Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
2 | * xfrm6_policy.c: based on xfrm4_policy.c | |
3 | * | |
4 | * Authors: | |
5 | * Mitsuru KANDA @USAGI | |
6 | * Kazunori MIYAZAWA @USAGI | |
7 | * Kunihiro Ishiguro <kunihiro@ipinfusion.com> | |
8 | * IPv6 support | |
9 | * YOSHIFUJI Hideaki | |
10 | * Split up af-specific portion | |
1ab1457c | 11 | * |
1da177e4 LT |
12 | */ |
13 | ||
66cdb3ca HX |
14 | #include <linux/err.h> |
15 | #include <linux/kernel.h> | |
aabc9761 HX |
16 | #include <linux/netdevice.h> |
17 | #include <net/addrconf.h> | |
45ff5a3f | 18 | #include <net/dst.h> |
1da177e4 LT |
19 | #include <net/xfrm.h> |
20 | #include <net/ip.h> | |
21 | #include <net/ipv6.h> | |
22 | #include <net/ip6_route.h> | |
59fbb3a6 | 23 | #if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE) |
2ce4272a MN |
24 | #include <net/mip6.h> |
25 | #endif | |
1da177e4 | 26 | |
1da177e4 LT |
27 | static struct xfrm_policy_afinfo xfrm6_policy_afinfo; |
28 | ||
c5b3cf46 | 29 | static struct dst_entry *xfrm6_dst_lookup(struct net *net, int tos, |
5e6b930f DM |
30 | const xfrm_address_t *saddr, |
31 | const xfrm_address_t *daddr) | |
1da177e4 | 32 | { |
7e1dc7b6 | 33 | struct flowi6 fl6; |
66cdb3ca HX |
34 | struct dst_entry *dst; |
35 | int err; | |
36 | ||
7e1dc7b6 DM |
37 | memset(&fl6, 0, sizeof(fl6)); |
38 | memcpy(&fl6.daddr, daddr, sizeof(fl6.daddr)); | |
66cdb3ca | 39 | if (saddr) |
7e1dc7b6 | 40 | memcpy(&fl6.saddr, saddr, sizeof(fl6.saddr)); |
66cdb3ca | 41 | |
4c9483b2 | 42 | dst = ip6_route_output(net, NULL, &fl6); |
66cdb3ca HX |
43 | |
44 | err = dst->error; | |
45 | if (dst->error) { | |
4251320f | 46 | dst_release(dst); |
66cdb3ca HX |
47 | dst = ERR_PTR(err); |
48 | } | |
49 | ||
50 | return dst; | |
1da177e4 LT |
51 | } |
52 | ||
fbda33b2 AD |
53 | static int xfrm6_get_saddr(struct net *net, |
54 | xfrm_address_t *saddr, xfrm_address_t *daddr) | |
a1e59abf | 55 | { |
66cdb3ca | 56 | struct dst_entry *dst; |
191cd582 | 57 | struct net_device *dev; |
66cdb3ca | 58 | |
fbda33b2 | 59 | dst = xfrm6_dst_lookup(net, 0, NULL, daddr); |
66cdb3ca HX |
60 | if (IS_ERR(dst)) |
61 | return -EHOSTUNREACH; | |
62 | ||
191cd582 BH |
63 | dev = ip6_dst_idev(dst)->dev; |
64 | ipv6_dev_get_saddr(dev_net(dev), dev, | |
7cbca67c | 65 | (struct in6_addr *)&daddr->a6, 0, |
5e5f3f0f | 66 | (struct in6_addr *)&saddr->a6); |
66cdb3ca HX |
67 | dst_release(dst); |
68 | return 0; | |
a1e59abf PM |
69 | } |
70 | ||
05d84025 | 71 | static int xfrm6_get_tos(const struct flowi *fl) |
1da177e4 | 72 | { |
25ee3286 HX |
73 | return 0; |
74 | } | |
1da177e4 | 75 | |
a1b05140 MN |
76 | static int xfrm6_init_path(struct xfrm_dst *path, struct dst_entry *dst, |
