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 AD |
29 | static struct dst_entry *xfrm6_dst_lookup(struct net *net, int tos, |
30 | xfrm_address_t *saddr, | |
66cdb3ca | 31 | xfrm_address_t *daddr) |
1da177e4 | 32 | { |
66cdb3ca HX |
33 | struct flowi fl = {}; |
34 | struct dst_entry *dst; | |
35 | int err; | |
36 | ||
37 | memcpy(&fl.fl6_dst, daddr, sizeof(fl.fl6_dst)); | |
38 | if (saddr) | |
39 | memcpy(&fl.fl6_src, saddr, sizeof(fl.fl6_src)); | |
40 | ||
c5b3cf46 | 41 | dst = ip6_route_output(net, NULL, &fl); |
66cdb3ca HX |
42 | |
43 | err = dst->error; | |
44 | if (dst->error) { | |
4251320f | 45 | dst_release(dst); |
66cdb3ca HX |
46 | dst = ERR_PTR(err); |
47 | } | |
48 | ||
49 | return dst; | |
1da177e4 LT |
50 | } |
51 | ||
fbda33b2 AD |
52 | static int xfrm6_get_saddr(struct net *net, |
53 | xfrm_address_t *saddr, xfrm_address_t *daddr) | |
a1e59abf | 54 | { |
66cdb3ca | 55 | struct dst_entry *dst; |
191cd582 | 56 | struct net_device *dev; |
66cdb3ca | 57 | |
fbda33b2 | 58 | dst = xfrm6_dst_lookup(net, 0, NULL, daddr); |
66cdb3ca HX |
59 | if (IS_ERR(dst)) |
60 | return -EHOSTUNREACH; | |
61 | ||
191cd582 BH |
62 | dev = ip6_dst_idev(dst)->dev; |
63 | ipv6_dev_get_saddr(dev_net(dev), dev, | |
7cbca67c | 64 | (struct in6_addr *)&daddr->a6, 0, |
5e5f3f0f | 65 | (struct in6_addr *)&saddr->a6); |
66cdb3ca HX |
66 | dst_release(dst); |
67 | return 0; | |
a1e59abf PM |
68 | } |
69 | ||
1da177e4 LT |
70 | static struct dst_entry * |
71 | __xfrm6_find_bundle(struct flowi *fl, struct xfrm_policy *policy) | |
72 | { | |
73 | struct dst_entry *dst; | |
74 | ||
75 | /* Still not clear if we should set fl->fl6_{src,dst}... */ | |
76 | read_lock_bh(&policy->lock); | |
77 | for (dst = policy->bundles; dst; dst = dst->next) { | |
78 | struct xfrm_dst *xdst = (struct xfrm_dst*)dst; | |
79 | struct in6_addr fl_dst_prefix, fl_src_prefix; | |
80 | ||
81 | ipv6_addr_prefix(&fl_dst_prefix, | |
82 | &fl->fl6_dst, | |
83 | xdst->u.rt6.rt6i_dst.plen); | |
84 | ipv6_addr_prefix(&fl_src_prefix, | |
85 | &fl->fl6_src, | |
86 | xdst->u.rt6.rt6i_src.plen); | |
87 | if (ipv6_addr_equal(&xdst->u.rt6.rt6i_dst.addr, &fl_dst_prefix) && | |
88 | ipv6_addr_equal(&xdst->u.rt6.rt6i_src.addr, &fl_src_prefix) && | |
5b368e61 | 89 | xfrm_bundle_ok(policy, xdst, fl, AF_INET6, |
e53820de MN |
90 | (xdst->u.rt6.rt6i_dst.plen != 128 || |
91 | xdst->u.rt6.rt6i_src.plen != 128))) { | |
1da177e4 LT |
92 | dst_clone(dst); |
93 | break; | |
94 | } | |
95 | } | |
96 | read_unlock_bh(&policy->lock); | |
97 | return dst; | |
98 | } | |
99 | ||
25ee3286 | 100 | static int xfrm6_get_tos(struct flowi *fl) |
1da177e4 | 101 | { |
25ee3286 HX |
102 | return 0; |
103 | } | |
1da177e4 | 104 | |
a1b05140 MN |
105 | static int xfrm6_init_path(struct xfrm_dst *path, struct dst_entry *dst, |
106 | int nfheader_len) | |
107 | { | |
108 | if (dst->ops->family == AF_INET6) { | |
109 | struct rt6_info *rt = (struct rt6_info*)dst; | |
110 | if (rt->rt6i_node) | |
111 | path->path_cookie = rt->rt6i_node->fn_sernum; | |
112 | } | |
113 | ||
114 | path->u.rt6.rt6i_nfheader_len = nfheader_len; | |
115 | ||
116 | return 0; | |
117 | } | |
118 | ||
25ee3286 HX |
