blob: 8ec3d45cd1d9cb627f08196676dbe06e97967a7d [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
27static struct dst_ops xfrm6_dst_ops;
28static struct xfrm_policy_afinfo xfrm6_policy_afinfo;
29
Alexey Dobriyanc5b3cf42008-11-25 17:51:25 -080030static struct dst_entry *xfrm6_dst_lookup(struct net *net, int tos,
31 xfrm_address_t *saddr,
Herbert Xu66cdb3c2007-11-13 21:37:28 -080032 xfrm_address_t *daddr)
Linus Torvalds1da177e2005-04-16 15:20:36 -070033{
Herbert Xu66cdb3c2007-11-13 21:37:28 -080034 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
Alexey Dobriyanc5b3cf42008-11-25 17:51:25 -080042 dst = ip6_route_output(net, NULL, &fl);
Herbert Xu66cdb3c2007-11-13 21:37:28 -080043
44 err = dst->error;
45 if (dst->error) {
Ville Nuorvala42513202006-10-16 22:10:05 -070046 dst_release(dst);
Herbert Xu66cdb3c2007-11-13 21:37:28 -080047 dst = ERR_PTR(err);
48 }
49
50 return dst;
Linus Torvalds1da177e2005-04-16 15:20:36 -070051}
52
Alexey Dobriyanfbda33b2008-11-25 17:56:49 -080053static int xfrm6_get_saddr(struct net *net,
54 xfrm_address_t *saddr, xfrm_address_t *daddr)
Patrick McHardya1e59ab2006-09-19 12:57:34 -070055{
Herbert Xu66cdb3c2007-11-13 21:37:28 -080056 struct dst_entry *dst;
Brian Haley191cd582008-08-14 15:33:21 -070057 struct net_device *dev;
Patrick McHardya1e59ab2006-09-19 12:57:34 -070058
Alexey Dobriyanfbda33b2008-11-25 17:56:49 -080059 dst = xfrm6_dst_lookup(net, 0, NULL, daddr);
Herbert Xu66cdb3c2007-11-13 21:37:28 -080060 if (IS_ERR(dst))
61 return -EHOSTUNREACH;
62
Brian Haley191cd582008-08-14 15:33:21 -070063 dev = ip6_dst_idev(dst)->dev;
64 ipv6_dev_get_saddr(dev_net(dev), dev,
YOSHIFUJI Hideaki7cbca672008-03-25 09:37:42 +090065 (struct in6_addr *)&daddr->a6, 0,
YOSHIFUJI Hideaki5e5f3f02008-03-03 21:44:34 +090066 (struct in6_addr *)&saddr->a6);
Herbert Xu66cdb3c2007-11-13 21:37:28 -080067 dst_release(dst);
68 return 0;
Patrick McHardya1e59ab2006-09-19 12:57:34 -070069}
70
Linus Torvalds1da177e2005-04-16 15:20:36 -070071static 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) &&
Venkat Yekkirala5b368e62006-10-05 15:42:18 -050090 xfrm_bundle_ok(policy, xdst, fl, AF_INET6,
Masahide NAKAMURAe53820d2006-08-23 19:12:01 -070091 (xdst->u.rt6.rt6i_dst.plen != 128 ||
92 xdst->u.rt6.rt6i_src.plen != 128))) {
Linus Torvalds1da177e2005-04-16 15:20:36 -070093 dst_clone(dst);
94 break;
95 }
96 }
97 read_unlock_bh(&policy->lock);
98 return dst;
99}
100
Herbert Xu25ee3282007-12-11 09:32:34 -0800101static int xfrm6_get_tos(struct flowi *fl)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700102{
Linus Torvalds1da177e2005-04-16 15:20:36 -0700103 return 0;
Herbert Xu25ee3282007-12-11 09:32:34 -0800104}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700105
Masahide NAKAMURAa1b05142007-12-20 20:41:12 -0800106static 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
Herbert Xu25ee3282007-12-11 09:32:34 -0800120static int xfrm6_fill_dst(struct xfrm_dst *xdst, struct net_device *dev)
121{
122 struct rt6_info *rt = (struct rt6_info*)xdst->route;
123
124 xdst->u.dst.dev = dev;
125 dev_hold(dev);
126
127 xdst->u.rt6.rt6i_idev = in6_dev_get(rt->u.dst.dev);
128 if (!xdst->u.rt6.rt6i_idev)
129 return -ENODEV;
130
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;
Masahide NAKAMURAa1b05142007-12-20 20:41:12 -0800137 if (rt->rt6i_node)
138 xdst->route_cookie = rt->rt6i_node->fn_sernum;
Herbert Xu25ee3282007-12-11 09:32:34 -0800139 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;
142
143 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700144}
145
146static inline void
Herbert Xud5422ef2007-12-12 10:44:16 -0800147_decode_session6(struct sk_buff *skb, struct flowi *fl, int reverse)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700148{
Nicolas Dichtel7e3a42a2008-11-01 21:12:07 -0700149 int onlyproto = 0;
Arnaldo Carvalho de Melocfe1fc72007-03-16 17:26:39 -0300150 u16 offset = skb_network_header_len(skb);
Arnaldo Carvalho de Melo0660e032007-04-25 17:54:47 -0700151 struct ipv6hdr *hdr = ipv6_hdr(skb);
