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 #include <net/l3mdev.h>
24 #if IS_ENABLED(CONFIG_IPV6_MIP6)
28 static struct xfrm_policy_afinfo xfrm6_policy_afinfo;
30 static struct dst_entry *xfrm6_dst_lookup(struct net *net, int tos, int oif,
31 const xfrm_address_t *saddr,
32 const xfrm_address_t *daddr,
36 struct dst_entry *dst;
39 memset(&fl6, 0, sizeof(fl6));
40 fl6.flowi6_oif = l3mdev_master_ifindex_by_index(net, oif);
41 fl6.flowi6_flags = FLOWI_FLAG_SKIP_NH_OIF;
42 fl6.flowi6_mark = mark;
43 memcpy(&fl6.daddr, daddr, sizeof(fl6.daddr));
45 memcpy(&fl6.saddr, saddr, sizeof(fl6.saddr));
47 dst = ip6_route_output(net, NULL, &fl6);
58 static int xfrm6_get_saddr(struct net *net, int oif,
59 xfrm_address_t *saddr, xfrm_address_t *daddr,
62 struct dst_entry *dst;
63 struct net_device *dev;
65 dst = xfrm6_dst_lookup(net, 0, oif, NULL, daddr, mark);
69 dev = ip6_dst_idev(dst)->dev;
70 ipv6_dev_get_saddr(dev_net(dev), dev, &daddr->in6, 0, &saddr->in6);
75 static int xfrm6_get_tos(const struct flowi *fl)
80 static int xfrm6_init_path(struct xfrm_dst *path, struct dst_entry *dst,
83 if (dst->ops->family == AF_INET6) {
84 struct rt6_info *rt = (struct rt6_info *)dst;
85 path->path_cookie = rt6_get_cookie(rt);
88 path->u.rt6.rt6i_nfheader_len = nfheader_len;
93 static int xfrm6_fill_dst(struct xfrm_dst *xdst, struct net_device *dev,
94 const struct flowi *fl)
96 struct rt6_info *rt = (struct rt6_info *)xdst->route;
98 xdst->u.dst.dev = dev;
101 xdst->u.rt6.rt6i_idev = in6_dev_get(dev);
102 if (!xdst->u.rt6.rt6i_idev) {
107 /* Sheit... I remember I did this right. Apparently,
108 * it was magically lost, so this code needs audit */
109 xdst->u.rt6.rt6i_flags = rt->rt6i_flags & (RTF_ANYCAST |
111 xdst->u.rt6.rt6i_metric = rt->rt6i_metric;
112 xdst->u.rt6.rt6i_node = rt->rt6i_node;
113 xdst->route_cookie = rt6_get_cookie(rt);
114 xdst->u.rt6.rt6i_gateway = rt->rt6i_gateway;
115 xdst->u.rt6.rt6i_dst = rt->rt6i_dst;
116 xdst->u.rt6.rt6i_src = rt->rt6i_src;
122 _decode_session6(struct sk_buff *skb, struct flowi *fl, int reverse)
124 struct flowi6 *fl6 = &fl->u.ip6;
126 const struct ipv6hdr *hdr = ipv6_hdr(skb);
127 u16 offset = sizeof(*hdr);
128 struct ipv6_opt_hdr *exthdr;
129 const unsigned char *nh = skb_network_header(skb);
130 u16 nhoff = IP6CB(skb)->nhoff;
135 nhoff = offsetof(struct ipv6hdr, nexthdr);
140 oif = l3mdev_fib_oif(skb_dst(skb)->dev);
142 memset(fl6, 0, sizeof(struct flowi6));
143 fl6->flowi6_mark = skb->mark;
144 fl6->flowi6_oif = reverse ? skb->skb_iif : oif;
146 fl6->daddr = reverse ? hdr->saddr : hdr->daddr;
147 fl6->saddr = reverse ? hdr->daddr : hdr->saddr;
149 while (nh + offset + 1 < skb->data ||
150 pskb_may_pull(skb, nh + offset + 1 - skb->data)) {
151 nh = skb_network_header(skb);