119 | static int xfrm6_fill_dst(struct xfrm_dst *xdst, struct net_device *dev) |
120 | { | |
121 | struct rt6_info *rt = (struct rt6_info*)xdst->route; | |
1da177e4 | 122 | |
25ee3286 HX |
123 | xdst->u.dst.dev = dev; |
124 | dev_hold(dev); | |
1da177e4 | 125 | |
25ee3286 HX |
126 | xdst->u.rt6.rt6i_idev = in6_dev_get(rt->u.dst.dev); |
127 | if (!xdst->u.rt6.rt6i_idev) | |
128 | return -ENODEV; | |
1da177e4 | 129 | |
25ee3286 HX |
130 | /* Sheit... I remember I did this right. Apparently, |
131 | * it was magically lost, so this code needs audit */ | |
132 | xdst->u.rt6.rt6i_flags = rt->rt6i_flags & (RTF_ANYCAST | | |
133 | RTF_LOCAL); | |
134 | xdst->u.rt6.rt6i_metric = rt->rt6i_metric; | |
135 | xdst->u.rt6.rt6i_node = rt->rt6i_node; | |
a1b05140 MN |
136 | if (rt->rt6i_node) |
137 | xdst->route_cookie = rt->rt6i_node->fn_sernum; | |
25ee3286 HX |
138 | xdst->u.rt6.rt6i_gateway = rt->rt6i_gateway; |
139 | xdst->u.rt6.rt6i_dst = rt->rt6i_dst; | |
140 | xdst->u.rt6.rt6i_src = rt->rt6i_src; | |
1da177e4 | 141 | |
1da177e4 | 142 | return 0; |
1da177e4 LT |
143 | } |
144 | ||
145 | static inline void | |
d5422efe | 146 | _decode_session6(struct sk_buff *skb, struct flowi *fl, int reverse) |
1da177e4 | 147 | { |
7e3a42a1 | 148 | int onlyproto = 0; |
cfe1fc77 | 149 | u16 offset = skb_network_header_len(skb); |
0660e03f | 150 | struct ipv6hdr *hdr = ipv6_hdr(skb); |
e3cae904 | 151 | struct ipv6_opt_hdr *exthdr; |
d56f90a7 ACM |
152 | const unsigned char *nh = skb_network_header(skb); |
153 | u8 nexthdr = nh[IP6CB(skb)->nhoff]; | |
1da177e4 LT |
154 | |
155 | memset(fl, 0, sizeof(struct flowi)); | |
d5422efe HX |
156 | ipv6_addr_copy(&fl->fl6_dst, reverse ? &hdr->saddr : &hdr->daddr); |
157 | ipv6_addr_copy(&fl->fl6_src, reverse ? &hdr->daddr : &hdr->saddr); | |
1da177e4 | 158 | |
59cae009 WY |
159 | while (nh + offset + 1 < skb->data || |
160 | pskb_may_pull(skb, nh + offset + 1 - skb->data)) { | |
d56f90a7 ACM |
161 | nh = skb_network_header(skb); |
162 | exthdr = (struct ipv6_opt_hdr *)(nh + offset); | |
e3cae904 | 163 | |
1da177e4 | 164 | switch (nexthdr) { |
7e3a42a1 ND |
165 | case NEXTHDR_FRAGMENT: |
166 | onlyproto = 1; | |
1da177e4 LT |
167 | case NEXTHDR_ROUTING: |
168 | case NEXTHDR_HOP: | |
169 | case NEXTHDR_DEST: | |
170 | offset += ipv6_optlen(exthdr); | |
171 | nexthdr = exthdr->nexthdr; | |
d56f90a7 | 172 | exthdr = (struct ipv6_opt_hdr *)(nh + offset); |
1da177e4 LT |
173 | break; |
174 | ||
175 | case IPPROTO_UDP: | |
ba4e58ec | 176 | case IPPROTO_UDPLITE: |
1da177e4 LT |
177 | case IPPROTO_TCP: |
178 | case IPPROTO_SCTP: | |
9e999993 | 179 | case IPPROTO_DCCP: |
59cae009 WY |
180 | if (!onlyproto && (nh + offset + 4 < skb->data || |
181 | pskb_may_pull(skb, nh + offset + 4 - skb->data))) { | |
8c689a6e | 182 | __be16 *ports = (__be16 *)exthdr; |
1da177e4 | 183 | |
d5422efe HX |
184 | fl->fl_ip_sport = ports[!!reverse]; |
185 | fl->fl_ip_dport = ports[!reverse]; | |
1da177e4 LT |
186 | } |
187 | fl->proto = nexthdr; | |
188 | return; | |
189 | ||
190 | case IPPROTO_ICMPV6: | |
7e3a42a1 | 191 | if (!onlyproto && pskb_may_pull(skb, nh + offset + 2 - skb->data)) { |
1da177e4 LT |
192 | u8 *icmp = (u8 *)exthdr; |
