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 #include "lib_errors.h"
  37
  38 #include "zebra/debug.h"
  39 #include "zebra/rib.h"
  40 #include "zebra/rt.h"
  41 #include "zebra/kernel_socket.h"
  42 #include "zebra/zebra_mpls.h"
  43 #include "zebra/zebra_errors.h"
  44
  45 extern struct zebra_privs_t zserv_privs;
  46
  47 #ifdef HAVE_STRUCT_SOCKADDR_IN_SIN_LEN
  48 /* Adjust netmask socket length. Return value is a adjusted sin_len
  49    value. */
  50 static int sin_masklen(struct in_addr mask)
  51 {
  52         char *p, *lim;
  53         int len;
  54         struct sockaddr_in sin;
  55
  56         if (mask.s_addr == 0)
  57                 return sizeof(long);
  58
  59         sin.sin_addr = mask;
  60         len = sizeof(struct sockaddr_in);
  61
  62         lim = (char *)&sin.sin_addr;
  63         p = lim + sizeof(sin.sin_addr);
  64
  65         while (*--p == 0 && p >= lim)
  66                 len--;
  67         return len;
  68 }
  69 #endif /* HAVE_STRUCT_SOCKADDR_IN_SIN_LEN */
  70
  71 #ifdef __OpenBSD__
  72 static int kernel_rtm_add_labels(struct mpls_label_stack *nh_label,
  73                                  struct sockaddr_mpls *smpls)
  74 {
  75         if (nh_label->num_labels > 1) {
  76                 flog_warn(EC_ZEBRA_MAX_LABELS_PUSH,
  77                           "%s: can't push %u labels at "
  78                           "once (maximum is 1)",
  79                           __func__, nh_label->num_labels);
  80                 return -1;
  81         }
  82
  83         memset(smpls, 0, sizeof(*smpls));
  84         smpls->smpls_len = sizeof(*smpls);
  85         smpls->smpls_family = AF_MPLS;
  86         smpls->smpls_label = htonl(nh_label->label[0] << MPLS_LABEL_OFFSET);
  87
  88         return 0;
  89 }
  90 #endif
  91
  92 /* Interface between zebra message and rtm message. */
  93 static int kernel_rtm_ipv4(int cmd, const struct prefix *p,
  94                            struct route_entry *re)
  95
  96 {
  97         struct sockaddr_in *mask = NULL;
  98         struct sockaddr_in sin_dest, sin_mask, sin_gate;
  99 #ifdef __OpenBSD__
 100         struct sockaddr_mpls smpls;
 101 #endif
 102         union sockunion *smplsp = NULL;
 103         struct nexthop *nexthop;
 104         int nexthop_num = 0;
 105         ifindex_t ifindex = 0;
 106         int gate = 0;
 107         int error;
 108         char prefix_buf[PREFIX_STRLEN];
 109         enum blackhole_type bh_type = BLACKHOLE_UNSPEC;
 110
 111         if (IS_ZEBRA_DEBUG_RIB)
 112                 prefix2str(p, prefix_buf, sizeof(prefix_buf));
 113         memset(&sin_dest, 0, sizeof(struct sockaddr_in));
 114         sin_dest.sin_family = AF_INET;
 115 #ifdef HAVE_STRUCT_SOCKADDR_IN_SIN_LEN
 116         sin_dest.sin_len = sizeof(struct sockaddr_in);
 117 #endif /* HAVE_STRUCT_SOCKADDR_IN_SIN_LEN */
 118         sin_dest.sin_addr = p->u.prefix4;
 119
 120         memset(&sin_mask, 0, sizeof(struct sockaddr_in));
 121
 122         memset(&sin_gate, 0, sizeof(struct sockaddr_in));
 123         sin_gate.sin_family = AF_INET;
 124 #ifdef HAVE_STRUCT_SOCKADDR_IN_SIN_LEN
 125         sin_gate.sin_len = sizeof(struct sockaddr_in);
 126 #endif /* HAVE_STRUCT_SOCKADDR_IN_SIN_LEN */
 127
 128         /* Make gateway. */
 129         for (ALL_NEXTHOPS(re->ng, nexthop)) {
 130                 if (CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_RECURSIVE))
 131                         continue;
 132
 133                 gate = 0;
 134                 char gate_buf[INET_ADDRSTRLEN] = "NULL";
 135
 136                 /*
 137                  * XXX We need to refrain from kernel operations in some cases,
 138                  * but this if statement seems overly cautious - what about
 139                  * other than ADD and DELETE?
