1 /*
2 * xfrm4_policy.c
3 *
4 * Changes:
5 * Kazunori MIYAZAWA @USAGI
6 * YOSHIFUJI Hideaki @USAGI
7 * Split up af-specific portion
8 *
9 */
10
11 #include <linux/err.h>
12 #include <linux/kernel.h>
13 #include <linux/inetdevice.h>
14 #include <linux/if_tunnel.h>
15 #include <net/dst.h>
16 #include <net/xfrm.h>
17 #include <net/ip.h>
18
19 static struct xfrm_policy_afinfo xfrm4_policy_afinfo;
20
__xfrm4_dst_lookup(struct net * net,struct flowi4 * fl4,int tos,const xfrm_address_t * saddr,const xfrm_address_t * daddr,u32 mark)21 static struct dst_entry *__xfrm4_dst_lookup(struct net *net, struct flowi4 *fl4,
22 int tos,
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_mark = mark;
33 if (saddr)
34 fl4->saddr = saddr->a4;
35
36 rt = __ip_route_output_key(net, fl4);
37 if (!IS_ERR(rt))
38 return &rt->dst;
39
40 return ERR_CAST(rt);
41 }
42
xfrm4_dst_lookup(struct net * net,int tos,const xfrm_address_t * saddr,const xfrm_address_t * daddr,u32 mark)43 static struct dst_entry *xfrm4_dst_lookup(struct net *net, int tos,
44 const xfrm_address_t *saddr,
45 const xfrm_address_t *daddr,
46 u32 mark)
47 {
48 struct flowi4 fl4;
49
50 return __xfrm4_dst_lookup(net, &fl4, tos, saddr, daddr, mark);
51 }
52
xfrm4_get_saddr(struct net * net,xfrm_address_t * saddr,xfrm_address_t * daddr,u32 mark)53 static int xfrm4_get_saddr(struct net *net,
54 xfrm_address_t *saddr, xfrm_address_t *daddr,
55 u32 mark)
56 {
57 struct dst_entry *dst;
58 struct flowi4 fl4;
59
60 dst = __xfrm4_dst_lookup(net, &fl4, 0, NULL, daddr, mark);
61 if (IS_ERR(dst))
62 return -EHOSTUNREACH;
63
64 saddr->a4 = fl4.saddr;
65 dst_release(dst);
66 return 0;
67 }
68
xfrm4_get_tos(const struct flowi * fl)69 static int xfrm4_get_tos(const struct flowi *fl)
70 {
71 return IPTOS_RT_MASK & fl->u.ip4.flowi4_tos; /* Strip ECN bits */
72 }
73
xfrm4_init_path(struct xfrm_dst * path,struct dst_entry * dst,int nfheader_len)74 static int xfrm4_init_path(struct xfrm_dst *path, struct dst_entry *dst,
75 int nfheader_len)
76 {
77 return 0;
78 }
79
xfrm4_fill_dst(struct xfrm_dst * xdst,struct net_device * dev,const struct flowi * fl)80 static int xfrm4_fill_dst(struct xfrm_dst *xdst, struct net_device *dev,
81 const struct flowi *fl)
82 {
83 struct rtable *rt = (struct rtable *)xdst->route;
84 const struct flowi4 *fl4 = &fl->u.ip4;
85
86 xdst->u.rt.rt_iif = fl4->flowi4_iif;
87
88 xdst->u.dst.dev = dev;
89 dev_hold(dev);
90
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 |
95 RTCF_LOCAL);
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 INIT_LIST_HEAD(&xdst->u.rt.rt_uncached);
101
102 return 0;
103 }
104
105 static void
_decode_session4(struct sk_buff * skb,struct flowi * fl,int reverse)106 _decode_session4(struct sk_buff *skb, struct flowi *fl, int reverse)
107 {
108 const struct iphdr *iph = ip_hdr(skb);
109 u8 *xprth = skb_network_header(skb) + iph->ihl * 4;
110 struct flowi4 *fl4 = &fl->u.ip4;
111
