Initial revision
diff --git a/zebra/zebra_rib.c b/zebra/zebra_rib.c
new file mode 100644
index 0000000..ec07e2e
--- /dev/null
+++ b/zebra/zebra_rib.c
@@ -0,0 +1,2199 @@
+/* Routing Information Base.
+ * Copyright (C) 1997, 98, 99, 2001 Kunihiro Ishiguro
+ *
+ * This file is part of GNU Zebra.
+ *
+ * GNU Zebra is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the
+ * Free Software Foundation; either version 2, or (at your option) any
+ * later version.
+ *
+ * GNU Zebra is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with GNU Zebra; see the file COPYING.  If not, write to the Free
+ * Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
+ * 02111-1307, USA.  
+ */
+
+#include <zebra.h>
+
+#include "prefix.h"
+#include "table.h"
+#include "memory.h"
+#include "str.h"
+#include "command.h"
+#include "if.h"
+#include "log.h"
+#include "sockunion.h"
+
+#include "zebra/rib.h"
+#include "zebra/rt.h"
+#include "zebra/zserv.h"
+#include "zebra/redistribute.h"
+#include "zebra/debug.h"
+
+/* Default rtm_table for all clients */
+extern int rtm_table_default;
+
+/* Each route type's string and default distance value. */
+struct
+{  
+  int key;
+  int distance;
+} route_info[] =
+{
+  {ZEBRA_ROUTE_SYSTEM,    0},
+  {ZEBRA_ROUTE_KERNEL,    0},
+  {ZEBRA_ROUTE_CONNECT,   0},
+  {ZEBRA_ROUTE_STATIC,    1},
+  {ZEBRA_ROUTE_RIP,     120},
+  {ZEBRA_ROUTE_RIPNG,   120},
+  {ZEBRA_ROUTE_OSPF,    110},
+  {ZEBRA_ROUTE_OSPF6,   110},
+  {ZEBRA_ROUTE_BGP,      20  /* IBGP is 200. */}
+};
+
+/* Vector for routing table.  */
+vector vrf_vector;
+
+/* Allocate new VRF.  */
+struct vrf *
+vrf_alloc (char *name)
+{
+  struct vrf *vrf;
+
+  vrf = XCALLOC (MTYPE_VRF, sizeof (struct vrf));
+
+  /* Put name.  */
+  if (name)
+    vrf->name = XSTRDUP (MTYPE_VRF_NAME, name);
+
+  /* Allocate routing table and static table.  */
+  vrf->table[AFI_IP][SAFI_UNICAST] = route_table_init ();
+  vrf->table[AFI_IP6][SAFI_UNICAST] = route_table_init ();
+  vrf->stable[AFI_IP][SAFI_UNICAST] = route_table_init ();
+  vrf->stable[AFI_IP6][SAFI_UNICAST] = route_table_init ();
+
+  return vrf;
+}
+
+/* Free VRF.  */
+void
+vrf_free (struct vrf *vrf)
+{
+  if (vrf->name)
+    XFREE (MTYPE_VRF_NAME, vrf->name);
+  XFREE (MTYPE_VRF, vrf);
+}
+
+/* Lookup VRF by identifier.  */
+struct vrf *
+vrf_lookup (u_int32_t id)
+{
+  return vector_lookup (vrf_vector, id);
+}
+
+/* Lookup VRF by name.  */
+struct vrf *
+vrf_lookup_by_name (char *name)
+{
+  int i;
+  struct vrf *vrf;
+
+  for (i = 0; i < vector_max (vrf_vector); i++)
+    if ((vrf = vector_slot (vrf_vector, i)) != NULL)
+      if (vrf->name && name && strcmp (vrf->name, name) == 0)
+	return vrf;
+  return NULL;
+}
+
+/* Initialize VRF.  */
+void
+vrf_init ()
+{
+  struct vrf *default_table;
+
+  /* Allocate VRF vector.  */
+  vrf_vector = vector_init (1);
+
+  /* Allocate default main table.  */
+  default_table = vrf_alloc ("Default-IP-Routing-Table");
+
+  /* Default table index must be 0.  */
+  vector_set_index (vrf_vector, 0, default_table);
+}
+
+/* Lookup route table.  */
+struct route_table *
+vrf_table (afi_t afi, safi_t safi, u_int32_t id)
+{
+  struct vrf *vrf;
+
+  vrf = vrf_lookup (id);
+  if (! vrf)
+    return NULL;
+
+  return vrf->table[afi][safi];
+}
+
+/* Lookup static route table.  */
+struct route_table *
+vrf_static_table (afi_t afi, safi_t safi, u_int32_t id)
+{
+  struct vrf *vrf;
+
+  vrf = vrf_lookup (id);
+  if (! vrf)
+    return NULL;
+
+  return vrf->stable[afi][safi];
+}
+
+/* Add nexthop to the end of the list.  */
+void
+nexthop_add (struct rib *rib, struct nexthop *nexthop)
+{
+  struct nexthop *last;
+
+  for (last = rib->nexthop; last && last->next; last = last->next)
+    ;
+  if (last)
+    last->next = nexthop;
+  else
+    rib->nexthop = nexthop;
+  nexthop->prev = last;
+
+  rib->nexthop_num++;
+}
+
+/* Delete specified nexthop from the list. */
+void
+nexthop_delete (struct rib *rib, struct nexthop *nexthop)
+{
+  if (nexthop->next)
+    nexthop->next->prev = nexthop->prev;
+  if (nexthop->prev)
+    nexthop->prev->next = nexthop->next;
+  else
+    rib->nexthop = nexthop->next;
+  rib->nexthop_num--;
+}
+
+/* Free nexthop. */
+void
+nexthop_free (struct nexthop *nexthop)
+{
+  if (nexthop->type == NEXTHOP_TYPE_IFNAME && nexthop->ifname)
+    free (nexthop->ifname);
+  XFREE (MTYPE_NEXTHOP, nexthop);
+}
+
+struct nexthop *
+nexthop_ifindex_add (struct rib *rib, unsigned int ifindex)
+{
+  struct nexthop *nexthop;
+
+  nexthop = XMALLOC (MTYPE_NEXTHOP, sizeof (struct nexthop));
+  memset (nexthop, 0, sizeof (struct nexthop));
+  nexthop->type = NEXTHOP_TYPE_IFINDEX;
+  nexthop->ifindex = ifindex;
+
+  nexthop_add (rib, nexthop);
+
+  return nexthop;
+}
+
+struct nexthop *
+nexthop_ifname_add (struct rib *rib, char *ifname)
+{
+  struct nexthop *nexthop;
+
+  nexthop = XMALLOC (MTYPE_NEXTHOP, sizeof (struct nexthop));
+  memset (nexthop, 0, sizeof (struct nexthop));
+  nexthop->type = NEXTHOP_TYPE_IFNAME;
+  nexthop->ifname = strdup (ifname);
+
+  nexthop_add (rib, nexthop);
+
+  return nexthop;
+}
+
+struct nexthop *
+nexthop_ipv4_add (struct rib *rib, struct in_addr *ipv4)
+{
+  struct nexthop *nexthop;
+
+  nexthop = XMALLOC (MTYPE_NEXTHOP, sizeof (struct nexthop));
+  memset (nexthop, 0, sizeof (struct nexthop));
+  nexthop->type = NEXTHOP_TYPE_IPV4;
+  nexthop->gate.ipv4 = *ipv4;
+
+  nexthop_add (rib, nexthop);
+
+  return nexthop;
+}
+
+struct nexthop *
+nexthop_ipv4_ifindex_add (struct rib *rib, struct in_addr *ipv4, 
+			  unsigned int ifindex)
+{
+  struct nexthop *nexthop;
+
+  nexthop = XMALLOC (MTYPE_NEXTHOP, sizeof (struct nexthop));
+  memset (nexthop, 0, sizeof (struct nexthop));
+  nexthop->type = NEXTHOP_TYPE_IPV4_IFINDEX;
+  nexthop->gate.ipv4 = *ipv4;
+  nexthop->ifindex = ifindex;
+
+  nexthop_add (rib, nexthop);
+
+  return nexthop;
+}
+
+#ifdef HAVE_IPV6
+struct nexthop *
+nexthop_ipv6_add (struct rib *rib, struct in6_addr *ipv6)
+{
+  struct nexthop *nexthop;
+
+  nexthop = XMALLOC (MTYPE_NEXTHOP, sizeof (struct nexthop));
+  memset (nexthop, 0, sizeof (struct nexthop));
+  nexthop->type = NEXTHOP_TYPE_IPV6;
+  nexthop->gate.ipv6 = *ipv6;
+
+  nexthop_add (rib, nexthop);
+
+  return nexthop;
+}
+
+struct nexthop *
+nexthop_ipv6_ifname_add (struct rib *rib, struct in6_addr *ipv6,
+			 char *ifname)
+{
+  struct nexthop *nexthop;
+
+  nexthop = XMALLOC (MTYPE_NEXTHOP, sizeof (struct nexthop));
+  memset (nexthop, 0, sizeof (struct nexthop));
+  nexthop->type = NEXTHOP_TYPE_IPV6_IFNAME;
+  nexthop->gate.ipv6 = *ipv6;
+  nexthop->ifname = XSTRDUP (0, ifname);
+
