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 #ifdef SIN6_LEN
  93 /* Calculate sin6_len value for netmask socket value. */
  94 static int sin6_masklen(struct in6_addr mask)
  95 {
  96         struct sockaddr_in6 sin6;
  97         char *p, *lim;
  98         int len;
  99
 100         if (IN6_IS_ADDR_UNSPECIFIED(&mask))
 101                 return sizeof(long);
 102
 103         sin6.sin6_addr = mask;
 104         len = sizeof(struct sockaddr_in6);
 105
 106         lim = (char *)&sin6.sin6_addr;
 107         p = lim + sizeof(sin6.sin6_addr);
 108
 109         while (*--p == 0 && p >= lim)
 110                 len--;
 111
 112         return len;
 113 }
 114 #endif /* SIN6_LEN */
 115
 116 /* Interface between zebra message and rtm message. */
 117 static int kernel_rtm(int cmd, const struct prefix *p,
 118                       const struct nexthop_group *ng, uint32_t metric)
 119
 120 {
 121         union sockunion sin_dest, sin_mask, sin_gate;
 122 #ifdef __OpenBSD__
 123         struct sockaddr_mpls smpls;
 124 #endif
 125         union sockunion *smplsp = NULL;
 126         struct nexthop *nexthop;
 127         int nexthop_num = 0;
 128         ifindex_t ifindex = 0;
 129         bool gate = false;
 130         int error;
 131         char prefix_buf[PREFIX_STRLEN];
 132         enum blackhole_type bh_type = BLACKHOLE_UNSPEC;
 133
 134         if (IS_ZEBRA_DEBUG_RIB)
 135                 prefix2str(p, prefix_buf, sizeof(prefix_buf));
 136
 137         /*
 138          * We only have the ability to ADD or DELETE at this point
 139          * in time.
 140          */
 141         if (cmd != RTM_ADD && cmd != RTM_DELETE) {
 142                 if (IS_ZEBRA_DEBUG_KERNEL)
 143                         zlog_debug("%s: %s odd command %s for flags %d",
 144                                    __func__, prefix_buf,
 145                                    lookup_msg(rtm_type_str, cmd, NULL),
 146                                    nexthop->flags);
 147                 return 0;
 148         }
 149
 150         memset(&sin_dest, 0, sizeof(sin_dest));
 151         memset(&sin_gate, 0, sizeof(sin_gate));
 152         memset(&sin_mask, 0, sizeof(sin_mask));
 153
 154         switch (p->family) {
 155         case AF_INET:
 156                 sin_dest.sin.sin_family = AF_INET;
 157 #ifdef HAVE_STRUCT_SOCKADDR_IN_SIN_LEN
 158                 sin_dest.sin.sin_len = sizeof(sin_dest);
 159                 sin_gate.sin.sin_len = sizeof(sin_gate);
 160 #endif /* HAVE_STRUCT_SOCKADDR_IN_SIN_LEN */
 161                 sin_dest.sin.sin_addr = p->u.prefix4;
 162                 sin_gate.sin.sin_family = AF_INET;
 163                 break;
 164         case AF_INET6:
 165                 sin_dest.sin6.sin6_family = AF_INET6;
 166 #ifdef SIN6_LEN
 167                 sin_dest.sin6.sin6_len = sizeof(sin_dest);
 168 #endif /* SIN6_LEN */
 169                 sin_dest.sin6.sin6_addr = p->u.prefix6;
 170                 sin_gate.sin6.sin6_family = AF_INET6;
 171 #ifdef HAVE_STRUCT_SOCKADDR_IN_SIN_LEN
 172                 sin_gate.sin6.sin6_len = sizeof(sin_gate);
 173 #endif /* HAVE_STRUCT_SOCKADDR_IN_SIN_LEN */
 174                 break;
 175         }
 176
 177         /* Make gateway. */
 178         for (ALL_NEXTHOPS_PTR(ng, nexthop)) {
 179                 /*
 180                  * We only want to use the actual good nexthops
 181                  */
 182                 if (CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_RECURSIVE) ||
 183                     !CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_ACTIVE))
 184                         continue;
 185
 186                 gate = false;
 187                 char gate_buf[INET_ADDRSTRLEN] = "NULL";
 188
 189                 switch (nexthop->type) {
 190                 case NEXTHOP_TYPE_IPV4:
 191                 case NEXTHOP_TYPE_IPV4_IFINDEX:
 192                         sin_gate.sin.sin_addr = nexthop->gate.ipv4;
 193                         sin_gate.sin.sin_family = AF_INET;
 194                         ifindex = nexthop->ifindex;
 195                         gate = true;
 196                         break;
 197                 case NEXTHOP_TYPE_IPV6:
 198                 case NEXTHOP_TYPE_IPV6_IFINDEX:
 199                         sin_gate.sin6.sin6_addr = nexthop->gate.ipv6;
 200                         sin_gate.sin6.sin6_family = AF_INET6;
 201                         ifindex = nexthop->ifindex;
 202 /* Under kame set interface index to link local address */
 203 #ifdef KAME
