_CB(arg->cb->skb).portid, arg->cb->nlh->nlmsg_seq, 2683 NETLINK_CB(arg->cb->skb).portid, arg->cb->nlh->nlmsg_seq,
2687 prefix, 0, NLM_F_MULTI); 2684 prefix, 0, NLM_F_MULTI);
2688 } 2685 }
2689 2686
2690 static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh) 2687 static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh)
2691 { 2688 {
2692 struct net *net = sock_net(in_skb->sk); 2689 struct net *net = sock_net(in_skb->sk);
2693 struct nlattr *tb[RTA_MAX+1]; 2690 struct nlattr *tb[RTA_MAX+1];
2694 struct rt6_info *rt; 2691 struct rt6_info *rt;
2695 struct sk_buff *skb; 2692 struct sk_buff *skb;
2696 struct rtmsg *rtm; 2693 struct rtmsg *rtm;
2697 struct flowi6 fl6; 2694 struct flowi6 fl6;
2698 int err, iif = 0, oif = 0; 2695 int err, iif = 0, oif = 0;
2699 2696
2700 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy); 2697 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
2701 if (err < 0) 2698 if (err < 0)
2702 goto errout; 2699 goto errout;
2703 2700
2704 err = -EINVAL; 2701 err = -EINVAL;
2705 memset(&fl6, 0, sizeof(fl6)); 2702 memset(&fl6, 0, sizeof(fl6));
2706 2703
2707 if (tb[RTA_SRC]) { 2704 if (tb[RTA_SRC]) {
2708 if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr)) 2705 if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr))
2709 goto errout; 2706 goto errout;
2710 2707
2711 fl6.saddr = *(struct in6_addr *)nla_data(tb[RTA_SRC]); 2708 fl6.saddr = *(struct in6_addr *)nla_data(tb[RTA_SRC]);
2712 } 2709 }
2713 2710
2714 if (tb[RTA_DST]) { 2711 if (tb[RTA_DST]) {
2715 if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr)) 2712 if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr))
2716 goto errout; 2713 goto errout;
2717 2714
2718 fl6.daddr = *(struct in6_addr *)nla_data(tb[RTA_DST]); 2715 fl6.daddr = *(struct in6_addr *)nla_data(tb[RTA_DST]);
2719 } 2716 }
2720 2717
2721 if (tb[RTA_IIF]) 2718 if (tb[RTA_IIF])
2722 iif = nla_get_u32(tb[RTA_IIF]); 2719 iif = nla_get_u32(tb[RTA_IIF]);
2723 2720
2724 if (tb[RTA_OIF]) 2721 if (tb[RTA_OIF])
2725 oif = nla_get_u32(tb[RTA_OIF]); 2722 oif = nla_get_u32(tb[RTA_OIF]);
2726 2723
2727 if (tb[RTA_MARK]) 2724 if (tb[RTA_MARK])
2728 fl6.flowi6_mark = nla_get_u32(tb[RTA_MARK]); 2725 fl6.flowi6_mark = nla_get_u32(tb[RTA_MARK]);
2729 2726
2730 if (iif) { 2727 if (iif) {
2731 struct net_device *dev; 2728 struct net_device *dev;
2732 int flags = 0; 2729 int flags = 0;
2733 2730
2734 dev = __dev_get_by_index(net, iif); 2731 dev = __dev_get_by_index(net, iif);
2735 if (!dev) { 2732 if (!dev) {
2736 err = -ENODEV; 2733 err = -ENODEV;
2737 goto errout; 2734 goto errout;
2738 } 2735 }
2739 2736
2740 fl6.flowi6_iif = iif; 2737 fl6.flowi6_iif = iif;
2741 2738
2742 if (!ipv6_addr_any(&fl6.saddr)) 2739 if (!ipv6_addr_any(&fl6.saddr))
2743 flags |= RT6_LOOKUP_F_HAS_SADDR; 2740 flags |= RT6_LOOKUP_F_HAS_SADDR;
2744 2741
2745 rt = (struct rt6_info *)ip6_route_input_lookup(net, dev, &fl6, 2742 rt = (struct rt6_info *)ip6_route_input_lookup(net, dev, &fl6,
2746 flags); 2743 flags);
2747 } else { 2744 } else {
2748 fl6.flowi6_oif = oif; 2745 fl6.flowi6_oif = oif;
2749 2746
2750 rt = (struct rt6_info *)ip6_route_output(net, NULL, &fl6); 2747 rt = (struct rt6_info *)ip6_route_output(net, NULL, &fl6);
2751 } 2748 }
2752 2749
2753 skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL); 2750 skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
2754 if (!skb) { 2751 if (!skb) {
2755 ip6_rt_put(rt); 2752 ip6_rt_put(rt);
2756 err = -ENOBUFS; 2753 err = -ENOBUFS;
2757 goto errout; 2754 goto errout;
2758 } 2755 }
2759 2756
2760 /* Reserve room for dummy headers, this skb can pass 2757 /* Reserve room for dummy headers, this skb can pass
2761 through good chunk of routing engine. 2758 through good chunk of routing engine.
