From fca0c242e4edfdb05231ef18d60c14273067ff0a Mon Sep 17 00:00:00 2001
From: Neale Ranns <nranns@cisco.com>
Date: Fri, 13 Jan 2017 07:57:46 -0800
Subject: DHCPv[46] proxy tests

Change-Id: I6aaf9c602cd515ed9d4416d286f9191d048c1a87
Signed-off-by: Neale Ranns <nranns@cisco.com>
---
 test/test_dhcp.py | 739 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 739 insertions(+)
 create mode 100644 test/test_dhcp.py

(limited to 'test/test_dhcp.py')

diff --git a/test/test_dhcp.py b/test/test_dhcp.py
new file mode 100644
index 00000000..bdff679c
--- /dev/null
+++ b/test/test_dhcp.py
@@ -0,0 +1,739 @@
+#!/usr/bin/env python
+
+import unittest
+import socket
+
+from framework import VppTestCase, VppTestRunner
+from vpp_ip_route import IpRoute, RoutePath
+from vpp_lo_interface import VppLoInterface
+
+from scapy.layers.l2 import Ether, getmacbyip
+from scapy.layers.inet import IP, UDP, ICMP
+from scapy.layers.inet6 import IPv6, in6_getnsmac, in6_mactoifaceid
+from scapy.layers.dhcp import DHCP, BOOTP, DHCPTypes
+from scapy.layers.dhcp6 import DHCP6, DHCP6_Solicit, DHCP6_RelayForward, \
+    DHCP6_RelayReply, DHCP6_Advertise, DHCP6OptRelayMsg, DHCP6OptIfaceId, \
+    DHCP6OptStatusCode, DHCP6OptVSS, DHCP6OptClientLinkLayerAddr
+from socket import AF_INET, AF_INET6
+from scapy.utils import inet_pton, inet_ntop
+from scapy.utils6 import in6_ptop
+
+DHCP4_CLIENT_PORT = 68
+DHCP4_SERVER_PORT = 67
+DHCP6_CLIENT_PORT = 547
+DHCP6_SERVER_PORT = 546
+
+
+def mk_ll_addr(mac):
+
+    euid = in6_mactoifaceid(mac)
+    addr = "fe80::" + euid
+    return addr
+
+
+class TestDHCP(VppTestCase):
+    """ DHCP Test Case """
+
+    def setUp(self):
+        super(TestDHCP, self).setUp()
+
+        # create 3 pg interfaces
+        self.create_pg_interfaces(range(4))
+
+        # pg0 and 1 are IP configured in VRF 0 and 1.
+        # pg2 and 3 are non IP-configured in VRF 0 and 1
+        table_id = 0
+        for i in self.pg_interfaces[:2]:
+            i.admin_up()
+            i.set_table_ip4(table_id)
+            i.set_table_ip6(table_id)
+            i.config_ip4()
+            i.resolve_arp()
+            i.config_ip6()
+            i.resolve_ndp()
+            table_id += 1
+
+        table_id = 0
+        for i in self.pg_interfaces[2:]:
+            i.admin_up()
+            i.set_table_ip4(table_id)
+            i.set_table_ip6(table_id)
+            table_id += 1
+
+    def send_and_assert_no_replies(self, intf, pkts, remark):
+        intf.add_stream(pkts)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+        for i in self.pg_interfaces:
+            i.assert_nothing_captured(remark=remark)
+
+    def validate_option_82(self, pkt, intf, ip_addr):
+        dhcp = pkt[DHCP]
+        found = 0
+        data = []
+
+        for i in dhcp.options:
+            if type(i) is tuple:
+                if i[0] == "relay_agent_Information":
+                    #
+                    # There are two sb-options present - each of length 6.
+                    #
+                    data = i[1]
+                    self.assertEqual(len(data), 12)
+
+                    #
+                    # First sub-option is ID 1, len 4, then encoded
+                    #  sw_if_index. This test uses low valued indicies
+                    # so [2:4] are 0.
+                    # The ID space is VPP internal - so no matching value
+                    # scapy
+                    #
+                    self.assertEqual(ord(data[0]), 1)
+                    self.assertEqual(ord(data[1]), 4)
+                    self.assertEqual(ord(data[2]), 0)
+                    self.assertEqual(ord(data[3]), 0)
+                    self.assertEqual(ord(data[4]), 0)
+                    self.assertEqual(ord(data[5]), intf._sw_if_index)
+
+                    #
+                    # next sub-option is the IP address of the client side
+                    # interface.
+                    # sub-option ID=5, length (of a v4 address)=4
+                    #
+                    claddr = socket.inet_pton(AF_INET, ip_addr)
+
+                    self.assertEqual(ord(data[6]), 5)
+                    self.assertEqual(ord(data[7]), 4)
+                    self.assertEqual(data[8], claddr[0])
+                    self.assertEqual(data[9], claddr[1])
+                    self.assertEqual(data[10], claddr[2])
+                    self.assertEqual(data[11], claddr[3])
+
+                    found = 1
+        self.assertTrue(found)
+
+        return data
+
+    def verify_dhcp_offer(self, pkt, intf, check_option_82=True):
+        ether = pkt[Ether]
+        self.assertEqual(ether.dst, "ff:ff:ff:ff:ff:ff")
+        self.assertEqual(ether.src, intf.local_mac)
+
+        ip = pkt[IP]
+        self.assertEqual(ip.dst, "255.255.255.255")
+        self.assertEqual(ip.src, intf.local_ip4)
+
+        udp = pkt[UDP]
+        self.assertEqual(udp.dport, DHCP4_CLIENT_PORT)
+        self.assertEqual(udp.sport, DHCP4_SERVER_PORT)
+
+        dhcp = pkt[DHCP]
+        is_offer = False
+        for o in dhcp.options:
+            if type(o) is tuple:
+                if o[0] == "message-type" \
+                   and DHCPTypes[o[1]] == "offer":
+                    is_offer = True
+        self.assertTrue(is_offer)
+
+        if check_option_82:
+            data = self.validate_option_82(pkt, intf, intf.local_ip4)
+
+    def verify_dhcp_discover(self, pkt, intf, src_intf=None,
+                             option_82_present=True):
+        ether = pkt[Ether]
+        self.assertEqual(ether.dst, intf.remote_mac)
+        self.assertEqual(ether.src, intf.local_mac)
+
+        ip = pkt[IP]
+        self.assertEqual(ip.dst, intf.remote_ip4)
+        self.assertEqual(ip.src, intf.local_ip4)
+
+        udp = pkt[UDP]
+        self.assertEqual(udp.dport, DHCP4_SERVER_PORT)
+        self.assertEqual(udp.sport, DHCP4_CLIENT_PORT)
+
+        dhcp = pkt[DHCP]
+
+        is_discover = False
+        for o in dhcp.options:
+            if type(o) is tuple:
+                if o[0] == "message-type" \
+                   and DHCPTypes[o[1]] == "discover":
+                    is_discover = True
+        self.assertTrue(is_discover)
+
+        if option_82_present:
+            data = self.validate_option_82(pkt, src_intf, src_intf.local_ip4)
+            return data
+        else:
+            for i in dhcp.options:
+                if type(i) is tuple:
+                    self.assertNotEqual(i[0], "relay_agent_Information")
+
+    def verify_dhcp6_solicit(self, pkt, intf,
+                             peer_ip, peer_mac,
+                             fib_id=0,
+                             oui=0):
+        ether = pkt[Ether]
+        self.assertEqual(ether.dst, intf.remote_mac)
+        self.assertEqual(ether.src, intf.local_mac)
+
+        ip = pkt[IPv6]
+        self.assertEqual(in6_ptop(ip.dst), in6_ptop(intf.remote_ip6))
+        self.assertEqual(in6_ptop(ip.src), in6_ptop(intf.local_ip6))
+
+        udp = pkt[UDP]
+        self.assertEqual(udp.dport, DHCP6_CLIENT_PORT)
+        self.assertEqual(udp.sport, DHCP6_SERVER_PORT)
+
+        relay = pkt[DHCP6_RelayForward]
+        self.assertEqual(in6_ptop(relay.peeraddr), in6_ptop(peer_ip))
+        oid = pkt[DHCP6OptIfaceId]
+        cll = pkt[DHCP6OptClientLinkLayerAddr]
+        self.assertEqual(cll.optlen, 8)
+        self.assertEqual(cll.lltype, 1)
+        self.assertEqual(cll.clladdr, peer_mac)
+
+        vss = pkt[DHCP6OptVSS]
+        self.assertEqual(vss.optlen, 8)
+        self.assertEqual(vss.type, 1)
+        # the OUI and FIB-id are really 3 and 4 bytes resp.
+        # but the tested range is small
+        self.assertEqual(ord(vss.data[0]), 0)
+        self.assertEqual(ord(vss.data[1]), 0)
+        self.assertEqual(ord(vss.data[2]), oui)
+        self.assertEqual(ord(vss.data[3]), 0)
+        self.assertEqual(ord(vss.data[4]), 0)
+        self.assertEqual(ord(vss.data[5]), 0)
+        self.assertEqual(ord(vss.data[6]), fib_id)
+
+        # the relay message should be an encoded Solicit
+        msg = pkt[DHCP6OptRelayMsg]
+        sol = DHCP6_Solicit()
+        self.assertEqual(msg.optlen, len(str(sol)))
+        self.assertEqual(str(sol), (str(msg[1]))[:msg.optlen])
+
+    def verify_dhcp6_advert(self, pkt, intf, peer):
+        ether = pkt[Ether]
+        self.assertEqual(ether.dst, "ff:ff:ff:ff:ff:ff")
+        self.assertEqual(ether.src, intf.local_mac)
+
+        ip = pkt[IPv6]
+        self.assertEqual(in6_ptop(ip.dst), in6_ptop(peer))
+        self.assertEqual(in6_ptop(ip.src), in6_ptop(intf.local_ip6))
+
+        udp = pkt[UDP]
+        self.assertEqual(udp.dport, DHCP6_SERVER_PORT)
+        self.assertEqual(udp.sport, DHCP6_CLIENT_PORT)
+
+        # not sure why this is not decoding
+        # adv = pkt[DHCP6_Advertise]
+
+    def test_dhcp_proxy(self):
+        """ DHCPv4 Proxy """
+
+        #
+        # Verify no response to DHCP request without DHCP config
+        #
+        p_disc_vrf0 = (Ether(dst="ff:ff:ff:ff:ff:ff",
+                             src=self.pg2.remote_mac) /
+                       IP(src="0.0.0.0", dst="255.255.255.255") /
+                       UDP(sport=DHCP4_CLIENT_PORT,
+                           dport=DHCP4_SERVER_PORT) /
+                       BOOTP(op=1) /
+                       DHCP(options=[('message-type', 'discover'), ('end')]))
+        pkts_disc_vrf0 = [p_disc_vrf0]
+        p_disc_vrf1 = (Ether(dst="ff:ff:ff:ff:ff:ff",
+                             src=self.pg3.remote_mac) /
+                       IP(src="0.0.0.0", dst="255.255.255.255") /
+                       UDP(sport=DHCP4_CLIENT_PORT,
+                           dport=DHCP4_SERVER_PORT) /
+                       BOOTP(op=1) /
+                       DHCP(options=[('message-type', 'discover'), ('end')]))
+        pkts_disc_vrf1 = [p_disc_vrf0]
+
+        self.send_and_assert_no_replies(self.pg2, pkts_disc_vrf0,
+                                        "DHCP with no configuration")
+        self.send_and_assert_no_replies(self.pg3, pkts_disc_vrf1,
+                                        "DHCP with no configuration")
+
+        #
+        # Enable DHCP proxy in VRF 0
+        #
+        server_addr = self.pg0.remote_ip4n
+        src_addr = self.pg0.local_ip4n
+
+        self.vapi.dhcp_proxy_config(server_addr,
+                                    src_addr,
+                                    rx_table_id=0)
+
+        #
+        # Now a DHCP request on pg2, which is in the same VRF
+        # as the DHCP config, will result in a relayed DHCP
+        # message to the [fake] server
+        #
+        self.pg2.add_stream(pkts_disc_vrf0)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg0.get_capture(1)
+        rx = rx[0]
+
+        #
+        # Rx'd packet should be to the server address and from the configured
+        # source address
+        # UDP source ports are unchanged
+        # we've no option 82 config so that should be absent
+        #
+        self.verify_dhcp_discover(rx, self.pg0, option_82_present=False)
+
+        #
+        # Inject a response from the server
+        #  VPP will only relay the offer if option 82 is present.
+        #  so this one is dropped
+        #
+        p = (Ether(dst=self.pg0.local_mac, src=self.pg0.remote_mac) /
+             IP(src=self.pg0.remote_ip4, dst=self.pg0.local_ip4) /
+             UDP(sport=DHCP4_SERVER_PORT, dport=DHCP4_SERVER_PORT) /
+             BOOTP(op=1) /
+             DHCP(options=[('message-type', 'offer'), ('end')]))
+        pkts = [p]
+
+        self.send_and_assert_no_replies(self.pg0, pkts,
+                                        "DHCP offer no option 82")
+
+        #
+        # Configure sending option 82 in relayed messages
+        #
+        self.vapi.dhcp_proxy_config(server_addr,
+                                    src_addr,
+                                    rx_table_id=0,
+                                    insert_circuit_id=1)
+
+        #
+        # Send a request:
+        #  again dropped, but ths time because there is no IP addrees on the
+        #  clinet interfce to fill in the option.
+        #
+        self.send_and_assert_no_replies(self.pg2, pkts_disc_vrf0,
+                                        "DHCP no relay address")
+
+        #
+        # configure an IP address on the client facing interface
+        #
+        self.pg2.config_ip4()
+
+        #
+        # Try again with a discover packet
+        # Rx'd packet should be to the server address and from the configured
+        # source address
+        # UDP source ports are unchanged
+        # we've no option 82 config so that should be absent
+        #
+        self.pg2.add_stream(pkts_disc_vrf0)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg0.get_capture(1)
+        rx = rx[0]
+
+        option_82 = self.verify_dhcp_discover(rx, self.pg0, src_intf=self.pg2)
+
+        #
+        # Create an DHCP offer reply from the server with a correctly formatted
+        # option 82. i.e. send back what we just captured
+        # The offer, sent mcast to the client, still has option 82.
+        #
+        p = (Ether(dst=self.pg0.local_mac, src=self.pg0.remote_mac) /
+             IP(src=self.pg0.remote_ip4, dst=self.pg0.local_ip4) /
+             UDP(sport=DHCP4_SERVER_PORT, dport=DHCP4_SERVER_PORT) /
+             BOOTP(op=1) /
+             DHCP(options=[('message-type', 'offer'),
+                           ('relay_agent_Information', option_82),
+                           ('end')]))
+        pkts = [p]
+
+        self.pg0.add_stream(pkts)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg2.get_capture(1)
+        rx = rx[0]
+
+        self.verify_dhcp_offer(rx, self.pg2)
+
+        #
+        # Bogus Option 82:
+        #
+        # 1. not our IP address = not checked by VPP? so offer is replayed
+        #    to client
+        bad_ip = option_82[0:8] + chr(33) + option_82[9:]
+
+        p = (Ether(dst=self.pg0.local_mac, src=self.pg0.remote_mac) /
+             IP(src=self.pg0.remote_ip4, dst=self.pg0.local_ip4) /
+             UDP(sport=DHCP4_SERVER_PORT, dport=DHCP4_SERVER_PORT) /
+             BOOTP(op=1) /
+             DHCP(options=[('message-type', 'offer'),
+                           ('relay_agent_Information', bad_ip),
+                           ('end')]))
+        pkts = [p]
+
+        self.pg0.add_stream(pkts)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+        rx = self.pg2.get_capture(1)
+        rx = rx[0]
+
+        self.verify_dhcp_offer(rx, self.pg2, check_option_82=False)
+        self.pg0.assert_nothing_captured(remark="")
+
+        # 2. Not a sw_if_index VPP knows
+        bad_if_index = option_82[0:2] + chr(33) + option_82[3:]
+
+        p = (Ether(dst=self.pg0.local_mac, src=self.pg0.remote_mac) /
+             IP(src=self.pg0.remote_ip4, dst=self.pg0.local_ip4) /
+             UDP(sport=DHCP4_SERVER_PORT, dport=DHCP4_SERVER_PORT) /
+             BOOTP(op=1) /
+             DHCP(options=[('message-type', 'offer'),
+                           ('relay_agent_Information', bad_if_index),
+                           ('end')]))
+        pkts = [p]
+        self.send_and_assert_no_replies(self.pg0, pkts,
+                                        "DHCP offer option 82 bad if index")
+
+        #
+        # Send a DHCP request in VRF 1. should be dropped.
+        #
+        self.send_and_assert_no_replies(self.pg3, pkts_disc_vrf1,
+                                        "DHCP with no configuration VRF 1")
+
+        #
+        # Delete the DHCP config in VRF 0
+        # Should now drop requests.
+        #
+        self.vapi.dhcp_proxy_config(server_addr,
+                                    src_addr,
+                                    rx_table_id=0,
+                                    is_add=0,
+                                    insert_circuit_id=1)
+
+        self.send_and_assert_no_replies(self.pg2, pkts_disc_vrf0,
+                                        "DHCP config removed VRF 0")
+        self.send_and_assert_no_replies(self.pg3, pkts_disc_vrf1,
+                                        "DHCP config removed VRF 1")
+
+        #
+        # Add DHCP config for VRF 1
+        #
+        server_addr = self.pg1.remote_ip4n
+        src_addr = self.pg1.local_ip4n
+        self.vapi.dhcp_proxy_config(server_addr,
+                                    src_addr,
+                                    rx_table_id=1,
+                                    server_table_id=1,
+                                    insert_circuit_id=1)
+
+        #
+        # Confim DHCP requests ok in VRF 1.
+        #  - dropped on IP config on client interface
+        #
+        self.send_and_assert_no_replies(self.pg3, pkts_disc_vrf1,
+                                        "DHCP config removed VRF 1")
+
+        #
+        # configure an IP address on the client facing interface
+        #
+        self.pg3.config_ip4()
+
+        self.pg3.add_stream(pkts_disc_vrf1)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg1.get_capture(1)
+        rx = rx[0]
+        self.verify_dhcp_discover(rx, self.pg1, src_intf=self.pg3)
+
+        #
+        # remove DHCP config to cleanup
+        #
+        self.vapi.dhcp_proxy_config(server_addr,
+                                    src_addr,
+                                    rx_table_id=1,
+                                    server_table_id=1,
+                                    insert_circuit_id=1,
+                                    is_add=0)
+
+        self.send_and_assert_no_replies(self.pg2, pkts_disc_vrf0,
+                                        "DHCP cleanup VRF 0")
+        self.send_and_assert_no_replies(self.pg3, pkts_disc_vrf1,
+                                        "DHCP cleanup VRF 1")
+
+    def test_dhcp6_proxy(self):
+        """ DHCPv6 Proxy"""
+        #
+        # Verify no response to DHCP request without DHCP config
+        #
+        dhcp_solicit_dst = "ff02::1:2"
+        dhcp_solicit_src_vrf0 = mk_ll_addr(self.pg2.remote_mac)
+        dhcp_solicit_src_vrf1 = mk_ll_addr(self.pg3.remote_mac)
+        server_addr_vrf0 = self.pg0.remote_ip6n
+        src_addr_vrf0 = self.pg0.local_ip6n
+        server_addr_vrf1 = self.pg1.remote_ip6n
+        src_addr_vrf1 = self.pg1.local_ip6n
+
+        #
+        # Add the Route to receive the DHCP packets
+        #
+        route_dhcp_vrf0 = IpRoute(self, dhcp_solicit_dst, 128,
+                                  [], is_local=1, is_ip6=1)
+        route_dhcp_vrf0.add_vpp_config()
+        route_dhcp_vrf1 = IpRoute(self, dhcp_solicit_dst, 128,
+                                  [], is_local=1, is_ip6=1,
+                                  table_id=1)
+        route_dhcp_vrf1.add_vpp_config()
+
+        dmac = in6_getnsmac(inet_pton(socket.AF_INET6, dhcp_solicit_dst))
+        p_solicit_vrf0 = (Ether(dst=dmac, src=self.pg2.remote_mac) /
+                          IPv6(src=dhcp_solicit_src_vrf0,
+                               dst=dhcp_solicit_dst) /
+                          UDP(sport=DHCP6_SERVER_PORT,
+                              dport=DHCP6_CLIENT_PORT) /
+                          DHCP6_Solicit())
+        pkts_solicit_vrf0 = [p_solicit_vrf0]
+        p_solicit_vrf1 = (Ether(dst=dmac, src=self.pg3.remote_mac) /
+                          IPv6(src=dhcp_solicit_src_vrf1,
+                               dst=dhcp_solicit_dst) /
+                          UDP(sport=DHCP6_SERVER_PORT,
+                              dport=DHCP6_CLIENT_PORT) /
+                          DHCP6_Solicit())
+        pkts_solicit_vrf1 = [p_solicit_vrf1]
+
+        self.send_and_assert_no_replies(self.pg2, pkts_solicit_vrf0,
+                                        "DHCP with no configuration")
+        self.send_and_assert_no_replies(self.pg3, pkts_solicit_vrf1,
+                                        "DHCP with no configuration")
+
+        #
+        # DHCPv6 config in VRF 0.
+        # Packets still dropped because the client facing interface has no
+        # IPv6 config
+        #
+        self.vapi.dhcp_proxy_config(server_addr_vrf0,
+                                    src_addr_vrf0,
+                                    rx_table_id=0,
+                                    server_table_id=0,
+                                    insert_circuit_id=1,
+                                    is_ipv6=1)
+
+        self.send_and_assert_no_replies(self.pg2, pkts_solicit_vrf0,
+                                        "DHCP with no configuration")
+        self.send_and_assert_no_replies(self.pg3, pkts_solicit_vrf1,
+                                        "DHCP with no configuration")
+
+        #
+        # configure an IP address on the client facing interface
+        #
+        self.pg2.config_ip6()
+
+        #
+        # Now the DHCP requests are relayed to the server
+        #
+        self.pg2.add_stream(pkts_solicit_vrf0)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg0.get_capture(1)
+        rx = rx[0]
+        self.verify_dhcp6_solicit(rx, self.pg0,
+                                  dhcp_solicit_src_vrf0,
+                                  self.pg2.remote_mac)
+
+        #
+        # Exception cases for rejected relay responses
+        #
+
+        # 1 - not a relay reply
+        p_adv_vrf0 = (Ether(dst=self.pg0.local_mac, src=self.pg0.remote_mac) /
+                      IPv6(dst=self.pg0.local_ip6, src=self.pg0.remote_ip6) /
+                      UDP(sport=DHCP6_SERVER_PORT, dport=DHCP6_SERVER_PORT) /
+                      DHCP6_Advertise())
+        pkts_adv_vrf0 = [p_adv_vrf0]
+        self.send_and_assert_no_replies(self.pg2, pkts_adv_vrf0,
+                                        "DHCP6 not a relay reply")
+
+        # 2 - no relay message option
+        p_adv_vrf0 = (Ether(dst=self.pg0.local_mac, src=self.pg0.remote_mac) /
+                      IPv6(dst=self.pg0.local_ip6, src=self.pg0.remote_ip6) /
+                      UDP(sport=DHCP6_SERVER_PORT, dport=DHCP6_SERVER_PORT) /
+                      DHCP6_RelayReply() /
+                      DHCP6_Advertise())
+        pkts_adv_vrf0 = [p_adv_vrf0]
+        self.send_and_assert_no_replies(self.pg2, pkts_adv_vrf0,
+                                        "DHCP not a relay message")
+
+        # 3 - no circuit ID
+        p_adv_vrf0 = (Ether(dst=self.pg0.local_mac, src=self.pg0.remote_mac) /
+                      IPv6(dst=self.pg0.local_ip6, src=self.pg0.remote_ip6) /
+                      UDP(sport=DHCP6_SERVER_PORT, dport=DHCP6_SERVER_PORT) /
+                      DHCP6_RelayReply() /
+                      DHCP6OptRelayMsg(optlen=0) /
+                      DHCP6_Advertise())
+        pkts_adv_vrf0 = [p_adv_vrf0]
+        self.send_and_assert_no_replies(self.pg2, pkts_adv_vrf0,
+                                        "DHCP6 no circuit ID")
+        # 4 - wrong circuit ID
+        p_adv_vrf0 = (Ether(dst=self.pg0.local_mac, src=self.pg0.remote_mac) /
+                      IPv6(dst=self.pg0.local_ip6, src=self.pg0.remote_ip6) /
+                      UDP(sport=DHCP6_SERVER_PORT, dport=DHCP6_SERVER_PORT) /
+                      DHCP6_RelayReply() /
+                      DHCP6OptIfaceId(optlen=4, ifaceid='\x00\x00\x00\x05') /
+                      DHCP6OptRelayMsg(optlen=0) /
+                      DHCP6_Advertise())
+        pkts_adv_vrf0 = [p_adv_vrf0]
+        self.send_and_assert_no_replies(self.pg2, pkts_adv_vrf0,
+                                        "DHCP6 wrong circuit ID")
+
+        #
+        # Send the relay response (the advertisement)
+        #   - no peer address
+        p_adv_vrf0 = (Ether(dst=self.pg0.local_mac, src=self.pg0.remote_mac) /
+                      IPv6(dst=self.pg0.local_ip6, src=self.pg0.remote_ip6) /
+                      UDP(sport=DHCP6_SERVER_PORT, dport=DHCP6_SERVER_PORT) /
+                      DHCP6_RelayReply() /
+                      DHCP6OptIfaceId(optlen=4, ifaceid='\x00\x00\x00\x03') /
+                      DHCP6OptRelayMsg(optlen=0) /
+                      DHCP6_Advertise(trid=1) /
+                      DHCP6OptStatusCode(statuscode=0))
+        pkts_adv_vrf0 = [p_adv_vrf0]
+
+        self.pg0.add_stream(pkts_adv_vrf0)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg2.get_capture(1)
+        rx = rx[0]
+        self.verify_dhcp6_advert(rx, self.pg2, "::")
+
+        #
+        # Send the relay response (the advertisement)
+        #   - with peer address
+        p_adv_vrf0 = (Ether(dst=self.pg0.local_mac, src=self.pg0.remote_mac) /
+                      IPv6(dst=self.pg0.local_ip6, src=self.pg0.remote_ip6) /
+                      UDP(sport=DHCP6_SERVER_PORT, dport=DHCP6_SERVER_PORT) /
+                      DHCP6_RelayReply(peeraddr=dhcp_solicit_src_vrf0) /
+                      DHCP6OptIfaceId(optlen=4, ifaceid='\x00\x00\x00\x03') /
+                      DHCP6OptRelayMsg(optlen=0) /
+                      DHCP6_Advertise(trid=1) /
+                      DHCP6OptStatusCode(statuscode=0))
+        pkts_adv_vrf0 = [p_adv_vrf0]
+
+        self.pg0.add_stream(pkts_adv_vrf0)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg2.get_capture(1)
+        rx = rx[0]
+        self.verify_dhcp6_advert(rx, self.pg2, dhcp_solicit_src_vrf0)
+
+        #
+        # Add all the config for VRF 1
+        #
+        self.vapi.dhcp_proxy_config(server_addr_vrf1,
+                                    src_addr_vrf1,
+                                    rx_table_id=1,
+                                    server_table_id=1,
+                                    insert_circuit_id=1,
+                                    is_ipv6=1)
+        self.pg3.config_ip6()
+
+        #
+        # VRF 1 solicit
+        #
+        self.pg3.add_stream(pkts_solicit_vrf1)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg1.get_capture(1)
+        rx = rx[0]
+        self.verify_dhcp6_solicit(rx, self.pg1,
+                                  dhcp_solicit_src_vrf1,
+                                  self.pg3.remote_mac)
+
+        #
+        # VRF 1 Advert
+        #
+        p_adv_vrf1 = (Ether(dst=self.pg1.local_mac, src=self.pg1.remote_mac) /
+                      IPv6(dst=self.pg1.local_ip6, src=self.pg1.remote_ip6) /
+                      UDP(sport=DHCP6_SERVER_PORT, dport=DHCP6_SERVER_PORT) /
+                      DHCP6_RelayReply(peeraddr=dhcp_solicit_src_vrf1) /
+                      DHCP6OptIfaceId(optlen=4, ifaceid='\x00\x00\x00\x04') /
+                      DHCP6OptRelayMsg(optlen=0) /
+                      DHCP6_Advertise(trid=1) /
+                      DHCP6OptStatusCode(statuscode=0))
+        pkts_adv_vrf1 = [p_adv_vrf1]
+
+        self.pg1.add_stream(pkts_adv_vrf1)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg3.get_capture(1)
+        rx = rx[0]
+        self.verify_dhcp6_advert(rx, self.pg3, dhcp_solicit_src_vrf1)
+
+        #
+        # Add VSS config
+        #  table=1, fib=id=1, oui=4
+        self.vapi.dhcp_proxy_set_vss(1, 1, 4, is_ip6=1)
+
+        self.pg3.add_stream(pkts_solicit_vrf1)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg1.get_capture(1)
+        rx = rx[0]
+        self.verify_dhcp6_solicit(rx, self.pg1,
+                                  dhcp_solicit_src_vrf1,
+                                  self.pg3.remote_mac,
+                                  fib_id=1,
+                                  oui=4)
+
+        #
+        # Remove the VSS config
+        #  relayed DHCP has default vlaues in the option.
+        #
+        self.vapi.dhcp_proxy_set_vss(1, 1, 4, is_ip6=1, is_add=0)
+
+        self.pg3.add_stream(pkts_solicit_vrf1)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg1.get_capture(1)
+        rx = rx[0]
+        self.verify_dhcp6_solicit(rx, self.pg1,
+                                  dhcp_solicit_src_vrf1,
+                                  self.pg3.remote_mac)
+
+        #
+        # Cleanup
+        #
+        self.vapi.dhcp_proxy_config(server_addr_vrf1,
+                                    src_addr_vrf1,
+                                    rx_table_id=1,
+                                    server_table_id=1,
+                                    insert_circuit_id=1,
+                                    is_ipv6=1,
+                                    is_add=0)
+        self.vapi.dhcp_proxy_config(server_addr_vrf1,
+                                    src_addr_vrf1,
+                                    rx_table_id=0,
+                                    server_table_id=0,
+                                    insert_circuit_id=1,
+                                    is_ipv6=1,
+                                    is_add=0)
+
+        route_dhcp_vrf0.remove_vpp_config()
+        route_dhcp_vrf1.remove_vpp_config()
+
+if __name__ == '__main__':
+    unittest.main(testRunner=VppTestRunner)
-- 
cgit 1.2.3-korg


From 32e1c010b0c34fd0984f7fc45fae648a182025c5 Mon Sep 17 00:00:00 2001
From: Neale Ranns <nranns@cisco.com>
Date: Tue, 22 Nov 2016 17:07:28 +0000
Subject: IP Multicast FIB (mfib)

 - IPv[46] mfib tables with support for (*,G/m), (*,G) and (S,G) exact and longest prefix match
 - Replication represented via a new replicate DPO.
 - RPF configuration and data-plane checking
 - data-plane signals sent to listening control planes.

The functions of multicast forwarding entries differ from their unicast conterparts, so we introduce a new mfib_table_t and mfib_entry_t objects. However, we re-use the fib_path_list to resolve and build the entry's output list. the fib_path_list provides the service to construct a replicate DPO for multicast.

'make tests' is added to with two new suites; TEST=mfib, this is invocation of the CLI command 'test mfib' which deals with many path add/remove, flag set/unset scenarios, TEST=ip-mcast, data-plane forwarding tests.

Updated applications to use the new MIFB functions;
  - IPv6 NS/RA.
  - DHCPv6
 unit tests for these are undated accordingly.

Change-Id: I49ec37b01f1b170335a5697541c8fd30e6d3a961
Signed-off-by: Neale Ranns <nranns@cisco.com>
---
 src/scripts/vnet/mcast/ip4         |   22 +
 src/vat/api_format.c               |  173 +++++
 src/vnet.am                        |   35 +-
 src/vnet/adj/adj.c                 |   11 +-
 src/vnet/adj/adj.h                 |    1 -
 src/vnet/adj/adj_internal.h        |    2 +
 src/vnet/adj/adj_mcast.c           |  346 ++++++++++
 src/vnet/adj/adj_mcast.h           |   78 +++
 src/vnet/adj/adj_nbr.c             |    2 +-
 src/vnet/adj/adj_rewrite.c         |   53 --
 src/vnet/adj/adj_rewrite.h         |   49 --
 src/vnet/dhcpv6/proxy_node.c       |   46 +-
 src/vnet/dpo/dpo.c                 |    2 +
 src/vnet/dpo/dpo.h                 |    8 +-
 src/vnet/dpo/load_balance.c        |   13 +-
 src/vnet/dpo/load_balance.h        |    8 +
 src/vnet/dpo/replicate_dpo.c       |  759 ++++++++++++++++++++++
 src/vnet/dpo/replicate_dpo.h       |  143 +++++
 src/vnet/ethernet/arp.c            |   84 ++-
 src/vnet/ethernet/ethernet.h       |    2 +
 src/vnet/ethernet/interface.c      |   20 +
 src/vnet/fib/fib_attached_export.c |    4 +-
 src/vnet/fib/fib_entry.h           |    2 +-
 src/vnet/fib/fib_entry_delegate.c  |    3 +
 src/vnet/fib/fib_entry_src.c       |    4 +
 src/vnet/fib/fib_node.h            |    2 +
 src/vnet/fib/fib_path.c            |   91 ++-
 src/vnet/fib/fib_path_list.c       |    9 +-
 src/vnet/fib/fib_path_list.h       |    6 +
 src/vnet/fib/fib_table.c           |   69 +-
 src/vnet/fib/fib_table.h           |   16 +
 src/vnet/fib/fib_test.c            |  207 +++---
 src/vnet/fib/fib_types.c           |    4 +
 src/vnet/fib/fib_types.h           |   14 +
 src/vnet/fib/fib_urpf_list.c       |   20 +-
 src/vnet/fib/ip4_fib.c             |   50 +-
 src/vnet/fib/ip4_fib.h             |    9 +
 src/vnet/fib/ip6_fib.c             |  117 ++--
 src/vnet/fib/ip6_fib.h             |   11 +-
 src/vnet/fib/mpls_fib.c            |   17 +-
 src/vnet/fib/mpls_fib.h            |    9 +
 src/vnet/ip/ip.api                 |   53 ++
 src/vnet/ip/ip4.h                  |   24 +
 src/vnet/ip/ip4_forward.c          |  498 +++++----------
 src/vnet/ip/ip4_input.c            |    4 +-
 src/vnet/ip/ip6.h                  |   27 +
 src/vnet/ip/ip6_forward.c          |   83 ++-
 src/vnet/ip/ip6_input.c            |   43 +-
 src/vnet/ip/ip6_neighbor.c         |  134 ++--
 src/vnet/ip/ip_api.c               |  210 +++++++
 src/vnet/ip/lookup.c               |  171 +++++
 src/vnet/ip/lookup.h               |   82 +--
 src/vnet/mcast/mcast.c             |  565 -----------------
 src/vnet/mcast/mcast.h             |   50 --
 src/vnet/mcast/mcast_test.c        |  149 -----
 src/vnet/mfib/ip4_mfib.c           |  465 ++++++++++++++
 src/vnet/mfib/ip4_mfib.h           |   95 +++
 src/vnet/mfib/ip6_mfib.c           |  663 +++++++++++++++++++
 src/vnet/mfib/ip6_mfib.h           |  109 ++++
 src/vnet/mfib/mfib_entry.c         | 1096 ++++++++++++++++++++++++++++++++
 src/vnet/mfib/mfib_entry.h         |  172 +++++
 src/vnet/mfib/mfib_forward.c       |  512 +++++++++++++++
 src/vnet/mfib/mfib_itf.c           |  119 ++++
 src/vnet/mfib/mfib_itf.h           |   63 ++
 src/vnet/mfib/mfib_signal.c        |  201 ++++++
 src/vnet/mfib/mfib_signal.h        |   59 ++
 src/vnet/mfib/mfib_table.c         |  489 ++++++++++++++
 src/vnet/mfib/mfib_table.h         |  331 ++++++++++
 src/vnet/mfib/mfib_test.c          | 1225 ++++++++++++++++++++++++++++++++++++
 src/vnet/mfib/mfib_types.c         |  213 +++++++
 src/vnet/mfib/mfib_types.h         |  185 ++++++
 src/vnet/misc.c                    |    3 +
 src/vnet/rewrite.h                 |   31 +
 src/vnet/sr/sr.c                   |    4 +-
 src/vnet/util/radix.c              | 1104 ++++++++++++++++++++++++++++++++
 src/vnet/util/radix.h              |  147 +++++
 src/vnet/vxlan/vxlan.c             |  112 +++-
 src/vpp/api/api.c                  |   14 +-
 src/vppinfra.am                    |    2 +-
 src/vppinfra/dlist.h               |    2 +-
 src/vppinfra/format.c              |    8 +-
 src/vppinfra/format.h              |    4 +-
 src/vppinfra/unformat.c            |   16 +-
 test/test_dhcp.py                  |   16 -
 test/test_ip6.py                   |  131 ++--
 test/test_ip_mcast.py              |  612 ++++++++++++++++++
 test/test_mfib.py                  |   23 +
 test/vpp_interface.py              |    3 +-
 test/vpp_ip_route.py               |  101 ++-
 test/vpp_papi_provider.py          |   34 +-
 90 files changed, 11211 insertions(+), 1767 deletions(-)
 create mode 100644 src/scripts/vnet/mcast/ip4
 create mode 100644 src/vnet/adj/adj_mcast.c
 create mode 100644 src/vnet/adj/adj_mcast.h
 delete mode 100644 src/vnet/adj/adj_rewrite.c
 delete mode 100644 src/vnet/adj/adj_rewrite.h
 create mode 100644 src/vnet/dpo/replicate_dpo.c
 create mode 100644 src/vnet/dpo/replicate_dpo.h
 delete mode 100644 src/vnet/mcast/mcast.c
 delete mode 100644 src/vnet/mcast/mcast.h
 delete mode 100644 src/vnet/mcast/mcast_test.c
 create mode 100644 src/vnet/mfib/ip4_mfib.c
 create mode 100644 src/vnet/mfib/ip4_mfib.h
 create mode 100644 src/vnet/mfib/ip6_mfib.c
 create mode 100644 src/vnet/mfib/ip6_mfib.h
 create mode 100644 src/vnet/mfib/mfib_entry.c
 create mode 100644 src/vnet/mfib/mfib_entry.h
 create mode 100644 src/vnet/mfib/mfib_forward.c
 create mode 100644 src/vnet/mfib/mfib_itf.c
 create mode 100644 src/vnet/mfib/mfib_itf.h
 create mode 100644 src/vnet/mfib/mfib_signal.c
 create mode 100644 src/vnet/mfib/mfib_signal.h
 create mode 100644 src/vnet/mfib/mfib_table.c
 create mode 100644 src/vnet/mfib/mfib_table.h
 create mode 100644 src/vnet/mfib/mfib_test.c
 create mode 100644 src/vnet/mfib/mfib_types.c
 create mode 100644 src/vnet/mfib/mfib_types.h
 create mode 100644 src/vnet/util/radix.c
 create mode 100644 src/vnet/util/radix.h
 create mode 100644 test/test_ip_mcast.py
 create mode 100644 test/test_mfib.py

(limited to 'test/test_dhcp.py')

diff --git a/src/scripts/vnet/mcast/ip4 b/src/scripts/vnet/mcast/ip4
new file mode 100644
index 00000000..69f1ee00
--- /dev/null
+++ b/src/scripts/vnet/mcast/ip4
@@ -0,0 +1,22 @@
+packet-generator new {
+  name x
+  limit 1
+  node ip4-input
+  size 64-64
+  no-recycle
+  data {
+    ICMP: 1.0.0.2 -> 232.1.1.1
+    ICMP echo_request
+    incrementing 100
+  }
+}
+
+trace add pg-input 100
+loop create
+loop create
+set int state loop0 up
+set int state loop1 up
+
+ip mroute add 232.1.1.1 via pg0 Accept
+ip mroute add 232.1.1.1 via loop0 Forward
+ip mroute add 232.1.1.1 via loop1 Forward
diff --git a/src/vat/api_format.c b/src/vat/api_format.c
index b83313de..4cfe4a58 100644
--- a/src/vat/api_format.c
+++ b/src/vat/api_format.c
@@ -48,6 +48,7 @@
 #include <vnet/span/span.h>
 #include <vnet/policer/policer.h>
 #include <vnet/policer/police.h>
+#include <vnet/mfib/mfib_types.h>
 
 #include "vat/json_format.h"
 
@@ -505,6 +506,53 @@ unformat_flow_classify_table_type (unformat_input_t * input, va_list * va)
   return 1;
 }
 
+static const char *mfib_flag_names[] = MFIB_ENTRY_NAMES_SHORT;
+static const char *mfib_flag_long_names[] = MFIB_ENTRY_NAMES_LONG;
+static const char *mfib_itf_flag_long_names[] = MFIB_ITF_NAMES_LONG;
+static const char *mfib_itf_flag_names[] = MFIB_ITF_NAMES_SHORT;
+
+uword
+unformat_mfib_itf_flags (unformat_input_t * input, va_list * args)
+{
+  mfib_itf_flags_t old, *iflags = va_arg (*args, mfib_itf_flags_t *);
+  mfib_itf_attribute_t attr;
+
+  old = *iflags;
+  FOR_EACH_MFIB_ITF_ATTRIBUTE (attr)
+  {
+    if (unformat (input, mfib_itf_flag_long_names[attr]))
+      *iflags |= (1 << attr);
+  }
+  FOR_EACH_MFIB_ITF_ATTRIBUTE (attr)
+  {
+    if (unformat (input, mfib_itf_flag_names[attr]))
+      *iflags |= (1 << attr);
+  }
+
+  return (old == *iflags ? 0 : 1);
+}
+
+uword
+unformat_mfib_entry_flags (unformat_input_t * input, va_list * args)
+{
+  mfib_entry_flags_t old, *eflags = va_arg (*args, mfib_entry_flags_t *);
+  mfib_entry_attribute_t attr;
+
+  old = *eflags;
+  FOR_EACH_MFIB_ATTRIBUTE (attr)
+  {
+    if (unformat (input, mfib_flag_long_names[attr]))
+      *eflags |= (1 << attr);
+  }
+  FOR_EACH_MFIB_ATTRIBUTE (attr)
+  {
+    if (unformat (input, mfib_flag_names[attr]))
+      *eflags |= (1 << attr);
+  }
+
+  return (old == *eflags ? 0 : 1);
+}
+
 #if (VPP_API_TEST_BUILTIN==0)
 u8 *
 format_ip4_address (u8 * s, va_list * args)
@@ -3592,6 +3640,7 @@ _(bridge_domain_add_del_reply)                          \
 _(sw_interface_set_l2_xconnect_reply)                   \
 _(l2fib_add_del_reply)                                  \
 _(ip_add_del_route_reply)                               \
+_(ip_mroute_add_del_reply)                              \
 _(mpls_route_add_del_reply)                             \
 _(mpls_ip_bind_unbind_reply)                            \
 _(proxy_arp_add_del_reply)                              \
@@ -3792,6 +3841,7 @@ _(TAP_MODIFY_REPLY, tap_modify_reply)					\
 _(TAP_DELETE_REPLY, tap_delete_reply)					\
 _(SW_INTERFACE_TAP_DETAILS, sw_interface_tap_details)                   \
 _(IP_ADD_DEL_ROUTE_REPLY, ip_add_del_route_reply)			\
+_(IP_MROUTE_ADD_DEL_REPLY, ip_mroute_add_del_reply)			\
 _(MPLS_ROUTE_ADD_DEL_REPLY, mpls_route_add_del_reply)			\
 _(MPLS_IP_BIND_UNBIND_REPLY, mpls_ip_bind_unbind_reply)			\
 _(PROXY_ARP_ADD_DEL_REPLY, proxy_arp_add_del_reply)                     \
@@ -6383,6 +6433,126 @@ api_ip_add_del_route (vat_main_t * vam)
   return (vam->retval);
 }
 
+static int
+api_ip_mroute_add_del (vat_main_t * vam)
+{
+  unformat_input_t *i = vam->input;
+  vl_api_ip_mroute_add_del_t *mp;
+  f64 timeout;
+  u32 sw_if_index = ~0, vrf_id = 0;
+  u8 is_ipv6 = 0;
+  u8 is_local = 0;
+  u8 create_vrf_if_needed = 0;
+  u8 is_add = 1;
+  u8 address_set = 0;
+  u32 grp_address_length = 0;
+  ip4_address_t v4_grp_address, v4_src_address;
+  ip6_address_t v6_grp_address, v6_src_address;
+  mfib_itf_flags_t iflags = 0;
+  mfib_entry_flags_t eflags = 0;
+
+  /* Parse args required to build the message */
+  while (unformat_check_input (i) != UNFORMAT_END_OF_INPUT)
+    {
+      if (unformat (i, "sw_if_index %d", &sw_if_index))
+	;
+      else if (unformat (i, "%U %U",
+			 unformat_ip4_address, &v4_src_address,
+			 unformat_ip4_address, &v4_grp_address))
+	{
+	  grp_address_length = 64;
+	  address_set = 1;
+	  is_ipv6 = 0;
+	}
+      else if (unformat (i, "%U %U",
+			 unformat_ip6_address, &v6_src_address,
+			 unformat_ip6_address, &v6_grp_address))
+	{
+	  grp_address_length = 256;
+	  address_set = 1;
+	  is_ipv6 = 1;
+	}
+      else if (unformat (i, "%U", unformat_ip4_address, &v4_grp_address))
+	{
+	  memset (&v4_src_address, 0, sizeof (v4_src_address));
+	  grp_address_length = 32;
+	  address_set = 1;
+	  is_ipv6 = 0;
+	}
+      else if (unformat (i, "%U", unformat_ip6_address, &v6_grp_address))
+	{
+	  memset (&v6_src_address, 0, sizeof (v6_src_address));
+	  grp_address_length = 128;
+	  address_set = 1;
+	  is_ipv6 = 1;
+	}
+      else if (unformat (i, "/%d", &grp_address_length))
+	;
+      else if (unformat (i, "local"))
+	{
+	  is_local = 1;
+	}
+      else if (unformat (i, "del"))
+	is_add = 0;
+      else if (unformat (i, "add"))
+	is_add = 1;
+      else if (unformat (i, "vrf %d", &vrf_id))
+	;
+      else if (unformat (i, "create-vrf"))
+	create_vrf_if_needed = 1;
+      else if (unformat (i, "%U", unformat_mfib_itf_flags, &iflags))
+	;
+      else if (unformat (i, "%U", unformat_mfib_entry_flags, &eflags))
+	;
+      else
+	{
+	  clib_warning ("parse error '%U'", format_unformat_error, i);
+	  return -99;
+	}
+    }
+
+  if (address_set == 0)
+    {
+      errmsg ("missing addresses\n");
+      return -99;
+    }
+
+  /* Construct the API message */
+  M (IP_MROUTE_ADD_DEL, ip_mroute_add_del);
+
+  mp->next_hop_sw_if_index = ntohl (sw_if_index);
+  mp->table_id = ntohl (vrf_id);
+  mp->create_vrf_if_needed = create_vrf_if_needed;
+
+  mp->is_add = is_add;
+  mp->is_ipv6 = is_ipv6;
+  mp->is_local = is_local;
+  mp->itf_flags = ntohl (iflags);
+  mp->entry_flags = ntohl (eflags);
+  mp->grp_address_length = grp_address_length;
+  mp->grp_address_length = ntohs (mp->grp_address_length);
+
+  if (is_ipv6)
+    {
+      clib_memcpy (mp->grp_address, &v6_grp_address, sizeof (v6_grp_address));
+      clib_memcpy (mp->src_address, &v6_src_address, sizeof (v6_src_address));
+    }
+  else
+    {
+      clib_memcpy (mp->grp_address, &v4_grp_address, sizeof (v4_grp_address));
+      clib_memcpy (mp->src_address, &v4_src_address, sizeof (v4_src_address));
+
+    }
+
+  /* send it... */
+  S;
+  /* Wait for a reply... */
+  W;
+
+  /* Return the good/bad news */
+  return (vam->retval);
+}
+
 static int
 api_mpls_route_add_del (vat_main_t * vam)
 {
@@ -17512,6 +17682,9 @@ _(ip_add_del_route,                                                     \
   "[<intfc> | sw_if_index <id>] [resolve-attempts <n>]\n"               \
   "[weight <n>] [drop] [local] [classify <n>] [del]\n"                  \
   "[multipath] [count <n>]")                                            \
+_(ip_mroute_add_del,                                                    \
+  "<src> <grp>/<mask> [table-id <n>]\n"                                 \
+  "[<intfc> | sw_if_index <id>] [local] [del]")                         \
 _(mpls_route_add_del,                                                   \
   "<label> <eos> via <addr> [table-id <n>]\n"                           \
   "[<intfc> | sw_if_index <id>] [resolve-attempts <n>]\n"               \
diff --git a/src/vnet.am b/src/vnet.am
index c6922493..6d043e32 100644
--- a/src/vnet.am
+++ b/src/vnet.am
@@ -946,16 +946,15 @@ nobase_include_HEADERS +=			\
 
 libvnet_la_SOURCES +=				\
   vnet/adj/adj_nbr.c				\
-  vnet/adj/adj_rewrite.c			\
   vnet/adj/adj_glean.c   			\
   vnet/adj/adj_midchain.c   			\
+  vnet/adj/adj_mcast.c   			\
   vnet/adj/adj_l2.c      			\
   vnet/adj/adj.c
 
 nobase_include_HEADERS +=			\
   vnet/adj/adj.h				\
   vnet/adj/adj_types.h				\
-  vnet/adj/adj_rewrite.h			\
   vnet/adj/adj_glean.h  			\
   vnet/adj/adj_nbr.h
 
@@ -971,8 +970,9 @@ libvnet_la_SOURCES +=				\
   vnet/dpo/receive_dpo.c			\
   vnet/dpo/load_balance.c			\
   vnet/dpo/load_balance_map.c			\
-  vnet/dpo/lookup_dpo.c				\
-  vnet/dpo/classify_dpo.c			\
+  vnet/dpo/lookup_dpo.c   			\
+  vnet/dpo/classify_dpo.c   			\
+  vnet/dpo/replicate_dpo.c   			\
   vnet/dpo/mpls_label_dpo.c
 
 nobase_include_HEADERS +=			\
@@ -985,6 +985,33 @@ nobase_include_HEADERS +=			\
   vnet/dpo/ip_null_dpo.h			\
   vnet/dpo/dpo.h
 
+########################################
+# Multicast FIB
+########################################
+
+libvnet_la_SOURCES +=				\
+  vnet/mfib/mfib_test.c                         \
+  vnet/mfib/mfib_forward.c                      \
+  vnet/mfib/ip4_mfib.c                          \
+  vnet/mfib/ip6_mfib.c                          \
+  vnet/mfib/mfib_types.c                        \
+  vnet/mfib/mfib_signal.c                       \
+  vnet/mfib/mfib_itf.c                          \
+  vnet/mfib/mfib_entry.c                        \
+  vnet/mfib/mfib_table.c
+
+nobase_include_HEADERS +=			\
+  vnet/mfib/ip4_mfib.h                          \
+  vnet/mfib/mfib_types.h                        \
+  vnet/mfib/mfib_table.h
+
+########################################
+# Utilities
+########################################
+
+libvnet_la_SOURCES +=                    \
+  vnet/util/radix.c
+
 ########################################
 # Plugin client library
 ########################################
diff --git a/src/vnet/adj/adj.c b/src/vnet/adj/adj.c
index d0be0f0e..a99f173f 100644
--- a/src/vnet/adj/adj.c
+++ b/src/vnet/adj/adj.c
@@ -17,6 +17,7 @@
 #include <vnet/adj/adj_internal.h>
 #include <vnet/adj/adj_glean.h>
 #include <vnet/adj/adj_midchain.h>
+#include <vnet/adj/adj_mcast.h>
 #include <vnet/fib/fib_node_list.h>
 
 /*
@@ -58,8 +59,6 @@ adj_alloc (fib_protocol_t proto)
                                    adj_get_index(adj));
 
     adj->rewrite_header.sw_if_index = ~0;
-    adj->mcast_group_index = ~0;
-    adj->saved_lookup_next_index = 0;
     adj->n_adj = 1;
     adj->lookup_next_index = 0;
 
@@ -116,6 +115,9 @@ format_ip_adjacency (u8 * s, va_list * args)
     case IP_LOOKUP_NEXT_MIDCHAIN:
 	s = format (s, "%U", format_adj_midchain, adj_index, 2);
 	break;
+    case IP_LOOKUP_NEXT_MCAST:
+	s = format (s, "%U", format_adj_mcast, adj_index, 0);
+	break;
     default:
 	break;
     }
@@ -179,6 +181,10 @@ adj_last_lock_gone (ip_adjacency_t *adj)
 	adj_glean_remove(adj->ia_nh_proto,
 			 adj->rewrite_header.sw_if_index);
 	break;
+    case IP_LOOKUP_NEXT_MCAST:
+	adj_mcast_remove(adj->ia_nh_proto,
+			 adj->rewrite_header.sw_if_index);
+	break;
     default:
 	/*
 	 * type not stored in any DB from which we need to remove it
@@ -350,6 +356,7 @@ adj_module_init (vlib_main_t * vm)
     adj_nbr_module_init();
     adj_glean_module_init();
     adj_midchain_module_init();
+    adj_mcast_module_init();
 
     /*
      * one special adj to reserve index 0
diff --git a/src/vnet/adj/adj.h b/src/vnet/adj/adj.h
index e85625db..29bae673 100644
--- a/src/vnet/adj/adj.h
+++ b/src/vnet/adj/adj.h
@@ -45,7 +45,6 @@
 #include <vnet/ip/lookup.h>
 #include <vnet/adj/adj_types.h>
 #include <vnet/adj/adj_nbr.h>
-#include <vnet/adj/adj_rewrite.h>
 #include <vnet/adj/adj_glean.h>
 
