1 /*
2 * xfrm6_policy.c: based on xfrm4_policy.c
3 *
4 * Authors:
5 * Mitsuru KANDA @USAGI
6 * Kazunori MIYAZAWA @USAGI
7 * Kunihiro Ishiguro <kunihiro@ipinfusion.com>
8 * IPv6 support
9 * YOSHIFUJI Hideaki
10 * Split up af-specific portion
11 *
12 */
13
14 #include <linux/err.h>
15 #include <linux/kernel.h>
16 #include <linux/netdevice.h>
17 #include <net/addrconf.h>
18 #include <net/dst.h>
19 #include <net/xfrm.h>
20 #include <net/ip.h>
21 #include <net/ipv6.h>
22 #include <net/ip6_route.h>
23 #if IS_ENABLED(CONFIG_IPV6_MIP6)
24 #include <net/mip6.h>
25 #endif
26
27 static struct xfrm_policy_afinfo xfrm6_policy_afinfo;
28
xfrm6_dst_lookup(struct net * net,int tos,const xfrm_address_t * saddr,const xfrm_address_t * daddr,u32 mark)29 static struct dst_entry *xfrm6_dst_lookup(struct net *net, int tos,
30 const xfrm_address_t *saddr,
31 const xfrm_address_t *daddr,
32 u32 mark)
33 {
34 struct flowi6 fl6;
35 struct dst_entry *dst;
36 int err;
37
38 memset(&fl6, 0, sizeof(fl6));
39 fl6.flowi6_mark = mark;
40 memcpy(&fl6.daddr, daddr, sizeof(fl6.daddr));
41 if (saddr)
42 memcpy(&fl6.saddr, saddr, sizeof(fl6.saddr));
43
44 dst = ip6_route_output(net, NULL, &fl6);
45
46 err = dst->error;
47 if (dst->error) {
48 dst_release(dst);
49 dst = ERR_PTR(err);
50 }
51
52 return dst;
53 }
54
xfrm6_get_saddr(struct net * net,xfrm_address_t * saddr,xfrm_address_t * daddr,u32 mark)55 static int xfrm6_get_saddr(struct net *net,
56 xfrm_address_t *saddr, xfrm_address_t *daddr,
57 u32 mark)
58 {
59 struct dst_entry *dst;
60 struct net_device *dev;
61
62 dst = xfrm6_dst_lookup(net, 0, NULL, daddr, mark);
63 if (IS_ERR(dst))
64 return -EHOSTUNREACH;
65
66 dev = ip6_dst_idev(dst)->dev;
67 ipv6_dev_get_saddr(dev_net(dev), dev,
68 (struct in6_addr *)&daddr->a6, 0,
69 (struct in6_addr *)&saddr->a6);
70 dst_release(dst);
71 return 0;
72 }
73
xfrm6_get_tos(const struct flowi * fl)74 static int xfrm6_get_tos(const struct flowi *fl)
75 {
76 return 0;
77 }
78
xfrm6_init_dst(struct net * net,struct xfrm_dst * xdst)79 static void xfrm6_init_dst(struct net *net, struct xfrm_dst *xdst)
80 {
81 struct rt6_info *rt = (struct rt6_info *)xdst;
82
83 rt6_init_peer(rt, net->ipv6.peers);
84 }
85
xfrm6_init_path(struct xfrm_dst * path,struct dst_entry * dst,int nfheader_len)86 static int xfrm6_init_path(struct xfrm_dst *path, struct dst_entry *dst,
87 int nfheader_len)
88 {
89 if (dst->ops->family == AF_INET6) {
90 struct rt6_info *rt = (struct rt6_info*)dst;
91 if (rt->rt6i_node)
92 path->path_cookie = rt->rt6i_node->fn_sernum;
93 }
94
95 path->u.rt6.rt6i_nfheader_len = nfheader_len;
96
97 return 0;
98 }
99
xfrm6_fill_dst(struct xfrm_dst * xdst,struct net_device * dev,const struct flowi * fl)100 static int xfrm6_fill_dst(struct xfrm_dst *xdst, struct net_device *dev,