2762 */ 2759 */
2763 skb_reset_mac_header(skb); 2760 skb_reset_mac_header(skb);
2764 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr)); 2761 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr));
2765 2762
2766 skb_dst_set(skb, &rt->dst); 2763 skb_dst_set(skb, &rt->dst);
2767 2764
2768 err = rt6_fill_node(net, skb, rt, &fl6.daddr, &fl6.saddr, iif, 2765 err = rt6_fill_node(net, skb, rt, &fl6.daddr, &fl6.saddr, iif,
2769 RTM_NEWROUTE, NETLINK_CB(in_skb).portid, 2766 RTM_NEWROUTE, NETLINK_CB(in_skb).portid,
2770 nlh->nlmsg_seq, 0, 0, 0); 2767 nlh->nlmsg_seq, 0, 0, 0);
2771 if (err < 0) { 2768 if (err < 0) {
2772 kfree_skb(skb); 2769 kfree_skb(skb);
2773 goto errout; 2770 goto errout;
2774 } 2771 }
2775 2772
2776 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid); 2773 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
2777 errout: 2774 errout:
2778 return err; 2775 return err;
2779 } 2776 }
2780 2777
2781 void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info) 2778 void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info)
2782 { 2779 {
2783 struct sk_buff *skb; 2780 struct sk_buff *skb;
2784 struct net *net = info->nl_net; 2781 struct net *net = info->nl_net;
2785 u32 seq; 2782 u32 seq;
2786 int err; 2783 int err;
2787 2784
2788 err = -ENOBUFS; 2785 err = -ENOBUFS;
2789 seq = info->nlh ? info->nlh->nlmsg_seq : 0; 2786 seq = info->nlh ? info->nlh->nlmsg_seq : 0;
2790 2787
2791 skb = nlmsg_new(rt6_nlmsg_size(), gfp_any()); 2788 skb = nlmsg_new(rt6_nlmsg_size(), gfp_any());
2792 if (!skb) 2789 if (!skb)
2793 goto errout; 2790 goto errout;
2794 2791
2795 err = rt6_fill_node(net, skb, rt, NULL, NULL, 0, 2792 err = rt6_fill_node(net, skb, rt, NULL, NULL, 0,
2796 event, info->portid, seq, 0, 0, 0); 2793 event, info->portid, seq, 0, 0, 0);
2797 if (err < 0) { 2794 if (err < 0) {
2798 /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */ 2795 /* -EMSGSIZE implies BUG in rt6_nlmsg_size() */
2799 WARN_ON(err == -EMSGSIZE); 2796 WARN_ON(err == -EMSGSIZE);
2800 kfree_skb(skb); 2797 kfree_skb(skb);
2801 goto errout; 2798 goto errout;
2802 } 2799 }
2803 rtnl_notify(skb, net, info->portid, RTNLGRP_IPV6_ROUTE, 2800 rtnl_notify(skb, net, info->portid, RTNLGRP_IPV6_ROUTE,
2804 info->nlh, gfp_any()); 2801 info->nlh, gfp_any());
2805 return; 2802 return;
2806 errout: 2803 errout:
2807 if (err < 0) 2804 if (err < 0)
2808 rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err); 2805 rtnl_set_sk_err(net, RTNLGRP_IPV6_ROUTE, err);
2809 } 2806 }
2810 2807
2811 static int ip6_route_dev_notify(struct notifier_block *this, 2808 static int ip6_route_dev_notify(struct notifier_block *this,
2812 unsigned long event, void *ptr) 2809 unsigned long event, void *ptr)
2813 { 2810 {
2814 struct net_device *dev = netdev_notifier_info_to_dev(ptr); 2811 struct net_device *dev = netdev_notifier_info_to_dev(ptr);
2815 struct net *net = dev_net(dev); 2812 struct net *net = dev_net(dev);
2816 2813
2817 if (event == NETDEV_REGISTER && (dev->flags & IFF_LOOPBACK)) { 2814 if (event == NETDEV_REGISTER && (dev->flags & IFF_LOOPBACK)) {
2818 net->ipv6.ip6_null_entry->dst.dev = dev; 2815 net->ipv6.ip6_null_entry->dst.dev = dev;
2819 net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev); 2816 net->ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(dev);
2820 #ifdef CONFIG_IPV6_MULTIPLE_TABLES 2817 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
2821 net->ipv6.ip6_prohibit_entry->dst.dev = dev; 2818 net->ipv6.ip6_prohibit_entry->dst.dev = dev;
2822 net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev); 2819 net->ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(dev);
2823 net->ipv6.ip6_blk_hole_entry->dst.dev = dev; 2820 net->ipv6.ip6_blk_hole_entry->dst.dev = dev;
2824 net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev); 2821 net->ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(dev);
2825 #endif 2822 #endif
2826 } 2823 }
2827 2824
2828 return NOTIFY_OK; 2825 return NOTIFY_OK;
2829 } 2826 }
2830 2827
2831 /* 2828 /*
2832 * /proc 2829 * /proc
2833 */ 2830 */
2834 2831
