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 LT |
26 | |
27 | static struct dst_ops xfrm6_dst_ops; | |
28 | static struct xfrm_policy_afinfo xfrm6_policy_afinfo; | |
29 | ||
c5b3cf46 AD |
30 | static struct dst_entry *xfrm6_dst_lookup(struct net *net, int tos, |
31 | xfrm_address_t *saddr, | |
66cdb3ca | 32 | xfrm_address_t *daddr) |
1da177e4 | 33 | { |
66cdb3ca HX |
34 | struct flowi fl = {}; |
35 | struct dst_entry *dst; | |
36 | int err; | |
37 | ||
38 | memcpy(&fl.fl6_dst, daddr, sizeof(fl.fl6_dst)); | |
39 | if (saddr) | |
40 | memcpy(&fl.fl6_src, saddr, sizeof(fl.fl6_src)); | |
41 | ||
c5b3cf46 | 42 | dst = ip6_route_output(net, NULL, &fl); |
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 | ||
1da177e4 LT |
71 | static struct dst_entry * |
72 | __xfrm6_find_bundle(struct flowi *fl, struct xfrm_policy *policy) | |
73 | { | |
74 | struct dst_entry *dst; | |
75 | ||
76 | /* Still not clear if we should set fl->fl6_{src,dst}... */ | |
77 | read_lock_bh(&policy->lock); | |
78 | for (dst = policy->bundles; dst; dst = dst->next) { | |
79 | struct xfrm_dst *xdst = (struct xfrm_dst*)dst; | |
80 | struct in6_addr fl_dst_prefix, fl_src_prefix; | |
81 | ||
82 | ipv6_addr_prefix(&fl_dst_prefix, | |
83 | &fl->fl6_dst, | |
84 | xdst->u.rt6.rt6i_dst.plen); | |
85 | ipv6_addr_prefix(&fl_src_prefix, | |
86 | &fl->fl6_src, | |
87 | xdst->u.rt6.rt6i_src.plen); | |
88 | if (ipv6_addr_equal(&xdst->u.rt6.rt6i_dst.addr, &fl_dst_prefix) && | |
89 | ipv6_addr_equal(&xdst->u.rt6.rt6i_src.addr, &fl_src_prefix) && | |
5b368e61 | 90 | xfrm_bundle_ok(policy, xdst, fl, AF_INET6, |
e53820de MN |
91 | (xdst->u.rt6.rt6i_dst.plen != 128 || |
92 | xdst->u.rt6.rt6i_src.plen != 128))) { | |
1da177e4 LT |
93 | dst_clone(dst); |
94 | break; | |
95 | } | |
96 | } | |
97 | read_unlock_bh(&policy->lock); | |
98 | return dst; | |
99 | } | |
100 | ||
25ee3286 | 101 | static int xfrm6_get_tos(struct flowi *fl) |
1da177e4 | 102 | { |
25ee3286 HX |
103 | return 0; |
104 | } | |
1da177e4 | 105 | |
a1b05140 MN |
106 | static int xfrm6_init_path(struct xfrm_dst *path, struct dst_entry *dst, |
107 | int nfheader_len) | |
108 | { | |
109 | if (dst->ops->family == AF_INET6) { | |
110 | struct rt6_info *rt = (struct rt6_info*)dst; | |
111 | if (rt->rt6i_node) | |
112 | path->path_cookie = rt->rt6i_node->fn_sernum; | |
113 | } | |
114 | ||
115 | path->u.rt6.rt6i_nfheader_len = nfheader_len; | |
116 | ||
117 | return 0; | |
118 | } | |
119 | ||
25ee3286 HX |
120 | static int xfrm6_fill_dst(struct xfrm_dst *xdst, struct net_device *dev) |
121 | { | |
122 | struct rt6_info *rt = (struct rt6_info*)xdst->route; | |
1da177e4 | 123 | |
25ee3286 HX |
124 | xdst->u.dst.dev = dev; |
125 | dev_hold(dev); | |
1da177e4 | 126 | |
25ee3286 HX |
127 | xdst->u.rt6.rt6i_idev = in6_dev_get(rt->u.dst.dev); |
128 | if (!xdst->u.rt6.rt6i_idev) | |
129 | return -ENODEV; | |
1da177e4 | 130 | |
25ee3286 HX |
131 | /* Sheit... I remember I did this right. Apparently, |
