bgp_route.c

大名鼎鼎的路由器源码。程序分ZEBRA、OSPFRIP等3个包。程序框架采用一个路由协议一个进程的方式

  /* If community is not disabled check the no-export and local. */
  if (! transparent && bgp_community_filter (peer, ri->attr)) 
    return 0;

  /* If the attribute has originator-id and it is same as remote
     peer's id. */
  if (ri->attr->flag & ATTR_FLAG_BIT (BGP_ATTR_ORIGINATOR_ID))
    {
      if (IPV4_ADDR_SAME (&peer->remote_id, &ri->attr->originator_id))
	{
	  if (BGP_DEBUG (filter, FILTER))  
	    zlog (peer->log, LOG_INFO,
		  "%s [Update:SEND] %s/%d originator-id is same as remote router-id",
		  peer->host,
		  inet_ntop(p->family, &p->u.prefix, buf, SU_ADDRSTRLEN),
		  p->prefixlen);
	  return 0;
	}
    }
 
  /* ORF prefix-list filter check */
  if (CHECK_FLAG (peer->af_cap[afi][safi], PEER_CAP_ORF_PREFIX_RM_ADV)
      && (CHECK_FLAG (peer->af_cap[afi][safi], PEER_CAP_ORF_PREFIX_SM_RCV)
	  || CHECK_FLAG (peer->af_cap[afi][safi], PEER_CAP_ORF_PREFIX_SM_OLD_RCV)))
    if (peer->orf_plist[afi][safi])
      {
	if (prefix_list_apply (peer->orf_plist[afi][safi], p) == PREFIX_DENY)
          return 0;
      }

  /* Output filter check. */
  if (bgp_output_filter (peer, p, ri->attr, afi, safi) == FILTER_DENY)
    {
      if (BGP_DEBUG (filter, FILTER))
	zlog (peer->log, LOG_INFO,
	      "%s [Update:SEND] %s/%d is filtered",
	      peer->host,
	      inet_ntop(p->family, &p->u.prefix, buf, SU_ADDRSTRLEN),
	      p->prefixlen);
      return 0;
    }

#ifdef BGP_SEND_ASPATH_CHECK
  /* AS path loop check. */
  if (aspath_loop_check (ri->attr->aspath, peer->as))
    {
      if (BGP_DEBUG (filter, FILTER))  
        zlog (peer->log, LOG_INFO, 
	      "%s [Update:SEND] suppress announcement to peer AS %d is AS path.",
	      peer->host, peer->as);
      return 0;
    }
#endif /* BGP_SEND_ASPATH_CHECK */

  /* If we're a CONFED we need to loop check the CONFED ID too */
  if (CHECK_FLAG(bgp->config, BGP_CONFIG_CONFEDERATION))
    {
      if (aspath_loop_check(ri->attr->aspath, bgp->confed_id))
	{
	  if (BGP_DEBUG (filter, FILTER))  
	    zlog (peer->log, LOG_INFO, 
		  "%s [Update:SEND] suppress announcement to peer AS %d is AS path.",
		  peer->host,
		  bgp->confed_id);
	  return 0;
	}      
    }

  /* Route-Reflect check. */
  if (peer_sort (from) == BGP_PEER_IBGP && peer_sort (peer) == BGP_PEER_IBGP)
    reflect = 1;
  else
    reflect = 0;

  /* IBGP reflection check. */
  if (reflect)
    {
      /* A route from a Client peer. */
      if (CHECK_FLAG (from->af_flags[afi][safi], PEER_FLAG_REFLECTOR_CLIENT))
	{
	  /* Reflect to all the Non-Client peers and also to the
             Client peers other than the originator.  Originator check
             is already done.  So there is noting to do. */
	  /* no bgp client-to-client reflection check. */
	  if (bgp_flag_check (bgp, BGP_FLAG_NO_CLIENT_TO_CLIENT))
	    if (CHECK_FLAG (peer->af_flags[afi][safi], PEER_FLAG_REFLECTOR_CLIENT))
	      return 0;
	}
      else
	{
	  /* A route from a Non-client peer. Reflect to all other
	     clients. */
	  if (! CHECK_FLAG (peer->af_flags[afi][safi], PEER_FLAG_REFLECTOR_CLIENT))
	    return 0;
	}
    }