2835 #ifdef CONFIG_PROC_FS 2832 #ifdef CONFIG_PROC_FS
2836 2833
2837 static const struct file_operations ipv6_route_proc_fops = { 2834 static const struct file_operations ipv6_route_proc_fops = {
2838 .owner = THIS_MODULE, 2835 .owner = THIS_MODULE,
2839 .open = ipv6_route_open, 2836 .open = ipv6_route_open,
2840 .read = seq_read, 2837 .read = seq_read,
2841 .llseek = seq_lseek, 2838 .llseek = seq_lseek,
2842 .release = seq_release_net, 2839 .release = seq_release_net,
2843 }; 2840 };
2844 2841
2845 static int rt6_stats_seq_show(struct seq_file *seq, void *v) 2842 static int rt6_stats_seq_show(struct seq_file *seq, void *v)
2846 { 2843 {
2847 struct net *net = (struct net *)seq->private; 2844 struct net *net = (struct net *)seq->private;
2848 seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n", 2845 seq_printf(seq, "%04x %04x %04x %04x %04x %04x %04x\n",
2849 net->ipv6.rt6_stats->fib_nodes, 2846 net->ipv6.rt6_stats->fib_nodes,
2850 net->ipv6.rt6_stats->fib_route_nodes, 2847 net->ipv6.rt6_stats->fib_route_nodes,
2851 net->ipv6.rt6_stats->fib_rt_alloc, 2848 net->ipv6.rt6_stats->fib_rt_alloc,
2852 net->ipv6.rt6_stats->fib_rt_entries, 2849 net->ipv6.rt6_stats->fib_rt_entries,
2853 net->ipv6.rt6_stats->fib_rt_cache, 2850 net->ipv6.rt6_stats->fib_rt_cache,
2854 dst_entries_get_slow(&net->ipv6.ip6_dst_ops), 2851 dst_entries_get_slow(&net->ipv6.ip6_dst_ops),
2855 net->ipv6.rt6_stats->fib_discarded_routes); 2852 net->ipv6.rt6_stats->fib_discarded_routes);
2856 2853
2857 return 0; 2854 return 0;
2858 } 2855 }
2859 2856
2860 static int rt6_stats_seq_open(struct inode *inode, struct file *file) 2857 static int rt6_stats_seq_open(struct inode *inode, struct file *file)
2861 { 2858 {
2862 return single_open_net(inode, file, rt6_stats_seq_show); 2859 return single_open_net(inode, file, rt6_stats_seq_show);
2863 } 2860 }
2864 2861
2865 static const struct file_operations rt6_stats_seq_fops = { 2862 static const struct file_operations rt6_stats_seq_fops = {
2866 .owner = THIS_MODULE, 2863 .owner = THIS_MODULE,
2867 .open = rt6_stats_seq_open, 2864 .open = rt6_stats_seq_open,
2868 .read = seq_read, 2865 .read = seq_read,
2869 .llseek = seq_lseek, 2866 .llseek = seq_lseek,
2870 .release = single_release_net, 2867 .release = single_release_net,
2871 }; 2868 };
2872 #endif /* CONFIG_PROC_FS */ 2869 #endif /* CONFIG_PROC_FS */
2873 2870
2874 #ifdef CONFIG_SYSCTL 2871 #ifdef CONFIG_SYSCTL
2875 2872
2876 static 2873 static
2877 int ipv6_sysctl_rtcache_flush(struct ctl_table *ctl, int write, 2874 int ipv6_sysctl_rtcache_flush(struct ctl_table *ctl, int write,
2878 void __user *buffer, size_t *lenp, loff_t *ppos) 2875 void __user *buffer, size_t *lenp, loff_t *ppos)
2879 { 2876 {
2880 struct net *net; 2877 struct net *net;
2881 int delay; 2878 int delay;
2882 if (!write) 2879 if (!write)
2883 return -EINVAL; 2880 return -EINVAL;
2884 2881
2885 net = (struct net *)ctl->extra1; 2882 net = (struct net *)ctl->extra1;
2886 delay = net->ipv6.sysctl.flush_delay; 2883 delay = net->ipv6.sysctl.flush_delay;
2887 proc_dointvec(ctl, write, buffer, lenp, ppos); 2884 proc_dointvec(ctl, write, buffer, lenp, ppos);
2888 fib6_run_gc(delay <= 0 ? 0 : (unsigned long)delay, net, delay > 0); 2885 fib6_run_gc(delay <= 0 ? 0 : (unsigned long)delay, net, delay > 0);
2889 return 0; 2886 return 0;
2890 } 2887 }
2891 2888
2892 struct ctl_table ipv6_route_table_template[] = { 2889 struct ctl_table ipv6_route_table_template[] = {
2893 { 2890 {
2894 .procname = "flush", 2891 .procname = "flush",
2895 .data = &init_net.ipv6.sysctl.flush_delay, 2892 .data = &init_net.ipv6.sysctl.flush_delay,
2896 .maxlen = sizeof(int), 2893 .maxlen = sizeof(int),
2897 .mode = 0200, 2894 .mode = 0200,
2898 .proc_handler = ipv6_sysctl_rtcache_flush 2895 .proc_handler = ipv6_sysctl_rtcache_flush
2899 }, 2896 },
2900 { 2897 {
2901 .procname = "gc_thresh", 2898 .procname = "gc_thresh",
2902 .data = &ip6_dst_ops_template.gc_thresh, 2899 .data = &ip6_dst_ops_template.gc_thresh,
2903 .maxlen = sizeof(int), 2900 .maxlen = sizeof(int),
2904 .mode = 0644, 2901 .mode = 0644,
2905 .proc_handler = proc_dointvec, 2902 .proc_handler = proc_dointvec,