+  nexthop_add (rib, nexthop);
+
+  return nexthop;
+}
+
+struct nexthop *
+nexthop_ipv6_ifindex_add (struct rib *rib, struct in6_addr *ipv6,
+			  unsigned int ifindex)
+{
+  struct nexthop *nexthop;
+
+  nexthop = XMALLOC (MTYPE_NEXTHOP, sizeof (struct nexthop));
+  memset (nexthop, 0, sizeof (struct nexthop));
+  nexthop->type = NEXTHOP_TYPE_IPV6_IFINDEX;
+  nexthop->gate.ipv6 = *ipv6;
+  nexthop->ifindex = ifindex;
+
+  nexthop_add (rib, nexthop);
+
+  return nexthop;
+}
+#endif /* HAVE_IPV6 */
+
+
+struct nexthop *
+nexthop_blackhole_add (struct rib *rib)
+{
+  struct nexthop *nexthop;
+
+  nexthop = XMALLOC (MTYPE_NEXTHOP, sizeof (struct nexthop));
+  memset (nexthop, 0, sizeof (struct nexthop));
+  nexthop->type = NEXTHOP_TYPE_BLACKHOLE;
+  SET_FLAG (rib->flags, ZEBRA_FLAG_BLACKHOLE);
+
+  nexthop_add (rib, nexthop);
+
+  return nexthop;
+}
+
+/* If force flag is not set, do not modify falgs at all for uninstall
+   the route from FIB. */
+int
+nexthop_active_ipv4 (struct rib *rib, struct nexthop *nexthop, int set,
+		     struct route_node *top)
+{
+  struct prefix_ipv4 p;
+  struct route_table *table;
+  struct route_node *rn;
+  struct rib *match;
+  struct nexthop *newhop;
+
+  if (nexthop->type == NEXTHOP_TYPE_IPV4)
+    nexthop->ifindex = 0;
+
+  if (set)
+    UNSET_FLAG (nexthop->flags, NEXTHOP_FLAG_RECURSIVE);
+
+  /* Make lookup prefix. */
+  memset (&p, 0, sizeof (struct prefix_ipv4));
+  p.family = AF_INET;
+  p.prefixlen = IPV4_MAX_PREFIXLEN;
+  p.prefix = nexthop->gate.ipv4;
+
+  /* Lookup table.  */
+  table = vrf_table (AFI_IP, SAFI_UNICAST, 0);
+  if (! table)
+    return 0;
+
+  rn = route_node_match (table, (struct prefix *) &p);
+  while (rn)
+    {
+      route_unlock_node (rn);
+      
+      /* If lookup self prefix return immidiately. */
+      if (rn == top)
+	return 0;
+
+      /* Pick up selected route. */
+      for (match = rn->info; match; match = match->next)
+	if (CHECK_FLAG (match->flags, ZEBRA_FLAG_SELECTED))
+	  break;
+
+      /* If there is no selected route or matched route is EGP, go up
+         tree. */
+      if (! match 
+	  || match->type == ZEBRA_ROUTE_BGP)
+	{
+	  do {
+	    rn = rn->parent;
+	  } while (rn && rn->info == NULL);
+	  if (rn)
+	    route_lock_node (rn);
+	}
+      else
+	{
+	  if (match->type == ZEBRA_ROUTE_CONNECT)
+	    {
+	      /* Directly point connected route. */
+	      newhop = match->nexthop;
+	      if (newhop && nexthop->type == NEXTHOP_TYPE_IPV4)
+		nexthop->ifindex = newhop->ifindex;
+	      
+	      return 1;
+	    }
+	  else if (CHECK_FLAG (rib->flags, ZEBRA_FLAG_INTERNAL))
+	    {
+	      for (newhop = match->nexthop; newhop; newhop = newhop->next)
+		if (CHECK_FLAG (newhop->flags, NEXTHOP_FLAG_FIB)
+		    && ! CHECK_FLAG (newhop->flags, NEXTHOP_FLAG_RECURSIVE))
+		  {
+		    if (set)
+		      {
+			SET_FLAG (nexthop->flags, NEXTHOP_FLAG_RECURSIVE);
+			nexthop->rtype = newhop->type;
+			if (newhop->type == NEXTHOP_TYPE_IPV4 ||
+			    newhop->type == NEXTHOP_TYPE_IPV4_IFINDEX)
+			  nexthop->rgate.ipv4 = newhop->gate.ipv4;
+			if (newhop->type == NEXTHOP_TYPE_IFINDEX
+			    || newhop->type == NEXTHOP_TYPE_IFNAME
+			    || newhop->type == NEXTHOP_TYPE_IPV4_IFINDEX)
+			  nexthop->rifindex = newhop->ifindex;
+		      }
+		    return 1;
+		  }
+	      return 0;
+	    }
+	  else
+	    {
+	      return 0;
+	    }
+	}
+    }
+  return 0;
+}
+
+#ifdef HAVE_IPV6
+/* If force flag is not set, do not modify falgs at all for uninstall
+   the route from FIB. */
+int
+nexthop_active_ipv6 (struct rib *rib, struct nexthop *nexthop, int set,
+		     struct route_node *top)
+{
+  struct prefix_ipv6 p;
+  struct route_table *table;
+  struct route_node *rn;
+  struct rib *match;
+  struct nexthop *newhop;
+
+  if (nexthop->type == NEXTHOP_TYPE_IPV6)
+    nexthop->ifindex = 0;
+
+  if (set)
+    UNSET_FLAG (nexthop->flags, NEXTHOP_FLAG_RECURSIVE);
+
+  /* Make lookup prefix. */
+  memset (&p, 0, sizeof (struct prefix_ipv6));
+  p.family = AF_INET6;
+  p.prefixlen = IPV6_MAX_PREFIXLEN;
+  p.prefix = nexthop->gate.ipv6;
+
+  /* Lookup table.  */
+  table = vrf_table (AFI_IP6, SAFI_UNICAST, 0);
+  if (! table)
+    return 0;
+
+  rn = route_node_match (table, (struct prefix *) &p);
+  while (rn)
+    {
+      route_unlock_node (rn);
+      
+      /* If lookup self prefix return immidiately. */
+      if (rn == top)
+	return 0;
+
+      /* Pick up selected route. */
+      for (match = rn->info; match; match = match->next)
+	if (CHECK_FLAG (match->flags, ZEBRA_FLAG_SELECTED))
+	  break;
+
+      /* If there is no selected route or matched route is EGP, go up
+         tree. */
+      if (! match
+	  || match->type == ZEBRA_ROUTE_BGP)
+	{
+	  do {
+	    rn = rn->parent;
+	  } while (rn && rn->info == NULL);
+	  if (rn)
+	    route_lock_node (rn);
+	}
+      else
+	{
+	  if (match->type == ZEBRA_ROUTE_CONNECT)
+	    {
+	      /* Directly point connected route. */
+	      newhop = match->nexthop;
+
+	      if (newhop && nexthop->type == NEXTHOP_TYPE_IPV6)
+		nexthop->ifindex = newhop->ifindex;
+	      
+	      return 1;
+	    }
+	  else if (CHECK_FLAG (rib->flags, ZEBRA_FLAG_INTERNAL))
+	    {
+	      for (newhop = match->nexthop; newhop; newhop = newhop->next)
+		if (CHECK_FLAG (newhop->flags, NEXTHOP_FLAG_FIB)
+		    && ! CHECK_FLAG (newhop->flags, NEXTHOP_FLAG_RECURSIVE))
+		  {
+		    if (set)
+		      {
+			SET_FLAG (nexthop->flags, NEXTHOP_FLAG_RECURSIVE);
+			nexthop->rtype = newhop->type;
+			if (newhop->type == NEXTHOP_TYPE_IPV6
+			    || newhop->type == NEXTHOP_TYPE_IPV6_IFINDEX
+			    || newhop->type == NEXTHOP_TYPE_IPV6_IFNAME)
+			  nexthop->rgate.ipv6 = newhop->gate.ipv6;
+			if (newhop->type == NEXTHOP_TYPE_IFINDEX
+			    || newhop->type == NEXTHOP_TYPE_IFNAME
+			    || newhop->type == NEXTHOP_TYPE_IPV6_IFINDEX
+			    || newhop->type == NEXTHOP_TYPE_IPV6_IFNAME)
+			  nexthop->rifindex = newhop->ifindex;
+		      }
+		    return 1;
+		  }
+	      return 0;
+	    }
+	  else
+	    {
+	      return 0;
+	    }
+	}
+    }
+  return 0;
+}
+#endif /* HAVE_IPV6 */
+
+struct rib *
+rib_match_ipv4 (struct in_addr addr)
+{
+  struct prefix_ipv4 p;
+  struct route_table *table;
+  struct route_node *rn;
+  struct rib *match;
+  struct nexthop *newhop;
+
+  /* Lookup table.  */
+  table = vrf_table (AFI_IP, SAFI_UNICAST, 0);
+  if (! table)
+    return 0;
+
+  memset (&p, 0, sizeof (struct prefix_ipv4));
+  p.family = AF_INET;
+  p.prefixlen = IPV4_MAX_PREFIXLEN;
+  p.prefix = addr;
+
+  rn = route_node_match (table, (struct prefix *) &p);
+
+  while (rn)
+    {
