1
2
3
4
5
6
7
8
9
10
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
27static struct xfrm_policy_afinfo xfrm6_policy_afinfo;
28
29static struct dst_entry *xfrm6_dst_lookup(struct net *net, int tos,
30 const xfrm_address_t *saddr,
31 const xfrm_address_t *daddr)
32{
33 struct flowi6 fl6;
34 struct dst_entry *dst;
35 int err;
36
37 memset(&fl6, 0, sizeof(fl6));
38 memcpy(&fl6.daddr, daddr, sizeof(fl6.daddr));
39 if (saddr)
40 memcpy(&fl6.saddr, saddr, sizeof(fl6.saddr));
41
42 dst = ip6_route_output(net, NULL, &fl6);
43
44 err = dst->error;
45 if (dst->error) {
46 dst_release(dst);
47 dst = ERR_PTR(err);
48 }
49
50 return dst;
51}
52
53static int xfrm6_get_saddr(struct net *net,
54 xfrm_address_t *saddr, xfrm_address_t *daddr)
55{
56 struct dst_entry *dst;
57 struct net_device *dev;
58
59 dst = xfrm6_dst_lookup(net, 0, NULL, daddr);
60 if (IS_ERR(dst))
61 return -EHOSTUNREACH;
62
63 dev = ip6_dst_idev(dst)->dev;
64 ipv6_dev_get_saddr(dev_net(dev), dev, &daddr->in6, 0, &saddr->in6);
65 dst_release(dst);
66 return 0;
67}
68
69static int xfrm6_get_tos(const struct flowi *fl)
70{
71 return 0;
72}
73
74static int xfrm6_init_path(struct xfrm_dst *path, struct dst_entry *dst,
75 int nfheader_len)
76{
77 if (dst->ops->family == AF_INET6) {
78 struct rt6_info *rt = (struct rt6_info *)dst;
79 path->path_cookie = rt6_get_cookie(rt);
80 }
81
82 path->u.rt6.rt6i_nfheader_len = nfheader_len;
83
84 return 0;
85}
86
87static int xfrm6_fill_dst(struct xfrm_dst *xdst, struct net_device *dev,
88 const struct flowi *fl)
89{
90 struct rt6_info *rt = (struct rt6_info*)xdst->route;
91
92 xdst->u.dst.dev = dev;
93 dev_hold(dev);
94
95 xdst->u.rt6.rt6i_idev = in6_dev_get(dev);
96 if (!xdst->u.rt6.rt6i_idev) {
97 dev_put(dev);
98 return -ENODEV;
99 }
100
101
102
103 xdst->u.rt6.rt6i_flags = rt->rt6i_flags & (RTF_ANYCAST |
104 RTF_LOCAL);
105 xdst->u.rt6.rt6i_metric = rt->rt6i_metric;
106 xdst->u.rt6.rt6i_node = rt->rt6i_node;
107 xdst->route_cookie = rt6_get_cookie(rt);
108 xdst->u.rt6.rt6i_gateway = rt->rt6i_gateway;
109 xdst->u.rt6.rt6i_dst = rt->rt6i_dst;
110 xdst->u.rt6.rt6i_src = rt->rt6i_src;
111
112 return 0;
113}
114
115static inline void
116_decode_session6(struct sk_buff *skb, struct flowi *fl, int reverse)
117{
118 struct flowi6 *fl6 = &fl->u.ip6;
119 int onlyproto = 0;
120 const struct ipv6hdr *hdr = ipv6_hdr(skb);
121 u32 offset = sizeof(*hdr);
122 struct ipv6_opt_hdr *exthdr;
123 const unsigned char *nh = skb_network_header(skb);
124 u8 nexthdr = nh[IP6CB(skb)->nhoff];
125 int oif = 0;
126
127 if (skb_dst(skb) && skb_dst(skb)->dev)
128 oif = skb_dst(skb)->dev->ifindex;
129
130 memset(fl6, 0, sizeof(struct flowi6));
131 fl6->flowi6_mark = skb->mark;
132 fl6->flowi6_oif = reverse ? skb->skb_iif : oif;
133
134 fl6->daddr = reverse ? hdr->saddr : hdr->daddr;
135 fl6->saddr = reverse ? hdr->daddr : hdr->saddr;
136
137 while (nh + offset + 1 < skb->data ||
138 pskb_may_pull(skb, nh + offset + 1 - skb->data)) {
139 nh = skb_network_header(skb);
140 exthdr = (struct ipv6_opt_hdr *)(nh + offset);
141
142 switch (nexthdr) {
143 case NEXTHDR_FRAGMENT:
144 onlyproto = 1;
145 case NEXTHDR_ROUTING:
146 case NEXTHDR_HOP:
147 case NEXTHDR_DEST:
148 offset += ipv6_optlen(exthdr);
149 nexthdr = exthdr->nexthdr;
150 exthdr = (struct ipv6_opt_hdr *)(nh + offset);