  /* For modify attribute, copy it to temporary structure. */
  *attr = *ri->attr;

  /* If local-preference is not set. */
  if ((peer_sort (peer) == BGP_PEER_IBGP 
       || peer_sort (peer) == BGP_PEER_CONFED) 
      && (! (attr->flag & ATTR_FLAG_BIT (BGP_ATTR_LOCAL_PREF))))
    {
      attr->flag |= ATTR_FLAG_BIT (BGP_ATTR_LOCAL_PREF);
      attr->local_pref = bgp->default_local_pref;
    }

  /* Remove MED if its an EBGP peer - will get overwritten by route-maps */
  if (peer_sort (peer) == BGP_PEER_EBGP 
      && attr->flag & ATTR_FLAG_BIT (BGP_ATTR_MULTI_EXIT_DISC))
    {
      if (ri->peer != bgp->peer_self && ! transparent
	  && ! CHECK_FLAG (peer->af_flags[afi][safi], PEER_FLAG_MED_UNCHANGED))
	attr->flag &= ~(ATTR_FLAG_BIT (BGP_ATTR_MULTI_EXIT_DISC));
    }

  /* next-hop-set */
  if (transparent || reflect
      || (CHECK_FLAG (peer->af_flags[afi][safi], PEER_FLAG_NEXTHOP_UNCHANGED)
	  && ((p->family == AF_INET && attr->nexthop.s_addr)
#ifdef HAVE_IPV6
	      || (p->family == AF_INET6 && ri->peer != bgp->peer_self)
#endif /* HAVE_IPV6 */
	      )))
    {
      /* NEXT-HOP Unchanged. */
    }
  else if (CHECK_FLAG (peer->af_flags[afi][safi], PEER_FLAG_NEXTHOP_SELF)
	   || (p->family == AF_INET && attr->nexthop.s_addr == 0)
#ifdef HAVE_IPV6
	   || (p->family == AF_INET6 && ri->peer == bgp->peer_self)
#endif /* HAVE_IPV6 */
	   || (peer_sort (peer) == BGP_PEER_EBGP
	       && bgp_multiaccess_check_v4 (attr->nexthop, peer->host) == 0))
    {
      /* Set IPv4 nexthop. */
      if (p->family == AF_INET)
	{
	  if (safi == SAFI_MPLS_VPN)
	    memcpy (&attr->mp_nexthop_global_in, &peer->nexthop.v4, IPV4_MAX_BYTELEN);
	  else
	    memcpy (&attr->nexthop, &peer->nexthop.v4, IPV4_MAX_BYTELEN);
	}
#ifdef HAVE_IPV6
      /* Set IPv6 nexthop. */
      if (p->family == AF_INET6)
	{
	  /* IPv6 global nexthop must be included. */
	  memcpy (&attr->mp_nexthop_global, &peer->nexthop.v6_global, 
		  IPV6_MAX_BYTELEN);
	  attr->mp_nexthop_len = 16;
	}
#endif /* HAVE_IPV6 */
    }

#ifdef HAVE_IPV6
  if (p->family == AF_INET6)
    {
      /* Link-local address should not be transit to different peer. */
      attr->mp_nexthop_len = 16;

      /* Set link-local address for shared network peer. */
      if (peer->shared_network 
	  && ! IN6_IS_ADDR_UNSPECIFIED (&peer->nexthop.v6_local))
	{
	  memcpy (&attr->mp_nexthop_local, &peer->nexthop.v6_local, 
		  IPV6_MAX_BYTELEN);
	  attr->mp_nexthop_len = 32;
	}

      /* If bgpd act as BGP-4+ route-reflector, do not send link-local
	 address.*/
      if (reflect)
	attr->mp_nexthop_len = 16;

      /* If BGP-4+ link-local nexthop is not link-local nexthop. */
      if (! IN6_IS_ADDR_LINKLOCAL (&peer->nexthop.v6_local))
	attr->mp_nexthop_len = 16;
    }
#endif /* HAVE_IPV6 */