101 const struct flowi *fl)
102 {
103 struct rt6_info *rt = (struct rt6_info*)xdst->route;
104
105 xdst->u.dst.dev = dev;
106 dev_hold(dev);
107
108 xdst->u.rt6.rt6i_idev = in6_dev_get(dev);
109 if (!xdst->u.rt6.rt6i_idev) {
110 dev_put(dev);
111 return -ENODEV;
112 }
113
114 rt6_transfer_peer(&xdst->u.rt6, rt);
115
116 /* Sheit... I remember I did this right. Apparently,
117 * it was magically lost, so this code needs audit */
118 xdst->u.rt6.rt6i_flags = rt->rt6i_flags & (RTF_ANYCAST |
119 RTF_LOCAL);
120 xdst->u.rt6.rt6i_metric = rt->rt6i_metric;
121 xdst->u.rt6.rt6i_node = rt->rt6i_node;
122 if (rt->rt6i_node)
123 xdst->route_cookie = rt->rt6i_node->fn_sernum;
124 xdst->u.rt6.rt6i_gateway = rt->rt6i_gateway;
125 xdst->u.rt6.rt6i_dst = rt->rt6i_dst;
126 xdst->u.rt6.rt6i_src = rt->rt6i_src;
127
128 return 0;
129 }
130
131 static inline void
_decode_session6(struct sk_buff * skb,struct flowi * fl,int reverse)132 _decode_session6(struct sk_buff *skb, struct flowi *fl, int reverse)
133 {
134 struct flowi6 *fl6 = &fl->u.ip6;
135 int onlyproto = 0;
136 u16 offset = skb_network_header_len(skb);
137 const struct ipv6hdr *hdr = ipv6_hdr(skb);
138 struct ipv6_opt_hdr *exthdr;
139 const unsigned char *nh = skb_network_header(skb);
140 u8 nexthdr = nh[IP6CB(skb)->nhoff];
141
142 memset(fl6, 0, sizeof(struct flowi6));
143 fl6->flowi6_mark = skb->mark;
144
145 fl6->daddr = reverse ? hdr->saddr : hdr->daddr;
146 fl6->saddr = reverse ? hdr->daddr : hdr->saddr;
147
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);
152
153 switch (nexthdr) {
154 case NEXTHDR_FRAGMENT:
155 onlyproto = 1;
156 case NEXTHDR_ROUTING:
157 case NEXTHDR_HOP:
158 case NEXTHDR_DEST:
159 offset += ipv6_optlen(exthdr);
160 nexthdr = exthdr->nexthdr;
161 exthdr = (struct ipv6_opt_hdr *)(nh + offset);
162 break;
163
164 case IPPROTO_UDP:
165 case IPPROTO_UDPLITE:
166 case IPPROTO_TCP:
167 case IPPROTO_SCTP:
168 case IPPROTO_DCCP:
169 if (!onlyproto && (nh + offset + 4 < skb->data ||
170 pskb_may_pull(skb, nh + offset + 4 - skb->data))) {
171 __be16 *ports = (__be16 *)exthdr;
172
173 fl6->fl6_sport = ports[!!reverse];
174 fl6->fl6_dport = ports[!reverse];
175 }
176 fl6->flowi6_proto = nexthdr;
177 return;
178
179 case IPPROTO_ICMPV6:
180 if (!onlyproto && pskb_may_pull(skb, nh + offset + 2 - skb->data)) {
181 u8 *icmp = (u8 *)exthdr;
182
183 fl6->fl6_icmp_type = icmp[0];
184 fl6->fl6_icmp_code = icmp[1];
185 }
186 fl6->flowi6_proto = nexthdr;
187 return;
188
189 #if IS_ENABLED(CONFIG_IPV6_MIP6)
190 case IPPROTO_MH:
191 if (!onlyproto && pskb_may_pull(skb, nh + offset + 3 - skb->data)) {
192 struct ip6_mh *mh;
193 mh = (struct ip6_mh *)exthdr;
194
195 fl6->fl6_mh_type = mh->ip6mh_type;
196 }
197 fl6->flowi6_proto = nexthdr;
198 return;
199 #endif
200
201 /* XXX Why are there these headers? */