 /**
diff --git a/src/vnet/adj/adj_internal.h b/src/vnet/adj/adj_internal.h
index 833bc7c9..ece59121 100644
--- a/src/vnet/adj/adj_internal.h
+++ b/src/vnet/adj/adj_internal.h
@@ -100,5 +100,7 @@ extern void adj_nbr_remove(adj_index_t ai,
 			   u32 sw_if_index);
 extern void adj_glean_remove(fib_protocol_t proto,
 			     u32 sw_if_index);
+extern void adj_mcast_remove(fib_protocol_t proto,
+			     u32 sw_if_index);
 
 #endif
diff --git a/src/vnet/adj/adj_mcast.c b/src/vnet/adj/adj_mcast.c
new file mode 100644
index 00000000..1345aedb
--- /dev/null
+++ b/src/vnet/adj/adj_mcast.c
@@ -0,0 +1,346 @@
+/*
+ * Copyright (c) 2016 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <vnet/adj/adj.h>
+#include <vnet/adj/adj_internal.h>
+#include <vnet/fib/fib_walk.h>
+#include <vnet/ip/ip.h>
+
+/*
+ * The 'DB' of all mcast adjs.
+ * There is only one mcast per-interface per-protocol, so this is a per-interface
+ * vector
+ */
+static adj_index_t *adj_mcasts[FIB_PROTOCOL_MAX];
+
+static u32
+adj_get_mcast_node (fib_protocol_t proto)
+{
+    switch (proto) {
+    case FIB_PROTOCOL_IP4:
+	return (ip4_rewrite_mcast_node.index);
+    case FIB_PROTOCOL_IP6:
+	return (ip6_rewrite_mcast_node.index);
+    case FIB_PROTOCOL_MPLS:
+	break;
+    }
+    ASSERT(0);
+    return (0);
+}
+
+/*
+ * adj_mcast_add_or_lock
+ *
+ * The next_hop address here is used for source address selection in the DP.
+ * The mcast adj is added to an interface's connected prefix, the next-hop
+ * passed here is the local prefix on the same interface.
+ */
+adj_index_t
+adj_mcast_add_or_lock (fib_protocol_t proto,
+                       vnet_link_t link_type,
+		       u32 sw_if_index)
+{
+    ip_adjacency_t * adj;
+
+    vec_validate_init_empty(adj_mcasts[proto], sw_if_index, ADJ_INDEX_INVALID);
+
+    if (ADJ_INDEX_INVALID == adj_mcasts[proto][sw_if_index])
+    {
+        vnet_main_t *vnm;
+
+        vnm = vnet_get_main();
+	adj = adj_alloc(proto);
+
+	adj->lookup_next_index = IP_LOOKUP_NEXT_MCAST;
+	adj->ia_nh_proto = proto;
+	adj->ia_link = link_type;
+	adj_mcasts[proto][sw_if_index] = adj_get_index(adj);
+        adj_lock(adj_get_index(adj));
+
+	vnet_rewrite_init(vnm, sw_if_index,
+			  adj_get_mcast_node(proto),
+			  vnet_tx_node_index_for_sw_interface(vnm, sw_if_index),
+			  &adj->rewrite_header);
+
+	/*
+	 * we need a rewrite where the destination IP address is converted
+	 * to the appropriate link-layer address. This is interface specific.
+	 * So ask the interface to do it.
+	 */
+	vnet_update_adjacency_for_sw_interface(vnm, sw_if_index,
+                                               adj_get_index(adj));
+    }
+    else
+    {
+	adj = adj_get(adj_mcasts[proto][sw_if_index]);
+        adj_lock(adj_get_index(adj));
+    }
+
+    return (adj_get_index(adj));
+}
+
+/**
+ * adj_mcast_update_rewrite
+ *
+ * Update the adjacency's rewrite string. A NULL string implies the
+ * rewirte is reset (i.e. when ARP/ND etnry is gone).
+ * NB: the adj being updated may be handling traffic in the DP.
+ */
+void
+adj_mcast_update_rewrite (adj_index_t adj_index,
+                          u8 *rewrite)
+{
+    ip_adjacency_t *adj;
+
+    ASSERT(ADJ_INDEX_INVALID != adj_index);
+
+    adj = adj_get(adj_index);
+
+    /*
+     * update the adj's rewrite string and build the arc
+     * from the rewrite node to the interface's TX node
+     */
+    adj_nbr_update_rewrite_internal(adj, IP_LOOKUP_NEXT_MCAST,
+                                    adj_get_mcast_node(adj->ia_nh_proto),
+                                    vnet_tx_node_index_for_sw_interface(
+                                        vnet_get_main(),
+                                        adj->rewrite_header.sw_if_index),
+                                    rewrite);
+}
+
+void
+adj_mcast_remove (fib_protocol_t proto,
+		  u32 sw_if_index)
+{
+    ASSERT(sw_if_index < vec_len(adj_mcasts[proto]));
+
+    adj_mcasts[proto][sw_if_index] = ADJ_INDEX_INVALID;
+}
+
+static clib_error_t *
+adj_mcast_interface_state_change (vnet_main_t * vnm,
+				  u32 sw_if_index,
+				  u32 flags)
+{
+    /*
+     * for each mcast on the interface trigger a walk back to the children
+     */
+    fib_protocol_t proto;
+    ip_adjacency_t *adj;
+
+
+    for (proto = FIB_PROTOCOL_IP4; proto <= FIB_PROTOCOL_IP6; proto++)
+    {
+	if (sw_if_index >= vec_len(adj_mcasts[proto]) ||
+	    ADJ_INDEX_INVALID == adj_mcasts[proto][sw_if_index])
+	    continue;
+
+	adj = adj_get(adj_mcasts[proto][sw_if_index]);
+
+	fib_node_back_walk_ctx_t bw_ctx = {
+	    .fnbw_reason = (flags & VNET_SW_INTERFACE_FLAG_ADMIN_UP ?
+			    FIB_NODE_BW_REASON_FLAG_INTERFACE_UP :
+			    FIB_NODE_BW_REASON_FLAG_INTERFACE_DOWN),
+	};
+
+	fib_walk_sync(FIB_NODE_TYPE_ADJ, adj_get_index(adj), &bw_ctx);
+    }
+
+    return (NULL);
+}
+
+VNET_SW_INTERFACE_ADMIN_UP_DOWN_FUNCTION(adj_mcast_interface_state_change);
+
+/**
+ * @brief Invoked on each SW interface of a HW interface when the
+ * HW interface state changes
+ */
+static void
+adj_nbr_hw_sw_interface_state_change (vnet_main_t * vnm,
+                                      u32 sw_if_index,
+                                      void *arg)
+{
+    adj_mcast_interface_state_change(vnm, sw_if_index, (uword) arg);
+}
+
+/**
+ * @brief Registered callback for HW interface state changes
+ */
+static clib_error_t *
+adj_mcast_hw_interface_state_change (vnet_main_t * vnm,
+                                     u32 hw_if_index,
+                                     u32 flags)
+{
+    /*
+     * walk SW interfaces on the HW
+     */
+    uword sw_flags;
+
+    sw_flags = ((flags & VNET_HW_INTERFACE_FLAG_LINK_UP) ?
+                VNET_SW_INTERFACE_FLAG_ADMIN_UP :
+                0);
+
+    vnet_hw_interface_walk_sw(vnm, hw_if_index,
+                              adj_nbr_hw_sw_interface_state_change,
+                              (void*) sw_flags);
+
+    return (NULL);
+}
+
+VNET_HW_INTERFACE_LINK_UP_DOWN_FUNCTION(
+    adj_mcast_hw_interface_state_change);
+
+static clib_error_t *
+adj_mcast_interface_delete (vnet_main_t * vnm,
+			    u32 sw_if_index,
+			    u32 is_add)
+{
+    /*
+     * for each mcast on the interface trigger a walk back to the children
+     */
+    fib_protocol_t proto;
+    ip_adjacency_t *adj;
+
+    if (is_add)
+    {
+	/*
+	 * not interested in interface additions. we will not back walk
+	 * to resolve paths through newly added interfaces. Why? The control
+	 * plane should have the brains to add interfaces first, then routes.
+	 * So the case where there are paths with a interface that matches
+	 * one just created is the case where the path resolved through an
+	 * interface that was deleted, and still has not been removed. The
+	 * new interface added, is NO GUARANTEE that the interface being
+	 * added now, even though it may have the same sw_if_index, is the
+	 * same interface that the path needs. So tough!
+	 * If the control plane wants these routes to resolve it needs to
+	 * remove and add them again.
+	 */
+	return (NULL);
+    }
+
+    for (proto = FIB_PROTOCOL_IP4; proto <= FIB_PROTOCOL_IP6; proto++)
+    {
+	if (sw_if_index >= vec_len(adj_mcasts[proto]) ||
+	    ADJ_INDEX_INVALID == adj_mcasts[proto][sw_if_index])
+	    continue;
+
+	adj = adj_get(adj_mcasts[proto][sw_if_index]);
+
+	fib_node_back_walk_ctx_t bw_ctx = {
+	    .fnbw_reason =  FIB_NODE_BW_REASON_FLAG_INTERFACE_DELETE,
+	};
+
+	fib_walk_sync(FIB_NODE_TYPE_ADJ, adj_get_index(adj), &bw_ctx);
+    }
+
+    return (NULL);
+}
+
+VNET_SW_INTERFACE_ADD_DEL_FUNCTION(adj_mcast_interface_delete);
+
+u8*
+format_adj_mcast (u8* s, va_list *ap)
+{
+    index_t index = va_arg(*ap, index_t);
+    CLIB_UNUSED(u32 indent) = va_arg(*ap, u32);
+    vnet_main_t * vnm = vnet_get_main();
+    ip_adjacency_t * adj = adj_get(index);
+
+    s = format(s, "%U-mcast: ",
+               format_fib_protocol, adj->ia_nh_proto);
+    s = format (s, "%U",
+		format_vnet_rewrite,
+		vnm->vlib_main, &adj->rewrite_header,
+                sizeof (adj->rewrite_data), 0);
+
+    return (s);
+}
+
+
+static void
+adj_dpo_lock (dpo_id_t *dpo)
+{
+    adj_lock(dpo->dpoi_index);
+}
+static void
+adj_dpo_unlock (dpo_id_t *dpo)
+{
+    adj_unlock(dpo->dpoi_index);
+}
+
+const static dpo_vft_t adj_mcast_dpo_vft = {
+    .dv_lock = adj_dpo_lock,
+    .dv_unlock = adj_dpo_unlock,
+    .dv_format = format_adj_mcast,
+};
+
+/**
+ * @brief The per-protocol VLIB graph nodes that are assigned to a mcast
+ *        object.
+ *
+ * this means that these graph nodes are ones from which a mcast is the
+ * parent object in the DPO-graph.
+ */
+const static char* const adj_mcast_ip4_nodes[] =
+{
+    "ip4-rewrite-mcast",
+    NULL,
+};
+const static char* const adj_mcast_ip6_nodes[] =
+{
+    "ip6-rewrite-mcast",
+    NULL,
+};
+
+const static char* const * const adj_mcast_nodes[DPO_PROTO_NUM] =
+{
+    [DPO_PROTO_IP4]  = adj_mcast_ip4_nodes,
+    [DPO_PROTO_IP6]  = adj_mcast_ip6_nodes,
+    [DPO_PROTO_MPLS] = NULL,
+};
+
+/**
+ * @brief Return the size of the adj DB.
+ * This is only for testing purposes so an efficient implementation is not needed
+ */
+u32
+adj_mcast_db_size (void)
+{
+    u32 n_adjs, sw_if_index;
+    fib_protocol_t proto;
+
+    n_adjs = 0;
+    for (proto = FIB_PROTOCOL_IP4; proto <= FIB_PROTOCOL_IP6; proto++)
+    {
+        for (sw_if_index = 0;
+             sw_if_index < vec_len(adj_mcasts[proto]);
+             sw_if_index++)
+        {
+            if (ADJ_INDEX_INVALID != adj_mcasts[proto][sw_if_index])
+            {
+                n_adjs++;
+            }
+        }
+    }
+    
+    return (n_adjs);
+}
+
+void
+adj_mcast_module_init (void)
+{
+    dpo_register(DPO_ADJACENCY_MCAST, &adj_mcast_dpo_vft, adj_mcast_nodes);
+}
diff --git a/src/vnet/adj/adj_mcast.h b/src/vnet/adj/adj_mcast.h
new file mode 100644
index 00000000..21c5a141
--- /dev/null
+++ b/src/vnet/adj/adj_mcast.h
@@ -0,0 +1,78 @@
+/*
+ * Copyright (c) 2016 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+/**
+ * @brief Mcast Adjacency
+ *
+ * The multicast adjacency forwards IP traffic on an interface toward a multicast
+ * group address. This is a different type of adjacency to a unicast adjacency
+ * since the application of the MAC header is different, and so the VLIB node
+ * visited is also different. DPO types have different VLIB nodes.
+ */
+
+#ifndef __ADJ_MCAST_H__
+#define __ADJ_MCAST_H__
+
+#include <vnet/adj/adj_types.h>
+
+/**
+ * @brief
+ *  Add (and lock) a new or lock an existing mcast adjacency
+ *
+ * @param proto
+ *  The protocol for the neighbours that we wish to mcast
+ *
+ * @param link_type
+ *  A description of the protocol of the packets that will forward
+ *  through this adj. On an ethernet interface this is the MAC header's
+ *  ether-type
+ *
+ * @param sw_if_index
+ *  The interface on which to mcast
+ */
+extern adj_index_t adj_mcast_add_or_lock(fib_protocol_t proto,
+                                         vnet_link_t link_type,
+					 u32 sw_if_index);
+
+/**
+ * @brief
+ *  Update the rewrite string for an existing adjacecny.
+ *
+ * @param
+ *  The index of the adj to update
+ *
+ * @param
+ *  The new rewrite
+ */
+extern void adj_mcast_update_rewrite(adj_index_t adj_index,
+                                     u8 *rewrite);
+
+/**
+ * @brief Format/display a mcast adjacency.
+ */
+extern u8* format_adj_mcast(u8* s, va_list *ap);
+
+/**
+ * @brief Get the sze of the mcast adj DB. Test purposes only.
+ */
+extern u32 adj_mcast_db_size(void);
+
+/**
+ * @brief
+ *  Module initialisation
+ */
+extern void adj_mcast_module_init(void);
+
+#endif
diff --git a/src/vnet/adj/adj_nbr.c b/src/vnet/adj/adj_nbr.c
index 1344bb67..9e8073d3 100644
--- a/src/vnet/adj/adj_nbr.c
+++ b/src/vnet/adj/adj_nbr.c
@@ -162,7 +162,7 @@ adj_nbr_alloc (fib_protocol_t nh_proto,
 }
 
 /*
- * adj_add_for_nbr
+ * adj_nbr_add_or_lock
  *
  * Add an adjacency for the neighbour requested.
  *
diff --git a/src/vnet/adj/adj_rewrite.c b/src/vnet/adj/adj_rewrite.c
deleted file mode 100644
index 7d792557..00000000
--- a/src/vnet/adj/adj_rewrite.c
+++ /dev/null
@@ -1,53 +0,0 @@
-/*
- * Copyright (c) 2016 Cisco and/or its affiliates.
- * Licensed under the Apache License, Version 2.0 (the "License");
- * you may not use this file except in compliance with the License.
- * You may obtain a copy of the License at:
- *
- *     http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-
-#include <vnet/adj/adj.h>
-#include <vnet/adj/adj_internal.h>
-
-/**
- * adj_rewrite_add_and_lock
- *
- * A rewrite sub-type has the rewrite string provided, but no key
- */
-adj_index_t
-adj_rewrite_add_and_lock (fib_protocol_t nh_proto,
-			  vnet_link_t link_type,
-			  u32 sw_if_index,
-			  u8 *rewrite)
-{
-    ip_adjacency_t *adj;
-
-    adj = adj_alloc(nh_proto);
-
-    adj->lookup_next_index = IP_LOOKUP_NEXT_REWRITE;
-    memset(&adj->sub_type.nbr.next_hop, 0, sizeof(adj->sub_type.nbr.next_hop));
-    adj->ia_link = link_type;
-    adj->ia_nh_proto = nh_proto;
-    adj->rewrite_header.sw_if_index = sw_if_index;
-
-    ASSERT(NULL != rewrite);
-
-    vnet_rewrite_for_sw_interface(vnet_get_main(),
-				  link_type,
-				  adj->rewrite_header.sw_if_index,
-				  adj_get_rewrite_node(link_type),
-				  rewrite,
-				  &adj->rewrite_header,
-				  sizeof (adj->rewrite_data));
-
-    adj_lock(adj_get_index(adj));
-
-    return (adj_get_index(adj));
-}
diff --git a/src/vnet/adj/adj_rewrite.h b/src/vnet/adj/adj_rewrite.h
deleted file mode 100644
index 25e6bba8..00000000
--- a/src/vnet/adj/adj_rewrite.h
+++ /dev/null
@@ -1,49 +0,0 @@
-/*
- * Copyright (c) 2016 Cisco and/or its affiliates.
- * Licensed under the Apache License, Version 2.0 (the "License");
- * you may not use this file except in compliance with the License.
- * You may obtain a copy of the License at:
- *
- *     http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-/**
- * @brief
- * A rewrite adjacency has no key, and thus cannot be 'found' from the
- * FIB resolution code. the client therefore needs to maange these adjacencies
- */
-
-#ifndef __ADJ_REWRITE_H__
-#define __ADJ_REWRITE_H__
-
-#include <vnet/adj/adj_types.h>
-
-/**
- * @brief
- *  Add (and lock) a new or lock an existing neighbour adjacency
- *
- * @param nh_proto
- *  The protocol for the next-hop address (v4 or v6)
- *
- * @param link_type
- *  A description of the protocol of the packets that will forward
- *  through this adj. On an ethernet interface this is the MAC header's
- *  ether-type
- *
- * @param sw_if_index
- *  The interface on which the peer resides
- *
- * @param rewrite
- *  The rewrite to prepend to packets
- */
-extern adj_index_t adj_rewrite_add_and_lock(fib_protocol_t nh_proto,
-					    vnet_link_t link_type,
-					    u32 sw_if_index,
-					    u8 *rewrite);
-
-#endif
diff --git a/src/vnet/dhcpv6/proxy_node.c b/src/vnet/dhcpv6/proxy_node.c
index 4dd2239f..77afef2a 100644
--- a/src/vnet/dhcpv6/proxy_node.c
+++ b/src/vnet/dhcpv6/proxy_node.c
@@ -19,6 +19,8 @@
 #include <vnet/pg/pg.h>
 #include <vnet/dhcpv6/proxy.h>
 #include <vnet/fib/ip6_fib.h>
+#include <vnet/mfib/mfib_table.h>
+#include <vnet/mfib/ip6_mfib.h>
 
 static char * dhcpv6_proxy_error_strings[] = {
 #define dhcpv6_proxy_error(n,s) s,
@@ -819,7 +821,7 @@ int dhcpv6_proxy_set_server_2 (ip6_address_t *addr, ip6_address_t *src_address,
   u32 server_fib_index = 0;
   u32 rx_fib_index = 0;
 
-  rx_fib_index = ip6_fib_table_find_or_create_and_lock(rx_fib_id);
+  rx_fib_index = ip6_mfib_table_find_or_create_and_lock(rx_fib_id);
   server_fib_index = ip6_fib_table_find_or_create_and_lock(server_fib_id);
 
   if (is_del)
@@ -848,8 +850,10 @@ int dhcpv6_proxy_set_server_2 (ip6_address_t *addr, ip6_address_t *src_address,
   if (rx_fib_id == 0)
     {
       server = pool_elt_at_index (dm->dhcp6_servers, 0);
-
-      goto initialize_it;
+      if (server->valid)
+          goto reconfigure_it;
+      else
+          goto initialize_it;
     }
 
   if (rx_fib_index < vec_len(dm->dhcp6_server_index_by_rx_fib_index))
@@ -866,6 +870,42 @@ int dhcpv6_proxy_set_server_2 (ip6_address_t *addr, ip6_address_t *src_address,
   pool_get (dm->dhcp6_servers, server);
 
   initialize_it:
+  {
+      const mfib_prefix_t all_dhcp_servers = {
+          .fp_len = 128,
+          .fp_proto = FIB_PROTOCOL_IP6,
+          .fp_grp_addr = {
+              .ip6 = dm->all_dhcpv6_server_relay_agent_address,
+          }
+      };
+      const fib_route_path_t path_for_us = {
+          .frp_proto = FIB_PROTOCOL_IP6,
+          .frp_addr = zero_addr,
+          .frp_sw_if_index = 0xffffffff,
+          .frp_fib_index = ~0,
+          .frp_weight = 0,
+          .frp_flags = FIB_ROUTE_PATH_LOCAL,
+      };
+      mfib_table_entry_path_update(rx_fib_index,
+                                   &all_dhcp_servers,
+                                   MFIB_SOURCE_DHCP,
+                                   &path_for_us,
+                                   MFIB_ITF_FLAG_FORWARD);
+      /*
+       * Each interface that is enabled in this table, needs to be added
+       * as an accepting interface, but this is not easily doable in VPP.
+       * So we cheat. Add a flag to the entry that indicates accept form
+       * any interface.
+       * We will still only accept on v6 enabled interfaces, since the input
+       * feature ensures this.
+       */
+      mfib_table_entry_update(rx_fib_index,
+                              &all_dhcp_servers,
+                              MFIB_SOURCE_DHCP,
+                              MFIB_ENTRY_FLAG_ACCEPT_ALL_ITF);
+  }
+
+reconfigure_it:
 
   copy_ip6_address(&server->dhcp6_server, addr);
   copy_ip6_address(&server->dhcp6_src_address, src_address);
diff --git a/src/vnet/dpo/dpo.c b/src/vnet/dpo/dpo.c
index 688d2892..cc2fa0eb 100644
--- a/src/vnet/dpo/dpo.c
+++ b/src/vnet/dpo/dpo.c
@@ -36,6 +36,7 @@
 #include <vnet/dpo/punt_dpo.h>
 #include <vnet/dpo/classify_dpo.h>
 #include <vnet/dpo/ip_null_dpo.h>
+#include <vnet/dpo/replicate_dpo.h>
 
 /**
  * Array of char* names for the DPO types and protos
@@ -449,6 +450,7 @@ dpo_module_init (vlib_main_t * vm)
     classify_dpo_module_init();
     lookup_dpo_module_init();
     ip_null_dpo_module_init();
+    replicate_module_init();
 
     return (NULL);
 }
diff --git a/src/vnet/dpo/dpo.h b/src/vnet/dpo/dpo.h
index 1efcbc88..aff4e1b8 100644
--- a/src/vnet/dpo/dpo.h
+++ b/src/vnet/dpo/dpo.h
@@ -100,15 +100,18 @@ typedef enum dpo_type_t_ {
      * @brief load-balancing over a choice of [un]equal cost paths
      */
     DPO_LOAD_BALANCE,
+    DPO_REPLICATE,
     DPO_ADJACENCY,
     DPO_ADJACENCY_INCOMPLETE,
     DPO_ADJACENCY_MIDCHAIN,
     DPO_ADJACENCY_GLEAN,
+    DPO_ADJACENCY_MCAST,
     DPO_RECEIVE,
     DPO_LOOKUP,
     DPO_LISP_CP,
     DPO_CLASSIFY,
     DPO_MPLS_LABEL,
+    DPO_MFIB_ENTRY,
     DPO_LAST,
 } __attribute__((packed)) dpo_type_t;
 
@@ -123,12 +126,15 @@ typedef enum dpo_type_t_ {
     [DPO_ADJACENCY_INCOMPLETE] = "dpo-adjacency-incomplete",	\
     [DPO_ADJACENCY_MIDCHAIN] = "dpo-adjacency-midcahin",	\
     [DPO_ADJACENCY_GLEAN] = "dpo-glean",	\
+    [DPO_ADJACENCY_MCAST] = "dpo-adj-mcast",	\
     [DPO_RECEIVE] = "dpo-receive",	\
     [DPO_LOOKUP] = "dpo-lookup",	\
     [DPO_LOAD_BALANCE] = "dpo-load-balance",	\
+    [DPO_REPLICATE] = "dpo-replicate",	\
     [DPO_LISP_CP] = "dpo-lisp-cp",	\
     [DPO_CLASSIFY] = "dpo-classify",	\
-    [DPO_MPLS_LABEL] = "dpo-mpls-label"	\
+    [DPO_MPLS_LABEL] = "dpo-mpls-label", \
+    [DPO_MFIB_ENTRY] = "dpo-mfib_entry"	\
 }
 
 /**
diff --git a/src/vnet/dpo/load_balance.c b/src/vnet/dpo/load_balance.c
index e70a7a30..f11b4e4d 100644
--- a/src/vnet/dpo/load_balance.c
+++ b/src/vnet/dpo/load_balance.c
@@ -238,6 +238,17 @@ load_balance_is_drop (const dpo_id_t *dpo)
     return (0);
 }
 
+void
+load_balance_set_fib_entry_flags (index_t lbi,
+                                  fib_entry_flag_t flags)
+{
+    load_balance_t *lb;
+
+    lb = load_balance_get(lbi);
+    lb->lb_fib_entry_flags = flags;
+}
+
+
 void
 load_balance_set_urpf (index_t lbi,
 		       index_t urpf)
@@ -683,7 +694,7 @@ load_balance_multipath_update (const dpo_id_t *dpo,
                                           buckets,
                                           n_buckets);
 
-                for (ii = old_n_buckets-n_buckets; ii < old_n_buckets; ii++)
+                for (ii = n_buckets; ii < old_n_buckets; ii++)
                 {
                     dpo_reset(&buckets[ii]);
                 }
diff --git a/src/vnet/dpo/load_balance.h b/src/vnet/dpo/load_balance.h
index 17996536..b901c5be 100644
--- a/src/vnet/dpo/load_balance.h
+++ b/src/vnet/dpo/load_balance.h
@@ -36,6 +36,7 @@
 #include <vnet/ip/lookup.h>
 #include <vnet/dpo/dpo.h>
 #include <vnet/fib/fib_types.h>
+#include <vnet/fib/fib_entry.h>
 
 /**
  * Load-balance main
@@ -98,6 +99,11 @@ typedef struct load_balance_t_ {
      */
     dpo_proto_t lb_proto;
 
+    /**
+     * Flags from the load-balance's associated fib_entry_t
+     */
+    fib_entry_flag_t lb_fib_entry_flags;
+
     /**
      * The number of locks, which is approximately the number of users,
      * of this load-balance.
@@ -167,6 +173,8 @@ extern void load_balance_set_bucket(index_t lbi,
 				    const dpo_id_t *next);
 extern void load_balance_set_urpf(index_t lbi,
 				  index_t urpf);
+extern void load_balance_set_fib_entry_flags(index_t lbi,
+                                             fib_entry_flag_t flags);
 extern index_t load_balance_get_urpf(index_t lbi);
 
 extern u8* format_load_balance(u8 * s, va_list * args);
diff --git a/src/vnet/dpo/replicate_dpo.c b/src/vnet/dpo/replicate_dpo.c
new file mode 100644
index 00000000..a2d5fdb6
--- /dev/null
+++ b/src/vnet/dpo/replicate_dpo.c
@@ -0,0 +1,759 @@
+/*
+ * Copyright (c) 2016 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <vnet/ip/lookup.h>
+#include <vnet/dpo/replicate_dpo.h>
+#include <vnet/dpo/drop_dpo.h>
+#include <vnet/adj/adj.h>
+
+#undef REP_DEBUG
+
+#ifdef REP_DEBUG
+#define REP_DBG(_rep, _fmt, _args...)                                   \
+{                                                                       \
+    u8* _tmp =NULL;                                                     \
+    clib_warning("rep:[%s]:" _fmt,                                      \
+                 replicate_format(replicate_get_index((_rep)),          \
+                                  0, _tmp),                             \
+                 ##_args);                                              \
+    vec_free(_tmp);                                                     \
+}
+#else
+#define REP_DBG(_p, _fmt, _args...)
+#endif
+
+
+/**
+ * Pool of all DPOs. It's not static so the DP can have fast access
+ */
+replicate_t *replicate_pool;
+
+/**
+ * The one instance of replicate main
+ */
+replicate_main_t replicate_main;
+
+static inline index_t
+replicate_get_index (const replicate_t *rep)
+{
+    return (rep - replicate_pool);
+}
+
+static inline dpo_id_t*
+replicate_get_buckets (replicate_t *rep)
+{
+    if (REP_HAS_INLINE_BUCKETS(rep))
+    {
+        return (rep->rep_buckets_inline);
+    }
+    else
+    {
+        return (rep->rep_buckets);
+    }
+}
+
+static replicate_t *
+replicate_alloc_i (void)
+{
+    replicate_t *rep;
+
+    pool_get_aligned(replicate_pool, rep, CLIB_CACHE_LINE_BYTES);
+    memset(rep, 0, sizeof(*rep));
+
+    vlib_validate_combined_counter(&(replicate_main.repm_counters),
+                                   replicate_get_index(rep));
+    vlib_zero_combined_counter(&(replicate_main.repm_counters),
+                               replicate_get_index(rep));
+
+    return (rep);
+}
+
+static u8*
+replicate_format (index_t repi,
+                  replicate_format_flags_t flags,
+                  u32 indent,
+                  u8 *s)
+{
+    vlib_counter_t to;
+    replicate_t *rep;
+    dpo_id_t *buckets;
+    u32 i;
+
+    rep = replicate_get(repi);
+    vlib_get_combined_counter(&(replicate_main.repm_counters), repi, &to);
+    buckets = replicate_get_buckets(rep);
+
+    s = format(s, "%U: ", format_dpo_type, DPO_REPLICATE);
+    s = format(s, "[index:%d buckets:%d ", repi, rep->rep_n_buckets);
+    s = format(s, "to:[%Ld:%Ld]]", to.packets, to.bytes);
+
+    for (i = 0; i < rep->rep_n_buckets; i++)
+    {
+        s = format(s, "\n%U", format_white_space, indent+2);
+        s = format(s, "[%d]", i);
+        s = format(s, " %U", format_dpo_id, &buckets[i], indent+6);
+    }
+    return (s);
+}
+
+u8*
+format_replicate (u8 * s, va_list * args)
+{
+    index_t repi = va_arg(*args, index_t);
+    replicate_format_flags_t flags = va_arg(*args, replicate_format_flags_t);
+
+    return (replicate_format(repi, flags, 0, s));
+}
+static u8*
+format_replicate_dpo (u8 * s, va_list * args)
+{
+    index_t repi = va_arg(*args, index_t);
+    u32 indent = va_arg(*args, u32);
+
+    return (replicate_format(repi, REPLICATE_FORMAT_DETAIL, indent, s));
+}
+
+
+static replicate_t *
+replicate_create_i (u32 num_buckets,
+                    dpo_proto_t rep_proto)
+{
+    replicate_t *rep;
+
+    rep = replicate_alloc_i();
+    rep->rep_n_buckets = num_buckets;
+    rep->rep_proto = rep_proto;
+
+    if (!REP_HAS_INLINE_BUCKETS(rep))
+    {
+        vec_validate_aligned(rep->rep_buckets,
+                             rep->rep_n_buckets - 1,
+                             CLIB_CACHE_LINE_BYTES);
+    }
+
+    REP_DBG(rep, "create");
+
+    return (rep);
+}
+
+index_t
+replicate_create (u32 n_buckets,
+                  dpo_proto_t rep_proto)
+{
+    return (replicate_get_index(replicate_create_i(n_buckets, rep_proto)));
+}
+
+static inline void
+replicate_set_bucket_i (replicate_t *rep,
+                        u32 bucket,
+                        dpo_id_t *buckets,
+                        const dpo_id_t *next)
+{
+    dpo_stack(DPO_REPLICATE, rep->rep_proto, &buckets[bucket], next);
+}
+
+void
+replicate_set_bucket (index_t repi,
+                      u32 bucket,
+                      const dpo_id_t *next)
+{
+    replicate_t *rep;
+    dpo_id_t *buckets;
+
+    rep = replicate_get(repi);
+    buckets = replicate_get_buckets(rep);
+
+    ASSERT(bucket < rep->rep_n_buckets);
+
+    replicate_set_bucket_i(rep, bucket, buckets, next);
+}
+
+int
+replicate_is_drop (const dpo_id_t *dpo)
+{
+    replicate_t *rep;
+
+    if (DPO_REPLICATE != dpo->dpoi_type)
+        return (0);
+
+    rep = replicate_get(dpo->dpoi_index);
+
+    if (1 == rep->rep_n_buckets)
+    {
+        return (dpo_is_drop(replicate_get_bucket_i(rep, 0)));
+    }
+    return (0);
+}
+
+const dpo_id_t *
+replicate_get_bucket (index_t repi,
+                      u32 bucket)
+{
+    replicate_t *rep;
+
+    rep = replicate_get(repi);
+
+    return (replicate_get_bucket_i(rep, bucket));
+}
+
+
+static load_balance_path_t *
+replicate_multipath_next_hop_fixup (load_balance_path_t *nhs,
+                                    dpo_proto_t drop_proto)
+{
+    if (0 == vec_len(nhs))
+    {
+        load_balance_path_t *nh;
+
+        /*
+         * we need something for the replicate. so use the drop
+         */
+        vec_add2(nhs, nh, 1);
+
+        nh->path_weight = 1;
+        dpo_copy(&nh->path_dpo, drop_dpo_get(drop_proto));
+    }
+
+    return (nhs);
+}
+
+/*
+ * Fill in adjacencies in block based on corresponding
+ * next hop adjacencies.
+ */
+static void
+replicate_fill_buckets (replicate_t *rep,
+                        load_balance_path_t *nhs,
+                        dpo_id_t *buckets,
+                        u32 n_buckets)
+{
+    load_balance_path_t * nh;
+    u16 ii, bucket;
+
+    bucket = 0;
+
+    /*
+     * the next-hops have normalised weights. that means their sum is the number
+     * of buckets we need to fill.
+     */
+    vec_foreach (nh, nhs)
+    {
+        for (ii = 0; ii < nh->path_weight; ii++)
+        {
+            ASSERT(bucket < n_buckets);
+            replicate_set_bucket_i(rep, bucket++, buckets, &nh->path_dpo);
+        }
+    }
+}
+
+static inline void
+replicate_set_n_buckets (replicate_t *rep,
+                         u32 n_buckets)
+{
+    rep->rep_n_buckets = n_buckets;
+}
+
+void
+replicate_multipath_update (const dpo_id_t *dpo,
+                            load_balance_path_t * next_hops)
+{
+    load_balance_path_t * nh, * nhs;
+    dpo_id_t *tmp_dpo;
+    u32 ii, n_buckets;
+    replicate_t *rep;
+
+    ASSERT(DPO_REPLICATE == dpo->dpoi_type);
+    rep = replicate_get(dpo->dpoi_index);
+    nhs = replicate_multipath_next_hop_fixup(next_hops,
+                                             rep->rep_proto);
+    n_buckets = vec_len(nhs);
+
+    if (0 == rep->rep_n_buckets)
+    {
+        /*
+         * first time initialisation. no packets inflight, so we can write
+         * at leisure.
+         */
+        replicate_set_n_buckets(rep, n_buckets);
+
+        if (!REP_HAS_INLINE_BUCKETS(rep))
+            vec_validate_aligned(rep->rep_buckets,
+                                 rep->rep_n_buckets - 1,
+                                 CLIB_CACHE_LINE_BYTES);
+
+        replicate_fill_buckets(rep, nhs,
+                               replicate_get_buckets(rep),
+                               n_buckets);
+    }
+    else
+    {
+        /*
+         * This is a modification of an existing replicate.
+         * We need to ensure that packets in flight see a consistent state, that
+         * is the number of reported buckets the REP has
+         * is not more than it actually has. So if the
+         * number of buckets is increasing, we must update the bucket array first,
+         * then the reported number. vice-versa if the number of buckets goes down.
+         */
+        if (n_buckets == rep->rep_n_buckets)
+        {
+            /*
+             * no change in the number of buckets. we can simply fill what
+             * is new over what is old.
+             */
+            replicate_fill_buckets(rep, nhs,
+                                   replicate_get_buckets(rep),
+                                   n_buckets);
+        }
+        else if (n_buckets > rep->rep_n_buckets)
+        {
+            /*
+             * we have more buckets. the old replicate map (if there is one)
+             * will remain valid, i.e. mapping to indices within range, so we
+             * update it last.
+             */
+            if (n_buckets > REP_NUM_INLINE_BUCKETS &&
+                rep->rep_n_buckets <= REP_NUM_INLINE_BUCKETS)
+            {
+                /*
+                 * the new increased number of buckets is crossing the threshold
+                 * from the inline storage to out-line. Alloc the outline buckets
+                 * first, then fixup the number. then reset the inlines.
+                 */
+                ASSERT(NULL == rep->rep_buckets);
+                vec_validate_aligned(rep->rep_buckets,
+                                     n_buckets - 1,
+                                     CLIB_CACHE_LINE_BYTES);
+
+                replicate_fill_buckets(rep, nhs,
+                                       rep->rep_buckets,
+                                       n_buckets);
+                CLIB_MEMORY_BARRIER();
+                replicate_set_n_buckets(rep, n_buckets);
+
+                CLIB_MEMORY_BARRIER();
+
+                for (ii = 0; ii < REP_NUM_INLINE_BUCKETS; ii++)
+                {
+                    dpo_reset(&rep->rep_buckets_inline[ii]);
+                }
+            }
+            else
+            {
+                if (n_buckets <= REP_NUM_INLINE_BUCKETS)
+                {
+                    /*
+                     * we are not crossing the threshold and it's still inline buckets.
+                     * we can write the new on the old..
+                     */
+                    replicate_fill_buckets(rep, nhs,
+                                           replicate_get_buckets(rep),
+                                           n_buckets);
+                    CLIB_MEMORY_BARRIER();
+                    replicate_set_n_buckets(rep, n_buckets);
+                }
+                else
+                {
+                    /*
+                     * we are not crossing the threshold. We need a new bucket array to
+                     * hold the increased number of choices.
+                     */
+                    dpo_id_t *new_buckets, *old_buckets, *tmp_dpo;
+
+                    new_buckets = NULL;
+                    old_buckets = replicate_get_buckets(rep);
+
+                    vec_validate_aligned(new_buckets,
+                                         n_buckets - 1,
+                                         CLIB_CACHE_LINE_BYTES);
+
+                    replicate_fill_buckets(rep, nhs, new_buckets, n_buckets);
+                    CLIB_MEMORY_BARRIER();
+                    rep->rep_buckets = new_buckets;
+                    CLIB_MEMORY_BARRIER();
+                    replicate_set_n_buckets(rep, n_buckets);
+
+                    vec_foreach(tmp_dpo, old_buckets)
+                    {
+                        dpo_reset(tmp_dpo);
+                    }
+                    vec_free(old_buckets);
+                }
+            }
+        }
+        else
+        {
+            /*
+             * bucket size shrinkage.
+             */
+            if (n_buckets <= REP_NUM_INLINE_BUCKETS &&
+                rep->rep_n_buckets > REP_NUM_INLINE_BUCKETS)
+            {
+                /*
+                 * the new decreased number of buckets is crossing the threshold
+                 * from out-line storage to inline:
+                 *   1 - Fill the inline buckets,
+                 *   2 - fixup the number (and this point the inline buckets are
+                 *       used).
+                 *   3 - free the outline buckets
+                 */
+                replicate_fill_buckets(rep, nhs,
+                                       rep->rep_buckets_inline,
+                                       n_buckets);
+                CLIB_MEMORY_BARRIER();
+                replicate_set_n_buckets(rep, n_buckets);
+                CLIB_MEMORY_BARRIER();
+
+                vec_foreach(tmp_dpo, rep->rep_buckets)
+                {
+                    dpo_reset(tmp_dpo);
+                }
+                vec_free(rep->rep_buckets);
+            }
+            else
+            {
+                /*
+                 * not crossing the threshold.
+                 *  1 - update the number to the smaller size
+                 *  2 - write the new buckets
+                 *  3 - reset those no longer used.
+                 */
+                dpo_id_t *buckets;
+                u32 old_n_buckets;
+
+                old_n_buckets = rep->rep_n_buckets;
+                buckets = replicate_get_buckets(rep);
+
+                replicate_set_n_buckets(rep, n_buckets);
+                CLIB_MEMORY_BARRIER();
+
+                replicate_fill_buckets(rep, nhs,
+                                       buckets,
+                                       n_buckets);
+
+                for (ii = n_buckets; ii < old_n_buckets; ii++)
+                {
+                    dpo_reset(&buckets[ii]);
+                }
+            }
+        }
+    }
+
+    vec_foreach (nh, nhs)
+    {
+        dpo_reset(&nh->path_dpo);
+    }
+    vec_free(nhs);
+}
+
+static void
+replicate_lock (dpo_id_t *dpo)
+{
+    replicate_t *rep;
+
+    rep = replicate_get(dpo->dpoi_index);
+
+    rep->rep_locks++;
+}
+
+static void
+replicate_destroy (replicate_t *rep)
+{
+    dpo_id_t *buckets;
+    int i;
+
+    buckets = replicate_get_buckets(rep);
+
+    for (i = 0; i < rep->rep_n_buckets; i++)
+    {
+        dpo_reset(&buckets[i]);
+    }
+
+    REP_DBG(rep, "destroy");
+    if (!REP_HAS_INLINE_BUCKETS(rep))
+    {
+        vec_free(rep->rep_buckets);
+    }
+
+    pool_put(replicate_pool, rep);
+}
+
+static void
+replicate_unlock (dpo_id_t *dpo)
+{
+    replicate_t *rep;
+
+    rep = replicate_get(dpo->dpoi_index);
+
+    rep->rep_locks--;
+
+    if (0 == rep->rep_locks)
+    {
+        replicate_destroy(rep);
+    }
+}
+
+static void
+replicate_mem_show (void)
+{
+    fib_show_memory_usage("replicate",
+			  pool_elts(replicate_pool),
+			  pool_len(replicate_pool),
+			  sizeof(replicate_t));
+}
+
+const static dpo_vft_t rep_vft = {
+    .dv_lock = replicate_lock,
+    .dv_unlock = replicate_unlock,
+    .dv_format = format_replicate_dpo,
+    .dv_mem_show = replicate_mem_show,
+};
+
+/**
+ * @brief The per-protocol VLIB graph nodes that are assigned to a replicate
+ *        object.
+ *
+ * this means that these graph nodes are ones from which a replicate is the
+ * parent object in the DPO-graph.
+ */
+const static char* const replicate_ip4_nodes[] =
+{
+    "ip4-replicate",
+    NULL,
+};
+const static char* const replicate_ip6_nodes[] =
+{
+    "ip6-replicate",
+    NULL,
+};
+const static char* const replicate_mpls_nodes[] =
+{
+    "mpls-replicate",
+    NULL,
+};
+
+const static char* const * const replicate_nodes[DPO_PROTO_NUM] =
+{
+    [DPO_PROTO_IP4]  = replicate_ip4_nodes,
+    [DPO_PROTO_IP6]  = replicate_ip6_nodes,
+    [DPO_PROTO_MPLS] = replicate_mpls_nodes,
+};
+
+void
+replicate_module_init (void)
+{
+    dpo_register(DPO_REPLICATE, &rep_vft, replicate_nodes);
+}
+
+static clib_error_t *
+replicate_show (vlib_main_t * vm,
+                unformat_input_t * input,
+                vlib_cli_command_t * cmd)
+{
+    index_t repi = INDEX_INVALID;
+
+    while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT)
+    {
+        if (unformat (input, "%d", &repi))
+            ;
+        else
+            break;
+    }
+
+    if (INDEX_INVALID != repi)
+    {
+        vlib_cli_output (vm, "%U", format_replicate, repi,
+                         REPLICATE_FORMAT_DETAIL);
+    }
+    else
+    {
+        replicate_t *rep;
+
+        pool_foreach(rep, replicate_pool,
+        ({
+            vlib_cli_output (vm, "%U", format_replicate,
+                             replicate_get_index(rep),
+                             REPLICATE_FORMAT_NONE);
+        }));
+    }
+
+    return 0;
+}
+
+VLIB_CLI_COMMAND (replicate_show_command, static) = {
+    .path = "show replicate",
+    .short_help = "show replicate [<index>]",
+    .function = replicate_show,
+};
+
+typedef struct replicate_trace_t_
+{
+    index_t rep_index;
+    index_t dpo_index;
+    dpo_type_t dpo_type;
+} replicate_trace_t;
+
+static uword
+replicate_inline (vlib_main_t * vm,
+                  vlib_node_runtime_t * node,
+                  vlib_frame_t * frame)
+{
+    vlib_combined_counter_main_t * cm = &replicate_main.repm_counters;
+    u32 n_left_from, * from, * to_next, next_index;
+    u32 cpu_index = os_get_cpu_number();
+
+    from = vlib_frame_vector_args (frame);
+    n_left_from = frame->n_vectors;
+    next_index = node->cached_next_index;
+  
+    while (n_left_from > 0)
+    {
+        u32 n_left_to_next;
+
+        vlib_get_next_frame (vm, node, next_index,
+                             to_next, n_left_to_next);
+
+        while (n_left_from > 0 && n_left_to_next > 0)
+	{
+            u32 next0, ci0, bi0, bucket, repi0;
+            const replicate_t *rep0;
+            vlib_buffer_t * b0, *c0;
+            const dpo_id_t *dpo0;
+
+            bi0 = from[0];
+            to_next[0] = bi0;
+            from += 1;
+            to_next += 1;
+            n_left_from -= 1;
+            n_left_to_next -= 1;
+
+            b0 = vlib_get_buffer (vm, bi0);
+            repi0 = vnet_buffer (b0)->ip.adj_index[VLIB_TX];
+            rep0 = replicate_get(repi0);
+
+            vlib_increment_combined_counter(
+                cm, cpu_index, repi0, 1,
+                vlib_buffer_length_in_chain(vm, b0));
+
+            /* ship the original to the first bucket */
+            dpo0 = replicate_get_bucket_i(rep0, 0);
+            next0 = dpo0->dpoi_next_node;
+            vnet_buffer (b0)->ip.adj_index[VLIB_TX] = dpo0->dpoi_index;
+
+            if (PREDICT_FALSE(b0->flags & VLIB_BUFFER_IS_TRACED))
+            {
+                replicate_trace_t *t = vlib_add_trace (vm, node, b0, sizeof (*t));
+                t->rep_index = repi0;
+                t->dpo_index = dpo0->dpoi_index;
+                t->dpo_type = dpo0->dpoi_type;
+            }
+            vlib_validate_buffer_enqueue_x1 (vm, node, next_index,
+                                             to_next, n_left_to_next,
+                                             bi0, next0);
+
+            /* ship copies to the rest of the buckets */
+            for (bucket = 1; bucket < rep0->rep_n_buckets; bucket++)
+            {
+                /* Make a copy */
+                c0 = vlib_buffer_copy(vm, b0);
+                ci0 = vlib_get_buffer_index(vm, c0);
+
+                to_next[0] = ci0;
+                to_next += 1;
+                n_left_to_next -= 1;
+
+                dpo0 = replicate_get_bucket_i(rep0, bucket);
+                next0 = dpo0->dpoi_next_node;
+                vnet_buffer (c0)->ip.adj_index[VLIB_TX] = dpo0->dpoi_index;
+
+                if (PREDICT_FALSE(b0->flags & VLIB_BUFFER_IS_TRACED))
+                {
+                    replicate_trace_t *t = vlib_add_trace (vm, node, b0, sizeof (*t));
+                    t->rep_index = repi0;
+                    t->dpo_index = dpo0->dpoi_index;
+                    t->dpo_type = dpo0->dpoi_type;
+                }
+
+                vlib_validate_buffer_enqueue_x1 (vm, node, next_index,
+                                                 to_next, n_left_to_next,
+                                                 ci0, next0);
+            }
+        }
+
+        vlib_put_next_frame (vm, node, next_index, n_left_to_next);
+    }
+
+    return frame->n_vectors;
+}
+
+static u8 *
+format_replicate_trace (u8 * s, va_list * args)
+{
+  CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *);
+  CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *);
+  replicate_trace_t *t = va_arg (*args, replicate_trace_t *);
+
+  s = format (s, "replicate: %d via %U:%d",
+              t->rep_index,
+              format_dpo_type, t->dpo_type,
+              t->dpo_index);
+  return s;
+}
+
+static uword
+ip4_replicate (vlib_main_t * vm,
+               vlib_node_runtime_t * node,
+               vlib_frame_t * frame)
+{
+    return (replicate_inline (vm, node, frame));
+}
+
+/**
+ * @brief
+ */
+VLIB_REGISTER_NODE (ip4_replicate_node) = {
+  .function = ip4_replicate,
+  .name = "ip4-replicate",
+  .vector_size = sizeof (u32),
+
+  .format_trace = format_replicate_trace,
+  .n_next_nodes = 1,
+  .next_nodes = {
+      [0] = "error-drop",
+  },
+};
+
+static uword
+ip6_replicate (vlib_main_t * vm,
+               vlib_node_runtime_t * node,
+               vlib_frame_t * frame)
+{
+    return (replicate_inline (vm, node, frame));
+}
+
+/**
+ * @brief
+ */
+VLIB_REGISTER_NODE (ip6_replicate_node) = {
+  .function = ip6_replicate,
+  .name = "ip6-replicate",
+  .vector_size = sizeof (u32),
+
+  .format_trace = format_replicate_trace,
+  .n_next_nodes = 1,
+  .next_nodes = {
+      [0] = "error-drop",
+  },
+};
diff --git a/src/vnet/dpo/replicate_dpo.h b/src/vnet/dpo/replicate_dpo.h
new file mode 100644
index 00000000..a564739c
--- /dev/null
+++ b/src/vnet/dpo/replicate_dpo.h
@@ -0,0 +1,143 @@
+/*
+ * Copyright (c) 2016 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+/**
+ * @brief
+ *
+ */
+
+#ifndef __REPLICATE_DPO_H__
+#define __REPLICATE_DPO_H__
+
+#include <vlib/vlib.h>
+#include <vnet/ip/lookup.h>
+#include <vnet/dpo/dpo.h>
+#include <vnet/dpo/load_balance.h>
+#include <vnet/fib/fib_types.h>
+
+/**
+ * replicate main
+ */
+typedef struct replicate_main_t_
+{
+    vlib_combined_counter_main_t repm_counters;
+} replicate_main_t;
+
+extern replicate_main_t replicate_main;
+
+/**
+ * The number of buckets that a load-balance object can have and still
+ * fit in one cache-line
+ */
+#define REP_NUM_INLINE_BUCKETS 4
+
+/**
+ * The FIB DPO provieds;
+ *  - load-balancing over the next DPOs in the chain/graph
+ *  - per-route counters
+ */
+typedef struct replicate_t_ {
+    /**
+     * number of buckets in the load-balance. always a power of 2.
+     */
+    u16 rep_n_buckets;
+
+   /**
+     * The protocol of packets that traverse this REP.
+     * need in combination with the flow hash config to determine how to hash.
+     * u8.
+     */
+    dpo_proto_t rep_proto;
+
+    /**
+     * The number of locks, which is approximately the number of users,
+     * of this load-balance.
+     * Load-balance objects of via-entries are heavily shared by recursives,
+     * so the lock count is a u32.
+     */
+    u32 rep_locks;
+
+    /**
+     * Vector of buckets containing the next DPOs, sized as repo_num
+     */
+    dpo_id_t *rep_buckets;
+
+    /**
+     * The rest of the cache line is used for buckets. In the common case
+     * where there there are less than 4 buckets, then the buckets are
+     * on the same cachlie and we save ourselves a pointer dereferance in 
+     * the data-path.
+     */
+    dpo_id_t rep_buckets_inline[REP_NUM_INLINE_BUCKETS];
+} replicate_t;
+
+STATIC_ASSERT(sizeof(replicate_t) <= CLIB_CACHE_LINE_BYTES,
+	      "A replicate object size exceeds one cachline");
+
+/**
+ * Flags controlling load-balance formatting/display
+ */
+typedef enum replicate_format_flags_t_ {
+    REPLICATE_FORMAT_NONE,
+    REPLICATE_FORMAT_DETAIL = (1 << 0),
+} replicate_format_flags_t;
+
+extern index_t replicate_create(u32 num_buckets,
+                                dpo_proto_t rep_proto);
+extern void replicate_multipath_update(
+    const dpo_id_t *dpo,
+    load_balance_path_t *next_hops);
+
+extern void replicate_set_bucket(index_t repi,
+				    u32 bucket,
+				    const dpo_id_t *next);
+
+extern u8* format_replicate(u8 * s, va_list * args);
+
+extern const dpo_id_t *replicate_get_bucket(index_t repi,
+					       u32 bucket);
+extern int replicate_is_drop(const dpo_id_t *dpo);
+
+/**
+ * The encapsulation breakages are for fast DP access
+ */
+extern replicate_t *replicate_pool;
+static inline replicate_t*
+replicate_get (index_t repi)
+{
+    return (pool_elt_at_index(replicate_pool, repi));
+}
+
+#define REP_HAS_INLINE_BUCKETS(_rep)		\
+    ((_rep)->rep_n_buckets <= REP_NUM_INLINE_BUCKETS)
+
+static inline const dpo_id_t *
+replicate_get_bucket_i (const replicate_t *rep,
+			   u32 bucket)
+{
+    ASSERT(bucket < rep->rep_n_buckets);
+
+    if (PREDICT_TRUE(REP_HAS_INLINE_BUCKETS(rep)))
+    {
+	return (&rep->rep_buckets_inline[bucket]);
+    }
+    else
+    {
+	return (&rep->rep_buckets[bucket]);
+    }
+}
+
+extern void replicate_module_init(void);
+
+#endif
diff --git a/src/vnet/ethernet/arp.c b/src/vnet/ethernet/arp.c
index 0298541b..8f07fa55 100644
--- a/src/vnet/ethernet/arp.c
+++ b/src/vnet/ethernet/arp.c
@@ -23,6 +23,7 @@
 #include <vppinfra/mhash.h>
 #include <vnet/fib/ip4_fib.h>
 #include <vnet/adj/adj_nbr.h>
+#include <vnet/adj/adj_mcast.h>
 #include <vnet/mpls/mpls.h>
 