+      route_unlock_node (rn);
+      
+      /* Pick up selected route. */
+      for (match = rn->info; match; match = match->next)
+	if (CHECK_FLAG (match->flags, ZEBRA_FLAG_SELECTED))
+	  break;
+
+      /* If there is no selected route or matched route is EGP, go up
+         tree. */
+      if (! match 
+	  || match->type == ZEBRA_ROUTE_BGP)
+	{
+	  do {
+	    rn = rn->parent;
+	  } while (rn && rn->info == NULL);
+	  if (rn)
+	    route_lock_node (rn);
+	}
+      else
+	{
+	  if (match->type == ZEBRA_ROUTE_CONNECT)
+	    /* Directly point connected route. */
+	    return match;
+	  else
+	    {
+	      for (newhop = match->nexthop; newhop; newhop = newhop->next)
+		if (CHECK_FLAG (newhop->flags, NEXTHOP_FLAG_FIB))
+		  return match;
+	      return NULL;
+	    }
+	}
+    }
+  return NULL;
+}
+
+struct rib *
+rib_lookup_ipv4 (struct prefix_ipv4 *p)
+{
+  struct route_table *table;
+  struct route_node *rn;
+  struct rib *match;
+  struct nexthop *nexthop;
+
+  /* Lookup table.  */
+  table = vrf_table (AFI_IP, SAFI_UNICAST, 0);
+  if (! table)
+    return 0;
+
+  rn = route_node_lookup (table, (struct prefix *) p);
+
+  /* No route for this prefix. */
+  if (! rn)
+    return NULL;
+
+  /* Unlock node. */
+  route_unlock_node (rn);
+
+  /* Pick up selected route. */
+  for (match = rn->info; match; match = match->next)
+    if (CHECK_FLAG (match->flags, ZEBRA_FLAG_SELECTED))
+      break;
+
+  if (! match || match->type == ZEBRA_ROUTE_BGP)
+    return NULL;
+
+  if (match->type == ZEBRA_ROUTE_CONNECT)
+    return match;
+  
+  for (nexthop = match->nexthop; nexthop; nexthop = nexthop->next)
+    if (CHECK_FLAG (nexthop->flags, NEXTHOP_FLAG_FIB))
+      return match;
+
+  return NULL;
+}
+
+#ifdef HAVE_IPV6
+struct rib *
+rib_match_ipv6 (struct in6_addr *addr)
+{
+  struct prefix_ipv6 p;
+  struct route_table *table;
+  struct route_node *rn;
+  struct rib *match;
+  struct nexthop *newhop;
+
+  /* Lookup table.  */
+  table = vrf_table (AFI_IP6, SAFI_UNICAST, 0);
+  if (! table)
+    return 0;
+
+  memset (&p, 0, sizeof (struct prefix_ipv6));
+  p.family = AF_INET6;
+  p.prefixlen = IPV6_MAX_PREFIXLEN;
+  IPV6_ADDR_COPY (&p.prefix, addr);
+
+  rn = route_node_match (table, (struct prefix *) &p);
+
+  while (rn)
+    {
+      route_unlock_node (rn);
+      
+      /* Pick up selected route. */
+      for (match = rn->info; match; match = match->next)
+	if (CHECK_FLAG (match->flags, ZEBRA_FLAG_SELECTED))
+	  break;
+
+      /* If there is no selected route or matched route is EGP, go up
+         tree. */
+      if (! match 
+	  || match->type == ZEBRA_ROUTE_BGP)
+	{
+	  do {
+	    rn = rn->parent;
+	  } while (rn && rn->info == NULL);
+	  if (rn)
+	    route_lock_node (rn);
+	}
+      else
+	{
+	  if (match->type == ZEBRA_ROUTE_CONNECT)
+	    /* Directly point connected route. */
+	    return match;
+	  else
+	    {
+	      for (newhop = match->nexthop; newhop; newhop = newhop->next)
+		if (CHECK_FLAG (newhop->flags, NEXTHOP_FLAG_FIB))
+		  return match;
+	      return NULL;
+	    }
+	}
+    }
+  return NULL;
+}
+#endif /* HAVE_IPV6 */
+
+int
+nexthop_active_check (struct route_node *rn, struct rib *rib,
+		      struct nexthop *nexthop, int set)
+{
+  struct interface *ifp;
+
+  switch (nexthop->type)
+    {
+    case NEXTHOP_TYPE_IFINDEX:
+      ifp = if_lookup_by_index (nexthop->ifindex);
+      if (ifp && if_is_up (ifp))
+	SET_FLAG (nexthop->flags, NEXTHOP_FLAG_ACTIVE);
+      else
+	UNSET_FLAG (nexthop->flags, NEXTHOP_FLAG_ACTIVE);
+      break;
+    case NEXTHOP_TYPE_IFNAME:
+    case NEXTHOP_TYPE_IPV6_IFNAME:
+      ifp = if_lookup_by_name (nexthop->ifname);
+      if (ifp && if_is_up (ifp))
+	{
+	  if (set)
+	    nexthop->ifindex = ifp->ifindex;
+	  SET_FLAG (nexthop->flags, NEXTHOP_FLAG_ACTIVE);
+	}
+      else
+	{
+	  if (set)
+	    nexthop->ifindex = 0;
+	  UNSET_FLAG (nexthop->flags, NEXTHOP_FLAG_ACTIVE);
+	}
+      break;
+    case NEXTHOP_TYPE_IPV4:
+    case NEXTHOP_TYPE_IPV4_IFINDEX:
+      if (nexthop_active_ipv4 (rib, nexthop, set, rn))
+	SET_FLAG (nexthop->flags, NEXTHOP_FLAG_ACTIVE);
+      else
+	UNSET_FLAG (nexthop->flags, NEXTHOP_FLAG_ACTIVE);
+      break;
+#ifdef HAVE_IPV6
+    case NEXTHOP_TYPE_IPV6:
+      if (nexthop_active_ipv6 (rib, nexthop, set, rn))
+	SET_FLAG (nexthop->flags, NEXTHOP_FLAG_ACTIVE);
+      else
+	UNSET_FLAG (nexthop->flags, NEXTHOP_FLAG_ACTIVE);
+      break;
+    case NEXTHOP_TYPE_IPV6_IFINDEX:
+      if (IN6_IS_ADDR_LINKLOCAL (&nexthop->gate.ipv6))
+	{
+	  ifp = if_lookup_by_index (nexthop->ifindex);
+	  if (ifp && if_is_up (ifp))
+	    SET_FLAG (nexthop->flags, NEXTHOP_FLAG_ACTIVE);
+	  else
+	    UNSET_FLAG (nexthop->flags, NEXTHOP_FLAG_ACTIVE);
+	}
+      else
+	{
+	  if (nexthop_active_ipv6 (rib, nexthop, set, rn))
+	    SET_FLAG (nexthop->flags, NEXTHOP_FLAG_ACTIVE);
+	  else
+	    UNSET_FLAG (nexthop->flags, NEXTHOP_FLAG_ACTIVE);
+	}
+      break;
+#endif /* HAVE_IPV6 */
+    case NEXTHOP_TYPE_BLACKHOLE:
+      SET_FLAG (nexthop->flags, NEXTHOP_FLAG_ACTIVE);
+      break;
+    default:
+      break;
+    }
+  return CHECK_FLAG (nexthop->flags, NEXTHOP_FLAG_ACTIVE);
+}
+
+int
+nexthop_active_update (struct route_node *rn, struct rib *rib, int set)
+{
+  struct nexthop *nexthop;
+  int active;
+
+  rib->nexthop_active_num = 0;
+  UNSET_FLAG (rib->flags, ZEBRA_FLAG_CHANGED);
+
+  for (nexthop = rib->nexthop; nexthop; nexthop = nexthop->next)
+    {
+      active = CHECK_FLAG (nexthop->flags, NEXTHOP_FLAG_ACTIVE);
+      rib->nexthop_active_num += nexthop_active_check (rn, rib, nexthop, set);
+      if (active != CHECK_FLAG (nexthop->flags, NEXTHOP_FLAG_ACTIVE))
+	SET_FLAG (rib->flags, ZEBRA_FLAG_CHANGED);
+    }
+  return rib->nexthop_active_num;
+}
+
+#define RIB_SYSTEM_ROUTE(R) \
+        ((R)->type == ZEBRA_ROUTE_KERNEL || (R)->type == ZEBRA_ROUTE_CONNECT)
+
+void
+newrib_free (struct rib *rib)
+{
+  struct nexthop *nexthop;
+  struct nexthop *next;
+
+  for (nexthop = rib->nexthop; nexthop; nexthop = next)
+    {
+      next = nexthop->next;
+      nexthop_free (nexthop);
+    }
+  XFREE (MTYPE_RIB, rib);
+}
+
+void
+rib_install_kernel (struct route_node *rn, struct rib *rib)
+{
+  int ret = 0;
+  struct nexthop *nexthop;
+
+  switch (PREFIX_FAMILY (&rn->p))
+    {
+    case AF_INET:
+      ret = kernel_add_ipv4 (&rn->p, rib);
+      break;
+#ifdef HAVE_IPV6
+    case AF_INET6:
+      ret = kernel_add_ipv6 (&rn->p, rib);
+      break;
+#endif /* HAVE_IPV6 */
+    }
+
+  if (ret < 0)
+    {
+      for (nexthop = rib->nexthop; nexthop; nexthop = nexthop->next)
+	UNSET_FLAG (nexthop->flags, NEXTHOP_FLAG_FIB);