YOSHIFUJI Hideakie3cae902006-04-18 14:46:52 -0700152 struct ipv6_opt_hdr *exthdr;
Arnaldo Carvalho de Melod56f90a2007-04-10 20:50:43 -0700153 const unsigned char *nh = skb_network_header(skb);
154 u8 nexthdr = nh[IP6CB(skb)->nhoff];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700155
156 memset(fl, 0, sizeof(struct flowi));
Herbert Xud5422ef2007-12-12 10:44:16 -0800157 ipv6_addr_copy(&fl->fl6_dst, reverse ? &hdr->saddr : &hdr->daddr);
158 ipv6_addr_copy(&fl->fl6_src, reverse ? &hdr->daddr : &hdr->saddr);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700159
Wei Yongjun59cae002009-07-02 16:59:49 +0000160 while (nh + offset + 1 < skb->data ||
161 pskb_may_pull(skb, nh + offset + 1 - skb->data)) {
Arnaldo Carvalho de Melod56f90a2007-04-10 20:50:43 -0700162 nh = skb_network_header(skb);
163 exthdr = (struct ipv6_opt_hdr *)(nh + offset);
YOSHIFUJI Hideakie3cae902006-04-18 14:46:52 -0700164
Linus Torvalds1da177e2005-04-16 15:20:36 -0700165 switch (nexthdr) {
Nicolas Dichtel7e3a42a2008-11-01 21:12:07 -0700166 case NEXTHDR_FRAGMENT:
167 onlyproto = 1;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700168 case NEXTHDR_ROUTING:
169 case NEXTHDR_HOP:
170 case NEXTHDR_DEST:
171 offset += ipv6_optlen(exthdr);
172 nexthdr = exthdr->nexthdr;
Arnaldo Carvalho de Melod56f90a2007-04-10 20:50:43 -0700173 exthdr = (struct ipv6_opt_hdr *)(nh + offset);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700174 break;
175
176 case IPPROTO_UDP:
Gerrit Renkerba4e58e2006-11-27 11:10:57 -0800177 case IPPROTO_UDPLITE:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700178 case IPPROTO_TCP:
179 case IPPROTO_SCTP:
Patrick McHardy9e999992005-12-19 14:03:46 -0800180 case IPPROTO_DCCP:
Wei Yongjun59cae002009-07-02 16:59:49 +0000181 if (!onlyproto && (nh + offset + 4 < skb->data ||
182 pskb_may_pull(skb, nh + offset + 4 - skb->data))) {
Al Viro8c689a62006-11-08 00:20:21 -0800183 __be16 *ports = (__be16 *)exthdr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700184
Herbert Xud5422ef2007-12-12 10:44:16 -0800185 fl->fl_ip_sport = ports[!!reverse];
186 fl->fl_ip_dport = ports[!reverse];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700187 }
188 fl->proto = nexthdr;
189 return;
190
191 case IPPROTO_ICMPV6:
Nicolas Dichtel7e3a42a2008-11-01 21:12:07 -0700192 if (!onlyproto && pskb_may_pull(skb, nh + offset + 2 - skb->data)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700193 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
Masahide NAKAMURA59fbb3a62007-06-26 23:56:32 -0700201#if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE)
Masahide NAKAMURA2ce42722006-08-23 20:39:03 -0700202 case IPPROTO_MH:
Nicolas Dichtel7e3a42a2008-11-01 21:12:07 -0700203 if (!onlyproto && pskb_may_pull(skb, nh + offset + 3 - skb->data)) {
Masahide NAKAMURA2ce42722006-08-23 20:39:03 -0700204 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
Linus Torvalds1da177e2005-04-16 15:20:36 -0700213 /* 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;
Stephen Hemminger3ff50b72007-04-20 17:09:22 -0700221 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700222 }
223}
224
Daniel Lezcano569d3642008-01-18 03:56:57 -0800225static inline int xfrm6_garbage_collect(struct dst_ops *ops)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700226{
Alexey Dobriyanddcfd792008-11-25 17:37:23 -0800227 xfrm6_policy_afinfo.garbage_collect(&init_net);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700228 return (atomic_read(&xfrm6_dst_ops.entries) > xfrm6_dst_ops.gc_thresh*2);
229}
230
231static 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
Herbert Xuaabc9762005-05-03 16:27:10 -0700239static 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
248static 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) {
Denis V. Lunev5a3e55d2007-12-07 00:38:10 -0800258 struct inet6_dev *loopback_idev =
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900259 in6_dev_get(dev_net(dev)->loopback_dev);
Herbert Xuaabc9762005-05-03 16:27:10 -0700260 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