193 | ||
194 | fl->fl_icmp_type = icmp[0]; | |
195 | fl->fl_icmp_code = icmp[1]; | |
196 | } | |
197 | fl->proto = nexthdr; | |
198 | return; | |
199 | ||
59fbb3a6 | 200 | #if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE) |
2ce4272a | 201 | case IPPROTO_MH: |
7e3a42a1 | 202 | if (!onlyproto && pskb_may_pull(skb, nh + offset + 3 - skb->data)) { |
2ce4272a MN |
203 | struct ip6_mh *mh; |
204 | mh = (struct ip6_mh *)exthdr; | |
205 | ||
206 | fl->fl_mh_type = mh->ip6mh_type; | |
207 | } | |
208 | fl->proto = nexthdr; | |
209 | return; | |
210 | #endif | |
211 | ||
1da177e4 LT |
212 | /* XXX Why are there these headers? */ |
213 | case IPPROTO_AH: | |
214 | case IPPROTO_ESP: | |
215 | case IPPROTO_COMP: | |
216 | default: | |
217 | fl->fl_ipsec_spi = 0; | |
218 | fl->proto = nexthdr; | |
219 | return; | |
3ff50b79 | 220 | } |
1da177e4 LT |
221 | } |
222 | } | |
223 | ||
569d3645 | 224 | static inline int xfrm6_garbage_collect(struct dst_ops *ops) |
1da177e4 | 225 | { |
d7c7544c AD |
226 | struct net *net = container_of(ops, struct net, xfrm.xfrm6_dst_ops); |
227 | ||
228 | xfrm6_policy_afinfo.garbage_collect(net); | |
229 | return (atomic_read(&ops->entries) > ops->gc_thresh * 2); | |
1da177e4 LT |
230 | } |
231 | ||
232 | static void xfrm6_update_pmtu(struct dst_entry *dst, u32 mtu) | |
233 | { | |
234 | struct xfrm_dst *xdst = (struct xfrm_dst *)dst; | |
235 | struct dst_entry *path = xdst->route; | |
236 | ||
237 | path->ops->update_pmtu(path, mtu); | |
238 | } | |
239 | ||
aabc9761 HX |
240 | static void xfrm6_dst_destroy(struct dst_entry *dst) |
241 | { | |
242 | struct xfrm_dst *xdst = (struct xfrm_dst *)dst; | |
243 | ||
244 | if (likely(xdst->u.rt6.rt6i_idev)) | |
245 | in6_dev_put(xdst->u.rt6.rt6i_idev); | |
246 | xfrm_dst_destroy(xdst); | |
247 | } | |
248 | ||
249 | static void xfrm6_dst_ifdown(struct dst_entry *dst, struct net_device *dev, | |
250 | int unregister) | |
251 | { | |
252 | struct xfrm_dst *xdst; | |
253 | ||
254 | if (!unregister) | |
255 | return; | |
256 | ||
257 | xdst = (struct xfrm_dst *)dst; | |
258 | if (xdst->u.rt6.rt6i_idev->dev == dev) { | |
5a3e55d6 | 259 | struct inet6_dev *loopback_idev = |
c346dca1 | 260 | in6_dev_get(dev_net(dev)->loopback_dev); |
aabc9761 HX |
261 | BUG_ON(!loopback_idev); |
262 | ||
263 | do { | |
264 | in6_dev_put(xdst->u.rt6.rt6i_idev); | |
265 | xdst->u.rt6.rt6i_idev = loopback_idev; | |
266 | in6_dev_hold(loopback_idev); | |
267 | xdst = (struct xfrm_dst *)xdst->u.dst.child; | |
268 | } while (xdst->u.dst.xfrm); | |
269 | ||
270 | __in6_dev_put(loopback_idev); | |
271 | } | |
272 | ||
273 | xfrm_dst_ifdown(dst, dev); | |
274 | } | |
275 | ||
1da177e4 LT |
276 | static struct dst_ops xfrm6_dst_ops = { |
277 | .family = AF_INET6, | |
09640e63 | 278 | .protocol = cpu_to_be16(ETH_P_IPV6), |
1da177e4 LT |
279 | .gc = xfrm6_garbage_collect, |
280 | .update_pmtu = xfrm6_update_pmtu, | |
aabc9761 HX |
281 | .destroy = xfrm6_dst_destroy, |
282 | .ifdown = xfrm6_dst_ifdown, | |
862b82c6 | 283 | .local_out = __ip6_local_out, |
1da177e4 | 284 | .gc_thresh = 1024, |
e2422970 | 285 | .entries = ATOMIC_INIT(0), |
1da177e4 LT |
286 | }; |