77 | int nfheader_len) | |
78 | { | |
79 | if (dst->ops->family == AF_INET6) { | |
80 | struct rt6_info *rt = (struct rt6_info*)dst; | |
81 | if (rt->rt6i_node) | |
82 | path->path_cookie = rt->rt6i_node->fn_sernum; | |
83 | } | |
84 | ||
85 | path->u.rt6.rt6i_nfheader_len = nfheader_len; | |
86 | ||
87 | return 0; | |
88 | } | |
89 | ||
87c1e12b | 90 | static int xfrm6_fill_dst(struct xfrm_dst *xdst, struct net_device *dev, |
0c7b3eef | 91 | const struct flowi *fl) |
25ee3286 HX |
92 | { |
93 | struct rt6_info *rt = (struct rt6_info*)xdst->route; | |
1da177e4 | 94 | |
25ee3286 HX |
95 | xdst->u.dst.dev = dev; |
96 | dev_hold(dev); | |
1da177e4 | 97 | |
bc8e4b95 | 98 | xdst->u.rt6.rt6i_idev = in6_dev_get(dev); |
25ee3286 HX |
99 | if (!xdst->u.rt6.rt6i_idev) |
100 | return -ENODEV; | |
1da177e4 | 101 | |
97bab73f | 102 | rt6_transfer_peer(&xdst->u.rt6, rt); |
7cc2edb8 | 103 | |
25ee3286 HX |
104 | /* Sheit... I remember I did this right. Apparently, |
105 | * it was magically lost, so this code needs audit */ | |
97cac082 | 106 | xdst->u.rt6.n = neigh_clone(rt->n); |
25ee3286 HX |
107 | xdst->u.rt6.rt6i_flags = rt->rt6i_flags & (RTF_ANYCAST | |
108 | RTF_LOCAL); | |
109 | xdst->u.rt6.rt6i_metric = rt->rt6i_metric; | |
110 | xdst->u.rt6.rt6i_node = rt->rt6i_node; | |
a1b05140 MN |
111 | if (rt->rt6i_node) |
112 | xdst->route_cookie = rt->rt6i_node->fn_sernum; | |
25ee3286 HX |
113 | xdst->u.rt6.rt6i_gateway = rt->rt6i_gateway; |
114 | xdst->u.rt6.rt6i_dst = rt->rt6i_dst; | |
115 | xdst->u.rt6.rt6i_src = rt->rt6i_src; | |
1da177e4 | 116 | |
1da177e4 | 117 | return 0; |
1da177e4 LT |
118 | } |
119 | ||
120 | static inline void | |
d5422efe | 121 | _decode_session6(struct sk_buff *skb, struct flowi *fl, int reverse) |
1da177e4 | 122 | { |
7e1dc7b6 | 123 | struct flowi6 *fl6 = &fl->u.ip6; |
7e3a42a1 | 124 | int onlyproto = 0; |
cfe1fc77 | 125 | u16 offset = skb_network_header_len(skb); |
b71d1d42 | 126 | const struct ipv6hdr *hdr = ipv6_hdr(skb); |
e3cae904 | 127 | struct ipv6_opt_hdr *exthdr; |
d56f90a7 ACM |
128 | const unsigned char *nh = skb_network_header(skb); |
129 | u8 nexthdr = nh[IP6CB(skb)->nhoff]; | |
1da177e4 | 130 | |
7e1dc7b6 DM |
131 | memset(fl6, 0, sizeof(struct flowi6)); |
132 | fl6->flowi6_mark = skb->mark; | |
44b451f1 | 133 | |
4e3fd7a0 AD |
134 | fl6->daddr = reverse ? hdr->saddr : hdr->daddr; |
135 | fl6->saddr = reverse ? hdr->daddr : hdr->saddr; | |
1da177e4 | 136 | |
59cae009 WY |
137 | while (nh + offset + 1 < skb->data || |
138 | pskb_may_pull(skb, nh + offset + 1 - skb->data)) { | |
d56f90a7 ACM |
139 | nh = skb_network_header(skb); |
140 | exthdr = (struct ipv6_opt_hdr *)(nh + offset); | |
e3cae904 | 141 | |
