blob: dbdc696f5fc5b5fbc207b48135fa768028f7e7f4 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
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
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +090011 *
Linus Torvalds1da177e2005-04-16 15:20:36 -070012 */
13
Herbert Xu66cdb3c2007-11-13 21:37:28 -080014#include <linux/err.h>
15#include <linux/kernel.h>
Herbert Xuaabc9762005-05-03 16:27:10 -070016#include <linux/netdevice.h>
17#include <net/addrconf.h>
Herbert Xu45ff5a32007-11-13 21:35:32 -080018#include <net/dst.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070019#include <net/xfrm.h>
20#include <net/ip.h>
21#include <net/ipv6.h>
22#include <net/ip6_route.h>
Masahide NAKAMURA59fbb3a62007-06-26 23:56:32 -070023#if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE)
Masahide NAKAMURA2ce42722006-08-23 20:39:03 -070024#include <net/mip6.h>
25#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -070026
Linus Torvalds1da177e2005-04-16 15:20:36 -070027static struct xfrm_policy_afinfo xfrm6_policy_afinfo;
28
Alexey Dobriyanc5b3cf42008-11-25 17:51:25 -080029static struct dst_entry *xfrm6_dst_lookup(struct net *net, int tos,
30 xfrm_address_t *saddr,
Herbert Xu66cdb3c2007-11-13 21:37:28 -080031 xfrm_address_t *daddr)
Linus Torvalds1da177e2005-04-16 15:20:36 -070032{
Herbert Xu66cdb3c2007-11-13 21:37:28 -080033 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
Alexey Dobriyanc5b3cf42008-11-25 17:51:25 -080041 dst = ip6_route_output(net, NULL, &fl);
Herbert Xu66cdb3c2007-11-13 21:37:28 -080042
43 err = dst->error;
44 if (dst->error) {
Ville Nuorvala42513202006-10-16 22:10:05 -070045 dst_release(dst);
Herbert Xu66cdb3c2007-11-13 21:37:28 -080046 dst = ERR_PTR(err);
47 }
48
49 return dst;
Linus Torvalds1da177e2005-04-16 15:20:36 -070050}
51
Alexey Dobriyanfbda33b2008-11-25 17:56:49 -080052static int xfrm6_get_saddr(struct net *net,
53 xfrm_address_t *saddr, xfrm_address_t *daddr)
Patrick McHardya1e59ab2006-09-19 12:57:34 -070054{
Herbert Xu66cdb3c2007-11-13 21:37:28 -080055 struct dst_entry *dst;
Brian Haley191cd582008-08-14 15:33:21 -070056 struct net_device *dev;
Patrick McHardya1e59ab2006-09-19 12:57:34 -070057
Alexey Dobriyanfbda33b2008-11-25 17:56:49 -080058 dst = xfrm6_dst_lookup(net, 0, NULL, daddr);
Herbert Xu66cdb3c2007-11-13 21:37:28 -080059 if (IS_ERR(dst))
60 return -EHOSTUNREACH;
61
Brian Haley191cd582008-08-14 15:33:21 -070062 dev = ip6_dst_idev(dst)->dev;
63 ipv6_dev_get_saddr(dev_net(dev), dev,
YOSHIFUJI Hideaki7cbca672008-03-25 09:37:42 +090064 (struct in6_addr *)&daddr->a6, 0,
YOSHIFUJI Hideaki5e5f3f02008-03-03 21:44:34 +090065 (struct in6_addr *)&saddr->a6);
Herbert Xu66cdb3c2007-11-13 21:37:28 -080066 dst_release(dst);
67 return 0;
Patrick McHardya1e59ab2006-09-19 12:57:34 -070068}
69
Linus Torvalds1da177e2005-04-16 15:20:36 -070070static 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) &&
Venkat Yekkirala5b368e62006-10-05 15:42:18 -050089 xfrm_bundle_ok(policy, xdst, fl, AF_INET6,
Masahide NAKAMURAe53820d2006-08-23 19:12:01 -070090 (xdst->u.rt6.rt6i_dst.plen != 128 ||
91 xdst->u.rt6.rt6i_src.plen != 128))) {
Linus Torvalds1da177e2005-04-16 15:20:36 -070092 dst_clone(dst);
93 break;
94 }