202 case IPPROTO_AH:
203 case IPPROTO_ESP:
204 case IPPROTO_COMP:
205 default:
206 fl6->fl6_ipsec_spi = 0;
207 fl6->flowi6_proto = nexthdr;
208 return;
209 }
210 }
211 }
212
xfrm6_garbage_collect(struct dst_ops * ops)213 static inline int xfrm6_garbage_collect(struct dst_ops *ops)
214 {
215 struct net *net = container_of(ops, struct net, xfrm.xfrm6_dst_ops);
216
217 xfrm6_policy_afinfo.garbage_collect(net);
218 return dst_entries_get_fast(ops) > ops->gc_thresh * 2;
219 }
220
xfrm6_update_pmtu(struct dst_entry * dst,struct sock * sk,struct sk_buff * skb,u32 mtu)221 static void xfrm6_update_pmtu(struct dst_entry *dst, struct sock *sk,
222 struct sk_buff *skb, u32 mtu)
223 {
224 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
225 struct dst_entry *path = xdst->route;
226
227 path->ops->update_pmtu(path, sk, skb, mtu);
228 }
229
xfrm6_redirect(struct dst_entry * dst,struct sock * sk,struct sk_buff * skb)230 static void xfrm6_redirect(struct dst_entry *dst, struct sock *sk,
231 struct sk_buff *skb)
232 {
233 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
234 struct dst_entry *path = xdst->route;
235
236 path->ops->redirect(path, sk, skb);
237 }
238
xfrm6_dst_destroy(struct dst_entry * dst)239 static void xfrm6_dst_destroy(struct dst_entry *dst)
240 {
241 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
242
243 if (likely(xdst->u.rt6.rt6i_idev))
244 in6_dev_put(xdst->u.rt6.rt6i_idev);
245 dst_destroy_metrics_generic(dst);
246 if (rt6_has_peer(&xdst->u.rt6)) {
247 struct inet_peer *peer = rt6_peer_ptr(&xdst->u.rt6);
248 inet_putpeer(peer);
249 }
250 xfrm_dst_destroy(xdst);
251 }
252
xfrm6_dst_ifdown(struct dst_entry * dst,struct net_device * dev,int unregister)253 static void xfrm6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
254 int unregister)
255 {
256 struct xfrm_dst *xdst;
257
258 if (!unregister)
259 return;
260
261 xdst = (struct xfrm_dst *)dst;
262 if (xdst->u.rt6.rt6i_idev->dev == dev) {
263 struct inet6_dev *loopback_idev =
264 in6_dev_get(dev_net(dev)->loopback_dev);
265 BUG_ON(!loopback_idev);
266
267 do {
268 in6_dev_put(xdst->u.rt6.rt6i_idev);
269 xdst->u.rt6.rt6i_idev = loopback_idev;
270 in6_dev_hold(loopback_idev);
271 xdst = (struct xfrm_dst *)xdst->u.dst.child;
272 } while (xdst->u.dst.xfrm);
273
274 __in6_dev_put(loopback_idev);
275 }
276
277 xfrm_dst_ifdown(dst, dev);
278 }
279
280 static struct dst_ops xfrm6_dst_ops = {
281 .family = AF_INET6,
282 .protocol = cpu_to_be16(ETH_P_IPV6),
283 .gc = xfrm6_garbage_collect,
284 .update_pmtu = xfrm6_update_pmtu,
285 .redirect = xfrm6_redirect,
286 .cow_metrics = dst_cow_metrics_generic,
287 .destroy = xfrm6_dst_destroy,
288 .ifdown = xfrm6_dst_ifdown,
289 .local_out = __ip6_local_out,
290 .gc_thresh = 1024,
291 };
292