2906 }, 2903 },
2907 { 2904 {
2908 .procname = "max_size", 2905 .procname = "max_size",
2909 .data = &init_net.ipv6.sysctl.ip6_rt_max_size, 2906 .data = &init_net.ipv6.sysctl.ip6_rt_max_size,
2910 .maxlen = sizeof(int), 2907 .maxlen = sizeof(int),
2911 .mode = 0644, 2908 .mode = 0644,
2912 .proc_handler = proc_dointvec, 2909 .proc_handler = proc_dointvec,
2913 }, 2910 },
2914 { 2911 {
2915 .procname = "gc_min_interval", 2912 .procname = "gc_min_interval",
2916 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval, 2913 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
2917 .maxlen = sizeof(int), 2914 .maxlen = sizeof(int),
2918 .mode = 0644, 2915 .mode = 0644,
2919 .proc_handler = proc_dointvec_jiffies, 2916 .proc_handler = proc_dointvec_jiffies,
2920 }, 2917 },
2921 { 2918 {
2922 .procname = "gc_timeout", 2919 .procname = "gc_timeout",
2923 .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout, 2920 .data = &init_net.ipv6.sysctl.ip6_rt_gc_timeout,
2924 .maxlen = sizeof(int), 2921 .maxlen = sizeof(int),
2925 .mode = 0644, 2922 .mode = 0644,
2926 .proc_handler = proc_dointvec_jiffies, 2923 .proc_handler = proc_dointvec_jiffies,
2927 }, 2924 },
2928 { 2925 {
2929 .procname = "gc_interval", 2926 .procname = "gc_interval",
2930 .data = &init_net.ipv6.sysctl.ip6_rt_gc_interval, 2927 .data = &init_net.ipv6.sysctl.ip6_rt_gc_interval,
2931 .maxlen = sizeof(int), 2928 .maxlen = sizeof(int),
2932 .mode = 0644, 2929 .mode = 0644,
2933 .proc_handler = proc_dointvec_jiffies, 2930 .proc_handler = proc_dointvec_jiffies,
2934 }, 2931 },
2935 { 2932 {
2936 .procname = "gc_elasticity", 2933 .procname = "gc_elasticity",
2937 .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity, 2934 .data = &init_net.ipv6.sysctl.ip6_rt_gc_elasticity,
2938 .maxlen = sizeof(int), 2935 .maxlen = sizeof(int),
2939 .mode = 0644, 2936 .mode = 0644,
2940 .proc_handler = proc_dointvec, 2937 .proc_handler = proc_dointvec,
2941 }, 2938 },
2942 { 2939 {
2943 .procname = "mtu_expires", 2940 .procname = "mtu_expires",
2944 .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires, 2941 .data = &init_net.ipv6.sysctl.ip6_rt_mtu_expires,
2945 .maxlen = sizeof(int), 2942 .maxlen = sizeof(int),
2946 .mode = 0644, 2943 .mode = 0644,
2947 .proc_handler = proc_dointvec_jiffies, 2944 .proc_handler = proc_dointvec_jiffies,
2948 }, 2945 },
2949 { 2946 {
2950 .procname = "min_adv_mss", 2947 .procname = "min_adv_mss",
2951 .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss, 2948 .data = &init_net.ipv6.sysctl.ip6_rt_min_advmss,
2952 .maxlen = sizeof(int), 2949 .maxlen = sizeof(int),
2953 .mode = 0644, 2950 .mode = 0644,
2954 .proc_handler = proc_dointvec, 2951 .proc_handler = proc_dointvec,
2955 }, 2952 },
2956 { 2953 {
2957 .procname = "gc_min_interval_ms", 2954 .procname = "gc_min_interval_ms",
2958 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval, 2955 .data = &init_net.ipv6.sysctl.ip6_rt_gc_min_interval,
2959 .maxlen = sizeof(int), 2956 .maxlen = sizeof(int),
2960 .mode = 0644, 2957 .mode = 0644,
2961 .proc_handler = proc_dointvec_ms_jiffies, 2958 .proc_handler = proc_dointvec_ms_jiffies,
2962 }, 2959 },
2963 { } 2960 { }
2964 }; 2961 };
2965 2962
2966 struct ctl_table * __net_init ipv6_route_sysctl_init(struct net *net) 2963 struct ctl_table * __net_init ipv6_route_sysctl_init(struct net *net)
2967 { 2964 {
2968 struct ctl_table *table; 2965 struct ctl_table *table;
2969 2966
2970 table = kmemdup(ipv6_route_table_template, 2967 table = kmemdup(ipv6_route_table_template,
2971 sizeof(ipv6_route_table_template), 2968 sizeof(ipv6_route_table_template),
2972 GFP_KERNEL); 2969 GFP_KERNEL);
2973 2970
2974 if (table) { 2971 if (table) {
2975 table[0].data = &net->ipv6.sysctl.flush_delay; 2972 table[0].data = &net->ipv6.sysctl.flush_delay;
2976 table[0].extra1 = net; 2973 table[0].extra1 = net;
2977 table[1].data = &net->ipv6.ip6_dst_ops.gc_thresh; 2974 table[1].data = &net->ipv6.ip6_dst_ops.gc_thresh;
2978 table[2].data = &net->ipv6.sysctl.ip6_rt_max_size; 2975 table[2].data = &net->ipv6.sysctl.ip6_rt_max_size;
