1
2
3
4
5
6
7
8
9
10
11
12#include <linux/err.h>
13#include <linux/kernel.h>
14#include <linux/inetdevice.h>
15#include <linux/if_tunnel.h>
16#include <net/dst.h>
17#include <net/xfrm.h>
18#include <net/ip.h>
19#include <net/l3mdev.h>
20
21static struct dst_entry *__xfrm4_dst_lookup(struct net *net, struct flowi4 *fl4,
22 int tos, int oif,
23 const xfrm_address_t *saddr,
24 const xfrm_address_t *daddr,
25 u32 mark)
26{
27 struct rtable *rt;
28
29 memset(fl4, 0, sizeof(*fl4));
30 fl4->daddr = daddr->a4;
31 fl4->flowi4_tos = tos;
32 fl4->flowi4_oif = l3mdev_master_ifindex_by_index(net, oif);
33 fl4->flowi4_mark = mark;
34 if (saddr)
35 fl4->saddr = saddr->a4;
36
37 fl4->flowi4_flags = FLOWI_FLAG_SKIP_NH_OIF;
38
39 rt = __ip_route_output_key(net, fl4);
40 if (!IS_ERR(rt))
41 return &rt->dst;
42
43 return ERR_CAST(rt);
44}
45
46static struct dst_entry *xfrm4_dst_lookup(struct net *net, int tos, int oif,
47 const xfrm_address_t *saddr,
48 const xfrm_address_t *daddr,
49 u32 mark)
50{
51 struct flowi4 fl4;
52
53 return __xfrm4_dst_lookup(net, &fl4, tos, oif, saddr, daddr, mark);
54}
55
56static int xfrm4_get_saddr(struct net *net, int oif,
57 xfrm_address_t *saddr, xfrm_address_t *daddr,
58 u32 mark)
59{
60 struct dst_entry *dst;
61 struct flowi4 fl4;
62
63 dst = __xfrm4_dst_lookup(net, &fl4, 0, oif, NULL, daddr, mark);
64 if (IS_ERR(dst))
65 return -EHOSTUNREACH;
66
67 saddr->a4 = fl4.saddr;
68 dst_release(dst);
69 return 0;
70}
71
72static int xfrm4_get_tos(const struct flowi *fl)
73{
74 return IPTOS_RT_MASK & fl->u.ip4.flowi4_tos;
75}
76
77static int xfrm4_init_path(struct xfrm_dst *path, struct dst_entry *dst,
78 int nfheader_len)
79{
80 return 0;
81}
82
83static int xfrm4_fill_dst(struct xfrm_dst *xdst, struct net_device *dev,
84 const struct flowi *fl)
85{
86 struct rtable *rt = (struct rtable *)xdst->route;
87 const struct flowi4 *fl4 = &fl->u.ip4;
88
89 xdst->u.rt.rt_iif = fl4->flowi4_iif;
90
91 xdst->u.dst.dev = dev;
92 dev_hold(dev);
93
94
95
96 xdst->u.rt.rt_is_input = rt->rt_is_input;
97 xdst->u.rt.rt_flags = rt->rt_flags & (RTCF_BROADCAST | RTCF_MULTICAST |
98 RTCF_LOCAL);
99 xdst->u.rt.rt_type = rt->rt_type;
100 xdst->u.rt.rt_gateway = rt->rt_gateway;
101 xdst->u.rt.rt_uses_gateway = rt->rt_uses_gateway;
102 xdst->u.rt.rt_pmtu = rt->rt_pmtu;
103 xdst->u.rt.rt_mtu_locked = rt->rt_mtu_locked;
104 INIT_LIST_HEAD(&xdst->u.rt.rt_uncached);
105 rt_add_uncached_list(&xdst->u.rt);
106
107 return 0;
108}
109
110static void
111_decode_session4(struct sk_buff *skb, struct flowi *fl, int reverse)
112{
113 const struct iphdr *iph = ip_hdr(skb);
114 int ihl = iph->ihl;
115 u8 *xprth = skb_network_header(skb) + ihl * 4;
116 struct flowi4 *fl4 = &fl->u.ip4;
117 int oif = 0;
118
119 if (skb_dst(skb))
120 oif = skb_dst(skb)->dev->ifindex;
121
122 memset(fl4, 0, sizeof(struct flowi4));
123 fl4->flowi4_mark = skb->mark;
124 fl4->flowi4_oif = reverse ? skb->skb_iif : oif;
125
126 fl4->flowi4_proto = iph->protocol;
127 fl4->daddr = reverse ? iph->saddr : iph->daddr;
128 fl4->saddr = reverse ? iph->daddr : iph->saddr;