152 exthdr = (struct ipv6_opt_hdr *)(nh + offset);
155 case NEXTHDR_FRAGMENT:
157 case NEXTHDR_ROUTING:
160 offset += ipv6_optlen(exthdr);
161 nexthdr = exthdr->nexthdr;
162 exthdr = (struct ipv6_opt_hdr *)(nh + offset);
166 case IPPROTO_UDPLITE:
170 if (!onlyproto && (nh + offset + 4 < skb->data ||
171 pskb_may_pull(skb, nh + offset + 4 - skb->data))) {
174 nh = skb_network_header(skb);
175 ports = (__be16 *)(nh + offset);
176 fl6->fl6_sport = ports[!!reverse];
177 fl6->fl6_dport = ports[!reverse];
179 fl6->flowi6_proto = nexthdr;
183 if (!onlyproto && (nh + offset + 2 < skb->data ||
184 pskb_may_pull(skb, nh + offset + 2 - skb->data))) {
187 nh = skb_network_header(skb);
188 icmp = (u8 *)(nh + offset);
189 fl6->fl6_icmp_type = icmp[0];
190 fl6->fl6_icmp_code = icmp[1];
192 fl6->flowi6_proto = nexthdr;
195 #if IS_ENABLED(CONFIG_IPV6_MIP6)
197 offset += ipv6_optlen(exthdr);
198 if (!onlyproto && (nh + offset + 3 < skb->data ||
199 pskb_may_pull(skb, nh + offset + 3 - skb->data))) {
202 nh = skb_network_header(skb);
203 mh = (struct ip6_mh *)(nh + offset);
204 fl6->fl6_mh_type = mh->ip6mh_type;
206 fl6->flowi6_proto = nexthdr;
210 /* XXX Why are there these headers? */
215 fl6->fl6_ipsec_spi = 0;
216 fl6->flowi6_proto = nexthdr;
222 static inline int xfrm6_garbage_collect(struct dst_ops *ops)
224 struct net *net = container_of(ops, struct net, xfrm.xfrm6_dst_ops);
226 xfrm6_policy_afinfo.garbage_collect(net);
227 return dst_entries_get_fast(ops) > ops->gc_thresh * 2;
230 static void xfrm6_update_pmtu(struct dst_entry *dst, struct sock *sk,
231 struct sk_buff *skb, u32 mtu)
233 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
234 struct dst_entry *path = xdst->route;
236 path->ops->update_pmtu(path, sk, skb, mtu);
239 static void xfrm6_redirect(struct dst_entry *dst, struct sock *sk,
242 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
243 struct dst_entry *path = xdst->route;
245 path->ops->redirect(path, sk, skb);
248 static void xfrm6_dst_destroy(struct dst_entry *dst)
250 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
252 if (likely(xdst->u.rt6.rt6i_idev))
253 in6_dev_put(xdst->u.rt6.rt6i_idev);
254 dst_destroy_metrics_generic(dst);
255 xfrm_dst_destroy(xdst);
258 static void xfrm6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
261 struct xfrm_dst *xdst;
266 xdst = (struct xfrm_dst *)dst;
267 if (xdst->u.rt6.rt6i_idev->dev == dev) {
268 struct inet6_dev *loopback_idev =
269 in6_dev_get(dev_net(dev)->loopback_dev);
270 BUG_ON(!loopback_idev);
273 in6_dev_put(xdst->u.rt6.rt6i_idev);
274 xdst->u.rt6.rt6i_idev = loopback_idev;
275 in6_dev_hold(loopback_idev);
276 xdst = (struct xfrm_dst *)xdst->u.dst.child;
277 } while (xdst->u.dst.xfrm);
279 __in6_dev_put(loopback_idev);
282 xfrm_dst_ifdown(dst, dev);