 /**
@@ -438,33 +439,74 @@ arp_update_adjacency (vnet_main_t * vnm, u32 sw_if_index, u32 ai)
   arp_int = &am->ethernet_arp_by_sw_if_index[sw_if_index];
   e = arp_entry_find (arp_int, &adj->sub_type.nbr.next_hop.ip4);
 
-  if (NULL != e)
-    {
-      adj_nbr_walk_nh4 (sw_if_index,
-			&e->ip4_address, arp_mk_complete_walk, e);
-    }
-  else
+  switch (adj->lookup_next_index)
     {
+    case IP_LOOKUP_NEXT_ARP:
+    case IP_LOOKUP_NEXT_GLEAN:
+      if (NULL != e)
+	{
+	  adj_nbr_walk_nh4 (sw_if_index,
+			    &e->ip4_address, arp_mk_complete_walk, e);
+	}
+      else
+	{
+	  /*
+	   * no matching ARP entry.
+	   * construct the rewrite required to for an ARP packet, and stick
+	   * that in the adj's pipe to smoke.
+	   */
+	  adj_nbr_update_rewrite
+	    (ai,
+	     ADJ_NBR_REWRITE_FLAG_INCOMPLETE,
+	     ethernet_build_rewrite
+	     (vnm,
+	      sw_if_index,
+	      VNET_LINK_ARP,
+	      VNET_REWRITE_FOR_SW_INTERFACE_ADDRESS_BROADCAST));
+
+	  /*
+	   * since the FIB has added this adj for a route, it makes sense it
+	   * may want to forward traffic sometime soon. Let's send a
+	   * speculative ARP. just one. If we were to do periodically that
+	   * wouldn't be bad either, but that's more code than i'm prepared to
+	   * write at this time for relatively little reward.
+	   */
+	  arp_nbr_probe (adj);
+	}
+      break;
+    case IP_LOOKUP_NEXT_MCAST:
       /*
-       * no matching ARP entry.
-       * construct the rewire required to for an ARP packet, and stick
-       * that in the adj's pipe to smoke.
+       * Construct a partial rewrite from the known ethernet mcast dest MAC
        */
-      adj_nbr_update_rewrite (ai,
-			      ADJ_NBR_REWRITE_FLAG_INCOMPLETE,
-			      ethernet_build_rewrite (vnm,
-						      sw_if_index,
-						      VNET_LINK_ARP,
-						      VNET_REWRITE_FOR_SW_INTERFACE_ADDRESS_BROADCAST));
+      adj_mcast_update_rewrite
+	(ai,
+	 ethernet_build_rewrite (vnm,
+				 sw_if_index,
+				 adj->ia_link,
+				 ethernet_ip4_mcast_dst_addr ()));
 
       /*
-       * since the FIB has added this adj for a route, it makes sense it may
-       * want to forward traffic sometime soon. Let's send a speculative ARP.
-       * just one. If we were to do periodically that wouldn't be bad either,
-       * but that's more code than i'm prepared to write at this time for
-       * relatively little reward.
+       * Complete the remaining fields of the adj's rewrite to direct the
+       * complete of the rewrite at switch time by copying in the IP
+       * dst address's bytes.
+       * Ofset is 11 bytes from the end of the MAC header - which is three
+       * bytes into the desintation address. And we write 3 bytes.
        */
-      arp_nbr_probe (adj);
+      adj->rewrite_header.dst_mcast_offset = 11;
+      adj->rewrite_header.dst_mcast_n_bytes = 3;
+
+      break;
+
+    case IP_LOOKUP_NEXT_DROP:
+    case IP_LOOKUP_NEXT_PUNT:
+    case IP_LOOKUP_NEXT_LOCAL:
+    case IP_LOOKUP_NEXT_REWRITE:
+    case IP_LOOKUP_NEXT_LOAD_BALANCE:
+    case IP_LOOKUP_NEXT_MIDCHAIN:
+    case IP_LOOKUP_NEXT_ICMP_ERROR:
+    case IP_LOOKUP_N_NEXT:
+      ASSERT (0);
+      break;
     }
 }
 
diff --git a/src/vnet/ethernet/ethernet.h b/src/vnet/ethernet/ethernet.h
index f88b0cf3..3acde421 100644
--- a/src/vnet/ethernet/ethernet.h
+++ b/src/vnet/ethernet/ethernet.h
@@ -547,6 +547,8 @@ void ethernet_update_adjacency (vnet_main_t * vnm, u32 sw_if_index, u32 ai);
 u8 *ethernet_build_rewrite (vnet_main_t * vnm,
 			    u32 sw_if_index,
 			    vnet_link_t link_type, const void *dst_address);
+const u8 *ethernet_ip4_mcast_dst_addr (void);
+const u8 *ethernet_ip6_mcast_dst_addr (void);
 
 extern vlib_node_registration_t ethernet_input_node;
 
diff --git a/src/vnet/ethernet/interface.c b/src/vnet/ethernet/interface.c
index 1c1f4353..95700309 100644
--- a/src/vnet/ethernet/interface.c
+++ b/src/vnet/ethernet/interface.c
@@ -51,6 +51,26 @@
  * This file contains code to manage loopback interfaces.
  */
 
+const u8 *
+ethernet_ip4_mcast_dst_addr (void)
+{
+  const static u8 ethernet_mcast_dst_mac[] = {
+    0x1, 0x0, 0x5e, 0x0, 0x0, 0x0,
+  };
+
+  return (ethernet_mcast_dst_mac);
+}
+
+const u8 *
+ethernet_ip6_mcast_dst_addr (void)
+{
+  const static u8 ethernet_mcast_dst_mac[] = {
+    0x33, 0x33, 0x00, 0x0, 0x0, 0x0,
+  };
+
+  return (ethernet_mcast_dst_mac);
+}
+
 /**
  * @brief build a rewrite string to use for sending packets of type 'link_type'
  * to 'dst_address'
diff --git a/src/vnet/fib/fib_attached_export.c b/src/vnet/fib/fib_attached_export.c
index fd597e3e..574131de 100644
--- a/src/vnet/fib/fib_attached_export.c
+++ b/src/vnet/fib/fib_attached_export.c
@@ -303,8 +303,8 @@ fib_attached_export_import (fib_entry_t *fib_entry,
      * may have realloc'd.
      */
     fib_entry = fib_entry_get(fei);
-        import->faei_export_sibling =
-	  fib_entry_cover_track(fib_entry_get(import->faei_export_entry), fei);
+    import->faei_export_sibling =
+	fib_entry_cover_track(fib_entry_get(import->faei_export_entry), fei);
 
     fed = fib_entry_delegate_find_or_add(fib_entry,
                                          FIB_ENTRY_DELEGATE_ATTACHED_IMPORT);
diff --git a/src/vnet/fib/fib_entry.h b/src/vnet/fib/fib_entry.h
index 44a5f2e6..f258b755 100644
--- a/src/vnet/fib/fib_entry.h
+++ b/src/vnet/fib/fib_entry.h
@@ -220,7 +220,7 @@ typedef enum fib_entry_flag_t_ {
     FIB_ENTRY_FLAG_EXCLUSIVE = (1 << FIB_ENTRY_ATTRIBUTE_EXCLUSIVE),
     FIB_ENTRY_FLAG_LOCAL     = (1 << FIB_ENTRY_ATTRIBUTE_LOCAL),
     FIB_ENTRY_FLAG_IMPORT    = (1 << FIB_ENTRY_ATTRIBUTE_IMPORT),
-} fib_entry_flag_t;
+} __attribute__((packed)) fib_entry_flag_t;
 
 /**
  * Flags for the source data
diff --git a/src/vnet/fib/fib_entry_delegate.c b/src/vnet/fib/fib_entry_delegate.c
index a0d45f97..efe402d1 100644
--- a/src/vnet/fib/fib_entry_delegate.c
+++ b/src/vnet/fib/fib_entry_delegate.c
@@ -119,6 +119,9 @@ fib_entry_chain_type_to_delegate_type (fib_forward_chain_type_t fct)
         return (FIB_ENTRY_DELEGATE_CHAIN_MPLS_NON_EOS);
     case FIB_FORW_CHAIN_TYPE_ETHERNET:
         return (FIB_ENTRY_DELEGATE_CHAIN_ETHERNET);
+    case FIB_FORW_CHAIN_TYPE_MCAST_IP4:
+    case FIB_FORW_CHAIN_TYPE_MCAST_IP6:
+        break;
     }
     ASSERT(0);
     return (FIB_ENTRY_DELEGATE_CHAIN_UNICAST_IP4);
diff --git a/src/vnet/fib/fib_entry_src.c b/src/vnet/fib/fib_entry_src.c
index 1fb04060..d54787cd 100644
--- a/src/vnet/fib/fib_entry_src.c
+++ b/src/vnet/fib/fib_entry_src.c
@@ -313,6 +313,8 @@ fib_entry_src_collect_forwarding (fib_node_index_t pl_index,
         {
         case FIB_FORW_CHAIN_TYPE_UNICAST_IP4:
         case FIB_FORW_CHAIN_TYPE_UNICAST_IP6:
+        case FIB_FORW_CHAIN_TYPE_MCAST_IP4:
+        case FIB_FORW_CHAIN_TYPE_MCAST_IP6:
             /*
              * EOS traffic with no label to stack, we need the IP Adj
              */
@@ -458,6 +460,8 @@ fib_entry_src_mk_lb (fib_entry_t *fib_entry,
     {
 	load_balance_set_urpf(dpo_lb->dpoi_index, ui);
     }
+    load_balance_set_fib_entry_flags(dpo_lb->dpoi_index,
+                                     fib_entry_get_flags_i(fib_entry));
 }
 
 void
diff --git a/src/vnet/fib/fib_node.h b/src/vnet/fib/fib_node.h
index 3ad8ee95..457dfb7a 100644
--- a/src/vnet/fib/fib_node.h
+++ b/src/vnet/fib/fib_node.h
@@ -31,6 +31,7 @@ typedef enum fib_node_type_t_ {
      */
     FIB_NODE_TYPE_WALK,
     FIB_NODE_TYPE_ENTRY,
+    FIB_NODE_TYPE_MFIB_ENTRY,
     FIB_NODE_TYPE_PATH_LIST,
     FIB_NODE_TYPE_PATH,
     FIB_NODE_TYPE_ADJ,
@@ -51,6 +52,7 @@ typedef enum fib_node_type_t_ {
 
 #define FIB_NODE_TYPES {                          \
     [FIB_NODE_TYPE_ENTRY]     = "entry",          \
+    [FIB_NODE_TYPE_MFIB_ENTRY] = "mfib-entry",    \
     [FIB_NODE_TYPE_WALK]      = "walk",           \
     [FIB_NODE_TYPE_PATH_LIST] = "path-list",      \
     [FIB_NODE_TYPE_PATH]      = "path",           \
diff --git a/src/vnet/fib/fib_path.c b/src/vnet/fib/fib_path.c
index 809e3e16..080057f3 100644
--- a/src/vnet/fib/fib_path.c
+++ b/src/vnet/fib/fib_path.c
@@ -23,6 +23,7 @@
 #include <vnet/dpo/lookup_dpo.h>
 
 #include <vnet/adj/adj.h>
+#include <vnet/adj/adj_mcast.h>
 
 #include <vnet/fib/fib_path.h>
 #include <vnet/fib/fib_node.h>
@@ -960,6 +961,8 @@ fib_path_route_flags_to_cfg_flags (const fib_route_path_t *rpath)
 	cfg_flags |= FIB_PATH_CFG_FLAG_RESOLVE_HOST;
     if (rpath->frp_flags & FIB_ROUTE_PATH_RESOLVE_VIA_ATTACHED)
 	cfg_flags |= FIB_PATH_CFG_FLAG_RESOLVE_ATTACHED;
+    if (rpath->frp_flags & FIB_ROUTE_PATH_LOCAL)
+	cfg_flags |= FIB_PATH_CFG_FLAG_LOCAL;
 
     return (cfg_flags);
 }
@@ -1003,28 +1006,25 @@ fib_path_create (fib_node_index_t pl_index,
     /*
      * deduce the path's tpye from the parementers and save what is needed.
      */
-    if (~0 != rpath->frp_sw_if_index)
+    if (path->fp_cfg_flags & FIB_PATH_CFG_FLAG_LOCAL)
     {
-	if (flags & FIB_PATH_CFG_FLAG_LOCAL)
-	{
-	    path->fp_type = FIB_PATH_TYPE_RECEIVE;
-	    path->receive.fp_interface = rpath->frp_sw_if_index;
-            path->receive.fp_addr = rpath->frp_addr;
-	}
-	else
-	{
-	    if (ip46_address_is_zero(&rpath->frp_addr))
-	    {
-		path->fp_type = FIB_PATH_TYPE_ATTACHED;
-		path->attached.fp_interface = rpath->frp_sw_if_index;
-	    }
-	    else
-	    {
-		path->fp_type = FIB_PATH_TYPE_ATTACHED_NEXT_HOP;
-		path->attached_next_hop.fp_interface = rpath->frp_sw_if_index;
-		path->attached_next_hop.fp_nh = rpath->frp_addr;
-	    }
-	}
+        path->fp_type = FIB_PATH_TYPE_RECEIVE;
+        path->receive.fp_interface = rpath->frp_sw_if_index;
+        path->receive.fp_addr = rpath->frp_addr;
+    }
+    else if (~0 != rpath->frp_sw_if_index)
+    {
+        if (ip46_address_is_zero(&rpath->frp_addr))
+        {
+            path->fp_type = FIB_PATH_TYPE_ATTACHED;
+            path->attached.fp_interface = rpath->frp_sw_if_index;
+        }
+        else
+        {
+            path->fp_type = FIB_PATH_TYPE_ATTACHED_NEXT_HOP;
+            path->attached_next_hop.fp_interface = rpath->frp_sw_if_index;
+            path->attached_next_hop.fp_nh = rpath->frp_addr;
+        }
     }
     else
     {
@@ -1199,7 +1199,7 @@ fib_path_cmp_i (const fib_path_t *path1,
     {
 	res = (path1->fp_type - path2->fp_type);
     }
-    if (path1->fp_nh_proto != path2->fp_nh_proto)
+    else if (path1->fp_nh_proto != path2->fp_nh_proto)
     {
 	res = (path1->fp_nh_proto - path2->fp_nh_proto);
     }
@@ -1770,8 +1770,11 @@ fib_path_contribute_forwarding (fib_node_index_t path_index,
 
 		break;
 	    }
-	    }
+	    case FIB_FORW_CHAIN_TYPE_MCAST_IP4:
+	    case FIB_FORW_CHAIN_TYPE_MCAST_IP6:
 	    break;
+            }
+            break;
 	case FIB_PATH_TYPE_RECURSIVE:
 	    switch (fct)
 	    {
@@ -1781,13 +1784,15 @@ fib_path_contribute_forwarding (fib_node_index_t path_index,
 	    case FIB_FORW_CHAIN_TYPE_MPLS_NON_EOS:
 		fib_path_recursive_adj_update(path, fct, dpo);
 		break;
+	    case FIB_FORW_CHAIN_TYPE_MCAST_IP4:
+	    case FIB_FORW_CHAIN_TYPE_MCAST_IP6:
 	    case FIB_FORW_CHAIN_TYPE_ETHERNET:
 		ASSERT(0);
 		break;
 	    }
 	    break;
 	case FIB_PATH_TYPE_DEAG:
-	    switch (fct)
+            switch (fct)
 	    {
 	    case FIB_FORW_CHAIN_TYPE_MPLS_NON_EOS:
                 lookup_dpo_add_or_lock_w_table_id(MPLS_FIB_DEFAULT_TABLE_ID,
@@ -1800,7 +1805,9 @@ fib_path_contribute_forwarding (fib_node_index_t path_index,
 	    case FIB_FORW_CHAIN_TYPE_UNICAST_IP6:
 	    case FIB_FORW_CHAIN_TYPE_MPLS_EOS:
 		dpo_copy(dpo, &path->fp_dpo);
-		break;		
+		break;
+	    case FIB_FORW_CHAIN_TYPE_MCAST_IP4:
+	    case FIB_FORW_CHAIN_TYPE_MCAST_IP6:
 	    case FIB_FORW_CHAIN_TYPE_ETHERNET:
 		ASSERT(0);
 		break;
@@ -1810,12 +1817,38 @@ fib_path_contribute_forwarding (fib_node_index_t path_index,
 	    dpo_copy(dpo, &path->exclusive.fp_ex_dpo);
 	    break;
         case FIB_PATH_TYPE_ATTACHED:
-	case FIB_PATH_TYPE_RECEIVE:
-	case FIB_PATH_TYPE_SPECIAL:
-	    ASSERT(0);
+	    switch (fct)
+	    {
+	    case FIB_FORW_CHAIN_TYPE_MPLS_NON_EOS:
+	    case FIB_FORW_CHAIN_TYPE_UNICAST_IP4:
+	    case FIB_FORW_CHAIN_TYPE_UNICAST_IP6:
+	    case FIB_FORW_CHAIN_TYPE_MPLS_EOS:
+	    case FIB_FORW_CHAIN_TYPE_ETHERNET:
+                break;
+	    case FIB_FORW_CHAIN_TYPE_MCAST_IP4:
+	    case FIB_FORW_CHAIN_TYPE_MCAST_IP6:
+                {
+                    adj_index_t ai;
+
+                    /*
+                     * Create the adj needed for sending IP multicast traffic
+                     */
+                    ai = adj_mcast_add_or_lock(path->fp_nh_proto,
+                                               fib_forw_chain_type_to_link_type(fct),
+                                               path->attached.fp_interface);
+                    dpo_set(dpo, DPO_ADJACENCY_MCAST,
+                            fib_forw_chain_type_to_dpo_proto(fct),
+                            ai);
+                    adj_unlock(ai);
+                }
+                break;
+            }
+            break;
+        case FIB_PATH_TYPE_RECEIVE:
+        case FIB_PATH_TYPE_SPECIAL:
+            dpo_copy(dpo, &path->fp_dpo);
             break;
 	}
-
     }
 }
 
diff --git a/src/vnet/fib/fib_path_list.c b/src/vnet/fib/fib_path_list.c
index db9d1af9..ce11cf45 100644
--- a/src/vnet/fib/fib_path_list.c
+++ b/src/vnet/fib/fib_path_list.c
@@ -585,8 +585,11 @@ fib_path_list_resolve (fib_path_list_t *path_list)
     path_list = fib_path_list_get(path_list_index);
 
     FIB_PATH_LIST_DBG(path_list, "resovled");
-    fib_path_list_mk_urpf(path_list);
 
+    if (!(path_list->fpl_flags & FIB_PATH_LIST_FLAG_NO_URPF))
+    {
+        fib_path_list_mk_urpf(path_list);
+    }
     return (path_list);
 }
 
@@ -1025,14 +1028,14 @@ fib_path_list_copy_and_path_remove (fib_node_index_t orig_path_list_index,
  */
 void
 fib_path_list_contribute_forwarding (fib_node_index_t path_list_index,
-				     fib_forward_chain_type_t type,
+				     fib_forward_chain_type_t fct,
 				     dpo_id_t *dpo)
 {
     fib_path_list_t *path_list;
 
     path_list = fib_path_list_get(path_list_index);
 
-    fib_path_list_mk_lb(path_list, type, dpo);
+    fib_path_list_mk_lb(path_list, fct, dpo);
 }
 
 /*
diff --git a/src/vnet/fib/fib_path_list.h b/src/vnet/fib/fib_path_list.h
index f4f94a1b..b4971add 100644
--- a/src/vnet/fib/fib_path_list.h
+++ b/src/vnet/fib/fib_path_list.h
@@ -60,6 +60,10 @@ typedef enum fib_path_list_attribute_t_ {
      * looped path-list. one path looped implies the whole list is
      */
     FIB_PATH_LIST_ATTRIBUTE_LOOPED,
+    /**
+     * no uRPF - do not generate unicast RPF list for this path-list
+     */
+    FIB_PATH_LIST_ATTRIBUTE_NO_URPF,
     /**
      * Marher. Add new flags before this one, and then update it.
      */
@@ -74,6 +78,7 @@ typedef enum fib_path_list_flags_t_ {
     FIB_PATH_LIST_FLAG_EXCLUSIVE = (1 << FIB_PATH_LIST_ATTRIBUTE_EXCLUSIVE),
     FIB_PATH_LIST_FLAG_RESOLVED  = (1 << FIB_PATH_LIST_ATTRIBUTE_RESOLVED),
     FIB_PATH_LIST_FLAG_LOOPED    = (1 << FIB_PATH_LIST_ATTRIBUTE_LOOPED),
+    FIB_PATH_LIST_FLAG_NO_URPF   = (1 << FIB_PATH_LIST_ATTRIBUTE_NO_URPF),
 } fib_path_list_flags_t;
 
 #define FIB_PATH_LIST_ATTRIBUTES {       		 \
@@ -83,6 +88,7 @@ typedef enum fib_path_list_flags_t_ {
     [FIB_PATH_LIST_ATTRIBUTE_EXCLUSIVE] = "exclusive",   \
     [FIB_PATH_LIST_ATTRIBUTE_LOCAL]     = "local",	 \
     [FIB_PATH_LIST_ATTRIBUTE_LOOPED]     = "looped",	 \
+    [FIB_PATH_LIST_ATTRIBUTE_NO_URPF]     = "no-uRPF",	 \
 }
 
 #define FOR_EACH_PATH_LIST_ATTRIBUTE(_item)		\
diff --git a/src/vnet/fib/fib_table.c b/src/vnet/fib/fib_table.c
index 76db42d0..57eb0ae8 100644
--- a/src/vnet/fib/fib_table.c
+++ b/src/vnet/fib/fib_table.c
@@ -1043,6 +1043,26 @@ fib_table_destroy (fib_table_t *fib_table)
 	break;
     }
 }
+void
+fib_table_walk (u32 fib_index,
+                fib_protocol_t proto,
+                fib_table_walk_fn_t fn,
+                void *ctx)
+{
+    switch (proto)
+    {
+    case FIB_PROTOCOL_IP4:
+	ip4_fib_table_walk(ip4_fib_get(fib_index), fn, ctx);
+	break;
+    case FIB_PROTOCOL_IP6:
+	ip6_fib_table_walk(fib_index, fn, ctx);
+	break;
+    case FIB_PROTOCOL_MPLS:
+	mpls_fib_table_walk(mpls_fib_get(fib_index), fn, ctx);
+	break;
+    }
+}
+
 
 void
 fib_table_unlock (u32 fib_index,
@@ -1094,11 +1114,56 @@ format_fib_table_name (u8* s, va_list ap)
     return (s);
 }
 
+/**
+ * @brief Table flush context. Store the indicies of matching FIB entries
+ * that need to be removed.
+ */
+typedef struct fib_table_flush_ctx_t_
+{
+    /**
+     * The list of entries to flush
+     */
+    fib_node_index_t *ftf_entries;
+
+    /**
+     * The source we are flushing
+     */
+    fib_source_t ftf_source;
+} fib_table_flush_ctx_t;
+
+static int
+fib_table_flush_cb (fib_node_index_t fib_entry_index,
+                    void *arg)
+{
+    fib_table_flush_ctx_t *ctx = arg;
+
+    if (fib_entry_is_sourced(fib_entry_index, ctx->ftf_source))
+    {
+        vec_add1(ctx->ftf_entries, fib_entry_index);
+    }
+    return (1);
+}
+
+
 void
 fib_table_flush (u32 fib_index,
 		 fib_protocol_t proto,
 		 fib_source_t source)
 {
-    // FIXME
-    ASSERT(0);
+    fib_node_index_t *fib_entry_index;
+    fib_table_flush_ctx_t ctx = {
+        .ftf_entries = NULL,
+        .ftf_source = source,
+    };
+
+    fib_table_walk(fib_index, proto,
+                   fib_table_flush_cb,
+                   &ctx);
+
+    vec_foreach(fib_entry_index, ctx.ftf_entries)
+    {
+        fib_entry_delete(*fib_entry_index, source);
+    }
+
+    vec_free(ctx.ftf_entries);
 }
diff --git a/src/vnet/fib/fib_table.h b/src/vnet/fib/fib_table.h
index cfec516d..e7e66acb 100644
--- a/src/vnet/fib/fib_table.h
+++ b/src/vnet/fib/fib_table.h
@@ -729,4 +729,20 @@ extern u32 fib_table_get_num_entries(u32 fib_index,
 extern fib_table_t *fib_table_get(fib_node_index_t index,
 				  fib_protocol_t proto);
 
+/**
+ * @brief Call back function when walking entries in a FIB table
+ */
+typedef int (*fib_table_walk_fn_t)(fib_node_index_t fei,
+                                   void *ctx);
+
+/**
+ * @brief Walk all entries in a FIB table
+ * N.B: This is NOT safe to deletes. If you need to delete walk the whole
+ * table and store elements in a vector, then delete the elements
+ */
+extern void fib_table_walk(u32 fib_index,
+                           fib_protocol_t proto,
+                           fib_table_walk_fn_t fn,
+                           void *ctx);
+
 #endif
diff --git a/src/vnet/fib/fib_test.c b/src/vnet/fib/fib_test.c
index 5083db26..1c4a63a2 100644
--- a/src/vnet/fib/fib_test.c
+++ b/src/vnet/fib/fib_test.c
@@ -663,14 +663,15 @@ fib_test_v4 (void)
     /*
      * at this stage there are 5 entries in the test FIB (plus 5 in the default),
      * all of which are special sourced and so none of which share path-lists.
-     * There are also 6 entries, and 6 non-shared path-lists, in the v6 default
-     * table
+     * There are also 2 entries, and 2 non-shared path-lists, in the v6 default
+     * table, and 4 path-lists in the v6 MFIB table
      */
-#define NBR (5+5+6)
+#define ENBR (5+5+2)
+#define PNBR (5+5+6)
     FIB_TEST((0 == fib_path_list_db_size()),   "path list DB is empty");
-    FIB_TEST((NBR == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR == fib_path_list_pool_size()), "path list pool size is %d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -754,9 +755,9 @@ fib_test_v4 (void)
      * +2 interface routes +2 non-shared path-lists
      */
     FIB_TEST((0 == fib_path_list_db_size()),   "path list DB is empty");
-    FIB_TEST((NBR+2 == fib_path_list_pool_size()), "path list pool size is%d",
+    FIB_TEST((PNBR+2 == fib_path_list_pool_size()), "path list pool size is%d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+2 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+2 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -808,9 +809,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((1 == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NBR+3 == fib_path_list_pool_size()), "path list pool size is%d",
+    FIB_TEST((PNBR+3 == fib_path_list_pool_size()), "path list pool size is%d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+2 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+2 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -838,9 +839,9 @@ fib_test_v4 (void)
      * -1 shared-path-list
      */
     FIB_TEST((0 == fib_path_list_db_size()),   "path list DB is empty");
-    FIB_TEST((NBR+2 == fib_path_list_pool_size()), "path list pool size is%d",
+    FIB_TEST((PNBR+2 == fib_path_list_pool_size()), "path list pool size is%d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+2 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+2 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -1018,9 +1019,9 @@ fib_test_v4 (void)
      * +2 adj-fibs, and their non-shared path-lists
      */
     FIB_TEST((0 == fib_path_list_db_size()),   "path list DB is empty");
-    FIB_TEST((NBR+4 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+4 == fib_path_list_pool_size()), "path list pool size is %d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+4 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+4 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -1054,9 +1055,9 @@ fib_test_v4 (void)
      * +1 entry and a shared path-list
      */
     FIB_TEST((1 == fib_path_list_db_size()),   "path list DB is empty");
-    FIB_TEST((NBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+5 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+5 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /* 1.1.2.0/24 */
@@ -1087,9 +1088,9 @@ fib_test_v4 (void)
      * +1 entry only
      */
     FIB_TEST((1 == fib_path_list_db_size()),   "path list DB is empty");
-    FIB_TEST((NBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+6 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+6 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -1127,9 +1128,9 @@ fib_test_v4 (void)
      * +1 shared-pathlist
      */
     FIB_TEST((2 == fib_path_list_db_size()),   "path list DB is empty");
-    FIB_TEST((NBR+6 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+6 == fib_path_list_pool_size()), "path list pool size is %d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+6 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+6 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -1158,9 +1159,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((1 == fib_path_list_db_size()),   "path list DB is %d",
 	     fib_path_list_db_size());
-    FIB_TEST((NBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+6 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+6 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -1203,9 +1204,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((2  == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NBR+6 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+6 == fib_path_list_pool_size()), "path list pool size is %d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+7 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+7 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     fib_prefix_t bgp_101_pfx = {
@@ -1239,9 +1240,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((2  == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NBR+6 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+6 == fib_path_list_pool_size()), "path list pool size is %d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+8 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+8 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -1368,9 +1369,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((3  == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NBR+7 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+7 == fib_path_list_pool_size()), "path list pool size is %d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+10 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+10 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -1983,9 +1984,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((4  == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NBR+8 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+8 == fib_path_list_pool_size()), "path list pool size is %d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+12 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+12 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -2030,9 +2031,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((4  == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NBR+8 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+8 == fib_path_list_pool_size()), "path list pool size is %d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+13 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+13 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -2080,9 +2081,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((5  == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NBR+9 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+9 == fib_path_list_pool_size()), "path list pool size is %d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+14 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+14 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -2118,9 +2119,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((4  == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NBR+8 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+8 == fib_path_list_pool_size()), "path list pool size is %d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+13 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+13 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -2154,9 +2155,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((4  == fib_path_list_db_size()),   "path list DB population:%d",
 	fib_path_list_db_size());
-    FIB_TEST((NBR+8 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+8 == fib_path_list_pool_size()), "path list pool size is %d",
 	fib_path_list_pool_size());
-    FIB_TEST((NBR+12 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+12 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -2185,9 +2186,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((4  == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NBR+8 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+8 == fib_path_list_pool_size()), "path list pool size is %d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+12 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+12 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -2215,9 +2216,9 @@ fib_test_v4 (void)
 
     FIB_TEST((3  == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NBR+7 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+7 == fib_path_list_pool_size()), "path list pool size is %d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+10 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+10 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -2245,9 +2246,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((2  == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NBR+6 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+6 == fib_path_list_pool_size()), "path list pool size is %d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+9 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+9 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -2355,9 +2356,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((1  == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+7 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+7 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -2380,9 +2381,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((2  == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NBR+6 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+6 == fib_path_list_pool_size()), "path list pool size is %d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+8 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+8 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     fei = fib_table_lookup_exact_match(fib_index, &bgp_200_pfx);
@@ -2428,9 +2429,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((3  == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NBR+8 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+8 == fib_path_list_pool_size()), "path list pool size is %d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NBR+10 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+10 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     ai_03 = adj_nbr_add_or_lock(FIB_PROTOCOL_IP4,
@@ -2492,9 +2493,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((1  == fib_path_list_db_size()),   "path list DB population:%d",
     	     fib_path_list_db_size());
-    FIB_TEST((NBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
     	     fib_path_list_pool_size());
-    FIB_TEST((NBR+7 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+7 == fib_entry_pool_size()), "entry pool size is %d",
     	     fib_entry_pool_size());
 
 
@@ -2562,9 +2563,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((4  == fib_path_list_db_size()),   "path list DB population:%d",
     	     fib_path_list_db_size());
-    FIB_TEST((NBR+8 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+8 == fib_path_list_pool_size()), "path list pool size is %d",
     	     fib_path_list_pool_size());
-    FIB_TEST((NBR+10 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+10 == fib_entry_pool_size()), "entry pool size is %d",
     	     fib_entry_pool_size());
 
     /*
@@ -2753,9 +2754,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((1  == fib_path_list_db_size()),   "path list DB population:%d",
     	     fib_path_list_db_size());
-    FIB_TEST((NBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
     	     fib_path_list_pool_size());
-    FIB_TEST((NBR+7 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+7 == fib_entry_pool_size()), "entry pool size is %d",
     	     fib_entry_pool_size());
 
     /*
@@ -2830,9 +2831,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((1  == fib_path_list_db_size()),   "path list DB population:%d",
     	     fib_path_list_db_size());
-    FIB_TEST((NBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
     	     fib_path_list_pool_size());
-    FIB_TEST((NBR+7 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+7 == fib_entry_pool_size()), "entry pool size is %d",
     	     fib_entry_pool_size());
 
     /*
@@ -3180,9 +3181,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((1  == fib_path_list_db_size()),   "path list DB population:%d",
     	     fib_path_list_db_size());
-    FIB_TEST((NBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
     	     fib_path_list_pool_size());
-    FIB_TEST((NBR+7 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+7 == fib_entry_pool_size()), "entry pool size is %d",
     	     fib_entry_pool_size());
 
     /*
@@ -3247,9 +3248,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((1  == fib_path_list_db_size()),   "path list DB population:%d",
     	     fib_path_list_db_size());
-    FIB_TEST((NBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
     	     fib_path_list_pool_size());
-    FIB_TEST((NBR+7 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+7 == fib_entry_pool_size()), "entry pool size is %d",
     	     fib_entry_pool_size());
 
     /*
@@ -3297,9 +3298,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((1  == fib_path_list_db_size()),   "path list DB population:%d",
     	     fib_path_list_db_size());
-    FIB_TEST((NBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
     	     fib_path_list_pool_size());
-    FIB_TEST((NBR+7 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+7 == fib_entry_pool_size()), "entry pool size is %d",
     	     fib_entry_pool_size());
 
     /*
@@ -3342,9 +3343,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((1  == fib_path_list_db_size()),   "path list DB population:%d",
     	     fib_path_list_db_size());
-    FIB_TEST((NBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
     	     fib_path_list_pool_size());
-    FIB_TEST((NBR+7 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+7 == fib_entry_pool_size()), "entry pool size is %d",
     	     fib_entry_pool_size());
 
     /*
@@ -3438,9 +3439,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((0  == fib_path_list_db_size()),   "path list DB population:%d",
     	     fib_path_list_db_size());
-    FIB_TEST((NBR+4 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+4 == fib_path_list_pool_size()), "path list pool size is %d",
     	     fib_path_list_pool_size());
-    FIB_TEST((NBR+4 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+4 == fib_entry_pool_size()), "entry pool size is %d",
     	     fib_entry_pool_size());
 
     /*
@@ -3482,9 +3483,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((1  == fib_path_list_db_size()),   "path list DB population:%d",
     	     fib_path_list_db_size());
-    FIB_TEST((NBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+5 == fib_path_list_pool_size()), "path list pool size is %d",
     	     fib_path_list_pool_size());
-    FIB_TEST((NBR+5 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+5 == fib_entry_pool_size()), "entry pool size is %d",
     	     fib_entry_pool_size());
 
     fib_table_entry_delete(fib_index,
@@ -3493,9 +3494,9 @@ fib_test_v4 (void)
 
     FIB_TEST((0  == fib_path_list_db_size()),   "path list DB population:%d",
     	     fib_path_list_db_size());
-    FIB_TEST((NBR+4 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+4 == fib_path_list_pool_size()), "path list pool size is %d",
     	     fib_path_list_pool_size());
-    FIB_TEST((NBR+4 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+4 == fib_entry_pool_size()), "entry pool size is %d",
     	     fib_entry_pool_size());
 
     /*
@@ -3577,9 +3578,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((0  == fib_path_list_db_size()),   "path list DB population:%d",
     	     fib_path_list_db_size());
-    FIB_TEST((NBR+2 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR+2 == fib_path_list_pool_size()), "path list pool size is %d",
     	     fib_path_list_pool_size());
-    FIB_TEST((NBR+2 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR+2 == fib_entry_pool_size()), "entry pool size is %d",
     	     fib_entry_pool_size());
 
     /*
@@ -3619,9 +3620,9 @@ fib_test_v4 (void)
      */
     FIB_TEST((0  == fib_path_list_db_size()),   "path list DB population:%d",
     	     fib_path_list_db_size());
-    FIB_TEST((NBR == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR == fib_path_list_pool_size()), "path list pool size is %d",
     	     fib_path_list_pool_size());
-    FIB_TEST((NBR == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR == fib_entry_pool_size()), "entry pool size is %d",
     	     fib_entry_pool_size());
 
     /*
@@ -3644,11 +3645,11 @@ fib_test_v4 (void)
 
     FIB_TEST((0  == fib_path_list_db_size()), "path list DB population:%d",
     	     fib_path_list_db_size());
-    FIB_TEST((NBR-5 == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNBR-5 == fib_path_list_pool_size()), "path list pool size is %d",
     	     fib_path_list_pool_size());
-    FIB_TEST((NBR-5 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENBR-5 == fib_entry_pool_size()), "entry pool size is %d",
     	     fib_entry_pool_size());
-    FIB_TEST((NBR-5 == pool_elts(fib_urpf_list_pool)), "uRPF pool size is %d",
+    FIB_TEST((ENBR-5 == pool_elts(fib_urpf_list_pool)), "uRPF pool size is %d",
     	     pool_elts(fib_urpf_list_pool));
 
     return 0;
@@ -3720,13 +3721,15 @@ fib_test_v6 (void)
 
     /*
      * At this stage there is one v4 FIB with 5 routes and two v6 FIBs
-     * each with 6 entries. All entries are special so no path-list sharing.
+     * each with 2 entries and a v6 mfib with 4 path-lists.
+     * All entries are special so no path-list sharing.
      */
-#define NPS (5+6+6)
+#define ENPS (5+4)
+#define PNPS (5+4+4)
     FIB_TEST((0 == fib_path_list_db_size()),   "path list DB is empty");
-    FIB_TEST((NPS == fib_path_list_pool_size()), "path list pool size is %d",
+    FIB_TEST((PNPS == fib_path_list_pool_size()), "path list pool size is %d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NPS == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENPS == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -3816,9 +3819,9 @@ fib_test_v6 (void)
      * +2 entries. +2 unshared path-lists
      */
     FIB_TEST((0 == fib_path_list_db_size()),   "path list DB is empty");
-    FIB_TEST((NPS+2 == fib_path_list_pool_size()), "path list pool size is%d",
+    FIB_TEST((PNPS+2 == fib_path_list_pool_size()), "path list pool size is%d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NPS+2 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENPS+2 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -3862,9 +3865,9 @@ fib_test_v6 (void)
      */
     FIB_TEST((1 == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NPS+3 == fib_path_list_pool_size()), "path list pool size is%d",
+    FIB_TEST((PNPS+3 == fib_path_list_pool_size()), "path list pool size is%d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NPS+2 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENPS+2 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -3890,9 +3893,9 @@ fib_test_v6 (void)
      */
     FIB_TEST((0 == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NPS+2 == fib_path_list_pool_size()), "path list pool size is%d",
+    FIB_TEST((PNPS+2 == fib_path_list_pool_size()), "path list pool size is%d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NPS+2 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENPS+2 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -4006,9 +4009,9 @@ fib_test_v6 (void)
      */
     FIB_TEST((0 == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NPS+4 == fib_path_list_pool_size()), "path list pool size is%d",
+    FIB_TEST((PNPS+4 == fib_path_list_pool_size()), "path list pool size is%d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NPS+4 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENPS+4 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -4073,9 +4076,9 @@ fib_test_v6 (void)
      */
     FIB_TEST((1 == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NPS+5 == fib_path_list_pool_size()), "path list pool size is%d",
+    FIB_TEST((PNPS+5 == fib_path_list_pool_size()), "path list pool size is%d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NPS+6 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENPS+6 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -4201,9 +4204,9 @@ fib_test_v6 (void)
      */
     FIB_TEST((1 == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NPS+5 == fib_path_list_pool_size()), "path list pool size is%d",
+    FIB_TEST((PNPS+5 == fib_path_list_pool_size()), "path list pool size is%d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NPS+6 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENPS+6 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -4275,9 +4278,9 @@ fib_test_v6 (void)
      */
     FIB_TEST((1 == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NPS+7 == fib_path_list_pool_size()), "path list pool size is%d",
+    FIB_TEST((PNPS+7 == fib_path_list_pool_size()), "path list pool size is%d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NPS+8 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENPS+8 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
 
@@ -4401,9 +4404,9 @@ fib_test_v6 (void)
      */
     FIB_TEST((1 == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NPS+7 == fib_path_list_pool_size()), "path list pool size is%d",
+    FIB_TEST((PNPS+7 == fib_path_list_pool_size()), "path list pool size is%d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NPS+8 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENPS+8 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -4501,9 +4504,9 @@ fib_test_v6 (void)
      */
     FIB_TEST((0 == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NPS == fib_path_list_pool_size()), "path list pool size is%d",
+    FIB_TEST((PNPS == fib_path_list_pool_size()), "path list pool size is%d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NPS == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENPS == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     /*
@@ -4513,9 +4516,9 @@ fib_test_v6 (void)
 
     FIB_TEST((0 == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
-    FIB_TEST((NPS-6 == fib_path_list_pool_size()), "path list pool size is%d",
+    FIB_TEST((PNPS-2 == fib_path_list_pool_size()), "path list pool size is%d",
 	     fib_path_list_pool_size());
-    FIB_TEST((NPS-6 == fib_entry_pool_size()), "entry pool size is %d",
+    FIB_TEST((ENPS-2 == fib_entry_pool_size()), "entry pool size is %d",
 	     fib_entry_pool_size());
 
     adj_unlock(ai_02);
diff --git a/src/vnet/fib/fib_types.c b/src/vnet/fib/fib_types.c
index b66e7194..3ecb38e8 100644
--- a/src/vnet/fib/fib_types.c
+++ b/src/vnet/fib/fib_types.c
@@ -290,8 +290,10 @@ fib_forw_chain_type_to_link_type (fib_forward_chain_type_t fct)
     switch (fct)
     {
     case FIB_FORW_CHAIN_TYPE_UNICAST_IP4:
+    case FIB_FORW_CHAIN_TYPE_MCAST_IP4:
 	return (VNET_LINK_IP4);
     case FIB_FORW_CHAIN_TYPE_UNICAST_IP6:
+    case FIB_FORW_CHAIN_TYPE_MCAST_IP6:
 	return (VNET_LINK_IP6);
     case FIB_FORW_CHAIN_TYPE_ETHERNET:
 	return (VNET_LINK_ETHERNET);
@@ -313,8 +315,10 @@ fib_forw_chain_type_to_dpo_proto (fib_forward_chain_type_t fct)
     switch (fct)
     {
     case FIB_FORW_CHAIN_TYPE_UNICAST_IP4:
+    case FIB_FORW_CHAIN_TYPE_MCAST_IP4:
 	return (DPO_PROTO_IP4);
     case FIB_FORW_CHAIN_TYPE_UNICAST_IP6:
+    case FIB_FORW_CHAIN_TYPE_MCAST_IP6:
 	return (DPO_PROTO_IP6);
     case FIB_FORW_CHAIN_TYPE_ETHERNET:
 	return (DPO_PROTO_ETHERNET);
diff --git a/src/vnet/fib/fib_types.h b/src/vnet/fib/fib_types.h
index 0a15fef1..c51bc9c0 100644
--- a/src/vnet/fib/fib_types.h
+++ b/src/vnet/fib/fib_types.h
@@ -95,6 +95,14 @@ typedef enum fib_forward_chain_type_t_ {
      * option is converted into one of the other three internally.
      */
     FIB_FORW_CHAIN_TYPE_MPLS_EOS,
+    /**
+     * Contribute an object that is to be used to forward IP4 packets
+     */
+    FIB_FORW_CHAIN_TYPE_MCAST_IP4,
+    /**
+     * Contribute an object that is to be used to forward IP6 packets
+     */
+    FIB_FORW_CHAIN_TYPE_MCAST_IP6,
     /**
      * Contribute an object that is to be used to forward Ethernet packets.
      * This is last in the list since it is not valid for many FIB objects,
@@ -107,6 +115,8 @@ typedef enum fib_forward_chain_type_t_ {
     [FIB_FORW_CHAIN_TYPE_ETHERNET]      = "ethernet",     	\
     [FIB_FORW_CHAIN_TYPE_UNICAST_IP4]   = "unicast-ip4",	\
     [FIB_FORW_CHAIN_TYPE_UNICAST_IP6]   = "unicast-ip6",	\
+    [FIB_FORW_CHAIN_TYPE_MCAST_IP4]     = "multicast-ip4",	\
+    [FIB_FORW_CHAIN_TYPE_MCAST_IP6]     = "multicast-ip6",	\
     [FIB_FORW_CHAIN_TYPE_MPLS_NON_EOS]  = "mpls-neos",	        \
     [FIB_FORW_CHAIN_TYPE_MPLS_EOS]      = "mpls-eos",	        \
 }
@@ -263,6 +273,10 @@ typedef enum fib_route_path_flags_t_
      * Recursion constraint of via an attahced prefix
      */
     FIB_ROUTE_PATH_RESOLVE_VIA_ATTACHED = (1 << 1),
+    /**
+     * A for-us/local path
+     */
+    FIB_ROUTE_PATH_LOCAL = (1 << 2),
 } fib_route_path_flags_t;
 
 /**
diff --git a/src/vnet/fib/fib_urpf_list.c b/src/vnet/fib/fib_urpf_list.c
index 263812ad..b4844420 100644
--- a/src/vnet/fib/fib_urpf_list.c
+++ b/src/vnet/fib/fib_urpf_list.c
@@ -29,16 +29,24 @@ format_fib_urpf_list (u8 *s, va_list args)
     u32 *swi;
 
     ui = va_arg(args, index_t);
-    urpf = fib_urpf_list_get(ui);
 
-    s = format(s, "uPRF-list:%d len:%d itfs:[",
-	       ui, vec_len(urpf->furpf_itfs));
+    if (INDEX_INVALID != ui)
+    {
+        urpf = fib_urpf_list_get(ui);
+
+        s = format(s, "uPRF-list:%d len:%d itfs:[",
+                   ui, vec_len(urpf->furpf_itfs));
 
-    vec_foreach(swi, urpf->furpf_itfs)
+        vec_foreach(swi, urpf->furpf_itfs)
+        {
+            s = format(s, "%d, ", *swi);
+        }
+        s = format(s, "]");
+    }
+    else
     {
-	s = format(s, "%d, ", *swi);
+        s = format(s, "uRPF-list: None");
     }
-    s = format(s, "]");
 
     return (s);
 }
diff --git a/src/vnet/fib/ip4_fib.c b/src/vnet/fib/ip4_fib.c
index f6ebce00..e8211c80 100644
--- a/src/vnet/fib/ip4_fib.c
+++ b/src/vnet/fib/ip4_fib.c
@@ -378,16 +378,13 @@ ip4_fib_table_fwding_dpo_remove (ip4_fib_t *fib,
     ip4_fib_mtrie_add_del_route(fib, *addr, len, dpo->dpoi_index, 1); // DELETE
 }
 
-static void
-ip4_fib_table_show_all (ip4_fib_t *fib,
-			vlib_main_t * vm)
+void
+ip4_fib_table_walk (ip4_fib_t *fib,
+                    fib_table_walk_fn_t fn,
+                    void *ctx)
 {
-    fib_node_index_t *fib_entry_indicies;
-    fib_node_index_t *fib_entry_index;
     int i;
 
-    fib_entry_indicies = NULL;
-
     for (i = 0; i < ARRAY_LEN (fib->fib_entry_by_dst_address); i++)
     {
 	uword * hash = fib->fib_entry_by_dst_address[i];
@@ -398,14 +395,45 @@ ip4_fib_table_show_all (ip4_fib_t *fib,
 
 	    hash_foreach_pair (p, hash,
 	    ({
-		vec_add1(fib_entry_indicies, p->value[0]);
+		fn(p->value[0], ctx);
 	    }));
 	}
     }
+}
+
+/**
+ * Walk show context
+ */
+typedef struct ip4_fib_show_walk_ctx_t_
+{
+    fib_node_index_t *ifsw_indicies;
+} ip4_fib_show_walk_ctx_t;
+
+static int
+ip4_fib_show_walk_cb (fib_node_index_t fib_entry_index,
+                      void *arg)
+{
+    ip4_fib_show_walk_ctx_t *ctx = arg;
+
+    vec_add1(ctx->ifsw_indicies, fib_entry_index);
+
+    return (1);
+}
+
+static void
+ip4_fib_table_show_all (ip4_fib_t *fib,
+			vlib_main_t * vm)
+{
+    ip4_fib_show_walk_ctx_t ctx = {
+        .ifsw_indicies = NULL,
+    };
+    fib_node_index_t *fib_entry_index;
 
-    vec_sort_with_function(fib_entry_indicies, fib_entry_cmp_for_sort);
+    ip4_fib_table_walk(fib, ip4_fib_show_walk_cb, &ctx);
+    vec_sort_with_function(ctx.ifsw_indicies,
+                           fib_entry_cmp_for_sort);
 
-    vec_foreach(fib_entry_index, fib_entry_indicies)
+    vec_foreach(fib_entry_index, ctx.ifsw_indicies)
     {
 	vlib_cli_output(vm, "%U",
                         format_fib_entry,
@@ -413,7 +441,7 @@ ip4_fib_table_show_all (ip4_fib_t *fib,
                         FIB_ENTRY_FORMAT_BRIEF);
     }
 
-    vec_free(fib_entry_indicies);
+    vec_free(ctx.ifsw_indicies);
 }
 
 static void
diff --git a/src/vnet/fib/ip4_fib.h b/src/vnet/fib/ip4_fib.h
index cf312cdc..a8dc68b5 100644
--- a/src/vnet/fib/ip4_fib.h
+++ b/src/vnet/fib/ip4_fib.h
@@ -64,6 +64,15 @@ extern void ip4_fib_table_fwding_dpo_remove(ip4_fib_t *fib,
 extern u32 ip4_fib_table_lookup_lb (ip4_fib_t *fib,
 				    const ip4_address_t * dst);
 
+/**
+ * @brief Walk all entries in a FIB table
+ * N.B: This is NOT safe to deletes. If you need to delete walk the whole
+ * table and store elements in a vector, then delete the elements
+ */
+extern void ip4_fib_table_walk(ip4_fib_t *fib,
+                               fib_table_walk_fn_t fn,
+                               void *ctx);
+
 /**
  * @brief Get the FIB at the given index
  */
diff --git a/src/vnet/fib/ip6_fib.c b/src/vnet/fib/ip6_fib.c
index d5b9bdcb..343ff55e 100644
--- a/src/vnet/fib/ip6_fib.c
+++ b/src/vnet/fib/ip6_fib.c
@@ -38,57 +38,6 @@ vnet_ip6_fib_init (u32 fib_index)
 				FIB_ENTRY_FLAG_DROP,
 				ADJ_INDEX_INVALID);
 
-    /*
-     * Add ff02::1:ff00:0/104 via local route for all tables.
-     *  This is required for neighbor discovery to work.
-     */
-    ip6_set_solicited_node_multicast_address(&pfx.fp_addr.ip6, 0);
-    pfx.fp_len = 104;
-    fib_table_entry_special_add(fib_index,
-				&pfx,
-				FIB_SOURCE_SPECIAL,
-				FIB_ENTRY_FLAG_LOCAL,
-				ADJ_INDEX_INVALID);
-
-    /*
-     * Add all-routers multicast address via local route for all tables
-     */
-    ip6_set_reserved_multicast_address (&pfx.fp_addr.ip6,
-					IP6_MULTICAST_SCOPE_link_local,
-					IP6_MULTICAST_GROUP_ID_all_routers);
-    pfx.fp_len = 128;
-    fib_table_entry_special_add(fib_index,
-				&pfx,
-				FIB_SOURCE_SPECIAL,
-				FIB_ENTRY_FLAG_LOCAL,
-				ADJ_INDEX_INVALID);
-
-    /*
-     * Add all-nodes multicast address via local route for all tables
-     */
-    ip6_set_reserved_multicast_address (&pfx.fp_addr.ip6,
-					IP6_MULTICAST_SCOPE_link_local,
-					IP6_MULTICAST_GROUP_ID_all_hosts);
-    pfx.fp_len = 128;
-    fib_table_entry_special_add(fib_index,
-				&pfx,
-				FIB_SOURCE_SPECIAL,
-				FIB_ENTRY_FLAG_LOCAL,
-				ADJ_INDEX_INVALID);
-
-    /*
-     *  Add all-mldv2  multicast address via local route for all tables
-     */
-    ip6_set_reserved_multicast_address (&pfx.fp_addr.ip6,
-					IP6_MULTICAST_SCOPE_link_local,
-					IP6_MULTICAST_GROUP_ID_mldv2_routers);
-    pfx.fp_len = 128;
-    fib_table_entry_special_add(fib_index,
-				&pfx,
-				FIB_SOURCE_SPECIAL,
-				FIB_ENTRY_FLAG_LOCAL,
-				ADJ_INDEX_INVALID);
-
     /*
      * all link local for us
      */
@@ -512,27 +461,68 @@ ip6_fib_table_fwding_dpo_remove (u32 fib_index,
     if (--table->dst_address_length_refcounts[len] == 0)
     {
 	table->non_empty_dst_address_length_bitmap =
-            clib_bitmap_set (table->non_empty_dst_address_length_bitmap, 
+            clib_bitmap_set (table->non_empty_dst_address_length_bitmap,
                              128 - len, 0);
 	compute_prefix_lengths_in_search_order (table);
     }
 }
 
+/**
+ * @brief Context when walking the IPv6 table. Since all VRFs are in the
+ * same hash table, we need to filter only those we need as we walk
+ */
+typedef struct ip6_fib_walk_ctx_t_
+{
+    u32 i6w_fib_index;
+    fib_table_walk_fn_t i6w_fn;
+    void *i6w_ctx;
+} ip6_fib_walk_ctx_t;
+
+static int
+ip6_fib_walk_cb (clib_bihash_kv_24_8_t * kvp,
+                 void *arg)
+{
+    ip6_fib_walk_ctx_t *ctx = arg;
+
+    if ((kvp->key[2] >> 32) == ctx->i6w_fib_index)
+    {
+        ctx->i6w_fn(kvp->value, ctx->i6w_ctx);
+    }
+
+    return (1);
+}
+
+void
+ip6_fib_table_walk (u32 fib_index,
+                    fib_table_walk_fn_t fn,
+                    void *arg)
+{
+    ip6_fib_walk_ctx_t ctx = {
+        .i6w_fib_index = fib_index,
+        .i6w_fn = fn,
+        .i6w_ctx = arg,
+    };
+    ip6_main_t *im = &ip6_main;
+
+    BV(clib_bihash_foreach_key_value_pair)(&im->ip6_table[IP6_FIB_TABLE_NON_FWDING].ip6_hash,
+					   ip6_fib_walk_cb,
+					   &ctx);
+
+}
+
 typedef struct ip6_fib_show_ctx_t_ {
-    u32 fib_index;
     fib_node_index_t *entries;
 } ip6_fib_show_ctx_t;
 