129 fl4->flowi4_tos = iph->tos;
130
131 if (!ip_is_fragment(iph)) {
132 switch (iph->protocol) {
133 case IPPROTO_UDP:
134 case IPPROTO_UDPLITE:
135 case IPPROTO_TCP:
136 case IPPROTO_SCTP:
137 case IPPROTO_DCCP:
138 if (xprth + 4 < skb->data ||
139 pskb_may_pull(skb, xprth + 4 - skb->data)) {
140 __be16 *ports;
141
142 xprth = skb_network_header(skb) + ihl * 4;
143 ports = (__be16 *)xprth;
144
145 fl4->fl4_sport = ports[!!reverse];
146 fl4->fl4_dport = ports[!reverse];
147 }
148 break;
149
150 case IPPROTO_ICMP:
151 if (xprth + 2 < skb->data ||
152 pskb_may_pull(skb, xprth + 2 - skb->data)) {
153 u8 *icmp;
154
155 xprth = skb_network_header(skb) + ihl * 4;
156 icmp = xprth;
157
158 fl4->fl4_icmp_type = icmp[0];
159 fl4->fl4_icmp_code = icmp[1];
160 }
161 break;
162
163 case IPPROTO_ESP:
164 if (xprth + 4 < skb->data ||
165 pskb_may_pull(skb, xprth + 4 - skb->data)) {
166 __be32 *ehdr;
167
168 xprth = skb_network_header(skb) + ihl * 4;
169 ehdr = (__be32 *)xprth;
170
171 fl4->fl4_ipsec_spi = ehdr[0];
172 }
173 break;
174
175 case IPPROTO_AH:
176 if (xprth + 8 < skb->data ||
177 pskb_may_pull(skb, xprth + 8 - skb->data)) {
178 __be32 *ah_hdr;
179
180 xprth = skb_network_header(skb) + ihl * 4;
181 ah_hdr = (__be32 *)xprth;
182
183 fl4->fl4_ipsec_spi = ah_hdr[1];
184 }
185 break;
186
187 case IPPROTO_COMP:
188 if (xprth + 4 < skb->data ||
189 pskb_may_pull(skb, xprth + 4 - skb->data)) {
190 __be16 *ipcomp_hdr;
191
192 xprth = skb_network_header(skb) + ihl * 4;
193 ipcomp_hdr = (__be16 *)xprth;
194
195 fl4->fl4_ipsec_spi = htonl(ntohs(ipcomp_hdr[1]));
196 }
197 break;
198
199 case IPPROTO_GRE:
200 if (xprth + 12 < skb->data ||
201 pskb_may_pull(skb, xprth + 12 - skb->data)) {
202 __be16 *greflags;
203 __be32 *gre_hdr;
204
205 xprth = skb_network_header(skb) + ihl * 4;
206 greflags = (__be16 *)xprth;
207 gre_hdr = (__be32 *)xprth;
208
209 if (greflags[0] & GRE_KEY) {
210 if (greflags[0] & GRE_CSUM)
211 gre_hdr++;
212 fl4->fl4_gre_key = gre_hdr[1];
213 }
214 }
215 break;
216
217 default:
218 fl4->fl4_ipsec_spi = 0;
219 break;
220 }
221 }
222}
223
224static void xfrm4_update_pmtu(struct dst_entry *dst, struct sock *sk,
225 struct sk_buff *skb, u32 mtu)
226{
227 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
228 struct dst_entry *path = xdst->route;
229
230 path->ops->update_pmtu(path, sk, skb, mtu);
231}
232
233static void xfrm4_redirect(struct dst_entry *dst, struct sock *sk,
234 struct sk_buff *skb)
235{
236 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
237 struct dst_entry *path = xdst->route;
238
239 path->ops->redirect(path, sk, skb);
240}
241
242static void xfrm4_dst_destroy(struct dst_entry *dst)
243{
244 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
245
246 dst_destroy_metrics_generic(dst);
247 if (xdst->u.rt.rt_uncached_list)
248 rt_del_uncached_list(&xdst->u.rt);
249 xfrm_dst_destroy(xdst);
250}
251
252static void xfrm4_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
253 int unregister)
254{
255 if (!unregister)
256 return;
257
258 xfrm_dst_ifdown(dst, dev);
259}