  /* If this is EBGP peer and remove-private-AS is set.  */
  if (peer_sort (peer) == BGP_PEER_EBGP
      && peer_af_flag_check (peer, afi, safi, PEER_FLAG_REMOVE_PRIVATE_AS)
      && aspath_private_as_check (attr->aspath))
    attr->aspath = aspath_empty_get ();

  /* Route map & unsuppress-map apply. */
  if (ROUTE_MAP_OUT_NAME (filter)
      || ri->suppress)
    {
      info.peer = peer;
      info.attr = attr;

      /* The route reflector is not allowed to modify the attributes
	 of the reflected IBGP routes. */
      if (peer_sort (from) == BGP_PEER_IBGP 
	  && peer_sort (peer) == BGP_PEER_IBGP)
	{
	  dummy_attr = *attr;
	  info.attr = &dummy_attr;
	}

      SET_FLAG (peer->rmap_type, PEER_RMAP_TYPE_OUT); 

      if (ri->suppress)
	ret = route_map_apply (UNSUPPRESS_MAP (filter), p, RMAP_BGP, &info);
      else
	ret = route_map_apply (ROUTE_MAP_OUT (filter), p, RMAP_BGP, &info);

      peer->rmap_type = 0;

      if (ret == RMAP_DENYMATCH)
	{
	  bgp_attr_flush (attr);
	  return 0;
	}
    }
  return 1;
}

int
bgp_process (struct bgp *bgp, struct bgp_node *rn, afi_t afi, safi_t safi)
{
  struct prefix *p;
  struct bgp_info *ri;
  struct bgp_info *new_select;
  struct bgp_info *old_select;
  struct listnode *nn;
  struct peer *peer;
  struct attr attr;
  struct bgp_info *ri1;
  struct bgp_info *ri2;

  p = &rn->p;

  /* bgp deterministic-med */
  new_select = NULL;
  if (bgp_flag_check (bgp, BGP_FLAG_DETERMINISTIC_MED))
    for (ri1 = rn->info; ri1; ri1 = ri1->next)
      {
	if (CHECK_FLAG (ri1->flags, BGP_INFO_DMED_CHECK))
	  continue;
	if (BGP_INFO_HOLDDOWN (ri1))
	  continue;

	new_select = ri1;
	if (ri1->next)
	  for (ri2 = ri1->next; ri2; ri2 = ri2->next)
	    {
	      if (CHECK_FLAG (ri2->flags, BGP_INFO_DMED_CHECK))
		continue;
	      if (BGP_INFO_HOLDDOWN (ri2))
		continue;

	      if (aspath_cmp_left (ri1->attr->aspath, ri2->attr->aspath)
		  || aspath_cmp_left_confed (ri1->attr->aspath,
					     ri2->attr->aspath))
		{
		  if (bgp_info_cmp (bgp, ri2, new_select))
		    {
		      UNSET_FLAG (new_select->flags, BGP_INFO_DMED_SELECTED);
		      new_select = ri2;
		    }

		  SET_FLAG (ri2->flags, BGP_INFO_DMED_CHECK);
		}
	    }
	SET_FLAG (new_select->flags, BGP_INFO_DMED_CHECK);
	SET_FLAG (new_select->flags, BGP_INFO_DMED_SELECTED);
      }

  /* Check old selected route and new selected route. */
  old_select = NULL;
  new_select = NULL;
  for (ri = rn->info; ri; ri = ri->next)
    {
      if (CHECK_FLAG (ri->flags, BGP_INFO_SELECTED))
	old_select = ri;

      if (BGP_INFO_HOLDDOWN (ri))
	continue;

      if (bgp_flag_check (bgp, BGP_FLAG_DETERMINISTIC_MED)
          && (! CHECK_FLAG (ri->flags, BGP_INFO_DMED_SELECTED)))
	{
	  UNSET_FLAG (ri->flags, BGP_INFO_DMED_CHECK);
	  continue;
        }
      UNSET_FLAG (ri->flags, BGP_INFO_DMED_CHECK);
      UNSET_FLAG (ri->flags, BGP_INFO_DMED_SELECTED);

