5 * Kazunori MIYAZAWA @USAGI
6 * YOSHIFUJI Hideaki @USAGI
7 * Split up af-specific portion
11 #include <linux/err.h>
12 #include <linux/kernel.h>
13 #include <linux/inetdevice.h>
14 #include <linux/if_tunnel.h>
18 #include <net/l3mdev.h>
20 static struct xfrm_policy_afinfo xfrm4_policy_afinfo;
22 static struct dst_entry *__xfrm4_dst_lookup(struct net *net, struct flowi4 *fl4,
24 const xfrm_address_t *saddr,
25 const xfrm_address_t *daddr)
29 memset(fl4, 0, sizeof(*fl4));
30 fl4->daddr = daddr->a4;
31 fl4->flowi4_tos = tos;
32 fl4->flowi4_oif = oif;
34 fl4->saddr = saddr->a4;
36 fl4->flowi4_flags = FLOWI_FLAG_SKIP_NH_OIF;
38 rt = __ip_route_output_key(net, fl4);
45 static struct dst_entry *xfrm4_dst_lookup(struct net *net, int tos, int oif,
46 const xfrm_address_t *saddr,
47 const xfrm_address_t *daddr)
51 return __xfrm4_dst_lookup(net, &fl4, tos, oif, saddr, daddr);
54 static int xfrm4_get_saddr(struct net *net, int oif,
55 xfrm_address_t *saddr, xfrm_address_t *daddr)
57 struct dst_entry *dst;
60 dst = __xfrm4_dst_lookup(net, &fl4, 0, oif, NULL, daddr);
64 saddr->a4 = fl4.saddr;
69 static int xfrm4_get_tos(const struct flowi *fl)
71 return IPTOS_RT_MASK & fl->u.ip4.flowi4_tos; /* Strip ECN bits */
74 static int xfrm4_init_path(struct xfrm_dst *path, struct dst_entry *dst,
80 static int xfrm4_fill_dst(struct xfrm_dst *xdst, struct net_device *dev,
81 const struct flowi *fl)
83 struct rtable *rt = (struct rtable *)xdst->route;
84 const struct flowi4 *fl4 = &fl->u.ip4;
86 xdst->u.rt.rt_iif = fl4->flowi4_iif;
88 xdst->u.dst.dev = dev;
91 /* Sheit... I remember I did this right. Apparently,
92 * it was magically lost, so this code needs audit */
93 xdst->u.rt.rt_is_input = rt->rt_is_input;
94 xdst->u.rt.rt_flags = rt->rt_flags & (RTCF_BROADCAST | RTCF_MULTICAST |
96 xdst->u.rt.rt_type = rt->rt_type;
97 xdst->u.rt.rt_gateway = rt->rt_gateway;
98 xdst->u.rt.rt_uses_gateway = rt->rt_uses_gateway;
99 xdst->u.rt.rt_pmtu = rt->rt_pmtu;
100 xdst->u.rt.rt_mtu_locked = rt->rt_mtu_locked;
101 xdst->u.rt.rt_table_id = rt->rt_table_id;
102 INIT_LIST_HEAD(&xdst->u.rt.rt_uncached);
108 _decode_session4(struct sk_buff *skb, struct flowi *fl, int reverse)
110 const struct iphdr *iph = ip_hdr(skb);
112 u8 *xprth = skb_network_header(skb) + ihl * 4;
113 struct flowi4 *fl4 = &fl->u.ip4;
117 oif = l3mdev_fib_oif(skb_dst(skb)->dev);
119 memset(fl4, 0, sizeof(struct flowi4));
120 fl4->flowi4_mark = skb->mark;
121 fl4->flowi4_oif = reverse ? skb->skb_iif : oif;
123 fl4->flowi4_proto = iph->protocol;
124 fl4->daddr = reverse ? iph->saddr : iph->daddr;
125 fl4->saddr = reverse ? iph->daddr : iph->saddr;
126 fl4->flowi4_tos = iph->tos;
128 if (!ip_is_fragment(iph)) {
129 switch (iph->protocol) {
131 case IPPROTO_UDPLITE:
135 if (xprth + 4 < skb->data ||
136 pskb_may_pull(skb, xprth + 4 - skb->data)) {
139 xprth = skb_network_header(skb) + ihl * 4;
140 ports = (__be16 *)xprth;
142 fl4->fl4_sport = ports[!!reverse];
143 fl4->fl4_dport = ports[!reverse];
148 if (xprth + 2 < skb->data ||
149 pskb_may_pull(skb, xprth + 2 - skb->data)) {
152 xprth = skb_network_header(skb) + ihl * 4;
155 fl4->fl4_icmp_type = icmp[0];
156 fl4->fl4_icmp_code = icmp[1];
161 if (xprth + 4 < skb->data ||
162 pskb_may_pull(skb, xprth + 4 - skb->data)) {
165 xprth = skb_network_header(skb) + ihl * 4;
166 ehdr = (__be32 *)xprth;
168 fl4->fl4_ipsec_spi = ehdr[0];
173 if (xprth + 8 < skb->data ||
174 pskb_may_pull(skb, xprth + 8 - skb->data)) {
177 xprth = skb_network_header(skb) + ihl * 4;
178 ah_hdr = (__be32 *)xprth;
180 fl4->fl4_ipsec_spi = ah_hdr[1];
185 if (xprth + 4 < skb->data ||
186 pskb_may_pull(skb, xprth + 4 - skb->data)) {
