net/ipv6/xfrm6_policy.c

   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 defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE)
  24 #include <net/mip6.h>
  25 #endif
  26
  27 static struct dst_ops xfrm6_dst_ops;
  28 static struct xfrm_policy_afinfo xfrm6_policy_afinfo;
  29
  30 static struct dst_entry *xfrm6_dst_lookup(struct net *net, int tos,
  31                                           xfrm_address_t *saddr,
  32                                           xfrm_address_t *daddr)
  33 {
  34         struct flowi fl = {};
  35         struct dst_entry *dst;
  36         int err;
  37
  38         memcpy(&fl.fl6_dst, daddr, sizeof(fl.fl6_dst));
  39         if (saddr)
  40                 memcpy(&fl.fl6_src, saddr, sizeof(fl.fl6_src));
  41
  42         dst = ip6_route_output(net, NULL, &fl);
  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
  53 static 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,
  65                            (struct in6_addr *)&daddr->a6, 0,
  66                            (struct in6_addr *)&saddr->a6);
  67         dst_release(dst);
  68         return 0;
  69 }
  70
  71 static struct dst_entry *
  72 __xfrm6_find_bundle(struct flowi *fl, struct xfrm_policy *policy)
  73 {
  74         struct dst_entry *dst;
  75
  76         /* Still not clear if we should set fl->fl6_{src,dst}... */
  77         read_lock_bh(&policy->lock);
  78         for (dst = policy->bundles; dst; dst = dst->next) {
  79                 struct xfrm_dst *xdst = (struct xfrm_dst*)dst;
  80                 struct in6_addr fl_dst_prefix, fl_src_prefix;
  81
  82                 ipv6_addr_prefix(&fl_dst_prefix,
  83                                  &fl->fl6_dst,
  84                                  xdst->u.rt6.rt6i_dst.plen);
  85                 ipv6_addr_prefix(&fl_src_prefix,
  86                                  &fl->fl6_src,
  87                                  xdst->u.rt6.rt6i_src.plen);
  88                 if (ipv6_addr_equal(&xdst->u.rt6.rt6i_dst.addr, &fl_dst_prefix) &&
  89                     ipv6_addr_equal(&xdst->u.rt6.rt6i_src.addr, &fl_src_prefix) &&
  90                     xfrm_bundle_ok(policy, xdst, fl, AF_INET6,
  91                                    (xdst->u.rt6.rt6i_dst.plen != 128 ||
  92                                     xdst->u.rt6.rt6i_src.plen != 128))) {
  93                         dst_clone(dst);
  94                         break;
  95                 }
  96         }
  97         read_unlock_bh(&policy->lock);
  98         return dst;
  99 }
 100
 101 static int xfrm6_get_tos(struct flowi *fl)
 102 {
 103         return 0;
 104 }
 105
 106 static int xfrm6_init_path(struct xfrm_dst *path, struct dst_entry *dst,
 107                            int nfheader_len)
 108 {
 109         if (dst->ops->family == AF_INET6) {
 110                 struct rt6_info *rt = (struct rt6_info*)dst;
 111                 if (rt->rt6i_node)
 112                         path->path_cookie = rt->rt6i_node->fn_sernum;
 113         }
 114
 115         path->u.rt6.rt6i_nfheader_len = nfheader_len;
 116
 117         return 0;
 118 }
 119
 120 static int xfrm6_fill_dst(struct xfrm_dst *xdst, struct net_device *dev)
 121 {
 122         struct rt6_info *rt = (struct rt6_info*)xdst->route;
 123
 124         xdst->u.dst.dev = dev;
 125         dev_hold(dev);
 126
 127         xdst->u.rt6.rt6i_idev = in6_dev_get(rt->u.dst.dev);
 128         if (!xdst->u.rt6.rt6i_idev)
 129                 return -ENODEV;
 130
 131         /* Sheit... I remember I did this right. Apparently,
 132          * it was magically lost, so this code needs audit */
 133         xdst->u.rt6.rt6i_flags = rt->rt6i_flags & (RTF_ANYCAST |
 134                                                    RTF_LOCAL);
 135         xdst->u.rt6.rt6i_metric = rt->rt6i_metric;
 136         xdst->u.rt6.rt6i_node = rt->rt6i_node;
 137         if (rt->rt6i_node)
 138                 xdst->route_cookie = rt->rt6i_node->fn_sernum;
 139         xdst->u.rt6.rt6i_gateway = rt->rt6i_gateway;
 140         xdst->u.rt6.rt6i_dst = rt->rt6i_dst;
 141         xdst->u.rt6.rt6i_src = rt->rt6i_src;
 142
 143         return 0;
 144 }
 145
 146 static inline void
 147 _decode_session6(struct sk_buff *skb, struct flowi *fl, int reverse)