2979 table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval; 2976 table[3].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
2980 table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout; 2977 table[4].data = &net->ipv6.sysctl.ip6_rt_gc_timeout;
2981 table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval; 2978 table[5].data = &net->ipv6.sysctl.ip6_rt_gc_interval;
2982 table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity; 2979 table[6].data = &net->ipv6.sysctl.ip6_rt_gc_elasticity;
2983 table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires; 2980 table[7].data = &net->ipv6.sysctl.ip6_rt_mtu_expires;
2984 table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss; 2981 table[8].data = &net->ipv6.sysctl.ip6_rt_min_advmss;
2985 table[9].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval; 2982 table[9].data = &net->ipv6.sysctl.ip6_rt_gc_min_interval;
2986 2983
2987 /* Don't export sysctls to unprivileged users */ 2984 /* Don't export sysctls to unprivileged users */
2988 if (net->user_ns != &init_user_ns) 2985 if (net->user_ns != &init_user_ns)
2989 table[0].procname = NULL; 2986 table[0].procname = NULL;
2990 } 2987 }
2991 2988
2992 return table; 2989 return table;
2993 } 2990 }
2994 #endif 2991 #endif
2995 2992
2996 static int __net_init ip6_route_net_init(struct net *net) 2993 static int __net_init ip6_route_net_init(struct net *net)
2997 { 2994 {
2998 int ret = -ENOMEM; 2995 int ret = -ENOMEM;
2999 2996
3000 memcpy(&net->ipv6.ip6_dst_ops, &ip6_dst_ops_template, 2997 memcpy(&net->ipv6.ip6_dst_ops, &ip6_dst_ops_template,
3001 sizeof(net->ipv6.ip6_dst_ops)); 2998 sizeof(net->ipv6.ip6_dst_ops));
3002 2999
3003 if (dst_entries_init(&net->ipv6.ip6_dst_ops) < 0) 3000 if (dst_entries_init(&net->ipv6.ip6_dst_ops) < 0)
3004 goto out_ip6_dst_ops; 3001 goto out_ip6_dst_ops;
3005 3002
3006 net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template, 3003 net->ipv6.ip6_null_entry = kmemdup(&ip6_null_entry_template,
3007 sizeof(*net->ipv6.ip6_null_entry), 3004 sizeof(*net->ipv6.ip6_null_entry),
3008 GFP_KERNEL); 3005 GFP_KERNEL);
3009 if (!net->ipv6.ip6_null_entry) 3006 if (!net->ipv6.ip6_null_entry)
3010 goto out_ip6_dst_entries; 3007 goto out_ip6_dst_entries;
3011 net->ipv6.ip6_null_entry->dst.path = 3008 net->ipv6.ip6_null_entry->dst.path =
3012 (struct dst_entry *)net->ipv6.ip6_null_entry; 3009 (struct dst_entry *)net->ipv6.ip6_null_entry;
3013 net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops; 3010 net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops;
3014 dst_init_metrics(&net->ipv6.ip6_null_entry->dst, 3011 dst_init_metrics(&net->ipv6.ip6_null_entry->dst,
3015 ip6_template_metrics, true); 3012 ip6_template_metrics, true);
3016 3013
3017 #ifdef CONFIG_IPV6_MULTIPLE_TABLES 3014 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
3018 net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template, 3015 net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template,
3019 sizeof(*net->ipv6.ip6_prohibit_entry), 3016 sizeof(*net->ipv6.ip6_prohibit_entry),
3020 GFP_KERNEL); 3017 GFP_KERNEL);
3021 if (!net->ipv6.ip6_prohibit_entry) 3018 if (!net->ipv6.ip6_prohibit_entry)
3022 goto out_ip6_null_entry; 3019 goto out_ip6_null_entry;
3023 net->ipv6.ip6_prohibit_entry->dst.path = 3020 net->ipv6.ip6_prohibit_entry->dst.path =
3024 (struct dst_entry *)net->ipv6.ip6_prohibit_entry; 3021 (struct dst_entry *)net->ipv6.ip6_prohibit_entry;
3025 net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops; 3022 net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops;
3026 dst_init_metrics(&net->ipv6.ip6_prohibit_entry->dst, 3023 dst_init_metrics(&net->ipv6.ip6_prohibit_entry->dst,
3027 ip6_template_metrics, true); 3024 ip6_template_metrics, true);
3028 3025
3029 net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template, 3026 net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template,
3030 sizeof(*net->ipv6.ip6_blk_hole_entry), 3027 sizeof(*net->ipv6.ip6_blk_hole_entry),
3031 GFP_KERNEL); 3028 GFP_KERNEL);
3032 if (!net->ipv6.ip6_blk_hole_entry) 3029 if (!net->ipv6.ip6_blk_hole_entry)