132 | * it was magically lost, so this code needs audit */ | |
133 | xdst->u.rt6.rt6i_flags = rt->rt6i_flags & (RTF_ANYCAST | | |
134 | RTF_LOCAL); | |
135 | xdst->u.rt6.rt6i_metric = rt->rt6i_metric; | |
136 | xdst->u.rt6.rt6i_node = rt->rt6i_node; | |
a1b05140 MN |
137 | if (rt->rt6i_node) |
138 | xdst->route_cookie = rt->rt6i_node->fn_sernum; | |
25ee3286 HX |
139 | xdst->u.rt6.rt6i_gateway = rt->rt6i_gateway; |
140 | xdst->u.rt6.rt6i_dst = rt->rt6i_dst; | |
141 | xdst->u.rt6.rt6i_src = rt->rt6i_src; | |
1da177e4 | 142 | |
1da177e4 | 143 | return 0; |
1da177e4 LT |
144 | } |
145 | ||
146 | static inline void | |
d5422efe | 147 | _decode_session6(struct sk_buff *skb, struct flowi *fl, int reverse) |
1da177e4 | 148 | { |
7e3a42a1 | 149 | int onlyproto = 0; |
cfe1fc77 | 150 | u16 offset = skb_network_header_len(skb); |
0660e03f | 151 | struct ipv6hdr *hdr = ipv6_hdr(skb); |
e3cae904 | 152 | struct ipv6_opt_hdr *exthdr; |
d56f90a7 ACM |
153 | const unsigned char *nh = skb_network_header(skb); |
154 | u8 nexthdr = nh[IP6CB(skb)->nhoff]; | |
1da177e4 LT |
155 | |
156 | memset(fl, 0, sizeof(struct flowi)); | |
d5422efe HX |
157 | ipv6_addr_copy(&fl->fl6_dst, reverse ? &hdr->saddr : &hdr->daddr); |
158 | ipv6_addr_copy(&fl->fl6_src, reverse ? &hdr->daddr : &hdr->saddr); | |
1da177e4 | 159 | |
59cae009 WY |
160 | while (nh + offset + 1 < skb->data || |
161 | pskb_may_pull(skb, nh + offset + 1 - skb->data)) { | |
d56f90a7 ACM |
162 | nh = skb_network_header(skb); |
163 | exthdr = (struct ipv6_opt_hdr *)(nh + offset); | |
e3cae904 | 164 | |
1da177e4 | 165 | switch (nexthdr) { |
7e3a42a1 ND |
166 | case NEXTHDR_FRAGMENT: |
167 | onlyproto = 1; | |
1da177e4 LT |
168 | case NEXTHDR_ROUTING: |
169 | case NEXTHDR_HOP: | |
170 | case NEXTHDR_DEST: | |
171 | offset += ipv6_optlen(exthdr); | |
172 | nexthdr = exthdr->nexthdr; | |
d56f90a7 | 173 | exthdr = (struct ipv6_opt_hdr *)(nh + offset); |
1da177e4 LT |
174 | break; |
175 | ||
176 | case IPPROTO_UDP: | |
ba4e58ec | 177 | case IPPROTO_UDPLITE: |
1da177e4 LT |
178 | case IPPROTO_TCP: |
179 | case IPPROTO_SCTP: | |
9e999993 | 180 | case IPPROTO_DCCP: |
59cae009 WY |
181 | if (!onlyproto && (nh + offset + 4 < skb->data || |
182 | pskb_may_pull(skb, nh + offset + 4 - skb->data))) { | |
8c689a6e | 183 | __be16 *ports = (__be16 *)exthdr; |
1da177e4 | 184 | |
d5422efe HX |
185 | fl->fl_ip_sport = ports[!!reverse]; |
186 | fl->fl_ip_dport = ports[!reverse]; | |
1da177e4 LT |
187 | } |
188 | fl->proto = nexthdr; | |
189 | return; | |
190 | ||
191 | case IPPROTO_ICMPV6: | |
7e3a42a1 | 192 | if (!onlyproto && pskb_may_pull(skb, nh + offset + 2 - skb->data)) { |
1da177e4 LT |
193 | u8 *icmp = (u8 *)exthdr; |
194 | ||
195 | fl->fl_icmp_type = icmp[0]; | |
196 | fl->fl_icmp_code = icmp[1]; | |
197 | } | |
198 | fl->proto = nexthdr; | |
199 | return; | |
200 | ||
59fbb3a6 | 201 | #if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE) |
2ce4272a | 202 | case IPPROTO_MH: |