      if (bgp_info_cmp (bgp, ri, new_select))
	new_select = ri;
    }

  /* Nothing to do. */
  if (old_select && old_select == new_select)
    {
      if (! CHECK_FLAG (old_select->flags, BGP_INFO_ATTR_CHANGED))
	{
	  if (CHECK_FLAG (old_select->flags, BGP_INFO_IGP_CHANGED))
	    bgp_zebra_announce (p, old_select, bgp);
	  return 0;
	}
    }

  if (old_select)
    UNSET_FLAG (old_select->flags, BGP_INFO_SELECTED);
  if (new_select)
    {
      SET_FLAG (new_select->flags, BGP_INFO_SELECTED);
      UNSET_FLAG (new_select->flags, BGP_INFO_ATTR_CHANGED);
    }

  /* Check each BGP peer. */
  LIST_LOOP (bgp->peer, peer, nn)
    {
      /* Announce route to Established peer. */
      if (peer->status != Established)
	continue;

      /* Address family configuration check. */
      if (! peer->afc_nego[afi][safi])
	continue;

      /* First update is deferred until ORF or ROUTE-REFRESH is received */
      if (CHECK_FLAG (peer->af_sflags[afi][safi], PEER_STATUS_ORF_WAIT_REFRESH))
	continue;

      /* Announcement to peer->conf.  If the route is filtered,
         withdraw it. */
      if (new_select 
	  && bgp_announce_check (new_select, peer, p, &attr, afi, safi))
	bgp_adj_out_set (rn, peer, p, &attr, afi, safi, new_select);
      else
	bgp_adj_out_unset (rn, peer, p, afi, safi);
    }

  /* FIB update. */
  if (safi == SAFI_UNICAST && ! bgp->name &&
      ! bgp_option_check (BGP_OPT_NO_FIB))
    {
      if (new_select 
	  && new_select->type == ZEBRA_ROUTE_BGP 
	  && new_select->sub_type == BGP_ROUTE_NORMAL)
	bgp_zebra_announce (p, new_select, bgp);
      else
	{
	  /* Withdraw the route from the kernel. */
	  if (old_select 
	      && old_select->type == ZEBRA_ROUTE_BGP
	      && old_select->sub_type == BGP_ROUTE_NORMAL)
	    bgp_zebra_withdraw (p, old_select);
	}
    }
  return 0;
}

int
bgp_maximum_prefix_overflow (struct peer *peer, afi_t afi, safi_t safi, int always)
{
  if (! CHECK_FLAG (peer->af_flags[afi][safi], PEER_FLAG_MAX_PREFIX))
    return 0;

  if (peer->pcount[afi][safi] > peer->pmax[afi][safi])
    {
      if (CHECK_FLAG (peer->af_sflags[afi][safi], PEER_STATUS_PREFIX_LIMIT)
	  && ! always)
	return 0;

      zlog (peer->log, LOG_INFO,
	    "%%MAXPFXEXCEED: No. of prefix received from %s (afi %d): %ld exceed limit %ld",
	    peer->host, afi, peer->pcount[afi][safi], peer->pmax[afi][safi]);
      SET_FLAG (peer->af_sflags[afi][safi], PEER_STATUS_PREFIX_LIMIT);

      if (CHECK_FLAG (peer->af_flags[afi][safi], PEER_FLAG_MAX_PREFIX_WARNING))
	return 0;

      {
	char ndata[7];

	ndata[0] = (u_char)(afi >>  8);
	ndata[1] = (u_char) afi;
	ndata[3] = (u_char)(peer->pmax[afi][safi] >> 24);
	ndata[4] = (u_char)(peer->pmax[afi][safi] >> 16);
	ndata[5] = (u_char)(peer->pmax[afi][safi] >> 8);
	ndata[6] = (u_char)(peer->pmax[afi][safi]);

	if (safi == SAFI_MPLS_VPN)
	  safi = BGP_SAFI_VPNV4;
	ndata[2] = (u_char) safi;