-static void
-ip6_fib_table_collect_entries (clib_bihash_kv_24_8_t * kvp,
-			       void *arg)
+static int
+ip6_fib_table_show_walk (fib_node_index_t fib_entry_index,
+                         void *arg)
 {
     ip6_fib_show_ctx_t *ctx = arg;
 
-    if ((kvp->key[2] >> 32) == ctx->fib_index)
-    {
-	vec_add1(ctx->entries, kvp->value);
-    }
+    vec_add1(ctx->entries, fib_entry_index);
+
+    return (1);
 }
 
 static void
@@ -541,15 +531,10 @@ ip6_fib_table_show_all (ip6_fib_t *fib,
 {
     fib_node_index_t *fib_entry_index;
     ip6_fib_show_ctx_t ctx = {
-	.fib_index = fib->index,
 	.entries = NULL,
     };
-    ip6_main_t *im = &ip6_main;
-
-    BV(clib_bihash_foreach_key_value_pair)(&im->ip6_table[IP6_FIB_TABLE_NON_FWDING].ip6_hash,
-					   ip6_fib_table_collect_entries,
-					   &ctx);
 
+    ip6_fib_table_walk(fib->index, ip6_fib_table_show_walk, &ctx);
     vec_sort_with_function(ctx.entries, fib_entry_cmp_for_sort);
 
     vec_foreach(fib_entry_index, ctx.entries)
diff --git a/src/vnet/fib/ip6_fib.h b/src/vnet/fib/ip6_fib.h
index f6af993a..78da3746 100644
--- a/src/vnet/fib/ip6_fib.h
+++ b/src/vnet/fib/ip6_fib.h
@@ -54,9 +54,18 @@ u32 ip6_fib_table_fwding_lookup_with_if_index(ip6_main_t * im,
 					      u32 sw_if_index,
 					      const ip6_address_t * dst);
 u32 ip6_fib_table_fwding_lookup(ip6_main_t * im,
-				u32 fib_index, 
+				u32 fib_index,
 				const ip6_address_t * dst);
 
+/**
+ * @brief Walk all entries in a FIB table
+ * N.B: This is NOT safe to deletes. If you need to delete walk the whole
+ * table and store elements in a vector, then delete the elements
+ */
+extern void ip6_fib_table_walk(u32 fib_index,
+                               fib_table_walk_fn_t fn,
+                               void *ctx);
+
 /**
  * @biref return the DPO that the LB stacks on.
  */
diff --git a/src/vnet/fib/mpls_fib.c b/src/vnet/fib/mpls_fib.c
index 6a9b1ac2..cc657975 100644
--- a/src/vnet/fib/mpls_fib.c
+++ b/src/vnet/fib/mpls_fib.c
@@ -44,10 +44,11 @@
  * Switching between schemes based on observed/measured action similarity is not
  * considered on the grounds of complexity and flip-flopping.
  *
- * VPP mantra - favour performance over memory. We choose a 21 bit key.  
+ * VPP mantra - favour performance over memory. We choose a 21 bit key.
  */
 
 #include <vnet/fib/fib_table.h>
+#include <vnet/fib/mpls_fib.h>
 #include <vnet/dpo/load_balance.h>
 #include <vnet/dpo/drop_dpo.h>
 #include <vnet/dpo/punt_dpo.h>
@@ -342,6 +343,20 @@ mpls_fib_table_get_flow_hash_config (u32 fib_index)
     return (0);
 }
 
+void
+mpls_fib_table_walk (mpls_fib_t *mpls_fib,
+                     fib_table_walk_fn_t fn,
+                     void *ctx)
+{
+    fib_node_index_t lfei;
+    mpls_label_t key;
+
+    hash_foreach(key, lfei, mpls_fib->mf_entries,
+    ({
+	fn(lfei, ctx);
+    }));
+}
+
 static void
 mpls_fib_table_show_all (const mpls_fib_t *mpls_fib,
 			 vlib_main_t * vm)
diff --git a/src/vnet/fib/mpls_fib.h b/src/vnet/fib/mpls_fib.h
index 93ae4623..e2ef9253 100644
--- a/src/vnet/fib/mpls_fib.h
+++ b/src/vnet/fib/mpls_fib.h
@@ -70,6 +70,15 @@ extern void mpls_fib_forwarding_table_reset(mpls_fib_t *mf,
 					    mpls_label_t label,
 					    mpls_eos_bit_t eos);
 
+/**
+ * @brief Walk all entries in a FIB table
+ * N.B: This is NOT safe to deletes. If you need to delete walk the whole
+ * table and store elements in a vector, then delete the elements
+ */
+extern void mpls_fib_table_walk(mpls_fib_t *fib,
+                                fib_table_walk_fn_t fn,
+                                void *ctx);
+
 /**
  * @brief
  *  Lookup a label and EOS bit in the MPLS_FIB table to retrieve the
diff --git a/src/vnet/ip/ip.api b/src/vnet/ip/ip.api
index 5371696c..65f6e7a7 100644
--- a/src/vnet/ip/ip.api
+++ b/src/vnet/ip/ip.api
@@ -395,6 +395,41 @@ define ip_add_del_route_reply
   i32 retval;
 };
 
+/** \brief Add / del route request
+    @param client_index - opaque cookie to identify the sender
+    @param context - sender context, to match reply w/ request
+    @param sw_if_index - software index of the new vlan's parent interface
+    @param vrf_id - fib table /vrf associated with the route
+
+    FIXME
+*/
+define ip_mroute_add_del
+{
+  u32 client_index;
+  u32 context;
+  u32 next_hop_sw_if_index;
+  u32 table_id;
+  u32 entry_flags;
+  u32 itf_flags;
+  u16 grp_address_length;
+  u8 create_vrf_if_needed;
+  u8 is_add;
+  u8 is_ipv6;
+  u8 is_local;
+  u8 grp_address[16];
+  u8 src_address[16];
+};
+
+/** \brief Reply for add / del mroute request
+    @param context - returned sender context, to match reply w/ request
+    @param retval - return code
+*/
+define ip_mroute_add_del_reply
+{
+  u32 context;
+  i32 retval;
+};
+
 define ip_address_details
 {
   u32 client_index;
@@ -424,6 +459,24 @@ define ip_dump
   u8 is_ipv6;
 };
 
+define mfib_signal_dump
+{
+  u32 client_index;
+  u32 context;
+};
+
+define mfib_signal_details
+{
+  u32 client_index;
+  u32 context;
+  u32 sw_if_index;
+  u32 table_id;
+  u16 grp_address_len;
+  u8 grp_address[16];
+  u8 src_address[16];
+  u16 ip_packet_len;
+  u8 ip_packet_data[256];
+};
 
 /*
  * Local Variables:
diff --git a/src/vnet/ip/ip4.h b/src/vnet/ip/ip4.h
index cc33dff4..b184fbae 100644
--- a/src/vnet/ip/ip4.h
+++ b/src/vnet/ip/ip4.h
@@ -68,6 +68,18 @@ typedef struct ip4_fib_t
 
 } ip4_fib_t;
 
+typedef struct ip4_mfib_t
+{
+  /* Hash table for each prefix length mapping. */
+  uword *fib_entry_by_dst_address[65];
+
+  /* Table ID (hash key) for this FIB. */
+  u32 table_id;
+
+  /* Index into FIB vector. */
+  u32 index;
+} ip4_mfib_t;
+
 struct ip4_main_t;
 
 typedef void (ip4_add_del_interface_address_function_t)
@@ -99,11 +111,17 @@ typedef struct ip4_main_t
   /** Vector of FIBs. */
   struct fib_table_t_ *fibs;
 
+  /** Vector of MFIBs. */
+  struct mfib_table_t_ *mfibs;
+
   u32 fib_masks[33];
 
   /** Table index indexed by software interface. */
   u32 *fib_index_by_sw_if_index;
 
+  /** Table index indexed by software interface. */
+  u32 *mfib_index_by_sw_if_index;
+
   /* IP4 enabled count by software interface */
   u8 *ip_enabled_by_sw_if_index;
 
@@ -111,6 +129,10 @@ typedef struct ip4_main_t
      ID space is not necessarily dense; index space is dense. */
   uword *fib_index_by_table_id;
 
+  /** Hash table mapping table id to multicast fib index.
+     ID space is not necessarily dense; index space is dense. */
+  uword *mfib_index_by_table_id;
+
   /** Functions to call when interface address changes. */
     ip4_add_del_interface_address_callback_t
     * add_del_interface_address_callbacks;
@@ -140,7 +162,9 @@ extern ip4_main_t ip4_main;
 /** Global ip4 input node.  Errors get attached to ip4 input node. */
 extern vlib_node_registration_t ip4_input_node;
 extern vlib_node_registration_t ip4_lookup_node;
+extern vlib_node_registration_t ip4_local_node;
 extern vlib_node_registration_t ip4_rewrite_node;
+extern vlib_node_registration_t ip4_rewrite_mcast_node;
 extern vlib_node_registration_t ip4_rewrite_local_node;
 extern vlib_node_registration_t ip4_arp_node;
 extern vlib_node_registration_t ip4_glean_node;
diff --git a/src/vnet/ip/ip4_forward.c b/src/vnet/ip/ip4_forward.c
index 87b345bd..8081b34b 100644
--- a/src/vnet/ip/ip4_forward.c
+++ b/src/vnet/ip/ip4_forward.c
@@ -50,6 +50,7 @@
 #include <vnet/fib/ip4_fib.h>
 #include <vnet/dpo/load_balance.h>
 #include <vnet/dpo/classify_dpo.h>
+#include <vnet/mfib/mfib_table.h>	/* for mFIB table and entry creation */
 
 /**
  * @file
@@ -714,16 +715,17 @@ ip4_interface_first_address (ip4_main_t * im, u32 sw_if_index,
   ip_interface_address_t *ia = 0;
   ip4_address_t *result = 0;
 
-  foreach_ip_interface_address (lm, ia, sw_if_index,
-				1 /* honor unnumbered */ ,
-				(
-				  {
-				  ip4_address_t * a =
-				  ip_interface_address_get_address (lm, ia);
-				  result = a;
-				  break;
-				  }
-				));
+  /* *INDENT-OFF* */
+  foreach_ip_interface_address
+    (lm, ia, sw_if_index,
+     1 /* honor unnumbered */ ,
+     ({
+       ip4_address_t * a =
+         ip_interface_address_get_address (lm, ia);
+       result = a;
+       break;
+     }));
+  /* *INDENT-OFF* */
   if (result_ia)
     *result_ia = result ? ia : 0;
   return result;
@@ -748,9 +750,19 @@ ip4_add_interface_routes (u32 sw_if_index,
     {
       fib_node_index_t fei;
 
-      fei = fib_table_entry_update_one_path (fib_index, &pfx, FIB_SOURCE_INTERFACE, (FIB_ENTRY_FLAG_CONNECTED | FIB_ENTRY_FLAG_ATTACHED), FIB_PROTOCOL_IP4, NULL,	/* No next-hop address */
-					     sw_if_index, ~0,	// invalid FIB index
-					     1, NULL,	// no out-label stack
+      fei = fib_table_entry_update_one_path (fib_index, &pfx,
+                                             FIB_SOURCE_INTERFACE,
+                                             (FIB_ENTRY_FLAG_CONNECTED |
+                                              FIB_ENTRY_FLAG_ATTACHED),
+                                             FIB_PROTOCOL_IP4,
+                                             /* No next-hop address */
+                                             NULL,
+					     sw_if_index,
+                                             // invalid FIB index
+                                             ~0,
+					     1,
+                                             // no out-label stack
+                                             NULL,
 					     FIB_ROUTE_PATH_FLAG_NONE);
       a->neighbor_probe_adj_index = fib_entry_get_adj (fei);
     }
@@ -778,8 +790,16 @@ ip4_add_interface_routes (u32 sw_if_index,
 	}
     }
 
-  fib_table_entry_update_one_path (fib_index, &pfx, FIB_SOURCE_INTERFACE, (FIB_ENTRY_FLAG_CONNECTED | FIB_ENTRY_FLAG_LOCAL), FIB_PROTOCOL_IP4, &pfx.fp_addr, sw_if_index, ~0,	// invalid FIB index
-				   1, NULL,	// no out-label stack
+  fib_table_entry_update_one_path (fib_index, &pfx,
+                                   FIB_SOURCE_INTERFACE,
+                                   (FIB_ENTRY_FLAG_CONNECTED |
+                                    FIB_ENTRY_FLAG_LOCAL),
+                                   FIB_PROTOCOL_IP4,
+                                   &pfx.fp_addr,
+                                   sw_if_index,
+                                   // invalid FIB index
+                                   ~0,
+				   1, NULL,
 				   FIB_ROUTE_PATH_FLAG_NONE);
 }
 
@@ -827,9 +847,10 @@ ip4_sw_interface_enable_disable (u32 sw_if_index, u32 is_enable)
   vnet_feature_enable_disable ("ip4-unicast", "ip4-drop", sw_if_index,
 			       !is_enable, 0, 0);
 
-  vnet_feature_enable_disable ("ip4-multicast", "ip4-drop", sw_if_index,
-			       !is_enable, 0, 0);
 
+  vnet_feature_enable_disable ("ip4-multicast",
+			       "ip4-mfib-forward-lookup",
+			       sw_if_index, is_enable, 0, 0);
 }
 
 static clib_error_t *
@@ -855,36 +876,37 @@ ip4_add_del_interface_address_internal (vlib_main_t * vm,
    * subnets on interfaces. Easy fix - disallow overlapping subnets, like
    * most routers do.
    */
+  /* *INDENT-OFF* */
   if (!is_del)
     {
       /* When adding an address check that it does not conflict
          with an existing address. */
       ip_interface_address_t *ia;
-      foreach_ip_interface_address (&im->lookup_main, ia, sw_if_index,
-				    0 /* honor unnumbered */ ,
-				    (
-				      {
-				      ip4_address_t * x =
-				      ip_interface_address_get_address
-				      (&im->lookup_main, ia);
-				      if (ip4_destination_matches_route
-					  (im, address, x, ia->address_length)
-					  ||
-					  ip4_destination_matches_route (im,
-									 x,
-									 address,
-									 address_length))
-				      return
-				      clib_error_create
-				      ("failed to add %U which conflicts with %U for interface %U",
-				       format_ip4_address_and_length, address,
-				       address_length,
-				       format_ip4_address_and_length, x,
-				       ia->address_length,
-				       format_vnet_sw_if_index_name, vnm,
-				       sw_if_index);}
-				    ));
+      foreach_ip_interface_address
+        (&im->lookup_main, ia, sw_if_index,
+         0 /* honor unnumbered */ ,
+         ({
+           ip4_address_t * x =
+             ip_interface_address_get_address
+             (&im->lookup_main, ia);
+           if (ip4_destination_matches_route
+               (im, address, x, ia->address_length) ||
+               ip4_destination_matches_route (im,
+                                              x,
+                                              address,
+                                              address_length))
+             return
+               clib_error_create
+               ("failed to add %U which conflicts with %U for interface %U",
+                format_ip4_address_and_length, address,
+                address_length,
+                format_ip4_address_and_length, x,
+                ia->address_length,
+                format_vnet_sw_if_index_name, vnm,
+                sw_if_index);
+         }));
     }
+  /* *INDENT-ON* */
 
   elts_before = pool_elts (lm->if_address_pool);
 
@@ -918,9 +940,10 @@ done:
 }
 
 clib_error_t *
-ip4_add_del_interface_address (vlib_main_t * vm, u32 sw_if_index,
-			       ip4_address_t * address, u32 address_length,
-			       u32 is_del)
+ip4_add_del_interface_address (vlib_main_t * vm,
+			       u32 sw_if_index,
+			       ip4_address_t * address,
+			       u32 address_length, u32 is_del)
 {
   return ip4_add_del_interface_address_internal
     (vm, sw_if_index, address, address_length, is_del);
@@ -1027,13 +1050,13 @@ VNET_FEATURE_INIT (ip4_vpath_mc, static) =
 {
   .arc_name = "ip4-multicast",
   .node_name = "vpath-input-ip4",
-  .runs_before = VNET_FEATURES ("ip4-lookup-multicast"),
+  .runs_before = VNET_FEATURES ("ip4-mfib-forward-lookup"),
 };
 
 VNET_FEATURE_INIT (ip4_lookup_mc, static) =
 {
   .arc_name = "ip4-multicast",
-  .node_name = "ip4-lookup-multicast",
+  .node_name = "ip4-mfib-forward-lookup",
   .runs_before = VNET_FEATURES ("ip4-drop"),
 };
 
@@ -1083,6 +1106,7 @@ ip4_sw_interface_add_del (vnet_main_t * vnm, u32 sw_if_index, u32 is_add)
 
   /* Fill in lookup tables with default table (0). */
   vec_validate (im->fib_index_by_sw_if_index, sw_if_index);
+  vec_validate (im->mfib_index_by_sw_if_index, sw_if_index);
 
   vnet_feature_enable_disable ("ip4-unicast", "ip4-drop", sw_if_index,
 			       is_add, 0, 0);
@@ -1123,6 +1147,7 @@ ip4_lookup_init (vlib_main_t * vm)
 
   /* Create FIB with index 0 and table id of 0. */
   fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, 0);
+  mfib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, 0);
 
   {
     pg_node_t *pn;
@@ -1341,27 +1366,35 @@ ip4_punt (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame)
   return ip4_drop_or_punt (vm, node, frame, IP4_ERROR_ADJACENCY_PUNT);
 }
 
+/* *INDENT-OFF* */
 VLIB_REGISTER_NODE (ip4_drop_node, static) =
 {
-  .function = ip4_drop,.name = "ip4-drop",.vector_size =
-    sizeof (u32),.format_trace = format_ip4_forward_next_trace,.n_next_nodes =
-    1,.next_nodes =
-  {
-  [0] = "error-drop",}
-,};
+  .function = ip4_drop,.
+  name = "ip4-drop",
+  .vector_size = sizeof (u32),
+  .format_trace = format_ip4_forward_next_trace,
+  .n_next_nodes = 1,
+  .next_nodes = {
+    [0] = "error-drop",
+  },
+};
 
 VLIB_NODE_FUNCTION_MULTIARCH (ip4_drop_node, ip4_drop);
 
 VLIB_REGISTER_NODE (ip4_punt_node, static) =
 {
-  .function = ip4_punt,.name = "ip4-punt",.vector_size =
-    sizeof (u32),.format_trace = format_ip4_forward_next_trace,.n_next_nodes =
-    1,.next_nodes =
-  {
-  [0] = "error-punt",}
-,};
+  .function = ip4_punt,
+  .name = "ip4-punt",
+  .vector_size = sizeof (u32),
+  .format_trace = format_ip4_forward_next_trace,
+  .n_next_nodes = 1,
+  .next_nodes = {
+    [0] = "error-punt",
+  },
+};
 
 VLIB_NODE_FUNCTION_MULTIARCH (ip4_punt_node, ip4_punt);
+/* *INDENT-ON */
 
 /* Compute TCP/UDP/ICMP4 checksum in software. */
 u16
@@ -1500,13 +1533,15 @@ ip4_local (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame)
 
 	  fib_index0 = vec_elt (im->fib_index_by_sw_if_index,
 				vnet_buffer (p0)->sw_if_index[VLIB_RX]);
-	  fib_index0 = (vnet_buffer (p0)->sw_if_index[VLIB_TX] == (u32) ~ 0) ?
-	    fib_index0 : vnet_buffer (p0)->sw_if_index[VLIB_TX];
+	  fib_index0 =
+	    (vnet_buffer (p0)->sw_if_index[VLIB_TX] ==
+	     (u32) ~ 0) ? fib_index0 : vnet_buffer (p0)->sw_if_index[VLIB_TX];
 
 	  fib_index1 = vec_elt (im->fib_index_by_sw_if_index,
 				vnet_buffer (p1)->sw_if_index[VLIB_RX]);
-	  fib_index1 = (vnet_buffer (p1)->sw_if_index[VLIB_TX] == (u32) ~ 0) ?
-	    fib_index1 : vnet_buffer (p1)->sw_if_index[VLIB_TX];
+	  fib_index1 =
+	    (vnet_buffer (p1)->sw_if_index[VLIB_TX] ==
+	     (u32) ~ 0) ? fib_index1 : vnet_buffer (p1)->sw_if_index[VLIB_TX];
 
 	  mtrie0 = &ip4_fib_get (fib_index0)->mtrie;
 	  mtrie1 = &ip4_fib_get (fib_index1)->mtrie;
@@ -1726,8 +1761,9 @@ ip4_local (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame)
 
 	  fib_index0 = vec_elt (im->fib_index_by_sw_if_index,
 				vnet_buffer (p0)->sw_if_index[VLIB_RX]);
-	  fib_index0 = (vnet_buffer (p0)->sw_if_index[VLIB_TX] == (u32) ~ 0) ?
-	    fib_index0 : vnet_buffer (p0)->sw_if_index[VLIB_TX];
+	  fib_index0 =
+	    (vnet_buffer (p0)->sw_if_index[VLIB_TX] ==
+	     (u32) ~ 0) ? fib_index0 : vnet_buffer (p0)->sw_if_index[VLIB_TX];
 
 	  mtrie0 = &ip4_fib_get (fib_index0)->mtrie;
 
@@ -1838,10 +1874,11 @@ ip4_local (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame)
   return frame->n_vectors;
 }
 
-VLIB_REGISTER_NODE (ip4_local_node, static) =
+VLIB_REGISTER_NODE (ip4_local_node) =
 {
   .function = ip4_local,.name = "ip4-local",.vector_size =
-    sizeof (u32),.format_trace = format_ip4_forward_next_trace,.n_next_nodes =
+    sizeof (u32),.format_trace =
+    format_ip4_forward_next_trace,.n_next_nodes =
     IP_LOCAL_N_NEXT,.next_nodes =
   {
   [IP_LOCAL_NEXT_DROP] = "error-drop",
@@ -2022,8 +2059,8 @@ ip4_arp_inline (vlib_main_t * vm,
 	   * Can happen if the control-plane is programming tables
 	   * with traffic flowing; at least that's today's lame excuse.
 	   */
-	  if ((is_glean && adj0->lookup_next_index != IP_LOOKUP_NEXT_GLEAN) ||
-	      (!is_glean && adj0->lookup_next_index != IP_LOOKUP_NEXT_ARP))
+	  if ((is_glean && adj0->lookup_next_index != IP_LOOKUP_NEXT_GLEAN)
+	      || (!is_glean && adj0->lookup_next_index != IP_LOOKUP_NEXT_ARP))
 	    {
 	      p0->error = node->errors[IP4_ARP_ERROR_NON_ARP_ADJ];
 	    }
@@ -2196,15 +2233,17 @@ ip4_probe_neighbor (vlib_main_t * vm, ip4_address_t * dst, u32 sw_if_index)
     {
       vnm->api_errno = VNET_API_ERROR_NO_MATCHING_INTERFACE;
       return clib_error_return
-	(0, "no matching interface address for destination %U (interface %U)",
-	 format_ip4_address, dst,
-	 format_vnet_sw_if_index_name, vnm, sw_if_index);
+	(0,
+	 "no matching interface address for destination %U (interface %U)",
+	 format_ip4_address, dst, format_vnet_sw_if_index_name, vnm,
+	 sw_if_index);
     }
 
   adj = ip_get_adjacency (&im->lookup_main, ia->neighbor_probe_adj_index);
 
   h =
-    vlib_packet_template_get_packet (vm, &im->ip4_arp_request_packet_template,
+    vlib_packet_template_get_packet (vm,
+				     &im->ip4_arp_request_packet_template,
 				     &bi);
 
   hi = vnet_get_sup_hw_interface (vnm, sw_if_index);
@@ -2243,7 +2282,7 @@ typedef enum
 always_inline uword
 ip4_rewrite_inline (vlib_main_t * vm,
 		    vlib_node_runtime_t * node,
-		    vlib_frame_t * frame, int is_midchain)
+		    vlib_frame_t * frame, int is_midchain, int is_mcast)
 {
   ip_lookup_main_t *lm = &ip4_main.lookup_main;
   u32 *from = vlib_frame_vector_args (frame);
@@ -2457,6 +2496,14 @@ ip4_rewrite_inline (vlib_main_t * vm,
 	      adj0->sub_type.midchain.fixup_func (vm, adj0, p0);
 	      adj1->sub_type.midchain.fixup_func (vm, adj1, p1);
 	    }
+	  if (is_mcast)
+	    {
+	      /*
+	       * copy bytes from the IP address into the MAC rewrite
+	       */
+	      vnet_fixup_one_header (adj0[0], &ip0->dst_address, ip0, 1);
+	      vnet_fixup_one_header (adj1[0], &ip1->dst_address, ip1, 1);
+	    }
 
 	  vlib_validate_buffer_enqueue_x2 (vm, node, next_index,
 					   to_next, n_left_to_next,
@@ -2530,6 +2577,13 @@ ip4_rewrite_inline (vlib_main_t * vm,
 
 	  /* Guess we are only writing on simple Ethernet header. */
 	  vnet_rewrite_one_header (adj0[0], ip0, sizeof (ethernet_header_t));
+	  if (is_mcast)
+	    {
+	      /*
+	       * copy bytes from the IP address into the MAC rewrite
+	       */
+	      vnet_fixup_one_header (adj0[0], &ip0->dst_address, ip0, 1);
+	    }
 
 	  /* Update packet buffer attributes/set output interface. */
 	  rw_len0 = adj0[0].rewrite_header.data_bytes;
@@ -2624,36 +2678,58 @@ static uword
 ip4_rewrite (vlib_main_t * vm,
 	     vlib_node_runtime_t * node, vlib_frame_t * frame)
 {
-  return ip4_rewrite_inline (vm, node, frame, 0);
+  return ip4_rewrite_inline (vm, node, frame, 0, 0);
 }
 
 static uword
 ip4_midchain (vlib_main_t * vm,
 	      vlib_node_runtime_t * node, vlib_frame_t * frame)
 {
-  return ip4_rewrite_inline (vm, node, frame, 1);
+  return ip4_rewrite_inline (vm, node, frame, 1, 0);
 }
 
-
-VLIB_REGISTER_NODE (ip4_rewrite_node) =
+static uword
+ip4_rewrite_mcast (vlib_main_t * vm,
+		   vlib_node_runtime_t * node, vlib_frame_t * frame)
 {
-  .function = ip4_rewrite,.name = "ip4-rewrite",.vector_size =
-    sizeof (u32),.format_trace = format_ip4_rewrite_trace,.n_next_nodes =
-    2,.next_nodes =
-  {
-  [IP4_REWRITE_NEXT_DROP] = "error-drop",
-      [IP4_REWRITE_NEXT_ICMP_ERROR] = "ip4-icmp-error",}
-,};
+  return ip4_rewrite_inline (vm, node, frame, 0, 1);
+}
 
-VLIB_NODE_FUNCTION_MULTIARCH (ip4_rewrite_node, ip4_rewrite);
+/* *INDENT-OFF* */
+VLIB_REGISTER_NODE (ip4_rewrite_node) = {
+  .function = ip4_rewrite,
+  .name = "ip4-rewrite",
+  .vector_size = sizeof (u32),
+
+  .format_trace = format_ip4_rewrite_trace,
+
+  .n_next_nodes = 2,
+  .next_nodes = {
+    [IP4_REWRITE_NEXT_DROP] = "error-drop",
+    [IP4_REWRITE_NEXT_ICMP_ERROR] = "ip4-icmp-error",
+  },
+};
+VLIB_NODE_FUNCTION_MULTIARCH (ip4_rewrite_node, ip4_rewrite)
 
-VLIB_REGISTER_NODE (ip4_midchain_node) =
-{
-.function = ip4_midchain,.name = "ip4-midchain",.vector_size =
-    sizeof (u32),.format_trace = format_ip4_forward_next_trace,.sibling_of =
-    "ip4-rewrite",};
+VLIB_REGISTER_NODE (ip4_rewrite_mcast_node) = {
+  .function = ip4_rewrite_mcast,
+  .name = "ip4-rewrite-mcast",
+  .vector_size = sizeof (u32),
 
+  .format_trace = format_ip4_rewrite_trace,
+  .sibling_of = "ip4-rewrite",
+};
+VLIB_NODE_FUNCTION_MULTIARCH (ip4_rewrite_mcast_node, ip4_rewrite_mcast)
+
+VLIB_REGISTER_NODE (ip4_midchain_node) = {
+  .function = ip4_midchain,
+  .name = "ip4-midchain",
+  .vector_size = sizeof (u32),
+  .format_trace = format_ip4_forward_next_trace,
+  .sibling_of =  "ip4-rewrite",
+};
 VLIB_NODE_FUNCTION_MULTIARCH (ip4_midchain_node, ip4_midchain);
+/* *INDENT-ON */
 
 static clib_error_t *
 add_del_interface_table (vlib_main_t * vm,
@@ -2695,6 +2771,11 @@ add_del_interface_table (vlib_main_t * vm,
     //
     vec_validate (im->fib_index_by_sw_if_index, sw_if_index);
     im->fib_index_by_sw_if_index[sw_if_index] = fib_index;
+
+    fib_index = mfib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4,
+                                                   table_id);
+    vec_validate (im->mfib_index_by_sw_if_index, sw_if_index);
+    im->mfib_index_by_sw_if_index[sw_if_index] = fib_index;
   }
 
 done:
@@ -2730,243 +2811,6 @@ VLIB_CLI_COMMAND (set_interface_ip_table_command, static) =
 };
 /* *INDENT-ON* */
 
-
-static uword
-ip4_lookup_multicast (vlib_main_t * vm,
-		      vlib_node_runtime_t * node, vlib_frame_t * frame)
-{
-  ip4_main_t *im = &ip4_main;
-  vlib_combined_counter_main_t *cm = &load_balance_main.lbm_to_counters;
-  u32 n_left_from, n_left_to_next, *from, *to_next;
-  ip_lookup_next_t next;
-  u32 cpu_index = os_get_cpu_number ();
-
-  from = vlib_frame_vector_args (frame);
-  n_left_from = frame->n_vectors;
-  next = node->cached_next_index;
-
-  while (n_left_from > 0)
-    {
-      vlib_get_next_frame (vm, node, next, to_next, n_left_to_next);
-
-      while (n_left_from >= 4 && n_left_to_next >= 2)
-	{
-	  vlib_buffer_t *p0, *p1;
-	  u32 pi0, pi1, lb_index0, lb_index1, wrong_next;
-	  ip_lookup_next_t next0, next1;
-	  ip4_header_t *ip0, *ip1;
-	  u32 fib_index0, fib_index1;
-	  const dpo_id_t *dpo0, *dpo1;
-	  const load_balance_t *lb0, *lb1;
-
-	  /* Prefetch next iteration. */
-	  {
-	    vlib_buffer_t *p2, *p3;
-
-	    p2 = vlib_get_buffer (vm, from[2]);
-	    p3 = vlib_get_buffer (vm, from[3]);
-
-	    vlib_prefetch_buffer_header (p2, LOAD);
-	    vlib_prefetch_buffer_header (p3, LOAD);
-
-	    CLIB_PREFETCH (p2->data, sizeof (ip0[0]), LOAD);
-	    CLIB_PREFETCH (p3->data, sizeof (ip0[0]), LOAD);
-	  }
-
-	  pi0 = to_next[0] = from[0];
-	  pi1 = to_next[1] = from[1];
-
-	  p0 = vlib_get_buffer (vm, pi0);
-	  p1 = vlib_get_buffer (vm, pi1);
-
-	  ip0 = vlib_buffer_get_current (p0);
-	  ip1 = vlib_buffer_get_current (p1);
-
-	  fib_index0 =
-	    vec_elt (im->fib_index_by_sw_if_index,
-		     vnet_buffer (p0)->sw_if_index[VLIB_RX]);
-	  fib_index1 =
-	    vec_elt (im->fib_index_by_sw_if_index,
-		     vnet_buffer (p1)->sw_if_index[VLIB_RX]);
-	  fib_index0 =
-	    (vnet_buffer (p0)->sw_if_index[VLIB_TX] ==
-	     (u32) ~ 0) ? fib_index0 : vnet_buffer (p0)->sw_if_index[VLIB_TX];
-	  fib_index1 =
-	    (vnet_buffer (p1)->sw_if_index[VLIB_TX] ==
-	     (u32) ~ 0) ? fib_index1 : vnet_buffer (p1)->sw_if_index[VLIB_TX];
-
-	  lb_index0 = ip4_fib_table_lookup_lb (ip4_fib_get (fib_index0),
-					       &ip0->dst_address);
-	  lb_index1 = ip4_fib_table_lookup_lb (ip4_fib_get (fib_index1),
-					       &ip1->dst_address);
-
-	  lb0 = load_balance_get (lb_index0);
-	  lb1 = load_balance_get (lb_index1);
-
-	  ASSERT (lb0->lb_n_buckets > 0);
-	  ASSERT (is_pow2 (lb0->lb_n_buckets));
-	  ASSERT (lb1->lb_n_buckets > 0);
-	  ASSERT (is_pow2 (lb1->lb_n_buckets));
-
-	  vnet_buffer (p0)->ip.flow_hash = ip4_compute_flow_hash
-	    (ip0, lb0->lb_hash_config);
-
-	  vnet_buffer (p1)->ip.flow_hash = ip4_compute_flow_hash
-	    (ip1, lb1->lb_hash_config);
-
-	  dpo0 = load_balance_get_bucket_i (lb0,
-					    (vnet_buffer (p0)->ip.flow_hash &
-					     (lb0->lb_n_buckets_minus_1)));
-	  dpo1 = load_balance_get_bucket_i (lb1,
-					    (vnet_buffer (p1)->ip.flow_hash &
-					     (lb1->lb_n_buckets_minus_1)));
-
-	  next0 = dpo0->dpoi_next_node;
-	  vnet_buffer (p0)->ip.adj_index[VLIB_TX] = dpo0->dpoi_index;
-	  next1 = dpo1->dpoi_next_node;
-	  vnet_buffer (p1)->ip.adj_index[VLIB_TX] = dpo1->dpoi_index;
-
-	  if (1)		/* $$$$$$ HACK FIXME */
-	    vlib_increment_combined_counter
-	      (cm, cpu_index, lb_index0, 1,
-	       vlib_buffer_length_in_chain (vm, p0));
-	  if (1)		/* $$$$$$ HACK FIXME */
-	    vlib_increment_combined_counter
-	      (cm, cpu_index, lb_index1, 1,
-	       vlib_buffer_length_in_chain (vm, p1));
-
-	  from += 2;
-	  to_next += 2;
-	  n_left_to_next -= 2;
-	  n_left_from -= 2;
-
-	  wrong_next = (next0 != next) + 2 * (next1 != next);
-	  if (PREDICT_FALSE (wrong_next != 0))
-	    {
-	      switch (wrong_next)
-		{
-		case 1:
-		  /* A B A */
-		  to_next[-2] = pi1;
-		  to_next -= 1;
-		  n_left_to_next += 1;
-		  vlib_set_next_frame_buffer (vm, node, next0, pi0);
-		  break;
-
-		case 2:
-		  /* A A B */
-		  to_next -= 1;
-		  n_left_to_next += 1;
-		  vlib_set_next_frame_buffer (vm, node, next1, pi1);
-		  break;
-
-		case 3:
-		  /* A B C */
-		  to_next -= 2;
-		  n_left_to_next += 2;
-		  vlib_set_next_frame_buffer (vm, node, next0, pi0);
-		  vlib_set_next_frame_buffer (vm, node, next1, pi1);
-		  if (next0 == next1)
-		    {
-		      /* A B B */
-		      vlib_put_next_frame (vm, node, next, n_left_to_next);
-		      next = next1;
-		      vlib_get_next_frame (vm, node, next, to_next,
-					   n_left_to_next);
-		    }
-		}
-	    }
-	}
-
-      while (n_left_from > 0 && n_left_to_next > 0)
-	{
-	  vlib_buffer_t *p0;
-	  ip4_header_t *ip0;
-	  u32 pi0, lb_index0;
-	  ip_lookup_next_t next0;
-	  u32 fib_index0;
-	  const dpo_id_t *dpo0;
-	  const load_balance_t *lb0;
-
-	  pi0 = from[0];
-	  to_next[0] = pi0;
-
-	  p0 = vlib_get_buffer (vm, pi0);
-
-	  ip0 = vlib_buffer_get_current (p0);
-
-	  fib_index0 = vec_elt (im->fib_index_by_sw_if_index,
-				vnet_buffer (p0)->sw_if_index[VLIB_RX]);
-	  fib_index0 = (vnet_buffer (p0)->sw_if_index[VLIB_TX] == (u32) ~ 0) ?
-	    fib_index0 : vnet_buffer (p0)->sw_if_index[VLIB_TX];
-
-	  lb_index0 = ip4_fib_table_lookup_lb (ip4_fib_get (fib_index0),
-					       &ip0->dst_address);
-
-	  lb0 = load_balance_get (lb_index0);
-
-	  ASSERT (lb0->lb_n_buckets > 0);
-	  ASSERT (is_pow2 (lb0->lb_n_buckets));
-
-	  vnet_buffer (p0)->ip.flow_hash = ip4_compute_flow_hash
-	    (ip0, lb0->lb_hash_config);
-
-	  dpo0 = load_balance_get_bucket_i (lb0,
-					    (vnet_buffer (p0)->ip.flow_hash &
-					     (lb0->lb_n_buckets_minus_1)));
-
-	  next0 = dpo0->dpoi_next_node;
-	  vnet_buffer (p0)->ip.adj_index[VLIB_TX] = dpo0->dpoi_index;
-
-	  if (1)		/* $$$$$$ HACK FIXME */
-	    vlib_increment_combined_counter
-	      (cm, cpu_index, lb_index0, 1,
-	       vlib_buffer_length_in_chain (vm, p0));
-
-	  from += 1;
-	  to_next += 1;
-	  n_left_to_next -= 1;
-	  n_left_from -= 1;
-
-	  if (PREDICT_FALSE (next0 != next))
-	    {
-	      n_left_to_next += 1;
-	      vlib_put_next_frame (vm, node, next, n_left_to_next);
-	      next = next0;
-	      vlib_get_next_frame (vm, node, next, to_next, n_left_to_next);
-	      to_next[0] = pi0;
-	      to_next += 1;
-	      n_left_to_next -= 1;
-	    }
-	}
-
-      vlib_put_next_frame (vm, node, next, n_left_to_next);
-    }
-
-  if (node->flags & VLIB_NODE_FLAG_TRACE)
-    ip4_forward_next_trace (vm, node, frame, VLIB_TX);
-
-  return frame->n_vectors;
-}
-
-VLIB_REGISTER_NODE (ip4_lookup_multicast_node, static) =
-{
-.function = ip4_lookup_multicast,.name =
-    "ip4-lookup-multicast",.vector_size = sizeof (u32),.sibling_of =
-    "ip4-lookup",.format_trace = format_ip4_lookup_trace,.n_next_nodes = 0,};
-
-VLIB_NODE_FUNCTION_MULTIARCH (ip4_lookup_multicast_node,
-			      ip4_lookup_multicast);
-
-VLIB_REGISTER_NODE (ip4_multicast_node, static) =
-{
-  .function = ip4_drop,.name = "ip4-multicast",.vector_size =
-    sizeof (u32),.format_trace = format_ip4_forward_next_trace,.n_next_nodes =
-    1,.next_nodes =
-  {
-  [0] = "error-drop",}
-,};
-
 int
 ip4_lookup_validate (ip4_address_t * a, u32 fib_index0)
 {
diff --git a/src/vnet/ip/ip4_input.c b/src/vnet/ip/ip4_input.c
index 1cf5e0b8..ba200a9f 100644
--- a/src/vnet/ip/ip4_input.c
+++ b/src/vnet/ip/ip4_input.c
@@ -426,7 +426,7 @@ VLIB_REGISTER_NODE (ip4_input_node) = {
     [IP4_INPUT_NEXT_DROP] = "error-drop",
     [IP4_INPUT_NEXT_PUNT] = "error-punt",
     [IP4_INPUT_NEXT_LOOKUP] = "ip4-lookup",
-    [IP4_INPUT_NEXT_LOOKUP_MULTICAST] = "ip4-lookup-multicast",
+    [IP4_INPUT_NEXT_LOOKUP_MULTICAST] = "ip4-mfib-forward-lookup",
     [IP4_INPUT_NEXT_ICMP_ERROR] = "ip4-icmp-error",
   },
 
@@ -448,7 +448,7 @@ VLIB_REGISTER_NODE (ip4_input_no_checksum_node,static) = {
     [IP4_INPUT_NEXT_DROP] = "error-drop",
     [IP4_INPUT_NEXT_PUNT] = "error-punt",
     [IP4_INPUT_NEXT_LOOKUP] = "ip4-lookup",
-    [IP4_INPUT_NEXT_LOOKUP_MULTICAST] = "ip4-lookup-multicast",
+    [IP4_INPUT_NEXT_LOOKUP_MULTICAST] = "ip4-mfib-forward-lookup",
     [IP4_INPUT_NEXT_ICMP_ERROR] = "ip4-icmp-error",
   },
 
diff --git a/src/vnet/ip/ip6.h b/src/vnet/ip/ip6.h
index f493db01..6fecd42d 100644
--- a/src/vnet/ip/ip6.h
+++ b/src/vnet/ip/ip6.h
@@ -49,6 +49,7 @@
 #include <stdbool.h>
 #include <vppinfra/bihash_24_8.h>
 #include <vppinfra/bihash_template.h>
+#include <vnet/util/radix.h>
 
 /*
  * Default size of the ip6 fib hash table
@@ -75,6 +76,21 @@ typedef struct
   flow_hash_config_t flow_hash_config;
 } ip6_fib_t;
 
+typedef struct ip6_mfib_t
+{
+  /* Table ID (hash key) for this FIB. */
+  u32 table_id;
+
+  /* Index into FIB vector. */
+  u32 index;
+
+  /*
+   *  Pointer to the top of a radix tree.
+   * This cannot be realloc'd, hence it cannot be inlined with this table
+   */
+  struct radix_node_head *rhead;
+} ip6_mfib_t;
+
 struct ip6_main_t;
 
 typedef void (ip6_add_del_interface_address_function_t)
@@ -137,12 +153,18 @@ typedef struct ip6_main_t
   /* Pool of FIBs. */
   struct fib_table_t_ *fibs;
 
+  /** Vector of MFIBs. */
+  struct mfib_table_t_ *mfibs;
+
   /* Network byte orders subnet mask for each prefix length */
   ip6_address_t fib_masks[129];
 
   /* Table index indexed by software interface. */
   u32 *fib_index_by_sw_if_index;
 
+  /** Table index indexed by software interface. */
+  u32 *mfib_index_by_sw_if_index;
+
   /* IP6 enabled count by software interface */
   u8 *ip_enabled_by_sw_if_index;
 
@@ -150,6 +172,10 @@ typedef struct ip6_main_t
      ID space is not necessarily dense; index space is dense. */
   uword *fib_index_by_table_id;
 
+  /** Hash table mapping table id to multicast fib index.
+     ID space is not necessarily dense; index space is dense. */
+  uword *mfib_index_by_table_id;
+
   /* Hash table mapping interface rewrite adjacency index by sw if index. */
   uword *interface_route_adj_index_by_sw_if_index;
 
@@ -185,6 +211,7 @@ extern ip6_main_t ip6_main;
 /* Global ip6 input node.  Errors get attached to ip6 input node. */
 extern vlib_node_registration_t ip6_input_node;
 extern vlib_node_registration_t ip6_rewrite_node;
+extern vlib_node_registration_t ip6_rewrite_mcast_node;
 extern vlib_node_registration_t ip6_rewrite_local_node;
 extern vlib_node_registration_t ip6_discover_neighbor_node;
 extern vlib_node_registration_t ip6_glean_node;
diff --git a/src/vnet/ip/ip6_forward.c b/src/vnet/ip/ip6_forward.c
index 232f7283..ac47b3ad 100644
--- a/src/vnet/ip/ip6_forward.c
+++ b/src/vnet/ip/ip6_forward.c
@@ -42,8 +42,8 @@
 #include <vnet/ethernet/ethernet.h>	/* for ethernet_header_t */
 #include <vnet/srp/srp.h>	/* for srp_hw_interface_class */
 #include <vppinfra/cache.h>
-#include <vnet/fib/fib_table.h>
 #include <vnet/fib/ip6_fib.h>
+#include <vnet/mfib/ip6_mfib.h>
 #include <vnet/dpo/load_balance.h>
 #include <vnet/dpo/classify_dpo.h>
 
@@ -411,11 +411,14 @@ ip6_sw_interface_enable_disable (u32 sw_if_index, u32 is_enable)
 	return;
     }
 
+  if (sw_if_index != 0)
+    ip6_mfib_interface_enable_disable (sw_if_index, is_enable);
+
   vnet_feature_enable_disable ("ip6-unicast", "ip6-lookup", sw_if_index,
 			       is_enable, 0, 0);
 
-  vnet_feature_enable_disable ("ip6-multicast", "ip6-lookup", sw_if_index,
-			       is_enable, 0, 0);
+  vnet_feature_enable_disable ("ip6-multicast", "ip6-mfib-forward-lookup",
+			       sw_if_index, is_enable, 0, 0);
 
 }
 
@@ -457,6 +460,8 @@ ip6_add_del_interface_address (vlib_main_t * vm,
   ip6_address_fib_t ip6_af, *addr_fib = 0;
 
   vec_validate (im->fib_index_by_sw_if_index, sw_if_index);
+  vec_validate (im->mfib_index_by_sw_if_index, sw_if_index);
+
   ip6_addr_fib_init (&ip6_af, address,
 		     vec_elt (im->fib_index_by_sw_if_index, sw_if_index));
   vec_add1 (addr_fib, ip6_af);
@@ -611,12 +616,12 @@ VNET_FEATURE_ARC_INIT (ip6_multicast, static) =
 VNET_FEATURE_INIT (ip6_vpath_mc, static) = {
   .arc_name = "ip6-multicast",
   .node_name = "vpath-input-ip6",
-  .runs_before = VNET_FEATURES ("ip6-lookup"),
+  .runs_before = VNET_FEATURES ("ip6-mfib-forward-lookup"),
 };
 
 VNET_FEATURE_INIT (ip6_mc_lookup, static) = {
   .arc_name = "ip6-multicast",
-  .node_name = "ip6-lookup",
+  .node_name = "ip6-mfib-forward-lookup",
   .runs_before = VNET_FEATURES ("ip6-drop"),
 };
 
@@ -1122,22 +1127,6 @@ VLIB_REGISTER_NODE (ip6_punt_node, static) =
 
 VLIB_NODE_FUNCTION_MULTIARCH (ip6_punt_node, ip6_punt);
 
-/* *INDENT-OFF* */
-VLIB_REGISTER_NODE (ip6_multicast_node, static) =
-{
-  .function = ip6_drop,
-  .name = "ip6-multicast",
-  .vector_size = sizeof (u32),
-  .format_trace = format_ip6_forward_next_trace,
-  .n_next_nodes = 1,
-  .next_nodes =
-  {
-    [0] = "error-drop",
-  },
-};
-
-/* *INDENT-ON* */
-
 /* Compute TCP/UDP/ICMP6 checksum in software. */
 u16
 ip6_tcp_udp_icmp_compute_checksum (vlib_main_t * vm, vlib_buffer_t * p0,
@@ -1977,7 +1966,7 @@ typedef enum
 always_inline uword
 ip6_rewrite_inline (vlib_main_t * vm,
 		    vlib_node_runtime_t * node,
-		    vlib_frame_t * frame, int is_midchain)
+		    vlib_frame_t * frame, int is_midchain, int is_mcast)
 {
   ip_lookup_main_t *lm = &ip6_main.lookup_main;
   u32 *from = vlib_frame_vector_args (frame);
@@ -2165,6 +2154,14 @@ ip6_rewrite_inline (vlib_main_t * vm,
 	      adj0->sub_type.midchain.fixup_func (vm, adj0, p0);
 	      adj1->sub_type.midchain.fixup_func (vm, adj1, p1);
 	    }
+	  if (is_mcast)
+	    {
+	      /*
+	       * copy bytes from the IP address into the MAC rewrite
+	       */
+	      vnet_fixup_one_header (adj0[0], &ip0->dst_address, ip0, 0);
+	      vnet_fixup_one_header (adj1[0], &ip1->dst_address, ip1, 0);
+	    }
 
 	  vlib_validate_buffer_enqueue_x2 (vm, node, next_index,
 					   to_next, n_left_to_next,
@@ -2265,6 +2262,10 @@ ip6_rewrite_inline (vlib_main_t * vm,
 	    {
 	      adj0->sub_type.midchain.fixup_func (vm, adj0, p0);
 	    }
+	  if (is_mcast)
+	    {
+	      vnet_fixup_one_header (adj0[0], &ip0->dst_address, ip0, 0);
+	    }
 
 	  p0->error = error_node->errors[error0];
 
@@ -2292,16 +2293,21 @@ static uword
 ip6_rewrite (vlib_main_t * vm,
 	     vlib_node_runtime_t * node, vlib_frame_t * frame)
 {
-  return ip6_rewrite_inline (vm, node, frame,
-			     /* midchain */ 0);
+  return ip6_rewrite_inline (vm, node, frame, 0, 0);
+}
+
+static uword
+ip6_rewrite_mcast (vlib_main_t * vm,
+		   vlib_node_runtime_t * node, vlib_frame_t * frame)
+{
+  return ip6_rewrite_inline (vm, node, frame, 0, 1);
 }
 
 static uword
 ip6_midchain (vlib_main_t * vm,
 	      vlib_node_runtime_t * node, vlib_frame_t * frame)
 {
-  return ip6_rewrite_inline (vm, node, frame,
-			     /* midchain */ 1);
+  return ip6_rewrite_inline (vm, node, frame, 1, 0);
 }
 
 /* *INDENT-OFF* */
@@ -2335,10 +2341,22 @@ VLIB_REGISTER_NODE (ip6_rewrite_node) =
 
 VLIB_NODE_FUNCTION_MULTIARCH (ip6_rewrite_node, ip6_rewrite);
 
+/* *INDENT-OFF* */
+VLIB_REGISTER_NODE (ip6_rewrite_mcast_node) =
+{
+  .function = ip6_rewrite_mcast,
+  .name = "ip6-rewrite-mcast",
+  .vector_size = sizeof (u32),
+  .format_trace = format_ip6_rewrite_trace,
+  .sibling_of = "ip6-rewrite",
+};
+/* *INDENT-ON* */
+
+VLIB_NODE_FUNCTION_MULTIARCH (ip6_rewrite_mcast_node, ip6_rewrite_mcast);
+
 /*
  * Hop-by-Hop handling
  */
-
 ip6_hop_by_hop_main_t ip6_hop_by_hop_main;
 
 #define foreach_ip6_hop_by_hop_error \
@@ -2346,13 +2364,15 @@ _(PROCESSED, "pkts with ip6 hop-by-hop options") \
 _(FORMAT, "incorrectly formatted hop-by-hop options") \
 _(UNKNOWN_OPTION, "unknown ip6 hop-by-hop options")
 
+/* *INDENT-OFF* */
 typedef enum
 {
 #define _(sym,str) IP6_HOP_BY_HOP_ERROR_##sym,
   foreach_ip6_hop_by_hop_error
 #undef _
-    IP6_HOP_BY_HOP_N_ERROR,
+  IP6_HOP_BY_HOP_N_ERROR,
 } ip6_hop_by_hop_error_t;
+/* *INDENT-ON* */
 
 /*
  * Primary h-b-h handler trace support
@@ -2878,6 +2898,7 @@ ip6_lookup_init (vlib_main_t * vm)
 
   /* Create FIB with index 0 and table id of 0. */
   fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6, 0);
+  mfib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6, 0);
 
   {
     pg_node_t *pn;
@@ -2955,6 +2976,12 @@ add_del_ip6_interface_table (vlib_main_t * vm,
 
     vec_validate (ip6_main.fib_index_by_sw_if_index, sw_if_index);
     ip6_main.fib_index_by_sw_if_index[sw_if_index] = fib_index;
+
+    fib_index = mfib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6,
+						    table_id);
+
+    vec_validate (ip6_main.mfib_index_by_sw_if_index, sw_if_index);
+    ip6_main.mfib_index_by_sw_if_index[sw_if_index] = fib_index;
   }
 
 
diff --git a/src/vnet/ip/ip6_input.c b/src/vnet/ip/ip6_input.c
index bbc2ceba..20306088 100644
--- a/src/vnet/ip/ip6_input.c
+++ b/src/vnet/ip/ip6_input.c
@@ -64,6 +64,7 @@ typedef enum
 {
   IP6_INPUT_NEXT_DROP,
   IP6_INPUT_NEXT_LOOKUP,
+  IP6_INPUT_NEXT_LOOKUP_MULTICAST,
   IP6_INPUT_NEXT_ICMP_ERROR,
   IP6_INPUT_N_NEXT,
 } ip6_input_next_t;
@@ -142,12 +143,27 @@ ip6_input (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame)
 	  sw_if_index0 = vnet_buffer (p0)->sw_if_index[VLIB_RX];
 	  sw_if_index1 = vnet_buffer (p1)->sw_if_index[VLIB_RX];
 
