zebra/rt_socket.c

   1 /*
   2  * Kernel routing table updates by routing socket.
   3  * Copyright (C) 1997, 98 Kunihiro Ishiguro
   4  *
   5  * This file is part of GNU Zebra.
   6  *
   7  * GNU Zebra is free software; you can redistribute it and/or modify it
   8  * under the terms of the GNU General Public License as published by the
   9  * Free Software Foundation; either version 2, or (at your option) any
  10  * later version.
  11  *
  12  * GNU Zebra is distributed in the hope that it will be useful, but
  13  * WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  15  * General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU General Public License along
  18  * with this program; see the file COPYING; if not, write to the Free Software
  19  * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
  20  */
  21
  22 #include <zebra.h>
  23
  24 #ifndef HAVE_NETLINK
  25
  26 #ifdef __OpenBSD__
  27 #include <netmpls/mpls.h>
  28 #endif
  29
  30 #include "if.h"
  31 #include "prefix.h"
  32 #include "sockunion.h"
  33 #include "log.h"
  34 #include "privs.h"
  35 #include "vxlan.h"
  36
  37 #include "zebra/debug.h"
  38 #include "zebra/rib.h"
  39 #include "zebra/rt.h"
  40 #include "zebra/kernel_socket.h"
  41 #include "zebra/zebra_mpls.h"
  42
  43 extern struct zebra_privs_t zserv_privs;
  44
  45 #ifdef HAVE_STRUCT_SOCKADDR_IN_SIN_LEN
  46 /* Adjust netmask socket length. Return value is a adjusted sin_len
  47    value. */
  48 static int sin_masklen(struct in_addr mask)
  49 {
  50         char *p, *lim;
  51         int len;
  52         struct sockaddr_in sin;
  53
  54         if (mask.s_addr == 0)
  55                 return sizeof(long);
  56
  57         sin.sin_addr = mask;
  58         len = sizeof(struct sockaddr_in);
  59
  60         lim = (char *)&sin.sin_addr;
  61         p = lim + sizeof(sin.sin_addr);
  62
  63         while (*--p == 0 && p >= lim)
  64                 len--;
  65         return len;
  66 }
  67 #endif /* HAVE_STRUCT_SOCKADDR_IN_SIN_LEN */
  68
  69 #ifdef __OpenBSD__
  70 static int kernel_rtm_add_labels(struct mpls_label_stack *nh_label,
  71                                  struct sockaddr_mpls *smpls)
  72 {
  73         if (nh_label->num_labels > 1) {
  74                 zlog_warn(
  75                         "%s: can't push %u labels at "
  76                         "once (maximum is 1)",
  77                         __func__, nh_label->num_labels);
  78                 return -1;
  79         }
  80
  81         memset(smpls, 0, sizeof(*smpls));
  82         smpls->smpls_len = sizeof(*smpls);
  83         smpls->smpls_family = AF_MPLS;
  84         smpls->smpls_label = htonl(nh_label->label[0] << MPLS_LABEL_OFFSET);
  85
  86         return 0;
  87 }
  88 #endif
  89
  90 /* Interface between zebra message and rtm message. */
  91 static int kernel_rtm_ipv4(int cmd, struct prefix *p, struct route_entry *re)
  92
  93 {
  94         struct sockaddr_in *mask = NULL;
  95         struct sockaddr_in sin_dest, sin_mask, sin_gate;
  96 #ifdef __OpenBSD__
  97         struct sockaddr_mpls smpls;
  98 #endif
  99         union sockunion *smplsp = NULL;
 100         struct nexthop *nexthop;
 101         int nexthop_num = 0;
 102         ifindex_t ifindex = 0;
 103         int gate = 0;
 104         int error;
 105         char prefix_buf[PREFIX_STRLEN];
 106         enum blackhole_type bh_type = BLACKHOLE_UNSPEC;
 107
 108         if (IS_ZEBRA_DEBUG_RIB)
 109                 prefix2str(p, prefix_buf, sizeof(prefix_buf));
 110         memset(&sin_dest, 0, sizeof(struct sockaddr_in));
 111         sin_dest.sin_family = AF_INET;
 112 #ifdef HAVE_STRUCT_SOCKADDR_IN_SIN_LEN
 113         sin_dest.sin_len = sizeof(struct sockaddr_in);
 114 #endif /* HAVE_STRUCT_SOCKADDR_IN_SIN_LEN */
 115         sin_dest.sin_addr = p->u.prefix4;
 116
 117         memset(&sin_mask, 0, sizeof(struct sockaddr_in));
 118
 119         memset(&sin_gate, 0, sizeof(struct sockaddr_in));
 120         sin_gate.sin_family = AF_INET;
 121 #ifdef HAVE_STRUCT_SOCKADDR_IN_SIN_LEN
 122         sin_gate.sin_len = sizeof(struct sockaddr_in);
 123 #endif /* HAVE_STRUCT_SOCKADDR_IN_SIN_LEN */
 124
 125         /* Make gateway. */
 126         for (ALL_NEXTHOPS(re->ng, nexthop)) {
 127                 if (CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_RECURSIVE))
 128                         continue;
 129
 130                 gate = 0;