+    }
+}
+
+/* Uninstall the route from kernel. */
+int
+rib_uninstall_kernel (struct route_node *rn, struct rib *rib)
+{
+  int ret = 0;
+  struct nexthop *nexthop;
+
+  switch (PREFIX_FAMILY (&rn->p))
+    {
+    case AF_INET:
+      ret = kernel_delete_ipv4 (&rn->p, rib);
+      break;
+#ifdef HAVE_IPV6
+    case AF_INET6:
+      ret = kernel_delete_ipv6 (&rn->p, rib);
+      break;
+#endif /* HAVE_IPV6 */
+    }
+
+  for (nexthop = rib->nexthop; nexthop; nexthop = nexthop->next)
+    UNSET_FLAG (nexthop->flags, NEXTHOP_FLAG_FIB);
+
+  return ret;
+}
+
+/* Uninstall the route from kernel. */
+void
+rib_uninstall (struct route_node *rn, struct rib *rib)
+{
+  if (CHECK_FLAG (rib->flags, ZEBRA_FLAG_SELECTED))
+    {
+      redistribute_delete (&rn->p, rib);
+      if (! RIB_SYSTEM_ROUTE (rib))
+	rib_uninstall_kernel (rn, rib);
+      UNSET_FLAG (rib->flags, ZEBRA_FLAG_SELECTED);
+    }
+}
+
+/* Core function for processing routing information base. */
+void
+rib_process (struct route_node *rn, struct rib *del)
+{
+  struct rib *rib;
+  struct rib *next;
+  struct rib *fib = NULL;
+  struct rib *select = NULL;
+
+  for (rib = rn->info; rib; rib = next)
+    {
+      next = rib->next;
+
+      /* Currently installed rib. */
+      if (CHECK_FLAG (rib->flags, ZEBRA_FLAG_SELECTED))
+	fib = rib;
+
+      /* Skip unreachable nexthop. */
+      if (! nexthop_active_update (rn, rib, 0))
+	continue;
+
+      /* Infinit distance. */
+      if (rib->distance == DISTANCE_INFINITY)
+	continue;
+
+      /* Newly selected rib. */
+      if (! select || rib->distance < select->distance 
+	  || rib->type == ZEBRA_ROUTE_CONNECT)
+	select = rib;
+    }
+
+  /* Deleted route check. */
+  if (del && CHECK_FLAG (del->flags, ZEBRA_FLAG_SELECTED))
+    fib = del;
+
+  /* Same route is selected. */
+  if (select && select == fib)
+    {
+      if (CHECK_FLAG (select->flags, ZEBRA_FLAG_CHANGED))
+	{
+	  redistribute_delete (&rn->p, select);
+	  if (! RIB_SYSTEM_ROUTE (select))
+	    rib_uninstall_kernel (rn, select);
+
+	  /* Set real nexthop. */
+	  nexthop_active_update (rn, select, 1);
+  
+	  if (! RIB_SYSTEM_ROUTE (select))
+	    rib_install_kernel (rn, select);
+	  redistribute_add (&rn->p, select);
+	}
+      return;
+    }
+
+  /* Uninstall old rib from forwarding table. */
+  if (fib)
+    {
+      redistribute_delete (&rn->p, fib);
+      if (! RIB_SYSTEM_ROUTE (fib))
+	rib_uninstall_kernel (rn, fib);
+      UNSET_FLAG (fib->flags, ZEBRA_FLAG_SELECTED);
+
+      /* Set real nexthop. */
+      nexthop_active_update (rn, fib, 1);
+    }
+
+  /* Install new rib into forwarding table. */
+  if (select)
+    {
+      /* Set real nexthop. */
+      nexthop_active_update (rn, select, 1);
+
+      if (! RIB_SYSTEM_ROUTE (select))
+	rib_install_kernel (rn, select);
+      SET_FLAG (select->flags, ZEBRA_FLAG_SELECTED);
+      redistribute_add (&rn->p, select);
+    }
+}
+
+/* Add RIB to head of the route node. */
+void
+rib_addnode (struct route_node *rn, struct rib *rib)
+{
+  struct rib *head;
+
+  head = rn->info;
+  if (head)
+    head->prev = rib;
+  rib->next = head;
+  rn->info = rib;
+}
+
+void
+rib_delnode (struct route_node *rn, struct rib *rib)
+{
+  if (rib->next)
+    rib->next->prev = rib->prev;
+  if (rib->prev)
+    rib->prev->next = rib->next;
+  else
+    rn->info = rib->next;
+}
+
+int
+rib_add_ipv4 (int type, int flags, struct prefix_ipv4 *p, 
+	      struct in_addr *gate, unsigned int ifindex, u_int32_t vrf_id,
+	      u_int32_t metric, u_char distance)
+{
+  struct rib *rib;
+  struct rib *same = NULL;
+  struct route_table *table;
+  struct route_node *rn;
+  struct nexthop *nexthop;
+
+  /* Lookup table.  */
+  table = vrf_table (AFI_IP, SAFI_UNICAST, 0);
+  if (! table)
+    return 0;
+
+  /* Make it sure prefixlen is applied to the prefix. */
+  apply_mask_ipv4 (p);
+
+  /* Set default distance by route type. */
+  if (distance == 0)
+    {
+      distance = route_info[type].distance;
+
+      /* iBGP distance is 200. */
+      if (type == ZEBRA_ROUTE_BGP && CHECK_FLAG (flags, ZEBRA_FLAG_IBGP))
+	distance = 200;
+    }
+
+  /* Lookup route node.*/
+  rn = route_node_get (table, (struct prefix *) p);
+
+  /* If same type of route are installed, treat it as a implicit
+     withdraw. */
+  for (rib = rn->info; rib; rib = rib->next)
+    {
+      if (rib->type == ZEBRA_ROUTE_CONNECT)
+	{
+	  nexthop = rib->nexthop;
+
+	  /* Duplicate connected route comes in. */
+	  if (rib->type == type
+	      && nexthop && nexthop->type == NEXTHOP_TYPE_IFINDEX
+	      && nexthop->ifindex == ifindex)
+	    {
+	      rib->refcnt++;
+	      return 0 ;
+	    }
+	}
+      else if (rib->type == type)
+	{
+	  same = rib;
+	  rib_delnode (rn, same);
+	  route_unlock_node (rn);
+	  break;
+	}
+    }
+
+  /* Allocate new rib structure. */
+  rib = XMALLOC (MTYPE_RIB, sizeof (struct rib));
+  memset (rib, 0, sizeof (struct rib));
+  rib->type = type;
+  rib->distance = distance;
+  rib->flags = flags;
+  rib->metric = metric;
+  rib->nexthop_num = 0;
+  rib->uptime = time (NULL);
+
+  /* Nexthop settings. */
+  if (gate)
+    {
+      if (ifindex)
+	nexthop_ipv4_ifindex_add (rib, gate, ifindex);
+      else
+	nexthop_ipv4_add (rib, gate);
+    }
+  else
+    nexthop_ifindex_add (rib, ifindex);
+
+  /* If this route is kernel route, set FIB flag to the route. */
+  if (type == ZEBRA_ROUTE_KERNEL || type == ZEBRA_ROUTE_CONNECT)
+    for (nexthop = rib->nexthop; nexthop; nexthop = nexthop->next)
+      SET_FLAG (nexthop->flags, NEXTHOP_FLAG_FIB);
+
+  /* Link new rib to node.*/
+  rib_addnode (rn, rib);
+
+  /* Process this route node. */
+  rib_process (rn, same);
+
+  /* Free implicit route.*/
+  if (same)
+    newrib_free (same);
+
+  return 0;
+}
+
+int
+rib_add_ipv4_multipath (struct prefix_ipv4 *p, struct rib *rib)
+{
+  struct route_table *table;
+  struct route_node *rn;
+  struct rib *same;
+  struct nexthop *nexthop;
+
+  /* Lookup table.  */
+  table = vrf_table (AFI_IP, SAFI_UNICAST, 0);
+  if (! table)
+    return 0;
+
+  /* Make it sure prefixlen is applied to the prefix. */
+  apply_mask_ipv4 (p);
+
+  /* Set default distance by route type. */
+  if (rib->distance == 0)
+    {
+      rib->distance = route_info[rib->type].distance;
+
+      /* iBGP distance is 200. */
+      if (rib->type == ZEBRA_ROUTE_BGP 
+	  && CHECK_FLAG (rib->flags, ZEBRA_FLAG_IBGP))
+	rib->distance = 200;
+    }
+
+  /* Lookup route node.*/
+  rn = route_node_get (table, (struct prefix *) p);
+
+  /* If same type of route are installed, treat it as a implicit
+     withdraw. */