1da177e4 | 142 | switch (nexthdr) { |
7e3a42a1 ND |
143 | case NEXTHDR_FRAGMENT: |
144 | onlyproto = 1; | |
1da177e4 LT |
145 | case NEXTHDR_ROUTING: |
146 | case NEXTHDR_HOP: | |
147 | case NEXTHDR_DEST: | |
148 | offset += ipv6_optlen(exthdr); | |
149 | nexthdr = exthdr->nexthdr; | |
d56f90a7 | 150 | exthdr = (struct ipv6_opt_hdr *)(nh + offset); |
1da177e4 LT |
151 | break; |
152 | ||
153 | case IPPROTO_UDP: | |
ba4e58ec | 154 | case IPPROTO_UDPLITE: |
1da177e4 LT |
155 | case IPPROTO_TCP: |
156 | case IPPROTO_SCTP: | |
9e999993 | 157 | case IPPROTO_DCCP: |
59cae009 WY |
158 | if (!onlyproto && (nh + offset + 4 < skb->data || |
159 | pskb_may_pull(skb, nh + offset + 4 - skb->data))) { | |
8c689a6e | 160 | __be16 *ports = (__be16 *)exthdr; |
1da177e4 | 161 | |
1958b856 DM |
162 | fl6->fl6_sport = ports[!!reverse]; |
163 | fl6->fl6_dport = ports[!reverse]; | |
1da177e4 | 164 | } |
7e1dc7b6 | 165 | fl6->flowi6_proto = nexthdr; |
1da177e4 LT |
166 | return; |
167 | ||
168 | case IPPROTO_ICMPV6: | |
7e3a42a1 | 169 | if (!onlyproto && pskb_may_pull(skb, nh + offset + 2 - skb->data)) { |
1da177e4 LT |
170 | u8 *icmp = (u8 *)exthdr; |
171 | ||
1958b856 DM |
172 | fl6->fl6_icmp_type = icmp[0]; |
173 | fl6->fl6_icmp_code = icmp[1]; | |
1da177e4 | 174 | } |
7e1dc7b6 | 175 | fl6->flowi6_proto = nexthdr; |
1da177e4 LT |
176 | return; |
177 | ||
59fbb3a6 | 178 | #if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE) |
2ce4272a | 179 | case IPPROTO_MH: |
7e3a42a1 | 180 | if (!onlyproto && pskb_may_pull(skb, nh + offset + 3 - skb->data)) { |
2ce4272a MN |
181 | struct ip6_mh *mh; |
182 | mh = (struct ip6_mh *)exthdr; | |
183 | ||
1958b856 | 184 | fl6->fl6_mh_type = mh->ip6mh_type; |
2ce4272a | 185 | } |
7e1dc7b6 | 186 | fl6->flowi6_proto = nexthdr; |
2ce4272a MN |
187 | return; |
188 | #endif | |
189 | ||
1da177e4 LT |
190 | /* XXX Why are there these headers? */ |
191 | case IPPROTO_AH: | |
192 | case IPPROTO_ESP: | |
193 | case IPPROTO_COMP: | |
194 | default: | |
1958b856 | 195 | fl6->fl6_ipsec_spi = 0; |
7e1dc7b6 | 196 | fl6->flowi6_proto = nexthdr; |
1da177e4 | 197 | return; |
3ff50b79 | 198 | } |
1da177e4 LT |
199 | } |
200 | } | |
201 | ||
569d3645 | 202 | static inline int xfrm6_garbage_collect(struct dst_ops *ops) |
1da177e4 | 203 | { |
d7c7544c AD |
204 | struct net *net = container_of(ops, struct net, xfrm.xfrm6_dst_ops); |
205 | ||
206 | xfrm6_policy_afinfo.garbage_collect(net); | |
fc66f95c | 207 | return dst_entries_get_fast(ops) > ops->gc_thresh * 2; |
1da177e4 LT |
208 | } |
209 | ||
210 | static void xfrm6_update_pmtu(struct dst_entry *dst, u32 mtu) | |
211 | { | |
212 | struct xfrm_dst *xdst = (struct xfrm_dst *)dst; | |