 131                 char gate_buf[INET_ADDRSTRLEN] = "NULL";
 132
 133                 /*
 134                  * XXX We need to refrain from kernel operations in some cases,
 135                  * but this if statement seems overly cautious - what about
 136                  * other than ADD and DELETE?
 137                  */
 138                 if ((cmd == RTM_ADD && NEXTHOP_IS_ACTIVE(nexthop->flags))
 139                     || (cmd == RTM_DELETE
 140                         && CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_FIB))) {
 141                         if (nexthop->type == NEXTHOP_TYPE_IPV4
 142                             || nexthop->type == NEXTHOP_TYPE_IPV4_IFINDEX) {
 143                                 sin_gate.sin_addr = nexthop->gate.ipv4;
 144                                 gate = 1;
 145                         }
 146                         if (nexthop->type == NEXTHOP_TYPE_IFINDEX
 147                             || nexthop->type == NEXTHOP_TYPE_IPV4_IFINDEX)
 148                                 ifindex = nexthop->ifindex;
 149                         if (nexthop->type == NEXTHOP_TYPE_BLACKHOLE) {
 150                                 struct in_addr loopback;
 151                                 loopback.s_addr = htonl(INADDR_LOOPBACK);
 152                                 sin_gate.sin_addr = loopback;
 153                                 bh_type = nexthop->bh_type;
 154                                 gate = 1;
 155                         }
 156
 157                         if (gate && p->prefixlen == 32)
 158                                 mask = NULL;
 159                         else {
 160                                 masklen2ip(p->prefixlen, &sin_mask.sin_addr);
 161                                 sin_mask.sin_family = AF_INET;
 162 #ifdef HAVE_STRUCT_SOCKADDR_IN_SIN_LEN
 163                                 sin_mask.sin_len =
 164                                         sin_masklen(sin_mask.sin_addr);
 165 #endif /* HAVE_STRUCT_SOCKADDR_IN_SIN_LEN */
 166                                 mask = &sin_mask;
 167                         }
 168
 169 #ifdef __OpenBSD__
 170                         if (nexthop->nh_label
 171                             && !kernel_rtm_add_labels(nexthop->nh_label,
 172                                                       &smpls))
 173                                 continue;
 174                         smplsp = (union sockunion *)&smpls;
 175 #endif
 176
 177                         error = rtm_write(cmd, (union sockunion *)&sin_dest,
 178                                           (union sockunion *)mask,
 179                                           gate ? (union sockunion *)&sin_gate
 180                                                : NULL,
 181                                           smplsp, ifindex, bh_type, re->metric);
 182
 183                         if (IS_ZEBRA_DEBUG_RIB) {
 184                                 if (!gate) {
 185                                         zlog_debug(
 186                                                 "%s: %s: attention! gate not found for re %p",
 187                                                 __func__, prefix_buf, re);
 188                                         route_entry_dump(p, NULL, re);
 189                                 } else
 190                                         inet_ntop(AF_INET, &sin_gate.sin_addr,
 191                                                   gate_buf, INET_ADDRSTRLEN);
 192                         }
 193
 194                         switch (error) {
 195                         /* We only flag nexthops as being in FIB if rtm_write()
 196                          * did its work. */
 197                         case ZEBRA_ERR_NOERROR:
 198                                 nexthop_num++;
 199                                 if (IS_ZEBRA_DEBUG_RIB)
 200                                         zlog_debug(
 201                                                 "%s: %s: successfully did NH %s",
 202                                                 __func__, prefix_buf, gate_buf);
 203                                 break;
 204
 205                         /* The only valid case for this error is kernel's
 206                          * failure to install
 207                          * a multipath route, which is common for FreeBSD. This
 208                          * should be
 209                          * ignored silently, but logged as an error otherwise.