-	  arc0 =
-	    ip6_address_is_multicast (&ip0->dst_address) ?
-	    lm->mcast_feature_arc_index : lm->ucast_feature_arc_index;
-	  arc1 =
-	    ip6_address_is_multicast (&ip1->dst_address) ?
-	    lm->mcast_feature_arc_index : lm->ucast_feature_arc_index;
+	  if (PREDICT_FALSE (ip6_address_is_multicast (&ip0->dst_address)))
+	    {
+	      arc0 = lm->mcast_feature_arc_index;
+	      next0 = IP6_INPUT_NEXT_LOOKUP_MULTICAST;
+	    }
+	  else
+	    {
+	      arc0 = lm->ucast_feature_arc_index;
+	      next0 = IP6_INPUT_NEXT_LOOKUP;
+	    }
+
+	  if (PREDICT_FALSE (ip6_address_is_multicast (&ip1->dst_address)))
+	    {
+	      arc1 = lm->mcast_feature_arc_index;
+	      next1 = IP6_INPUT_NEXT_LOOKUP_MULTICAST;
+	    }
+	  else
+	    {
+	      arc1 = lm->ucast_feature_arc_index;
+	      next1 = IP6_INPUT_NEXT_LOOKUP;
+	    }
 
 	  vnet_buffer (p0)->ip.adj_index[VLIB_RX] = ~0;
 	  vnet_buffer (p1)->ip.adj_index[VLIB_RX] = ~0;
@@ -240,9 +256,17 @@ ip6_input (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame)
 	  ip0 = vlib_buffer_get_current (p0);
 
 	  sw_if_index0 = vnet_buffer (p0)->sw_if_index[VLIB_RX];
-	  arc0 =
-	    ip6_address_is_multicast (&ip0->dst_address) ?
-	    lm->mcast_feature_arc_index : lm->ucast_feature_arc_index;
+	  if (PREDICT_FALSE (ip6_address_is_multicast (&ip0->dst_address)))
+	    {
+	      arc0 = lm->mcast_feature_arc_index;
+	      next0 = IP6_INPUT_NEXT_LOOKUP_MULTICAST;
+	    }
+	  else
+	    {
+	      arc0 = lm->ucast_feature_arc_index;
+	      next0 = IP6_INPUT_NEXT_LOOKUP;
+	    }
+
 	  vnet_buffer (p0)->ip.adj_index[VLIB_RX] = ~0;
 	  vnet_feature_arc_start (arc0, sw_if_index0, &next0, p0);
 
@@ -313,6 +337,7 @@ VLIB_REGISTER_NODE (ip6_input_node) = {
     [IP6_INPUT_NEXT_DROP] = "error-drop",
     [IP6_INPUT_NEXT_LOOKUP] = "ip6-lookup",
     [IP6_INPUT_NEXT_ICMP_ERROR] = "ip6-icmp-error",
+    [IP6_INPUT_NEXT_LOOKUP_MULTICAST] = "ip6-mfib-forward-lookup",
   },
 
   .format_buffer = format_ip6_header,
diff --git a/src/vnet/ip/ip6_neighbor.c b/src/vnet/ip/ip6_neighbor.c
index 46c0e316..46d04769 100644
--- a/src/vnet/ip/ip6_neighbor.c
+++ b/src/vnet/ip/ip6_neighbor.c
@@ -21,6 +21,7 @@
 #include <vppinfra/mhash.h>
 #include <vppinfra/md5.h>
 #include <vnet/adj/adj.h>
+#include <vnet/adj/adj_mcast.h>
 #include <vnet/fib/fib_table.h>
 #include <vnet/fib/ip6_fib.h>
 
@@ -116,9 +117,7 @@ typedef struct
   u32 seed;
   u64 randomizer;
   int ref_count;
-  adj_index_t all_nodes_adj_index;
-  adj_index_t all_routers_adj_index;
-  adj_index_t all_mldv2_routers_adj_index;
+  adj_index_t mcast_adj_index;
 
   /* timing information */
 #define DEF_MAX_RADV_INTERVAL 200
@@ -474,33 +473,72 @@ ip6_ethernet_update_adjacency (vnet_main_t * vnm, u32 sw_if_index, u32 ai)
 
   nbr = ip6_nd_find (sw_if_index, &adj->sub_type.nbr.next_hop.ip6);
 
-  if (NULL != nbr)
-    {
-      adj_nbr_walk_nh6 (sw_if_index, &nbr->key.ip6_address,
-			ip6_nd_mk_complete_walk, nbr);
-    }
-  else
+  switch (adj->lookup_next_index)
     {
+    case IP_LOOKUP_NEXT_ARP:
+    case IP_LOOKUP_NEXT_GLEAN:
+      if (NULL != nbr)
+	{
+	  adj_nbr_walk_nh6 (sw_if_index, &nbr->key.ip6_address,
+			    ip6_nd_mk_complete_walk, nbr);
+	}
+      else
+	{
+	  /*
+	   * no matching ND entry.
+	   * construct the rewrite required to for an ND packet, and stick
+	   * that in the adj's pipe to smoke.
+	   */
+	  adj_nbr_update_rewrite (ai,
+				  ADJ_NBR_REWRITE_FLAG_INCOMPLETE,
+				  ethernet_build_rewrite (vnm,
+							  sw_if_index,
+							  VNET_LINK_IP6,
+							  VNET_REWRITE_FOR_SW_INTERFACE_ADDRESS_BROADCAST));
+
+	  /*
+	   * since the FIB has added this adj for a route, it makes sense it may
+	   * want to forward traffic sometime soon. Let's send a speculative ND.
+	   * just one. If we were to do periodically that wouldn't be bad either,
+	   * but that's more code than i'm prepared to write at this time for
+	   * relatively little reward.
+	   */
+	  ip6_nbr_probe (adj);
+	}
+      break;
+    case IP_LOOKUP_NEXT_MCAST:
       /*
-       * no matching ND entry.
-       * construct the rewrite required to for an ND packet, and stick
-       * that in the adj's pipe to smoke.
+       * Construct a partial rewrite from the known ethernet mcast dest MAC
        */
-      adj_nbr_update_rewrite (ai,
-			      ADJ_NBR_REWRITE_FLAG_INCOMPLETE,
-			      ethernet_build_rewrite (vnm,
-						      sw_if_index,
-						      VNET_LINK_IP6,
-						      VNET_REWRITE_FOR_SW_INTERFACE_ADDRESS_BROADCAST));
+      adj_mcast_update_rewrite
+	(ai,
+	 ethernet_build_rewrite (vnm,
+				 sw_if_index,
+				 adj->ia_link,
+				 ethernet_ip6_mcast_dst_addr ()));
 
       /*
-       * since the FIB has added this adj for a route, it makes sense it may
-       * want to forward traffic sometime soon. Let's send a speculative ND.
-       * just one. If we were to do periodically that wouldn't be bad either,
-       * but that's more code than i'm prepared to write at this time for
-       * relatively little reward.
+       * Complete the remaining fields of the adj's rewrite to direct the
+       * complete of the rewrite at switch time by copying in the IP
+       * dst address's bytes.
+       * Ofset is 12 bytes from the end of the MAC header - which is 2
+       * bytes into the desintation address. And we write 4 bytes.
        */
-      ip6_nbr_probe (adj);
+      adj->rewrite_header.dst_mcast_offset = 12;
+      adj->rewrite_header.dst_mcast_n_bytes = 4;
+
+      break;
+
+    case IP_LOOKUP_NEXT_DROP:
+    case IP_LOOKUP_NEXT_PUNT:
+    case IP_LOOKUP_NEXT_LOCAL:
+    case IP_LOOKUP_NEXT_REWRITE:
+    case IP_LOOKUP_NEXT_LOAD_BALANCE:
+    case IP_LOOKUP_NEXT_MIDCHAIN:
+    case IP_LOOKUP_NEXT_ICMP_ERROR:
+    case IP_LOOKUP_N_NEXT:
+      ASSERT (0);
+      break;
     }
 }
 
@@ -1517,7 +1555,7 @@ icmp6_router_solicitation (vlib_main_t * vm,
 			}
 		      else
 			{
-			  adj_index0 = radv_info->all_nodes_adj_index;
+			  adj_index0 = radv_info->mcast_adj_index;
 			  if (adj_index0 == 0)
 			    error0 = ICMP6_ERROR_DST_LOOKUP_MISS;
 			  else
@@ -1918,10 +1956,8 @@ ip6_neighbor_sw_interface_add_del (vnet_main_t * vnm,
 	  ip6_radv_prefix_t *p;
 	  ip6_mldp_group_t *m;
 
-	  /* remove adjacencies */
-	  adj_unlock (a->all_nodes_adj_index);
-	  adj_unlock (a->all_routers_adj_index);
-	  adj_unlock (a->all_mldv2_routers_adj_index);
+	  /* release the lock on the interface's mcast adj */
+	  adj_unlock (a->mcast_adj_index);
 
 	  /* clean up prefix_pool */
 	  /* *INDENT-OFF* */
@@ -2017,36 +2053,9 @@ ip6_neighbor_sw_interface_add_del (vnet_main_t * vnm,
 	  mhash_init (&a->address_to_mldp_index, sizeof (uword),
 		      sizeof (ip6_address_t));
 
-	  {
-	    u8 link_layer_address[6] = { 0x33, 0x33, 0x00, 0x00, 0x00,
-	      IP6_MULTICAST_GROUP_ID_all_hosts
-	    };
-
-	    a->all_nodes_adj_index =
-	      adj_rewrite_add_and_lock (FIB_PROTOCOL_IP6, VNET_LINK_IP6,
-					sw_if_index, link_layer_address);
-	  }
-
-	  {
-	    u8 link_layer_address[6] = { 0x33, 0x33, 0x00, 0x00, 0x00,
-	      IP6_MULTICAST_GROUP_ID_all_routers
-	    };
-
-	    a->all_routers_adj_index =
-	      adj_rewrite_add_and_lock (FIB_PROTOCOL_IP6, VNET_LINK_IP6,
-					sw_if_index, link_layer_address);
-	  }
-
-	  {
-	    u8 link_layer_address[6] = { 0x33, 0x33, 0x00, 0x00, 0x00,
-	      IP6_MULTICAST_GROUP_ID_mldv2_routers
-	    };
-
-	    a->all_mldv2_routers_adj_index =
-	      adj_rewrite_add_and_lock (FIB_PROTOCOL_IP6,
-					VNET_LINK_IP6,
-					sw_if_index, link_layer_address);
-	  }
+	  a->mcast_adj_index = adj_mcast_add_or_lock (FIB_PROTOCOL_IP6,
+						      VNET_LINK_IP6,
+						      sw_if_index);
 
 	  /* add multicast groups we will always be reporting  */
 	  ip6_address_t addr;
@@ -2273,11 +2282,10 @@ ip6_neighbor_send_mldpv2_report (u32 sw_if_index)
   vnet_buffer (b0)->sw_if_index[VLIB_RX] =
     vnet_main.local_interface_sw_if_index;
 
-  vnet_buffer (b0)->ip.adj_index[VLIB_TX] =
-    radv_info->all_mldv2_routers_adj_index;
+  vnet_buffer (b0)->ip.adj_index[VLIB_TX] = radv_info->mcast_adj_index;
   b0->flags |= VNET_BUFFER_LOCALLY_ORIGINATED;
 
-  vlib_node_t *node = vlib_get_node_by_name (vm, (u8 *) "ip6-rewrite");
+  vlib_node_t *node = vlib_get_node_by_name (vm, (u8 *) "ip6-rewrite-mcast");
 
   f = vlib_get_frame_to_node (vm, node->index);
   to_next = vlib_frame_vector_args (f);
@@ -2301,7 +2309,7 @@ VLIB_REGISTER_NODE (ip6_icmp_router_solicitation_node,static) =
   .n_next_nodes = ICMP6_ROUTER_SOLICITATION_N_NEXT,
   .next_nodes = {
     [ICMP6_ROUTER_SOLICITATION_NEXT_DROP] = "error-drop",
-    [ICMP6_ROUTER_SOLICITATION_NEXT_REPLY_RW] = "ip6-rewrite",
+    [ICMP6_ROUTER_SOLICITATION_NEXT_REPLY_RW] = "ip6-rewrite-mcast",
     [ICMP6_ROUTER_SOLICITATION_NEXT_REPLY_TX] = "interface-output",
   },
 };
diff --git a/src/vnet/ip/ip_api.c b/src/vnet/ip/ip_api.c
index aafde464..437d2674 100644
--- a/src/vnet/ip/ip_api.c
+++ b/src/vnet/ip/ip_api.c
@@ -33,6 +33,9 @@
 #include <vnet/dpo/classify_dpo.h>
 #include <vnet/dpo/ip_null_dpo.h>
 #include <vnet/ethernet/arp_packet.h>
+//#include <vnet/mfib/ip6_mfib.h>
+#include <vnet/mfib/ip4_mfib.h>
+#include <vnet/mfib/mfib_signal.h>
 
 #include <vnet/vnet_msg_enum.h>
 
@@ -58,6 +61,8 @@ _(IP_FIB_DETAILS, ip_fib_details)                                       \
 _(IP6_FIB_DUMP, ip6_fib_dump)                                           \
 _(IP6_FIB_DETAILS, ip6_fib_details)                                     \
 _(IP_NEIGHBOR_DUMP, ip_neighbor_dump)                                   \
+_(IP_MROUTE_ADD_DEL, ip_mroute_add_del)                                 \
+_(MFIB_SIGNAL_DUMP, mfib_signal_dump)                                    \
 _(IP_NEIGHBOR_DETAILS, ip_neighbor_details)                             \
 _(IP_ADDRESS_DUMP, ip_address_dump)                                     \
 _(IP_DUMP, ip_dump)                                                     \
@@ -845,6 +850,144 @@ vl_api_ip_add_del_route_t_handler (vl_api_ip_add_del_route_t * mp)
   REPLY_MACRO (VL_API_IP_ADD_DEL_ROUTE_REPLY);
 }
 
+static int
+add_del_mroute_check (fib_protocol_t table_proto,
+		      u32 table_id,
+		      u32 next_hop_sw_if_index,
+		      u8 is_local, u8 create_missing_tables, u32 * fib_index)
+{
+  vnet_main_t *vnm = vnet_get_main ();
+
+  *fib_index = mfib_table_find (table_proto, ntohl (table_id));
+  if (~0 == *fib_index)
+    {
+      if (create_missing_tables)
+	{
+	  *fib_index = mfib_table_find_or_create_and_lock (table_proto,
+							   ntohl (table_id));
+	}
+      else
+	{
+	  /* No such VRF, and we weren't asked to create one */
+	  return VNET_API_ERROR_NO_SUCH_FIB;
+	}
+    }
+
+  if (~0 != ntohl (next_hop_sw_if_index))
+    {
+      if (pool_is_free_index (vnm->interface_main.sw_interfaces,
+			      ntohl (next_hop_sw_if_index)))
+	{
+	  return VNET_API_ERROR_NO_MATCHING_INTERFACE;
+	}
+    }
+
+  return (0);
+}
+
+static int
+mroute_add_del_handler (u8 is_add,
+			u8 is_local,
+			u32 fib_index,
+			const mfib_prefix_t * prefix,
+			u32 entry_flags,
+			u32 next_hop_sw_if_index, u32 itf_flags)
+{
+  stats_dslock_with_hint (1 /* release hint */ , 2 /* tag */ );
+
+  fib_route_path_t path = {
+    .frp_sw_if_index = next_hop_sw_if_index,
+    .frp_proto = prefix->fp_proto,
+  };
+
+  if (is_local)
+    path.frp_flags |= FIB_ROUTE_PATH_LOCAL;
+
+
+  if (!is_local && ~0 == next_hop_sw_if_index)
+    {
+      mfib_table_entry_update (fib_index, prefix,
+			       MFIB_SOURCE_API, entry_flags);
+    }
+  else
+    {
+      if (is_add)
+	{
+	  mfib_table_entry_path_update (fib_index, prefix,
+					MFIB_SOURCE_API, &path, itf_flags);
+	}
+      else
+	{
+	  mfib_table_entry_path_remove (fib_index, prefix,
+					MFIB_SOURCE_API, &path);
+	}
+    }
+
+  stats_dsunlock ();
+  return (0);
+}
+
+static int
+api_mroute_add_del_t_handler (vl_api_ip_mroute_add_del_t * mp)
+{
+  fib_protocol_t fproto;
+  u32 fib_index;
+  int rv;
+
+  fproto = (mp->is_ipv6 ? FIB_PROTOCOL_IP6 : FIB_PROTOCOL_IP4);
+  rv = add_del_mroute_check (fproto,
+			     mp->table_id,
+			     mp->next_hop_sw_if_index,
+			     mp->is_local,
+			     mp->create_vrf_if_needed, &fib_index);
+
+  if (0 != rv)
+    return (rv);
+
+  mfib_prefix_t pfx = {
+    .fp_len = ntohs (mp->grp_address_length),
+    .fp_proto = fproto,
+  };
+
+  if (FIB_PROTOCOL_IP4 == fproto)
+    {
+      clib_memcpy (&pfx.fp_grp_addr.ip4, mp->grp_address,
+		   sizeof (pfx.fp_grp_addr.ip4));
+      clib_memcpy (&pfx.fp_src_addr.ip4, mp->src_address,
+		   sizeof (pfx.fp_src_addr.ip4));
+    }
+  else
+    {
+      clib_memcpy (&pfx.fp_grp_addr.ip6, mp->grp_address,
+		   sizeof (pfx.fp_grp_addr.ip6));
+      clib_memcpy (&pfx.fp_src_addr.ip6, mp->src_address,
+		   sizeof (pfx.fp_src_addr.ip6));
+    }
+
+  return (mroute_add_del_handler (mp->is_add,
+				  mp->is_local,
+				  fib_index, &pfx,
+				  ntohl (mp->entry_flags),
+				  ntohl (mp->next_hop_sw_if_index),
+				  ntohl (mp->itf_flags)));
+}
+
+void
+vl_api_ip_mroute_add_del_t_handler (vl_api_ip_mroute_add_del_t * mp)
+{
+  vl_api_ip_mroute_add_del_reply_t *rmp;
+  int rv;
+  vnet_main_t *vnm = vnet_get_main ();
+
+  vnm->api_errno = 0;
+
+  rv = api_mroute_add_del_t_handler (mp);
+
+  rv = (rv == 0) ? vnm->api_errno : rv;
+
+  REPLY_MACRO (VL_API_IP_MROUTE_ADD_DEL_REPLY);
+}
+
 static void
 send_ip_details (vpe_api_main_t * am,
 		 unix_shared_memory_queue_t * q, u32 sw_if_index, u32 context)
@@ -1148,6 +1291,73 @@ static void
   REPLY_MACRO (VL_API_SW_INTERFACE_IP6_SET_LINK_LOCAL_ADDRESS_REPLY);
 }
 
+void
+vl_mfib_signal_send_one (unix_shared_memory_queue_t * q,
+			 u32 context, const mfib_signal_t * mfs)
+{
+  vl_api_mfib_signal_details_t *mp;
+  mfib_prefix_t prefix;
+  mfib_table_t *mfib;
+  mfib_itf_t *mfi;
+
+  mp = vl_msg_api_alloc (sizeof (*mp));
+
+  memset (mp, 0, sizeof (*mp));
+  mp->_vl_msg_id = ntohs (VL_API_MFIB_SIGNAL_DETAILS);
+  mp->context = context;
+
+  mfi = mfib_itf_get (mfs->mfs_itf);
+  mfib_entry_get_prefix (mfs->mfs_entry, &prefix);
+  mfib = mfib_table_get (mfib_entry_get_fib_index (mfs->mfs_entry),
+			 prefix.fp_proto);
+  mp->table_id = ntohl (mfib->mft_table_id);
+  mp->sw_if_index = ntohl (mfi->mfi_sw_if_index);
+
+  if (FIB_PROTOCOL_IP4 == prefix.fp_proto)
+    {
+      mp->grp_address_len = ntohs (prefix.fp_len);
+
+      memcpy (mp->grp_address, &prefix.fp_grp_addr.ip4, 4);
+      if (prefix.fp_len > 32)
+	{
+	  memcpy (mp->src_address, &prefix.fp_src_addr.ip4, 4);
+	}
+    }
+  else
+    {
+      mp->grp_address_len = ntohs (prefix.fp_len);
+
+      ASSERT (0);
+    }
+
+  if (0 != mfs->mfs_buffer_len)
+    {
+      mp->ip_packet_len = ntohs (mfs->mfs_buffer_len);
+
+      memcpy (mp->ip_packet_data, mfs->mfs_buffer, mfs->mfs_buffer_len);
+    }
+  else
+    {
+      mp->ip_packet_len = 0;
+    }
+
+  vl_msg_api_send_shmem (q, (u8 *) & mp);
+}
+
+static void
+vl_api_mfib_signal_dump_t_handler (vl_api_mfib_signal_dump_t * mp)
+{
+  unix_shared_memory_queue_t *q;
+
+  q = vl_api_client_index_to_input_queue (mp->client_index);
+  if (q == 0)
+    {
+      return;
+    }
+
+  while (q->cursize < q->maxsize && mfib_signal_send_one (q, mp->context))
+    ;
+}
 
 #define vl_msg_name_crc_list
 #include <vnet/ip/ip.api.h>
diff --git a/src/vnet/ip/lookup.c b/src/vnet/ip/lookup.c
index 734a4cd7..6c5611d3 100644
--- a/src/vnet/ip/lookup.c
+++ b/src/vnet/ip/lookup.c
@@ -43,6 +43,7 @@
 #include <vnet/fib/ip4_fib.h>
 #include <vnet/fib/ip6_fib.h>
 #include <vnet/mpls/mpls.h>
+#include <vnet/mfib/mfib_table.h>
 #include <vnet/dpo/drop_dpo.h>
 #include <vnet/dpo/classify_dpo.h>
 #include <vnet/dpo/punt_dpo.h>
@@ -258,6 +259,9 @@ format_ip_lookup_next (u8 * s, va_list * args)
     case IP_LOOKUP_NEXT_GLEAN:
       t = "glean";
       break;
+    case IP_LOOKUP_NEXT_MCAST:
+      t = "mcast";
+      break;
     case IP_LOOKUP_NEXT_REWRITE:
       break;
     }
@@ -767,6 +771,173 @@ VLIB_CLI_COMMAND (ip_route_command, static) = {
 };
 /* *INDENT-ON* */
 
+clib_error_t *
+vnet_ip_mroute_cmd (vlib_main_t * vm,
+		    unformat_input_t * main_input, vlib_cli_command_t * cmd)
+{
+  unformat_input_t _line_input, *line_input = &_line_input;
+  clib_error_t *error = NULL;
+  fib_route_path_t rpath;
+  u32 table_id, is_del;
+  vnet_main_t *vnm;
+  mfib_prefix_t pfx;
+  u32 fib_index;
+  mfib_itf_flags_t iflags = 0;
+  mfib_entry_flags_t eflags = 0;
+
+  vnm = vnet_get_main ();
+  is_del = 0;
+  table_id = 0;
+  memset (&pfx, 0, sizeof (pfx));
+  memset (&rpath, 0, sizeof (rpath));
+  rpath.frp_sw_if_index = ~0;
+
+  /* Get a line of input. */
+  if (!unformat_user (main_input, unformat_line_input, line_input))
+    return 0;
+
+  while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
+    {
+      if (unformat (line_input, "table %d", &table_id))
+	;
+      else if (unformat (line_input, "del"))
+	is_del = 1;
+      else if (unformat (line_input, "add"))
+	is_del = 0;
+      else if (unformat (line_input, "%U %U",
+			 unformat_ip4_address,
+			 &pfx.fp_src_addr.ip4,
+			 unformat_ip4_address, &pfx.fp_grp_addr.ip4))
+	{
+	  pfx.fp_proto = FIB_PROTOCOL_IP4;
+	  pfx.fp_len = 64;
+	}
+      else if (unformat (line_input, "%U %U",
+			 unformat_ip6_address,
+			 &pfx.fp_src_addr.ip6,
+			 unformat_ip6_address, &pfx.fp_grp_addr.ip6))
+	{
+	  pfx.fp_proto = FIB_PROTOCOL_IP6;
+	  pfx.fp_len = 256;
+	}
+      else if (unformat (line_input, "%U/%d",
+			 unformat_ip4_address,
+			 &pfx.fp_grp_addr.ip4, &pfx.fp_len))
+	{
+	  pfx.fp_proto = FIB_PROTOCOL_IP4;
+	}
+      else if (unformat (line_input, "%U/%d",
+			 unformat_ip6_address,
+			 &pfx.fp_grp_addr.ip6, &pfx.fp_len))
+	{
+	  pfx.fp_proto = FIB_PROTOCOL_IP6;
+	}
+      else if (unformat (line_input, "%U",
+			 unformat_ip4_address, &pfx.fp_grp_addr.ip4))
+	{
+	  memset (&pfx.fp_src_addr.ip4, 0, sizeof (pfx.fp_src_addr.ip4));
+	  pfx.fp_proto = FIB_PROTOCOL_IP4;
+	  pfx.fp_len = 32;
+	}
+      else if (unformat (line_input, "%U",
+			 unformat_ip6_address, &pfx.fp_grp_addr.ip6))
+	{
+	  memset (&pfx.fp_src_addr.ip6, 0, sizeof (pfx.fp_src_addr.ip6));
+	  pfx.fp_proto = FIB_PROTOCOL_IP6;
+	  pfx.fp_len = 128;
+	}
+      else if (unformat (line_input, "via %U",
+			 unformat_vnet_sw_interface, vnm,
+			 &rpath.frp_sw_if_index))
+	{
+	  rpath.frp_weight = 1;
+	  rpath.frp_proto = FIB_PROTOCOL_IP4;
+	}
+      else if (unformat (line_input, "%U", unformat_mfib_itf_flags, &iflags))
+	;
+      else if (unformat (line_input, "%U",
+			 unformat_mfib_entry_flags, &eflags))
+	;
+      else
+	{
+	  error = unformat_parse_error (line_input);
+	  goto done;
+	}
+    }
+
+  unformat_free (line_input);
+
+  if (~0 == table_id)
+    {
+      /*
+       * if no table_id is passed we will manipulate the default
+       */
+      fib_index = 0;
+    }
+  else
+    {
+      fib_index = mfib_table_find (pfx.fp_proto, table_id);
+
+      if (~0 == fib_index)
+	{
+	  error = clib_error_return (0, "Nonexistent table id %d", table_id);
+	  goto done;
+	}
+    }
+
+  if (is_del && 0 == rpath.frp_weight)
+    {
+      mfib_table_entry_delete (fib_index, &pfx, MFIB_SOURCE_CLI);
+    }
+  else if (eflags)
+    {
+      mfib_table_entry_update (fib_index, &pfx, MFIB_SOURCE_CLI, eflags);
+    }
+  else
+    {
+      if (is_del)
+	mfib_table_entry_path_remove (fib_index,
+				      &pfx, MFIB_SOURCE_CLI, &rpath);
+      else
+	mfib_table_entry_path_update (fib_index,
+				      &pfx, MFIB_SOURCE_CLI, &rpath, iflags);
+    }
+
+done:
+  return error;
+}
+
+/*?
+ * This command is used to add or delete IPv4 or IPv6  multicastroutes. All
+ * IP Addresses ('<em><dst-ip-addr>/<width></em>',
+ * '<em><next-hop-ip-addr></em>' and '<em><adj-hop-ip-addr></em>')
+ * can be IPv4 or IPv6, but all must be of the same form in a single
+ * command. To display the current set of routes, use the commands
+ * '<em>show ip mfib</em>' and '<em>show ip6 mfib</em>'.
+ * The full set of support flags for interfaces and route is shown via;
+ * '<em>show mfib route flags</em>' and '<em>show mfib itf flags</em>'
+ * respectively.
+ * @cliexpar
+ * Example of how to add a forwarding interface to a route (and create the
+ * route if it does not exist)
+ * @cliexcmd{ip mroute add 232.1.1.1 via GigabitEthernet2/0/0 Forward}
+ * Example of how to add an accepting interface to a route (and create the
+ * route if it does not exist)
+ * @cliexcmd{ip mroute add 232.1.1.1 via GigabitEthernet2/0/1 Accept}
+ * Example of changing the route's flags to send signals via the API
+ * @cliexcmd{ip mroute add 232.1.1.1 Signal}
+
+ ?*/
+/* *INDENT-OFF* */
+VLIB_CLI_COMMAND (ip_mroute_command, static) =
+{
+  .path = "ip mroute",
+  .short_help = "ip mroute [add|del] <dst-ip-addr>/<width> [table <table-id>] [via <next-hop-ip-addr> [<interface>],",
+  .function = vnet_ip_mroute_cmd,
+  .is_mp_safe = 1,
+};
+/* *INDENT-ON* */
+
 /*
  * The next two routines address a longstanding script hemorrhoid.
  * Probing a v4 or v6 neighbor needs to appear to be synchronous,
diff --git a/src/vnet/ip/lookup.h b/src/vnet/ip/lookup.h
index 3dbd7b3b..27c70943 100644
--- a/src/vnet/ip/lookup.h
+++ b/src/vnet/ip/lookup.h
@@ -91,6 +91,9 @@ typedef enum
   /** This packets needs to go to ICMP error */
   IP_LOOKUP_NEXT_ICMP_ERROR,
 
+  /** Multicast Adjacency. */
+  IP_LOOKUP_NEXT_MCAST,
+
   IP_LOOKUP_N_NEXT,
 } ip_lookup_next_t;
 
@@ -115,6 +118,7 @@ typedef enum
     [IP_LOOKUP_NEXT_ARP] = "ip4-arp",				\
     [IP_LOOKUP_NEXT_GLEAN] = "ip4-glean",			\
     [IP_LOOKUP_NEXT_REWRITE] = "ip4-rewrite",    		\
+    [IP_LOOKUP_NEXT_MCAST] = "ip4-rewrite-mcast",	        \
     [IP_LOOKUP_NEXT_MIDCHAIN] = "ip4-midchain",		        \
     [IP_LOOKUP_NEXT_LOAD_BALANCE] = "ip4-load-balance",		\
     [IP_LOOKUP_NEXT_ICMP_ERROR] = "ip4-icmp-error",		\
@@ -127,6 +131,7 @@ typedef enum
     [IP_LOOKUP_NEXT_ARP] = "ip6-discover-neighbor",		\
     [IP_LOOKUP_NEXT_GLEAN] = "ip6-glean",			\
     [IP_LOOKUP_NEXT_REWRITE] = "ip6-rewrite",			\
+    [IP_LOOKUP_NEXT_MCAST] = "ip6-rewrite-mcast",		\
     [IP_LOOKUP_NEXT_MIDCHAIN] = "ip6-midchain",			\
     [IP_LOOKUP_NEXT_LOAD_BALANCE] = "ip6-load-balance",		\
     [IP_LOOKUP_NEXT_ICMP_ERROR] = "ip6-icmp-error",		\
@@ -203,12 +208,6 @@ typedef struct ip_adjacency_t_
   /** Interface address index for this local/arp adjacency. */
   u32 if_address_index;
 
-  /** Force re-lookup in a different FIB. ~0 => normal behavior */
-  u16 mcast_group_index;
-
-  /** Highest possible perf subgraph arc interposition, e.g. for ip6 ioam */
-  u16 saved_lookup_next_index;
-
   /*
    * link/ether-type
    */
@@ -236,28 +235,28 @@ typedef struct ip_adjacency_t_
        */
     struct
     {
-	  /**
-	   * The recursive next-hop
-	   */
+      /**
+       * The recursive next-hop
+       */
       ip46_address_t next_hop;
-	  /**
-	   * The node index of the tunnel's post rewrite/TX function.
-	   */
+      /**
+       * The node index of the tunnel's post rewrite/TX function.
+       */
       u32 tx_function_node;
-	  /**
-	   * The next DPO to use
-	   */
+      /**
+       * The next DPO to use
+       */
       dpo_id_t next_dpo;
-	  /**
-	   * A function to perform the post-rewrite fixup
-	   */
-      adj_midchain_fixup_t fixup_func;
-    } midchain;
       /**
-       * IP_LOOKUP_NEXT_GLEAN
-       *
-       * Glean the address to ARP for from the packet's destination
+       * A function to perform the post-rewrite fixup
        */
+      adj_midchain_fixup_t fixup_func;
+    } midchain;
+    /**
+     * IP_LOOKUP_NEXT_GLEAN
+     *
+     * Glean the address to ARP for from the packet's destination
+     */
     struct
     {
       ip46_address_t receive_addr;
@@ -291,43 +290,6 @@ STATIC_ASSERT ((STRUCT_OFFSET_OF (ip_adjacency_t, cacheline1) ==
 /* An all zeros address */
 extern const ip46_address_t zero_addr;
 
-/* IP multicast adjacency. */
-typedef struct
-{
-  /* Handle for this adjacency in adjacency heap. */
-  u32 heap_handle;
-
-  /* Number of adjecencies in block. */
-  u32 n_adj;
-
-  /* Rewrite string. */
-    vnet_declare_rewrite (64 - 2 * sizeof (u32));
-}
-ip_multicast_rewrite_t;
-
-typedef struct
-{
-  /* ip4-multicast-rewrite next index. */
-  u32 next_index;
-
-  u8 n_rewrite_bytes;
-
-  u8 rewrite_string[64 - 1 * sizeof (u32) - 1 * sizeof (u8)];
-}
-ip_multicast_rewrite_string_t;
-
-typedef struct
-{
-  ip_multicast_rewrite_t *rewrite_heap;
-
-  ip_multicast_rewrite_string_t *rewrite_strings;
-
-  /* Negative rewrite string index; >= 0 sw_if_index.
-     Sorted.  Used to hash. */
-  i32 **adjacency_id_vector;
-
-  uword *adjacency_by_id_vector;
-} ip_multicast_lookup_main_t;
 