+  for (same = rn->info; same; same = same->next)
+    {
+      if (same->type == rib->type && same->table == rib->table
+	  && same->type != ZEBRA_ROUTE_CONNECT)
+	{
+	  rib_delnode (rn, same);
+	  route_unlock_node (rn);
+	  break;
+	}
+    }
+
+  /* If this route is kernel route, set FIB flag to the route. */
+  if (rib->type == ZEBRA_ROUTE_KERNEL || rib->type == ZEBRA_ROUTE_CONNECT)
+    for (nexthop = rib->nexthop; nexthop; nexthop = nexthop->next)
+      SET_FLAG (nexthop->flags, NEXTHOP_FLAG_FIB);
+
+  /* Link new rib to node.*/
+  rib_addnode (rn, rib);
+
+  /* Process this route node. */
+  rib_process (rn, same);
+
+  /* Free implicit route.*/
+  if (same)
+    newrib_free (same);
+
+  return 0;
+}
+
+int
+rib_delete_ipv4 (int type, int flags, struct prefix_ipv4 *p,
+		 struct in_addr *gate, unsigned int ifindex, u_int32_t vrf_id)
+{
+  struct route_table *table;
+  struct route_node *rn;
+  struct rib *rib;
+  struct rib *fib = NULL;
+  struct rib *same = NULL;
+  struct nexthop *nexthop;
+  char buf1[BUFSIZ];
+  char buf2[BUFSIZ];
+
+  /* Lookup table.  */
+  table = vrf_table (AFI_IP, SAFI_UNICAST, 0);
+  if (! table)
+    return 0;
+
+  /* Apply mask. */
+  apply_mask_ipv4 (p);
+
+  /* Lookup route node. */
+  rn = route_node_lookup (table, (struct prefix *) p);
+  if (! rn)
+    {
+      if (IS_ZEBRA_DEBUG_KERNEL)
+	{
+	  if (gate)
+	    zlog_info ("route %s/%d via %s ifindex %d doesn't exist in rib",
+		       inet_ntop (AF_INET, &p->prefix, buf1, BUFSIZ),
+		       p->prefixlen,
+		       inet_ntop (AF_INET, gate, buf2, BUFSIZ),
+		       ifindex);
+	  else
+	    zlog_info ("route %s/%d ifindex %d doesn't exist in rib",
+		       inet_ntop (AF_INET, &p->prefix, buf1, BUFSIZ),
+		       p->prefixlen,
+		       ifindex);
+	}
+      return ZEBRA_ERR_RTNOEXIST;
+    }
+
+  /* Lookup same type route. */
+  for (rib = rn->info; rib; rib = rib->next)
+    {
+      if (CHECK_FLAG (rib->flags, ZEBRA_FLAG_SELECTED))
+	fib = rib;
+
+      if (rib->type == ZEBRA_ROUTE_CONNECT)
+	{
+	  nexthop = rib->nexthop;
+
+	  if (rib->type == type
+	      && nexthop && nexthop->type == NEXTHOP_TYPE_IFINDEX
+	      && nexthop->ifindex == ifindex)
+	    {
+	      if (rib->refcnt)
+		{
+		  rib->refcnt--;
+		  route_unlock_node (rn);
+		  route_unlock_node (rn);
+		  return 0;
+		}
+	      same = rib;
+	      break;
+	    }
+	}
+      else
+	{
+	  if (rib->type == type)
+	    {
+	      same = rib;
+	      break;
+	    }
+	}
+    }
+
+  /* If same type of route can't be found and this message is from
+     kernel. */
+  if (! same)
+    {
+      if (fib && type == ZEBRA_ROUTE_KERNEL)
+	{
+	  /* Unset flags. */
+	  for (nexthop = fib->nexthop; nexthop; nexthop = nexthop->next)
+	    UNSET_FLAG (nexthop->flags, NEXTHOP_FLAG_FIB);
+
+	  UNSET_FLAG (fib->flags, ZEBRA_FLAG_SELECTED);
+	}
+      else
+	{
+	  if (IS_ZEBRA_DEBUG_KERNEL)
+	    {
+	      if (gate)
+		zlog_info ("route %s/%d via %s ifindex %d type %d doesn't exist in rib",
+			   inet_ntop (AF_INET, &p->prefix, buf1, BUFSIZ),
+			   p->prefixlen,
+			   inet_ntop (AF_INET, gate, buf2, BUFSIZ),
+			   ifindex,
+			   type);
+	      else
+		zlog_info ("route %s/%d ifindex %d type %d doesn't exist in rib",
+			   inet_ntop (AF_INET, &p->prefix, buf1, BUFSIZ),
+			   p->prefixlen,
+			   ifindex,
+			   type);
+	    }
+	  route_unlock_node (rn);
+	  return ZEBRA_ERR_RTNOEXIST;
+	}
+    }
+
+  if (same)
+    rib_delnode (rn, same);
+
+  /* Process changes. */
+  rib_process (rn, same);
+
+  if (same)
+    {
+      newrib_free (same);
+      route_unlock_node (rn);
+    }
+
+  route_unlock_node (rn);
+
+  return 0;
+}
+
+/* Install static route into rib. */
+void
+static_install_ipv4 (struct prefix *p, struct static_ipv4 *si)
+{
+  struct rib *rib;
+  struct route_node *rn;
+  struct route_table *table;
+
+  /* Lookup table.  */
+  table = vrf_table (AFI_IP, SAFI_UNICAST, 0);
+  if (! table)
+    return;
+
+  /* Lookup existing route */
+  rn = route_node_get (table, p);
+  for (rib = rn->info; rib; rib = rib->next)
+    if (rib->type == ZEBRA_ROUTE_STATIC && rib->distance == si->distance)
+      break;
+
+  if (rib)
+    {
+      /* Same distance static route is there.  Update it with new
+         nexthop. */
+      rib_uninstall (rn, rib);
+      route_unlock_node (rn);
+
+      switch (si->type)
+	{
+	case STATIC_IPV4_GATEWAY:
+	  nexthop_ipv4_add (rib, &si->gate.ipv4);
+	  break;
+	case STATIC_IPV4_IFNAME:
+	  nexthop_ifname_add (rib, si->gate.ifname);
+	  break;
+	case STATIC_IPV4_BLACKHOLE:
+	  nexthop_blackhole_add (rib);
+	  break;
+	}
+      rib_process (rn, NULL);
+    }
+  else
+    {
+      /* This is new static route. */
+      rib = XMALLOC (MTYPE_RIB, sizeof (struct rib));
+      memset (rib, 0, sizeof (struct rib));
+
+      rib->type = ZEBRA_ROUTE_STATIC;
+      rib->distance = si->distance;
+      rib->metric = 0;
+      rib->nexthop_num = 0;
+
+      switch (si->type)
+	{
+	case STATIC_IPV4_GATEWAY:
+	  nexthop_ipv4_add (rib, &si->gate.ipv4);
+	  break;
+	case STATIC_IPV4_IFNAME:
+	  nexthop_ifname_add (rib, si->gate.ifname);
+	  break;
+	case STATIC_IPV4_BLACKHOLE:
+	  nexthop_blackhole_add (rib);
+	  break;
+	}
+
+      /* Link this rib to the tree. */
+      rib_addnode (rn, rib);
+
+      /* Process this prefix. */
+      rib_process (rn, NULL);
+    }
+}
+
+int
+static_ipv4_nexthop_same (struct nexthop *nexthop, struct static_ipv4 *si)
+{
+  if (nexthop->type == NEXTHOP_TYPE_IPV4
+      && si->type == STATIC_IPV4_GATEWAY
+      && IPV4_ADDR_SAME (&nexthop->gate.ipv4, &si->gate.ipv4))
+    return 1;
+  if (nexthop->type == NEXTHOP_TYPE_IFNAME
+      && si->type == STATIC_IPV4_IFNAME
+      && strcmp (nexthop->ifname, si->gate.ifname) == 0)
+    return 1;
+  if (nexthop->type == NEXTHOP_TYPE_BLACKHOLE
+      && si->type == STATIC_IPV4_BLACKHOLE)
+    return 1;
+  return 0;;
+}
+
+/* Uninstall static route from RIB. */
+void
+static_uninstall_ipv4 (struct prefix *p, struct static_ipv4 *si)
+{
+  struct route_node *rn;
+  struct rib *rib;
+  struct nexthop *nexthop;
+  struct route_table *table;
+
+  /* Lookup table.  */
+  table = vrf_table (AFI_IP, SAFI_UNICAST, 0);
+  if (! table)
+    return;
+
+  /* Lookup existing route with type and distance. */
+  rn = route_node_lookup (table, p);
+  if (! rn)
+    return;
+
+  for (rib = rn->info; rib; rib = rib->next)
+    if (rib->type == ZEBRA_ROUTE_STATIC && rib->distance == si->distance)
+      break;
+
+  if (! rib)
+    {
+      route_unlock_node (rn);
+      return;
+    }
+
+  /* Lookup nexthop. */
