Merge tag 'cleanup2' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
[GitHub/mt8127/android_kernel_alcatel_ttab.git] / net / ipv6 / xfrm6_policy.c
CommitLineData
1da177e4
LT
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
1ab1457c 11 *
1da177e4
LT
12 */
13
66cdb3ca
HX
14#include <linux/err.h>
15#include <linux/kernel.h>
aabc9761
HX
16#include <linux/netdevice.h>
17#include <net/addrconf.h>
45ff5a3f 18#include <net/dst.h>
1da177e4
LT
19#include <net/xfrm.h>
20#include <net/ip.h>
21#include <net/ipv6.h>
22#include <net/ip6_route.h>
59fbb3a6 23#if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE)
2ce4272a
MN
24#include <net/mip6.h>
25#endif
1da177e4 26
1da177e4
LT
27static struct xfrm_policy_afinfo xfrm6_policy_afinfo;
28
c5b3cf46 29static struct dst_entry *xfrm6_dst_lookup(struct net *net, int tos,
5e6b930f
DM
30 const xfrm_address_t *saddr,
31 const xfrm_address_t *daddr)
1da177e4 32{
7e1dc7b6 33 struct flowi6 fl6;
66cdb3ca
HX
34 struct dst_entry *dst;
35 int err;
36
7e1dc7b6
DM
37 memset(&fl6, 0, sizeof(fl6));
38 memcpy(&fl6.daddr, daddr, sizeof(fl6.daddr));
66cdb3ca 39 if (saddr)
7e1dc7b6 40 memcpy(&fl6.saddr, saddr, sizeof(fl6.saddr));
66cdb3ca 41
4c9483b2 42 dst = ip6_route_output(net, NULL, &fl6);
66cdb3ca
HX
43
44 err = dst->error;
45 if (dst->error) {
4251320f 46 dst_release(dst);
66cdb3ca
HX
47 dst = ERR_PTR(err);
48 }
49
50 return dst;
1da177e4
LT
51}
52
fbda33b2
AD
53static int xfrm6_get_saddr(struct net *net,
54 xfrm_address_t *saddr, xfrm_address_t *daddr)
a1e59abf 55{
66cdb3ca 56 struct dst_entry *dst;
191cd582 57 struct net_device *dev;
66cdb3ca 58
fbda33b2 59 dst = xfrm6_dst_lookup(net, 0, NULL, daddr);
66cdb3ca
HX
60 if (IS_ERR(dst))
61 return -EHOSTUNREACH;
62
191cd582
BH
63 dev = ip6_dst_idev(dst)->dev;
64 ipv6_dev_get_saddr(dev_net(dev), dev,
7cbca67c 65 (struct in6_addr *)&daddr->a6, 0,
5e5f3f0f 66 (struct in6_addr *)&saddr->a6);
66cdb3ca
HX
67 dst_release(dst);
68 return 0;
a1e59abf
PM
69}
70
05d84025 71static int xfrm6_get_tos(const struct flowi *fl)
1da177e4 72{
25ee3286
HX
73 return 0;
74}
1da177e4 75
a1b05140
MN
76static int xfrm6_init_path(struct xfrm_dst *path, struct dst_entry *dst,
77 int nfheader_len)
78{
79 if (dst->ops->family == AF_INET6) {
80 struct rt6_info *rt = (struct rt6_info*)dst;
81 if (rt->rt6i_node)
82 path->path_cookie = rt->rt6i_node->fn_sernum;
83 }
84
85 path->u.rt6.rt6i_nfheader_len = nfheader_len;
86
87 return 0;
88}
89
87c1e12b 90static int xfrm6_fill_dst(struct xfrm_dst *xdst, struct net_device *dev,
0c7b3eef 91 const struct flowi *fl)
25ee3286
HX
92{
93 struct rt6_info *rt = (struct rt6_info*)xdst->route;
1da177e4 94
25ee3286
HX
95 xdst->u.dst.dev = dev;
96 dev_hold(dev);
1da177e4 97
bc8e4b95 98 xdst->u.rt6.rt6i_idev = in6_dev_get(dev);
25ee3286
HX
99 if (!xdst->u.rt6.rt6i_idev)