 typedef struct
 {
diff --git a/src/vnet/mcast/mcast.c b/src/vnet/mcast/mcast.c
deleted file mode 100644
index 55be89ae..00000000
--- a/src/vnet/mcast/mcast.c
+++ /dev/null
@@ -1,565 +0,0 @@
-/*
- * Copyright (c) 2015 Cisco and/or its affiliates.
- * Licensed under the Apache License, Version 2.0 (the "License");
- * you may not use this file except in compliance with the License.
- * You may obtain a copy of the License at:
- *
- *     http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-#include <vnet/mcast/mcast.h>
-
-#include <vlib/vlib.h>
-#include <vnet/vnet.h>
-#include <vnet/pg/pg.h>
-#include <vppinfra/error.h>
-#include <vnet/ip/ip4_packet.h>
-#include <vnet/ip/icmp46_packet.h>
-#include <vnet/ip/ip4.h>
-
-typedef struct {
-  u32 sw_if_index;
-  u32 next_index;
-  u32 group_index;
-} mcast_prep_trace_t;
-
-/* packet trace format function */
-static u8 * format_mcast_prep_trace (u8 * s, va_list * args)
-{
-  CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *);
-  CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *);
-  mcast_prep_trace_t * t = va_arg (*args, mcast_prep_trace_t *);
-  
-  s = format (s, "MCAST_PREP: group %d, next index %d, tx_sw_if_index %d",
-              t->group_index, t->next_index, t->sw_if_index);
-  return s;
-}
-
-mcast_main_t mcast_main;
-vlib_node_registration_t mcast_prep_node;
-vlib_node_registration_t mcast_recycle_node;
-
-#define foreach_mcast_prep_error \
-_(MCASTS, "Multicast Packets")
-
-typedef enum {
-#define _(sym,str) MCAST_PREP_ERROR_##sym,
-  foreach_mcast_prep_error
-#undef _
-  MCAST_PREP_N_ERROR,
-} mcast_prep_error_t;
-
-static char * mcast_prep_error_strings[] = {
-#define _(sym,string) string,
-  foreach_mcast_prep_error
-#undef _
-};
-
-typedef enum {
-  MCAST_PREP_NEXT_DROP,
-  MCAST_PREP_N_NEXT,
-} mcast_prep_next_t;
-
-static uword
-mcast_prep_node_fn (vlib_main_t * vm,
-		  vlib_node_runtime_t * node,
-		  vlib_frame_t * frame)
-{
-  u32 n_left_from, * from, * to_next;
-  mcast_prep_next_t next_index;
-  mcast_main_t * mcm = &mcast_main;
-  vlib_node_t *n = vlib_get_node (vm, mcast_prep_node.index);
-  u32 node_counter_base_index = n->error_heap_index;
-  vlib_error_main_t * em = &vm->error_main;
-  ip4_main_t * im = &ip4_main;
-  ip_lookup_main_t * lm = &im->lookup_main;
-
-  from = vlib_frame_vector_args (frame);
-  n_left_from = frame->n_vectors;
-  next_index = node->cached_next_index;
-
-  while (n_left_from > 0)
-    {
-      u32 n_left_to_next;
-
-      vlib_get_next_frame (vm, node, next_index,
-			   to_next, n_left_to_next);
-
-      while (0 && n_left_from >= 4 && n_left_to_next >= 2)
-	{
-          u32 bi0, bi1;
-	  vlib_buffer_t * b0, * b1;
-          u32 next0, next1;
-          u32 sw_if_index0, sw_if_index1;
-          
-	  /* Prefetch next iteration. */
-	  {
-	    vlib_buffer_t * p2, * p3;
-            
-	    p2 = vlib_get_buffer (vm, from[2]);
-	    p3 = vlib_get_buffer (vm, from[3]);
-            
-	    vlib_prefetch_buffer_header (p2, LOAD);
-	    vlib_prefetch_buffer_header (p3, LOAD);
-
-	    CLIB_PREFETCH (p2->data, CLIB_CACHE_LINE_BYTES, STORE);
-	    CLIB_PREFETCH (p3->data, CLIB_CACHE_LINE_BYTES, STORE);
-	  }
-
-          /* speculatively enqueue b0 and b1 to the current next frame */
-	  to_next[0] = bi0 = from[0];
-	  to_next[1] = bi1 = from[1];
-	  from += 2;
-	  to_next += 2;
-	  n_left_from -= 2;
-	  n_left_to_next -= 2;
-
-	  b0 = vlib_get_buffer (vm, bi0);
-	  b1 = vlib_get_buffer (vm, bi1);
-
-          sw_if_index0 = vnet_buffer(b0)->sw_if_index[VLIB_RX];
-          next0 = 0;
-          sw_if_index1 = vnet_buffer(b1)->sw_if_index[VLIB_RX];
-          next1 = 0;
-
-          /* $$$$ your message in this space. Process 2 x pkts */
-
-          if (PREDICT_FALSE((node->flags & VLIB_NODE_FLAG_TRACE)))
-            {
-              if (b0->flags & VLIB_BUFFER_IS_TRACED) 
-                {
-                    mcast_prep_trace_t *t = 
-                      vlib_add_trace (vm, node, b0, sizeof (*t));
-                    t->sw_if_index = sw_if_index0;
-                    t->next_index = next0;
-                  }
-                if (b1->flags & VLIB_BUFFER_IS_TRACED) 
-                  {
-                    mcast_prep_trace_t *t = 
-                      vlib_add_trace (vm, node, b1, sizeof (*t));
-                    t->sw_if_index = sw_if_index1;
-                    t->next_index = next1;
-                  }
-              }
-            
-            /* verify speculative enqueues, maybe switch current next frame */
-            vlib_validate_buffer_enqueue_x2 (vm, node, next_index,
-                                             to_next, n_left_to_next,
-                                             bi0, bi1, next0, next1);
-        }
-      
-      while (n_left_from > 0 && n_left_to_next > 0)
-	{
-          u32 bi0;
-	  vlib_buffer_t * b0;
-          u32 next0, adj_index0;
-          mcast_group_t * g0;
-          ip_adjacency_t * adj0;
-          
-          /* speculatively enqueue b0 to the current next frame */
-	  bi0 = from[0];
-	  to_next[0] = bi0;
-	  from += 1;
-	  to_next += 1;
-	  n_left_from -= 1;
-	  n_left_to_next -= 1;
-
-	  b0 = vlib_get_buffer (vm, bi0);
-
-          adj_index0 = vnet_buffer (b0)->ip.adj_index[VLIB_TX];
-          adj0 = ip_get_adjacency (lm, adj_index0);
-          vnet_buffer(b0)->mcast.mcast_group_index = adj0->mcast_group_index;
-          g0 = pool_elt_at_index (mcm->groups, adj0->mcast_group_index);
-
-          /* 
-           * Handle the degenerate single-copy case 
-           * If we don't change the freelist, the packet will never
-           * make it to the recycle node...
-           */
-          if (PREDICT_TRUE(vec_len (g0->members) > 1))
-            {
-              /* Save the original free list index */
-              vnet_buffer(b0)->mcast.original_free_list_index =
-                b0->free_list_index;
-              
-              /* Swap in the multicast recycle list */
-              b0->free_list_index = mcm->mcast_recycle_list_index;
-              
-              /* 
-               * Make sure that intermediate "frees" don't screw up 
-               */
-              b0->recycle_count = vec_len (g0->members);
-              b0->flags |= VLIB_BUFFER_RECYCLE;
-
-              /* Set up for the recycle node */
-              vnet_buffer(b0)->mcast.mcast_current_index = 1;
-            }
-
-          /* Transmit the pkt on the first interface */
-          next0 = g0->members[0].prep_and_recycle_node_next_index;
-          vnet_buffer(b0)->sw_if_index[VLIB_TX] = 
-            g0->members[0].tx_sw_if_index;
-
-          if (PREDICT_FALSE((node->flags & VLIB_NODE_FLAG_TRACE) 
-                            && (b0->flags & VLIB_BUFFER_IS_TRACED))) {
-            mcast_prep_trace_t *t = 
-               vlib_add_trace (vm, node, b0, sizeof (*t));
-            t->next_index = next0;
-            t->sw_if_index = vnet_buffer(b0)->sw_if_index[VLIB_TX];
-            t->group_index = vnet_buffer(b0)->mcast.mcast_group_index;
-            }
-            
-          /* verify speculative enqueue, maybe switch current next frame */
-	  vlib_validate_buffer_enqueue_x1 (vm, node, next_index,
-					   to_next, n_left_to_next,
-					   bi0, next0);
-	}
-
-      vlib_put_next_frame (vm, node, next_index, n_left_to_next);
-    }
-
-  em->counters[node_counter_base_index + MCAST_PREP_ERROR_MCASTS] += 
-      frame->n_vectors;
-
-  return frame->n_vectors;
-}
-
-VLIB_REGISTER_NODE (mcast_prep_node) = {
-  .function = mcast_prep_node_fn,
-  .name = "mcast_prep",
-  .vector_size = sizeof (u32),
-  .format_trace = format_mcast_prep_trace,
-  .type = VLIB_NODE_TYPE_INTERNAL,
-  
-  .n_errors = ARRAY_LEN(mcast_prep_error_strings),
-  .error_strings = mcast_prep_error_strings,
-
-  .n_next_nodes = MCAST_PREP_N_NEXT,
-
-  /* edit / add dispositions here */
-  .next_nodes = {
-        [MCAST_PREP_NEXT_DROP] = "error-drop",
-  },
-};
-
-typedef struct {
-  u32 sw_if_index;
-  u32 next_index;
-  u32 current_member;
-  u32 group_index;
-} mcast_recycle_trace_t;
-
-static u8 * format_mcast_recycle_trace (u8 * s, va_list * args)
-{
-  CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *);
-  CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *);
-  mcast_recycle_trace_t * t = va_arg (*args, mcast_recycle_trace_t *);
-  
-  s = format (s, 
-"MCAST_R: group %d, current member %d next (node) index %d, tx_sw_if_index %d",
-              t->group_index, t->current_member, t->next_index, t->sw_if_index);
-  return s;
-}
-
-#define foreach_mcast_recycle_error \
-_(RECYCLES, "Multicast Recycles")
-
-typedef enum {
-#define _(sym,str) MCAST_RECYCLE_ERROR_##sym,
-  foreach_mcast_recycle_error
-#undef _
-  MCAST_RECYCLE_N_ERROR,
-} mcast_recycle_error_t;
-
-static char * mcast_recycle_error_strings[] = {
-#define _(sym,string) string,
-  foreach_mcast_recycle_error
-#undef _
-};
-
-typedef enum {
-  MCAST_RECYCLE_NEXT_DROP,
-  MCAST_RECYCLE_N_NEXT,
-} mcast_recycle_next_t;
-
-static uword
-mcast_recycle_node_fn (vlib_main_t * vm,
-		  vlib_node_runtime_t * node,
-		  vlib_frame_t * frame)
-{
-  u32 n_left_from, * from, * to_next;
-  mcast_recycle_next_t next_index;
-  mcast_main_t * mcm = &mcast_main;
-  vlib_node_t *n = vlib_get_node (vm, mcast_recycle_node.index);
-  u32 node_counter_base_index = n->error_heap_index;
-  vlib_error_main_t * em = &vm->error_main;
-
-  from = vlib_frame_vector_args (frame);
-  n_left_from = frame->n_vectors;
-  next_index = node->cached_next_index;
-
-  while (n_left_from > 0)
-    {
-      u32 n_left_to_next;
-
-      vlib_get_next_frame (vm, node, next_index,
-			   to_next, n_left_to_next);
-
-      while (0 && n_left_from >= 4 && n_left_to_next >= 2)
-	{
-          u32 bi0, bi1;
-	  vlib_buffer_t * b0, * b1;
-          u32 next0, next1;
-          u32 sw_if_index0, sw_if_index1;
-          
-	  /* Prefetch next iteration. */
-	  {
-	    vlib_buffer_t * p2, * p3;
-            
-	    p2 = vlib_get_buffer (vm, from[2]);
-	    p3 = vlib_get_buffer (vm, from[3]);
-            
-	    vlib_prefetch_buffer_header (p2, LOAD);
-	    vlib_prefetch_buffer_header (p3, LOAD);
-
-	    CLIB_PREFETCH (p2->data, CLIB_CACHE_LINE_BYTES, STORE);
-	    CLIB_PREFETCH (p3->data, CLIB_CACHE_LINE_BYTES, STORE);
-	  }
-
-          /* speculatively enqueue b0 and b1 to the current next frame */
-	  to_next[0] = bi0 = from[0];
-	  to_next[1] = bi1 = from[1];
-	  from += 2;
-	  to_next += 2;
-	  n_left_from -= 2;
-	  n_left_to_next -= 2;
-
-	  b0 = vlib_get_buffer (vm, bi0);
-	  b1 = vlib_get_buffer (vm, bi1);
-
-          sw_if_index0 = vnet_buffer(b0)->sw_if_index[VLIB_RX];
-          next0 = 0;
-          sw_if_index1 = vnet_buffer(b1)->sw_if_index[VLIB_RX];
-          next1 = 0;
-
-          /* $$$$ your message in this space. Process 2 x pkts */
-
-          if (PREDICT_FALSE((node->flags & VLIB_NODE_FLAG_TRACE)))
-            {
-              if (b0->flags & VLIB_BUFFER_IS_TRACED) 
-                {
-                    mcast_recycle_trace_t *t = 
-                      vlib_add_trace (vm, node, b0, sizeof (*t));
-                    t->sw_if_index = sw_if_index0;
-                    t->next_index = next0;
-                  }
-                if (b1->flags & VLIB_BUFFER_IS_TRACED) 
-                  {
-                    mcast_recycle_trace_t *t = 
-                      vlib_add_trace (vm, node, b1, sizeof (*t));
-                    t->sw_if_index = sw_if_index1;
-                    t->next_index = next1;
-                  }
-              }
-            
-            /* verify speculative enqueues, maybe switch current next frame */
-            vlib_validate_buffer_enqueue_x2 (vm, node, next_index,
-                                             to_next, n_left_to_next,
-                                             bi0, bi1, next0, next1);
-        }
-      
-      while (n_left_from > 0 && n_left_to_next > 0)
-	{
-          u32 bi0;
-	  vlib_buffer_t * b0;
-          u32 next0;
-          u32 current_member0;
-          mcast_group_t * g0;
-          
-          /* speculatively enqueue b0 to the current next frame */
-	  bi0 = from[0];
-	  to_next[0] = bi0;
-	  from += 1;
-	  to_next += 1;
-	  n_left_from -= 1;
-	  n_left_to_next -= 1;
-
-	  b0 = vlib_get_buffer (vm, bi0);
-
-          g0 = pool_elt_at_index (mcm->groups, 
-                                  vnet_buffer(b0)->mcast.mcast_group_index);
-
-          /* No more replicas? */
-          if (b0->recycle_count == 1)
-            {
-              /* Restore the original free list index */
-              b0->free_list_index = 
-                vnet_buffer(b0)->mcast.original_free_list_index;
-              b0->flags &= ~(VLIB_BUFFER_RECYCLE);
-            }
-          current_member0 = vnet_buffer(b0)->mcast.mcast_current_index;
-          
-          next0 = 
-            g0->members[current_member0].prep_and_recycle_node_next_index;
-          vnet_buffer(b0)->sw_if_index[VLIB_TX] = 
-            g0->members[current_member0].tx_sw_if_index;
-          
-          vnet_buffer(b0)->mcast.mcast_current_index = 
-            current_member0 + 1;
-          
-          if (PREDICT_FALSE((node->flags & VLIB_NODE_FLAG_TRACE) 
-                            && (b0->flags & VLIB_BUFFER_IS_TRACED))) {
-            mcast_recycle_trace_t *t = 
-               vlib_add_trace (vm, node, b0, sizeof (*t));
-            t->next_index = next0;
-            t->sw_if_index = vnet_buffer(b0)->sw_if_index[VLIB_TX];
-            t->group_index = vnet_buffer(b0)->mcast.mcast_group_index;
-            t->current_member = current_member0;
-            }
-            
-          /* verify speculative enqueue, maybe switch current next frame */
-	  vlib_validate_buffer_enqueue_x1 (vm, node, next_index,
-					   to_next, n_left_to_next,
-					   bi0, next0);
-	}
-      vlib_put_next_frame (vm, node, next_index, n_left_to_next);
-    }
-
-  em->counters[node_counter_base_index + MCAST_RECYCLE_ERROR_RECYCLES] += 
-      frame->n_vectors;
-
-  return frame->n_vectors;
-}
-
-VLIB_REGISTER_NODE (mcast_recycle_node) = {
-  .function = mcast_recycle_node_fn,
-  .name = "mcast-recycle",
-  .vector_size = sizeof (u32),
-  .format_trace = format_mcast_recycle_trace,
-  .type = VLIB_NODE_TYPE_INTERNAL,
-  
-  .n_errors = ARRAY_LEN(mcast_recycle_error_strings),
-  .error_strings = mcast_recycle_error_strings,
-
-  .n_next_nodes = MCAST_RECYCLE_N_NEXT,
-
-  /* edit / add dispositions here */
-  .next_nodes = {
-        [MCAST_RECYCLE_NEXT_DROP] = "error-drop",
-  },
-};
-
-/*
- * fish pkts back from the recycle queue/freelist
- * un-flatten the context chains
- */
-static void mcast_recycle_callback (vlib_main_t *vm, 
-                                    vlib_buffer_free_list_t * fl)
-{
-  vlib_frame_t * f = 0;
-  u32 n_left_from;
-  u32 n_left_to_next = 0;
-  u32 n_this_frame = 0;
-  u32 * from;
-  u32 * to_next;
-  u32 bi0, pi0;
-  vlib_buffer_t *b0;
-  vlib_buffer_t *bnext0;
-  int i;
-
-  /* aligned, unaligned buffers */
-  for (i = 0; i < 2; i++) 
-    {
-      if (i == 0)
-        {
-          from = fl->aligned_buffers;
-          n_left_from = vec_len (from);
-        }
-      else
-        {
-          from = fl->unaligned_buffers;
-          n_left_from = vec_len (from);
-        }
-    
-      while (n_left_from > 0)
-        {
-          if (PREDICT_FALSE(n_left_to_next == 0)) 
-            {
-              if (f)
-                {
-                  f->n_vectors = n_this_frame;
-                  vlib_put_frame_to_node (vm, mcast_recycle_node.index, f);
-                }
-              
-              f = vlib_get_frame_to_node (vm, mcast_recycle_node.index);
-              to_next = vlib_frame_vector_args (f);
-              n_left_to_next = VLIB_FRAME_SIZE;
-              n_this_frame = 0;
-            }
-          
-          bi0 = from[0];
-          if (PREDICT_TRUE(n_left_from > 1))
-            {
-              pi0 = from[1];
-              vlib_prefetch_buffer_with_index(vm,pi0,LOAD);
-            }
-        
-          bnext0 = b0 = vlib_get_buffer (vm, bi0);
-          
-          while (bnext0->flags & VLIB_BUFFER_NEXT_PRESENT)
-            {
-              from += 1;
-              n_left_from -= 1;
-              bnext0 = vlib_get_buffer (vm, bnext0->next_buffer);
-            }
-          to_next[0] = bi0;
-
-          if (CLIB_DEBUG > 0)
-            vlib_buffer_set_known_state (vm, bi0, VLIB_BUFFER_KNOWN_ALLOCATED);
-
-          from++;
-          to_next++;
-          n_this_frame++;
-          n_left_to_next--;
-          n_left_from--;
-        }
-    }
-  
-  vec_reset_length (fl->aligned_buffers);
-  vec_reset_length (fl->unaligned_buffers);
-
-  if (f)
-    {
-      ASSERT(n_this_frame);
-      f->n_vectors = n_this_frame;
-      vlib_put_frame_to_node (vm, mcast_recycle_node.index, f);
-    }
-}
-
-clib_error_t *mcast_init (vlib_main_t *vm)
-{
-  mcast_main_t * mcm = &mcast_main;
-  vlib_buffer_main_t * bm = vm->buffer_main;
-  vlib_buffer_free_list_t * fl;
-    
-  mcm->vlib_main = vm;
-  mcm->vnet_main = vnet_get_main();
-  mcm->mcast_recycle_list_index = 
-    vlib_buffer_create_free_list (vm, 1024 /* fictional */, "mcast-recycle");
-
-  fl = pool_elt_at_index (bm->buffer_free_list_pool, 
-                          mcm->mcast_recycle_list_index);
-
-  fl->buffers_added_to_freelist_function = mcast_recycle_callback;
-
-  return 0;
-}
-
-VLIB_INIT_FUNCTION (mcast_init);
-
-
diff --git a/src/vnet/mcast/mcast.h b/src/vnet/mcast/mcast.h
deleted file mode 100644
index 96e51442..00000000
--- a/src/vnet/mcast/mcast.h
+++ /dev/null
@@ -1,50 +0,0 @@
-/*
- * Copyright (c) 2015 Cisco and/or its affiliates.
- * Licensed under the Apache License, Version 2.0 (the "License");
- * you may not use this file except in compliance with the License.
- * You may obtain a copy of the License at:
- *
- *     http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-#ifndef __included_vnet_mcast_h__
-#define __included_vnet_mcast_h__
-
-#include <vnet/vnet.h>
-#include <vlib/buffer.h>
-#include <vlib/buffer_funcs.h>
-
-typedef struct {
-  /* Arrange for both prep and recycle nodes to have identical
-     next indices for a given output interface */
-  u32 prep_and_recycle_node_next_index;
-
-  /* Show command, etc. */
-  u32 tx_sw_if_index;
-} mcast_group_member_t;
-
-typedef struct {
-  /* vector of group members */
-  mcast_group_member_t * members;
-} mcast_group_t;
-
-typedef struct {
-  /* pool of multicast (interface) groups */
-  mcast_group_t * groups;
-
-  /* multicast "free" list, aka recycle list */
-  u32 mcast_recycle_list_index;
-
-  /* convenience */
-  vlib_main_t * vlib_main;
-  vnet_main_t * vnet_main;
-} mcast_main_t;
-
-mcast_main_t mcast_main;
-
-#endif /* __included_vnet_mcast_h__ */
diff --git a/src/vnet/mcast/mcast_test.c b/src/vnet/mcast/mcast_test.c
deleted file mode 100644
index be80c9fc..00000000
--- a/src/vnet/mcast/mcast_test.c
+++ /dev/null
@@ -1,149 +0,0 @@
-/*
- * Copyright (c) 2015 Cisco and/or its affiliates.
- * Licensed under the Apache License, Version 2.0 (the "License");
- * you may not use this file except in compliance with the License.
- * You may obtain a copy of the License at:
- *
- *     http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-#include <vnet/mcast/mcast.h>
-
-#include <vlib/vlib.h>
-#include <vnet/vnet.h>
-#include <vnet/pg/pg.h>
-#include <vppinfra/error.h>
-#include <vnet/ip/lookup.h>
-#include <vnet/ip/ip4_packet.h>
-#include <vnet/ip/icmp46_packet.h>
-#include <vnet/ip/ip4.h>
-#include <vnet/mcast/mcast.h>
-
-typedef struct {
-  /* convenience */
-  vlib_main_t * vlib_main;
-  vnet_main_t * vnet_main;
-  mcast_main_t * mcast_main;
-} mcast_test_main_t;
-
-mcast_test_main_t mcast_test_main;
-vlib_node_registration_t mcast_prep_node;
-vlib_node_registration_t mcast_recycle_node;
-
-static clib_error_t *
-mcast_test_command_fn (vlib_main_t * vm,
-		 unformat_input_t * input,
-		 vlib_cli_command_t * cmd)
-{
-  /* u8 *rewrite_data; */
-  /* mcast_test_main_t * mtm = &mcast_test_main; */
-  /* mcast_main_t * mcm = mtm->mcast_main; */
-  /* ip_adjacency_t adj; */
-  /* u32 adj_index; */
-  /* mcast_group_t * g; */
-  /* mcast_group_member_t * member; */
-  /* unformat_input_t _line_input, * line_input = &_line_input; */
-  /* ip4_address_t dst_addr, zero; */
-  /* ip4_main_t * im = &ip4_main; */
-  /* ip_lookup_main_t * lm = &im->lookup_main; */
-
-  /* /\* Get a line of input. *\/ */
-  /* if (! unformat_user (input, unformat_line_input, line_input)) */
-  /*   return 0; */
-
-  /* pool_get (mcm->groups, g); */
-  /* memset (g, 0, sizeof (*g)); */
-
-  /* while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) */
-  /*   { */
-  /*     vnet_hw_interface_t *hw; */
-  /*     u32 next, sw_if_index; */
-
-  /*     if (unformat (line_input, "%U", unformat_vnet_sw_interface,  */
-  /*                   mtm->vnet_main, &sw_if_index))  */
-  /*       { */
-  /*         vec_add2 (g->members, member, 1); */
-  /*         member->tx_sw_if_index = sw_if_index; */
-          
-  /*         hw = vnet_get_sup_hw_interface (mtm->vnet_main,  */
-  /*                                         sw_if_index); */
-          
-  /*         next = vlib_node_add_next (mtm->vlib_main,  */
-  /*                                    mcast_prep_node.index, */
-  /*                                    hw->output_node_index); */
-          
-  /*         /\* Required to be the same next index... *\/ */
-  /*         vlib_node_add_next_with_slot (mtm->vlib_main, */
-  /*                                       mcast_recycle_node.index, */
-  /*                                       hw->output_node_index, next); */
-  /*         member->prep_and_recycle_node_next_index = next; */
-  /*       } */
-  /*     else */
-  /*       { */
-  /*         return unformat_parse_error (line_input); */
-  /*       } */
-  /*   } */
-
-  /* if (vec_len (g->members) == 0) */
-  /*   { */
-  /*     pool_put (mcm->groups, g); */
-  /*     vlib_cli_output (vm, "no group members specified"); */
-  /*     return 0; */
-  /*   } */
-
-
-  /* adj.lookup_next_index = IP_LOOKUP_NEXT_REWRITE; */
-  /* adj.mcast_group_index = g - mcm->groups; */
-  /* rewrite_data = format (0, "abcdefg"); */
-
-  /* vnet_rewrite_for_tunnel */
-  /*   (mtm->vnet_main, */
-  /*    (u32)~0, /\* tx_sw_if_index, we dont know yet *\/ */
-  /*    ip4_rewrite_node.index, */
-  /*    mcast_prep_node.index, */
-  /*    &adj.rewrite_header, */
-  /*    rewrite_data, vec_len(rewrite_data)); */
-
-  /* ip_add_adjacency (lm, &adj, 1 /\* one adj *\/, */
-  /*                   &adj_index); */
-  
-  /* dst_addr.as_u32 = clib_host_to_net_u32 (0x0a000002); */
-  /* zero.as_u32 = 0; */
-
-  /* ip4_add_del_route_next_hop (im, */
-  /*                             IP4_ROUTE_FLAG_ADD, */
-  /*                             &dst_addr, */
-  /*                             24 /\* mask width *\/, */
-  /*                             &zero /\* no next hop *\/, */
-                          
-  /*                             0, // next hop sw if index */
-  /*                             1, // weight */
-  /*                             adj_index, */
-  /*                             0 /\* explicit fib 0 *\/); */
-
-  return 0;
-}
-
-static VLIB_CLI_COMMAND (mcast_test_command) = {
-  .path = "test mc",
-  .short_help = "test mc",
-  .function = mcast_test_command_fn,
-};
-
-clib_error_t *mcast_test_init (vlib_main_t *vm)
-{
-  mcast_test_main_t * mtm = &mcast_test_main;
-    
-  mtm->vlib_main = vm;
-  mtm->vnet_main = vnet_get_main();
-  mtm->mcast_main = &mcast_main;
-
-  return 0;
-}
-
-VLIB_INIT_FUNCTION (mcast_test_init);
diff --git a/src/vnet/mfib/ip4_mfib.c b/src/vnet/mfib/ip4_mfib.c
new file mode 100644
index 00000000..08001c3f
--- /dev/null
+++ b/src/vnet/mfib/ip4_mfib.c
@@ -0,0 +1,465 @@
+/*
+ * Copyright (c) 2016 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <vnet/mfib/ip4_mfib.h>
+
+#include <vnet/mfib/mfib_table.h>
+#include <vnet/mfib/mfib_entry.h>
+
+static const mfib_prefix_t ip4_specials[] = {
+    {
+        /* (*,*)/0 */
+        .fp_src_addr = {
+            .ip4.data_u32 = 0,
+        },
+        .fp_grp_addr = {
+            .ip4.data_u32 = 0,
+        },
+        .fp_len  = 0,
+        .fp_proto = FIB_PROTOCOL_IP4,
+    },
+};
+
+static u32
+ip4_create_mfib_with_table_id (u32 table_id)
+{
+    mfib_table_t *mfib_table;
+
+    pool_get_aligned(ip4_main.mfibs, mfib_table, CLIB_CACHE_LINE_BYTES);
+    memset(mfib_table, 0, sizeof(*mfib_table));
+
+    mfib_table->mft_proto = FIB_PROTOCOL_IP4;
+    mfib_table->mft_index =
+        mfib_table->v4.index =
+            (mfib_table - ip4_main.mfibs);
+
+    hash_set (ip4_main.mfib_index_by_table_id,
+              table_id,
+              mfib_table->mft_index);
+
+    mfib_table->mft_table_id =
+        mfib_table->v4.table_id =
+            table_id;
+
+    mfib_table_lock(mfib_table->mft_index, FIB_PROTOCOL_IP4);
+
+    /*
+     * add the special entries into the new FIB
+     */
+    int ii;
+
+    for (ii = 0; ii < ARRAY_LEN(ip4_specials); ii++)
+    {
+        mfib_prefix_t prefix = ip4_specials[ii];
+
+        prefix.fp_src_addr.ip4.data_u32 =
+            clib_host_to_net_u32(prefix.fp_src_addr.ip4.data_u32);
+        prefix.fp_grp_addr.ip4.data_u32 =
+            clib_host_to_net_u32(prefix.fp_grp_addr.ip4.data_u32);
+
+        mfib_table_entry_update(mfib_table->mft_index,
+                                &prefix,
+                                MFIB_SOURCE_DEFAULT_ROUTE,
+                                MFIB_ENTRY_FLAG_DROP);
+    }
+
+    return (mfib_table->mft_index);
+}
+
+void
+ip4_mfib_table_destroy (ip4_mfib_t *mfib)
+{
+    mfib_table_t *mfib_table = (mfib_table_t*)mfib;
+    int ii;
+
+    /*
+     * remove all the specials we added when the table was created.
+     */
+    for (ii = 0; ii < ARRAY_LEN(ip4_specials); ii++)
+    {
+        fib_node_index_t mfei;
+        mfib_prefix_t prefix = ip4_specials[ii];
+
+        prefix.fp_src_addr.ip4.data_u32 =
+            clib_host_to_net_u32(prefix.fp_src_addr.ip4.data_u32);
+        prefix.fp_grp_addr.ip4.data_u32 =
+            clib_host_to_net_u32(prefix.fp_grp_addr.ip4.data_u32);
+
+        mfei = mfib_table_lookup(mfib_table->mft_index, &prefix);
+        mfib_table_entry_delete_index(mfei, MFIB_SOURCE_DEFAULT_ROUTE);
+    }
+
+    /*
+     * validate no more routes.
+     */
+    ASSERT(0 == mfib_table->mft_total_route_counts);
+    ASSERT(~0 != mfib_table->mft_table_id);
+
+    hash_unset (ip4_main.mfib_index_by_table_id, mfib_table->mft_table_id);
+    pool_put(ip4_main.mfibs, mfib_table);
+}
+
+u32
+ip4_mfib_table_find_or_create_and_lock (u32 table_id)
+{
+    u32 index;
+
+    index = ip4_mfib_index_from_table_id(table_id);
+    if (~0 == index)
+        return ip4_create_mfib_with_table_id(table_id);
+    mfib_table_lock(index, FIB_PROTOCOL_IP4);
+
+    return (index);
+}
+
+u32
+ip4_mfib_table_get_index_for_sw_if_index (u32 sw_if_index)
+{
+    if (sw_if_index >= vec_len(ip4_main.mfib_index_by_sw_if_index))
+    {
+        /*
+         * This is the case for interfaces that are not yet mapped to
+         * a IP table
+         */
+        return (~0);
+    }
+    return (ip4_main.mfib_index_by_sw_if_index[sw_if_index]);
+}
+
+#define IPV4_MFIB_GRP_LEN(_len)\
+    (_len > 32 ? 32 : _len)
+
+#define IP4_MFIB_MK_KEY(_grp, _src, _len, _key)                         \
+{                                                                       \
+    _key  = ((u64)(_grp->data_u32 &                                     \
+                   ip4_main.fib_masks[IPV4_MFIB_GRP_LEN(_len)])) << 32; \
+    _key |= _src->data_u32;                                             \
+}
+#define IP4_MFIB_MK_GRP_KEY(_grp, _len, _key)                           \
+{                                                                       \
+    _key  = ((u64)(_grp->data_u32 &                                     \
+                   ip4_main.fib_masks[IPV4_MFIB_GRP_LEN(_len)])) << 32; \
+}
+
+/*
+ * ip4_fib_table_lookup_exact_match
+ *
+ * Exact match prefix lookup
+ */
+fib_node_index_t
+ip4_mfib_table_lookup_exact_match (const ip4_mfib_t *mfib,
+                                   const ip4_address_t *grp,
+                                   const ip4_address_t *src,
+                                   u32 len)
+{
+    uword * hash, * result;
+    u64 key;
+
+    hash = mfib->fib_entry_by_dst_address[len];
+    IP4_MFIB_MK_KEY(grp, src, len, key);
+
+    result = hash_get(hash, key);
+
+    if (NULL != result) {
+        return (result[0]);
+    }
+    return (FIB_NODE_INDEX_INVALID);
+}
+
+/*
+ * ip4_fib_table_lookup
+ *
+ * Longest prefix match
+ */
+fib_node_index_t
+ip4_mfib_table_lookup (const ip4_mfib_t *mfib,
+                       const ip4_address_t *src,
+                       const ip4_address_t *grp,
+                       u32 len)
+{
+    uword * hash, * result;
+    i32 mask_len;
+    u64 key;
+
+    mask_len = len;
+
+    if (PREDICT_TRUE(64 == mask_len))
+    {
+        hash = mfib->fib_entry_by_dst_address[mask_len];
+        IP4_MFIB_MK_KEY(grp, src, mask_len, key);
+
+        result = hash_get (hash, key);
+
+        if (NULL != result) {
+            return (result[0]);
+        }
+    }
+
+    for (mask_len = 32; mask_len >= 0; mask_len--)
+    {
+        hash = mfib->fib_entry_by_dst_address[mask_len];
+        IP4_MFIB_MK_GRP_KEY(grp, mask_len, key);
+
+        result = hash_get (hash, key);
+
+        if (NULL != result) {
+            return (result[0]);
+        }
+    }
+    return (FIB_NODE_INDEX_INVALID);
+}
+
+void
+ip4_mfib_table_entry_insert (ip4_mfib_t *mfib,
+                             const ip4_address_t *grp,
+                             const ip4_address_t *src,
+                             u32 len,
+                             fib_node_index_t fib_entry_index)
+{
+    uword * hash, * result;
+    u64 key;
+
+    IP4_MFIB_MK_KEY(grp, src, len, key);
+    hash = mfib->fib_entry_by_dst_address[len];
+    result = hash_get (hash, key);
+
+    if (NULL == result) {
+        /*
+         * adding a new entry
+         */
+        if (NULL == hash) {
+            hash = hash_create (32 /* elts */, sizeof (uword));
+            hash_set_flags (hash, HASH_FLAG_NO_AUTO_SHRINK);
+        }
+        hash = hash_set(hash, key, fib_entry_index);
+        mfib->fib_entry_by_dst_address[len] = hash;
+    }
+    else
+    {
+        ASSERT(0);
+    }
+}
+
+void
+ip4_mfib_table_entry_remove (ip4_mfib_t *mfib,
+                             const ip4_address_t *grp,
+                             const ip4_address_t *src,
+                             u32 len)
+{
+    uword * hash, * result;
+    u64 key;
+
+    IP4_MFIB_MK_KEY(grp, src, len, key);
+    hash = mfib->fib_entry_by_dst_address[len];
+    result = hash_get (hash, key);
+
+    if (NULL == result)
+    {
+        /*
+         * removing a non-existant entry. i'll allow it.
+         */
+    }
+    else
+    {
+        hash_unset(hash, key);
+    }
+
+    mfib->fib_entry_by_dst_address[len] = hash;
+}
+
+static void
+ip4_mfib_table_show_all (ip4_mfib_t *mfib,
+                         vlib_main_t * vm)
+{
+    fib_node_index_t *mfib_entry_indicies;
+    fib_node_index_t *mfib_entry_index;
+    int i;
+
+    mfib_entry_indicies = NULL;
+
+    for (i = 0; i < ARRAY_LEN (mfib->fib_entry_by_dst_address); i++)
+    {
+        uword * hash = mfib->fib_entry_by_dst_address[i];
+
+        if (NULL != hash)
+        {
+            hash_pair_t * p;
+
+            hash_foreach_pair (p, hash,
+            ({
+                vec_add1(mfib_entry_indicies, p->value[0]);
+            }));
+        }
+    }
+
+    vec_sort_with_function(mfib_entry_indicies, mfib_entry_cmp_for_sort);
+
+    vec_foreach(mfib_entry_index, mfib_entry_indicies)
+    {
+        vlib_cli_output(vm, "%U",
+                        format_mfib_entry,
+                        *mfib_entry_index,
+                        MFIB_ENTRY_FORMAT_BRIEF);
+    }
+
+    vec_free(mfib_entry_indicies);
+}
+
+static void
+ip4_mfib_table_show_one (ip4_mfib_t *mfib,
+                         vlib_main_t * vm,
+                         ip4_address_t *src,
+                         ip4_address_t *grp,
+                         u32 mask_len)
+{
+    vlib_cli_output(vm, "%U",
+                    format_mfib_entry,
+                    ip4_mfib_table_lookup(mfib, src, grp, mask_len),
+                    MFIB_ENTRY_FORMAT_DETAIL);
+}
+
+static clib_error_t *
+ip4_show_mfib (vlib_main_t * vm,
+               unformat_input_t * input,
+               vlib_cli_command_t * cmd)
+{
+    ip4_main_t * im4 = &ip4_main;
+    mfib_table_t *mfib_table;
+    int verbose, matching;
+    ip4_address_t grp, src = {{0}};
+    u32 mask = 32;
+    int i, table_id = -1, fib_index = ~0;
+
+    verbose = 1;
+    matching = 0;
+
+    while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT)
+    {
+        if (unformat (input, "brief") || unformat (input, "summary")
+            || unformat (input, "sum"))
+            verbose = 0;
+
+        else if (unformat (input, "%U %U",
+                           unformat_ip4_address, &src,
+                           unformat_ip4_address, &grp))
+        {
+            matching = 1;
+            mask = 64;
+        }
+        else if (unformat (input, "%U", unformat_ip4_address, &grp))
+        {
+            matching = 1;
+            mask = 32;
+        }
+        else if (unformat (input, "%U/%d",
+                           unformat_ip4_address, &grp, &mask))
+            matching = 1;
+        else if (unformat (input, "table %d", &table_id))
+            ;
+        else if (unformat (input, "index %d", &fib_index))
+            ;
+        else
+            break;
+    }
+
+    pool_foreach (mfib_table, im4->mfibs,
+    ({
+        ip4_mfib_t *mfib = &mfib_table->v4;
+
+        if (table_id >= 0 && table_id != (int)mfib->table_id)
+            continue;
+        if (fib_index != ~0 && fib_index != (int)mfib->index)
+            continue;
+
+        vlib_cli_output (vm, "%U, fib_index %d",
+                         format_mfib_table_name, mfib->index, FIB_PROTOCOL_IP4,
+                         mfib->index);
+
+        /* Show summary? */
+        if (! verbose)
+        {
+            vlib_cli_output (vm, "%=20s%=16s", "Prefix length", "Count");
+            for (i = 0; i < ARRAY_LEN (mfib->fib_entry_by_dst_address); i++)
+            {
+                uword * hash = mfib->fib_entry_by_dst_address[i];
+                uword n_elts = hash_elts (hash);
+                if (n_elts > 0)
+                    vlib_cli_output (vm, "%20d%16d", i, n_elts);
+            }
+            continue;
+        }
+
+        if (!matching)
+        {
+            ip4_mfib_table_show_all(mfib, vm);
+        }
+        else
+        {
+            ip4_mfib_table_show_one(mfib, vm, &src, &grp, mask);
+        }
+    }));
+
+    return 0;
+}
+
+/*?
+ * This command displays the IPv4 MulticasrFIB Tables (VRF Tables) and
+ * the route entries for each table.
+ *
+ * @note This command will run for a long time when the FIB tables are
+ * comprised of millions of entries. For those senarios, consider displaying
+ * a single table or summary mode.
+ *
+ * @cliexpar
+ * Example of how to display all the IPv4 Multicast FIB tables:
+ * @cliexstart{show ip fib}
+ * ipv4-VRF:0, fib_index 0
+ * (*, 0.0.0.0/0):  flags:D,
+ *  Interfaces:
+ *  multicast-ip4-chain
+ *   [@1]: dpo-drop ip4
+ * (*, 232.1.1.1/32):
+ * Interfaces:
+ *  test-eth1: Forward,
+ *  test-eth2: Forward,
+ *  test-eth0: Accept,
+ * multicast-ip4-chain
+ * [@2]: dpo-replicate: [index:1 buckets:2 to:[0:0]]
+ *   [0] [@1]: ipv4-mcast: test-eth1: IP4: d0:d1:d2:d3:d4:01 -> 01:00:05:00:00:00
+ *   [1] [@1]: ipv4-mcast: test-eth2: IP4: d0:d1:d2:d3:d4:02 -> 01:00:05:00:00:00
+ *
+ * @cliexend
+ * Example of how to display a summary of all IPv4 FIB tables:
+ * @cliexstart{show ip fib summary}
+ * ipv4-VRF:0, fib_index 0, flow hash: src dst sport dport proto
+ *     Prefix length         Count
+ *                    0               1
+ *                    8               2
+ *                   32               4
+ * ipv4-VRF:7, fib_index 1, flow hash: src dst sport dport proto
+ *     Prefix length         Count
+ *                    0               1
+ *                    8               2
+ *                   24               2
+ *                   32               4
+ * @cliexend
+ ?*/
+/* *INDENT-OFF* */
+VLIB_CLI_COMMAND (ip4_show_fib_command, static) = {
+    .path = "show ip mfib",
+    .short_help = "show ip mfib [summary] [table <table-id>] [index <fib-id>] [<gre-addr>[/<mask>]] [<gre-addr>] [<src-addr> <gre-addr>]",
+    .function = ip4_show_mfib,
+};
+/* *INDENT-ON* */
diff --git a/src/vnet/mfib/ip4_mfib.h b/src/vnet/mfib/ip4_mfib.h
new file mode 100644
index 00000000..6fc74a36
--- /dev/null
+++ b/src/vnet/mfib/ip4_mfib.h
@@ -0,0 +1,95 @@
+/*
+ * Copyright (c) 2016 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+/**
+ * @brief The IPv4 Multicast-FIB
+ *
+ * FIXME
+ *
+ * This IPv4 FIB is used by the protocol independent FIB. So directly using
+ * this APIs in client code is not encouraged. However, this IPv4 FIB can be
+ * used if all the client wants is an IPv4 prefix data-base
+ */
+
+#ifndef __IP4_MFIB_H__
+#define __IP4_MFIB_H__
+
+#include <vlib/vlib.h>
+#include <vnet/ip/ip.h>
+
+#include <vnet/mfib/mfib_table.h>
+
+extern fib_node_index_t ip4_mfib_table_lookup(const ip4_mfib_t *fib,
+                                              const ip4_address_t *src,
+                                              const ip4_address_t *grp,
+                                              u32 len);
+extern fib_node_index_t ip4_mfib_table_lookup_exact_match(const ip4_mfib_t *fib,
+                                                          const ip4_address_t *grp,
+                                                          const ip4_address_t *src,
+                                                          u32 len);
+
+extern void ip4_mfib_table_entry_remove(ip4_mfib_t *fib,
+                                        const ip4_address_t *grp,
+                                        const ip4_address_t *src,
+                                        u32 len);
+
+extern void ip4_mfib_table_entry_insert(ip4_mfib_t *fib,
+                                        const ip4_address_t *grp,
+                                        const ip4_address_t *src,
+                                        u32 len,
+                                        fib_node_index_t fib_entry_index);
+extern void ip4_mfib_table_destroy(ip4_mfib_t *fib);
+
+/**
+ * @brief Get the FIB at the given index
+ */
+static inline ip4_mfib_t *
+ip4_mfib_get (u32 index)
+{
+    return (&(pool_elt_at_index(ip4_main.mfibs, index)->v4));
+}
+
+/**
+ * @brief Get or create an IPv4 fib.
+ *
+ * Get or create an IPv4 fib with the provided table ID.
+ *
+ * @param table_id
+ *      When set to \c ~0, an arbitrary and unused fib ID is picked
+ *      and can be retrieved with \c ret->table_id.
+ *      Otherwise, the fib ID to be used to retrieve or create the desired fib.
+ * @returns A pointer to the retrieved or created fib.
+ *
+ */
+extern u32 ip4_mfib_table_find_or_create_and_lock(u32 table_id);
+extern u32 ip4_mfib_table_create_and_lock(void);
+
+static inline
+u32 ip4_mfib_index_from_table_id (u32 table_id)
+{
+  ip4_main_t * im = &ip4_main;
+  uword * p;
+
+  p = hash_get (im->mfib_index_by_table_id, table_id);
+  if (!p)
+    return ~0;
+
+  return p[0];
+}
+
+extern u32 ip4_mfib_table_get_index_for_sw_if_index(u32 sw_if_index);
+
+
+#endif
+
diff --git a/src/vnet/mfib/ip6_mfib.c b/src/vnet/mfib/ip6_mfib.c
new file mode 100644
index 00000000..0c2e4c7b
--- /dev/null
+++ b/src/vnet/mfib/ip6_mfib.c
@@ -0,0 +1,663 @@
+/*
+ * Copyright (c) 2016 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <vnet/mfib/ip6_mfib.h>
+
+#include <vnet/mfib/mfib_table.h>
+#include <vnet/mfib/mfib_entry.h>
+#include <vnet/fib/ip6_fib.h>
+
+/**
+ * The number of bytes in an address/ask key in the radix tree
+ * First byte is the length in bytes.
+ */
+#define IP6_MFIB_KEY_LEN 33
+
+/**
+ * Key and mask for radix
+ */
+typedef struct ip6_mfib_key_t_
+{
+    u8 key[IP6_MFIB_KEY_LEN];
+    u8 mask[IP6_MFIB_KEY_LEN];
+} ip6_mfib_key_t;
+
+/**
+ * An object that is inserted into the radix tree.
+ * Since it's in the tree and has pointers, it cannot realloc and so cannot
+ * come from a vlib pool.
+ */
+typedef struct ip6_mfib_node_t_
+{
+    struct radix_node i6mn_nodes[2];
+    ip6_mfib_key_t i6mn_key;
+    index_t i6mn_entry;
+} ip6_mfib_node_t;
+
+static const mfib_prefix_t all_zeros = {
+    /* (*,*) */
+    .fp_src_addr = {
+        .ip6.as_u64 = {0, 0},
+    },
+    .fp_grp_addr = {
+        .ip6.as_u64 = {0, 0},
+    },
+    .fp_len  = 0,
+    .fp_proto = FIB_PROTOCOL_IP6,
+};
+
+typedef enum ip6_mfib_special_type_t_ {
+    IP6_MFIB_SPECIAL_TYPE_NONE,
+    IP6_MFIB_SPECIAL_TYPE_SOLICITED,
+} ip6_mfib_special_type_t;
+
+typedef struct ip6_mfib_special_t_ {
+    /**
+     * @brief solicited or not
+     */
+    ip6_mfib_special_type_t ims_type;
+
+    /**
+     * @brief the Prefix length
+     */
+    u8 ims_len;
+
+    /**
+     * @brief The last byte of the mcast address
+     */
+    u8 ims_byte;
+    /**
+     * @brief The scope of the address
+     */
+    u8 ims_scope;
+} ip6_mfib_special_t;
+
+static const ip6_mfib_special_t ip6_mfib_specials[] =
+{
+    {
+        /*
+         * Add ff02::1:ff00:0/104 via local route for all tables.
+         *  This is required for neighbor discovery to work.
+         */
+        .ims_type = IP6_MFIB_SPECIAL_TYPE_SOLICITED,
+        .ims_len = 104,
+    },
+    {
+        /*
+         * all-routers multicast address
+         */
+        .ims_type = IP6_MFIB_SPECIAL_TYPE_NONE,
+        .ims_scope = IP6_MULTICAST_SCOPE_link_local,
+        .ims_byte = IP6_MULTICAST_GROUP_ID_all_routers,
+        .ims_len = 128,
+    },
+    {
+        /*
+         * all-nodes multicast address
+         */
+        .ims_type = IP6_MFIB_SPECIAL_TYPE_NONE,
+        .ims_scope = IP6_MULTICAST_SCOPE_link_local,
+        .ims_byte = IP6_MULTICAST_GROUP_ID_all_hosts,
+        .ims_len = 128,
+    },
+    {
+        /*
+         *  Add all-mldv2  multicast address via local route for all tables
+         */
+        .ims_type = IP6_MFIB_SPECIAL_TYPE_NONE,
+        .ims_len = 128,
+        .ims_scope = IP6_MULTICAST_SCOPE_link_local,
+        .ims_byte = IP6_MULTICAST_GROUP_ID_mldv2_routers,
+    }
+};
+
+#define FOR_EACH_IP6_SPECIAL(_pfx, _body)                               \
+{                                                                       \
+    const ip6_mfib_special_t *_spec;                                    \
+    u8 _ii;                                                             \
+    for (_ii = 0;                                                       \
+         _ii < ARRAY_LEN(ip6_mfib_specials);                            \
+         _ii++)                                                         \
+    {                                                                   \
+        _spec = &ip6_mfib_specials[_ii];                                \
+        if (IP6_MFIB_SPECIAL_TYPE_SOLICITED == _spec->ims_type)         \
+        {                                                               \
+            ip6_set_solicited_node_multicast_address(                   \
+                &(_pfx)->fp_grp_addr.ip6, 0);                           \
+        }                                                               \
+        else                                                            \
+        {                                                               \
+            ip6_set_reserved_multicast_address (                        \
+                &(_pfx)->fp_grp_addr.ip6,                               \
+                _spec->ims_scope,                                       \
+                _spec->ims_byte);                                       \
+        }                                                               \
+        (_pfx)->fp_len = _spec->ims_len;                                \
+        do { _body; } while (0);                                        \
+    }                                                                   \
+}
+
+
+static u32
+ip6_create_mfib_with_table_id (u32 table_id)
+{
+    mfib_table_t *mfib_table;
+    mfib_prefix_t pfx = {
+        .fp_proto = FIB_PROTOCOL_IP6,
+    };
+    const fib_route_path_t path_for_us = {
+        .frp_proto = FIB_PROTOCOL_IP6,
+        .frp_addr = zero_addr,
+        .frp_sw_if_index = 0xffffffff,
+        .frp_fib_index = ~0,
+        .frp_weight = 0,
+        .frp_flags = FIB_ROUTE_PATH_LOCAL,
+    };
+
+    pool_get_aligned(ip6_main.mfibs, mfib_table, CLIB_CACHE_LINE_BYTES);
+    memset(mfib_table, 0, sizeof(*mfib_table));
+
+    mfib_table->mft_proto = FIB_PROTOCOL_IP6;
+    mfib_table->mft_index =
+        mfib_table->v6.index =
+            (mfib_table - ip6_main.mfibs);
+
+    hash_set (ip6_main.mfib_index_by_table_id,
+              table_id,
+              mfib_table->mft_index);
+
+    mfib_table->mft_table_id =
+        mfib_table->v6.table_id =
+            table_id;
+
+    mfib_table_lock(mfib_table->mft_index, FIB_PROTOCOL_IP6);
+
+    mfib_table->v6.rhead =
+        clib_mem_alloc_aligned (sizeof(*mfib_table->v6.rhead),
+                                CLIB_CACHE_LINE_BYTES);
+    rn_inithead0(mfib_table->v6.rhead, 8);
+
+    /*
+     * add the special entries into the new FIB
+     */
+    mfib_table_entry_update(mfib_table->mft_index,
+                            &all_zeros,
+                            MFIB_SOURCE_DEFAULT_ROUTE,
+                            MFIB_ENTRY_FLAG_DROP);
+
+    /*
+     * Add each of the specials
+     */
+    FOR_EACH_IP6_SPECIAL(&pfx,
+    ({
+        mfib_table_entry_path_update(mfib_table->mft_index,
+                                     &pfx,
+                                     MFIB_SOURCE_SPECIAL,
+                                     &path_for_us,
+                                     MFIB_ITF_FLAG_FORWARD);
+    }));
+
+    return (mfib_table->mft_index);
+}
+
+void
+ip6_mfib_table_destroy (ip6_mfib_t *mfib)
+{
+    mfib_table_t *mfib_table = (mfib_table_t*)mfib;
+    fib_node_index_t mfei;
+    mfib_prefix_t pfx = {
+        .fp_proto = FIB_PROTOCOL_IP6,
+    };
+    const fib_route_path_t path_for_us = {
+        .frp_proto = FIB_PROTOCOL_IP6,
+        .frp_addr = zero_addr,
+        .frp_sw_if_index = 0xffffffff,
+        .frp_fib_index = ~0,
+        .frp_weight = 0,
+        .frp_flags = FIB_ROUTE_PATH_LOCAL,
+    };
+
+    /*
+     * remove all the specials we added when the table was created.
+     */
+    FOR_EACH_IP6_SPECIAL(&pfx,
+    {
+        mfib_table_entry_path_remove(mfib_table->mft_index,
+                                     &pfx,
+                                     MFIB_SOURCE_SPECIAL,
+                                     &path_for_us);
+    });
+
+    mfei = mfib_table_lookup_exact_match(mfib_table->mft_index, &all_zeros);
+    mfib_table_entry_delete_index(mfei, MFIB_SOURCE_DEFAULT_ROUTE);
+
+    /*
+     * validate no more routes.
+     */
+    ASSERT(0 == mfib_table->mft_total_route_counts);
+    ASSERT(~0 != mfib_table->mft_table_id);
+
+    hash_unset (ip6_main.mfib_index_by_table_id, mfib_table->mft_table_id);
+    clib_mem_free(mfib_table->v6.rhead);
+    pool_put(ip6_main.mfibs, mfib_table);
+}
+
+void
+ip6_mfib_interface_enable_disable (u32 sw_if_index, int is_enable)
+{
+    const fib_route_path_t path = {
+        .frp_proto = FIB_PROTOCOL_IP6,
+        .frp_addr = zero_addr,
+        .frp_sw_if_index = sw_if_index,
+        .frp_fib_index = ~0,
+        .frp_weight = 0,
+    };
+    mfib_prefix_t pfx = {
+        .fp_proto = FIB_PROTOCOL_IP6,
+    };
+    u32 mfib_index;
+
+    vec_validate (ip6_main.mfib_index_by_sw_if_index, sw_if_index);
+    mfib_index = ip6_mfib_table_get_index_for_sw_if_index(sw_if_index);
+
+    if (is_enable)
+    {
+        FOR_EACH_IP6_SPECIAL(&pfx,
+        {
+            mfib_table_entry_path_update(mfib_index,
+                                         &pfx,
+                                         MFIB_SOURCE_SPECIAL,
+                                         &path,
+                                         MFIB_ITF_FLAG_ACCEPT);
+        });
+    }
+    else
+    {
+        FOR_EACH_IP6_SPECIAL(&pfx,
+        {
+            mfib_table_entry_path_remove(mfib_index,
+                                         &pfx,
+                                         MFIB_SOURCE_SPECIAL,
+                                         &path);
+        });
+    }
+}
+
+u32
+ip6_mfib_table_find_or_create_and_lock (u32 table_id)
+{
+    u32 index;
+
+    index = ip6_mfib_index_from_table_id(table_id);
+    if (~0 == index)
+        return ip6_create_mfib_with_table_id(table_id);
+    mfib_table_lock(index, FIB_PROTOCOL_IP6);
+
+    return (index);
+}
+
+u32
+ip6_mfib_table_get_index_for_sw_if_index (u32 sw_if_index)
+{
+    if (sw_if_index >= vec_len(ip6_main.mfib_index_by_sw_if_index))
+    {
+        /*
+         * This is the case for interfaces that are not yet mapped to
+         * a IP table
+         */
+        return (~0);
+    }
+    return (ip6_main.mfib_index_by_sw_if_index[sw_if_index]);
+}
+
+#define IP6_MFIB_MK_KEY(_grp, _src, _key)                           \
+{                                                                   \
+    (_key)->key[0] = 33;                                            \
+    memcpy((_key)->key+1, _grp, 16);                                \
+    memcpy((_key)->key+17, _src, 16);                               \
+}
+
+#define IP6_MFIB_MK_KEY_MASK(_grp, _src, _len, _key)                \
+{                                                                   \
+    IP6_MFIB_MK_KEY(_grp, _src, _key);                              \
+                                                                    \
+    (_key)->mask[0] = 33;                                           \
+    if (_len <= 128)                                                \
+    {                                                               \
+        memcpy((_key)->mask+1, &ip6_main.fib_masks[_len], 16);      \
+        memset((_key)->mask+17, 0, 16);                             \
+    }                                                               \
+    else                                                            \
+    {                                                               \
+        ASSERT(_len == 256);                                        \
+        memcpy((_key)->mask+1, &ip6_main.fib_masks[128], 16);       \
+        memcpy((_key)->mask+17, &ip6_main.fib_masks[128], 16);      \
+    }                                                               \
+}
+
+/*
+ * ip6_fib_table_lookup_exact_match
+ *
+ * Exact match prefix lookup
+ */
+fib_node_index_t
+ip6_mfib_table_lookup_exact_match (const ip6_mfib_t *mfib,
+                                   const ip6_address_t *grp,
+                                   const ip6_address_t *src,
+                                   u32 len)
+{
+    ip6_mfib_node_t *i6mn;
+    ip6_mfib_key_t key;
+
+    IP6_MFIB_MK_KEY_MASK(grp, src, len, &key);
+
+    i6mn = (ip6_mfib_node_t*) rn_lookup(key.key, key.mask,
+                                        (struct radix_node_head *)mfib->rhead);
+
+    if (NULL == i6mn)
+    {
+        return (INDEX_INVALID);
+    }
+
+    return (i6mn->i6mn_entry);
+}
+
+/*
+ * ip6_fib_table_lookup
+ *
+ * Longest prefix match
+ */
+fib_node_index_t
+ip6_mfib_table_lookup (const ip6_mfib_t *mfib,
+                       const ip6_address_t *src,
+                       const ip6_address_t *grp,
+                       u32 len)
+{
+    ip6_mfib_node_t *i6mn;
+    ip6_mfib_key_t key;
+
+    IP6_MFIB_MK_KEY_MASK(grp, src, len, &key);
+
+    i6mn = (ip6_mfib_node_t*) rn_search_m(key.key,
+                                          mfib->rhead->rnh_treetop,
+                                          key.mask);
+
+    ASSERT(NULL != i6mn);
+
+    return (i6mn->i6mn_entry);
+}
+
+/*
+ * ip6_fib_table_lookup
+ *
+ * Longest prefix match no mask
+ */
+fib_node_index_t
+ip6_mfib_table_lookup2 (const ip6_mfib_t *mfib,
+                        const ip6_address_t *src,
+                        const ip6_address_t *grp)
+{
+    ip6_mfib_node_t *i6mn;
+    ip6_mfib_key_t key;
+
+    IP6_MFIB_MK_KEY(grp, src, &key);
+
+    i6mn = (ip6_mfib_node_t*) rn_match(key.key,
+                                       (struct radix_node_head *)mfib->rhead); // const cast
+
+    ASSERT(NULL != i6mn);
+
+    return (i6mn->i6mn_entry);
+}
+
+void
+ip6_mfib_table_entry_insert (ip6_mfib_t *mfib,
+                             const ip6_address_t *grp,