 204
 205 #define SET_IN6_LINKLOCAL_IFINDEX(a, i)                                        \
 206         do {                                                                   \
 207                 (a).s6_addr[2] = ((i) >> 8) & 0xff;                            \
 208                 (a).s6_addr[3] = (i)&0xff;                                     \
 209         } while (0)
 210
 211                         if (IN6_IS_ADDR_LINKLOCAL(&sin_gate.sin6.sin6_addr))
 212                                 SET_IN6_LINKLOCAL_IFINDEX(
 213                                         sin_gate.sin6.sin6_addr,
 214                                         ifindex);
 215 #endif /* KAME */
 216
 217                         gate = true;
 218                         break;
 219                 case NEXTHOP_TYPE_IFINDEX:
 220                         ifindex = nexthop->ifindex;
 221                         break;
 222                 case NEXTHOP_TYPE_BLACKHOLE:
 223                         bh_type = nexthop->bh_type;
 224                         switch (p->family) {
 225                         case AFI_IP: {
 226                                 struct in_addr loopback;
 227                                 loopback.s_addr = htonl(INADDR_LOOPBACK);
 228                                 sin_gate.sin.sin_addr = loopback;
 229                                 gate = true;
 230                         }
 231                                 break;
 232                         case AFI_IP6:
 233                                 break;
 234                         }
 235                 }
 236
 237                 switch (p->family) {
 238                 case AF_INET:
 239                         masklen2ip(p->prefixlen, &sin_mask.sin.sin_addr);
 240                         sin_mask.sin.sin_family = AF_INET;
 241 #ifdef HAVE_STRUCT_SOCKADDR_IN_SIN_LEN
 242                         sin_mask.sin.sin_len = sin_masklen(
 243                                 sin_mask.sin.sin_addr);
 244 #endif /* HAVE_STRUCT_SOCKADDR_IN_SIN_LEN */
 245                         break;
 246                 case AF_INET6:
 247                         masklen2ip6(p->prefixlen, &sin_mask.sin6.sin6_addr);
 248                         sin_mask.sin6.sin6_family = AF_INET6;
 249 #ifdef SIN6_LEN
 250                         sin_mask.sin6.sin6_len = sin6_masklen(
 251                                 sin_mask.sin6.sin6_addr);
 252 #endif /* SIN6_LEN */
 253                         break;
 254                 }
 255
 256 #ifdef __OpenBSD__
 257                 if (nexthop->nh_label
 258                     && !kernel_rtm_add_labels(nexthop->nh_label, &smpls))
 259                         continue;
 260                 smplsp = (union sockunion *)&smpls;
 261 #endif
 262                 error = rtm_write(cmd, &sin_dest, &sin_mask,
 263                                   gate ? &sin_gate : NULL, smplsp,
 264                                   ifindex, bh_type, metric);
 265
 266                 if (IS_ZEBRA_DEBUG_KERNEL) {
 267                         if (!gate) {
 268                                 zlog_debug("%s: %s: attention! gate not found for re",
 269                                            __func__, prefix_buf);
 270                         } else
 271                                 inet_ntop(p->family == AFI_IP ? AF_INET
 272                                           : AF_INET6,
 273                                           &sin_gate.sin.sin_addr,
 274                                           gate_buf, INET_ADDRSTRLEN);
 275                 }
 276                 switch (error) {
 277                         /* We only flag nexthops as being in FIB if
 278                          * rtm_write() did its work. */
 279                 case ZEBRA_ERR_NOERROR:
 280                         nexthop_num++;
 281                         if (IS_ZEBRA_DEBUG_KERNEL)
 282                                 zlog_debug("%s: %s: successfully did NH %s",
 283                                            __func__, prefix_buf, gate_buf);
 284                         if (cmd == RTM_ADD)
 285                                 SET_FLAG(nexthop->flags, NEXTHOP_FLAG_FIB);
 286                         break;
 287
 288                         /* The only valid case for this error is
 289                          * kernel's failure to install a multipath
 290                          * route, which is common for FreeBSD. This
 291                          * should be ignored silently, but logged as an error
 292                          * otherwise.
