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>
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)
35 struct dst_entry *dst;
38 memset(&fl6, 0, sizeof(fl6));
40 fl6.flowi6_flags = FLOWI_FLAG_SKIP_NH_OIF;
41 memcpy(&fl6.daddr, daddr, sizeof(fl6.daddr));
43 memcpy(&fl6.saddr, saddr, sizeof(fl6.saddr));
45 dst = ip6_route_output(net, NULL, &fl6);
56 static int xfrm6_get_saddr(struct net *net, int oif,
57 xfrm_address_t *saddr, xfrm_address_t *daddr)
59 struct dst_entry *dst;
60 struct net_device *dev;
62 dst = xfrm6_dst_lookup(net, 0, oif, NULL, daddr);
66 dev = ip6_dst_idev(dst)->dev;
67 ipv6_dev_get_saddr(dev_net(dev), dev, &daddr->in6, 0, &saddr->in6);
72 static int xfrm6_get_tos(const struct flowi *fl)
77 static int xfrm6_init_path(struct xfrm_dst *path, struct dst_entry *dst,
80 if (dst->ops->family == AF_INET6) {
81 struct rt6_info *rt = (struct rt6_info *)dst;
82 path->path_cookie = rt6_get_cookie(rt);
85 path->u.rt6.rt6i_nfheader_len = nfheader_len;
90 static int xfrm6_fill_dst(struct xfrm_dst *xdst, struct net_device *dev,
91 const struct flowi *fl)
93 struct rt6_info *rt = (struct rt6_info *)xdst->route;
95 xdst->u.dst.dev = dev;
98 xdst->u.rt6.rt6i_idev = in6_dev_get(dev);
99 if (!xdst->u.rt6.rt6i_idev) {
104 /* Sheit... I remember I did this right. Apparently,
105 * it was magically lost, so this code needs audit */
106 xdst->u.rt6.rt6i_flags = rt->rt6i_flags & (RTF_ANYCAST |
108 xdst->u.rt6.rt6i_metric = rt->rt6i_metric;
109 xdst->u.rt6.rt6i_node = rt->rt6i_node;
110 xdst->route_cookie = rt6_get_cookie(rt);
111 xdst->u.rt6.rt6i_gateway = rt->rt6i_gateway;
112 xdst->u.rt6.rt6i_dst = rt->rt6i_dst;
113 xdst->u.rt6.rt6i_src = rt->rt6i_src;
119 _decode_session6(struct sk_buff *skb, struct flowi *fl, int reverse)
121 struct flowi6 *fl6 = &fl->u.ip6;
123 const struct ipv6hdr *hdr = ipv6_hdr(skb);
124 u16 offset = sizeof(*hdr);
125 struct ipv6_opt_hdr *exthdr;
126 const unsigned char *nh = skb_network_header(skb);
127 u16 nhoff = IP6CB(skb)->nhoff;
132 nhoff = offsetof(struct ipv6hdr, nexthdr);
137 oif = vrf_master_ifindex(skb_dst(skb)->dev) ?
