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>
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_table_id = rt->rt_table_id;
101 INIT_LIST_HEAD(&xdst->u.rt.rt_uncached);
107 _decode_session4(struct sk_buff *skb, struct flowi *fl, int reverse)
109 const struct iphdr *iph = ip_hdr(skb);
110 u8 *xprth = skb_network_header(skb) + iph->ihl * 4;
111 struct flowi4 *fl4 = &fl->u.ip4;
115 oif = vrf_master_ifindex(skb_dst(skb)->dev) ?
116 : skb_dst(skb)->dev->ifindex;
119 memset(fl4, 0, sizeof(struct flowi4));
120 fl4->flowi4_mark = skb->mark;
121 fl4->flowi4_oif = reverse ? skb->skb_iif : oif;
123 if (!ip_is_fragment(iph)) {
124 switch (iph->protocol) {
126 case IPPROTO_UDPLITE:
130 if (xprth + 4 < skb->data ||
131 pskb_may_pull(skb, xprth + 4 - skb->data)) {
132 __be16 *ports = (__be16 *)xprth;
134 fl4->fl4_sport = ports[!!reverse];
135 fl4->fl4_dport = ports[!reverse];
140 if (xprth + 2 < skb->data ||
141 pskb_may_pull(skb, xprth + 2 - skb->data)) {
144 fl4->fl4_icmp_type = icmp[0];
145 fl4->fl4_icmp_code = icmp[1];
150 if (xprth + 4 < skb->data ||
151 pskb_may_pull(skb, xprth + 4 - skb->data)) {
152 __be32 *ehdr = (__be32 *)xprth;
154 fl4->fl4_ipsec_spi = ehdr[0];
159 if (xprth + 8 < skb->data ||
160 pskb_may_pull(skb, xprth + 8 - skb->data)) {
161 __be32 *ah_hdr = (__be32 *)xprth;
163 fl4->fl4_ipsec_spi = ah_hdr[1];
168 if (xprth + 4 < skb->data ||
169 pskb_may_pull(skb, xprth + 4 - skb->data)) {
170 __be16 *ipcomp_hdr = (__be16 *)xprth;
172 fl4->fl4_ipsec_spi = htonl(ntohs(ipcomp_hdr[1]));
177 if (xprth + 12 < skb->data ||
178 pskb_may_pull(skb, xprth + 12 - skb->data)) {
179 __be16 *greflags = (__be16 *)xprth;
180 __be32 *gre_hdr = (__be32 *)xprth;
182 if (greflags[0] & GRE_KEY) {
183 if (greflags[0] & GRE_CSUM)
185 fl4->fl4_gre_key = gre_hdr[1];
191 fl4->fl4_ipsec_spi = 0;
195 fl4->flowi4_proto = iph->protocol;
196 fl4->daddr = reverse ? iph->saddr : iph->daddr;
197 fl4->saddr = reverse ? iph->daddr : iph->saddr;
198 fl4->flowi4_tos = iph->tos;
201 static inline int xfrm4_garbage_collect(struct dst_ops *ops)
203 struct net *net = container_of(ops, struct net, xfrm.xfrm4_dst_ops);
205 xfrm4_policy_afinfo.garbage_collect(net);
206 return (dst_entries_get_slow(ops) > ops->gc_thresh * 2);
209 static void xfrm4_update_pmtu(struct dst_entry *dst, struct sock *sk,
210 struct sk_buff *skb, u32 mtu)
212 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
213 struct dst_entry *path = xdst->route;
215 path->ops->update_pmtu(path, sk, skb, mtu);
218 static void xfrm4_redirect(struct dst_entry *dst, struct sock *sk,
221 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
222 struct dst_entry *path = xdst->route;
224 path->ops->redirect(path, sk, skb);
227 static void xfrm4_dst_destroy(struct dst_entry *dst)
229 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
231 dst_destroy_metrics_generic(dst);
233 xfrm_dst_destroy(xdst);
236 static void xfrm4_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
242 xfrm_dst_ifdown(dst, dev);
245 static struct dst_ops xfrm4_dst_ops = {
247 .gc = xfrm4_garbage_collect,
248 .update_pmtu = xfrm4_update_pmtu,
249 .redirect = xfrm4_redirect,
250 .cow_metrics = dst_cow_metrics_generic,
251 .destroy = xfrm4_dst_destroy,
252 .ifdown = xfrm4_dst_ifdown,
253 .local_out = __ip_local_out,
254 .gc_thresh = INT_MAX,
257 static struct xfrm_policy_afinfo xfrm4_policy_afinfo = {
259 .dst_ops = &xfrm4_dst_ops,
260 .dst_lookup = xfrm4_dst_lookup,
261 .get_saddr = xfrm4_get_saddr,
262 .decode_session = _decode_session4,
263 .get_tos = xfrm4_get_tos,
264 .init_path = xfrm4_init_path,
265 .fill_dst = xfrm4_fill_dst,
266 .blackhole_route = ipv4_blackhole_route,
270 static struct ctl_table xfrm4_policy_table[] = {
272 .procname = "xfrm4_gc_thresh",
273 .data = &init_net.xfrm.xfrm4_dst_ops.gc_thresh,
274 .maxlen = sizeof(int),
276 .proc_handler = proc_dointvec,
281 static int __net_init xfrm4_net_init(struct net *net)
283 struct ctl_table *table;
284 struct ctl_table_header *hdr;
286 table = xfrm4_policy_table;
287 if (!net_eq(net, &init_net)) {
288 table = kmemdup(table, sizeof(xfrm4_policy_table), GFP_KERNEL);
292 table[0].data = &net->xfrm.xfrm4_dst_ops.gc_thresh;
295 hdr = register_net_sysctl(net, "net/ipv4", table);
299 net->ipv4.xfrm4_hdr = hdr;
303 if (!net_eq(net, &init_net))
309 static void __net_exit xfrm4_net_exit(struct net *net)
311 struct ctl_table *table;
313 if (!net->ipv4.xfrm4_hdr)
316 table = net->ipv4.xfrm4_hdr->ctl_table_arg;
317 unregister_net_sysctl_table(net->ipv4.xfrm4_hdr);
318 if (!net_eq(net, &init_net))
322 static struct pernet_operations __net_initdata xfrm4_net_ops = {
323 .init = xfrm4_net_init,
324 .exit = xfrm4_net_exit,
328 static void __init xfrm4_policy_init(void)
330 xfrm_policy_register_afinfo(&xfrm4_policy_afinfo);
333 void __init xfrm4_init(void)
335 dst_entries_init(&xfrm4_dst_ops);
339 xfrm4_protocol_init();
341 register_pernet_subsys(&xfrm4_net_ops);