 148 {
 149         int onlyproto = 0;
 150         u16 offset = skb_network_header_len(skb);
 151         struct ipv6hdr *hdr = ipv6_hdr(skb);
 152         struct ipv6_opt_hdr *exthdr;
 153         const unsigned char *nh = skb_network_header(skb);
 154         u8 nexthdr = nh[IP6CB(skb)->nhoff];
 155
 156         memset(fl, 0, sizeof(struct flowi));
 157         ipv6_addr_copy(&fl->fl6_dst, reverse ? &hdr->saddr : &hdr->daddr);
 158         ipv6_addr_copy(&fl->fl6_src, reverse ? &hdr->daddr : &hdr->saddr);
 159
 160         while (nh + offset + 1 < skb->data ||
 161                pskb_may_pull(skb, nh + offset + 1 - skb->data)) {
 162                 nh = skb_network_header(skb);
 163                 exthdr = (struct ipv6_opt_hdr *)(nh + offset);
 164
 165                 switch (nexthdr) {
 166                 case NEXTHDR_FRAGMENT:
 167                         onlyproto = 1;
 168                 case NEXTHDR_ROUTING:
 169                 case NEXTHDR_HOP:
 170                 case NEXTHDR_DEST:
 171                         offset += ipv6_optlen(exthdr);
 172                         nexthdr = exthdr->nexthdr;
 173                         exthdr = (struct ipv6_opt_hdr *)(nh + offset);
 174                         break;
 175
 176                 case IPPROTO_UDP:
 177                 case IPPROTO_UDPLITE:
 178                 case IPPROTO_TCP:
 179                 case IPPROTO_SCTP:
 180                 case IPPROTO_DCCP:
 181                         if (!onlyproto && (nh + offset + 4 < skb->data ||
 182                              pskb_may_pull(skb, nh + offset + 4 - skb->data))) {
 183                                 __be16 *ports = (__be16 *)exthdr;
 184
 185                                 fl->fl_ip_sport = ports[!!reverse];
 186                                 fl->fl_ip_dport = ports[!reverse];
 187                         }
 188                         fl->proto = nexthdr;
 189                         return;
 190
 191                 case IPPROTO_ICMPV6:
 192                         if (!onlyproto && pskb_may_pull(skb, nh + offset + 2 - skb->data)) {
 193                                 u8 *icmp = (u8 *)exthdr;
 194
 195                                 fl->fl_icmp_type = icmp[0];
 196                                 fl->fl_icmp_code = icmp[1];
 197                         }
 198                         fl->proto = nexthdr;
 199                         return;
 200
 201 #if defined(CONFIG_IPV6_MIP6) || defined(CONFIG_IPV6_MIP6_MODULE)
 202                 case IPPROTO_MH:
 203                         if (!onlyproto && pskb_may_pull(skb, nh + offset + 3 - skb->data)) {
 204                                 struct ip6_mh *mh;
 205                                 mh = (struct ip6_mh *)exthdr;
 206
 207                                 fl->fl_mh_type = mh->ip6mh_type;
 208                         }
 209                         fl->proto = nexthdr;
 210                         return;
 211 #endif
 212
 213                 /* XXX Why are there these headers? */
 214                 case IPPROTO_AH:
 215                 case IPPROTO_ESP:
 216                 case IPPROTO_COMP:
 217                 default:
 218                         fl->fl_ipsec_spi = 0;
 219                         fl->proto = nexthdr;
 220                         return;
 221                 }
 222         }
 223 }
 224
 225 static inline int xfrm6_garbage_collect(struct dst_ops *ops)
 226 {
 227         xfrm6_policy_afinfo.garbage_collect(&init_net);
 228         return (atomic_read(&xfrm6_dst_ops.entries) > xfrm6_dst_ops.gc_thresh*2);
 229 }
 230
 231 static void xfrm6_update_pmtu(struct dst_entry *dst, u32 mtu)
 232 {
 233         struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
 234         struct dst_entry *path = xdst->route;
 235
 236         path->ops->update_pmtu(path, mtu);
 237 }
 238
 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         xfrm_dst_destroy(xdst);
 246 }
 247
 248 static void xfrm6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
 249                              int unregister)
 250 {
 251         struct xfrm_dst *xdst;
 252
 253         if (!unregister)
 254                 return;
 255
 256         xdst = (struct xfrm_dst *)dst;