151 break;
152
153 case IPPROTO_UDP:
154 case IPPROTO_UDPLITE:
155 case IPPROTO_TCP:
156 case IPPROTO_SCTP:
157 case IPPROTO_DCCP:
158 if (!onlyproto && (nh + offset + 4 < skb->data ||
159 pskb_may_pull(skb, nh + offset + 4 - skb->data))) {
160 __be16 *ports = (__be16 *)exthdr;
161
162 fl6->fl6_sport = ports[!!reverse];
163 fl6->fl6_dport = ports[!reverse];
164 }
165 fl6->flowi6_proto = nexthdr;
166 return;
167
168 case IPPROTO_ICMPV6:
169 if (!onlyproto && pskb_may_pull(skb, nh + offset + 2 - skb->data)) {
170 u8 *icmp = (u8 *)exthdr;
171
172 fl6->fl6_icmp_type = icmp[0];
173 fl6->fl6_icmp_code = icmp[1];
174 }
175 fl6->flowi6_proto = nexthdr;
176 return;
177
178#if IS_ENABLED(CONFIG_IPV6_MIP6)
179 case IPPROTO_MH:
180 offset += ipv6_optlen(exthdr);
181 if (!onlyproto && pskb_may_pull(skb, nh + offset + 3 - skb->data)) {
182 struct ip6_mh *mh;
183 mh = (struct ip6_mh *)exthdr;
184
185 fl6->fl6_mh_type = mh->ip6mh_type;
186 }
187 fl6->flowi6_proto = nexthdr;
188 return;
189#endif
190
191
192 case IPPROTO_AH:
193 case IPPROTO_ESP:
194 case IPPROTO_COMP:
195 default:
196 fl6->fl6_ipsec_spi = 0;
197 fl6->flowi6_proto = nexthdr;
198 return;
199 }
200 }
201}
202
203static inline int xfrm6_garbage_collect(struct dst_ops *ops)
204{
205 struct net *net = container_of(ops, struct net, xfrm.xfrm6_dst_ops);
206
207 xfrm6_policy_afinfo.garbage_collect(net);
208 return dst_entries_get_fast(ops) > ops->gc_thresh * 2;
209}
210
211static void xfrm6_update_pmtu(struct dst_entry *dst, struct sock *sk,
212 struct sk_buff *skb, u32 mtu)
213{
214 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
215 struct dst_entry *path = xdst->route;
216
217 path->ops->update_pmtu(path, sk, skb, mtu);
218}
219
220static void xfrm6_redirect(struct dst_entry *dst, struct sock *sk,
221 struct sk_buff *skb)
222{
223 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
224 struct dst_entry *path = xdst->route;
225
226 path->ops->redirect(path, sk, skb);
227}
228
229static void xfrm6_dst_destroy(struct dst_entry *dst)
230{
231 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
232
233 if (likely(xdst->u.rt6.rt6i_idev))
234 in6_dev_put(xdst->u.rt6.rt6i_idev);
235 dst_destroy_metrics_generic(dst);
236 xfrm_dst_destroy(xdst);
237}
238
239static void xfrm6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
240 int unregister)
241{
242 struct xfrm_dst *xdst;
243
244 if (!unregister)
245 return;
246
247 xdst = (struct xfrm_dst *)dst;
248 if (xdst->u.rt6.rt6i_idev->dev == dev) {
249 struct inet6_dev *loopback_idev =
250 in6_dev_get(dev_net(dev)->loopback_dev);
251 BUG_ON(!loopback_idev);
252
253 do {
254 in6_dev_put(xdst->u.rt6.rt6i_idev);
255 xdst->u.rt6.rt6i_idev = loopback_idev;
256 in6_dev_hold(loopback_idev);
257 xdst = (struct xfrm_dst *)xdst->u.dst.child;
258 } while (xdst->u.dst.xfrm);
259
260 __in6_dev_put(loopback_idev);
261 }
262
263 xfrm_dst_ifdown(dst, dev);
264}
265
266static struct dst_ops xfrm6_dst_ops_template = {
267 .family = AF_INET6,
268 .protocol = cpu_to_be16(ETH_P_IPV6),
269 .gc = xfrm6_garbage_collect,
270 .update_pmtu = xfrm6_update_pmtu,
271 .redirect = xfrm6_redirect,
272 .cow_metrics = dst_cow_metrics_generic,
273 .destroy = xfrm6_dst_destroy,
274 .ifdown = xfrm6_dst_ifdown,
275 .local_out = __ip6_local_out,
276 .gc_thresh = 32768,
277};