 140                  */
 141                 if ((cmd == RTM_ADD && NEXTHOP_IS_ACTIVE(nexthop->flags))
 142                     || (cmd == RTM_DELETE
 143                         && CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_FIB))) {
 144                         if (nexthop->type == NEXTHOP_TYPE_IPV4
 145                             || nexthop->type == NEXTHOP_TYPE_IPV4_IFINDEX) {
 146                                 sin_gate.sin_addr = nexthop->gate.ipv4;
 147                                 gate = 1;
 148                         }
 149                         if (nexthop->type == NEXTHOP_TYPE_IFINDEX
 150                             || nexthop->type == NEXTHOP_TYPE_IPV4_IFINDEX)
 151                                 ifindex = nexthop->ifindex;
 152                         if (nexthop->type == NEXTHOP_TYPE_BLACKHOLE) {
 153                                 struct in_addr loopback;
 154                                 loopback.s_addr = htonl(INADDR_LOOPBACK);
 155                                 sin_gate.sin_addr = loopback;
 156                                 bh_type = nexthop->bh_type;
 157                                 gate = 1;
 158                         }
 159
 160                         if (gate && p->prefixlen == 32)
 161                                 mask = NULL;
 162                         else {
 163                                 masklen2ip(p->prefixlen, &sin_mask.sin_addr);
 164                                 sin_mask.sin_family = AF_INET;
 165 #ifdef HAVE_STRUCT_SOCKADDR_IN_SIN_LEN
 166                                 sin_mask.sin_len =
 167                                         sin_masklen(sin_mask.sin_addr);
 168 #endif /* HAVE_STRUCT_SOCKADDR_IN_SIN_LEN */
 169                                 mask = &sin_mask;
 170                         }
 171
 172 #ifdef __OpenBSD__
 173                         if (nexthop->nh_label
 174                             && !kernel_rtm_add_labels(nexthop->nh_label,
 175                                                       &smpls))
 176                                 continue;
 177                         smplsp = (union sockunion *)&smpls;
 178 #endif
 179
 180                         error = rtm_write(cmd, (union sockunion *)&sin_dest,
 181                                           (union sockunion *)mask,
 182                                           gate ? (union sockunion *)&sin_gate
 183                                                : NULL,
 184                                           smplsp, ifindex, bh_type, re->metric);
 185
 186                         if (IS_ZEBRA_DEBUG_RIB) {
 187                                 if (!gate) {
 188                                         zlog_debug(
 189                                                 "%s: %s: attention! gate not found for re %p",
 190                                                 __func__, prefix_buf, re);
 191                                         route_entry_dump(p, NULL, re);
 192                                 } else
 193                                         inet_ntop(AF_INET, &sin_gate.sin_addr,
 194                                                   gate_buf, INET_ADDRSTRLEN);
 195                         }
 196
 197                         switch (error) {
 198                         /* We only flag nexthops as being in FIB if rtm_write()
 199                          * did its work. */
 200                         case ZEBRA_ERR_NOERROR:
 201                                 nexthop_num++;
 202                                 if (IS_ZEBRA_DEBUG_RIB)
 203                                         zlog_debug(
 204                                                 "%s: %s: successfully did NH %s",
 205                                                 __func__, prefix_buf, gate_buf);
 206                                 break;
 207
 208                         /* The only valid case for this error is kernel's
 209                          * failure to install
 210                          * a multipath route, which is common for FreeBSD. This
 211                          * should be
 212                          * ignored silently, but logged as an error otherwise.
 213                          */
 214                         case ZEBRA_ERR_RTEXIST:
 215                                 if (cmd != RTM_ADD)
 216                                         flog_err(
 217                                                 EC_LIB_SYSTEM_CALL,
 218                                                 "%s: rtm_write() returned %d for command %d",
 219                                                 __func__, error, cmd);
 220                                 continue;
 221                                 break;
 222
 223                         /* Given that our NEXTHOP_FLAG_FIB matches real kernel
 224                          * FIB, it isn't
 225                          * normal to get any other messages in ANY case.