100 return -ENODEV;
1da177e4 101
7cc2edb8
DM
102 xdst->u.rt6.rt6i_peer = rt->rt6i_peer;
103 if (rt->rt6i_peer)
104 atomic_inc(&rt->rt6i_peer->refcnt);
105
25ee3286
HX
106 /* Sheit... I remember I did this right. Apparently,
107 * it was magically lost, so this code needs audit */
108 xdst->u.rt6.rt6i_flags = rt->rt6i_flags & (RTF_ANYCAST |
109 RTF_LOCAL);
110 xdst->u.rt6.rt6i_metric = rt->rt6i_metric;
111 xdst->u.rt6.rt6i_node = rt->rt6i_node;
a1b05140
MN
112 if (rt->rt6i_node)
113 xdst->route_cookie = rt->rt6i_node->fn_sernum;
25ee3286
HX
114 xdst->u.rt6.rt6i_gateway = rt->rt6i_gateway;
115 xdst->u.rt6.rt6i_dst = rt->rt6i_dst;
116 xdst->u.rt6.rt6i_src = rt->rt6i_src;
1da177e4 117
1da177e4 118 return 0;
1da177e4
LT
119}
120
121static inline void
d5422efe 122_decode_session6(struct sk_buff *skb, struct flowi *fl, int reverse)
1da177e4 123{
7e1dc7b6 124 struct flowi6 *fl6 = &fl->u.ip6;
7e3a42a1 125 int onlyproto = 0;
cfe1fc77 126 u16 offset = skb_network_header_len(skb);
b71d1d42 127 const struct ipv6hdr *hdr = ipv6_hdr(skb);
e3cae904 128 struct ipv6_opt_hdr *exthdr;
d56f90a7
ACM
129 const unsigned char *nh = skb_network_header(skb);
130 u8 nexthdr = nh[IP6CB(skb)->nhoff];
1da177e4 131
7e1dc7b6
DM
132 memset(fl6, 0, sizeof(struct flowi6));
133 fl6->flowi6_mark = skb->mark;
44b451f1 134
4e3fd7a0
AD
135 fl6->daddr = reverse ? hdr->saddr : hdr->daddr;
136 fl6->saddr = reverse ? hdr->daddr : hdr->saddr;
1da177e4 137
59cae009
WY
138 while (nh + offset + 1 < skb->data ||
139 pskb_may_pull(skb, nh + offset + 1 - skb->data)) {
d56f90a7
ACM
140 nh = skb_network_header(skb);
141 exthdr = (struct ipv6_opt_hdr *)(nh + offset);
e3cae904 142
1da177e4 143 switch (nexthdr) {
7e3a42a1
ND
144 case NEXTHDR_FRAGMENT:
145 onlyproto = 1;
1da177e4
LT
146 case NEXTHDR_ROUTING:
147 case NEXTHDR_HOP:
148 case NEXTHDR_DEST:
149 offset += ipv6_optlen(exthdr);
150 nexthdr = exthdr->nexthdr;
d56f90a7 151 exthdr = (struct ipv6_opt_hdr *)(nh + offset);
1da177e4
LT
152 break;
153
154 case IPPROTO_UDP:
ba4e58ec 155 case IPPROTO_UDPLITE:
1da177e4
LT
156 case IPPROTO_TCP:
157 case IPPROTO_SCTP:
9e999993 158 case IPPROTO_DCCP:
59cae009
WY
159 if (!onlyproto && (nh + offset + 4 < skb->data ||
160 pskb_may_pull(skb, nh + offset + 4 - skb->data))) {
8c689a6e 161 __be16 *ports = (__be16 *)exthdr;
1da177e4 162
1958b856
DM
163 fl6->fl6_sport = ports[!!reverse];
164 fl6->fl6_dport = ports[!reverse];
1da177e4 165 }
7e1dc7b6 166 fl6->flowi6_proto = nexthdr;
1da177e4
LT
167 return;
168
169 case IPPROTO_ICMPV6:
7e3a42a1 170 if (!onlyproto && pskb_may_pull(skb, nh + offset + 2 - skb->data)) {
1da177e4
LT
171 u8 *icmp = (u8 *)exthdr;
172
1958b856
DM
173 fl6->fl6_icmp_type = icmp[0];
174 fl6->fl6_icmp_code = icmp[1];
1da177e4 175 }
7e1dc7b6 176 fl6->flowi6_proto = nexthdr;
1da177e4