95 }
96 read_unlock_bh(&policy->lock);
97 return dst;
98}
99
Herbert Xu25ee3282007-12-11 09:32:34 -0800100static int xfrm6_get_tos(struct flowi *fl)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700101{
Linus Torvalds1da177e2005-04-16 15:20:36 -0700102 return 0;
Herbert Xu25ee3282007-12-11 09:32:34 -0800103}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700104
Masahide NAKAMURAa1b05142007-12-20 20:41:12 -0800105static 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
Herbert Xu25ee3282007-12-11 09:32:34 -0800119static int xfrm6_fill_dst(struct xfrm_dst *xdst, struct net_device *dev)
120{
121 struct rt6_info *rt = (struct rt6_info*)xdst->route;
122
123 xdst->u.dst.dev = dev;
124 dev_hold(dev);
125
126 xdst->u.rt6.rt6i_idev = in6_dev_get(rt->u.dst.dev);
127 if (!xdst->u.rt6.rt6i_idev)
128 return -ENODEV;
129
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;
Masahide NAKAMURAa1b05142007-12-20 20:41:12 -0800136 if (rt->rt6i_node)
137 xdst->route_cookie = rt->rt6i_node->fn_sernum;
Herbert Xu25ee3282007-12-11 09:32:34 -0800138 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;
141
142 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700143}
144
145static inline void
Herbert Xud5422ef2007-12-12 10:44:16 -0800146_decode_session6(struct sk_buff *skb, struct flowi *fl, int reverse)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700147{
Nicolas Dichtel7e3a42a2008-11-01 21:12:07 -0700148 int onlyproto = 0;
Arnaldo Carvalho de Melocfe1fc72007-03-16 17:26:39 -0300149 u16 offset = skb_network_header_len(skb);
Arnaldo Carvalho de Melo0660e032007-04-25 17:54:47 -0700150 struct ipv6hdr *hdr = ipv6_hdr(skb);
YOSHIFUJI Hideakie3cae902006-04-18 14:46:52 -0700151 struct ipv6_opt_hdr *exthdr;
Arnaldo Carvalho de Melod56f90a2007-04-10 20:50:43 -0700152 const unsigned char *nh = skb_network_header(skb);
153 u8 nexthdr = nh[IP6CB(skb)->nhoff];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700154
155 memset(fl, 0, sizeof(struct flowi));
Herbert Xud5422ef2007-12-12 10:44:16 -0800156 ipv6_addr_copy(&fl->fl6_dst, reverse ? &hdr->saddr : &hdr->daddr);
157 ipv6_addr_copy(&fl->fl6_src, reverse ? &hdr->daddr : &hdr->saddr);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700158
Wei Yongjun59cae002009-07-02 16:59:49 +0000159 while (nh + offset + 1 < skb->data ||
160 pskb_may_pull(skb, nh + offset + 1 - skb->data)) {
Arnaldo Carvalho de Melod56f90a2007-04-10 20:50:43 -0700161 nh = skb_network_header(skb);
162 exthdr = (struct ipv6_opt_hdr *)(nh + offset);
YOSHIFUJI Hideakie3cae902006-04-18 14:46:52 -0700163
Linus Torvalds1da177e2005-04-16 15:20:36 -0700164 switch (nexthdr) {
Nicolas Dichtel7e3a42a2008-11-01 21:12:07 -0700165 case NEXTHDR_FRAGMENT:
166 onlyproto = 1;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700167 case NEXTHDR_ROUTING:
168 case NEXTHDR_HOP:
169 case NEXTHDR_DEST:
170 offset += ipv6_optlen(exthdr);
171 nexthdr = exthdr->nexthdr;
Arnaldo Carvalho de Melod56f90a2007-04-10 20:50:43 -0700172 exthdr = (struct ipv6_opt_hdr *)(nh + offset);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700173 break;
174
175 case IPPROTO_UDP:
Gerrit Renkerba4e58e2006-11-27 11:10:57 -0800176 case IPPROTO_UDPLITE:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700177 case IPPROTO_TCP:
178 case IPPROTO_SCTP:
Patrick McHardy9e999992005-12-19 14:03:46 -0800179 case IPPROTO_DCCP:
Wei Yongjun59cae002009-07-02 16:59:49 +0000180 if (!onlyproto && (nh + offset + 4 < skb->data ||
181 pskb_may_pull(skb, nh + offset + 4 - skb->data))) {
Al Viro8c689a62006-11-08 00:20:21 -0800182 __be16 *ports = (__be16 *)exthdr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700183
Herbert Xud5422ef2007-12-12 10:44:16 -0800184 fl->fl_ip_sport = ports[!!reverse];
185 fl->fl_ip_dport = ports[!reverse];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700186 }
187 fl->proto = nexthdr;
188 return;
189
190 case IPPROTO_ICMPV6:
Nicolas Dichtel7e3a42a2008-11-01 21:12:07 -0700191 if (!onlyproto && pskb_may_pull(skb, nh + offset + 2 - skb->data)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700192 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
Masahide NAKAMURA59fbb3a62007-06-26 23:56:32 -0700200#if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE)
Masahide NAKAMURA2ce42722006-08-23 20:39:03 -0700201 case IPPROTO_MH:
Nicolas Dichtel7e3a42a2008-11-01 21:12:07 -0700202 if (!onlyproto && pskb_may_pull(skb, nh + offset + 3 - skb->data)) {
Masahide NAKAMURA2ce42722006-08-23 20:39:03 -0700203 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
Linus Torvalds1da177e2005-04-16 15:20:36 -0700212 /* 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;
Stephen Hemminger3ff50b72007-04-20 17:09:22 -0700220 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700221 }
222}
223
Daniel Lezcano569d3642008-01-18 03:56:57 -0800224static inline int xfrm6_garbage_collect(struct dst_ops *ops)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700225{
Alexey Dobriyand7c75442010-01-24 22:47:53 -0800226 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);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700230}
231
232static 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
Herbert Xuaabc9762005-05-03 16:27:10 -0700240static 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
249static 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) {
Denis V. Lunev5a3e55d2007-12-07 00:38:10 -0800259 struct inet6_dev *loopback_idev =
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900260 in6_dev_get(dev_net(dev)->loopback_dev);
Herbert Xuaabc9762005-05-03 16:27:10 -0700261 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
Linus Torvalds1da177e2005-04-16 15:20:36 -0700276static struct dst_ops xfrm6_dst_ops = {
277 .family = AF_INET6,
Harvey Harrison09640e62009-02-01 00:45:17 -0800278 .protocol = cpu_to_be16(ETH_P_IPV6),
Linus Torvalds1da177e2005-04-16 15:20:36 -0700279 .gc = xfrm6_garbage_collect,
280 .update_pmtu = xfrm6_update_pmtu,
Herbert Xuaabc9762005-05-03 16:27:10 -0700281 .destroy = xfrm6_dst_destroy,
282 .ifdown = xfrm6_dst_ifdown,
Herbert Xu862b82c2007-11-13 21:43:11 -0800283 .local_out = __ip6_local_out,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700284 .gc_thresh = 1024,
Eric Dumazete2422972008-01-30 20:07:45 -0800285 .entries = ATOMIC_INIT(0),
Linus Torvalds1da177e2005-04-16 15:20:36 -0700286};
287
288static struct xfrm_policy_afinfo xfrm6_policy_afinfo = {