213 | struct dst_entry *path = xdst->route; | |
214 | ||
215 | path->ops->update_pmtu(path, mtu); | |
216 | } | |
217 | ||
aabc9761 HX |
218 | static void xfrm6_dst_destroy(struct dst_entry *dst) |
219 | { | |
220 | struct xfrm_dst *xdst = (struct xfrm_dst *)dst; | |
221 | ||
222 | if (likely(xdst->u.rt6.rt6i_idev)) | |
223 | in6_dev_put(xdst->u.rt6.rt6i_idev); | |
62fa8a84 | 224 | dst_destroy_metrics_generic(dst); |
97bab73f DM |
225 | if (rt6_has_peer(&xdst->u.rt6)) { |
226 | struct inet_peer *peer = rt6_peer_ptr(&xdst->u.rt6); | |
227 | inet_putpeer(peer); | |
228 | } | |
aabc9761 HX |
229 | xfrm_dst_destroy(xdst); |
230 | } | |
231 | ||
232 | static void xfrm6_dst_ifdown(struct dst_entry *dst, struct net_device *dev, | |
233 | int unregister) | |
234 | { | |
235 | struct xfrm_dst *xdst; | |
236 | ||
237 | if (!unregister) | |
238 | return; | |
239 | ||
240 | xdst = (struct xfrm_dst *)dst; | |
241 | if (xdst->u.rt6.rt6i_idev->dev == dev) { | |
5a3e55d6 | 242 | struct inet6_dev *loopback_idev = |
c346dca1 | 243 | in6_dev_get(dev_net(dev)->loopback_dev); |
aabc9761 HX |
244 | BUG_ON(!loopback_idev); |
245 | ||
246 | do { | |
247 | in6_dev_put(xdst->u.rt6.rt6i_idev); | |
248 | xdst->u.rt6.rt6i_idev = loopback_idev; | |
249 | in6_dev_hold(loopback_idev); | |
250 | xdst = (struct xfrm_dst *)xdst->u.dst.child; | |
251 | } while (xdst->u.dst.xfrm); | |
252 | ||
253 | __in6_dev_put(loopback_idev); | |
254 | } | |
255 | ||
256 | xfrm_dst_ifdown(dst, dev); | |
257 | } | |
258 | ||
1da177e4 LT |
259 | static struct dst_ops xfrm6_dst_ops = { |
260 | .family = AF_INET6, | |
09640e63 | 261 | .protocol = cpu_to_be16(ETH_P_IPV6), |
1da177e4 LT |
262 | .gc = xfrm6_garbage_collect, |
263 | .update_pmtu = xfrm6_update_pmtu, | |
62fa8a84 | 264 | .cow_metrics = dst_cow_metrics_generic, |
aabc9761 HX |
265 | .destroy = xfrm6_dst_destroy, |
266 | .ifdown = xfrm6_dst_ifdown, | |
862b82c6 | 267 | .local_out = __ip6_local_out, |
1da177e4 | 268 | .gc_thresh = 1024, |
1da177e4 LT |
269 | }; |
270 | ||
271 | static struct xfrm_policy_afinfo xfrm6_policy_afinfo = { | |
272 | .family = AF_INET6, | |
1da177e4 LT |
273 | .dst_ops = &xfrm6_dst_ops, |
274 | .dst_lookup = xfrm6_dst_lookup, | |
a1e59abf | 275 | .get_saddr = xfrm6_get_saddr, |
1da177e4 | 276 | .decode_session = _decode_session6, |
25ee3286 | 277 | .get_tos = xfrm6_get_tos, |
a1b05140 | 278 | .init_path = xfrm6_init_path, |
25ee3286 | 279 | .fill_dst = xfrm6_fill_dst, |
2774c131 | 280 | .blackhole_route = ip6_blackhole_route, |
1da177e4 LT |
281 | }; |
282 | ||
0013caba | 283 | static int __init xfrm6_policy_init(void) |
1da177e4 | 284 | { |
0013caba | 285 | return xfrm_policy_register_afinfo(&xfrm6_policy_afinfo); |