278
279static struct xfrm_policy_afinfo xfrm6_policy_afinfo = {
280 .family = AF_INET6,
281 .dst_ops = &xfrm6_dst_ops_template,
282 .dst_lookup = xfrm6_dst_lookup,
283 .get_saddr = xfrm6_get_saddr,
284 .decode_session = _decode_session6,
285 .get_tos = xfrm6_get_tos,
286 .init_path = xfrm6_init_path,
287 .fill_dst = xfrm6_fill_dst,
288 .blackhole_route = ip6_blackhole_route,
289};
290
291static int __init xfrm6_policy_init(void)
292{
293 return xfrm_policy_register_afinfo(&xfrm6_policy_afinfo);
294}
295
296static void xfrm6_policy_fini(void)
297{
298 xfrm_policy_unregister_afinfo(&xfrm6_policy_afinfo);
299}
300
301#ifdef CONFIG_SYSCTL
302static struct ctl_table xfrm6_policy_table[] = {
303 {
304 .procname = "xfrm6_gc_thresh",
305 .data = &init_net.xfrm.xfrm6_dst_ops.gc_thresh,
306 .maxlen = sizeof(int),
307 .mode = 0644,
308 .proc_handler = proc_dointvec,
309 },
310 { }
311};
312
313static int __net_init xfrm6_net_sysctl_init(struct net *net)
314{
315 struct ctl_table *table;
316 struct ctl_table_header *hdr;
317
318 table = xfrm6_policy_table;
319 if (!net_eq(net, &init_net)) {
320 table = kmemdup(table, sizeof(xfrm6_policy_table), GFP_KERNEL);
321 if (!table)
322 goto err_alloc;
323
324 table[0].data = &net->xfrm.xfrm6_dst_ops.gc_thresh;
325 }
326
327 hdr = register_net_sysctl(net, "net/ipv6", table);
328 if (!hdr)
329 goto err_reg;
330
331 net->ipv6.sysctl.xfrm6_hdr = hdr;
332 return 0;
333
334err_reg:
335 if (!net_eq(net, &init_net))
336 kfree(table);
337err_alloc:
338 return -ENOMEM;
339}
340
341static void __net_exit xfrm6_net_sysctl_exit(struct net *net)
342{
343 struct ctl_table *table;
344
345 if (net->ipv6.sysctl.xfrm6_hdr == NULL)
346 return;
347
348 table = net->ipv6.sysctl.xfrm6_hdr->ctl_table_arg;
349 unregister_net_sysctl_table(net->ipv6.sysctl.xfrm6_hdr);
350 if (!net_eq(net, &init_net))
351 kfree(table);
352}
353#else
354static int inline xfrm6_net_sysctl_init(struct net *net)
355{
356 return 0;
357}
358
359static void inline xfrm6_net_sysctl_exit(struct net *net)
360{
361}
362#endif
363
364static int __net_init xfrm6_net_init(struct net *net)
365{
366 int ret;
367
368 memcpy(&net->xfrm.xfrm6_dst_ops, &xfrm6_dst_ops_template,
369 sizeof(xfrm6_dst_ops_template));
370 ret = dst_entries_init(&net->xfrm.xfrm6_dst_ops);
371 if (ret)
372 return ret;
373
374 ret = xfrm6_net_sysctl_init(net);
375 if (ret)
376 dst_entries_destroy(&net->xfrm.xfrm6_dst_ops);
377
378 return ret;
379}
380
381static void __net_exit xfrm6_net_exit(struct net *net)
382{
383 xfrm6_net_sysctl_exit(net);
384 dst_entries_destroy(&net->xfrm.xfrm6_dst_ops);
385}
386
387static struct pernet_operations xfrm6_net_ops = {
388 .init = xfrm6_net_init,
389 .exit = xfrm6_net_exit,
390};
391
392int __init xfrm6_init(void)
393{
394 int ret;
395
396 ret = xfrm6_policy_init();
397 if (ret)
398 goto out;
399 ret = xfrm6_state_init();
400 if (ret)
401 goto out_policy;
402
403 ret = xfrm6_protocol_init();
404 if (ret)
405 goto out_state;
406
407 register_pernet_subsys(&xfrm6_net_ops);
408out:
409 return ret;
410out_state:
411 xfrm6_state_fini();
412out_policy:
413 xfrm6_policy_fini();
414 goto out;
415}
416
417void xfrm6_fini(void)
418{
419 unregister_pernet_subsys(&xfrm6_net_ops);
420 xfrm6_protocol_fini();
421 xfrm6_policy_fini();
422 xfrm6_state_fini();
423}
424