LT
177 return;
178
59fbb3a6 179#if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE)
2ce4272a 180 case IPPROTO_MH:
7e3a42a1 181 if (!onlyproto && pskb_may_pull(skb, nh + offset + 3 - skb->data)) {
2ce4272a
MN
182 struct ip6_mh *mh;
183 mh = (struct ip6_mh *)exthdr;
184
1958b856 185 fl6->fl6_mh_type = mh->ip6mh_type;
2ce4272a 186 }
7e1dc7b6 187 fl6->flowi6_proto = nexthdr;
2ce4272a
MN
188 return;
189#endif
190
1da177e4
LT
191 /* XXX Why are there these headers? */
192 case IPPROTO_AH:
193 case IPPROTO_ESP:
194 case IPPROTO_COMP:
195 default:
1958b856 196 fl6->fl6_ipsec_spi = 0;
7e1dc7b6 197 fl6->flowi6_proto = nexthdr;
1da177e4 198 return;
3ff50b79 199 }
1da177e4
LT
200 }
201}
202
569d3645 203static inline int xfrm6_garbage_collect(struct dst_ops *ops)
1da177e4 204{
d7c7544c
AD
205 struct net *net = container_of(ops, struct net, xfrm.xfrm6_dst_ops);
206
207 xfrm6_policy_afinfo.garbage_collect(net);
fc66f95c 208 return dst_entries_get_fast(ops) > ops->gc_thresh * 2;
1da177e4
LT
209}
210
211static void xfrm6_update_pmtu(struct dst_entry *dst, u32 mtu)
212{
213 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
214 struct dst_entry *path = xdst->route;
215
216 path->ops->update_pmtu(path, mtu);
217}
218
aabc9761
HX
219static void xfrm6_dst_destroy(struct dst_entry *dst)
220{
221 struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
222
223 if (likely(xdst->u.rt6.rt6i_idev))
224 in6_dev_put(xdst->u.rt6.rt6i_idev);
62fa8a84 225 dst_destroy_metrics_generic(dst);
7cc2edb8
DM
226 if (likely(xdst->u.rt6.rt6i_peer))
227 inet_putpeer(xdst->u.rt6.rt6i_peer);
aabc9761
HX
228 xfrm_dst_destroy(xdst);
229}
230
231static void xfrm6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
232 int unregister)
233{
234 struct xfrm_dst *xdst;
235
236 if (!unregister)
237 return;
238
239 xdst = (struct xfrm_dst *)dst;
240 if (xdst->u.rt6.rt6i_idev->dev == dev) {
5a3e55d6 241 struct inet6_dev *loopback_idev =
c346dca1 242 in6_dev_get(dev_net(dev)->loopback_dev);
aabc9761
HX
243 BUG_ON(!loopback_idev);
244
245 do {
246 in6_dev_put(xdst->u.rt6.rt6i_idev);
247 xdst->u.rt6.rt6i_idev = loopback_idev;
248 in6_dev_hold(loopback_idev);
249 xdst = (struct xfrm_dst *)xdst->u.dst.child;
250 } while (xdst->u.dst.xfrm);
251
252 __in6_dev_put(loopback_idev);
253 }
254
255 xfrm_dst_ifdown(dst, dev);
256}
257
1da177e4
LT
258static struct dst_ops xfrm6_dst_ops = {
259 .family = AF_INET6,
09640e63 260 .protocol = cpu_to_be16(ETH_P_IPV6),
1da177e4
LT
261 .gc = xfrm6_garbage_collect,
262 .update_pmtu = xfrm6_update_pmtu,
62fa8a84 263 .cow_metrics = dst_cow_metrics_generic,
aabc9761
HX
264 .destroy = xfrm6_dst_destroy,
265 .ifdown = xfrm6_dst_ifdown,
862b82c6 266 .local_out = __ip6_local_out,
1da177e4 267 .gc_thresh = 1024,
1da177e4
LT
268};
269