138 : skb_dst(skb)->dev->ifindex;
141 memset(fl6, 0, sizeof(struct flowi6));
142 fl6->flowi6_mark = skb->mark;
143 fl6->flowi6_oif = reverse ? skb->skb_iif : oif;
145 fl6->daddr = reverse ? hdr->saddr : hdr->daddr;
146 fl6->saddr = reverse ? hdr->daddr : hdr->saddr;
148 while (nh + offset + 1 < skb->data ||
149 pskb_may_pull(skb, nh + offset + 1 - skb->data)) {
150 nh = skb_network_header(skb);
151 exthdr = (struct ipv6_opt_hdr *)(nh + offset);
154 case NEXTHDR_FRAGMENT:
156 case NEXTHDR_ROUTING:
159 offset += ipv6_optlen(exthdr);
160 nexthdr = exthdr->nexthdr;
161 exthdr = (struct ipv6_opt_hdr *)(nh + offset);
165 case IPPROTO_UDPLITE:
169 if (!onlyproto && (nh + offset + 4 < skb->data ||
170 pskb_may_pull(skb, nh + offset + 4 - skb->data))) {
173 nh = skb_network_header(skb);
174 ports = (__be16 *)(nh + offset);
175 fl6->fl6_sport = ports[!!reverse];
176 fl6->fl6_dport = ports[!reverse];
178 fl6->flowi6_proto = nexthdr;
182 if (!onlyproto && (nh + offset + 2 < skb->data ||
183 pskb_may_pull(skb, nh + offset + 2 - skb->data))) {
186 nh = skb_network_header(skb);
187 icmp = (u8 *)(nh + offset);
188 fl6->fl6_icmp_type = icmp[0];
189 fl6->fl6_icmp_code = icmp[1];
191 fl6->flowi6_proto = nexthdr;
194 #if IS_ENABLED(CONFIG_IPV6_MIP6)
196 offset += ipv6_optlen(exthdr);
197 if (!onlyproto && (nh + offset + 3 < skb->data ||
198 pskb_may_pull(skb, nh + offset + 3 - skb->data))) {
201 nh = skb_network_header(skb);
202 mh = (struct ip6_mh *)(nh + offset);
203 fl6->fl6_mh_type = mh->ip6mh_type;
205 fl6->flowi6_proto = nexthdr;
209 /* XXX Why are there these headers? */
214 fl6->fl6_ipsec_spi = 0;
215 fl6->flowi6_proto = nexthdr;
221 static inline int xfrm6_garbage_collect(struct dst_ops *ops)
223 struct net *net = container_of(ops, struct net, xfrm.xfrm6_dst_ops);
225 xfrm6_policy_afinfo.garbage_collect(net);
226 return dst_entries_get_fast(ops) > ops->gc_thresh * 2;
229 static void xfrm6_update_pmtu(struct dst_entry *dst, struct sock *sk,
230 struct sk_buff *skb, u32 mtu)
232 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
233 struct dst_entry *path = xdst->route;
235 path->ops->update_pmtu(path, sk, skb, mtu);
238 static void xfrm6_redirect(struct dst_entry *dst, struct sock *sk,
241 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
242 struct dst_entry *path = xdst->route;
244 path->ops->redirect(path, sk, skb);
247 static void xfrm6_dst_destroy(struct dst_entry *dst)
249 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
251 if (likely(xdst->u.rt6.rt6i_idev))
252 in6_dev_put(xdst->u.rt6.rt6i_idev);
253 dst_destroy_metrics_generic(dst);
254 xfrm_dst_destroy(xdst);
257 static void xfrm6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
260 struct xfrm_dst *xdst;
265 xdst = (struct xfrm_dst *)dst;
266 if (xdst->u.rt6.rt6i_idev->dev == dev) {
267 struct inet6_dev *loopback_idev =
268 in6_dev_get(dev_net(dev)->loopback_dev);
269 BUG_ON(!loopback_idev);
272 in6_dev_put(xdst->u.rt6.rt6i_idev);
273 xdst->u.rt6.rt6i_idev = loopback_idev;
274 in6_dev_hold(loopback_idev);
275 xdst = (struct xfrm_dst *)xdst->u.dst.child;
276 } while (xdst->u.dst.xfrm);
278 __in6_dev_put(loopback_idev);
281 xfrm_dst_ifdown(dst, dev);
284 static struct dst_ops xfrm6_dst_ops = {
286 .gc = xfrm6_garbage_collect,
287 .update_pmtu = xfrm6_update_pmtu,
288 .redirect = xfrm6_redirect,
289 .cow_metrics = dst_cow_metrics_generic,
290 .destroy = xfrm6_dst_destroy,
291 .ifdown = xfrm6_dst_ifdown,
292 .local_out = __ip6_local_out,
296 static struct xfrm_policy_afinfo xfrm6_policy_afinfo = {
298 .dst_ops = &xfrm6_dst_ops,
299 .dst_lookup = xfrm6_dst_lookup,
300 .get_saddr = xfrm6_get_saddr,
301 .decode_session = _decode_session6,
302 .get_tos = xfrm6_get_tos,
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)
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);