2 * xfrm6_policy.c: based on xfrm4_policy.c
6 * Kazunori MIYAZAWA @USAGI
7 * Kunihiro Ishiguro <kunihiro@ipinfusion.com>
10 * Split up af-specific portion
14 #include <linux/err.h>
15 #include <linux/kernel.h>
16 #include <linux/netdevice.h>
17 #include <net/addrconf.h>
22 #include <net/ip6_route.h>
23 #if IS_ENABLED(CONFIG_IPV6_MIP6)
27 static struct xfrm_policy_afinfo xfrm6_policy_afinfo;
29 static struct dst_entry *xfrm6_dst_lookup(struct net *net, int tos, int oif,
30 const xfrm_address_t *saddr,
31 const xfrm_address_t *daddr)
34 struct dst_entry *dst;
37 memset(&fl6, 0, sizeof(fl6));
39 memcpy(&fl6.daddr, daddr, sizeof(fl6.daddr));
41 memcpy(&fl6.saddr, saddr, sizeof(fl6.saddr));
43 dst = ip6_route_output(net, NULL, &fl6);
54 static int xfrm6_get_saddr(struct net *net, int oif,
55 xfrm_address_t *saddr, xfrm_address_t *daddr)
57 struct dst_entry *dst;
58 struct net_device *dev;
60 dst = xfrm6_dst_lookup(net, 0, oif, NULL, daddr);
64 dev = ip6_dst_idev(dst)->dev;
65 ipv6_dev_get_saddr(dev_net(dev), dev, &daddr->in6, 0, &saddr->in6);
70 static int xfrm6_get_tos(const struct flowi *fl)
75 static int xfrm6_init_path(struct xfrm_dst *path, struct dst_entry *dst,
78 if (dst->ops->family == AF_INET6) {
79 struct rt6_info *rt = (struct rt6_info *)dst;
80 path->path_cookie = rt6_get_cookie(rt);
83 path->u.rt6.rt6i_nfheader_len = nfheader_len;
88 static int xfrm6_fill_dst(struct xfrm_dst *xdst, struct net_device *dev,
89 const struct flowi *fl)
91 struct rt6_info *rt = (struct rt6_info *)xdst->route;
93 xdst->u.dst.dev = dev;
96 xdst->u.rt6.rt6i_idev = in6_dev_get(dev);
97 if (!xdst->u.rt6.rt6i_idev) {
102 /* Sheit... I remember I did this right. Apparently,
103 * it was magically lost, so this code needs audit */
104 xdst->u.rt6.rt6i_flags = rt->rt6i_flags & (RTF_ANYCAST |
106 xdst->u.rt6.rt6i_metric = rt->rt6i_metric;
107 xdst->u.rt6.rt6i_node = rt->rt6i_node;
108 xdst->route_cookie = rt6_get_cookie(rt);
109 xdst->u.rt6.rt6i_gateway = rt->rt6i_gateway;
110 xdst->u.rt6.rt6i_dst = rt->rt6i_dst;
111 xdst->u.rt6.rt6i_src = rt->rt6i_src;
117 _decode_session6(struct sk_buff *skb, struct flowi *fl, int reverse)
119 struct flowi6 *fl6 = &fl->u.ip6;
121 const struct ipv6hdr *hdr = ipv6_hdr(skb);
122 u16 offset = sizeof(*hdr);
123 struct ipv6_opt_hdr *exthdr;
124 const unsigned char *nh = skb_network_header(skb);
125 u16 nhoff = IP6CB(skb)->nhoff;
130 nhoff = offsetof(struct ipv6hdr, nexthdr);
135 oif = skb_dst(skb)->dev->ifindex;
137 memset(fl6, 0, sizeof(struct flowi6));
138 fl6->flowi6_mark = skb->mark;
139 fl6->flowi6_oif = reverse ? skb->skb_iif : oif;
141 fl6->daddr = reverse ? hdr->saddr : hdr->daddr;
142 fl6->saddr = reverse ? hdr->daddr : hdr->saddr;
144 while (nh + offset + 1 < skb->data ||
145 pskb_may_pull(skb, nh + offset + 1 - skb->data)) {
146 nh = skb_network_header(skb);
147 exthdr = (struct ipv6_opt_hdr *)(nh + offset);
150 case NEXTHDR_FRAGMENT:
152 case NEXTHDR_ROUTING:
155 offset += ipv6_optlen(exthdr);
156 nexthdr = exthdr->nexthdr;
157 exthdr = (struct ipv6_opt_hdr *)(nh + offset);
161 case IPPROTO_UDPLITE:
165 if (!onlyproto && (nh + offset + 4 < skb->data ||
166 pskb_may_pull(skb, nh + offset + 4 - skb->data))) {
169 nh = skb_network_header(skb);
170 ports = (__be16 *)(nh + offset);
171 fl6->fl6_sport = ports[!!reverse];
172 fl6->fl6_dport = ports[!reverse];
174 fl6->flowi6_proto = nexthdr;
178 if (!onlyproto && pskb_may_pull(skb, nh + offset + 2 - skb->data)) {
181 nh = skb_network_header(skb);
182 icmp = (u8 *)(nh + offset);
183 fl6->fl6_icmp_type = icmp[0];
184 fl6->fl6_icmp_code = icmp[1];
186 fl6->flowi6_proto = nexthdr;
189 #if IS_ENABLED(CONFIG_IPV6_MIP6)
191 offset += ipv6_optlen(exthdr);
192 if (!onlyproto && pskb_may_pull(skb, nh + offset + 3 - skb->data)) {
195 nh = skb_network_header(skb);
196 mh = (struct ip6_mh *)(nh + offset);
197 fl6->fl6_mh_type = mh->ip6mh_type;
199 fl6->flowi6_proto = nexthdr;