7e3a42a1 | 203 | if (!onlyproto && pskb_may_pull(skb, nh + offset + 3 - skb->data)) { |
2ce4272a MN |
204 | struct ip6_mh *mh; |
205 | mh = (struct ip6_mh *)exthdr; | |
206 | ||
207 | fl->fl_mh_type = mh->ip6mh_type; | |
208 | } | |
209 | fl->proto = nexthdr; | |
210 | return; | |
211 | #endif | |
212 | ||
1da177e4 LT |
213 | /* XXX Why are there these headers? */ |
214 | case IPPROTO_AH: | |
215 | case IPPROTO_ESP: | |
216 | case IPPROTO_COMP: | |
217 | default: | |
218 | fl->fl_ipsec_spi = 0; | |
219 | fl->proto = nexthdr; | |
220 | return; | |
3ff50b79 | 221 | } |
1da177e4 LT |
222 | } |
223 | } | |
224 | ||
569d3645 | 225 | static inline int xfrm6_garbage_collect(struct dst_ops *ops) |
1da177e4 | 226 | { |
ddcfd796 | 227 | xfrm6_policy_afinfo.garbage_collect(&init_net); |
1da177e4 LT |
228 | return (atomic_read(&xfrm6_dst_ops.entries) > xfrm6_dst_ops.gc_thresh*2); |
229 | } | |
230 | ||
231 | static void xfrm6_update_pmtu(struct dst_entry *dst, u32 mtu) | |
232 | { | |
233 | struct xfrm_dst *xdst = (struct xfrm_dst *)dst; | |
234 | struct dst_entry *path = xdst->route; | |
235 | ||
236 | path->ops->update_pmtu(path, mtu); | |
237 | } | |
238 | ||
aabc9761 HX |
239 | static void xfrm6_dst_destroy(struct dst_entry *dst) |
240 | { | |
241 | struct xfrm_dst *xdst = (struct xfrm_dst *)dst; | |
242 | ||
243 | if (likely(xdst->u.rt6.rt6i_idev)) | |
244 | in6_dev_put(xdst->u.rt6.rt6i_idev); | |
245 | xfrm_dst_destroy(xdst); | |
246 | } | |
247 | ||
248 | static void xfrm6_dst_ifdown(struct dst_entry *dst, struct net_device *dev, | |
249 | int unregister) | |
250 | { | |
251 | struct xfrm_dst *xdst; | |
252 | ||
253 | if (!unregister) | |
254 | return; | |
255 | ||
256 | xdst = (struct xfrm_dst *)dst; | |
257 | if (xdst->u.rt6.rt6i_idev->dev == dev) { | |
5a3e55d6 | 258 | struct inet6_dev *loopback_idev = |
c346dca1 | 259 | in6_dev_get(dev_net(dev)->loopback_dev); |
aabc9761 HX |
260 | BUG_ON(!loopback_idev); |
261 | ||
262 | do { | |
263 | in6_dev_put(xdst->u.rt6.rt6i_idev); | |
264 | xdst->u.rt6.rt6i_idev = loopback_idev; | |
265 | in6_dev_hold(loopback_idev); | |
266 | xdst = (struct xfrm_dst *)xdst->u.dst.child; | |
267 | } while (xdst->u.dst.xfrm); | |
268 | ||
269 | __in6_dev_put(loopback_idev); | |
270 | } | |
271 | ||
272 | xfrm_dst_ifdown(dst, dev); | |
273 | } | |
274 | ||
1da177e4 LT |
275 | static struct dst_ops xfrm6_dst_ops = { |
276 | .family = AF_INET6, | |
09640e63 | 277 | .protocol = cpu_to_be16(ETH_P_IPV6), |
1da177e4 LT |
278 | .gc = xfrm6_garbage_collect, |
279 | .update_pmtu = xfrm6_update_pmtu, | |
aabc9761 HX |
280 | .destroy = xfrm6_dst_destroy, |
281 | .ifdown = xfrm6_dst_ifdown, | |
862b82c6 | 282 | .local_out = __ip6_local_out, |
1da177e4 | 283 | .gc_thresh = 1024, |
e2422970 | 284 | .entries = ATOMIC_INIT(0), |
1da177e4 LT |
285 | }; |
286 | ||
287 | static struct xfrm_policy_afinfo xfrm6_policy_afinfo = { | |
288 | .family = AF_INET6, | |
1da177e4 LT |
289 | .dst_ops = &xfrm6_dst_ops, |