112 memset(fl4, 0, sizeof(struct flowi4));
113 fl4->flowi4_mark = skb->mark;
114
115 if (!ip_is_fragment(iph)) {
116 switch (iph->protocol) {
117 case IPPROTO_UDP:
118 case IPPROTO_UDPLITE:
119 case IPPROTO_TCP:
120 case IPPROTO_SCTP:
121 case IPPROTO_DCCP:
122 if (xprth + 4 < skb->data ||
123 pskb_may_pull(skb, xprth + 4 - skb->data)) {
124 __be16 *ports = (__be16 *)xprth;
125
126 fl4->fl4_sport = ports[!!reverse];
127 fl4->fl4_dport = ports[!reverse];
128 }
129 break;
130
131 case IPPROTO_ICMP:
132 if (pskb_may_pull(skb, xprth + 2 - skb->data)) {
133 u8 *icmp = xprth;
134
135 fl4->fl4_icmp_type = icmp[0];
136 fl4->fl4_icmp_code = icmp[1];
137 }
138 break;
139
140 case IPPROTO_ESP:
141 if (pskb_may_pull(skb, xprth + 4 - skb->data)) {
142 __be32 *ehdr = (__be32 *)xprth;
143
144 fl4->fl4_ipsec_spi = ehdr[0];
145 }
146 break;
147
148 case IPPROTO_AH:
149 if (pskb_may_pull(skb, xprth + 8 - skb->data)) {
150 __be32 *ah_hdr = (__be32 *)xprth;
151
152 fl4->fl4_ipsec_spi = ah_hdr[1];
153 }
154 break;
155
156 case IPPROTO_COMP:
157 if (pskb_may_pull(skb, xprth + 4 - skb->data)) {
158 __be16 *ipcomp_hdr = (__be16 *)xprth;
159
160 fl4->fl4_ipsec_spi = htonl(ntohs(ipcomp_hdr[1]));
161 }
162 break;
163
164 case IPPROTO_GRE:
165 if (pskb_may_pull(skb, xprth + 12 - skb->data)) {
166 __be16 *greflags = (__be16 *)xprth;
167 __be32 *gre_hdr = (__be32 *)xprth;
168
169 if (greflags[0] & GRE_KEY) {
170 if (greflags[0] & GRE_CSUM)
171 gre_hdr++;
172 fl4->fl4_gre_key = gre_hdr[1];
173 }
174 }
175 break;
176
177 default:
178 fl4->fl4_ipsec_spi = 0;
179 break;
180 }
181 }
182 fl4->flowi4_proto = iph->protocol;
183 fl4->daddr = reverse ? iph->saddr : iph->daddr;
184 fl4->saddr = reverse ? iph->daddr : iph->saddr;
185 fl4->flowi4_tos = iph->tos;
186 }
187
xfrm4_garbage_collect(struct dst_ops * ops)188 static inline int xfrm4_garbage_collect(struct dst_ops *ops)
189 {
190 struct net *net = container_of(ops, struct net, xfrm.xfrm4_dst_ops);
191
192 xfrm4_policy_afinfo.garbage_collect(net);
193 return (dst_entries_get_slow(ops) > ops->gc_thresh * 2);
194 }
195
xfrm4_update_pmtu(struct dst_entry * dst,struct sock * sk,struct sk_buff * skb,u32 mtu)196 static void xfrm4_update_pmtu(struct dst_entry *dst, struct sock *sk,
197 struct sk_buff *skb, u32 mtu)
198 {
199 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
200 struct dst_entry *path = xdst->route;
201
202 path->ops->update_pmtu(path, sk, skb, mtu);
203 }
204
xfrm4_redirect(struct dst_entry * dst,struct sock * sk,struct sk_buff * skb)205 static void xfrm4_redirect(struct dst_entry *dst, struct sock *sk,
206 struct sk_buff *skb)
207 {
208 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
209 struct dst_entry *path = xdst->route;
210
211 path->ops->redirect(path, sk, skb);
212 }
213
xfrm4_dst_destroy(struct dst_entry * dst)214 static void xfrm4_dst_destroy(struct dst_entry *dst)
215 {
216 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