 210                          */
 211                         case ZEBRA_ERR_RTEXIST:
 212                                 if (cmd != RTM_ADD)
 213                                         zlog_err(
 214                                                 "%s: rtm_write() returned %d for command %d",
 215                                                 __func__, error, cmd);
 216                                 continue;
 217                                 break;
 218
 219                         /* Given that our NEXTHOP_FLAG_FIB matches real kernel
 220                          * FIB, it isn't
 221                          * normal to get any other messages in ANY case.
 222                          */
 223                         case ZEBRA_ERR_RTNOEXIST:
 224                         case ZEBRA_ERR_RTUNREACH:
 225                         default:
 226                                 zlog_err(
 227                                         "%s: %s: rtm_write() unexpectedly returned %d for command %s",
 228                                         __func__,
 229                                         prefix2str(p, prefix_buf,
 230                                                    sizeof(prefix_buf)),
 231                                         error,
 232                                         lookup_msg(rtm_type_str, cmd, NULL));
 233                                 break;
 234                         }
 235                 } /* if (cmd and flags make sense) */
 236                 else if (IS_ZEBRA_DEBUG_RIB)
 237                         zlog_debug("%s: odd command %s for flags %d", __func__,
 238                                    lookup_msg(rtm_type_str, cmd, NULL),
 239                                    nexthop->flags);
 240         } /* for (ALL_NEXTHOPS(...))*/
 241
 242         /* If there was no useful nexthop, then complain. */
 243         if (nexthop_num == 0 && IS_ZEBRA_DEBUG_KERNEL)
 244                 zlog_debug("%s: No useful nexthops were found in RIB entry %p",
 245                            __func__, re);
 246
 247         return 0; /*XXX*/
 248 }
 249
 250 #ifdef SIN6_LEN
 251 /* Calculate sin6_len value for netmask socket value. */
 252 static int sin6_masklen(struct in6_addr mask)
 253 {
 254         struct sockaddr_in6 sin6;
 255         char *p, *lim;
 256         int len;
 257
 258         if (IN6_IS_ADDR_UNSPECIFIED(&mask))
 259                 return sizeof(long);
 260
 261         sin6.sin6_addr = mask;
 262         len = sizeof(struct sockaddr_in6);
 263
 264         lim = (char *)&sin6.sin6_addr;
 265         p = lim + sizeof(sin6.sin6_addr);
 266
 267         while (*--p == 0 && p >= lim)
 268                 len--;
 269
 270         return len;
 271 }
 272 #endif /* SIN6_LEN */
 273
 274 /* Interface between zebra message and rtm message. */
 275 static int kernel_rtm_ipv6(int cmd, struct prefix *p, struct route_entry *re)
 276 {
 277         struct sockaddr_in6 *mask;
 278         struct sockaddr_in6 sin_dest, sin_mask, sin_gate;
 279 #ifdef __OpenBSD__
 280         struct sockaddr_mpls smpls;
 281 #endif
 282         union sockunion *smplsp = NULL;
 283         struct nexthop *nexthop;
 284         int nexthop_num = 0;
 285         ifindex_t ifindex = 0;
 286         int gate = 0;
 287         int error;
 288         enum blackhole_type bh_type = BLACKHOLE_UNSPEC;
 289
 290         memset(&sin_dest, 0, sizeof(struct sockaddr_in6));
 291         sin_dest.sin6_family = AF_INET6;
 292 #ifdef SIN6_LEN
 293         sin_dest.sin6_len = sizeof(struct sockaddr_in6);