Linus Torvalds1da177e2005-04-16 15:20:36 -0700275static struct dst_ops xfrm6_dst_ops = {
276 .family = AF_INET6,
Harvey Harrison09640e62009-02-01 00:45:17 -0800277 .protocol = cpu_to_be16(ETH_P_IPV6),
Linus Torvalds1da177e2005-04-16 15:20:36 -0700278 .gc = xfrm6_garbage_collect,
279 .update_pmtu = xfrm6_update_pmtu,
Herbert Xuaabc9762005-05-03 16:27:10 -0700280 .destroy = xfrm6_dst_destroy,
281 .ifdown = xfrm6_dst_ifdown,
Herbert Xu862b82c2007-11-13 21:43:11 -0800282 .local_out = __ip6_local_out,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700283 .gc_thresh = 1024,
Eric Dumazete2422972008-01-30 20:07:45 -0800284 .entries = ATOMIC_INIT(0),
Linus Torvalds1da177e2005-04-16 15:20:36 -0700285};
286
287static struct xfrm_policy_afinfo xfrm6_policy_afinfo = {
288 .family = AF_INET6,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700289 .dst_ops = &xfrm6_dst_ops,
290 .dst_lookup = xfrm6_dst_lookup,
Patrick McHardya1e59ab2006-09-19 12:57:34 -0700291 .get_saddr = xfrm6_get_saddr,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700292 .find_bundle = __xfrm6_find_bundle,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700293 .decode_session = _decode_session6,
Herbert Xu25ee3282007-12-11 09:32:34 -0800294 .get_tos = xfrm6_get_tos,
Masahide NAKAMURAa1b05142007-12-20 20:41:12 -0800295 .init_path = xfrm6_init_path,
Herbert Xu25ee3282007-12-11 09:32:34 -0800296 .fill_dst = xfrm6_fill_dst,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700297};
298
Daniel Lezcano0013cab2007-12-07 00:42:11 -0800299static int __init xfrm6_policy_init(void)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700300{
Daniel Lezcano0013cab2007-12-07 00:42:11 -0800301 return xfrm_policy_register_afinfo(&xfrm6_policy_afinfo);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700302}
303
304static void xfrm6_policy_fini(void)
305{
306 xfrm_policy_unregister_afinfo(&xfrm6_policy_afinfo);
307}
308
David S. Millerdb717892009-08-04 20:32:16 -0700309#ifdef CONFIG_SYSCTL
Neil Hormana44a4a02009-07-27 08:22:46 +0000310static struct ctl_table xfrm6_policy_table[] = {
311 {
312 .ctl_name = CTL_UNNUMBERED,
313 .procname = "xfrm6_gc_thresh",
314 .data = &xfrm6_dst_ops.gc_thresh,
315 .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
330 ret = xfrm6_policy_init();
331 if (ret)
332 goto out;
333
334 ret = xfrm6_state_init();
335 if (ret)
336 goto out_policy;
Neil Hormana33bc5c2009-07-30 18:52:15 -0700337 /*
338 * We need a good default value for the xfrm6 gc threshold.
339 * In ipv4 we set it to the route hash table size * 8, which
340 * is half the size of the maximaum route cache for ipv4. It
341 * would be good to do the same thing for v6, except the table is
342 * constructed differently here. Here each table for a net namespace
343 * can have FIB_TABLE_HASHSZ entries, so lets go with the same
344 * computation that we used for ipv4 here. Also, lets keep the initial
345 * gc_thresh to a minimum of 1024, since, the ipv6 route cache defaults
346 * to that as a minimum as well
347 */
348 gc_thresh = FIB6_TABLE_HASHSZ * 8;
349 xfrm6_dst_ops.gc_thresh = (gc_thresh < 1024) ? 1024 : gc_thresh;
David S. Millerdb717892009-08-04 20:32:16 -0700350#ifdef CONFIG_SYSCTL
Neil Hormana44a4a02009-07-27 08:22:46 +0000351 sysctl_hdr = register_net_sysctl_table(&init_net, net_ipv6_ctl_path,
352 xfrm6_policy_table);
David S. Millerdb717892009-08-04 20:32:16 -0700353#endif
Daniel Lezcano0013cab2007-12-07 00:42:11 -0800354out:
355 return ret;
356out_policy:
357 xfrm6_policy_fini();
358 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700359}
360
361void xfrm6_fini(void)
362{
David S. Millerdb717892009-08-04 20:32:16 -0700363#ifdef CONFIG_SYSCTL
Neil Hormana44a4a02009-07-27 08:22:46 +0000364 if (sysctl_hdr)
365 unregister_net_sysctl_table(sysctl_hdr);
David S. Millerdb717892009-08-04 20:32:16 -0700366#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700367 //xfrm6_input_fini();
368 xfrm6_policy_fini();
369 xfrm6_state_fini();
370}