3033 goto out_ip6_prohibit_entry; 3030 goto out_ip6_prohibit_entry;
3034 net->ipv6.ip6_blk_hole_entry->dst.path = 3031 net->ipv6.ip6_blk_hole_entry->dst.path =
3035 (struct dst_entry *)net->ipv6.ip6_blk_hole_entry; 3032 (struct dst_entry *)net->ipv6.ip6_blk_hole_entry;
3036 net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops; 3033 net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops;
3037 dst_init_metrics(&net->ipv6.ip6_blk_hole_entry->dst, 3034 dst_init_metrics(&net->ipv6.ip6_blk_hole_entry->dst,
3038 ip6_template_metrics, true); 3035 ip6_template_metrics, true);
3039 #endif 3036 #endif
3040 3037
3041 net->ipv6.sysctl.flush_delay = 0; 3038 net->ipv6.sysctl.flush_delay = 0;
3042 net->ipv6.sysctl.ip6_rt_max_size = 4096; 3039 net->ipv6.sysctl.ip6_rt_max_size = 4096;
3043 net->ipv6.sysctl.ip6_rt_gc_min_interval = HZ / 2; 3040 net->ipv6.sysctl.ip6_rt_gc_min_interval = HZ / 2;
3044 net->ipv6.sysctl.ip6_rt_gc_timeout = 60*HZ; 3041 net->ipv6.sysctl.ip6_rt_gc_timeout = 60*HZ;
3045 net->ipv6.sysctl.ip6_rt_gc_interval = 30*HZ; 3042 net->ipv6.sysctl.ip6_rt_gc_interval = 30*HZ;
3046 net->ipv6.sysctl.ip6_rt_gc_elasticity = 9; 3043 net->ipv6.sysctl.ip6_rt_gc_elasticity = 9;
3047 net->ipv6.sysctl.ip6_rt_mtu_expires = 10*60*HZ; 3044 net->ipv6.sysctl.ip6_rt_mtu_expires = 10*60*HZ;
3048 net->ipv6.sysctl.ip6_rt_min_advmss = IPV6_MIN_MTU - 20 - 40; 3045 net->ipv6.sysctl.ip6_rt_min_advmss = IPV6_MIN_MTU - 20 - 40;
3049 3046
3050 net->ipv6.ip6_rt_gc_expire = 30*HZ; 3047 net->ipv6.ip6_rt_gc_expire = 30*HZ;
3051 3048
3052 ret = 0; 3049 ret = 0;
3053 out: 3050 out:
3054 return ret; 3051 return ret;
3055 3052
3056 #ifdef CONFIG_IPV6_MULTIPLE_TABLES 3053 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
3057 out_ip6_prohibit_entry: 3054 out_ip6_prohibit_entry:
3058 kfree(net->ipv6.ip6_prohibit_entry); 3055 kfree(net->ipv6.ip6_prohibit_entry);
3059 out_ip6_null_entry: 3056 out_ip6_null_entry:
3060 kfree(net->ipv6.ip6_null_entry); 3057 kfree(net->ipv6.ip6_null_entry);
3061 #endif 3058 #endif
3062 out_ip6_dst_entries: 3059 out_ip6_dst_entries:
3063 dst_entries_destroy(&net->ipv6.ip6_dst_ops); 3060 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
3064 out_ip6_dst_ops: 3061 out_ip6_dst_ops:
3065 goto out; 3062 goto out;
3066 } 3063 }
3067 3064
3068 static void __net_exit ip6_route_net_exit(struct net *net) 3065 static void __net_exit ip6_route_net_exit(struct net *net)
3069 { 3066 {
3070 kfree(net->ipv6.ip6_null_entry); 3067 kfree(net->ipv6.ip6_null_entry);
3071 #ifdef CONFIG_IPV6_MULTIPLE_TABLES 3068 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
3072 kfree(net->ipv6.ip6_prohibit_entry); 3069 kfree(net->ipv6.ip6_prohibit_entry);
3073 kfree(net->ipv6.ip6_blk_hole_entry); 3070 kfree(net->ipv6.ip6_blk_hole_entry);
3074 #endif 3071 #endif
3075 dst_entries_destroy(&net->ipv6.ip6_dst_ops); 3072 dst_entries_destroy(&net->ipv6.ip6_dst_ops);
3076 } 3073 }
3077 3074
3078 static int __net_init ip6_route_net_init_late(struct net *net) 3075 static int __net_init ip6_route_net_init_late(struct net *net)
3079 { 3076 {
3080 #ifdef CONFIG_PROC_FS 3077 #ifdef CONFIG_PROC_FS
3081 proc_create("ipv6_route", 0, net->proc_net, &ipv6_route_proc_fops); 3078 proc_create("ipv6_route", 0, net->proc_net, &ipv6_route_proc_fops);
3082 proc_create("rt6_stats", S_IRUGO, net->proc_net, &rt6_stats_seq_fops); 3079 proc_create("rt6_stats", S_IRUGO, net->proc_net, &rt6_stats_seq_fops);
3083 #endif 3080 #endif
3084 return 0; 3081 return 0;
3085 } 3082 }
3086 3083
3087 static void __net_exit ip6_route_net_exit_late(struct net *net) 3084 static void __net_exit ip6_route_net_exit_late(struct net *net)
3088 { 3085 {
3089 #ifdef CONFIG_PROC_FS 3086 #ifdef CONFIG_PROC_FS
3090 remove_proc_entry("ipv6_route", net->proc_net); 3087 remove_proc_entry("ipv6_route", net->proc_net);
3091 remove_proc_entry("rt6_stats", net->proc_net); 3088 remove_proc_entry("rt6_stats", net->proc_net);
3092 #endif 3089 #endif
3093 } 3090 }
3094 3091
3095 static struct pernet_operations ip6_route_net_ops = { 3092 static struct pernet_operations ip6_route_net_ops = {