 226                          */
 227                         case ZEBRA_ERR_RTNOEXIST:
 228                         case ZEBRA_ERR_RTUNREACH:
 229                         default:
 230                                 flog_err(
 231                                         EC_LIB_SYSTEM_CALL,
 232                                         "%s: %s: rtm_write() unexpectedly returned %d for command %s",
 233                                         __func__,
 234                                         prefix2str(p, prefix_buf,
 235                                                    sizeof(prefix_buf)),
 236                                         error,
 237                                         lookup_msg(rtm_type_str, cmd, NULL));
 238                                 break;
 239                         }
 240                 } /* if (cmd and flags make sense) */
 241                 else if (IS_ZEBRA_DEBUG_RIB)
 242                         zlog_debug("%s: odd command %s for flags %d", __func__,
 243                                    lookup_msg(rtm_type_str, cmd, NULL),
 244                                    nexthop->flags);
 245         } /* for (ALL_NEXTHOPS(...))*/
 246
 247         /* If there was no useful nexthop, then complain. */
 248         if (nexthop_num == 0 && IS_ZEBRA_DEBUG_KERNEL)
 249                 zlog_debug("%s: No useful nexthops were found in RIB entry %p",
 250                            __func__, re);
 251
 252         return 0; /*XXX*/
 253 }
 254
 255 #ifdef SIN6_LEN
 256 /* Calculate sin6_len value for netmask socket value. */
 257 static int sin6_masklen(struct in6_addr mask)
 258 {
 259         struct sockaddr_in6 sin6;
 260         char *p, *lim;
 261         int len;
 262
 263         if (IN6_IS_ADDR_UNSPECIFIED(&mask))
 264                 return sizeof(long);
 265
 266         sin6.sin6_addr = mask;
 267         len = sizeof(struct sockaddr_in6);
 268
 269         lim = (char *)&sin6.sin6_addr;
 270         p = lim + sizeof(sin6.sin6_addr);
 271
 272         while (*--p == 0 && p >= lim)
 273                 len--;
 274
 275         return len;
 276 }
 277 #endif /* SIN6_LEN */
 278
 279 /* Interface between zebra message and rtm message. */
 280 static int kernel_rtm_ipv6(int cmd, const struct prefix *p,
 281                            struct route_entry *re)
 282 {
 283         struct sockaddr_in6 *mask;
 284         struct sockaddr_in6 sin_dest, sin_mask, sin_gate;
 285 #ifdef __OpenBSD__
 286         struct sockaddr_mpls smpls;
 287 #endif
 288         union sockunion *smplsp = NULL;
 289         struct nexthop *nexthop;
 290         int nexthop_num = 0;
 291         ifindex_t ifindex = 0;
 292         int gate = 0;
 293         int error;
 294         enum blackhole_type bh_type = BLACKHOLE_UNSPEC;
 295
 296         memset(&sin_dest, 0, sizeof(struct sockaddr_in6));
 297         sin_dest.sin6_family = AF_INET6;
 298 #ifdef SIN6_LEN
 299         sin_dest.sin6_len = sizeof(struct sockaddr_in6);
 300 #endif /* SIN6_LEN */
 301         sin_dest.sin6_addr = p->u.prefix6;
 302
 303         memset(&sin_mask, 0, sizeof(struct sockaddr_in6));
 304
 305         memset(&sin_gate, 0, sizeof(struct sockaddr_in6));
 306         sin_gate.sin6_family = AF_INET6;
 307 #ifdef HAVE_STRUCT_SOCKADDR_IN_SIN_LEN
 308         sin_gate.sin6_len = sizeof(struct sockaddr_in6);
 309 #endif /* HAVE_STRUCT_SOCKADDR_IN_SIN_LEN */
 310
 311         /* Make gateway. */
 312         for (ALL_NEXTHOPS(re->ng, nexthop)) {
 313                 if (CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_RECURSIVE))
 314                         continue;
 315
 316                 gate = 0;
 317
 318                 if ((cmd == RTM_ADD && NEXTHOP_IS_ACTIVE(nexthop->flags))
 319                     || (cmd == RTM_DELETE)) {
 320                         if (nexthop->type == NEXTHOP_TYPE_IPV6
 321                             || nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX) {
 322                                 sin_gate.sin6_addr = nexthop->gate.ipv6;
 323                                 gate = 1;
 324                         }
 325                         if (nexthop->type == NEXTHOP_TYPE_IFINDEX
 326                             || nexthop->type == NEXTHOP_TYPE_IPV6_IFINDEX)
 327                                 ifindex = nexthop->ifindex;
 328
 329                         if (nexthop->type == NEXTHOP_TYPE_BLACKHOLE)
 330                                 bh_type = nexthop->bh_type;
 331                 }
 332
 333 /* Under kame set interface index to link local address. */
 334 #ifdef KAME
 335
 336 #define SET_IN6_LINKLOCAL_IFINDEX(a, i)                                        \
 337         do {                                                                   \
 338                 (a).s6_addr[2] = ((i) >> 8) & 0xff;                            \
 339                 (a).s6_addr[3] = (i)&0xff;                                     \
 340         } while (0)
 341