+  for (nexthop = rib->nexthop; nexthop; nexthop = nexthop->next)
+    if (static_ipv4_nexthop_same (nexthop, si))
+      break;
+
+  /* Can't find nexthop. */
+  if (! nexthop)
+    {
+      route_unlock_node (rn);
+      return;
+    }
+  
+  /* Check nexthop. */
+  if (rib->nexthop_num == 1)
+    {
+      rib_delnode (rn, rib);
+      rib_process (rn, rib);
+      newrib_free (rib);
+      route_unlock_node (rn);
+    }
+  else
+    {
+      rib_uninstall (rn, rib);
+      nexthop_delete (rib, nexthop);
+      nexthop_free (nexthop);
+      rib_process (rn, rib);
+    }
+
+  /* Unlock node. */
+  route_unlock_node (rn);
+}
+
+/* Add static route into static route configuration. */
+int
+static_add_ipv4 (struct prefix *p, struct in_addr *gate, char *ifname,
+		 u_char distance, u_int32_t vrf_id)
+{
+  u_char type = 0;
+  struct route_node *rn;
+  struct static_ipv4 *si;
+  struct static_ipv4 *pp;
+  struct static_ipv4 *cp;
+  struct static_ipv4 *update = NULL;
+  struct route_table *stable;
+
+  /* Lookup table.  */
+  stable = vrf_static_table (AFI_IP, SAFI_UNICAST, vrf_id);
+  if (! stable)
+    return -1;
+  
+  /* Lookup static route prefix. */
+  rn = route_node_get (stable, p);
+
+  /* Make flags. */
+  if (gate)
+    type = STATIC_IPV4_GATEWAY;
+  else if (ifname)
+    type = STATIC_IPV4_IFNAME;
+  else
+    type = STATIC_IPV4_BLACKHOLE;
+
+  /* Do nothing if there is a same static route.  */
+  for (si = rn->info; si; si = si->next)
+    {
+      if (type == si->type
+	  && (! gate || IPV4_ADDR_SAME (gate, &si->gate.ipv4))
+	  && (! ifname || strcmp (ifname, si->gate.ifname) == 0))
+	{
+	  if (distance == si->distance)
+	    {
+	      route_unlock_node (rn);
+	      return 0;
+	    }
+	  else
+	    update = si;
+	}
+    }
+
+  /* Distance chaged.  */
+  if (update)
+    static_delete_ipv4 (p, gate, ifname, update->distance, vrf_id);
+
+  /* Make new static route structure. */
+  si = XMALLOC (MTYPE_STATIC_IPV4, sizeof (struct static_ipv4));
+  memset (si, 0, sizeof (struct static_ipv4));
+
+  si->type = type;
+  si->distance = distance;
+
+  if (gate)
+    si->gate.ipv4 = *gate;
+  if (ifname)
+    si->gate.ifname = XSTRDUP (0, ifname);
+
+  /* Add new static route information to the tree with sort by
+     distance value and gateway address. */
+  for (pp = NULL, cp = rn->info; cp; pp = cp, cp = cp->next)
+    {
+      if (si->distance < cp->distance)
+	break;
+      if (si->distance > cp->distance)
+	continue;
+      if (si->type == STATIC_IPV4_GATEWAY && cp->type == STATIC_IPV4_GATEWAY)
+	{
+	  if (ntohl (si->gate.ipv4.s_addr) < ntohl (cp->gate.ipv4.s_addr))
+	    break;
+	  if (ntohl (si->gate.ipv4.s_addr) > ntohl (cp->gate.ipv4.s_addr))
+	    continue;
+	}
+    }
+
+  /* Make linked list. */
+  if (pp)
+    pp->next = si;
+  else
+    rn->info = si;
+  if (cp)
+    cp->prev = si;
+  si->prev = pp;
+  si->next = cp;
+
+  /* Install into rib. */
+  static_install_ipv4 (p, si);
+
+  return 1;
+}
+
+/* Delete static route from static route configuration. */
+int
+static_delete_ipv4 (struct prefix *p, struct in_addr *gate, char *ifname,
+		    u_char distance, u_int32_t vrf_id)
+{
+  u_char type = 0;
+  struct route_node *rn;
+  struct static_ipv4 *si;
+  struct route_table *stable;
+
+  /* Lookup table.  */
+  stable = vrf_static_table (AFI_IP, SAFI_UNICAST, vrf_id);
+  if (! stable)
+    return -1;
+
+  /* Lookup static route prefix. */
+  rn = route_node_lookup (stable, p);
+  if (! rn)
+    return 0;
+
+  /* Make flags. */
+  if (gate)
+    type = STATIC_IPV4_GATEWAY;
+  else if (ifname)
+    type = STATIC_IPV4_IFNAME;
+  else
+    type = STATIC_IPV4_BLACKHOLE;
+
+  /* Find same static route is the tree */
+  for (si = rn->info; si; si = si->next)
+    if (type == si->type
+	&& (! gate || IPV4_ADDR_SAME (gate, &si->gate.ipv4))
+	&& (! ifname || strcmp (ifname, si->gate.ifname) == 0))
+      break;
+
+  /* Can't find static route. */
+  if (! si)
+    {
+      route_unlock_node (rn);
+      return 0;
+    }
+
+  /* Install into rib. */
+  static_uninstall_ipv4 (p, si);
+
+  /* Unlink static route from linked list. */
+  if (si->prev)
+    si->prev->next = si->next;
+  else
+    rn->info = si->next;
+  if (si->next)
+    si->next->prev = si->prev;
+  
+  /* Free static route configuration. */
+  XFREE (MTYPE_STATIC_IPV4, si);
+
+  return 1;
+}
+
+
+#ifdef HAVE_IPV6
+int
+rib_bogus_ipv6 (int type, struct prefix_ipv6 *p,
+		struct in6_addr *gate, unsigned int ifindex, int table)
+{
+  if (type == ZEBRA_ROUTE_CONNECT && IN6_IS_ADDR_UNSPECIFIED (&p->prefix))
+    return 1;
+  if (type == ZEBRA_ROUTE_KERNEL && IN6_IS_ADDR_UNSPECIFIED (&p->prefix)
+      && p->prefixlen == 96 && gate && IN6_IS_ADDR_UNSPECIFIED (gate))
+    {
+      kernel_delete_ipv6_old (p, gate, ifindex, 0, table);
+      return 1;
+    }
+  return 0;
+}
+
+int
+rib_add_ipv6 (int type, int flags, struct prefix_ipv6 *p,
+	      struct in6_addr *gate, unsigned int ifindex, u_int32_t vrf_id)
+{
+  struct rib *rib;
+  struct rib *same = NULL;
+  struct route_table *table;
+  struct route_node *rn;
+  struct nexthop *nexthop;
+
+  int distance;
+  u_int32_t metric = 0;
+
+  /* Lookup table.  */
+  table = vrf_table (AFI_IP6, SAFI_UNICAST, 0);
+  if (! table)
+    return 0;
+
+  /* Make sure mask is applied. */
+  apply_mask_ipv6 (p);
+
+  /* Set default distance by route type. */
+  distance = route_info[type].distance;
+  
+  if (type == ZEBRA_ROUTE_BGP && CHECK_FLAG (flags, ZEBRA_FLAG_IBGP))
+    distance = 200;
+
+  /* Filter bogus route. */
+  if (rib_bogus_ipv6 (type, p, gate, ifindex, 0))
+    return 0;
+
+  /* Lookup route node.*/
+  rn = route_node_get (table, (struct prefix *) p);
+
+  /* If same type of route are installed, treat it as a implicit
+     withdraw. */
+  for (rib = rn->info; rib; rib = rib->next)
+    {
+      if (rib->type == ZEBRA_ROUTE_CONNECT)
+	{
+	  nexthop = rib->nexthop;
+
+	  if (rib->type == type
+	      && nexthop && nexthop->type == NEXTHOP_TYPE_IFINDEX
+	      && nexthop->ifindex == ifindex)
+	  {
+	    rib->refcnt++;
+	    return 0;
+	  }
+	}
+      else if (rib->type == type)
+	{
+	  same = rib;
+	  rib_delnode (rn, same);
+	  route_unlock_node (rn);
+	  break;
+	}
+    }
+
+  /* Allocate new rib structure. */
+  rib = XMALLOC (MTYPE_RIB, sizeof (struct rib));
+  memset (rib, 0, sizeof (struct rib));
+  rib->type = type;
+  rib->distance = distance;
+  rib->flags = flags;
+  rib->metric = metric;
+  rib->nexthop_num = 0;
+  rib->uptime = time (NULL);
+
+  /* Nexthop settings. */
+  if (gate)
+    {
+      if (ifindex)
+	nexthop_ipv6_ifindex_add (rib, gate, ifindex);