189 xprth = skb_network_header(skb) + ihl * 4;
190 ipcomp_hdr = (__be16 *)xprth;
192 fl4->fl4_ipsec_spi = htonl(ntohs(ipcomp_hdr[1]));
197 if (xprth + 12 < skb->data ||
198 pskb_may_pull(skb, xprth + 12 - skb->data)) {
202 xprth = skb_network_header(skb) + ihl * 4;
203 greflags = (__be16 *)xprth;
204 gre_hdr = (__be32 *)xprth;
206 if (greflags[0] & GRE_KEY) {
207 if (greflags[0] & GRE_CSUM)
209 fl4->fl4_gre_key = gre_hdr[1];
215 fl4->fl4_ipsec_spi = 0;
221 static inline int xfrm4_garbage_collect(struct dst_ops *ops)
223 struct net *net = container_of(ops, struct net, xfrm.xfrm4_dst_ops);
225 xfrm4_policy_afinfo.garbage_collect(net);
226 return (dst_entries_get_slow(ops) > ops->gc_thresh * 2);
229 static void xfrm4_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 xfrm4_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 xfrm4_dst_destroy(struct dst_entry *dst)
249 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
251 dst_destroy_metrics_generic(dst);
253 xfrm_dst_destroy(xdst);
256 static void xfrm4_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
262 xfrm_dst_ifdown(dst, dev);
265 static struct dst_ops xfrm4_dst_ops_template = {
267 .gc = xfrm4_garbage_collect,
268 .update_pmtu = xfrm4_update_pmtu,
269 .redirect = xfrm4_redirect,
270 .cow_metrics = dst_cow_metrics_generic,
271 .destroy = xfrm4_dst_destroy,
272 .ifdown = xfrm4_dst_ifdown,
273 .local_out = __ip_local_out,
274 .gc_thresh = INT_MAX,
277 static struct xfrm_policy_afinfo xfrm4_policy_afinfo = {
279 .dst_ops = &xfrm4_dst_ops_template,
280 .dst_lookup = xfrm4_dst_lookup,
281 .get_saddr = xfrm4_get_saddr,
282 .decode_session = _decode_session4,
283 .get_tos = xfrm4_get_tos,
284 .init_path = xfrm4_init_path,
285 .fill_dst = xfrm4_fill_dst,
286 .blackhole_route = ipv4_blackhole_route,
290 static struct ctl_table xfrm4_policy_table[] = {
292 .procname = "xfrm4_gc_thresh",
293 .data = &init_net.xfrm.xfrm4_dst_ops.gc_thresh,
294 .maxlen = sizeof(int),
296 .proc_handler = proc_dointvec,
301 static int __net_init xfrm4_net_sysctl_init(struct net *net)
303 struct ctl_table *table;
304 struct ctl_table_header *hdr;
306 table = xfrm4_policy_table;
307 if (!net_eq(net, &init_net)) {
308 table = kmemdup(table, sizeof(xfrm4_policy_table), GFP_KERNEL);
312 table[0].data = &net->xfrm.xfrm4_dst_ops.gc_thresh;
315 hdr = register_net_sysctl(net, "net/ipv4", table);
319 net->ipv4.xfrm4_hdr = hdr;
323 if (!net_eq(net, &init_net))
329 static void __net_exit xfrm4_net_sysctl_exit(struct net *net)
331 struct ctl_table *table;
333 if (!net->ipv4.xfrm4_hdr)
336 table = net->ipv4.xfrm4_hdr->ctl_table_arg;
337 unregister_net_sysctl_table(net->ipv4.xfrm4_hdr);
338 if (!net_eq(net, &init_net))
341 #else /* CONFIG_SYSCTL */
342 static int inline xfrm4_net_sysctl_init(struct net *net)
347 static void inline xfrm4_net_sysctl_exit(struct net *net)
352 static int __net_init xfrm4_net_init(struct net *net)
356 memcpy(&net->xfrm.xfrm4_dst_ops, &xfrm4_dst_ops_template,
357 sizeof(xfrm4_dst_ops_template));
358 ret = dst_entries_init(&net->xfrm.xfrm4_dst_ops);
362 ret = xfrm4_net_sysctl_init(net);
364 dst_entries_destroy(&net->xfrm.xfrm4_dst_ops);
369 static void __net_exit xfrm4_net_exit(struct net *net)
371 xfrm4_net_sysctl_exit(net);
372 dst_entries_destroy(&net->xfrm.xfrm4_dst_ops);
375 static struct pernet_operations __net_initdata xfrm4_net_ops = {
376 .init = xfrm4_net_init,
377 .exit = xfrm4_net_exit,
380 static void __init xfrm4_policy_init(void)
382 xfrm_policy_register_afinfo(&xfrm4_policy_afinfo);
385 void __init xfrm4_init(void)
389 xfrm4_protocol_init();
390 register_pernet_subsys(&xfrm4_net_ops);