289 .family = AF_INET6,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700290 .dst_ops = &xfrm6_dst_ops,
291 .dst_lookup = xfrm6_dst_lookup,
Patrick McHardya1e59ab2006-09-19 12:57:34 -0700292 .get_saddr = xfrm6_get_saddr,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700293 .find_bundle = __xfrm6_find_bundle,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700294 .decode_session = _decode_session6,
Herbert Xu25ee3282007-12-11 09:32:34 -0800295 .get_tos = xfrm6_get_tos,
Masahide NAKAMURAa1b05142007-12-20 20:41:12 -0800296 .init_path = xfrm6_init_path,
Herbert Xu25ee3282007-12-11 09:32:34 -0800297 .fill_dst = xfrm6_fill_dst,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700298};
299
Daniel Lezcano0013cab2007-12-07 00:42:11 -0800300static int __init xfrm6_policy_init(void)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700301{
Daniel Lezcano0013cab2007-12-07 00:42:11 -0800302 return xfrm_policy_register_afinfo(&xfrm6_policy_afinfo);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700303}
304
305static void xfrm6_policy_fini(void)
306{
307 xfrm_policy_unregister_afinfo(&xfrm6_policy_afinfo);
308}
309
David S. Millerdb717892009-08-04 20:32:16 -0700310#ifdef CONFIG_SYSCTL
Neil Hormana44a4a02009-07-27 08:22:46 +0000311static struct ctl_table xfrm6_policy_table[] = {
312 {
Neil Hormana44a4a02009-07-27 08:22:46 +0000313 .procname = "xfrm6_gc_thresh",
Alexey Dobriyand7c75442010-01-24 22:47:53 -0800314 .data = &init_net.xfrm.xfrm6_dst_ops.gc_thresh,
Neil Hormana44a4a02009-07-27 08:22:46 +0000315 .maxlen = sizeof(int),
316 .mode = 0644,
317 .proc_handler = proc_dointvec,
318 },
319 { }
320};
321
322static struct ctl_table_header *sysctl_hdr;
David S. Millerdb717892009-08-04 20:32:16 -0700323#endif
Neil Hormana44a4a02009-07-27 08:22:46 +0000324
Daniel Lezcano0013cab2007-12-07 00:42:11 -0800325int __init xfrm6_init(void)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700326{
Daniel Lezcano0013cab2007-12-07 00:42:11 -0800327 int ret;
Neil Hormana33bc5c2009-07-30 18:52:15 -0700328 unsigned int gc_thresh;
Daniel Lezcano0013cab2007-12-07 00:42:11 -0800329
Neil Hormana33bc5c2009-07-30 18:52:15 -0700330 /*
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;
Alexey Dobriyand7c75442010-01-24 22:47:53 -0800343
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
David S. Millerdb717892009-08-04 20:32:16 -0700352#ifdef CONFIG_SYSCTL
Neil Hormana44a4a02009-07-27 08:22:46 +0000353 sysctl_hdr = register_net_sysctl_table(&init_net, net_ipv6_ctl_path,
354 xfrm6_policy_table);
David S. Millerdb717892009-08-04 20:32:16 -0700355#endif
Daniel Lezcano0013cab2007-12-07 00:42:11 -0800356out:
357 return ret;
358out_policy:
359 xfrm6_policy_fini();
360 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700361}
362
363void xfrm6_fini(void)
364{
David S. Millerdb717892009-08-04 20:32:16 -0700365#ifdef CONFIG_SYSCTL
Neil Hormana44a4a02009-07-27 08:22:46 +0000366 if (sysctl_hdr)
367 unregister_net_sysctl_table(sysctl_hdr);
David S. Millerdb717892009-08-04 20:32:16 -0700368#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700369 //xfrm6_input_fini();
370 xfrm6_policy_fini();
371 xfrm6_state_fini();
372}