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,
30 const xfrm_address_t *saddr,
31 const xfrm_address_t *daddr)
34 struct dst_entry *dst;
37 memset(&fl6, 0, sizeof(fl6));
38 memcpy(&fl6.daddr, daddr, sizeof(fl6.daddr));
40 memcpy(&fl6.saddr, saddr, sizeof(fl6.saddr));
42 dst = ip6_route_output(net, NULL, &fl6);
53 static int xfrm6_get_saddr(struct net *net,
54 xfrm_address_t *saddr, xfrm_address_t *daddr)
56 struct dst_entry *dst;
57 struct net_device *dev;
59 dst = xfrm6_dst_lookup(net, 0, NULL, daddr);
63 dev = ip6_dst_idev(dst)->dev;
64 ipv6_dev_get_saddr(dev_net(dev), dev,
65 (struct in6_addr *)&daddr->a6, 0,
66 (struct in6_addr *)&saddr->a6);
71 static int xfrm6_get_tos(const struct flowi *fl)
76 static void xfrm6_init_dst(struct net *net, struct xfrm_dst *xdst)
78 struct rt6_info *rt = (struct rt6_info *)xdst;
80 rt6_init_peer(rt, net->ipv6.peers);
83 static int xfrm6_init_path(struct xfrm_dst *path, struct dst_entry *dst,
86 if (dst->ops->family == AF_INET6) {
87 struct rt6_info *rt = (struct rt6_info*)dst;
89 path->path_cookie = rt->rt6i_node->fn_sernum;
92 path->u.rt6.rt6i_nfheader_len = nfheader_len;
97 static int xfrm6_fill_dst(struct xfrm_dst *xdst, struct net_device *dev,
98 const struct flowi *fl)
100 struct rt6_info *rt = (struct rt6_info*)xdst->route;
102 xdst->u.dst.dev = dev;
105 xdst->u.rt6.rt6i_idev = in6_dev_get(dev);
106 if (!xdst->u.rt6.rt6i_idev) {
111 rt6_transfer_peer(&xdst->u.rt6, rt);
113 /* Sheit... I remember I did this right. Apparently,
114 * it was magically lost, so this code needs audit */
115 xdst->u.rt6.rt6i_flags = rt->rt6i_flags & (RTF_ANYCAST |
117 xdst->u.rt6.rt6i_metric = rt->rt6i_metric;
118 xdst->u.rt6.rt6i_node = rt->rt6i_node;
120 xdst->route_cookie = rt->rt6i_node->fn_sernum;
121 xdst->u.rt6.rt6i_gateway = rt->rt6i_gateway;
122 xdst->u.rt6.rt6i_dst = rt->rt6i_dst;
123 xdst->u.rt6.rt6i_src = rt->rt6i_src;
129 _decode_session6(struct sk_buff *skb, struct flowi *fl, int reverse)
131 struct flowi6 *fl6 = &fl->u.ip6;
133 u16 offset = skb_network_header_len(skb);
134 const struct ipv6hdr *hdr = ipv6_hdr(skb);
135 struct ipv6_opt_hdr *exthdr;
136 const unsigned char *nh = skb_network_header(skb);
137 u8 nexthdr = nh[IP6CB(skb)->nhoff];
141 oif = skb_dst(skb)->dev->ifindex;
143 memset(fl6, 0, sizeof(struct flowi6));
144 fl6->flowi6_mark = skb->mark;
145 fl6->flowi6_oif = reverse ? skb->skb_iif : oif;
147 fl6->daddr = reverse ? hdr->saddr : hdr->daddr;
148 fl6->saddr = reverse ? hdr->daddr : hdr->saddr;
150 while (nh + offset + 1 < skb->data ||
151 pskb_may_pull(skb, nh + offset + 1 - skb->data)) {
152 nh = skb_network_header(skb);
153 exthdr = (struct ipv6_opt_hdr *)(nh + offset);
156 case NEXTHDR_FRAGMENT:
158 case NEXTHDR_ROUTING:
161 offset += ipv6_optlen(exthdr);
162 nexthdr = exthdr->nexthdr;
163 exthdr = (struct ipv6_opt_hdr *)(nh + offset);
167 case IPPROTO_UDPLITE:
171 if (!onlyproto && (nh + offset + 4 < skb->data ||
172 pskb_may_pull(skb, nh + offset + 4 - skb->data))) {
173 __be16 *ports = (__be16 *)exthdr;
175 fl6->fl6_sport = ports[!!reverse];
176 fl6->fl6_dport = ports[!reverse];
178 fl6->flowi6_proto = nexthdr;
182 if (!onlyproto && pskb_may_pull(skb, nh + offset + 2 - skb->data)) {
183 u8 *icmp = (u8 *)exthdr;
185 fl6->fl6_icmp_type = icmp[0];