+      else
+	nexthop_ipv6_add (rib, gate);
+    }
+  else
+    nexthop_ifindex_add (rib, ifindex);
+
+  /* If this route is kernel route, set FIB flag to the route. */
+  if (type == ZEBRA_ROUTE_KERNEL || type == ZEBRA_ROUTE_CONNECT)
+    for (nexthop = rib->nexthop; nexthop; nexthop = nexthop->next)
+      SET_FLAG (nexthop->flags, NEXTHOP_FLAG_FIB);
+
+  /* Link new rib to node.*/
+  rib_addnode (rn, rib);
+
+  /* Process this route node. */
+  rib_process (rn, same);
+
+  /* Free implicit route.*/
+  if (same)
+    newrib_free (same);
+
+  return 0;
+}
+
+int
+rib_delete_ipv6 (int type, int flags, struct prefix_ipv6 *p,
+		 struct in6_addr *gate, unsigned int ifindex, u_int32_t vrf_id)
+{
+  struct route_table *table;
+  struct route_node *rn;
+  struct rib *rib;
+  struct rib *fib = NULL;
+  struct rib *same = NULL;
+  struct nexthop *nexthop;
+  char buf1[BUFSIZ];
+  char buf2[BUFSIZ];
+
+  /* Apply mask. */
+  apply_mask_ipv6 (p);
+
+  /* Lookup table.  */
+  table = vrf_table (AFI_IP6, SAFI_UNICAST, 0);
+  if (! table)
+    return 0;
+
+  /* Lookup route node. */
+  rn = route_node_lookup (table, (struct prefix *) p);
+  if (! rn)
+    {
+      if (IS_ZEBRA_DEBUG_KERNEL)
+	{
+	  if (gate)
+	    zlog_info ("route %s/%d via %s ifindex %d doesn't exist in rib",
+		       inet_ntop (AF_INET6, &p->prefix, buf1, BUFSIZ),
+		       p->prefixlen,
+		       inet_ntop (AF_INET6, gate, buf2, BUFSIZ),
+		       ifindex);
+	  else
+	    zlog_info ("route %s/%d ifindex %d doesn't exist in rib",
+		       inet_ntop (AF_INET6, &p->prefix, buf1, BUFSIZ),
+		       p->prefixlen,
+		       ifindex);
+	}
+      return ZEBRA_ERR_RTNOEXIST;
+    }
+
+  /* Lookup same type route. */
+  for (rib = rn->info; rib; rib = rib->next)
+    {
+      if (CHECK_FLAG (rib->flags, ZEBRA_FLAG_SELECTED))
+	fib = rib;
+
+      if (rib->type == ZEBRA_ROUTE_CONNECT)
+	{
+	  nexthop = rib->nexthop;
+
+	  if (rib->type == type
+	      && nexthop && nexthop->type == NEXTHOP_TYPE_IFINDEX
+	      && nexthop->ifindex == ifindex)
+	    {
+	      if (rib->refcnt)
+		{
+		  rib->refcnt--;
+		  route_unlock_node (rn);
+		  route_unlock_node (rn);
+		  return 0;
+		}
+	      same = rib;
+	      break;
+	    }
+	}
+      else
+	{
+	  if (rib->type == type)
+	    {
+	      same = rib;
+	      break;
+	    }
+	}
+    }
+
+  /* If same type of route can't be found and this message is from
+     kernel. */
+  if (! same)
+    {
+      if (fib && type == ZEBRA_ROUTE_KERNEL)
+	{
+	  /* Unset flags. */
+	  for (nexthop = fib->nexthop; nexthop; nexthop = nexthop->next)
+	    UNSET_FLAG (nexthop->flags, NEXTHOP_FLAG_FIB);
+
+	  UNSET_FLAG (fib->flags, ZEBRA_FLAG_SELECTED);
+	}
+      else
+	{
+	  if (IS_ZEBRA_DEBUG_KERNEL)
+	    {
+	      if (gate)
+		zlog_info ("route %s/%d via %s ifindex %d type %d doesn't exist in rib",
+			   inet_ntop (AF_INET6, &p->prefix, buf1, BUFSIZ),
+			   p->prefixlen,
+			   inet_ntop (AF_INET6, gate, buf2, BUFSIZ),
+			   ifindex,
+			   type);
+	      else
+		zlog_info ("route %s/%d ifindex %d type %d doesn't exist in rib",
+			   inet_ntop (AF_INET6, &p->prefix, buf1, BUFSIZ),
+			   p->prefixlen,
+			   ifindex,
+			   type);
+	    }
+	  route_unlock_node (rn);
+	  return ZEBRA_ERR_RTNOEXIST;
+	}
+    }
+
+  if (same)
+    rib_delnode (rn, same);
+
+  /* Process changes. */
+  rib_process (rn, same);
+
+  if (same)
+    {
+      newrib_free (same);
+      route_unlock_node (rn);
+    }
+
+  route_unlock_node (rn);
+
+  return 0;
+}
+
+/* Install static route into rib. */
+void
+static_install_ipv6 (struct prefix *p, struct static_ipv6 *si)
+{
+  struct rib *rib;
+  struct route_table *table;
+  struct route_node *rn;
+
+  /* Lookup table.  */
+  table = vrf_table (AFI_IP6, SAFI_UNICAST, 0);
+  if (! table)
+    return;
+
+  /* Lookup existing route */
+  rn = route_node_get (table, p);
+  for (rib = rn->info; rib; rib = rib->next)
+    if (rib->type == ZEBRA_ROUTE_STATIC && rib->distance == si->distance)
+      break;
+
+  if (rib)
+    {
+      /* Same distance static route is there.  Update it with new
+         nexthop. */
+      rib_uninstall (rn, rib);
+      route_unlock_node (rn);
+
+      switch (si->type)
+	{
+	case STATIC_IPV6_GATEWAY:
+	  nexthop_ipv6_add (rib, &si->ipv6);
+	  break;
+	case STATIC_IPV6_IFNAME:
+	  nexthop_ifname_add (rib, si->ifname);
+	  break;
+	case STATIC_IPV6_GATEWAY_IFNAME:
+	  nexthop_ipv6_ifname_add (rib, &si->ipv6, si->ifname);
+	  break;
+	}
+      rib_process (rn, NULL);
+    }
+  else
+    {
+      /* This is new static route. */
+      rib = XMALLOC (MTYPE_RIB, sizeof (struct rib));
+      memset (rib, 0, sizeof (struct rib));
+
+      rib->type = ZEBRA_ROUTE_STATIC;
+      rib->distance = si->distance;
+      rib->metric = 0;
+      rib->nexthop_num = 0;
+
+      switch (si->type)
+	{
+	case STATIC_IPV6_GATEWAY:
+	  nexthop_ipv6_add (rib, &si->ipv6);
+	  break;
+	case STATIC_IPV6_IFNAME:
+	  nexthop_ifname_add (rib, si->ifname);
+	  break;
+	case STATIC_IPV6_GATEWAY_IFNAME:
+	  nexthop_ipv6_ifname_add (rib, &si->ipv6, si->ifname);
+	  break;
+	}
+
+      /* Link this rib to the tree. */
+      rib_addnode (rn, rib);
+
+      /* Process this prefix. */
+      rib_process (rn, NULL);
+    }
+}
+
+int
+static_ipv6_nexthop_same (struct nexthop *nexthop, struct static_ipv6 *si)
+{
+  if (nexthop->type == NEXTHOP_TYPE_IPV6
+      && si->type == STATIC_IPV6_GATEWAY
+      && IPV6_ADDR_SAME (&nexthop->gate.ipv6, &si->ipv6))
+    return 1;
+  if (nexthop->type == NEXTHOP_TYPE_IFNAME
+      && si->type == STATIC_IPV6_IFNAME
+      && strcmp (nexthop->ifname, si->ifname) == 0)
+    return 1;
+  if (nexthop->type == NEXTHOP_TYPE_IPV6_IFNAME
+      && si->type == STATIC_IPV6_GATEWAY_IFNAME
+      && IPV6_ADDR_SAME (&nexthop->gate.ipv6, &si->ipv6)
+      && strcmp (nexthop->ifname, si->ifname) == 0)
+    return 1;
+  return 0;;
+}
+
+void
+static_uninstall_ipv6 (struct prefix *p, struct static_ipv6 *si)
+{
+  struct route_table *table;
+  struct route_node *rn;
+  struct rib *rib;
+  struct nexthop *nexthop;
+
+  /* Lookup table.  */
+  table = vrf_table (AFI_IP6, SAFI_UNICAST, 0);
+  if (! table)
+    return;
+
+  /* Lookup existing route with type and distance. */
+  rn = route_node_lookup (table, (struct prefix *) p);
+  if (! rn)
+    return;
+
+  for (rib = rn->info; rib; rib = rib->next)
+    if (rib->type == ZEBRA_ROUTE_STATIC && rib->distance == si->distance)
+      break;
+  if (! rib)
+    {
+      route_unlock_node (rn);
+      return;
+    }
+
+  /* Lookup nexthop. */
+  for (nexthop = rib->nexthop; nexthop; nexthop = nexthop->next)