1da177e4 LT |
286 | } |
287 | ||
288 | static void xfrm6_policy_fini(void) | |
289 | { | |
290 | xfrm_policy_unregister_afinfo(&xfrm6_policy_afinfo); | |
291 | } | |
292 | ||
db71789c | 293 | #ifdef CONFIG_SYSCTL |
a44a4a00 NH |
294 | static struct ctl_table xfrm6_policy_table[] = { |
295 | { | |
a44a4a00 | 296 | .procname = "xfrm6_gc_thresh", |
d7c7544c | 297 | .data = &init_net.xfrm.xfrm6_dst_ops.gc_thresh, |
a44a4a00 NH |
298 | .maxlen = sizeof(int), |
299 | .mode = 0644, | |
300 | .proc_handler = proc_dointvec, | |
301 | }, | |
302 | { } | |
303 | }; | |
304 | ||
305 | static struct ctl_table_header *sysctl_hdr; | |
db71789c | 306 | #endif |
a44a4a00 | 307 | |
0013caba | 308 | int __init xfrm6_init(void) |
1da177e4 | 309 | { |
0013caba | 310 | int ret; |
a33bc5c1 | 311 | unsigned int gc_thresh; |
0013caba | 312 | |
a33bc5c1 NH |
313 | /* |
314 | * We need a good default value for the xfrm6 gc threshold. | |
315 | * In ipv4 we set it to the route hash table size * 8, which | |
316 | * is half the size of the maximaum route cache for ipv4. It | |
317 | * would be good to do the same thing for v6, except the table is | |
318 | * constructed differently here. Here each table for a net namespace | |
319 | * can have FIB_TABLE_HASHSZ entries, so lets go with the same | |
320 | * computation that we used for ipv4 here. Also, lets keep the initial | |
321 | * gc_thresh to a minimum of 1024, since, the ipv6 route cache defaults | |
322 | * to that as a minimum as well | |
323 | */ | |
324 | gc_thresh = FIB6_TABLE_HASHSZ * 8; | |
325 | xfrm6_dst_ops.gc_thresh = (gc_thresh < 1024) ? 1024 : gc_thresh; | |
fc66f95c | 326 | dst_entries_init(&xfrm6_dst_ops); |
d7c7544c AD |
327 | |
328 | ret = xfrm6_policy_init(); | |
fc66f95c ED |
329 | if (ret) { |
330 | dst_entries_destroy(&xfrm6_dst_ops); | |
d7c7544c | 331 | goto out; |
fc66f95c | 332 | } |
d7c7544c AD |
333 | ret = xfrm6_state_init(); |
334 | if (ret) | |
335 | goto out_policy; | |
336 | ||
db71789c | 337 | #ifdef CONFIG_SYSCTL |
ec8f23ce EB |
338 | sysctl_hdr = register_net_sysctl(&init_net, "net/ipv6", |
339 | xfrm6_policy_table); | |
db71789c | 340 | #endif |
0013caba DL |
341 | out: |
342 | return ret; | |
343 | out_policy: | |
344 | xfrm6_policy_fini(); | |
345 | goto out; | |
1da177e4 LT |
346 | } |
347 | ||
348 | void xfrm6_fini(void) | |
349 | { | |
db71789c | 350 | #ifdef CONFIG_SYSCTL |
a44a4a00 NH |
351 | if (sysctl_hdr) |
352 | unregister_net_sysctl_table(sysctl_hdr); | |
db71789c | 353 | #endif |
1da177e4 LT |
354 | //xfrm6_input_fini(); |
355 | xfrm6_policy_fini(); | |
356 | xfrm6_state_fini(); | |
fc66f95c | 357 | dst_entries_destroy(&xfrm6_dst_ops); |
1da177e4 | 358 | } |