290 | .dst_lookup = xfrm6_dst_lookup, | |
a1e59abf | 291 | .get_saddr = xfrm6_get_saddr, |
1da177e4 | 292 | .find_bundle = __xfrm6_find_bundle, |
1da177e4 | 293 | .decode_session = _decode_session6, |
25ee3286 | 294 | .get_tos = xfrm6_get_tos, |
a1b05140 | 295 | .init_path = xfrm6_init_path, |
25ee3286 | 296 | .fill_dst = xfrm6_fill_dst, |
1da177e4 LT |
297 | }; |
298 | ||
0013caba | 299 | static int __init xfrm6_policy_init(void) |
1da177e4 | 300 | { |
0013caba | 301 | return xfrm_policy_register_afinfo(&xfrm6_policy_afinfo); |
1da177e4 LT |
302 | } |
303 | ||
304 | static void xfrm6_policy_fini(void) | |
305 | { | |
306 | xfrm_policy_unregister_afinfo(&xfrm6_policy_afinfo); | |
307 | } | |
308 | ||
db71789c | 309 | #ifdef CONFIG_SYSCTL |
a44a4a00 NH |
310 | static struct ctl_table xfrm6_policy_table[] = { |
311 | { | |
a44a4a00 NH |
312 | .procname = "xfrm6_gc_thresh", |
313 | .data = &xfrm6_dst_ops.gc_thresh, | |
314 | .maxlen = sizeof(int), | |
315 | .mode = 0644, | |
316 | .proc_handler = proc_dointvec, | |
317 | }, | |
318 | { } | |
319 | }; | |
320 | ||
321 | static struct ctl_table_header *sysctl_hdr; | |
db71789c | 322 | #endif |
a44a4a00 | 323 | |
0013caba | 324 | int __init xfrm6_init(void) |
1da177e4 | 325 | { |
0013caba | 326 | int ret; |
a33bc5c1 | 327 | unsigned int gc_thresh; |
0013caba DL |
328 | |
329 | ret = xfrm6_policy_init(); | |
330 | if (ret) | |
331 | goto out; | |
332 | ||
333 | ret = xfrm6_state_init(); | |
334 | if (ret) | |
335 | goto out_policy; | |
a33bc5c1 NH |
336 | /* |
337 | * We need a good default value for the xfrm6 gc threshold. | |
338 | * In ipv4 we set it to the route hash table size * 8, which | |
339 | * is half the size of the maximaum route cache for ipv4. It | |
340 | * would be good to do the same thing for v6, except the table is | |
341 | * constructed differently here. Here each table for a net namespace | |
342 | * can have FIB_TABLE_HASHSZ entries, so lets go with the same | |
343 | * computation that we used for ipv4 here. Also, lets keep the initial | |
344 | * gc_thresh to a minimum of 1024, since, the ipv6 route cache defaults | |
345 | * to that as a minimum as well | |
346 | */ | |
347 | gc_thresh = FIB6_TABLE_HASHSZ * 8; | |
348 | xfrm6_dst_ops.gc_thresh = (gc_thresh < 1024) ? 1024 : gc_thresh; | |
db71789c | 349 | #ifdef CONFIG_SYSCTL |
a44a4a00 NH |
350 | sysctl_hdr = register_net_sysctl_table(&init_net, net_ipv6_ctl_path, |
351 | xfrm6_policy_table); | |
db71789c | 352 | #endif |
0013caba DL |
353 | out: |
354 | return ret; | |
355 | out_policy: | |
356 | xfrm6_policy_fini(); | |
357 | goto out; | |
1da177e4 LT |
358 | } |
359 | ||
360 | void xfrm6_fini(void) | |
361 | { | |
db71789c | 362 | #ifdef CONFIG_SYSCTL |
a44a4a00 NH |
363 | if (sysctl_hdr) |
364 | unregister_net_sysctl_table(sysctl_hdr); | |
db71789c | 365 | #endif |
1da177e4 LT |
366 | //xfrm6_input_fini(); |
367 | xfrm6_policy_fini(); | |
368 | xfrm6_state_fini(); | |
369 | } |