	SET_FLAG (peer->sflags, PEER_STATUS_PREFIX_OVERFLOW);
	bgp_notify_send_with_data (peer, BGP_NOTIFY_CEASE,
				   BGP_NOTIFY_CEASE_MAX_PREFIX, ndata, 7);
      }
      return 1;
    }
  else
    UNSET_FLAG (peer->af_sflags[afi][safi], PEER_STATUS_PREFIX_LIMIT);

  if (peer->pcount[afi][safi] > (peer->pmax[afi][safi] * peer->pmax_threshold[afi][safi] / 100))
    {
      if (CHECK_FLAG (peer->af_sflags[afi][safi], PEER_STATUS_PREFIX_THRESHOLD)
	  && ! always)
	return 0;

      zlog (peer->log, LOG_INFO,
	    "%%MAXPFX: No. of prefix received from %s (afi %d) reaches %ld, max %ld",
	    peer->host, afi, peer->pcount[afi][safi], peer->pmax[afi][safi]);
      SET_FLAG (peer->af_sflags[afi][safi], PEER_STATUS_PREFIX_THRESHOLD);
    }
  else
    UNSET_FLAG (peer->af_sflags[afi][safi], PEER_STATUS_PREFIX_THRESHOLD);

  return 0;
}

void
bgp_rib_remove (struct bgp_node *rn, struct bgp_info *ri, struct peer *peer,
		afi_t afi, safi_t safi)
{
  if (! CHECK_FLAG (ri->flags, BGP_INFO_HISTORY))
    {
      peer->pcount[afi][safi]--;
      bgp_aggregate_decrement (peer->bgp, &rn->p, ri, afi, safi);
      UNSET_FLAG (ri->flags, BGP_INFO_VALID);
      bgp_process (peer->bgp, rn, afi, safi);
    }
  bgp_info_delete (rn, ri);
  bgp_info_free (ri);
  bgp_unlock_node (rn);
}

void
bgp_rib_withdraw (struct bgp_node *rn, struct bgp_info *ri, struct peer *peer,
		  afi_t afi, safi_t safi, int force)
{
  int valid;
  int status = BGP_DAMP_NONE;

  if (! CHECK_FLAG (ri->flags, BGP_INFO_HISTORY))
    {
      peer->pcount[afi][safi]--;
      bgp_aggregate_decrement (peer->bgp, &rn->p, ri, afi, safi);
    }

  if (! force)
    {
      if (CHECK_FLAG (peer->bgp->af_flags[afi][safi], BGP_CONFIG_DAMPENING)
	  && peer_sort (peer) == BGP_PEER_EBGP)
	status = bgp_damp_withdraw (ri, rn, afi, safi, 0);

      if (status == BGP_DAMP_SUPPRESSED)
	return;
    }

  valid = CHECK_FLAG (ri->flags, BGP_INFO_VALID);
  UNSET_FLAG (ri->flags, BGP_INFO_VALID);
  bgp_process (peer->bgp, rn, afi, safi);

  if (valid)
    SET_FLAG (ri->flags, BGP_INFO_VALID);

  if (status != BGP_DAMP_USED)
    {
      bgp_info_delete (rn, ri);
      bgp_info_free (ri);
      bgp_unlock_node (rn);
    }
}

int
bgp_update (struct peer *peer, struct prefix *p, struct attr *attr, 
	    afi_t afi, safi_t safi, int type, int sub_type,
	    struct prefix_rd *prd, u_char *tag, int soft_reconfig)
{
  int ret;
  int aspath_loop_count = 0;
  struct bgp_node *rn;
  struct bgp *bgp;
  struct attr new_attr;
  struct attr *attr_new;
  struct bgp_info *ri;
  struct bgp_info *new;
  char *reason;
  char buf[SU_ADDRSTRLEN];

  bgp = peer->bgp;
  rn = bgp_afi_node_get (bgp, afi, safi, p, prd);

  /* When peer's soft reconfiguration enabled.  Record input packet in
     Adj-RIBs-In.  */
  if (CHECK_FLAG (peer->af_flags[afi][safi], PEER_FLAG_SOFT_RECONFIG)
      && peer != bgp->peer_self && ! soft_reconfig)
    bgp_adj_in_set (rn, peer, attr);

  /* Check previously received route. */
  for (ri = rn->info; ri; ri = ri->next)
    if (ri->peer == peer && ri->type == type && ri->sub_type == sub_type)