 257         if (xdst->u.rt6.rt6i_idev->dev == dev) {
 258                 struct inet6_dev *loopback_idev =
 259                         in6_dev_get(dev_net(dev)->loopback_dev);
 260                 BUG_ON(!loopback_idev);
 261
 262                 do {
 263                         in6_dev_put(xdst->u.rt6.rt6i_idev);
 264                         xdst->u.rt6.rt6i_idev = loopback_idev;
 265                         in6_dev_hold(loopback_idev);
 266                         xdst = (struct xfrm_dst *)xdst->u.dst.child;
 267                 } while (xdst->u.dst.xfrm);
 268
 269                 __in6_dev_put(loopback_idev);
 270         }
 271
 272         xfrm_dst_ifdown(dst, dev);
 273 }
 274
 275 static struct dst_ops xfrm6_dst_ops = {
 276         .family =               AF_INET6,
 277         .protocol =             cpu_to_be16(ETH_P_IPV6),
 278         .gc =                   xfrm6_garbage_collect,
 279         .update_pmtu =          xfrm6_update_pmtu,
 280         .destroy =              xfrm6_dst_destroy,
 281         .ifdown =               xfrm6_dst_ifdown,
 282         .local_out =            __ip6_local_out,
 283         .gc_thresh =            1024,
 284         .entries =              ATOMIC_INIT(0),
 285 };
 286
 287 static struct xfrm_policy_afinfo xfrm6_policy_afinfo = {
 288         .family =               AF_INET6,
 289         .dst_ops =              &xfrm6_dst_ops,
 290         .dst_lookup =           xfrm6_dst_lookup,
 291         .get_saddr =            xfrm6_get_saddr,
 292         .find_bundle =          __xfrm6_find_bundle,
 293         .decode_session =       _decode_session6,
 294         .get_tos =              xfrm6_get_tos,
 295         .init_path =            xfrm6_init_path,
 296         .fill_dst =             xfrm6_fill_dst,
 297 };
 298
 299 static int __init xfrm6_policy_init(void)
 300 {
 301         return xfrm_policy_register_afinfo(&xfrm6_policy_afinfo);
 302 }
 303
 304 static void xfrm6_policy_fini(void)
 305 {
 306         xfrm_policy_unregister_afinfo(&xfrm6_policy_afinfo);
 307 }
 308
 309 #ifdef CONFIG_SYSCTL
 310 static struct ctl_table xfrm6_policy_table[] = {
 311         {
 312                 .procname       = "xfrm6_gc_thresh",
 313                 .data           = &xfrm6_dst_ops.gc_thresh,
 314                 .maxlen         = sizeof(int),
 315                 .mode           = 0644,
 316                 .proc_handler   = proc_dointvec,
 317         },
 318         { }
 319 };
 320
 321 static struct ctl_table_header *sysctl_hdr;
 322 #endif
 323
 324 int __init xfrm6_init(void)
 325 {
 326         int ret;
 327         unsigned int gc_thresh;
 328
 329         ret = xfrm6_policy_init();
 330         if (ret)
 331                 goto out;
 332
 333         ret = xfrm6_state_init();
 334         if (ret)
 335                 goto out_policy;
 336         /*
 337          * We need a good default value for the xfrm6 gc threshold.
 338          * In ipv4 we set it to the route hash table size * 8, which
 339          * is half the size of the maximaum route cache for ipv4.  It
 340          * would be good to do the same thing for v6, except the table is
 341          * constructed differently here.  Here each table for a net namespace
 342          * can have FIB_TABLE_HASHSZ entries, so lets go with the same
 343          * computation that we used for ipv4 here.  Also, lets keep the initial
 344          * gc_thresh to a minimum of 1024, since, the ipv6 route cache defaults
 345          * to that as a minimum as well
 346          */
 347         gc_thresh = FIB6_TABLE_HASHSZ * 8;
 348         xfrm6_dst_ops.gc_thresh = (gc_thresh < 1024) ? 1024 : gc_thresh;
 349 #ifdef CONFIG_SYSCTL
 350         sysctl_hdr = register_net_sysctl_table(&init_net, net_ipv6_ctl_path,
 351                                                 xfrm6_policy_table);
 352 #endif
 353 out:
 354         return ret;
 355 out_policy:
 356         xfrm6_policy_fini();
 357         goto out;
 358 }
 359
 360 void xfrm6_fini(void)
 361 {
 362 #ifdef CONFIG_SYSCTL
 363         if (sysctl_hdr)
 364                 unregister_net_sysctl_table(sysctl_hdr);
 365 #endif
 366         //xfrm6_input_fini();
 367         xfrm6_policy_fini();
 368         xfrm6_state_fini();
 369 }