203 /* XXX Why are there these headers? */
208 fl6->fl6_ipsec_spi = 0;
209 fl6->flowi6_proto = nexthdr;
215 static inline int xfrm6_garbage_collect(struct dst_ops *ops)
217 struct net *net = container_of(ops, struct net, xfrm.xfrm6_dst_ops);
219 xfrm6_policy_afinfo.garbage_collect(net);
220 return dst_entries_get_fast(ops) > ops->gc_thresh * 2;
223 static void xfrm6_update_pmtu(struct dst_entry *dst, struct sock *sk,
224 struct sk_buff *skb, u32 mtu)
226 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
227 struct dst_entry *path = xdst->route;
229 path->ops->update_pmtu(path, sk, skb, mtu);
232 static void xfrm6_redirect(struct dst_entry *dst, struct sock *sk,
235 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
236 struct dst_entry *path = xdst->route;
238 path->ops->redirect(path, sk, skb);
241 static void xfrm6_dst_destroy(struct dst_entry *dst)
243 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
245 if (likely(xdst->u.rt6.rt6i_idev))
246 in6_dev_put(xdst->u.rt6.rt6i_idev);
247 dst_destroy_metrics_generic(dst);
248 xfrm_dst_destroy(xdst);
251 static void xfrm6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
254 struct xfrm_dst *xdst;
259 xdst = (struct xfrm_dst *)dst;
260 if (xdst->u.rt6.rt6i_idev->dev == dev) {
261 struct inet6_dev *loopback_idev =
262 in6_dev_get(dev_net(dev)->loopback_dev);
263 BUG_ON(!loopback_idev);
266 in6_dev_put(xdst->u.rt6.rt6i_idev);
267 xdst->u.rt6.rt6i_idev = loopback_idev;
268 in6_dev_hold(loopback_idev);
269 xdst = (struct xfrm_dst *)xdst->u.dst.child;
270 } while (xdst->u.dst.xfrm);
272 __in6_dev_put(loopback_idev);
275 xfrm_dst_ifdown(dst, dev);
278 static struct dst_ops xfrm6_dst_ops = {
280 .gc = xfrm6_garbage_collect,
281 .update_pmtu = xfrm6_update_pmtu,
282 .redirect = xfrm6_redirect,
283 .cow_metrics = dst_cow_metrics_generic,
284 .destroy = xfrm6_dst_destroy,
285 .ifdown = xfrm6_dst_ifdown,
286 .local_out = __ip6_local_out,
290 static struct xfrm_policy_afinfo xfrm6_policy_afinfo = {
292 .dst_ops = &xfrm6_dst_ops,
293 .dst_lookup = xfrm6_dst_lookup,
294 .get_saddr = xfrm6_get_saddr,
295 .decode_session = _decode_session6,
296 .get_tos = xfrm6_get_tos,
297 .init_path = xfrm6_init_path,
298 .fill_dst = xfrm6_fill_dst,
299 .blackhole_route = ip6_blackhole_route,
302 static int __init xfrm6_policy_init(void)
304 return xfrm_policy_register_afinfo(&xfrm6_policy_afinfo);
307 static void xfrm6_policy_fini(void)
309 xfrm_policy_unregister_afinfo(&xfrm6_policy_afinfo);
313 static struct ctl_table xfrm6_policy_table[] = {
315 .procname = "xfrm6_gc_thresh",
316 .data = &init_net.xfrm.xfrm6_dst_ops.gc_thresh,
317 .maxlen = sizeof(int),
319 .proc_handler = proc_dointvec,
324 static int __net_init xfrm6_net_init(struct net *net)
326 struct ctl_table *table;
327 struct ctl_table_header *hdr;
329 table = xfrm6_policy_table;
330 if (!net_eq(net, &init_net)) {
331 table = kmemdup(table, sizeof(xfrm6_policy_table), GFP_KERNEL);
335 table[0].data = &net->xfrm.xfrm6_dst_ops.gc_thresh;
338 hdr = register_net_sysctl(net, "net/ipv6", table);
342 net->ipv6.sysctl.xfrm6_hdr = hdr;
346 if (!net_eq(net, &init_net))
352 static void __net_exit xfrm6_net_exit(struct net *net)
354 struct ctl_table *table;
356 if (!net->ipv6.sysctl.xfrm6_hdr)
359 table = net->ipv6.sysctl.xfrm6_hdr->ctl_table_arg;
360 unregister_net_sysctl_table(net->ipv6.sysctl.xfrm6_hdr);
361 if (!net_eq(net, &init_net))
365 static struct pernet_operations xfrm6_net_ops = {
366 .init = xfrm6_net_init,
367 .exit = xfrm6_net_exit,
371 int __init xfrm6_init(void)
375 dst_entries_init(&xfrm6_dst_ops);
377 ret = xfrm6_policy_init();
379 dst_entries_destroy(&xfrm6_dst_ops);
382 ret = xfrm6_state_init();
386 ret = xfrm6_protocol_init();
391 register_pernet_subsys(&xfrm6_net_ops);
402 void xfrm6_fini(void)
405 unregister_pernet_subsys(&xfrm6_net_ops);
407 xfrm6_protocol_fini();
410 dst_entries_destroy(&xfrm6_dst_ops);