+                             const ip6_address_t *src,
+                             u32 len,
+                             fib_node_index_t mfib_entry_index)
+{
+    ip6_mfib_node_t *i6mn = clib_mem_alloc(sizeof(*i6mn));
+
+    memset(i6mn, 0, sizeof(*i6mn));
+
+    IP6_MFIB_MK_KEY_MASK(grp, src, len, &i6mn->i6mn_key);
+    i6mn->i6mn_entry = mfib_entry_index;
+
+    if (NULL == rn_addroute(i6mn->i6mn_key.key,
+                            i6mn->i6mn_key.mask,
+                            mfib->rhead,
+                            i6mn->i6mn_nodes))
+    {
+        ASSERT(0);
+    }
+}
+
+void
+ip6_mfib_table_entry_remove (ip6_mfib_t *mfib,
+                             const ip6_address_t *grp,
+                             const ip6_address_t *src,
+                             u32 len)
+{
+    ip6_mfib_node_t *i6mn;
+    ip6_mfib_key_t key;
+
+    IP6_MFIB_MK_KEY_MASK(grp, src, len, &key);
+
+    i6mn = (ip6_mfib_node_t*) rn_delete(key.key, key.mask, mfib->rhead);
+
+    clib_mem_free(i6mn);
+}
+
+static clib_error_t *
+ip6_mfib_module_init (vlib_main_t * vm)
+{
+    return (NULL);
+}
+
+VLIB_INIT_FUNCTION(ip6_mfib_module_init);
+
+static void
+ip6_mfib_table_show_one (ip6_mfib_t *mfib,
+                         vlib_main_t * vm,
+                         ip6_address_t *src,
+                         ip6_address_t *grp,
+                         u32 mask_len)
+{
+    vlib_cli_output(vm, "%U",
+                    format_mfib_entry,
+                    ip6_mfib_table_lookup(mfib, src, grp, mask_len),
+                    MFIB_ENTRY_FORMAT_DETAIL);
+}
+
+typedef struct ip6_mfib_show_ctx_t_ {
+    u32 fib_index;
+    fib_node_index_t *entries;
+} ip6_mfib_show_ctx_t;
+
+
+static int
+ip6_mfib_table_collect_entries (struct radix_node *rn, void *arg)
+{
+    ip6_mfib_show_ctx_t *ctx = arg;
+    ip6_mfib_node_t *i6mn;
+
+    i6mn = (ip6_mfib_node_t*) rn;
+
+    vec_add1(ctx->entries, i6mn->i6mn_entry);
+
+    return (0);
+}
+
+static void
+ip6_mfib_table_show_all (ip6_mfib_t *mfib,
+                         vlib_main_t * vm)
+{
+    fib_node_index_t *mfib_entry_index;
+    ip6_mfib_show_ctx_t ctx = {
+        .fib_index = mfib->index,
+        .entries = NULL,
+    };
+
+    rn_walktree(mfib->rhead,
+                ip6_mfib_table_collect_entries,
+                &ctx);
+
+    vec_sort_with_function(ctx.entries, mfib_entry_cmp_for_sort);
+
+    vec_foreach(mfib_entry_index, ctx.entries)
+    {
+        vlib_cli_output(vm, "%U",
+                        format_mfib_entry,
+                        *mfib_entry_index,
+                        MFIB_ENTRY_FORMAT_BRIEF);
+    }
+
+    vec_free(ctx.entries);
+}
+
+static clib_error_t *
+ip6_show_mfib (vlib_main_t * vm,
+               unformat_input_t * input,
+               vlib_cli_command_t * cmd)
+{
+    ip6_main_t * im4 = &ip6_main;
+    mfib_table_t *mfib_table;
+    int verbose, matching;
+    ip6_address_t grp, src = {{0}};
+    u32 mask = 32;
+    int table_id = -1, fib_index = ~0;
+
+    verbose = 1;
+    matching = 0;
+
+    while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT)
+    {
+        if (unformat (input, "brief") || unformat (input, "summary")
+            || unformat (input, "sum"))
+            verbose = 0;
+
+        else if (unformat (input, "%U %U",
+                           unformat_ip6_address, &src,
+                           unformat_ip6_address, &grp))
+        {
+            matching = 1;
+            mask = 64;
+        }
+        else if (unformat (input, "%U", unformat_ip6_address, &grp))
+        {
+            matching = 1;
+            mask = 32;
+        }
+        else if (unformat (input, "%U/%d",
+                           unformat_ip6_address, &grp, &mask))
+            matching = 1;
+        else if (unformat (input, "table %d", &table_id))
+            ;
+        else if (unformat (input, "index %d", &fib_index))
+            ;
+        else
+            break;
+    }
+
+    pool_foreach (mfib_table, im4->mfibs,
+    ({
+        ip6_mfib_t *mfib = &mfib_table->v6;
+
+        if (table_id >= 0 && table_id != (int)mfib->table_id)
+            continue;
+        if (fib_index != ~0 && fib_index != (int)mfib->index)
+            continue;
+
+        vlib_cli_output (vm, "%U, fib_index %d",
+                         format_mfib_table_name, mfib->index, FIB_PROTOCOL_IP6,
+                         mfib->index);
+
+        /* Show summary? */
+        if (! verbose)
+        {
+            /* vlib_cli_output (vm, "%=20s%=16s", "Prefix length", "Count"); */
+            /* for (i = 0; i < ARRAY_LEN (mfib->fib_entry_by_dst_address); i++) */
+            /* { */
+            /*     uword * hash = mfib->fib_entry_by_dst_address[i]; */
+            /*     uword n_elts = hash_elts (hash); */
+            /*     if (n_elts > 0) */
+            /*         vlib_cli_output (vm, "%20d%16d", i, n_elts); */
+            /* } */
+            continue;
+        }
+
+        if (!matching)
+        {
+            ip6_mfib_table_show_all(mfib, vm);
+        }
+        else
+        {
+            ip6_mfib_table_show_one(mfib, vm, &src, &grp, mask);
+        }
+    }));
+
+    return 0;
+}
+
+/*
+ * This command displays the IPv4 MulticasrFIB Tables (VRF Tables) and
+ * the route entries for each table.
+ *
+ * @note This command will run for a long time when the FIB tables are
+ * comprised of millions of entries. For those senarios, consider displaying
+ * a single table or summary mode.
+ *
+ * @cliexpar
+ * Example of how to display all the IPv4 Multicast FIB tables:
+ * @cliexstart{show ip fib}
+ * ipv4-VRF:0, fib_index 0
+ * (*, 0.0.0.0/0):  flags:D,
+ *  Interfaces:
+ *  multicast-ip6-chain
+ *   [@1]: dpo-drop ip6
+ * (*, 232.1.1.1/32):
+ * Interfaces:
+ *  test-eth1: Forward,
+ *  test-eth2: Forward,
+ *  test-eth0: Accept,
+ * multicast-ip6-chain
+ * [@2]: dpo-replicate: [index:1 buckets:2 to:[0:0]]
+ *   [0] [@1]: ipv4-mcast: test-eth1: IP6: d0:d1:d2:d3:d4:01 -> 01:00:05:00:00:00
+ *   [1] [@1]: ipv4-mcast: test-eth2: IP6: d0:d1:d2:d3:d4:02 -> 01:00:05:00:00:00
+ *
+ * @cliexend
+ * Example of how to display a summary of all IPv4 FIB tables:
+ * @cliexstart{show ip fib summary}
+ * ipv4-VRF:0, fib_index 0, flow hash: src dst sport dport proto
+ *     Prefix length         Count
+ *                    0               1
+ *                    8               2
+ *                   32               4
+ * ipv4-VRF:7, fib_index 1, flow hash: src dst sport dport proto
+ *     Prefix length         Count
+ *                    0               1
+ *                    8               2
+ *                   24               2
+ *                   32               4
+ * @cliexend
+ */
+/* *INDENT-OFF* */
+VLIB_CLI_COMMAND (ip6_show_fib_command, static) = {
+    .path = "show ip6 mfib",
+    .short_help = "show ip mfib [summary] [table <table-id>] [index <fib-id>] [<grp-addr>[/<mask>]] [<grp-addr>] [<src-addr> <grp-addr>]",
+    .function = ip6_show_mfib,
+};
+/* *INDENT-ON* */
diff --git a/src/vnet/mfib/ip6_mfib.h b/src/vnet/mfib/ip6_mfib.h
new file mode 100644
index 00000000..d91af46d
--- /dev/null
+++ b/src/vnet/mfib/ip6_mfib.h
@@ -0,0 +1,109 @@
+/*
+ * Copyright (c) 2016 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+/**
+ * @brief The IPv4 Multicast-FIB
+ *
+ * FIXME
+ *
+ * This IPv4 FIB is used by the protocol independent FIB. So directly using
+ * this APIs in client code is not encouraged. However, this IPv4 FIB can be
+ * used if all the client wants is an IPv4 prefix data-base
+ */
+
+#ifndef __IP6_MFIB_H__
+#define __IP6_MFIB_H__
+
+#include <vlib/vlib.h>
+#include <vnet/ip/ip.h>
+
+#include <vnet/mfib/mfib_table.h>
+
+extern fib_node_index_t ip6_mfib_table_lookup(const ip6_mfib_t *fib,
+                                              const ip6_address_t *src,
+                                              const ip6_address_t *grp,
+                                              u32 len);
+extern fib_node_index_t ip6_mfib_table_lookup_exact_match(const ip6_mfib_t *fib,
+                                                          const ip6_address_t *grp,
+                                                          const ip6_address_t *src,
+                                                          u32 len);
+
+extern void ip6_mfib_table_entry_remove(ip6_mfib_t *fib,
+                                        const ip6_address_t *grp,
+                                        const ip6_address_t *src,
+                                        u32 len);
+
+extern void ip6_mfib_table_entry_insert(ip6_mfib_t *fib,
+                                        const ip6_address_t *grp,
+                                        const ip6_address_t *src,
+                                        u32 len,
+                                        fib_node_index_t fib_entry_index);
+extern void ip6_mfib_table_destroy(ip6_mfib_t *fib);
+
+/**
+ * @brief
+ *  Add/remove the interface from the accepting list of the special MFIB entries
+ */
+extern void ip6_mfib_interface_enable_disable(u32 sw_if_index,
+                                              int is_enable);
+
+/**
+ * @brief Get the FIB at the given index
+ */
+static inline ip6_mfib_t *
+ip6_mfib_get (u32 index)
+{
+    return (&(pool_elt_at_index(ip6_main.mfibs, index)->v6));
+}
+
+/**
+ * @brief Get or create an IPv4 fib.
+ *
+ * Get or create an IPv4 fib with the provided table ID.
+ *
+ * @param table_id
+ *      When set to \c ~0, an arbitrary and unused fib ID is picked
+ *      and can be retrieved with \c ret->table_id.
+ *      Otherwise, the fib ID to be used to retrieve or create the desired fib.
+ * @returns A pointer to the retrieved or created fib.
+ *
+ */
+extern u32 ip6_mfib_table_find_or_create_and_lock(u32 table_id);
+extern u32 ip6_mfib_table_create_and_lock(void);
+
+
+static inline
+u32 ip6_mfib_index_from_table_id (u32 table_id)
+{
+  ip6_main_t * im = &ip6_main;
+  uword * p;
+
+  p = hash_get (im->mfib_index_by_table_id, table_id);
+  if (!p)
+    return ~0;
+
+  return p[0];
+}
+
+extern u32 ip6_mfib_table_get_index_for_sw_if_index(u32 sw_if_index);
+
+/**
+ * @brief Data-plane lookup function
+ */
+extern fib_node_index_t ip6_mfib_table_lookup2(const ip6_mfib_t *mfib,
+                                               const ip6_address_t *src,
+                                               const ip6_address_t *grp);
+
+#endif
+
diff --git a/src/vnet/mfib/mfib_entry.c b/src/vnet/mfib/mfib_entry.c
new file mode 100644
index 00000000..479ce5f1
--- /dev/null
+++ b/src/vnet/mfib/mfib_entry.c
@@ -0,0 +1,1096 @@
+/*
+ * Copyright (c) 2016 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <vlib/vlib.h>
+
+#include <vnet/mfib/mfib_entry.h>
+#include <vnet/fib/fib_path_list.h>
+
+#include <vnet/dpo/drop_dpo.h>
+#include <vnet/dpo/replicate_dpo.h>
+
+/**
+ * Debug macro
+ */
+#ifdef MFIB_DEBUG
+#DEFIne MFIB_ENTRY_DBG(_e, _fmt, _args...)		\
+{                                                       \
+    u8*__tmp = NULL;					\
+    __tmp = format(__tmp, "e:[%d:%U",                   \
+                   mfib_entry_get_index(_e),		\
+                   format_ip46_address,			\
+                   &_e->mfe_prefix.fp_grp_addr,		\
+                   IP46_TYPE_ANY);			\
+    __tmp = format(__tmp, "/%d,",			\
+                   _e->mfe_prefix.fp_len);		\
+    __tmp = format(__tmp, "%U]",                        \
+                   mfib_entry_get_index(_e),		\
+                   format_ip46_address,			\
+                   &_e->mfe_prefix.fp_src_addr,		\
+                   IP46_TYPE_ANY);			\
+    __tmp = format(__tmp, _fmt, ##_args);		\
+    clib_warning("%s", __tmp);				\
+    vec_free(__tmp);					\
+}
+#else
+#define MFIB_ENTRY_DBG(_e, _fmt, _args...)
+#endif
+
+/**
+ * The source of an MFIB entry
+ */
+typedef struct mfib_entry_src_t_
+{
+    /**
+     * Which source this is
+     */
+    mfib_source_t mfes_src;
+
+    /**
+     * The path-list of forwarding interfaces
+     */
+    fib_node_index_t mfes_pl;
+
+    /**
+     * Route flags
+     */
+    mfib_entry_flags_t mfes_flags;
+
+    /**
+     * The hash table of all interfaces
+     */
+    mfib_itf_t *mfes_itfs;
+} mfib_entry_src_t;
+
+/**
+ * String names for each source
+ */
+static const char *mfib_source_names[] = MFIB_SOURCE_NAMES;
+
+/*
+ * Pool for all fib_entries
+ */
+mfib_entry_t *mfib_entry_pool;
+
+static fib_node_t *
+mfib_entry_get_node (fib_node_index_t index)
+{
+    return ((fib_node_t*)mfib_entry_get(index));
+}
+
+static fib_protocol_t
+mfib_entry_get_proto (const mfib_entry_t * mfib_entry)
+{
+    return (mfib_entry->mfe_prefix.fp_proto);
+}
+
+fib_forward_chain_type_t
+mfib_entry_get_default_chain_type (const mfib_entry_t *mfib_entry)
+{
+    switch (mfib_entry->mfe_prefix.fp_proto)
+    {
+    case FIB_PROTOCOL_IP4:
+        return (FIB_FORW_CHAIN_TYPE_MCAST_IP4);
+    case FIB_PROTOCOL_IP6:
+        return (FIB_FORW_CHAIN_TYPE_MCAST_IP6);
+    case FIB_PROTOCOL_MPLS:
+        ASSERT(0);
+        break;
+    }
+    return (FIB_FORW_CHAIN_TYPE_MCAST_IP4);
+}
+
+static u8 *
+format_mfib_entry_dpo (u8 * s, va_list * args)
+{
+    index_t fei = va_arg(*args, index_t);
+    CLIB_UNUSED(u32 indent) = va_arg(*args, u32);
+
+    return (format(s, "%U",
+                   format_mfib_entry, fei,
+                   MFIB_ENTRY_FORMAT_BRIEF));
+}
+
+u8 *
+format_mfib_entry (u8 * s, va_list * args)
+{
+    fib_node_index_t fei, mfi;
+    mfib_entry_t *mfib_entry;
+    mfib_entry_src_t *msrc;
+    u32 sw_if_index;
+    int level;
+
+    fei = va_arg (*args, fib_node_index_t);
+    level = va_arg (*args, int);
+    mfib_entry = mfib_entry_get(fei);
+
+    s = format (s, "%U", format_mfib_prefix, &mfib_entry->mfe_prefix);
+    s = format (s, ": %U", format_mfib_entry_flags, mfib_entry->mfe_flags);
+
+    if (level >= MFIB_ENTRY_FORMAT_DETAIL)
+    {
+        s = format (s, "\n");
+        s = format (s, " fib:%d", mfib_entry->mfe_fib_index);
+        s = format (s, " index:%d", mfib_entry_get_index(mfib_entry));
+        s = format (s, " locks:%d\n", mfib_entry->mfe_node.fn_locks);
+        vec_foreach(msrc, mfib_entry->mfe_srcs)
+        {
+            s = format (s, "  src:%s", mfib_source_names[msrc->mfes_src]);
+            s = format (s, ": %U\n", format_mfib_entry_flags, msrc->mfes_flags);
+            if (FIB_NODE_INDEX_INVALID != msrc->mfes_pl)
+            {
+                s = fib_path_list_format(msrc->mfes_pl, s);
+            }
+            hash_foreach(sw_if_index, mfi, msrc->mfes_itfs,
+            ({
+                s = format(s, "    %U\n", format_mfib_itf, mfi);
+            }));
+        }
+    }
+
+    s = format(s, "\n  Interfaces:");
+    hash_foreach(sw_if_index, mfi, mfib_entry->mfe_itfs,
+    ({
+        s = format(s, "\n  %U", format_mfib_itf, mfi);
+    }));
+
+    s = format(s, "\n  %U-chain\n  %U",
+               format_fib_forw_chain_type,
+               mfib_entry_get_default_chain_type(mfib_entry),
+               format_dpo_id,
+               &mfib_entry->mfe_rep,
+               2);
+    s = format(s, "\n");
+
+    if (level >= MFIB_ENTRY_FORMAT_DETAIL2)
+    {
+        s = format(s, "\nchildren:");
+        s = fib_node_children_format(mfib_entry->mfe_node.fn_children, s);
+    }
+
+    return (s);
+}
+
+static mfib_entry_t*
+mfib_entry_from_fib_node (fib_node_t *node)
+{
+#if CLIB_DEBUG > 0
+    ASSERT(FIB_NODE_TYPE_MFIB_ENTRY == node->fn_type);
+#endif
+    return ((mfib_entry_t*)node);
+}
+
+static int
+mfib_entry_src_cmp_for_sort (void * v1,
+                             void * v2)
+{
+    mfib_entry_src_t *esrc1 = v1, *esrc2 = v2;
+
+    return (esrc1->mfes_src - esrc2->mfes_src);
+}
+
+static void
+mfib_entry_src_init (mfib_entry_t *mfib_entry,
+                     mfib_source_t source)
+
+{
+    mfib_entry_src_t esrc = {
+        .mfes_pl = FIB_NODE_INDEX_INVALID,
+        .mfes_flags = MFIB_ENTRY_FLAG_NONE,
+        .mfes_src = source,
+    };
+
+    vec_add1(mfib_entry->mfe_srcs, esrc);
+    vec_sort_with_function(mfib_entry->mfe_srcs,
+                           mfib_entry_src_cmp_for_sort);
+}
+
+static mfib_entry_src_t *
+mfib_entry_src_find (const mfib_entry_t *mfib_entry,
+                    mfib_source_t source,
+                    u32 *index)
+
+{
+    mfib_entry_src_t *esrc;
+    int ii;
+
+    ii = 0;
+    vec_foreach(esrc, mfib_entry->mfe_srcs)
+    {
+        if (esrc->mfes_src == source)
+        {
+            if (NULL != index)
+            {
+                *index = ii;
+            }
+            return (esrc);
+        }
+        else
+        {
+            ii++;
+        }
+    }
+
+    return (NULL);
+}
+
+static mfib_entry_src_t *
+mfib_entry_src_find_or_create (mfib_entry_t *mfib_entry,
+                              mfib_source_t source)
+{
+    mfib_entry_src_t *esrc;
+
+    esrc = mfib_entry_src_find(mfib_entry, source, NULL);
+
+    if (NULL == esrc)
+    {
+        mfib_entry_src_init(mfib_entry, source);
+    }
+
+    return (mfib_entry_src_find(mfib_entry, source, NULL));
+}
+
+static mfib_entry_src_t*
+mfib_entry_get_best_src (const mfib_entry_t *mfib_entry)
+{
+    mfib_entry_src_t *bsrc;
+
+    /*
+     * the enum of sources is deliberately arranged in priority order
+     */
+    if (0 == vec_len(mfib_entry->mfe_srcs))
+    {
+        bsrc = NULL;
+    }
+    else
+    {
+        bsrc = vec_elt_at_index(mfib_entry->mfe_srcs, 0);
+    }
+
+    return (bsrc);
+}
+
+static void
+mfib_entry_src_flush (mfib_entry_src_t *msrc)
+{
+    u32 sw_if_index;
+    index_t mfii;
+
+    hash_foreach(sw_if_index, mfii, msrc->mfes_itfs,
+    ({
+        mfib_itf_delete(mfib_itf_get(mfii));
+    }));
+}
+
+static void
+mfib_entry_src_remove (mfib_entry_t *mfib_entry,
+                       mfib_source_t source)
+
+{
+    mfib_entry_src_t *msrc;
+    u32 index = ~0;
+
+    msrc = mfib_entry_src_find(mfib_entry, source, &index);
+
+    if (NULL != msrc)
+    {
+        mfib_entry_src_flush(msrc);
+        vec_del1(mfib_entry->mfe_srcs, index);
+    }
+}
+
+static int
+mfib_entry_src_n_itfs (const mfib_entry_src_t *msrc)
+{
+    return (hash_elts(msrc->mfes_itfs));
+}
+
+
+static void
+mfib_entry_last_lock_gone (fib_node_t *node)
+{
+    mfib_entry_t *mfib_entry;
+    mfib_entry_src_t *msrc;
+
+    mfib_entry = mfib_entry_from_fib_node(node);
+
+    dpo_reset(&mfib_entry->mfe_rep);
+
+    MFIB_ENTRY_DBG(mfib_entry, "last-lock");
+
+    vec_foreach(msrc, mfib_entry->mfe_srcs)
+    {
+        mfib_entry_src_flush(msrc);
+    }
+
+    fib_path_list_unlock(mfib_entry->mfe_parent);
+    vec_free(mfib_entry->mfe_srcs);
+
+    fib_node_deinit(&mfib_entry->mfe_node);
+    pool_put(mfib_entry_pool, mfib_entry);
+}
+
+/*
+ * mfib_entry_back_walk_notify
+ *
+ * A back walk has reach this entry.
+ */
+static fib_node_back_walk_rc_t
+mfib_entry_back_walk_notify (fib_node_t *node,
+                            fib_node_back_walk_ctx_t *ctx)
+{
+    // FIXME - re-evalute
+
+    return (FIB_NODE_BACK_WALK_CONTINUE);
+}
+
+static void
+mfib_entry_show_memory (void)
+{
+    fib_show_memory_usage("multicast-Entry",
+                          pool_elts(mfib_entry_pool),
+                          pool_len(mfib_entry_pool),
+                          sizeof(mfib_entry_t));
+}
+
+/*
+ * The MFIB entry's graph node virtual function table
+ */
+static const fib_node_vft_t mfib_entry_vft = {
+    .fnv_get = mfib_entry_get_node,
+    .fnv_last_lock = mfib_entry_last_lock_gone,
+    .fnv_back_walk = mfib_entry_back_walk_notify,
+    .fnv_mem_show = mfib_entry_show_memory,
+};
+
+u32
+mfib_entry_child_add (fib_node_index_t mfib_entry_index,
+                      fib_node_type_t child_type,
+                      fib_node_index_t child_index)
+{
+    return (fib_node_child_add(FIB_NODE_TYPE_MFIB_ENTRY,
+                               mfib_entry_index,
+                               child_type,
+                               child_index));
+};
+
+void
+mfib_entry_child_remove (fib_node_index_t mfib_entry_index,
+                         u32 sibling_index)
+{
+    fib_node_child_remove(FIB_NODE_TYPE_MFIB_ENTRY,
+                          mfib_entry_index,
+                          sibling_index);
+}
+
+static mfib_entry_t *
+mfib_entry_alloc (u32 fib_index,
+                  const mfib_prefix_t *prefix,
+                  fib_node_index_t *mfib_entry_index)
+{
+    mfib_entry_t *mfib_entry;
+
+    pool_get(mfib_entry_pool, mfib_entry);
+    memset(mfib_entry, 0, sizeof(*mfib_entry));
+
+    fib_node_init(&mfib_entry->mfe_node,
+                  FIB_NODE_TYPE_MFIB_ENTRY);
+
+    mfib_entry->mfe_fib_index = fib_index;
+    mfib_entry->mfe_prefix = *prefix;
+    mfib_entry->mfe_parent = FIB_NODE_INDEX_INVALID;
+
+    dpo_reset(&mfib_entry->mfe_rep);
+
+    *mfib_entry_index = mfib_entry_get_index(mfib_entry);
+
+    MFIB_ENTRY_DBG(mfib_entry, "alloc");
+
+    return (mfib_entry);
+}
+
+typedef struct mfib_entry_collect_forwarding_ctx_t_
+{
+    load_balance_path_t * next_hops;
+    fib_forward_chain_type_t fct;
+} mfib_entry_collect_forwarding_ctx_t;
+
+static int
+mfib_entry_src_collect_forwarding (fib_node_index_t pl_index,
+                                   fib_node_index_t path_index,
+                                   void *arg)
+{
+    mfib_entry_collect_forwarding_ctx_t *ctx;
+    load_balance_path_t *nh;
+
+    ctx = arg;
+
+    /*
+     * if the path is not resolved, don't include it.
+     */
+    if (!fib_path_is_resolved(path_index))
+    {
+        return (!0);
+    }
+
+    switch (ctx->fct)
+    {
+    case FIB_FORW_CHAIN_TYPE_MCAST_IP4:
+    case FIB_FORW_CHAIN_TYPE_MCAST_IP6:
+        /*
+         * EOS traffic with no label to stack, we need the IP Adj
+         */
+        vec_add2(ctx->next_hops, nh, 1);
+
+        nh->path_index = path_index;
+        nh->path_weight = fib_path_get_weight(path_index);
+        fib_path_contribute_forwarding(path_index, ctx->fct, &nh->path_dpo);
+        break;
+
+    case FIB_FORW_CHAIN_TYPE_UNICAST_IP4:
+    case FIB_FORW_CHAIN_TYPE_UNICAST_IP6:
+    case FIB_FORW_CHAIN_TYPE_MPLS_NON_EOS:
+    case FIB_FORW_CHAIN_TYPE_MPLS_EOS:
+    case FIB_FORW_CHAIN_TYPE_ETHERNET:
+        ASSERT(0);
+        break;
+    }
+
+    return (!0);
+}
+
+static void
+mfib_entry_stack (mfib_entry_t *mfib_entry)
+{
+    mfib_entry_collect_forwarding_ctx_t ctx = {
+        .next_hops = NULL,
+        .fct = mfib_entry_get_default_chain_type(mfib_entry),
+    };
+    dpo_proto_t dp;
+
+    dp = fib_proto_to_dpo(mfib_entry_get_proto(mfib_entry));
+
+    if (FIB_NODE_INDEX_INVALID != mfib_entry->mfe_parent)
+    {
+        fib_path_list_walk(mfib_entry->mfe_parent,
+                           mfib_entry_src_collect_forwarding,
+                           &ctx);
+
+        if (!dpo_id_is_valid(&mfib_entry->mfe_rep) ||
+            dpo_is_drop(&mfib_entry->mfe_rep))
+        {
+            dpo_id_t tmp_dpo = DPO_INVALID;
+
+            dpo_set(&tmp_dpo,
+                    DPO_REPLICATE, dp,
+                    replicate_create(0, dp));
+
+            dpo_stack(DPO_MFIB_ENTRY, dp,
+                      &mfib_entry->mfe_rep,
+                      &tmp_dpo);
+
+            dpo_reset(&tmp_dpo);
+        }
+        replicate_multipath_update(&mfib_entry->mfe_rep,
+                                   ctx.next_hops);
+    }
+    else
+    {
+        dpo_stack(DPO_MFIB_ENTRY, dp,
+                  &mfib_entry->mfe_rep,
+                  drop_dpo_get(dp));
+    }
+}
+
+static void
+mfib_entry_forwarding_path_add (mfib_entry_src_t *msrc,
+                                const fib_route_path_t *rpath)
+{
+    fib_node_index_t old_pl_index;
+    fib_route_path_t *rpaths;
+
+    /*
+     * path-lists require a vector of paths
+     */
+    rpaths = NULL;
+    vec_add1(rpaths, rpath[0]);
+
+    old_pl_index = msrc->mfes_pl;
+
+    if (FIB_NODE_INDEX_INVALID == msrc->mfes_pl)
+    {
+        msrc->mfes_pl =
+            fib_path_list_create(FIB_PATH_LIST_FLAG_NO_URPF,
+                                 rpaths);
+    }
+    else
+    {
+        msrc->mfes_pl =
+            fib_path_list_copy_and_path_add(msrc->mfes_pl,
+                                            FIB_PATH_LIST_FLAG_NO_URPF,
+                                            rpaths);
+    }
+    fib_path_list_lock(msrc->mfes_pl);
+    fib_path_list_unlock(old_pl_index);
+
+    vec_free(rpaths);
+}
+
+static int
+mfib_entry_forwarding_path_remove (mfib_entry_src_t *msrc,
+                                   const fib_route_path_t *rpath)
+{
+    fib_node_index_t old_pl_index;
+    fib_route_path_t *rpaths;
+
+    /*
+     * path-lists require a vector of paths
+     */
+    rpaths = NULL;
+    vec_add1(rpaths, rpath[0]);
+
+    old_pl_index = msrc->mfes_pl;
+
+    msrc->mfes_pl =
+        fib_path_list_copy_and_path_remove(msrc->mfes_pl,
+                                           FIB_PATH_LIST_FLAG_NONE,
+                                           rpaths);
+
+    fib_path_list_lock(msrc->mfes_pl);
+    fib_path_list_unlock(old_pl_index);
+
+    vec_free(rpaths);
+
+    return (FIB_NODE_INDEX_INVALID != msrc->mfes_pl);
+}
+
+static void
+mfib_entry_recalculate_forwarding (mfib_entry_t *mfib_entry)
+{
+    fib_node_index_t old_pl_index;
+    mfib_entry_src_t *bsrc;
+
+    old_pl_index = mfib_entry->mfe_parent;
+
+    /*
+     * copy the forwarding data from the bast source
+     */
+    bsrc = mfib_entry_get_best_src(mfib_entry);
+
+    if (NULL == bsrc)
+    {
+        mfib_entry->mfe_parent = FIB_NODE_INDEX_INVALID;
+    }
+    else
+    {
+        mfib_entry->mfe_parent = bsrc->mfes_pl;
+        mfib_entry->mfe_flags = bsrc->mfes_flags;
+        mfib_entry->mfe_itfs = bsrc->mfes_itfs;
+    }
+
+    /*
+     * re-stack the entry on the best forwarding info.
+     */
+    if (old_pl_index != mfib_entry->mfe_parent ||
+        FIB_NODE_INDEX_INVALID == old_pl_index)
+    {
+        mfib_entry_stack(mfib_entry);
+
+        fib_path_list_lock(mfib_entry->mfe_parent);
+        fib_path_list_unlock(old_pl_index);
+    }
+}
+
+
+fib_node_index_t
+mfib_entry_create (u32 fib_index,
+                   mfib_source_t source,
+                   const mfib_prefix_t *prefix,
+                   mfib_entry_flags_t entry_flags)
+{
+    fib_node_index_t mfib_entry_index;
+    mfib_entry_t *mfib_entry;
+    mfib_entry_src_t *msrc;
+
+    mfib_entry = mfib_entry_alloc(fib_index, prefix,
+                                  &mfib_entry_index);
+    msrc = mfib_entry_src_find_or_create(mfib_entry, source);
+    msrc->mfes_flags = entry_flags;
+
+    mfib_entry_recalculate_forwarding(mfib_entry);
+
+    return (mfib_entry_index);
+}
+
+static int
+mfib_entry_ok_for_delete (mfib_entry_t *mfib_entry)
+{
+    return (0 == vec_len(mfib_entry->mfe_srcs));
+}
+
+static int
+mfib_entry_src_ok_for_delete (const mfib_entry_src_t *msrc)
+{
+    return ((MFIB_ENTRY_FLAG_NONE == msrc->mfes_flags &&
+             0 == mfib_entry_src_n_itfs(msrc)));
+}
+
+int
+mfib_entry_update (fib_node_index_t mfib_entry_index,
+                   mfib_source_t source,
+                   mfib_entry_flags_t entry_flags)
+{
+    mfib_entry_t *mfib_entry;
+    mfib_entry_src_t *msrc;
+
+    mfib_entry = mfib_entry_get(mfib_entry_index);
+    msrc = mfib_entry_src_find_or_create(mfib_entry, source);
+    msrc->mfes_flags = entry_flags;
+
+    if (mfib_entry_src_ok_for_delete(msrc))
+    {
+        /*
+         * this source has no interfaces and no flags.
+         * it has nothing left to give - remove it
+         */
+        mfib_entry_src_remove(mfib_entry, source);
+    }
+
+    mfib_entry_recalculate_forwarding(mfib_entry);
+
+    return (mfib_entry_ok_for_delete(mfib_entry));
+}
+
+static void
+mfib_entry_itf_add (mfib_entry_src_t *msrc,
+                    u32 sw_if_index,
+                    index_t mi)
+{
+    hash_set(msrc->mfes_itfs, sw_if_index, mi);
+}
+
+static void
+mfib_entry_itf_remove (mfib_entry_src_t *msrc,
+                       u32 sw_if_index)
+{
+    mfib_itf_t *mfi;
+
+    mfi = mfib_entry_itf_find(msrc->mfes_itfs, sw_if_index);
+
+    mfib_itf_delete(mfi);
+
+    hash_unset(msrc->mfes_itfs, sw_if_index);
+}
+
+void
+mfib_entry_path_update (fib_node_index_t mfib_entry_index,
+                        mfib_source_t source,
+                        const fib_route_path_t *rpath,
+                        mfib_itf_flags_t itf_flags)
+{
+    mfib_entry_t *mfib_entry;
+    mfib_entry_src_t *msrc;
+    mfib_itf_t *mfib_itf;
+
+    mfib_entry = mfib_entry_get(mfib_entry_index);
+    ASSERT(NULL != mfib_entry);
+    msrc = mfib_entry_src_find_or_create(mfib_entry, source);
+
+    /*
+     * search for the interface in the current set
+     */
+    mfib_itf = mfib_entry_itf_find(msrc->mfes_itfs,
+                                   rpath[0].frp_sw_if_index);
+
+    if (NULL == mfib_itf)
+    {
+        /*
+         * this is a path we do not yet have. If it is forwarding then we
+         * add it to the replication set
+         */
+        if (itf_flags & MFIB_ITF_FLAG_FORWARD)
+        {
+            mfib_entry_forwarding_path_add(msrc, rpath);
+        }
+        /*
+         * construct a new ITF for this entry's list
+         */
+        mfib_entry_itf_add(msrc,
+                           rpath[0].frp_sw_if_index,
+                           mfib_itf_create(rpath[0].frp_sw_if_index,
+                                           itf_flags));
+    }
+    else
+    {
+        int was_forwarding = !!(mfib_itf->mfi_flags & MFIB_ITF_FLAG_FORWARD);
+        int is_forwarding  = !!(itf_flags & MFIB_ITF_FLAG_FORWARD);
+
+        if (!was_forwarding && is_forwarding)
+        {
+            mfib_entry_forwarding_path_add(msrc, rpath);
+        }
+        else if (was_forwarding && !is_forwarding)
+        {
+            mfib_entry_forwarding_path_remove(msrc, rpath);
+        }
+        /*
+         * packets in flight see these updates.
+         */
+        mfib_itf->mfi_flags = itf_flags;
+    }
+
+    mfib_entry_recalculate_forwarding(mfib_entry);
+}
+
+/*
+ * mfib_entry_path_remove
+ *
+ * remove a path from the entry.
+ * return the mfib_entry's index if it is still present, INVALID otherwise.
+ */
+int
+mfib_entry_path_remove (fib_node_index_t mfib_entry_index,
+                        mfib_source_t source,
+                        const fib_route_path_t *rpath)
+{
+    mfib_entry_t *mfib_entry;
+    mfib_entry_src_t *msrc;
+    mfib_itf_t *mfib_itf;
+
+    mfib_entry = mfib_entry_get(mfib_entry_index);
+    ASSERT(NULL != mfib_entry);
+    msrc = mfib_entry_src_find(mfib_entry, source, NULL);
+
+    if (NULL == msrc)
+    {
+        /*
+         * there are no paths left for this source
+         */
+        return (mfib_entry_ok_for_delete(mfib_entry));
+    }
+
+    /*
+     * search for the interface in the current set
+     */
+    mfib_itf = mfib_entry_itf_find(msrc->mfes_itfs,
+                                   rpath[0].frp_sw_if_index);
+
+    if (NULL == mfib_itf)
+    {
+        /*
+         * removing a path that does not exist
+         */
+        return (mfib_entry_ok_for_delete(mfib_entry));
+    }
+
+    /*
+     * we have this path. If it is forwarding then we
+     * remove it to the replication set
+     */
+    if (mfib_itf->mfi_flags & MFIB_ITF_FLAG_FORWARD)
+    {
+        mfib_entry_forwarding_path_remove(msrc, rpath);
+    }
+
+    /*
+     * remove the interface/path from this entry's list
+     */
+    mfib_entry_itf_remove(msrc, rpath[0].frp_sw_if_index);
+
+    if (mfib_entry_src_ok_for_delete(msrc))
+    {
+        /*
+         * this source has no interfaces and no flags.
+         * it has nothing left to give - remove it
+         */
+        mfib_entry_src_remove(mfib_entry, source);
+    }
+
+    mfib_entry_recalculate_forwarding(mfib_entry);
+
+    return (mfib_entry_ok_for_delete(mfib_entry));
+}
+
+/**
+ * mfib_entry_delete
+ *
+ * The source is withdrawing all the paths it provided
+ */
+int
+mfib_entry_delete (fib_node_index_t mfib_entry_index,
+                   mfib_source_t source)
+{
+    mfib_entry_t *mfib_entry;
+
+    mfib_entry = mfib_entry_get(mfib_entry_index);
+    mfib_entry_src_remove(mfib_entry, source);
+
+    mfib_entry_recalculate_forwarding(mfib_entry);
+
+    return (mfib_entry_ok_for_delete(mfib_entry));
+}
+
+static int
+fib_ip4_address_compare (ip4_address_t * a1,
+                         ip4_address_t * a2)
+{
+    /*
+     * IP addresses are unsiged ints. the return value here needs to be signed
+     * a simple subtraction won't cut it.
+     * If the addresses are the same, the sort order is undefiend, so phoey.
+     */
+    return ((clib_net_to_host_u32(a1->data_u32) >
+             clib_net_to_host_u32(a2->data_u32) ) ?
+            1 : -1);
+}
+
+static int
+fib_ip6_address_compare (ip6_address_t * a1,
+                         ip6_address_t * a2)
+{
+  int i;
+  for (i = 0; i < ARRAY_LEN (a1->as_u16); i++)
+  {
+      int cmp = (clib_net_to_host_u16 (a1->as_u16[i]) -
+                 clib_net_to_host_u16 (a2->as_u16[i]));
+      if (cmp != 0)
+          return cmp;
+  }
+  return 0;
+}
+
+static int
+mfib_entry_cmp (fib_node_index_t mfib_entry_index1,
+                fib_node_index_t mfib_entry_index2)
+{
+    mfib_entry_t *mfib_entry1, *mfib_entry2;
+    int cmp = 0;
+
+    mfib_entry1 = mfib_entry_get(mfib_entry_index1);
+    mfib_entry2 = mfib_entry_get(mfib_entry_index2);
+
+    switch (mfib_entry1->mfe_prefix.fp_proto)
+    {
+    case FIB_PROTOCOL_IP4:
+        cmp = fib_ip4_address_compare(&mfib_entry1->mfe_prefix.fp_grp_addr.ip4,
+                                      &mfib_entry2->mfe_prefix.fp_grp_addr.ip4);
+
+        if (0 == cmp)
+        {
+            cmp = fib_ip4_address_compare(&mfib_entry1->mfe_prefix.fp_src_addr.ip4,
+                                          &mfib_entry2->mfe_prefix.fp_src_addr.ip4);
+        }
+        break;
+    case FIB_PROTOCOL_IP6:
+        cmp = fib_ip6_address_compare(&mfib_entry1->mfe_prefix.fp_grp_addr.ip6,
+                                      &mfib_entry2->mfe_prefix.fp_grp_addr.ip6);
+
+        if (0 == cmp)
+        {
+            cmp = fib_ip6_address_compare(&mfib_entry1->mfe_prefix.fp_src_addr.ip6,
+                                          &mfib_entry2->mfe_prefix.fp_src_addr.ip6);
+        }
+        break;
+    case FIB_PROTOCOL_MPLS:
+        ASSERT(0);
+        cmp = 0;
+        break;
+    }
+
+    if (0 == cmp) {
+        cmp = (mfib_entry1->mfe_prefix.fp_len - mfib_entry2->mfe_prefix.fp_len);
+    }
+    return (cmp);
+}
+
+int
+mfib_entry_cmp_for_sort (void *i1, void *i2)
+{
+    fib_node_index_t *mfib_entry_index1 = i1, *mfib_entry_index2 = i2;
+
+    return (mfib_entry_cmp(*mfib_entry_index1,
+                           *mfib_entry_index2));
+}
+
+void
+mfib_entry_lock (fib_node_index_t mfib_entry_index)
+{
+    mfib_entry_t *mfib_entry;
+
+    mfib_entry = mfib_entry_get(mfib_entry_index);
+
+    fib_node_lock(&mfib_entry->mfe_node);
+}
+
+void
+mfib_entry_unlock (fib_node_index_t mfib_entry_index)
+{
+    mfib_entry_t *mfib_entry;
+
+    mfib_entry = mfib_entry_get(mfib_entry_index);
+
+    fib_node_unlock(&mfib_entry->mfe_node);
+}
+
+static void
+mfib_entry_dpo_lock (dpo_id_t *dpo)
+{
+}
+static void
+mfib_entry_dpo_unlock (dpo_id_t *dpo)
+{
+}
+
+const static dpo_vft_t mfib_entry_dpo_vft = {
+    .dv_lock = mfib_entry_dpo_lock,
+    .dv_unlock = mfib_entry_dpo_unlock,
+    .dv_format = format_mfib_entry_dpo,
+    .dv_mem_show = mfib_entry_show_memory,
+};
+
+const static char* const mfib_entry_ip4_nodes[] =
+{
+    "ip4-mfib-forward-rpf",
+    NULL,
+};
+const static char* const mfib_entry_ip6_nodes[] =
+{
+    "ip6-mfib-forward-rpf",
+    NULL,
+};
+
+const static char* const * const mfib_entry_nodes[DPO_PROTO_NUM] =
+{
+    [DPO_PROTO_IP4]  = mfib_entry_ip4_nodes,
+    [DPO_PROTO_IP6]  = mfib_entry_ip6_nodes,
+};
+
+void
+mfib_entry_module_init (void)
+{
+    fib_node_register_type (FIB_NODE_TYPE_MFIB_ENTRY, &mfib_entry_vft);
+    dpo_register(DPO_MFIB_ENTRY, &mfib_entry_dpo_vft, mfib_entry_nodes);
+}
+
+void
+mfib_entry_encode (fib_node_index_t mfib_entry_index,
+                  fib_route_path_encode_t **api_rpaths)
+{
+    mfib_entry_t *mfib_entry;
+
+    mfib_entry = mfib_entry_get(mfib_entry_index);
+    fib_path_list_walk(mfib_entry->mfe_parent, fib_path_encode, api_rpaths);
+}
+
+void
+mfib_entry_get_prefix (fib_node_index_t mfib_entry_index,
+                      mfib_prefix_t *pfx)
+{
+    mfib_entry_t *mfib_entry;
+
+    mfib_entry = mfib_entry_get(mfib_entry_index);
+    *pfx = mfib_entry->mfe_prefix;
+}
+
+u32
+mfib_entry_get_fib_index (fib_node_index_t mfib_entry_index)
+{
+    mfib_entry_t *mfib_entry;
+
+    mfib_entry = mfib_entry_get(mfib_entry_index);
+
+    return (mfib_entry->mfe_fib_index);
+}
+
+void
+mfib_entry_contribute_forwarding (fib_node_index_t mfib_entry_index,
+                                  fib_forward_chain_type_t type,
+                                  dpo_id_t *dpo)
+{
+    /*
+     * An IP mFIB entry can only provide a forwarding chain that
+     * is the same IP proto as the prefix.
+     * No use-cases (i know of) for other combinations.
+     */
+    mfib_entry_t *mfib_entry;
+    dpo_proto_t dp;
+
+    mfib_entry = mfib_entry_get(mfib_entry_index);
+
+    dp = fib_proto_to_dpo(mfib_entry->mfe_prefix.fp_proto);
+
+    if (type == fib_forw_chain_type_from_dpo_proto(dp))
+    {
+        dpo_copy(dpo, &mfib_entry->mfe_rep);
+    }
+    else
+    {
+        dpo_copy(dpo, drop_dpo_get(dp));
+    }
+}
+
+u32
+mfib_entry_pool_size (void)
+{
+    return (pool_elts(mfib_entry_pool));
+}
+
+static clib_error_t *
+show_mfib_entry_command (vlib_main_t * vm,
+                        unformat_input_t * input,
+                        vlib_cli_command_t * cmd)
+{
+    fib_node_index_t fei;
+
+    if (unformat (input, "%d", &fei))
+    {
+        /*
+         * show one in detail
+         */
+        if (!pool_is_free_index(mfib_entry_pool, fei))
+        {
+            vlib_cli_output (vm, "%d@%U",
+                             fei,
+                             format_mfib_entry, fei,
+                             MFIB_ENTRY_FORMAT_DETAIL2);
+        }
+        else
+        {
+            vlib_cli_output (vm, "entry %d invalid", fei);
+        }
+    }
+    else
+    {
+        /*
+         * show all
+         */
+        vlib_cli_output (vm, "FIB Entries:");
+        pool_foreach_index(fei, mfib_entry_pool,
+        ({
+            vlib_cli_output (vm, "%d@%U",
+                             fei,
+                             format_mfib_entry, fei,
+                             MFIB_ENTRY_FORMAT_BRIEF);
+        }));
+    }
+
+    return (NULL);
+}
+
+VLIB_CLI_COMMAND (show_mfib_entry, static) = {
+  .path = "show mfib entry",
+  .function = show_mfib_entry_command,
+  .short_help = "show mfib entry",
+};
diff --git a/src/vnet/mfib/mfib_entry.h b/src/vnet/mfib/mfib_entry.h
new file mode 100644
index 00000000..cc5d5326
--- /dev/null
+++ b/src/vnet/mfib/mfib_entry.h
@@ -0,0 +1,172 @@
+/*
+ * Copyright (c) 2016 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef __MFIB_ENTRY_H__
+#define __MFIB_ENTRY_H__
+
+#include <vnet/fib/fib_node.h>
+#include <vnet/mfib/mfib_types.h>
+#include <vnet/mfib/mfib_itf.h>
+#include <vnet/ip/ip.h>
+#include <vnet/dpo/dpo.h>
+
+/**
+ * An entry in a FIB table.
+ *
+ * This entry represents a route added to the FIB that is stored
+ * in one of the FIB tables.
+ */
+typedef struct mfib_entry_t_ {
+    CLIB_CACHE_LINE_ALIGN_MARK(cacheline0);
+    /**
+     * Base class. The entry's node representation in the graph.
+     */
+    fib_node_t mfe_node;
+    /**
+     * The prefix of the route
+     */
+    mfib_prefix_t mfe_prefix;
+    /**
+     * The index of the FIB table this entry is in
+     */
+    u32 mfe_fib_index;
+    /**
+     * the path-list for which this entry is a child. This is also the path-list
+     * that is contributing forwarding for this entry.
+     */
+    fib_node_index_t mfe_parent;
+    /**
+     * index of this entry in the parent's child list.
+     * This is set when this entry is added as a child, but can also
+     * be changed by the parent as it manages its list.
+     */
+    u32 mfe_sibling;
+
+    /**
+     * A vector of sources contributing forwarding
+     */
+    struct mfib_entry_src_t_ *mfe_srcs;
+
+    /**
+     * 2nd cache line has the members used in the data plane
+     */
+    CLIB_CACHE_LINE_ALIGN_MARK(cacheline1);
+
+    /**
+     * The Replicate used for forwarding.
+     */
+    dpo_id_t mfe_rep;
+
+    /**
+     * Route flags
+     */
+    mfib_entry_flags_t mfe_flags;
+
+    /**
+     * A hash table of interfaces
+     */
+    mfib_itf_t *mfe_itfs;
+} mfib_entry_t;
+
+#define MFIB_ENTRY_FORMAT_BRIEF   (0x0)
+#define MFIB_ENTRY_FORMAT_DETAIL  (0x1)
+#define MFIB_ENTRY_FORMAT_DETAIL2 (0x2)
+
+extern u8 *format_mfib_entry(u8 * s, va_list * args);
+
+
+extern fib_node_index_t mfib_entry_create(u32 fib_index,
+                                          mfib_source_t source,
+                                          const mfib_prefix_t *prefix,
+                                          mfib_entry_flags_t entry_flags);
+
+extern int mfib_entry_update(fib_node_index_t fib_entry_index,
+                             mfib_source_t source,
+                             mfib_entry_flags_t entry_flags);
+
+extern void mfib_entry_path_update(fib_node_index_t fib_entry_index,
+                                   mfib_source_t source,
+                                   const fib_route_path_t *rpath,
+                                   mfib_itf_flags_t itf_flags);
+
+
+extern int mfib_entry_path_remove(fib_node_index_t fib_entry_index,
+                                  mfib_source_t source,
+                                  const fib_route_path_t *rpath);
+
+extern int mfib_entry_delete(fib_node_index_t mfib_entry_index,
+                             mfib_source_t source);
+
+extern int mfib_entry_cmp_for_sort(void *i1, void *i2);
+
+extern u32 mfib_entry_child_add(fib_node_index_t mfib_entry_index,
+                                fib_node_type_t type,
+                                fib_node_index_t child_index);
+extern void mfib_entry_child_remove(fib_node_index_t mfib_entry_index,
+                                    u32 sibling_index);
+
+extern void mfib_entry_lock(fib_node_index_t fib_entry_index);
+extern void mfib_entry_unlock(fib_node_index_t fib_entry_index);
+
+extern void mfib_entry_get_prefix(fib_node_index_t fib_entry_index,
+                                  mfib_prefix_t *pfx);
+extern u32 mfib_entry_get_fib_index(fib_node_index_t fib_entry_index);
+
+extern void mfib_entry_contribute_forwarding(
+    fib_node_index_t mfib_entry_index,
+    fib_forward_chain_type_t type,
+    dpo_id_t *dpo);
+
+extern void mfib_entry_module_init(void);
+
+
+extern mfib_entry_t *mfib_entry_pool;
+
+static inline mfib_entry_t *
+mfib_entry_get (fib_node_index_t index)
+{
+    return (pool_elt_at_index(mfib_entry_pool, index));
+}
+static inline fib_node_index_t
+mfib_entry_get_index (const mfib_entry_t *mfe)
+{
+    return (mfe - mfib_entry_pool);
+}
+
+
+static inline mfib_itf_t *
+mfib_entry_itf_find (mfib_itf_t *itfs,
+                     u32 sw_if_index)
+{
+    uword *p;
+
+    p = hash_get(itfs, sw_if_index);
+
+    if (NULL != p)
+    {
+        return (mfib_itf_get(p[0]));
+    }
+
+    return (NULL);
+}
+
+static inline mfib_itf_t *
+mfib_entry_get_itf (const mfib_entry_t *mfe,
+                    u32 sw_if_index)
+{
+    return (mfib_entry_itf_find(mfe->mfe_itfs, sw_if_index));
+}
+
+#endif
diff --git a/src/vnet/mfib/mfib_forward.c b/src/vnet/mfib/mfib_forward.c
new file mode 100644
index 00000000..5fe0a57c
--- /dev/null
+++ b/src/vnet/mfib/mfib_forward.c
@@ -0,0 +1,512 @@
+/*
+ * Copyright (c) 2016 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <vnet/mfib/mfib_itf.h>
+#include <vnet/mfib/mfib_entry.h>
+#include <vnet/dpo/replicate_dpo.h>
+#include <vnet/mfib/ip4_mfib.h>
+#include <vnet/mfib/ip6_mfib.h>
+#include <vnet/mfib/mfib_signal.h>
+#include <vnet/fib/ip4_fib.h>
+#include <vnet/fib/ip6_fib.h>
+
+#include <vnet/ip/ip4.h>
+#include <vnet/vnet.h>
+
+typedef struct mfib_forward_lookup_trace_t_ {
+    u32 entry_index;
+    u32 fib_index;
+} mfib_forward_lookup_trace_t;
+
+static u8 *
+format_mfib_forward_lookup_trace (u8 * s, va_list * args)
+{
+    CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *);
+    CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *);
+    mfib_forward_lookup_trace_t * t = va_arg (*args, mfib_forward_lookup_trace_t *);
+
+    s = format (s, "fib %d entry %d", t->fib_index, t->entry_index);
+    return s;
+}
+
+/* Common trace function for all ip4-forward next nodes. */
+void
+mfib_forward_lookup_trace (vlib_main_t * vm,
+                           vlib_node_runtime_t * node,
+                           vlib_frame_t * frame)
+{
+    u32 * from, n_left;
+    ip4_main_t * im = &ip4_main;
+
+    n_left = frame->n_vectors;
+    from = vlib_frame_vector_args (frame);
+
+    while (n_left >= 4)
+    {
+        mfib_forward_lookup_trace_t * t0, * t1;
+        vlib_buffer_t * b0, * b1;
+        u32 bi0, bi1;
+
+        /* Prefetch next iteration. */
+        vlib_prefetch_buffer_with_index (vm, from[2], LOAD);
+        vlib_prefetch_buffer_with_index (vm, from[3], LOAD);
+
+        bi0 = from[0];
+        bi1 = from[1];
+
+        b0 = vlib_get_buffer (vm, bi0);
+        b1 = vlib_get_buffer (vm, bi1);
+
+        if (b0->flags & VLIB_BUFFER_IS_TRACED)
+        {
+            t0 = vlib_add_trace (vm, node, b0, sizeof (t0[0]));
+            t0->entry_index = vnet_buffer (b0)->ip.adj_index[VLIB_TX];
+            t0->fib_index = vec_elt (im->mfib_index_by_sw_if_index,
+                                     vnet_buffer(b1)->sw_if_index[VLIB_RX]);
+        }
+        if (b1->flags & VLIB_BUFFER_IS_TRACED)
+        {
+            t1 = vlib_add_trace (vm, node, b1, sizeof (t1[0]));
+            t1->entry_index = vnet_buffer (b1)->ip.adj_index[VLIB_TX];
+            t1->fib_index = vec_elt (im->mfib_index_by_sw_if_index,
+                                     vnet_buffer(b1)->sw_if_index[VLIB_RX]);
+        }
+        from += 2;
+        n_left -= 2;
+    }
+
+    while (n_left >= 1)
+    {
+        mfib_forward_lookup_trace_t * t0;
+        vlib_buffer_t * b0;
+        u32 bi0;
+
+        bi0 = from[0];
+
+        b0 = vlib_get_buffer (vm, bi0);
+
+        if (b0->flags & VLIB_BUFFER_IS_TRACED)
+        {
+            t0 = vlib_add_trace (vm, node, b0, sizeof (t0[0]));
+            t0->entry_index = vnet_buffer (b0)->ip.adj_index[VLIB_TX];
+            t0->fib_index = vec_elt (im->mfib_index_by_sw_if_index,
+                                     vnet_buffer(b0)->sw_if_index[VLIB_RX]);
+        }
+        from += 1;
+        n_left -= 1;
+    }
+}
+
+typedef enum mfib_forward_lookup_next_t_ {
+    MFIB_FORWARD_LOOKUP_NEXT_RPF,
+    MFIB_FORWARD_LOOKUP_N_NEXT,
+} mfib_forward_lookup_next_t;
+
+static uword
+mfib_forward_lookup (vlib_main_t * vm,
+                     vlib_node_runtime_t * node,
+                     vlib_frame_t * frame,
+                     int is_v4)
+{
+    u32 n_left_from, n_left_to_next, * from, * to_next;
+
+    from = vlib_frame_vector_args (frame);
+    n_left_from = frame->n_vectors;
+
+    while (n_left_from > 0)
+    {
+        vlib_get_next_frame (vm, node, MFIB_FORWARD_LOOKUP_NEXT_RPF,
+                             to_next, n_left_to_next);
+
+        while (n_left_from > 0 && n_left_to_next > 0)
+        {
+            fib_node_index_t mfei0;
+            vlib_buffer_t * p0;
+            u32 fib_index0;
+            u32 pi0;
+
+            pi0 = from[0];
+            to_next[0] = pi0;
+            from += 1;
+            to_next += 1;
+            n_left_to_next -= 1;
+            n_left_from -= 1;
+
+            p0 = vlib_get_buffer (vm, pi0);
+
+            if (is_v4)
+            {
+                ip4_header_t * ip0;
+
+                fib_index0 = vec_elt (ip4_main.mfib_index_by_sw_if_index,
+                                      vnet_buffer(p0)->sw_if_index[VLIB_RX]);
+                ip0 = vlib_buffer_get_current (p0);
+                mfei0 = ip4_mfib_table_lookup(ip4_mfib_get(fib_index0),
+                                              &ip0->src_address,
+                                              &ip0->dst_address,
+                                              64);
+            }
+            else
+            {
+                ip6_header_t * ip0;
+
+                fib_index0 = vec_elt (ip6_main.mfib_index_by_sw_if_index,
+                                      vnet_buffer(p0)->sw_if_index[VLIB_RX]);
+                ip0 = vlib_buffer_get_current (p0);
+                mfei0 = ip6_mfib_table_lookup2(ip6_mfib_get(fib_index0),
+                                               &ip0->src_address,
+                                               &ip0->dst_address);
+            }
+
+            vnet_buffer (p0)->ip.adj_index[VLIB_TX] = mfei0;
+        }
+
+        vlib_put_next_frame(vm, node,
+                            MFIB_FORWARD_LOOKUP_NEXT_RPF,
+                            n_left_to_next);
+    }
+
+    if (node->flags & VLIB_NODE_FLAG_TRACE)
+        mfib_forward_lookup_trace(vm, node, frame);
+
+    return frame->n_vectors;
+}
+
+static uword
+ip4_mfib_forward_lookup (vlib_main_t * vm,
+                         vlib_node_runtime_t * node,
+                         vlib_frame_t * frame)
+{
+    return (mfib_forward_lookup (vm, node, frame, 1));
+}
+
+VLIB_REGISTER_NODE (ip4_mfib_forward_lookup_node, static) = {
+    .function = ip4_mfib_forward_lookup,
+    .name = "ip4-mfib-forward-lookup",
+    .vector_size = sizeof (u32),
+
+    .format_trace = format_mfib_forward_lookup_trace,
+
+    .n_next_nodes = MFIB_FORWARD_LOOKUP_N_NEXT,
+    .next_nodes = {
+        [MFIB_FORWARD_LOOKUP_NEXT_RPF] = "ip4-mfib-forward-rpf",
+    },
+};
+
+VLIB_NODE_FUNCTION_MULTIARCH (ip4_mfib_forward_lookup_node,
+                              ip4_mfib_forward_lookup)
+
+static uword
+ip6_mfib_forward_lookup (vlib_main_t * vm,
+                         vlib_node_runtime_t * node,
+                         vlib_frame_t * frame)
+{
+    return (mfib_forward_lookup (vm, node, frame, 0));
+}
+
+VLIB_REGISTER_NODE (ip6_mfib_forward_lookup_node, static) = {
+    .function = ip6_mfib_forward_lookup,
+    .name = "ip6-mfib-forward-lookup",
+    .vector_size = sizeof (u32),
+
+    .format_trace = format_mfib_forward_lookup_trace,
+
+    .n_next_nodes = MFIB_FORWARD_LOOKUP_N_NEXT,
+    .next_nodes = {
+        [MFIB_FORWARD_LOOKUP_NEXT_RPF] = "ip6-mfib-forward-rpf",
+    },
+};
+
+VLIB_NODE_FUNCTION_MULTIARCH (ip6_mfib_forward_lookup_node,
+                              ip6_mfib_forward_lookup)
+
+
+typedef struct mfib_forward_rpf_trace_t_ {
+    u32 entry_index;
+    u32 sw_if_index;
+    mfib_itf_flags_t itf_flags;
+} mfib_forward_rpf_trace_t;
+
+typedef enum mfib_forward_rpf_next_t_ {
+    MFIB_FORWARD_RPF_NEXT_DROP,
+    MFIB_FORWARD_RPF_N_NEXT,
+} mfib_forward_rpf_next_t;
+
+static u8 *
+format_mfib_forward_rpf_trace (u8 * s, va_list * args)
+{
+    CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *);
+    CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *);
+    mfib_forward_rpf_trace_t * t = va_arg (*args, mfib_forward_rpf_trace_t *);
+
+    s = format (s, "entry %d", t->entry_index);
+    s = format (s, " %d", t->sw_if_index);
+    s = format (s, " %U", format_mfib_itf_flags, t->itf_flags);
+
+    return s;
+}
+
+static int
+mfib_forward_connected_check (vlib_buffer_t * b0,
+                              u32 sw_if_index,
+                              int is_v4)
+{
+    /*
+     * Lookup the source of the IP packet in the
+     * FIB. return true if the entry is attached.
+     */
+    index_t lbi0;
+
+    if (is_v4)
+    {
+        load_balance_t *lb0;
+        ip4_header_t *ip0;
+
+        ip0 = vlib_buffer_get_current(b0);
+
+        lbi0 = ip4_fib_forwarding_lookup(
+                   ip4_fib_table_get_index_for_sw_if_index(
+                       sw_if_index),
+                   &ip0->src_address);
+        lb0 = load_balance_get(lbi0);
+
+        return (FIB_ENTRY_FLAG_ATTACHED &
+                lb0->lb_fib_entry_flags);
+    }
+    else
+    {
+        ASSERT(0);
+    }
+    return (0);
+}
+
+static void
+mfib_forward_itf_signal (vlib_main_t *vm,
+                         const mfib_entry_t *mfe,
+                         mfib_itf_t *mfi,
+                         vlib_buffer_t *b0)
+{
+    mfib_itf_flags_t old_flags;
+
+    old_flags = __sync_fetch_and_or(&mfi->mfi_flags,
+                                    MFIB_ITF_FLAG_SIGNAL_PRESENT);
+
+    if (!(old_flags & MFIB_ITF_FLAG_SIGNAL_PRESENT))
+    {
+        /*
+         * we were the lucky ones to set the signal present flag
+         */
+        if (!(old_flags & MFIB_ITF_FLAG_DONT_PRESERVE))
+        {
+            /*
+             * preserve a copy of the packet for the control