270static struct xfrm_policy_afinfo xfrm6_policy_afinfo = {
271 .family = AF_INET6,
1da177e4
LT
272 .dst_ops = &xfrm6_dst_ops,
273 .dst_lookup = xfrm6_dst_lookup,
a1e59abf 274 .get_saddr = xfrm6_get_saddr,
1da177e4 275 .decode_session = _decode_session6,
25ee3286 276 .get_tos = xfrm6_get_tos,
a1b05140 277 .init_path = xfrm6_init_path,
25ee3286 278 .fill_dst = xfrm6_fill_dst,
2774c131 279 .blackhole_route = ip6_blackhole_route,
1da177e4
LT
280};
281
0013caba 282static int __init xfrm6_policy_init(void)
1da177e4 283{
0013caba 284 return xfrm_policy_register_afinfo(&xfrm6_policy_afinfo);
1da177e4
LT
285}
286
287static void xfrm6_policy_fini(void)
288{
289 xfrm_policy_unregister_afinfo(&xfrm6_policy_afinfo);
290}
291
db71789c 292#ifdef CONFIG_SYSCTL
a44a4a00
NH
293static struct ctl_table xfrm6_policy_table[] = {
294 {
a44a4a00 295 .procname = "xfrm6_gc_thresh",
d7c7544c 296 .data = &init_net.xfrm.xfrm6_dst_ops.gc_thresh,
a44a4a00
NH
297 .maxlen = sizeof(int),
298 .mode = 0644,
299 .proc_handler = proc_dointvec,
300 },
301 { }
302};
303
304static struct ctl_table_header *sysctl_hdr;
db71789c 305#endif
a44a4a00 306
0013caba 307int __init xfrm6_init(void)
1da177e4 308{
0013caba 309 int ret;
a33bc5c1 310 unsigned int gc_thresh;
0013caba 311
a33bc5c1
NH
312 /*
313 * We need a good default value for the xfrm6 gc threshold.
314 * In ipv4 we set it to the route hash table size * 8, which
315 * is half the size of the maximaum route cache for ipv4. It
316 * would be good to do the same thing for v6, except the table is
317 * constructed differently here. Here each table for a net namespace
318 * can have FIB_TABLE_HASHSZ entries, so lets go with the same
319 * computation that we used for ipv4 here. Also, lets keep the initial
320 * gc_thresh to a minimum of 1024, since, the ipv6 route cache defaults
321 * to that as a minimum as well
322 */
323 gc_thresh = FIB6_TABLE_HASHSZ * 8;
324 xfrm6_dst_ops.gc_thresh = (gc_thresh < 1024) ? 1024 : gc_thresh;
fc66f95c 325 dst_entries_init(&xfrm6_dst_ops);
d7c7544c
AD
326
327 ret = xfrm6_policy_init();
fc66f95c
ED
328 if (ret) {
329 dst_entries_destroy(&xfrm6_dst_ops);
d7c7544c 330 goto out;
fc66f95c 331 }
d7c7544c
AD
332 ret = xfrm6_state_init();
333 if (ret)
334 goto out_policy;
335
db71789c 336#ifdef CONFIG_SYSCTL
ec8f23ce
EB
337 sysctl_hdr = register_net_sysctl(&init_net, "net/ipv6",
338 xfrm6_policy_table);
db71789c 339#endif
0013caba
DL
340out:
341 return ret;
342out_policy:
343 xfrm6_policy_fini();
344 goto out;
1da177e4
LT
345}
346
347void xfrm6_fini(void)
348{
db71789c 349#ifdef CONFIG_SYSCTL
a44a4a00
NH
350 if (sysctl_hdr)
351 unregister_net_sysctl_table(sysctl_hdr);
db71789c 352#endif
1da177e4
LT
353 //xfrm6_input_fini();
354 xfrm6_policy_fini();
355 xfrm6_state_fini();
fc66f95c 356 dst_entries_destroy(&xfrm6_dst_ops);
1da177e4 357}