287 | ||
288 | static struct xfrm_policy_afinfo xfrm6_policy_afinfo = { | |
289 | .family = AF_INET6, | |
1da177e4 LT |
290 | .dst_ops = &xfrm6_dst_ops, |
291 | .dst_lookup = xfrm6_dst_lookup, | |
a1e59abf | 292 | .get_saddr = xfrm6_get_saddr, |
1da177e4 | 293 | .find_bundle = __xfrm6_find_bundle, |
1da177e4 | 294 | .decode_session = _decode_session6, |
25ee3286 | 295 | .get_tos = xfrm6_get_tos, |
a1b05140 | 296 | .init_path = xfrm6_init_path, |
25ee3286 | 297 | .fill_dst = xfrm6_fill_dst, |
1da177e4 LT |
298 | }; |
299 | ||
0013caba | 300 | static int __init xfrm6_policy_init(void) |
1da177e4 | 301 | { |
0013caba | 302 | return xfrm_policy_register_afinfo(&xfrm6_policy_afinfo); |
1da177e4 LT |
303 | } |
304 | ||
305 | static void xfrm6_policy_fini(void) | |
306 | { | |
307 | xfrm_policy_unregister_afinfo(&xfrm6_policy_afinfo); | |
308 | } | |
309 | ||
db71789c | 310 | #ifdef CONFIG_SYSCTL |
a44a4a00 NH |
311 | static struct ctl_table xfrm6_policy_table[] = { |
312 | { | |
a44a4a00 | 313 | .procname = "xfrm6_gc_thresh", |
d7c7544c | 314 | .data = &init_net.xfrm.xfrm6_dst_ops.gc_thresh, |
a44a4a00 NH |
315 | .maxlen = sizeof(int), |
316 | .mode = 0644, | |
317 | .proc_handler = proc_dointvec, | |
318 | }, | |
319 | { } | |
320 | }; | |
321 | ||
322 | static struct ctl_table_header *sysctl_hdr; | |
db71789c | 323 | #endif |
a44a4a00 | 324 | |
0013caba | 325 | int __init xfrm6_init(void) |
1da177e4 | 326 | { |
0013caba | 327 | int ret; |
a33bc5c1 | 328 | unsigned int gc_thresh; |
0013caba | 329 | |
a33bc5c1 NH |
330 | /* |
331 | * We need a good default value for the xfrm6 gc threshold. | |
332 | * In ipv4 we set it to the route hash table size * 8, which | |
333 | * is half the size of the maximaum route cache for ipv4. It | |
334 | * would be good to do the same thing for v6, except the table is | |
335 | * constructed differently here. Here each table for a net namespace | |
336 | * can have FIB_TABLE_HASHSZ entries, so lets go with the same | |
337 | * computation that we used for ipv4 here. Also, lets keep the initial | |
338 | * gc_thresh to a minimum of 1024, since, the ipv6 route cache defaults | |
339 | * to that as a minimum as well | |
340 | */ | |
341 | gc_thresh = FIB6_TABLE_HASHSZ * 8; | |
342 | xfrm6_dst_ops.gc_thresh = (gc_thresh < 1024) ? 1024 : gc_thresh; | |
d7c7544c AD |
343 | |
344 | ret = xfrm6_policy_init(); | |
345 | if (ret) | |
346 | goto out; | |
347 | ||
348 | ret = xfrm6_state_init(); | |
349 | if (ret) | |
350 | goto out_policy; | |
351 | ||
db71789c | 352 | #ifdef CONFIG_SYSCTL |
a44a4a00 NH |
353 | sysctl_hdr = register_net_sysctl_table(&init_net, net_ipv6_ctl_path, |
354 | xfrm6_policy_table); | |
db71789c | 355 | #endif |
0013caba DL |
356 | out: |
357 | return ret; | |
358 | out_policy: | |
359 | xfrm6_policy_fini(); | |
360 | goto out; | |
1da177e4 LT |
361 | } |
362 | ||
363 | void xfrm6_fini(void) | |
364 | { | |
db71789c | 365 | #ifdef CONFIG_SYSCTL |
a44a4a00 NH |
366 | if (sysctl_hdr) |
367 | unregister_net_sysctl_table(sysctl_hdr); | |
db71789c | 368 | #endif |
1da177e4 LT |
369 | //xfrm6_input_fini(); |
370 | xfrm6_policy_fini(); | |
371 | xfrm6_state_fini(); | |
372 | } |