3096 .init = ip6_route_net_init, 3093 .init = ip6_route_net_init,
3097 .exit = ip6_route_net_exit, 3094 .exit = ip6_route_net_exit,
3098 }; 3095 };
3099 3096
3100 static int __net_init ipv6_inetpeer_init(struct net *net) 3097 static int __net_init ipv6_inetpeer_init(struct net *net)
3101 { 3098 {
3102 struct inet_peer_base *bp = kmalloc(sizeof(*bp), GFP_KERNEL); 3099 struct inet_peer_base *bp = kmalloc(sizeof(*bp), GFP_KERNEL);
3103 3100
3104 if (!bp) 3101 if (!bp)
3105 return -ENOMEM; 3102 return -ENOMEM;
3106 inet_peer_base_init(bp); 3103 inet_peer_base_init(bp);
3107 net->ipv6.peers = bp; 3104 net->ipv6.peers = bp;
3108 return 0; 3105 return 0;
3109 } 3106 }
3110 3107
3111 static void __net_exit ipv6_inetpeer_exit(struct net *net) 3108 static void __net_exit ipv6_inetpeer_exit(struct net *net)
3112 { 3109 {
3113 struct inet_peer_base *bp = net->ipv6.peers; 3110 struct inet_peer_base *bp = net->ipv6.peers;
3114 3111
3115 net->ipv6.peers = NULL; 3112 net->ipv6.peers = NULL;
3116 inetpeer_invalidate_tree(bp); 3113 inetpeer_invalidate_tree(bp);
3117 kfree(bp); 3114 kfree(bp);
3118 } 3115 }
3119 3116
3120 static struct pernet_operations ipv6_inetpeer_ops = { 3117 static struct pernet_operations ipv6_inetpeer_ops = {
3121 .init = ipv6_inetpeer_init, 3118 .init = ipv6_inetpeer_init,
3122 .exit = ipv6_inetpeer_exit, 3119 .exit = ipv6_inetpeer_exit,
3123 }; 3120 };
3124 3121
3125 static struct pernet_operations ip6_route_net_late_ops = { 3122 static struct pernet_operations ip6_route_net_late_ops = {
3126 .init = ip6_route_net_init_late, 3123 .init = ip6_route_net_init_late,
3127 .exit = ip6_route_net_exit_late, 3124 .exit = ip6_route_net_exit_late,
3128 }; 3125 };
3129 3126
3130 static struct notifier_block ip6_route_dev_notifier = { 3127 static struct notifier_block ip6_route_dev_notifier = {
3131 .notifier_call = ip6_route_dev_notify, 3128 .notifier_call = ip6_route_dev_notify,
3132 .priority = 0, 3129 .priority = 0,
3133 }; 3130 };
3134 3131
3135 int __init ip6_route_init(void) 3132 int __init ip6_route_init(void)
3136 { 3133 {
3137 int ret; 3134 int ret;
3138 3135
3139 ret = -ENOMEM; 3136 ret = -ENOMEM;
3140 ip6_dst_ops_template.kmem_cachep = 3137 ip6_dst_ops_template.kmem_cachep =
3141 kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0, 3138 kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0,
3142 SLAB_HWCACHE_ALIGN, NULL); 3139 SLAB_HWCACHE_ALIGN, NULL);
3143 if (!ip6_dst_ops_template.kmem_cachep) 3140 if (!ip6_dst_ops_template.kmem_cachep)
3144 goto out; 3141 goto out;
3145 3142
3146 ret = dst_entries_init(&ip6_dst_blackhole_ops); 3143 ret = dst_entries_init(&ip6_dst_blackhole_ops);
3147 if (ret) 3144 if (ret)
3148 goto out_kmem_cache; 3145 goto out_kmem_cache;
3149 3146
3150 ret = register_pernet_subsys(&ipv6_inetpeer_ops); 3147 ret = register_pernet_subsys(&ipv6_inetpeer_ops);
3151 if (ret) 3148 if (ret)
3152 goto out_dst_entries; 3149 goto out_dst_entries;
3153 3150
3154 ret = register_pernet_subsys(&ip6_route_net_ops); 3151 ret = register_pernet_subsys(&ip6_route_net_ops);
3155 if (ret) 3152 if (ret)
3156 goto out_register_inetpeer; 3153 goto out_register_inetpeer;
3157 3154
3158 ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep; 3155 ip6_dst_blackhole_ops.kmem_cachep = ip6_dst_ops_template.kmem_cachep;
3159 3156
3160 /* Registering of the loopback is done before this portion of code, 3157 /* Registering of the loopback is done before this portion of code,
3161 * the loopback reference in rt6_info will not be taken, do it 3158 * the loopback reference in rt6_info will not be taken, do it
3162 * manually for init_net */ 3159 * manually for init_net */
3163 init_net.ipv6.ip6_null_entry->dst.dev = init_net.loopback_dev; 3160 init_net.ipv6.ip6_null_entry->dst.dev = init_net.loopback_dev;
3164 init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev); 3161 init_net.ipv6.ip6_null_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3165 #ifdef CONFIG_IPV6_MULTIPLE_TABLES 3162 #ifdef CONFIG_IPV6_MULTIPLE_TABLES