 342                 if (gate && IN6_IS_ADDR_LINKLOCAL(&sin_gate.sin6_addr))
 343                         SET_IN6_LINKLOCAL_IFINDEX(sin_gate.sin6_addr, ifindex);
 344 #endif /* KAME */
 345
 346                 if (gate && p->prefixlen == 128)
 347                         mask = NULL;
 348                 else {
 349                         masklen2ip6(p->prefixlen, &sin_mask.sin6_addr);
 350                         sin_mask.sin6_family = AF_INET6;
 351 #ifdef SIN6_LEN
 352                         sin_mask.sin6_len = sin6_masklen(sin_mask.sin6_addr);
 353 #endif /* SIN6_LEN */
 354                         mask = &sin_mask;
 355                 }
 356
 357 #ifdef __OpenBSD__
 358                 if (nexthop->nh_label
 359                     && !kernel_rtm_add_labels(nexthop->nh_label, &smpls))
 360                         continue;
 361                 smplsp = (union sockunion *)&smpls;
 362 #endif
 363
 364                 error = rtm_write(cmd, (union sockunion *)&sin_dest,
 365                                   (union sockunion *)mask,
 366                                   gate ? (union sockunion *)&sin_gate : NULL,
 367                                   smplsp, ifindex, bh_type, re->metric);
 368                 (void)error;
 369
 370                 nexthop_num++;
 371         }
 372
 373         /* If there is no useful nexthop then return. */
 374         if (nexthop_num == 0) {
 375                 if (IS_ZEBRA_DEBUG_KERNEL)
 376                         zlog_debug("kernel_rtm_ipv6(): No useful nexthop.");
 377                 return 0;
 378         }
 379
 380         return 0; /*XXX*/
 381 }
 382
 383 static int kernel_rtm(int cmd, const struct prefix *p, struct route_entry *re)
 384 {
 385         switch (PREFIX_FAMILY(p)) {
 386         case AF_INET:
 387                 return kernel_rtm_ipv4(cmd, p, re);
 388         case AF_INET6:
 389                 return kernel_rtm_ipv6(cmd, p, re);
 390         }
 391         return 0;
 392 }
 393
 394 enum dp_req_result kernel_route_rib(struct route_node *rn,
 395                                     const struct prefix *p,
 396                                     const struct prefix *src_p,
 397                                     struct route_entry *old,
 398                                     struct route_entry *new)
 399 {
 400         int route = 0;
 401
 402         if (src_p && src_p->prefixlen) {
 403                 flog_warn(EC_ZEBRA_UNSUPPORTED_V6_SRCDEST,
 404                           "%s: IPv6 sourcedest routes unsupported!", __func__);
 405                 return DP_REQUEST_FAILURE;
 406         }
 407
 408         frr_elevate_privs(&zserv_privs) {
 409
 410                 if (old)
 411                         route |= kernel_rtm(RTM_DELETE, p, old);
 412
 413                 if (new)
 414                         route |= kernel_rtm(RTM_ADD, p, new);
 415
 416         }
 417
 418         if (new) {
 419                 kernel_route_rib_pass_fail(
 420                         rn, p, new,
 421                         (!route) ? DP_INSTALL_SUCCESS
 422                                  : DP_INSTALL_FAILURE);
 423         } else {
 424                 kernel_route_rib_pass_fail(rn, p, old,
 425                                            (!route)
 426                                                    ? DP_DELETE_SUCCESS
 427                                                    : DP_DELETE_FAILURE);
 428         }
 429
 430         return DP_REQUEST_SUCCESS;
 431 }
 432
 433 int kernel_neigh_update(int add, int ifindex, uint32_t addr, char *lla,
 434                         int llalen, ns_id_t ns_id)
 435 {
 436         /* TODO */
 437         return 0;
 438 }
 439
 440 extern int kernel_get_ipmr_sg_stats(struct zebra_vrf *zvrf, void *mroute)
 441 {
 442         return 0;
 443 }
 444
 445 int kernel_add_vtep(vni_t vni, struct interface *ifp, struct in_addr *vtep_ip)
 446 {
 447         return 0;
 448 }
 449
 450 int kernel_del_vtep(vni_t vni, struct interface *ifp, struct in_addr *vtep_ip)
 451 {
 452         return 0;
 453 }
 454
 455 int kernel_add_mac(struct interface *ifp, vlanid_t vid, struct ethaddr *mac,
 456                    struct in_addr vtep_ip, bool sticky)
 457 {
 458         return 0;
 459 }
 460
 461 int kernel_del_mac(struct interface *ifp, vlanid_t vid, struct ethaddr *mac,
 462                    struct in_addr vtep_ip)
 463 {
 464         return 0;
 465 }
 466
 467 int kernel_add_neigh(struct interface *ifp, struct ipaddr *ip,
 468                      struct ethaddr *mac, uint8_t flags)
 469 {
 470         return 0;
 471 }
 472
 473 int kernel_del_neigh(struct interface *ifp, struct ipaddr *ip)
 474 {
 475         return 0;
 476 }
 477
 478 extern int kernel_interface_set_master(struct interface *master,
 479                                        struct interface *slave)
 480 {
 481         return 0;
 482 }
 483
 484 uint32_t kernel_get_speed(struct interface *ifp)
 485 {
 486         return ifp->speed;
 487 }
 488
 489 #endif /* !HAVE_NETLINK */