+    if (static_ipv6_nexthop_same (nexthop, si))
+      break;
+
+  /* Can't find nexthop. */
+  if (! nexthop)
+    {
+      route_unlock_node (rn);
+      return;
+    }
+  
+  /* Check nexthop. */
+  if (rib->nexthop_num == 1)
+    {
+      rib_delnode (rn, rib);
+      rib_process (rn, rib);
+      newrib_free (rib);
+      route_unlock_node (rn);
+    }
+  else
+    {
+      rib_uninstall (rn, rib);
+      nexthop_delete (rib, nexthop);
+      nexthop_free (nexthop);
+      rib_process (rn, rib);
+    }
+
+  /* Unlock node. */
+  route_unlock_node (rn);
+}
+
+/* Add static route into static route configuration. */
+int
+static_add_ipv6 (struct prefix *p, u_char type, struct in6_addr *gate,
+		 char *ifname, u_char distance, u_int32_t vrf_id)
+{
+  struct route_node *rn;
+  struct static_ipv6 *si;
+  struct static_ipv6 *pp;
+  struct static_ipv6 *cp;
+  struct route_table *stable;
+
+  /* Lookup table.  */
+  stable = vrf_static_table (AFI_IP6, SAFI_UNICAST, vrf_id);
+  if (! stable)
+    return -1;
+
+  /* Lookup static route prefix. */
+  rn = route_node_get (stable, p);
+
+  /* Do nothing if there is a same static route.  */
+  for (si = rn->info; si; si = si->next)
+    {
+      if (distance == si->distance 
+	  && type == si->type
+	  && (! gate || IPV6_ADDR_SAME (gate, &si->ipv6))
+	  && (! ifname || strcmp (ifname, si->ifname) == 0))
+	{
+	  route_unlock_node (rn);
+	  return 0;
+	}
+    }
+
+  /* Make new static route structure. */
+  si = XMALLOC (MTYPE_STATIC_IPV6, sizeof (struct static_ipv6));
+  memset (si, 0, sizeof (struct static_ipv6));
+
+  si->type = type;
+  si->distance = distance;
+
+  switch (type)
+    {
+    case STATIC_IPV6_GATEWAY:
+      si->ipv6 = *gate;
+      break;
+    case STATIC_IPV6_IFNAME:
+      si->ifname = XSTRDUP (0, ifname);
+      break;
+    case STATIC_IPV6_GATEWAY_IFNAME:
+      si->ipv6 = *gate;
+      si->ifname = XSTRDUP (0, ifname);
+      break;
+    }
+
+  /* Add new static route information to the tree with sort by
+     distance value and gateway address. */
+  for (pp = NULL, cp = rn->info; cp; pp = cp, cp = cp->next)
+    {
+      if (si->distance < cp->distance)
+	break;
+      if (si->distance > cp->distance)
+	continue;
+    }
+
+  /* Make linked list. */
+  if (pp)
+    pp->next = si;
+  else
+    rn->info = si;
+  if (cp)
+    cp->prev = si;
+  si->prev = pp;
+  si->next = cp;
+
+  /* Install into rib. */
+  static_install_ipv6 (p, si);
+
+  return 1;
+}
+
+/* Delete static route from static route configuration. */
+int
+static_delete_ipv6 (struct prefix *p, u_char type, struct in6_addr *gate,
+		    char *ifname, u_char distance, u_int32_t vrf_id)
+{
+  struct route_node *rn;
+  struct static_ipv6 *si;
+  struct route_table *stable;
+
+  /* Lookup table.  */
+  stable = vrf_static_table (AFI_IP6, SAFI_UNICAST, vrf_id);
+  if (! stable)
+    return -1;
+
+  /* Lookup static route prefix. */
+  rn = route_node_lookup (stable, p);
+  if (! rn)
+    return 0;
+
+  /* Find same static route is the tree */
+  for (si = rn->info; si; si = si->next)
+    if (distance == si->distance 
+	&& type == si->type
+	&& (! gate || IPV6_ADDR_SAME (gate, &si->ipv6))
+	&& (! ifname || strcmp (ifname, si->ifname) == 0))
+      break;
+
+  /* Can't find static route. */
+  if (! si)
+    {
+      route_unlock_node (rn);
+      return 0;
+    }
+
+  /* Install into rib. */
+  static_uninstall_ipv6 (p, si);
+
+  /* Unlink static route from linked list. */
+  if (si->prev)
+    si->prev->next = si->next;
+  else
+    rn->info = si->next;
+  if (si->next)
+    si->next->prev = si->prev;
+  
+  /* Free static route configuration. */
+  XFREE (MTYPE_STATIC_IPV6, si);
+
+  return 1;
+}
+#endif /* HAVE_IPV6 */
+
+/* RIB update function. */
+void
+rib_update ()
+{
+  struct route_node *rn;
+  struct route_table *table;
+
+  table = vrf_table (AFI_IP, SAFI_UNICAST, 0);
+  if (table)
+    for (rn = route_top (table); rn; rn = route_next (rn))
+      rib_process (rn, NULL);
+
+  table = vrf_table (AFI_IP6, SAFI_UNICAST, 0);
+  if (table)
+    for (rn = route_top (table); rn; rn = route_next (rn))
+      rib_process (rn, NULL);
+}
+
+/* Interface goes up. */
+void
+rib_if_up (struct interface *ifp)
+{
+  rib_update ();
+}
+
+/* Interface goes down. */
+void
+rib_if_down (struct interface *ifp)
+{
+  rib_update ();
+}
+
+/* Remove all routes which comes from non main table.  */
+void
+rib_weed_table (struct route_table *table)
+{
+  struct route_node *rn;
+  struct rib *rib;
+  struct rib *next;
+
+  if (table)
+    for (rn = route_top (table); rn; rn = route_next (rn))
+      for (rib = rn->info; rib; rib = next)
+	{
+	  next = rib->next;
+
+	  if (rib->table != rtm_table_default &&
+	      rib->table != RT_TABLE_MAIN)
+	    {
+	      rib_delnode (rn, rib);
+	      newrib_free (rib);
+	      route_unlock_node (rn);
+	    }
+	}
+}
+
+/* Delete all routes from non main table. */
+void
+rib_weed_tables ()
+{
+  rib_weed_table (vrf_table (AFI_IP, SAFI_UNICAST, 0));
+  rib_weed_table (vrf_table (AFI_IP6, SAFI_UNICAST, 0));
+}
+
+/* Delete self installed routes after zebra is relaunched.  */
+void
+rib_sweep_table (struct route_table *table)
+{
+  struct route_node *rn;
+  struct rib *rib;
+  struct rib *next;
+  int ret = 0;
+
+  if (table)
+    for (rn = route_top (table); rn; rn = route_next (rn))
+      for (rib = rn->info; rib; rib = next)
+	{
+	  next = rib->next;
+
+	  if (rib->type == ZEBRA_ROUTE_KERNEL && 
+	      CHECK_FLAG (rib->flags, ZEBRA_FLAG_SELFROUTE))
+	    {
+	      ret = rib_uninstall_kernel (rn, rib);
+	      if (! ret)
+		{
+		  rib_delnode (rn, rib);
+		  newrib_free (rib);
+		  route_unlock_node (rn);
+		}
+	    }
+	}
+}
+
+/* Sweep all RIB tables.  */
+void
+rib_sweep_route ()
+{
+  rib_sweep_table (vrf_table (AFI_IP, SAFI_UNICAST, 0));
+  rib_sweep_table (vrf_table (AFI_IP6, SAFI_UNICAST, 0));
+}
+
+/* Close RIB and clean up kernel routes. */
+void
+rib_close_table (struct route_table *table)
+{
+  struct route_node *rn;
+  struct rib *rib;
+
+  if (table)
+    for (rn = route_top (table); rn; rn = route_next (rn))
+      for (rib = rn->info; rib; rib = rib->next)
+	if (! RIB_SYSTEM_ROUTE (rib)
+	    && CHECK_FLAG (rib->flags, ZEBRA_FLAG_SELECTED))
+	  rib_uninstall_kernel (rn, rib);
+}
+
+/* Close all RIB tables.  */
+void
+rib_close ()
+{
+  rib_close_table (vrf_table (AFI_IP, SAFI_UNICAST, 0));
+  rib_close_table (vrf_table (AFI_IP6, SAFI_UNICAST, 0));
+}
+
+/* Routing information base initialize. */
+void
+rib_init ()
+{
+  /* VRF initialization.  */
+  vrf_init ();
+}