260
261static struct dst_ops xfrm4_dst_ops_template = {
262 .family = AF_INET,
263 .update_pmtu = xfrm4_update_pmtu,
264 .redirect = xfrm4_redirect,
265 .cow_metrics = dst_cow_metrics_generic,
266 .destroy = xfrm4_dst_destroy,
267 .ifdown = xfrm4_dst_ifdown,
268 .local_out = __ip_local_out,
269 .gc_thresh = 32768,
270};
271
272static const struct xfrm_policy_afinfo xfrm4_policy_afinfo = {
273 .dst_ops = &xfrm4_dst_ops_template,
274 .dst_lookup = xfrm4_dst_lookup,
275 .get_saddr = xfrm4_get_saddr,
276 .decode_session = _decode_session4,
277 .get_tos = xfrm4_get_tos,
278 .init_path = xfrm4_init_path,
279 .fill_dst = xfrm4_fill_dst,
280 .blackhole_route = ipv4_blackhole_route,
281};
282
283#ifdef CONFIG_SYSCTL
284static struct ctl_table xfrm4_policy_table[] = {
285 {
286 .procname = "xfrm4_gc_thresh",
287 .data = &init_net.xfrm.xfrm4_dst_ops.gc_thresh,
288 .maxlen = sizeof(int),
289 .mode = 0644,
290 .proc_handler = proc_dointvec,
291 },
292 { }
293};
294
295static __net_init int xfrm4_net_sysctl_init(struct net *net)
296{
297 struct ctl_table *table;
298 struct ctl_table_header *hdr;
299
300 table = xfrm4_policy_table;
301 if (!net_eq(net, &init_net)) {
302 table = kmemdup(table, sizeof(xfrm4_policy_table), GFP_KERNEL);
303 if (!table)
304 goto err_alloc;
305
306 table[0].data = &net->xfrm.xfrm4_dst_ops.gc_thresh;
307 }
308
309 hdr = register_net_sysctl(net, "net/ipv4", table);
310 if (!hdr)
311 goto err_reg;
312
313 net->ipv4.xfrm4_hdr = hdr;
314 return 0;
315
316err_reg:
317 if (!net_eq(net, &init_net))
318 kfree(table);
319err_alloc:
320 return -ENOMEM;
321}
322
323static __net_exit void xfrm4_net_sysctl_exit(struct net *net)
324{
325 struct ctl_table *table;
326
327 if (!net->ipv4.xfrm4_hdr)
328 return;
329
330 table = net->ipv4.xfrm4_hdr->ctl_table_arg;
331 unregister_net_sysctl_table(net->ipv4.xfrm4_hdr);
332 if (!net_eq(net, &init_net))
333 kfree(table);
334}
335#else
336static inline int xfrm4_net_sysctl_init(struct net *net)
337{
338 return 0;
339}
340
341static inline void xfrm4_net_sysctl_exit(struct net *net)
342{
343}
344#endif
345
346static int __net_init xfrm4_net_init(struct net *net)
347{
348 int ret;
349
350 memcpy(&net->xfrm.xfrm4_dst_ops, &xfrm4_dst_ops_template,
351 sizeof(xfrm4_dst_ops_template));
352 ret = dst_entries_init(&net->xfrm.xfrm4_dst_ops);
353 if (ret)
354 return ret;
355
356 ret = xfrm4_net_sysctl_init(net);
357 if (ret)
358 dst_entries_destroy(&net->xfrm.xfrm4_dst_ops);
359
360 return ret;
361}
362
363static void __net_exit xfrm4_net_exit(struct net *net)
364{
365 xfrm4_net_sysctl_exit(net);
366 dst_entries_destroy(&net->xfrm.xfrm4_dst_ops);
367}
368
369static struct pernet_operations __net_initdata xfrm4_net_ops = {
370 .init = xfrm4_net_init,
371 .exit = xfrm4_net_exit,
372};
373
374static void __init xfrm4_policy_init(void)
375{
376 xfrm_policy_register_afinfo(&xfrm4_policy_afinfo, AF_INET);
377}
378
379void __init xfrm4_init(void)
380{
381 xfrm4_state_init();
382 xfrm4_policy_init();
383 xfrm4_protocol_init();
384 register_pernet_subsys(&xfrm4_net_ops);
385}
386