 293                          */
 294                 case ZEBRA_ERR_RTEXIST:
 295                         if (cmd != RTM_ADD)
 296                                 flog_err(EC_LIB_SYSTEM_CALL,
 297                                          "%s: rtm_write() returned %d for command %d",
 298                                          __func__, error, cmd);
 299                         continue;
 300
 301                         /* Note any unexpected status returns */
 302                 default:
 303                         flog_err(EC_LIB_SYSTEM_CALL,
 304                                  "%s: %s: rtm_write() unexpectedly returned %d for command %s",
 305                                  __func__,
 306                                  prefix2str(p, prefix_buf,
 307                                             sizeof(prefix_buf)),
 308                                  error, lookup_msg(rtm_type_str, cmd, NULL));
 309                         break;
 310                 }
 311         } /* for (ALL_NEXTHOPS(...))*/
 312
 313         /* If there was no useful nexthop, then complain. */
 314         if (nexthop_num == 0) {
 315                 if (IS_ZEBRA_DEBUG_KERNEL)
 316                         zlog_debug("%s: No useful nexthops were found in RIB prefix %s",
 317                                    __func__, prefix2str(p, prefix_buf,
 318                                                         sizeof(prefix_buf)));
 319                 return 1;
 320         }
 321
 322         return 0; /*XXX*/
 323 }
 324
 325 /*
 326  * Update or delete a prefix from the kernel,
 327  * using info from a dataplane context struct.
 328  */
 329 enum zebra_dplane_result kernel_route_update(struct zebra_dplane_ctx *ctx)
 330 {
 331         enum zebra_dplane_result res = ZEBRA_DPLANE_REQUEST_SUCCESS;
 332
 333         if (dplane_ctx_get_src(ctx) != NULL) {
 334                 zlog_err("route add: IPv6 sourcedest routes unsupported!");
 335                 res = ZEBRA_DPLANE_REQUEST_FAILURE;
 336                 goto done;
 337         }
 338
 339         frr_elevate_privs(&zserv_privs) {
 340
 341                 if (dplane_ctx_get_op(ctx) == DPLANE_OP_ROUTE_DELETE)
 342                         kernel_rtm(RTM_DELETE, dplane_ctx_get_dest(ctx),
 343                                    dplane_ctx_get_ng(ctx),
 344                                    dplane_ctx_get_metric(ctx));
 345                 else if (dplane_ctx_get_op(ctx) == DPLANE_OP_ROUTE_INSTALL)
 346                         kernel_rtm(RTM_ADD, dplane_ctx_get_dest(ctx),
 347                                    dplane_ctx_get_ng(ctx),
 348                                    dplane_ctx_get_metric(ctx));
 349                 else if (dplane_ctx_get_op(ctx) == DPLANE_OP_ROUTE_UPDATE) {
 350                         /* Must do delete and add separately -
 351                          * no update available
 352                          */
 353                         kernel_rtm(RTM_DELETE, dplane_ctx_get_dest(ctx),
 354                                    dplane_ctx_get_old_ng(ctx),
 355                                    dplane_ctx_get_old_metric(ctx));
 356
 357                         kernel_rtm(RTM_ADD, dplane_ctx_get_dest(ctx),
 358                                    dplane_ctx_get_ng(ctx),
 359                                    dplane_ctx_get_metric(ctx));
 360                 } else {
 361                         zlog_err("Invalid routing socket update op %s (%u)",
 362                                  dplane_op2str(dplane_ctx_get_op(ctx)),
 363                                  dplane_ctx_get_op(ctx));
 364                         res = ZEBRA_DPLANE_REQUEST_FAILURE;
 365                 }
 366         } /* Elevated privs */
 367
 368 done:
 369
 370         return res;
 371 }
 372
 373 int kernel_neigh_update(int add, int ifindex, uint32_t addr, char *lla,
 374                         int llalen, ns_id_t ns_id)
 375 {
 376         /* TODO */
 377         return 0;
 378 }
 379
 380 extern int kernel_get_ipmr_sg_stats(struct zebra_vrf *zvrf, void *mroute)
 381 {
 382         return 0;
 383 }
 384
 385 int kernel_add_vtep(vni_t vni, struct interface *ifp, struct in_addr *vtep_ip)
 386 {
 387         return 0;
 388 }
 389
 390 int kernel_del_vtep(vni_t vni, struct interface *ifp, struct in_addr *vtep_ip)
 391 {
 392         return 0;
 393 }
 394
 395 int kernel_add_mac(struct interface *ifp, vlanid_t vid, struct ethaddr *mac,
 396                    struct in_addr vtep_ip, bool sticky)
 397 {
 398         return 0;
 399 }
 400
 401 int kernel_del_mac(struct interface *ifp, vlanid_t vid, struct ethaddr *mac,
 402                    struct in_addr vtep_ip)
 403 {
 404         return 0;
 405 }
 406
 407 int kernel_add_neigh(struct interface *ifp, struct ipaddr *ip,
 408                      struct ethaddr *mac, uint8_t flags)
 409 {
 410         return 0;
 411 }
 412
 413 int kernel_del_neigh(struct interface *ifp, struct ipaddr *ip)
 414 {
 415         return 0;
 416 }
 417
 418 extern int kernel_interface_set_master(struct interface *master,
 419                                        struct interface *slave)
 420 {
 421         return 0;
 422 }
 423
 424 uint32_t kernel_get_speed(struct interface *ifp)
 425 {
 426         return ifp->speed;
 427 }
 428
 429 #endif /* !HAVE_NETLINK */