186 fl6->fl6_icmp_code = icmp[1];
188 fl6->flowi6_proto = nexthdr;
191 #if IS_ENABLED(CONFIG_IPV6_MIP6)
193 if (!onlyproto && pskb_may_pull(skb, nh + offset + 3 - skb->data)) {
195 mh = (struct ip6_mh *)exthdr;
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 if (rt6_has_peer(&xdst->u.rt6)) {
249 struct inet_peer *peer = rt6_peer_ptr(&xdst->u.rt6);
252 xfrm_dst_destroy(xdst);
255 static void xfrm6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
258 struct xfrm_dst *xdst;
263 xdst = (struct xfrm_dst *)dst;
264 if (xdst->u.rt6.rt6i_idev->dev == dev) {
265 struct inet6_dev *loopback_idev =
266 in6_dev_get(dev_net(dev)->loopback_dev);
267 BUG_ON(!loopback_idev);
270 in6_dev_put(xdst->u.rt6.rt6i_idev);
271 xdst->u.rt6.rt6i_idev = loopback_idev;
272 in6_dev_hold(loopback_idev);
273 xdst = (struct xfrm_dst *)xdst->u.dst.child;
274 } while (xdst->u.dst.xfrm);
276 __in6_dev_put(loopback_idev);
279 xfrm_dst_ifdown(dst, dev);
282 static struct dst_ops xfrm6_dst_ops = {
284 .protocol = cpu_to_be16(ETH_P_IPV6),
285 .gc = xfrm6_garbage_collect,
286 .update_pmtu = xfrm6_update_pmtu,
287 .redirect = xfrm6_redirect,
288 .cow_metrics = dst_cow_metrics_generic,
289 .destroy = xfrm6_dst_destroy,
290 .ifdown = xfrm6_dst_ifdown,
291 .local_out = __ip6_local_out,
295 static struct xfrm_policy_afinfo xfrm6_policy_afinfo = {
297 .dst_ops = &xfrm6_dst_ops,
298 .dst_lookup = xfrm6_dst_lookup,
299 .get_saddr = xfrm6_get_saddr,
300 .decode_session = _decode_session6,
301 .get_tos = xfrm6_get_tos,
302 .init_dst = xfrm6_init_dst,
303 .init_path = xfrm6_init_path,
304 .fill_dst = xfrm6_fill_dst,
305 .blackhole_route = ip6_blackhole_route,
308 static int __init xfrm6_policy_init(void)
310 return xfrm_policy_register_afinfo(&xfrm6_policy_afinfo);
313 static void xfrm6_policy_fini(void)
315 xfrm_policy_unregister_afinfo(&xfrm6_policy_afinfo);
319 static struct ctl_table xfrm6_policy_table[] = {
321 .procname = "xfrm6_gc_thresh",
322 .data = &init_net.xfrm.xfrm6_dst_ops.gc_thresh,
323 .maxlen = sizeof(int),
325 .proc_handler = proc_dointvec,
330 static int __net_init xfrm6_net_init(struct net *net)
332 struct ctl_table *table;
333 struct ctl_table_header *hdr;
335 table = xfrm6_policy_table;
336 if (!net_eq(net, &init_net)) {
337 table = kmemdup(table, sizeof(xfrm6_policy_table), GFP_KERNEL);
341 table[0].data = &net->xfrm.xfrm6_dst_ops.gc_thresh;
344 hdr = register_net_sysctl(net, "net/ipv6", table);
348 net->ipv6.sysctl.xfrm6_hdr = hdr;
352 if (!net_eq(net, &init_net))
358 static void __net_exit xfrm6_net_exit(struct net *net)
360 struct ctl_table *table;
362 if (net->ipv6.sysctl.xfrm6_hdr == NULL)
365 table = net->ipv6.sysctl.xfrm6_hdr->ctl_table_arg;
366 unregister_net_sysctl_table(net->ipv6.sysctl.xfrm6_hdr);
367 if (!net_eq(net, &init_net))
371 static struct pernet_operations xfrm6_net_ops = {
372 .init = xfrm6_net_init,
373 .exit = xfrm6_net_exit,
377 int __init xfrm6_init(void)
381 dst_entries_init(&xfrm6_dst_ops);
383 ret = xfrm6_policy_init();
385 dst_entries_destroy(&xfrm6_dst_ops);
388 ret = xfrm6_state_init();
392 ret = xfrm6_protocol_init();
397 register_pernet_subsys(&xfrm6_net_ops);
408 void xfrm6_fini(void)
411 unregister_pernet_subsys(&xfrm6_net_ops);
413 xfrm6_protocol_fini();
416 dst_entries_destroy(&xfrm6_dst_ops);