3166 init_net.ipv6.ip6_prohibit_entry->dst.dev = init_net.loopback_dev; 3163 init_net.ipv6.ip6_prohibit_entry->dst.dev = init_net.loopback_dev;
3167 init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev); 3164 init_net.ipv6.ip6_prohibit_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3168 init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev; 3165 init_net.ipv6.ip6_blk_hole_entry->dst.dev = init_net.loopback_dev;
3169 init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev); 3166 init_net.ipv6.ip6_blk_hole_entry->rt6i_idev = in6_dev_get(init_net.loopback_dev);
3170 #endif 3167 #endif
3171 ret = fib6_init(); 3168 ret = fib6_init();
3172 if (ret) 3169 if (ret)
3173 goto out_register_subsys; 3170 goto out_register_subsys;
3174 3171
3175 ret = xfrm6_init(); 3172 ret = xfrm6_init();
3176 if (ret) 3173 if (ret)
3177 goto out_fib6_init; 3174 goto out_fib6_init;
3178 3175
3179 ret = fib6_rules_init(); 3176 ret = fib6_rules_init();
3180 if (ret) 3177 if (ret)
3181 goto xfrm6_init; 3178 goto xfrm6_init;
3182 3179
3183 ret = register_pernet_subsys(&ip6_route_net_late_ops); 3180 ret = register_pernet_subsys(&ip6_route_net_late_ops);
3184 if (ret) 3181 if (ret)
3185 goto fib6_rules_init; 3182 goto fib6_rules_init;
3186 3183
3187 ret = -ENOBUFS; 3184 ret = -ENOBUFS;
3188 if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL, NULL) || 3185 if (__rtnl_register(PF_INET6, RTM_NEWROUTE, inet6_rtm_newroute, NULL, NULL) ||
3189 __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL, NULL) || 3186 __rtnl_register(PF_INET6, RTM_DELROUTE, inet6_rtm_delroute, NULL, NULL) ||
3190 __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL, NULL)) 3187 __rtnl_register(PF_INET6, RTM_GETROUTE, inet6_rtm_getroute, NULL, NULL))
3191 goto out_register_late_subsys; 3188 goto out_register_late_subsys;
3192 3189
3193 ret = register_netdevice_notifier(&ip6_route_dev_notifier); 3190 ret = register_netdevice_notifier(&ip6_route_dev_notifier);
3194 if (ret) 3191 if (ret)
3195 goto out_register_late_subsys; 3192 goto out_register_late_subsys;
3196 3193
3197 out: 3194 out:
3198 return ret; 3195 return ret;
3199 3196
3200 out_register_late_subsys: 3197 out_register_late_subsys:
3201 unregister_pernet_subsys(&ip6_route_net_late_ops); 3198 unregister_pernet_subsys(&ip6_route_net_late_ops);
3202 fib6_rules_init: 3199 fib6_rules_init:
3203 fib6_rules_cleanup(); 3200 fib6_rules_cleanup();
3204 xfrm6_init: 3201 xfrm6_init:
3205 xfrm6_fini(); 3202 xfrm6_fini();
3206 out_fib6_init: 3203 out_fib6_init:
3207 fib6_gc_cleanup(); 3204 fib6_gc_cleanup();
3208 out_register_subsys: 3205 out_register_subsys:
3209 unregister_pernet_subsys(&ip6_route_net_ops); 3206 unregister_pernet_subsys(&ip6_route_net_ops);
3210 out_register_inetpeer: 3207 out_register_inetpeer:
3211 unregister_pernet_subsys(&ipv6_inetpeer_ops); 3208 unregister_pernet_subsys(&ipv6_inetpeer_ops);
3212 out_dst_entries: 3209 out_dst_entries:
3213 dst_entries_destroy(&ip6_dst_blackhole_ops); 3210 dst_entries_destroy(&ip6_dst_blackhole_ops);
3214 out_kmem_cache: 3211 out_kmem_cache:
3215 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep); 3212 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
3216 goto out; 3213 goto out;
3217 } 3214 }
3218 3215
3219 void ip6_route_cleanup(void) 3216 void ip6_route_cleanup(void)
3220 { 3217 {
3221 unregister_netdevice_notifier(&ip6_route_dev_notifier); 3218 unregister_netdevice_notifier(&ip6_route_dev_notifier);
3222 unregister_pernet_subsys(&ip6_route_net_late_ops); 3219 unregister_pernet_subsys(&ip6_route_net_late_ops);
3223 fib6_rules_cleanup(); 3220 fib6_rules_cleanup();
3224 xfrm6_fini(); 3221 xfrm6_fini();
3225 fib6_gc_cleanup(); 3222 fib6_gc_cleanup();
3226 unregister_pernet_subsys(&ipv6_inetpeer_ops); 3223 unregister_pernet_subsys(&ipv6_inetpeer_ops);
3227 unregister_pernet_subsys(&ip6_route_net_ops); 3224 unregister_pernet_subsys(&ip6_route_net_ops);
3228 dst_entries_destroy(&ip6_dst_blackhole_ops); 3225 dst_entries_destroy(&ip6_dst_blackhole_ops);
3229 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep); 3226 kmem_cache_destroy(ip6_dst_ops_template.kmem_cachep);
3230 } 3227 }
3231 3228