217
218 dst_destroy_metrics_generic(dst);
219
220 xfrm_dst_destroy(xdst);
221 }
222
xfrm4_dst_ifdown(struct dst_entry * dst,struct net_device * dev,int unregister)223 static void xfrm4_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
224 int unregister)
225 {
226 if (!unregister)
227 return;
228
229 xfrm_dst_ifdown(dst, dev);
230 }
231
232 static struct dst_ops xfrm4_dst_ops = {
233 .family = AF_INET,
234 .protocol = cpu_to_be16(ETH_P_IP),
235 .gc = xfrm4_garbage_collect,
236 .update_pmtu = xfrm4_update_pmtu,
237 .redirect = xfrm4_redirect,
238 .cow_metrics = dst_cow_metrics_generic,
239 .destroy = xfrm4_dst_destroy,
240 .ifdown = xfrm4_dst_ifdown,
241 .local_out = __ip_local_out,
242 .gc_thresh = 1024,
243 };
244
245 static struct xfrm_policy_afinfo xfrm4_policy_afinfo = {
246 .family = AF_INET,
247 .dst_ops = &xfrm4_dst_ops,
248 .dst_lookup = xfrm4_dst_lookup,
249 .get_saddr = xfrm4_get_saddr,
250 .decode_session = _decode_session4,
251 .get_tos = xfrm4_get_tos,
252 .init_path = xfrm4_init_path,
253 .fill_dst = xfrm4_fill_dst,
254 .blackhole_route = ipv4_blackhole_route,
255 };
256
257 #ifdef CONFIG_SYSCTL
258 static struct ctl_table xfrm4_policy_table[] = {
259 {
260 .procname = "xfrm4_gc_thresh",
261 .data = &init_net.xfrm.xfrm4_dst_ops.gc_thresh,
262 .maxlen = sizeof(int),
263 .mode = 0644,
264 .proc_handler = proc_dointvec,
265 },
266 { }
267 };
268
xfrm4_net_init(struct net * net)269 static int __net_init xfrm4_net_init(struct net *net)
270 {
271 struct ctl_table *table;
272 struct ctl_table_header *hdr;
273
274 table = xfrm4_policy_table;
275 if (!net_eq(net, &init_net)) {
276 table = kmemdup(table, sizeof(xfrm4_policy_table), GFP_KERNEL);
277 if (!table)
278 goto err_alloc;
279
280 table[0].data = &net->xfrm.xfrm4_dst_ops.gc_thresh;
281 }
282
283 hdr = register_net_sysctl(net, "net/ipv4", table);
284 if (!hdr)
285 goto err_reg;
286
287 net->ipv4.xfrm4_hdr = hdr;
288 return 0;
289
290 err_reg:
291 if (!net_eq(net, &init_net))
292 kfree(table);
293 err_alloc:
294 return -ENOMEM;
295 }
296
xfrm4_net_exit(struct net * net)297 static void __net_exit xfrm4_net_exit(struct net *net)
298 {
299 struct ctl_table *table;
300
301 if (net->ipv4.xfrm4_hdr == NULL)
302 return;
303
304 table = net->ipv4.xfrm4_hdr->ctl_table_arg;
305 unregister_net_sysctl_table(net->ipv4.xfrm4_hdr);
306 if (!net_eq(net, &init_net))
307 kfree(table);
308 }
309
310 static struct pernet_operations __net_initdata xfrm4_net_ops = {
311 .init = xfrm4_net_init,
312 .exit = xfrm4_net_exit,
313 };
314 #endif
315
xfrm4_policy_init(void)316 static void __init xfrm4_policy_init(void)
317 {
318 xfrm_policy_register_afinfo(&xfrm4_policy_afinfo);
319 }
320
xfrm4_init(void)321 void __init xfrm4_init(void)
322 {
323 dst_entries_init(&xfrm4_dst_ops);
324
325 xfrm4_state_init();
326 xfrm4_policy_init();
327 #ifdef CONFIG_SYSCTL
328 register_pernet_subsys(&xfrm4_net_ops);
329 #endif
330 }
331
332