 294 #endif /* SIN6_LEN */
 295         sin_dest.sin6_addr = p->u.prefix6;
 296
 297         memset(&sin_mask, 0, sizeof(struct sockaddr_in6));
 298
 299         memset(&sin_gate, 0, sizeof(struct sockaddr_in6));
 300         sin_gate.sin6_family = AF_INET6;
 301 #ifdef HAVE_STRUCT_SOCKADDR_IN_SIN_LEN
 302         sin_gate.sin6_len = sizeof(struct sockaddr_in6);
 303 #endif /* HAVE_STRUCT_SOCKADDR_IN_SIN_LEN */
 304
 305         /* Make gateway. */
 306         for (ALL_NEXTHOPS(re->ng, nexthop)) {
 307                 if (CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_RECURSIVE))
 308                         continue;
 309
 310                 gate = 0;
 311
 312                 if ((cmd == RTM_ADD && NEXTHOP_IS_ACTIVE(nexthop->flags))
 313                     || (cmd == RTM_DELETE)) {
 314                         if (nexthop->type == NEXTHOP_TYPE_IPV6
 315                             || nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX) {
 316                                 sin_gate.sin6_addr = nexthop->gate.ipv6;
 317                                 gate = 1;
 318                         }
 319                         if (nexthop->type == NEXTHOP_TYPE_IFINDEX
 320                             || nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX)
 321                                 ifindex = nexthop->ifindex;
 322
 323                         if (nexthop->type == NEXTHOP_TYPE_BLACKHOLE)
 324                                 bh_type = nexthop->bh_type;
 325                 }
 326
 327 /* Under kame set interface index to link local address. */
 328 #ifdef KAME
 329
 330 #define SET_IN6_LINKLOCAL_IFINDEX(a, i)                                        \
 331         do {                                                                   \
 332                 (a).s6_addr[2] = ((i) >> 8) & 0xff;                            \
 333                 (a).s6_addr[3] = (i)&0xff;                                     \
 334         } while (0)
 335
 336                 if (gate && IN6_IS_ADDR_LINKLOCAL(&sin_gate.sin6_addr))
 337                         SET_IN6_LINKLOCAL_IFINDEX(sin_gate.sin6_addr, ifindex);
 338 #endif /* KAME */
 339
 340                 if (gate && p->prefixlen == 128)
 341                         mask = NULL;
 342                 else {
 343                         masklen2ip6(p->prefixlen, &sin_mask.sin6_addr);
 344                         sin_mask.sin6_family = AF_INET6;
 345 #ifdef SIN6_LEN
 346                         sin_mask.sin6_len = sin6_masklen(sin_mask.sin6_addr);
 347 #endif /* SIN6_LEN */
 348                         mask = &sin_mask;
 349                 }
 350
 351 #ifdef __OpenBSD__
 352                 if (nexthop->nh_label
 353                     && !kernel_rtm_add_labels(nexthop->nh_label, &smpls))
 354                         continue;
 355                 smplsp = (union sockunion *)&smpls;
 356 #endif
 357
 358                 error = rtm_write(cmd, (union sockunion *)&sin_dest,
 359                                   (union sockunion *)mask,
 360                                   gate ? (union sockunion *)&sin_gate : NULL,
 361                                   smplsp, ifindex, bh_type, re->metric);
 362                 (void)error;
 363
 364                 nexthop_num++;
 365         }
 366
 367         /* If there is no useful nexthop then return. */
 368         if (nexthop_num == 0) {
 369                 if (IS_ZEBRA_DEBUG_KERNEL)
 370                         zlog_debug("kernel_rtm_ipv6(): No useful nexthop.");