293 static struct xfrm_policy_afinfo xfrm6_policy_afinfo = {
294 .family = AF_INET6,
295 .dst_ops = &xfrm6_dst_ops,
296 .dst_lookup = xfrm6_dst_lookup,
297 .get_saddr = xfrm6_get_saddr,
298 .decode_session = _decode_session6,
299 .get_tos = xfrm6_get_tos,
300 .init_dst = xfrm6_init_dst,
301 .init_path = xfrm6_init_path,
302 .fill_dst = xfrm6_fill_dst,
303 .blackhole_route = ip6_blackhole_route,
304 };
305
xfrm6_policy_init(void)306 static int __init xfrm6_policy_init(void)
307 {
308 return xfrm_policy_register_afinfo(&xfrm6_policy_afinfo);
309 }
310
xfrm6_policy_fini(void)311 static void xfrm6_policy_fini(void)
312 {
313 xfrm_policy_unregister_afinfo(&xfrm6_policy_afinfo);
314 }
315
316 #ifdef CONFIG_SYSCTL
317 static struct ctl_table xfrm6_policy_table[] = {
318 {
319 .procname = "xfrm6_gc_thresh",
320 .data = &init_net.xfrm.xfrm6_dst_ops.gc_thresh,
321 .maxlen = sizeof(int),
322 .mode = 0644,
323 .proc_handler = proc_dointvec,
324 },
325 { }
326 };
327
xfrm6_net_init(struct net * net)328 static int __net_init xfrm6_net_init(struct net *net)
329 {
330 struct ctl_table *table;
331 struct ctl_table_header *hdr;
332
333 table = xfrm6_policy_table;
334 if (!net_eq(net, &init_net)) {
335 table = kmemdup(table, sizeof(xfrm6_policy_table), GFP_KERNEL);
336 if (!table)
337 goto err_alloc;
338
339 table[0].data = &net->xfrm.xfrm6_dst_ops.gc_thresh;
340 }
341
342 hdr = register_net_sysctl(net, "net/ipv6", table);
343 if (!hdr)
344 goto err_reg;
345
346 net->ipv6.sysctl.xfrm6_hdr = hdr;
347 return 0;
348
349 err_reg:
350 if (!net_eq(net, &init_net))
351 kfree(table);
352 err_alloc:
353 return -ENOMEM;
354 }
355
xfrm6_net_exit(struct net * net)356 static void __net_exit xfrm6_net_exit(struct net *net)
357 {
358 struct ctl_table *table;
359
360 if (net->ipv6.sysctl.xfrm6_hdr == NULL)
361 return;
362
363 table = net->ipv6.sysctl.xfrm6_hdr->ctl_table_arg;
364 unregister_net_sysctl_table(net->ipv6.sysctl.xfrm6_hdr);
365 if (!net_eq(net, &init_net))
366 kfree(table);
367 }
368
369 static struct pernet_operations xfrm6_net_ops = {
370 .init = xfrm6_net_init,
371 .exit = xfrm6_net_exit,
372 };
373 #endif
374
xfrm6_init(void)375 int __init xfrm6_init(void)
376 {
377 int ret;
378
379 dst_entries_init(&xfrm6_dst_ops);
380
381 ret = xfrm6_policy_init();
382 if (ret) {
383 dst_entries_destroy(&xfrm6_dst_ops);
384 goto out;
385 }
386 ret = xfrm6_state_init();
387 if (ret)
388 goto out_policy;
389
390 #ifdef CONFIG_SYSCTL
391 register_pernet_subsys(&xfrm6_net_ops);
392 #endif
393 out:
394 return ret;
395 out_policy:
396 xfrm6_policy_fini();
397 goto out;
398 }
399
xfrm6_fini(void)400 void xfrm6_fini(void)
401 {
402 #ifdef CONFIG_SYSCTL
403 unregister_pernet_subsys(&xfrm6_net_ops);
404 #endif
405 xfrm6_policy_fini();
406 xfrm6_state_fini();
407 dst_entries_destroy(&xfrm6_dst_ops);
408 }
409