285 static struct dst_ops xfrm6_dst_ops_template = {
287 .gc = xfrm6_garbage_collect,
288 .update_pmtu = xfrm6_update_pmtu,
289 .redirect = xfrm6_redirect,
290 .cow_metrics = dst_cow_metrics_generic,
291 .destroy = xfrm6_dst_destroy,
292 .ifdown = xfrm6_dst_ifdown,
293 .local_out = __ip6_local_out,
294 .gc_thresh = INT_MAX,
297 static struct xfrm_policy_afinfo xfrm6_policy_afinfo = {
299 .dst_ops = &xfrm6_dst_ops_template,
300 .dst_lookup = xfrm6_dst_lookup,
301 .get_saddr = xfrm6_get_saddr,
302 .decode_session = _decode_session6,
303 .get_tos = xfrm6_get_tos,
304 .init_path = xfrm6_init_path,
305 .fill_dst = xfrm6_fill_dst,
306 .blackhole_route = ip6_blackhole_route,
309 static int __init xfrm6_policy_init(void)
311 return xfrm_policy_register_afinfo(&xfrm6_policy_afinfo);
314 static void xfrm6_policy_fini(void)
316 xfrm_policy_unregister_afinfo(&xfrm6_policy_afinfo);
320 static struct ctl_table xfrm6_policy_table[] = {
322 .procname = "xfrm6_gc_thresh",
323 .data = &init_net.xfrm.xfrm6_dst_ops.gc_thresh,
324 .maxlen = sizeof(int),
326 .proc_handler = proc_dointvec,
331 static int __net_init xfrm6_net_sysctl_init(struct net *net)
333 struct ctl_table *table;
334 struct ctl_table_header *hdr;
336 table = xfrm6_policy_table;
337 if (!net_eq(net, &init_net)) {
338 table = kmemdup(table, sizeof(xfrm6_policy_table), GFP_KERNEL);
342 table[0].data = &net->xfrm.xfrm6_dst_ops.gc_thresh;
345 hdr = register_net_sysctl(net, "net/ipv6", table);
349 net->ipv6.sysctl.xfrm6_hdr = hdr;
353 if (!net_eq(net, &init_net))
359 static void __net_exit xfrm6_net_sysctl_exit(struct net *net)
361 struct ctl_table *table;
363 if (!net->ipv6.sysctl.xfrm6_hdr)
366 table = net->ipv6.sysctl.xfrm6_hdr->ctl_table_arg;
367 unregister_net_sysctl_table(net->ipv6.sysctl.xfrm6_hdr);
368 if (!net_eq(net, &init_net))
371 #else /* CONFIG_SYSCTL */
372 static int inline xfrm6_net_sysctl_init(struct net *net)
377 static void inline xfrm6_net_sysctl_exit(struct net *net)
382 static int __net_init xfrm6_net_init(struct net *net)
386 memcpy(&net->xfrm.xfrm6_dst_ops, &xfrm6_dst_ops_template,
387 sizeof(xfrm6_dst_ops_template));
388 ret = dst_entries_init(&net->xfrm.xfrm6_dst_ops);
392 ret = xfrm6_net_sysctl_init(net);
394 dst_entries_destroy(&net->xfrm.xfrm6_dst_ops);
399 static void __net_exit xfrm6_net_exit(struct net *net)
401 xfrm6_net_sysctl_exit(net);
402 dst_entries_destroy(&net->xfrm.xfrm6_dst_ops);
405 static struct pernet_operations xfrm6_net_ops = {
406 .init = xfrm6_net_init,
407 .exit = xfrm6_net_exit,
410 int __init xfrm6_init(void)
414 ret = xfrm6_policy_init();
417 ret = xfrm6_state_init();
421 ret = xfrm6_protocol_init();
425 register_pernet_subsys(&xfrm6_net_ops);
435 void xfrm6_fini(void)
437 unregister_pernet_subsys(&xfrm6_net_ops);
438 xfrm6_protocol_fini();