 371                 return 0;
 372         }
 373
 374         return 0; /*XXX*/
 375 }
 376
 377 static int kernel_rtm(int cmd, struct prefix *p, struct route_entry *re)
 378 {
 379         switch (PREFIX_FAMILY(p)) {
 380         case AF_INET:
 381                 return kernel_rtm_ipv4(cmd, p, re);
 382         case AF_INET6:
 383                 return kernel_rtm_ipv6(cmd, p, re);
 384         }
 385         return 0;
 386 }
 387
 388 enum dp_req_result kernel_route_rib(struct route_node *rn,
 389                                     const struct prefix *p,
 390                                     const struct prefix *src_p,
 391                                     struct route_entry *old,
 392                                     struct route_entry *new)
 393 {
 394         int route = 0;
 395
 396         if (src_p && src_p->prefixlen) {
 397                 zlog_err("route add: IPv6 sourcedest routes unsupported!");
 398                 return DP_REQUEST_FAILURE;
 399         }
 400
 401         if (zserv_privs.change(ZPRIVS_RAISE))
 402                 zlog_err("Can't raise privileges");
 403
 404         if (old)
 405                 route |= kernel_rtm(RTM_DELETE, p, old);
 406
 407         if (new)
 408                 route |= kernel_rtm(RTM_ADD, p, new);
 409
 410         if (zserv_privs.change(ZPRIVS_LOWER))
 411                 zlog_err("Can't lower privileges");
 412
 413         if (new) {
 414                 kernel_route_rib_pass_fail(
 415                         rn, p, new,
 416                         (!route) ? DP_INSTALL_SUCCESS
 417                                  : DP_INSTALL_FAILURE);
 418         } else {
 419                 kernel_route_rib_pass_fail(rn, p, old,
 420                                            (!route)
 421                                                    ? DP_DELETE_SUCCESS
 422                                                    : DP_DELETE_FAILURE);
 423         }
 424
 425         return DP_REQUEST_SUCCESS;
 426 }
 427
 428 int kernel_neigh_update(int add, int ifindex, uint32_t addr, char *lla,
 429                         int llalen, ns_id_t ns_id)
 430 {
 431         /* TODO */
 432         return 0;
 433 }
 434
 435 extern int kernel_get_ipmr_sg_stats(struct zebra_vrf *zvrf, void *mroute)
 436 {
 437         return 0;
 438 }
 439
 440 int kernel_add_vtep(vni_t vni, struct interface *ifp, struct in_addr *vtep_ip)
 441 {
 442         return 0;
 443 }
 444
 445 int kernel_del_vtep(vni_t vni, struct interface *ifp, struct in_addr *vtep_ip)
 446 {
 447         return 0;
 448 }
 449
 450 int kernel_add_mac(struct interface *ifp, vlanid_t vid, struct ethaddr *mac,
 451                    struct in_addr vtep_ip, uint8_t sticky)
 452 {
 453         return 0;
 454 }
 455
 456 int kernel_del_mac(struct interface *ifp, vlanid_t vid, struct ethaddr *mac,
 457                    struct in_addr vtep_ip, int local)
 458 {
 459         return 0;
 460 }
 461
 462 int kernel_add_neigh(struct interface *ifp, struct ipaddr *ip,
 463                      struct ethaddr *mac)
 464 {
 465         return 0;
 466 }
 467
 468 int kernel_del_neigh(struct interface *ifp, struct ipaddr *ip)
 469 {
 470         return 0;
 471 }
 472
 473 extern int kernel_interface_set_master(struct interface *master,
 474                                        struct interface *slave)
 475 {
 476         return 0;
 477 }
 478
 479 uint32_t kernel_get_speed(struct interface *ifp)
 480 {
 481         return ifp->speed;
 482 }
 483
 484 #endif /* !HAVE_NETLINK */