+             * plane to examine.
+             * Only allow one preserved packet at at time, since
+             * when the signal present flag is cleared so is the
+             * preserved packet.
+             */
+            mfib_signal_push(mfe, mfi, b0);
+        }
+        else
+        {
+            /*
+             *  The control plane just wants the signal, not the packet as well
+             */
+            mfib_signal_push(mfe, mfi, NULL);
+        }
+    }
+    /*
+     * else
+     *   there is already a signal present on this interface that the
+     *   control plane has not yet acknowledged
+     */
+}
+
+always_inline uword
+mfib_forward_rpf (vlib_main_t * vm,
+                  vlib_node_runtime_t * node,
+                  vlib_frame_t * frame,
+                  int is_v4)
+{
+    u32 n_left_from, n_left_to_next, * from, * to_next;
+    mfib_forward_rpf_next_t next;
+
+    from = vlib_frame_vector_args (frame);
+    n_left_from = frame->n_vectors;
+    next = MFIB_FORWARD_RPF_NEXT_DROP;
+
+    while (n_left_from > 0)
+    {
+        vlib_get_next_frame (vm, node, next,
+                             to_next, n_left_to_next);
+
+        while (n_left_from > 0 && n_left_to_next > 0)
+        {
+            fib_node_index_t mfei0;
+            const mfib_entry_t *mfe0;
+            mfib_itf_t *mfi0;
+            vlib_buffer_t * b0;
+            u32 pi0, next0;
+            mfib_itf_flags_t iflags0;
+            mfib_entry_flags_t eflags0;
+
+            pi0 = from[0];
+            to_next[0] = pi0;
+            from += 1;
+            to_next += 1;
+            n_left_to_next -= 1;
+            n_left_from -= 1;
+
+            b0 = vlib_get_buffer (vm, pi0);
+            mfei0 = vnet_buffer (b0)->ip.adj_index[VLIB_TX];
+            mfe0 = mfib_entry_get(mfei0);
+            mfi0 = mfib_entry_get_itf(mfe0,
+                                      vnet_buffer(b0)->sw_if_index[VLIB_RX]);
+
+            /*
+             * throughout this function we are 'PREDICT' optimising
+             * for the case of throughput traffic that is not replicated
+             * to the host stack nor sets local flags
+             */
+            if (PREDICT_TRUE(NULL != mfi0))
+            {
+                iflags0 = mfi0->mfi_flags;
+            }
+            else
+            {
+                iflags0 = MFIB_ITF_FLAG_NONE;
+            }
+            eflags0 = mfe0->mfe_flags;
+
+            if (PREDICT_FALSE(eflags0 & MFIB_ENTRY_FLAG_CONNECTED))
+            {
+                /*
+                 * lookup the source in the unicast FIB - check it
+                 * matches a connected.
+                 */
+                if (mfib_forward_connected_check(
+                        b0,
+                        vnet_buffer(b0)->sw_if_index[VLIB_RX],
+                        is_v4))
+                {
+                    mfib_forward_itf_signal(vm, mfe0, mfi0, b0);
+                }
+            }
+            if (PREDICT_FALSE((eflags0 & MFIB_ENTRY_FLAG_SIGNAL) ^
+                              (iflags0 & MFIB_ITF_FLAG_NEGATE_SIGNAL)))
+            {
+                /*
+                 * Entry signal XOR interface negate-signal
+                 */
+                if (NULL != mfi0)
+                {
+                    mfib_forward_itf_signal(vm, mfe0, mfi0, b0);
+                }
+            }
+
+            if (PREDICT_TRUE((iflags0 & MFIB_ITF_FLAG_ACCEPT) ||
+                             (eflags0 & MFIB_ENTRY_FLAG_ACCEPT_ALL_ITF)))
+            {
+                /*
+                 * This interface is accepting packets for the matching entry
+                 */
+                next0 = mfe0->mfe_rep.dpoi_next_node;
+
+                vnet_buffer(b0)->ip.adj_index[VLIB_TX] =
+                    mfe0->mfe_rep.dpoi_index;
+            }
+            else
+            {
+                next0 = MFIB_FORWARD_RPF_NEXT_DROP;
+            }
+
+            if (b0->flags & VLIB_BUFFER_IS_TRACED)
+            {
+                mfib_forward_rpf_trace_t *t0;
+
+                t0 = vlib_add_trace (vm, node, b0, sizeof (t0[0]));
+                t0->entry_index = mfei0;
+                if (NULL == mfi0)
+                {
+                    t0->sw_if_index = ~0;
+                    t0->itf_flags = MFIB_ITF_FLAG_NONE;
+                }
+                else
+                {
+                    t0->sw_if_index = mfi0->mfi_sw_if_index;
+                    t0->itf_flags = mfi0->mfi_flags;
+                }
+            }
+            vlib_validate_buffer_enqueue_x1 (vm, node, next,
+                                             to_next, n_left_to_next,
+                                             pi0, next0);
+        }
+
+        vlib_put_next_frame(vm, node, next, n_left_to_next);
+    }
+
+    return frame->n_vectors;
+}
+
+static uword
+ip4_mfib_forward_rpf (vlib_main_t * vm,
+                      vlib_node_runtime_t * node,
+                      vlib_frame_t * frame)
+{
+    return (mfib_forward_rpf(vm, node, frame, 1));
+}
+
+
+VLIB_REGISTER_NODE (ip4_mfib_forward_rpf_node, static) = {
+    .function = ip4_mfib_forward_rpf,
+    .name = "ip4-mfib-forward-rpf",
+    .vector_size = sizeof (u32),
+
+    .format_trace = format_mfib_forward_rpf_trace,
+
+    .n_next_nodes = MFIB_FORWARD_RPF_N_NEXT,
+    .next_nodes = {
+        [MFIB_FORWARD_RPF_NEXT_DROP] = "error-drop",
+    },
+};
+
+VLIB_NODE_FUNCTION_MULTIARCH (ip4_mfib_forward_rpf_node,
+                              ip4_mfib_forward_rpf)
+
+static uword
+ip6_mfib_forward_rpf (vlib_main_t * vm,
+                      vlib_node_runtime_t * node,
+                      vlib_frame_t * frame)
+{
+    return (mfib_forward_rpf(vm, node, frame, 1));
+}
+
+
+VLIB_REGISTER_NODE (ip6_mfib_forward_rpf_node, static) = {
+    .function = ip6_mfib_forward_rpf,
+    .name = "ip6-mfib-forward-rpf",
+    .vector_size = sizeof (u32),
+
+    .format_trace = format_mfib_forward_rpf_trace,
+
+    .n_next_nodes = MFIB_FORWARD_RPF_N_NEXT,
+    .next_nodes = {
+        [MFIB_FORWARD_RPF_NEXT_DROP] = "error-drop",
+    },
+};
+
+VLIB_NODE_FUNCTION_MULTIARCH (ip6_mfib_forward_rpf_node,
+                              ip6_mfib_forward_rpf)
+
diff --git a/src/vnet/mfib/mfib_itf.c b/src/vnet/mfib/mfib_itf.c
new file mode 100644
index 00000000..b9fa1ec6
--- /dev/null
+++ b/src/vnet/mfib/mfib_itf.c
@@ -0,0 +1,119 @@
+/*
+ * Copyright (c) 2016 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <vnet/vnet.h>
+
+#include <vnet/mfib/mfib_itf.h>
+#include <vnet/mfib/mfib_signal.h>
+
+mfib_itf_t *mfib_itf_pool;
+
+index_t
+mfib_itf_create (u32 sw_if_index,
+                 mfib_itf_flags_t mfi_flags)
+{
+    mfib_itf_t *mfib_itf;
+
+    pool_get_aligned(mfib_itf_pool, mfib_itf,
+                     CLIB_CACHE_LINE_BYTES);
+
+    mfib_itf->mfi_sw_if_index = sw_if_index;
+    mfib_itf->mfi_flags = mfi_flags;
+    mfib_itf->mfi_si = INDEX_INVALID;
+
+    return (mfib_itf - mfib_itf_pool);
+}
+
+void
+mfib_itf_delete (mfib_itf_t *mfi)
+{
+    mfib_signal_remove_itf(mfi);
+    pool_put(mfib_itf_pool, mfi);
+}
+
+u8 *
+format_mfib_itf (u8 * s, va_list * args)
+{
+    mfib_itf_t *mfib_itf;
+    vnet_main_t *vnm;
+    index_t mfi;
+
+    mfi = va_arg (*args, index_t);
+
+    vnm = vnet_get_main();
+    mfib_itf = mfib_itf_get(mfi);
+
+    if (~0 != mfib_itf->mfi_sw_if_index)
+    {
+        return (format(s, " %U: %U",
+                       format_vnet_sw_interface_name,
+                       vnm,
+                       vnet_get_sw_interface(vnm,
+                                             mfib_itf->mfi_sw_if_index),
+                       format_mfib_itf_flags, mfib_itf->mfi_flags));
+    }
+    else
+    {
+        return (format(s, " local: %U",
+                       format_mfib_itf_flags, mfib_itf->mfi_flags));
+    }
+    return (s);
+}
+
+static clib_error_t *
+show_mfib_itf_command (vlib_main_t * vm,
+                        unformat_input_t * input,
+                        vlib_cli_command_t * cmd)
+{
+    index_t mfii;
+
+    if (unformat (input, "%d", &mfii))
+    {
+        /*
+         * show one in detail
+         */
+        if (!pool_is_free_index(mfib_itf_pool, mfii))
+        {
+            vlib_cli_output (vm, "%d@%U",
+                             mfii,
+                             format_mfib_itf, mfii);
+        }
+        else
+        {
+            vlib_cli_output (vm, "itf %d invalid", mfii);
+        }
+    }
+    else
+    {
+        /*
+         * show all
+         */
+        vlib_cli_output (vm, "mFIB interfaces::");
+        pool_foreach_index(mfii, mfib_itf_pool,
+        ({
+            vlib_cli_output (vm, "%d@%U",
+                             mfii,
+                             format_mfib_itf, mfii);
+        }));
+    }
+
+    return (NULL);
+}
+
+VLIB_CLI_COMMAND (show_mfib_itf, static) = {
+  .path = "show mfib interface",
+  .function = show_mfib_itf_command,
+  .short_help = "show mfib interface",
+};
diff --git a/src/vnet/mfib/mfib_itf.h b/src/vnet/mfib/mfib_itf.h
new file mode 100644
index 00000000..5f26a476
--- /dev/null
+++ b/src/vnet/mfib/mfib_itf.h
@@ -0,0 +1,63 @@
+/*
+ * Copyright (c) 2016 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef __MFIB_ITF_H__
+#define __MFIB_ITF_H__
+
+#include <vlib/vlib.h>
+#include <vnet/mfib/mfib_types.h>
+
+/**
+ * @brief An interface associated with a particular MFIB entry
+ */
+typedef struct mfib_itf_t_
+{
+    /**
+     * @brief Falags on the entry
+     */
+    mfib_itf_flags_t mfi_flags;
+
+    /**
+     * The SW IF index that this MFIB interface represents
+     */
+    u32 mfi_sw_if_index;
+
+    /**
+     * The index of the signal in the pending list
+     */
+    u32 mfi_si;
+} mfib_itf_t;
+
+
+extern index_t mfib_itf_create(u32 sw_if_index,
+                               mfib_itf_flags_t mfi_flags);
+extern void mfib_itf_delete(mfib_itf_t *mfi);
+
+extern u8 *format_mfib_itf(u8 * s, va_list * args);
+
+extern mfib_itf_t *mfib_itf_pool;
+
+static inline mfib_itf_t *
+mfib_itf_get (index_t mi)
+{
+    return (pool_elt_at_index(mfib_itf_pool, mi));
+}
+static inline index_t
+mfib_itf_get_index (const mfib_itf_t *mfi)
+{
+    return (mfi - mfib_itf_pool);
+}
+
+#endif
diff --git a/src/vnet/mfib/mfib_signal.c b/src/vnet/mfib/mfib_signal.c
new file mode 100644
index 00000000..9f6205de
--- /dev/null
+++ b/src/vnet/mfib/mfib_signal.c
@@ -0,0 +1,201 @@
+/*
+ * Copyright (c) 2016 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+
+#include <vnet/vnet.h>
+#include <vnet/mfib/mfib_signal.h>
+#include <vppinfra/dlist.h>
+
+/**
+ * @brief Pool of signals
+ */
+static mfib_signal_t *mfib_signal_pool;
+
+/**
+ * @brief pool of dlist elements
+ */
+static dlist_elt_t *mfib_signal_dlist_pool;
+
+/**
+ * the list/set of interfaces with signals pending
+ */
+typedef struct mfib_signal_q_t_
+{
+    /**
+     * the dlist indext that is the head of the list
+     */
+    u32 mip_head;
+
+    /**
+     * Spin lock to protect the list
+     */
+    int mip_lock;
+} mfib_signal_q_t;
+
+/**
+ * @brief The pending queue of signals to deliver to the control plane
+ */
+static mfib_signal_q_t mfib_signal_pending ;
+
+static void
+mfib_signal_list_init (void)
+{
+    dlist_elt_t *head;
+    u32 hi;
+
+    pool_get(mfib_signal_dlist_pool, head);
+    hi = head - mfib_signal_dlist_pool;
+
+    mfib_signal_pending.mip_head = hi;
+    clib_dlist_init(mfib_signal_dlist_pool, hi);
+}
+
+void
+mfib_signal_module_init (void)
+{
+    mfib_signal_list_init();
+}
+
+int
+mfib_signal_send_one (struct _unix_shared_memory_queue *q,
+                      u32 context)
+{
+    u32 li, si;
+
+    /*
+     * with the lock held, pop a signal from the q.
+     */
+    while (__sync_lock_test_and_set (&mfib_signal_pending.mip_lock, 1))
+        ;
+    {
+        li = clib_dlist_remove_head(mfib_signal_dlist_pool,
+                                    mfib_signal_pending.mip_head);
+    }
+    mfib_signal_pending.mip_lock = 0;
+
+    if (~0 != li)
+    {
+        mfib_signal_t *mfs;
+        mfib_itf_t *mfi;
+        dlist_elt_t *elt;
+
+        elt = pool_elt_at_index(mfib_signal_dlist_pool, li);
+        si = elt->value;
+
+        mfs = pool_elt_at_index(mfib_signal_pool, si);
+        mfi = mfib_itf_get(mfs->mfs_itf);
+        mfi->mfi_si = INDEX_INVALID;
+        __sync_fetch_and_and(&mfi->mfi_flags,
+                             ~MFIB_ITF_FLAG_SIGNAL_PRESENT);
+
+
+        vl_mfib_signal_send_one(q, context, mfs);
+
+        /*
+         * with the lock held, return the resoruces of the signals posted
+         */
+        while (__sync_lock_test_and_set(&mfib_signal_pending.mip_lock, 1))
+            ;
+        {
+            pool_put_index(mfib_signal_pool, si);
+            pool_put_index(mfib_signal_dlist_pool, li);
+        }
+        mfib_signal_pending.mip_lock = 0;
+
+        return (1);
+    }
+    return (0);
+}
+
+void
+mfib_signal_push (const mfib_entry_t *mfe,
+                  mfib_itf_t *mfi,
+                  vlib_buffer_t *b0)
+{
+    mfib_signal_t *mfs;
+    dlist_elt_t *elt;
+    u32 si, li;
+
+    while (__sync_lock_test_and_set (&mfib_signal_pending.mip_lock, 1))
+        ;
+    {
+        pool_get(mfib_signal_pool, mfs);
+        pool_get(mfib_signal_dlist_pool, elt);
+
+        si = mfs - mfib_signal_pool;
+        li = elt - mfib_signal_dlist_pool;
+
+        elt->value = si;
+        mfi->mfi_si = li;
+
+        clib_dlist_addhead(mfib_signal_dlist_pool,
+                           mfib_signal_pending.mip_head,
+                           li);
+    }
+    mfib_signal_pending.mip_lock = 0;
+
+    mfs->mfs_entry = mfib_entry_get_index(mfe);
+    mfs->mfs_itf = mfib_itf_get_index(mfi);
+
+    if (NULL != b0)
+    {
+        mfs->mfs_buffer_len = b0->current_length;
+        memcpy(mfs->mfs_buffer,
+               vlib_buffer_get_current(b0),
+               (mfs->mfs_buffer_len > MFIB_SIGNAL_BUFFER_SIZE ?
+                MFIB_SIGNAL_BUFFER_SIZE :
+                mfs->mfs_buffer_len));
+    }
+    else
+    {
+        mfs->mfs_buffer_len = 0;
+    }
+}
+
+void
+mfib_signal_remove_itf (const mfib_itf_t *mfi)
+{
+    u32 li;
+
+    /*
+     * lock the queue to prevent further additions while we fiddle.
+     */
+    li = mfi->mfi_si;
+
+    if (INDEX_INVALID != li)
+    {
+        /*
+         * it's in the pending q
+         */
+        while (__sync_lock_test_and_set (&mfib_signal_pending.mip_lock, 1))
+            ;
+        {
+            dlist_elt_t *elt;
+
+            /*
+             * with the lock held;
+             *  - remove the signal from the pending list
+             *  - free up the signal and list entry obejcts
+             */
+            clib_dlist_remove(mfib_signal_dlist_pool, li);
+
+            elt = pool_elt_at_index(mfib_signal_dlist_pool, li);
+            pool_put_index(mfib_signal_pool, elt->value);
+            pool_put(mfib_signal_dlist_pool, elt);
+        }
+
+        mfib_signal_pending.mip_lock = 0;
+    }
+}
diff --git a/src/vnet/mfib/mfib_signal.h b/src/vnet/mfib/mfib_signal.h
new file mode 100644
index 00000000..732d8aff
--- /dev/null
+++ b/src/vnet/mfib/mfib_signal.h
@@ -0,0 +1,59 @@
+/*
+ * Copyright (c) 2016 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef __MFIB_SIGNAL_H__
+#define __MFIB_SIGNAL_H__
+
+#include <vlib/vlib.h>
+#include <vnet/mfib/mfib_types.h>
+#include <vnet/mfib/mfib_itf.h>
+#include <vnet/mfib/mfib_entry.h>
+
+#define MFIB_SIGNAL_BUFFER_SIZE 255
+
+/**
+ * A pair of indicies, for the entry and interface resp.
+ */
+typedef struct mfib_signal_t_
+{
+    fib_node_index_t mfs_entry;
+    index_t mfs_itf;
+
+    /**
+     * @brief A buffer copied from the DP plane that triggered the signal
+     */
+    u8 mfs_buffer[MFIB_SIGNAL_BUFFER_SIZE];
+
+    u8 mfs_buffer_len;
+} mfib_signal_t;
+
+
+extern void mfib_signal_push(const mfib_entry_t *mfe,
+                             mfib_itf_t *mfi,
+                             vlib_buffer_t *b0);
+extern void mfib_signal_remove_itf(const mfib_itf_t *mfi);
+
+extern void mfib_signal_module_init(void);
+
+struct _unix_shared_memory_queue;
+
+extern void vl_mfib_signal_send_one(struct _unix_shared_memory_queue *q,
+                                    u32 context,
+                                    const mfib_signal_t *mfs);
+extern int mfib_signal_send_one(struct _unix_shared_memory_queue *q,
+                                u32 context);
+
+#endif
+
diff --git a/src/vnet/mfib/mfib_table.c b/src/vnet/mfib/mfib_table.c
new file mode 100644
index 00000000..e4c0936d
--- /dev/null
+++ b/src/vnet/mfib/mfib_table.c
@@ -0,0 +1,489 @@
+/*
+ * Copyright (c) 2016 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <vlib/vlib.h>
+#include <vnet/dpo/drop_dpo.h>
+
+#include <vnet/mfib/mfib_table.h>
+#include <vnet/mfib/ip4_mfib.h>
+#include <vnet/mfib/ip6_mfib.h>
+#include <vnet/mfib/mfib_entry.h>
+#include <vnet/mfib/mfib_signal.h>
+
+mfib_table_t *
+mfib_table_get (fib_node_index_t index,
+                fib_protocol_t proto)
+{
+    switch (proto)
+    {
+    case FIB_PROTOCOL_IP4:
+        return (pool_elt_at_index(ip4_main.mfibs, index));
+    case FIB_PROTOCOL_IP6:
+        return (pool_elt_at_index(ip6_main.mfibs, index));
+    case FIB_PROTOCOL_MPLS:
+        break;
+    }
+    ASSERT(0);
+    return (NULL);
+}
+
+static inline fib_node_index_t
+mfib_table_lookup_i (const mfib_table_t *mfib_table,
+                     const mfib_prefix_t *prefix)
+{
+    switch (prefix->fp_proto)
+    {
+    case FIB_PROTOCOL_IP4:
+        return (ip4_mfib_table_lookup(&mfib_table->v4,
+                                      &prefix->fp_src_addr.ip4,
+                                      &prefix->fp_grp_addr.ip4,
+                                      prefix->fp_len));
+    case FIB_PROTOCOL_IP6:
+        return (ip6_mfib_table_lookup(&mfib_table->v6,
+                                      &prefix->fp_src_addr.ip6,
+                                      &prefix->fp_grp_addr.ip6,
+                                      prefix->fp_len));
+    case FIB_PROTOCOL_MPLS:
+        break;
+    }
+    return (FIB_NODE_INDEX_INVALID);
+}
+
+fib_node_index_t
+mfib_table_lookup (u32 fib_index,
+                   const mfib_prefix_t *prefix)
+{
+    return (mfib_table_lookup_i(mfib_table_get(fib_index, prefix->fp_proto), prefix));
+}
+
+static inline fib_node_index_t
+mfib_table_lookup_exact_match_i (const mfib_table_t *mfib_table,
+                                 const mfib_prefix_t *prefix)
+{
+    switch (prefix->fp_proto)
+    {
+    case FIB_PROTOCOL_IP4:
+        return (ip4_mfib_table_lookup_exact_match(&mfib_table->v4,
+                                                  &prefix->fp_grp_addr.ip4,
+                                                  &prefix->fp_src_addr.ip4,
+                                                  prefix->fp_len));
+    case FIB_PROTOCOL_IP6:
+        return (ip6_mfib_table_lookup_exact_match(&mfib_table->v6,
+                                                  &prefix->fp_grp_addr.ip6,
+                                                  &prefix->fp_src_addr.ip6,
+                                                  prefix->fp_len));
+    case FIB_PROTOCOL_MPLS:
+        break;
+    }
+    return (FIB_NODE_INDEX_INVALID);
+}
+
+fib_node_index_t
+mfib_table_lookup_exact_match (u32 fib_index,
+                              const mfib_prefix_t *prefix)
+{
+    return (mfib_table_lookup_exact_match_i(mfib_table_get(fib_index,
+                                                          prefix->fp_proto),
+                                            prefix));
+}
+
+static void
+mfib_table_entry_remove (mfib_table_t *mfib_table,
+                         const mfib_prefix_t *prefix,
+                         fib_node_index_t fib_entry_index)
+{
+    vlib_smp_unsafe_warning();
+
+    mfib_table->mft_total_route_counts--;
+
+    switch (prefix->fp_proto)
+    {
+    case FIB_PROTOCOL_IP4:
+        ip4_mfib_table_entry_remove(&mfib_table->v4,
+                                    &prefix->fp_grp_addr.ip4,
+                                    &prefix->fp_src_addr.ip4,
+                                    prefix->fp_len);
+        break;
+    case FIB_PROTOCOL_IP6:
+        ip6_mfib_table_entry_remove(&mfib_table->v6,
+                                    &prefix->fp_grp_addr.ip6,
+                                    &prefix->fp_src_addr.ip6,
+                                    prefix->fp_len);
+        break;
+    case FIB_PROTOCOL_MPLS:
+        ASSERT(0);
+        break;
+    }
+
+    mfib_entry_unlock(fib_entry_index);
+}
+
+static void
+mfib_table_entry_insert (mfib_table_t *mfib_table,
+                         const mfib_prefix_t *prefix,
+                         fib_node_index_t mfib_entry_index)
+{
+    vlib_smp_unsafe_warning();
+
+    mfib_entry_lock(mfib_entry_index);
+    mfib_table->mft_total_route_counts++;
+
+    switch (prefix->fp_proto)
+    {
+    case FIB_PROTOCOL_IP4:
+        ip4_mfib_table_entry_insert(&mfib_table->v4,
+                                    &prefix->fp_grp_addr.ip4,
+                                    &prefix->fp_src_addr.ip4,
+                                    prefix->fp_len,
+                                    mfib_entry_index);
+        break;
+    case FIB_PROTOCOL_IP6:
+        ip6_mfib_table_entry_insert(&mfib_table->v6,
+                                    &prefix->fp_grp_addr.ip6,
+                                    &prefix->fp_src_addr.ip6,
+                                    prefix->fp_len,
+                                    mfib_entry_index);
+        break;
+    case FIB_PROTOCOL_MPLS:
+        break;
+    }
+}
+
+fib_node_index_t
+mfib_table_entry_update (u32 fib_index,
+                         const mfib_prefix_t *prefix,
+                         mfib_source_t source,
+                         mfib_entry_flags_t entry_flags)
+{
+    fib_node_index_t mfib_entry_index;
+    mfib_table_t *mfib_table;
+
+    mfib_table = mfib_table_get(fib_index, prefix->fp_proto);
+    mfib_entry_index = mfib_table_lookup_exact_match_i(mfib_table, prefix);
+
+    if (FIB_NODE_INDEX_INVALID == mfib_entry_index)
+    {
+        if (MFIB_ENTRY_FLAG_NONE != entry_flags)
+        {
+            /*
+             * update to a non-existing entry with non-zero flags
+             */
+            mfib_entry_index = mfib_entry_create(fib_index, source,
+                                                 prefix, entry_flags);
+
+            mfib_table_entry_insert(mfib_table, prefix, mfib_entry_index);
+        }
+        /*
+         * else
+         *   the entry doesn't exist and the request is to set no flags
+         *   the result would be an entry that doesn't exist - so do nothing
+         */
+    }
+    else
+    {
+        mfib_entry_lock(mfib_entry_index);
+
+        if (mfib_entry_update(mfib_entry_index, source, entry_flags))
+        {
+            /*
+             * this update means we can now remove the entry.
+             */
+            mfib_table_entry_remove(mfib_table, prefix, mfib_entry_index);
+        }
+
+        mfib_entry_unlock(mfib_entry_index);
+    }
+
+    return (mfib_entry_index);
+}
+
+fib_node_index_t
+mfib_table_entry_path_update (u32 fib_index,
+                              const mfib_prefix_t *prefix,
+                              mfib_source_t source,
+                              const fib_route_path_t *rpath,
+                              mfib_itf_flags_t itf_flags)
+{
+    fib_node_index_t mfib_entry_index;
+    mfib_table_t *mfib_table;
+
+    mfib_table = mfib_table_get(fib_index, prefix->fp_proto);
+    mfib_entry_index = mfib_table_lookup_exact_match_i(mfib_table, prefix);
+
+    if (FIB_NODE_INDEX_INVALID == mfib_entry_index)
+    {
+        mfib_entry_index = mfib_entry_create(fib_index,
+                                             source,
+                                             prefix,
+                                             MFIB_ENTRY_FLAG_NONE);
+
+        mfib_table_entry_insert(mfib_table, prefix, mfib_entry_index);
+    }
+
+    mfib_entry_path_update(mfib_entry_index,
+                           source,
+                           rpath,
+                           itf_flags);
+
+    return (mfib_entry_index);
+}
+
+void
+mfib_table_entry_path_remove (u32 fib_index,
+                              const mfib_prefix_t *prefix,
+                              mfib_source_t source,
+                              const fib_route_path_t *rpath)
+{
+    fib_node_index_t mfib_entry_index;
+    mfib_table_t *mfib_table;
+
+    mfib_table = mfib_table_get(fib_index, prefix->fp_proto);
+    mfib_entry_index = mfib_table_lookup_exact_match_i(mfib_table, prefix);
+
+    if (FIB_NODE_INDEX_INVALID == mfib_entry_index)
+    {
+        /*
+         * removing an etry that does not exist. i'll allow it.
+         */
+    }
+    else
+    {
+        int no_more_sources;
+
+        /*
+         * don't nobody go nowhere
+         */
+        mfib_entry_lock(mfib_entry_index);
+
+        no_more_sources = mfib_entry_path_remove(mfib_entry_index,
+                                                 source,
+                                                 rpath);
+
+        if (no_more_sources)
+        {
+            /*
+             * last source gone. remove from the table
+             */
+            mfib_table_entry_remove(mfib_table, prefix, mfib_entry_index);
+        }
+
+        mfib_entry_unlock(mfib_entry_index);
+    }
+}
+
+static void
+mfib_table_entry_delete_i (u32 fib_index,
+                           fib_node_index_t mfib_entry_index,
+                           const mfib_prefix_t *prefix,
+                           mfib_source_t source)
+{
+    mfib_table_t *mfib_table;
+
+    mfib_table = mfib_table_get(fib_index, prefix->fp_proto);
+
+    /*
+     * don't nobody go nowhere
+     */
+    mfib_entry_lock(mfib_entry_index);
+
+    if (mfib_entry_delete(mfib_entry_index, source))
+    {
+        /*
+         * last source gone. remove from the table
+         */
+        mfib_table_entry_remove(mfib_table, prefix, mfib_entry_index);
+    }
+    /*
+     * else
+     *   still has sources, leave it be.
+     */
+
+    mfib_entry_unlock(mfib_entry_index);
+}
+
+void
+mfib_table_entry_delete (u32 fib_index,
+                         const mfib_prefix_t *prefix,
+                         mfib_source_t source)
+{
+    fib_node_index_t mfib_entry_index;
+
+    mfib_entry_index = mfib_table_lookup_exact_match(fib_index, prefix);
+
+    if (FIB_NODE_INDEX_INVALID == mfib_entry_index)
+    {
+        /*
+         * removing an etry that does not exist.
+         * i'll allow it, but i won't like it.
+         */
+        clib_warning("%U not in FIB", format_mfib_prefix, prefix);
+    }
+    else
+    {
+        mfib_table_entry_delete_i(fib_index, mfib_entry_index,
+                                  prefix, source);
+    }
+}
+
+void
+mfib_table_entry_delete_index (fib_node_index_t mfib_entry_index,
+                               mfib_source_t source)
+{
+    mfib_prefix_t prefix;
+
+    mfib_entry_get_prefix(mfib_entry_index, &prefix);
+
+    mfib_table_entry_delete_i(mfib_entry_get_fib_index(mfib_entry_index),
+                              mfib_entry_index, &prefix, source);
+}
+
+u32
+mfib_table_get_index_for_sw_if_index (fib_protocol_t proto,
+                                      u32 sw_if_index)
+{
+    switch (proto)
+    {
+    case FIB_PROTOCOL_IP4:
+        return (ip4_mfib_table_get_index_for_sw_if_index(sw_if_index));
+    case FIB_PROTOCOL_IP6:
+        return (ip6_mfib_table_get_index_for_sw_if_index(sw_if_index));
+    case FIB_PROTOCOL_MPLS:
+        ASSERT(0);
+        break;
+    }
+    return (~0);
+}
+
+u32
+mfib_table_find (fib_protocol_t proto,
+                 u32 table_id)
+{
+    switch (proto)
+    {
+    case FIB_PROTOCOL_IP4:
+        return (ip4_mfib_index_from_table_id(table_id));
+    case FIB_PROTOCOL_IP6:
+        return (ip6_mfib_index_from_table_id(table_id));
+    case FIB_PROTOCOL_MPLS:
+        ASSERT(0);
+        break;
+    }
+    return (~0);
+}
+
+u32
+mfib_table_find_or_create_and_lock (fib_protocol_t proto,
+                                    u32 table_id)
+{
+    mfib_table_t *mfib_table;
+    fib_node_index_t fi;
+
+    switch (proto)
+    {
+    case FIB_PROTOCOL_IP4:
+        fi = ip4_mfib_table_find_or_create_and_lock(table_id);
+        break;
+    case FIB_PROTOCOL_IP6:
+        fi = ip6_mfib_table_find_or_create_and_lock(table_id);
+        break;
+    case FIB_PROTOCOL_MPLS:
+    default:
+        return (~0);
+    }
+
+    mfib_table = mfib_table_get(fi, proto);
+
+    mfib_table->mft_desc = format(NULL, "%U-VRF:%d",
+                                  format_fib_protocol, proto,
+                                  table_id);
+
+    return (fi);
+}
+
+static void
+mfib_table_destroy (mfib_table_t *mfib_table)
+{
+    vec_free(mfib_table->mft_desc);
+
+    switch (mfib_table->mft_proto)
+    {
+    case FIB_PROTOCOL_IP4:
+        ip4_mfib_table_destroy(&mfib_table->v4);
+        break;
+    case FIB_PROTOCOL_IP6:
+        ip6_mfib_table_destroy(&mfib_table->v6);
+        break;
+    case FIB_PROTOCOL_MPLS:
+        ASSERT(0);
+        break;
+    }
+}
+
+void
+mfib_table_unlock (u32 fib_index,
+                   fib_protocol_t proto)
+{
+    mfib_table_t *mfib_table;
+
+    mfib_table = mfib_table_get(fib_index, proto);
+    mfib_table->mft_locks--;
+
+    if (0 == mfib_table->mft_locks)
+    {
+        mfib_table_destroy(mfib_table);
+    }
+}
+
+void
+mfib_table_lock (u32 fib_index,
+                 fib_protocol_t proto)
+{
+    mfib_table_t *mfib_table;
+
+    mfib_table = mfib_table_get(fib_index, proto);
+    mfib_table->mft_locks++;
+}
+
+u8*
+format_mfib_table_name (u8* s, va_list ap)
+{
+    fib_node_index_t fib_index = va_arg(ap, fib_node_index_t);
+    fib_protocol_t proto = va_arg(ap, int); // int promotion
+    mfib_table_t *mfib_table;
+
+    mfib_table = mfib_table_get(fib_index, proto);
+
+    s = format(s, "%v", mfib_table->mft_desc);
+
+    return (s);
+}
+
+static clib_error_t *
+mfib_module_init (vlib_main_t * vm)
+{
+    clib_error_t * error;
+
+    if ((error = vlib_call_init_function (vm, fib_module_init)))
+        return (error);
+    if ((error = vlib_call_init_function (vm, rn_module_init)))
+        return (error);
+
+    mfib_entry_module_init();
+    mfib_signal_module_init();
+
+    return (error);
+}
+
+VLIB_INIT_FUNCTION(mfib_module_init);
diff --git a/src/vnet/mfib/mfib_table.h b/src/vnet/mfib/mfib_table.h
new file mode 100644
index 00000000..4faa69ee
--- /dev/null
+++ b/src/vnet/mfib/mfib_table.h
@@ -0,0 +1,331 @@
+/*
+ * Copyright (c) 2016 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef __MFIB_TABLE_H__
+#define __MFIB_TABLE_H__
+
+#include <vnet/ip/ip.h>
+#include <vnet/adj/adj.h>
+
+#include <vnet/mfib/mfib_types.h>
+
+/**
+ * @brief
+ *   A protocol Independent IP multicast FIB table
+ */
+typedef struct mfib_table_t_
+{
+    /**
+     * A union of the protocol specific FIBs that provide the
+     * underlying LPM mechanism.
+     * This element is first in the struct so that it is in the
+     * first cache line.
+     */
+    union {
+        ip4_mfib_t v4;
+        ip6_mfib_t v6;
+    };
+
+    /**
+     * Which protocol this table serves. Used to switch on the union above.
+     */
+    fib_protocol_t mft_proto;
+
+    /**
+     * number of locks on the table
+     */
+    u16 mft_locks;
+
+    /**
+     * Table ID (hash key) for this FIB.
+     */
+    u32 mft_table_id;
+
+    /**
+     * Index into FIB vector.
+     */
+    fib_node_index_t mft_index;
+
+    /**
+     * Total route counters
+     */
+    u32 mft_total_route_counts;
+
+    /**
+     * Table description
+     */
+    u8* mft_desc;
+} mfib_table_t;
+
+/**
+ * @brief
+ *  Format the description/name of the table
+ */
+extern u8* format_mfib_table_name(u8* s, va_list ap);
+
+/**
+ * @brief
+ *  Perfom a longest prefix match in the non-forwarding table
+ *
+ * @param fib_index
+ *  The index of the FIB
+ *
+ * @param prefix
+ *  The prefix to lookup
+ *
+ * @return
+ *  The index of the fib_entry_t for the best match, which may be the default route
+ */
+extern fib_node_index_t mfib_table_lookup(u32 fib_index,
+                                         const mfib_prefix_t *prefix);
+
+/**
+ * @brief
+ *  Perfom an exact match in the non-forwarding table
+ *
+ * @param fib_index
+ *  The index of the FIB
+ *
+ * @param prefix
+ *  The prefix to lookup
+ *
+ * @return
+ *  The index of the fib_entry_t for the exact match, or INVALID
+ *  is there is no match.
+ */
+extern fib_node_index_t mfib_table_lookup_exact_match(u32 fib_index,
+                                                      const mfib_prefix_t *prefix);
+
+/**
+ * @brief
+ * Add a new (with no replication) or lock an existing entry
+ *
+ * @param prefix
+ *  The prefix for the entry to add
+ *
+ * @return
+ *  the index of the fib_entry_t that is created (or existed already).
+ */
+extern fib_node_index_t mfib_table_entry_update(u32 fib_index,
+                                                const mfib_prefix_t *prefix,
+                                                mfib_source_t source,
+                                                mfib_entry_flags_t flags);
+
+/**
+ * @brief
+ *  Add n paths to an entry (aka route) in the FIB. If the entry does not
+ *  exist, it will be created.
+ * See the documentation for fib_route_path_t for more descirptions of
+ * the path parameters.
+ *
+ * @param fib_index
+ *  The index of the FIB
+ *
+ * @param prefix
+ *  The prefix for the entry to add
+ *
+ * @param source
+ *  The ID of the client/source adding the entry.
+ *
+ * @param flags
+ *  Flags for the entry.
+ *
+ * @param rpaths
+ *  A vector of paths.
+ *
+ * @return
+ *  the index of the fib_entry_t that is created (or existed already).
+ */
+extern fib_node_index_t mfib_table_entry_path_update(u32 fib_index,
+                                                     const mfib_prefix_t *prefix,
+                                                     mfib_source_t source,
+                                                     const fib_route_path_t *rpath,
+                                                     mfib_itf_flags_t flags);
+
+/**
+ * @brief
+ * Remove n paths to an entry (aka route) in the FIB. If this is the entry's
+ * last path, then the entry will be removed, unless it has other sources.
+ * See the documentation for fib_route_path_t for more descirptions of
+ * the path parameters.
+ *
+ * @param fib_index
+ *  The index of the FIB
+ *
+ * @param prefix
+ *  The prefix for the entry to add
+ *
+ * @param source
+ *  The ID of the client/source adding the entry.
+ *
+ * @param rpaths
+ *  A vector of paths.
+ */
+extern void mfib_table_entry_path_remove(u32 fib_index,
+                                         const mfib_prefix_t *prefix,
+                                         mfib_source_t source,
+                                         const fib_route_path_t *paths);
+
+
+
+/**
+ * @brief
+ *  Delete a FIB entry. If the entry has no more sources, then it is
+ * removed from the table.
+ *
+ * @param fib_index
+ *  The index of the FIB
+ *
+ * @param prefix
+ *  The prefix for the entry to remove
+ *
+ * @param source
+ *  The ID of the client/source adding the entry.
+ */
+extern void mfib_table_entry_delete(u32 fib_index,
+                                    const mfib_prefix_t *prefix,
+                                    mfib_source_t source);
+
+/**
+ * @brief
+ *  Delete a FIB entry. If the entry has no more sources, then it is
+ * removed from the table.
+ *
+ * @param entry_index
+ *  The index of the FIB entry
+ *
+ * @param source
+ *  The ID of the client/source adding the entry.
+ */
+extern void mfib_table_entry_delete_index(fib_node_index_t entry_index,
+                                          mfib_source_t source);
+
+/**
+ * @brief
+ *  Flush all entries from a table for the source
+ *
+ * @param fib_index
+ *  The index of the FIB
+ *
+ * @paran proto
+ *  The protocol of the entries in the table
+ *
+ * @param source
+ *  the source to flush
+ */
+extern void mfib_table_flush(u32 fib_index,
+                             fib_protocol_t proto);
+
+/**
+ * @brief
+ *  Get the index of the FIB bound to the interface
+ *
+ * @paran proto
+ *  The protocol of the FIB (and thus the entries therein)
+ *
+ * @param sw_if_index
+ *  The interface index
+ *
+ * @return fib_index
+ *  The index of the FIB
+ */
+extern u32 mfib_table_get_index_for_sw_if_index(fib_protocol_t proto,
+                                                u32 sw_if_index);
+
+/**
+ * @brief
+ *  Get the index of the FIB for a Table-ID. This DOES NOT create the
+ * FIB if it does not exist.
+ *
+ * @paran proto
+ *  The protocol of the FIB (and thus the entries therein)
+ *
+ * @param table-id
+ *  The Table-ID
+ *
+ * @return fib_index
+ *  The index of the FIB, which may be INVALID.
+ */
+extern u32 mfib_table_find(fib_protocol_t proto, u32 table_id);
+
+
+/**
+ * @brief
+ *  Get the index of the FIB for a Table-ID. This DOES create the
+ * FIB if it does not exist.
+ *
+ * @paran proto
+ *  The protocol of the FIB (and thus the entries therein)
+ *
+ * @param table-id
+ *  The Table-ID
+ *
+ * @return fib_index
+ *  The index of the FIB
+ */
+extern u32 mfib_table_find_or_create_and_lock(fib_protocol_t proto,
+                                              u32 table_id);
+
+
+/**
+ * @brief
+ * Take a reference counting lock on the table
+ *
+ * @param fib_index
+ *  The index of the FIB
+ *
+ * @paran proto
+ *  The protocol of the FIB (and thus the entries therein)
+ */
+extern void mfib_table_unlock(u32 fib_index,
+                              fib_protocol_t proto);
+
+/**
+ * @brief
+ * Release a reference counting lock on the table. When the last lock
+ * has gone. the FIB is deleted.
+ *
+ * @param fib_index
+ *  The index of the FIB
+ *
+ * @paran proto
+ *  The protocol of the FIB (and thus the entries therein)
+ */
+extern void mfib_table_lock(u32 fib_index,
+                            fib_protocol_t proto);
+
+/**
+ * @brief
+ * Return the number of entries in the FIB added by a given source.
+ *
+ * @param fib_index
+ *  The index of the FIB
+ *
+ * @paran proto
+ *  The protocol of the FIB (and thus the entries therein)
+ *
+ * @return number of sourced entries.
+ */
+extern u32 mfib_table_get_num_entries(u32 fib_index,
+                                      fib_protocol_t proto);
+
+/**
+ * @brief
+ * Get a pointer to a FIB table
+ */
+extern mfib_table_t *mfib_table_get(fib_node_index_t index,
+                                    fib_protocol_t proto);
+
+#endif
diff --git a/src/vnet/mfib/mfib_test.c b/src/vnet/mfib/mfib_test.c
new file mode 100644
index 00000000..8735bfa7
--- /dev/null
+++ b/src/vnet/mfib/mfib_test.c
@@ -0,0 +1,1225 @@
+/*
+ * Copyright (c) 2016 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <vnet/mpls/mpls_types.h>
+
+#include <vnet/mfib/mfib_table.h>
+#include <vnet/mfib/mfib_entry.h>
+#include <vnet/mfib/mfib_signal.h>
+#include <vnet/mfib/ip6_mfib.h>
+
+#include <vnet/dpo/replicate_dpo.h>
+#include <vnet/adj/adj_mcast.h>
+
+#define MFIB_TEST_I(_cond, _comment, _args...)			\
+({								\
+    int _evald = (_cond);					\
+    if (!(_evald)) {						\
+        fformat(stderr, "FAIL:%d: " _comment "\n",		\
+                __LINE__, ##_args);				\
+    } else {							\
+        fformat(stderr, "PASS:%d: " _comment "\n",		\
+                __LINE__, ##_args);				\
+    }								\
+    _evald;							\
+})
+#define MFIB_TEST(_cond, _comment, _args...)			\
+{								\
+    if (!MFIB_TEST_I(_cond, _comment, ##_args)) {		\
+        return 1;\
+        ASSERT(!("FAIL: " _comment));				\
+    }								\
+}
+#define MFIB_TEST_NS(_cond)                                     \
+{								\
+    if (!MFIB_TEST_I(_cond, "")) {                              \
+        return 1;\
+        ASSERT(!("FAIL: "));                                    \
+    }								\
+}
+
+/**
+ * A 'i'm not fussed is this is not efficient' store of test data
+ */
+typedef struct test_main_t_ {
+    /**
+     * HW if indicies
+     */
+    u32 hw_if_indicies[4];
+    /**
+     * HW interfaces
+     */
+    vnet_hw_interface_t * hw[4];
+
+} test_main_t;
+static test_main_t test_main;
+
+/* fake ethernet device class, distinct from "fake-ethX" */
+static u8 * format_test_interface_name (u8 * s, va_list * args)
+{
+  u32 dev_instance = va_arg (*args, u32);
+  return format (s, "test-eth%d", dev_instance);
+}
+
+static uword dummy_interface_tx (vlib_main_t * vm,
+                                 vlib_node_runtime_t * node,
+                                 vlib_frame_t * frame)
+{
+  clib_warning ("you shouldn't be here, leaking buffers...");
+  return frame->n_vectors;
+}
+
+static clib_error_t *
+test_interface_admin_up_down (vnet_main_t * vnm,
+                              u32 hw_if_index,
+                              u32 flags)
+{
+  u32 hw_flags = (flags & VNET_SW_INTERFACE_FLAG_ADMIN_UP) ?
+    VNET_HW_INTERFACE_FLAG_LINK_UP : 0;
+  vnet_hw_interface_set_flags (vnm, hw_if_index, hw_flags);
+  return 0;
+}
+
+VNET_DEVICE_CLASS (test_interface_device_class,static) = {
+  .name = "Test interface",
+  .format_device_name = format_test_interface_name,
+  .tx_function = dummy_interface_tx,
+  .admin_up_down_function = test_interface_admin_up_down,
+};
+
+static u8 *hw_address;
+
+static int
+mfib_test_mk_intf (u32 ninterfaces)
+{
+    clib_error_t * error = NULL;
+    test_main_t *tm = &test_main;
+    u8 byte;
+    u32 i;
+
+    ASSERT(ninterfaces <= ARRAY_LEN(tm->hw_if_indicies));
+
+    for (i=0; i<6; i++)
+    {
+        byte = 0xd0+i;
+        vec_add1(hw_address, byte);
+    }
+
+    for (i = 0; i < ninterfaces; i++)
+    {
+        hw_address[5] = i;
+
+        error = ethernet_register_interface(vnet_get_main(),
+                                            test_interface_device_class.index,
+                                            i /* instance */,
+                                            hw_address,
+                                            &tm->hw_if_indicies[i],
+                                            /* flag change */ 0);
+
+        MFIB_TEST((NULL == error), "ADD interface %d", i);
+
+        error = vnet_hw_interface_set_flags(vnet_get_main(),
+                                            tm->hw_if_indicies[i],
+                                            VNET_HW_INTERFACE_FLAG_LINK_UP);
+        tm->hw[i] = vnet_get_hw_interface(vnet_get_main(),
+                                          tm->hw_if_indicies[i]);
+        vec_validate (ip4_main.fib_index_by_sw_if_index,
+                      tm->hw[i]->sw_if_index);
+        vec_validate (ip6_main.fib_index_by_sw_if_index,
+                      tm->hw[i]->sw_if_index);
+        ip4_main.fib_index_by_sw_if_index[tm->hw[i]->sw_if_index] = 0;
+        ip6_main.fib_index_by_sw_if_index[tm->hw[i]->sw_if_index] = 0;
+
+        vec_validate (ip4_main.mfib_index_by_sw_if_index,
+                      tm->hw[i]->sw_if_index);
+        vec_validate (ip6_main.mfib_index_by_sw_if_index,
+                      tm->hw[i]->sw_if_index);
+        ip4_main.mfib_index_by_sw_if_index[tm->hw[i]->sw_if_index] = 0;
+        ip6_main.mfib_index_by_sw_if_index[tm->hw[i]->sw_if_index] = 0;
+
+        error = vnet_sw_interface_set_flags(vnet_get_main(),
+                                            tm->hw[i]->sw_if_index,
+                                            VNET_SW_INTERFACE_FLAG_ADMIN_UP);
+        MFIB_TEST((NULL == error), "UP interface %d", i);
+    }
+    /*
+     * re-eval after the inevitable realloc
+     */
+    for (i = 0; i < ninterfaces; i++)
+    {
+        tm->hw[i] = vnet_get_hw_interface(vnet_get_main(),
+                                          tm->hw_if_indicies[i]);
+    }
+
+    return (0);
+}
+
+#define MFIB_TEST_REP(_cond, _comment, _args...)		\
+{								\
+    if (!MFIB_TEST_I(_cond, _comment, ##_args)) {		\
+        return (0);						\
+    }								\
+}
+
+static int
+mfib_test_validate_rep_v (const replicate_t *rep,
+                          u16 n_buckets,
+                          va_list ap)
+{
+    const dpo_id_t *dpo;
+    adj_index_t ai;
+    dpo_type_t dt;
+    int bucket;
+
+    MFIB_TEST_REP((n_buckets == rep->rep_n_buckets),
+                  "n_buckets = %d", rep->rep_n_buckets);
+
+    for (bucket = 0; bucket < n_buckets; bucket++)
+    {
+        dt = va_arg(ap, int);  // type promotion
+        ai = va_arg(ap, adj_index_t);
+        dpo = replicate_get_bucket_i(rep, bucket);
+
+        MFIB_TEST_REP((dt == dpo->dpoi_type),
+                      "bucket %d stacks on %U",
+                      bucket,
+                      format_dpo_type, dpo->dpoi_type);
+
+        if (DPO_RECEIVE != dt)
+        {
+            MFIB_TEST_REP((ai == dpo->dpoi_index),
+                          "bucket %d stacks on %U",
+                          bucket,
+                          format_dpo_id, dpo, 0);
+        }
+    }
+    return (!0);
+}
+
+static fib_forward_chain_type_t
+fib_forw_chain_type_from_fib_proto (fib_protocol_t proto)
+{
+    switch (proto)
+    {
+    case FIB_PROTOCOL_IP4:
+        return (FIB_FORW_CHAIN_TYPE_UNICAST_IP4);
+    case FIB_PROTOCOL_IP6:
+        return (FIB_FORW_CHAIN_TYPE_UNICAST_IP6);
+    default:
+        break;
+    }
+    ASSERT(0);
+    return (0);
+}
+
+
+static int
+mfib_test_entry (fib_node_index_t fei,
+                 mfib_entry_flags_t eflags,
+                 u16 n_buckets,
+                 ...)
+{
+    const mfib_entry_t *mfe;
+    const replicate_t *rep;
+    mfib_prefix_t pfx;
+    va_list ap;
+
+    va_start(ap, n_buckets);
+
+    mfe = mfib_entry_get(fei);
+    mfib_entry_get_prefix(fei, &pfx);
+
+    MFIB_TEST_REP((eflags == mfe->mfe_flags),
+                  "%U has %U expect %U",
+                  format_mfib_prefix, &pfx,
+                  format_mfib_entry_flags, mfe->mfe_flags,
+                  format_mfib_entry_flags, eflags);
+
+    if (0 == n_buckets)
+    {
+        MFIB_TEST_REP((DPO_DROP == mfe->mfe_rep.dpoi_type),
+                      "%U links to %U",
+                      format_mfib_prefix, &pfx,
+                      format_dpo_id, &mfe->mfe_rep, 0);
+        return (!0);
+    }
+    else
+    {
+        dpo_id_t tmp = DPO_INVALID;
+        int res;
+
+        mfib_entry_contribute_forwarding(
+            fei,
+            fib_forw_chain_type_from_fib_proto(pfx.fp_proto),
+            &tmp);
+        rep = replicate_get(tmp.dpoi_index);
+
+        MFIB_TEST_REP((DPO_REPLICATE == tmp.dpoi_type),
+                      "%U links to %U",
+                      format_mfib_prefix, &pfx,
+                      format_dpo_type, tmp.dpoi_type);
+
+        res = mfib_test_validate_rep_v(rep, n_buckets, ap);
+
+        dpo_reset(&tmp);
+
+        return (res);
+    }
+}
+
+static int
+mfib_test_entry_itf (fib_node_index_t fei,
+                     u32 sw_if_index,
+                     mfib_itf_flags_t flags)
+{
+    const mfib_entry_t *mfe;
+    const mfib_itf_t *mfi;
+    mfib_prefix_t pfx;
+
+    mfe = mfib_entry_get(fei);
+    mfi = mfib_entry_get_itf(mfe, sw_if_index);
+    mfib_entry_get_prefix(fei, &pfx);
+
+    MFIB_TEST_REP((NULL != mfi),
+                  "%U has interface %d",
+                  format_mfib_prefix, &pfx, sw_if_index);
+
+    MFIB_TEST_REP((flags == mfi->mfi_flags),
+                  "%U interface %d has flags %U expect %U",
+                  format_mfib_prefix, &pfx, sw_if_index,
+                  format_mfib_itf_flags, flags,
+                  format_mfib_itf_flags, mfi->mfi_flags);
+
+    return (!0);
+}
+
+static int
+mfib_test_entry_no_itf (fib_node_index_t fei,
+                        u32 sw_if_index)
+{
+    const mfib_entry_t *mfe;
+    const mfib_itf_t *mfi;
+    mfib_prefix_t pfx;
+
+    mfe = mfib_entry_get(fei);
+    mfi = mfib_entry_get_itf(mfe, sw_if_index);
+    mfib_entry_get_prefix(fei, &pfx);
+
+    MFIB_TEST_REP((NULL == mfi),
+                  "%U has no interface %d",
+                  format_mfib_prefix, &pfx, sw_if_index);
+
+    return (!0);
+}
+
+static int
+mfib_test_i (fib_protocol_t PROTO,
+             vnet_link_t LINKT,
+             const mfib_prefix_t *pfx_no_forward,
+             const mfib_prefix_t *pfx_s_g,
+             const mfib_prefix_t *pfx_star_g_1,
+             const mfib_prefix_t *pfx_star_g_2,
+             const mfib_prefix_t *pfx_star_g_3,
+             const mfib_prefix_t *pfx_star_g_slash_m)
+{
+    fib_node_index_t mfei, mfei_dflt, mfei_no_f, mfei_s_g, mfei_g_1, mfei_g_2, mfei_g_3, mfei_g_m;
+    u32 fib_index, n_entries, n_itfs, n_reps;
+    fib_node_index_t ai_1, ai_2, ai_3;
+    test_main_t *tm;
+
+    mfib_prefix_t all_1s;
+    memset(&all_1s, 0xfd, sizeof(all_1s));
+
+    n_entries = pool_elts(mfib_entry_pool);
+    n_itfs = pool_elts(mfib_itf_pool);
+    n_reps = pool_elts(replicate_pool);
+    tm = &test_main;
+
+    ai_1 = adj_mcast_add_or_lock(PROTO,
+                                 LINKT,
+                                 tm->hw[1]->sw_if_index);
+    ai_2 = adj_mcast_add_or_lock(PROTO,
+                                 LINKT,
+                                 tm->hw[2]->sw_if_index);
+    ai_3 = adj_mcast_add_or_lock(PROTO,
+                                 LINKT,
+                                 tm->hw[3]->sw_if_index);
+
+    MFIB_TEST(3 == adj_mcast_db_size(), "3 MCAST adjs");
+
+    /* Find or create FIB table 11 */
+    fib_index = mfib_table_find_or_create_and_lock(PROTO, 11);
+
+    mfib_prefix_t pfx_dft = {
+        .fp_len = 0,
+        .fp_proto = PROTO,
+    };
+    mfei_dflt = mfib_table_lookup_exact_match(fib_index, &pfx_dft);
+    MFIB_TEST(FIB_NODE_INDEX_INVALID != mfei_dflt, "(*,*) presnet");
+    MFIB_TEST(mfib_test_entry(mfei_dflt,
+                              MFIB_ENTRY_FLAG_DROP,
+                              0),
+              "(*,*) no replcaitions");
+
+    MFIB_TEST(FIB_NODE_INDEX_INVALID != mfei_dflt, "(*,*) presnet");
+    MFIB_TEST(mfib_test_entry(mfei_dflt,
+                              MFIB_ENTRY_FLAG_DROP,
+                              0),
+              "(*,*) no replcaitions");
+
+
+    fib_route_path_t path_via_if0 = {
+        .frp_proto = PROTO,
+        .frp_addr = zero_addr,
+        .frp_sw_if_index = tm->hw[0]->sw_if_index,
+        .frp_fib_index = ~0,
+        .frp_weight = 0,
+        .frp_flags = 0,
+    };
+
+    mfib_table_entry_path_update(fib_index,
+                                 pfx_no_forward,
+                                 MFIB_SOURCE_API,
+                                 &path_via_if0,
+                                 MFIB_ITF_FLAG_ACCEPT);
+
+    mfei_no_f = mfib_table_lookup_exact_match(fib_index, pfx_no_forward);
+    MFIB_TEST(mfib_test_entry(mfei_no_f,
+                              MFIB_ENTRY_FLAG_NONE,
+                              0),
+              "%U no replcaitions",
+              format_mfib_prefix, pfx_no_forward);
+    MFIB_TEST_NS(mfib_test_entry_itf(mfei_no_f, tm->hw[0]->sw_if_index,
+                                     MFIB_ITF_FLAG_ACCEPT));
+
+    fib_route_path_t path_via_if1 = {
+        .frp_proto = PROTO,
+        .frp_addr = zero_addr,
+        .frp_sw_if_index = tm->hw[1]->sw_if_index,
+        .frp_fib_index = ~0,
+        .frp_weight = 0,
+        .frp_flags = 0,
+    };
+    fib_route_path_t path_via_if2 = {
+        .frp_proto = PROTO,
+        .frp_addr = zero_addr,
+        .frp_sw_if_index = tm->hw[2]->sw_if_index,
+        .frp_fib_index = ~0,
+        .frp_weight = 0,
+        .frp_flags = 0,
+    };
+    fib_route_path_t path_via_if3 = {
+        .frp_proto = PROTO,
+        .frp_addr = zero_addr,
+        .frp_sw_if_index = tm->hw[3]->sw_if_index,
+        .frp_fib_index = ~0,
+        .frp_weight = 0,
+        .frp_flags = 0,
+    };
+    fib_route_path_t path_for_us = {
+        .frp_proto = PROTO,
+        .frp_addr = zero_addr,
+        .frp_sw_if_index = 0xffffffff,
+        .frp_fib_index = ~0,
+        .frp_weight = 0,
+        .frp_flags = FIB_ROUTE_PATH_LOCAL,
+    };
+
+    /*
+     * An (S,G) with 1 accepting and 3 forwarding paths
+     */
+    mfib_table_entry_path_update(fib_index,
+                                 pfx_s_g,
+                                 MFIB_SOURCE_API,
+                                 &path_via_if0,
+                                 MFIB_ITF_FLAG_ACCEPT);
+    mfib_table_entry_path_update(fib_index,
+                                 pfx_s_g,
+                                 MFIB_SOURCE_API,
+                                 &path_via_if1,
+                                 MFIB_ITF_FLAG_FORWARD);
+    mfib_table_entry_path_update(fib_index,
+                                 pfx_s_g,
+                                 MFIB_SOURCE_API,
+                                 &path_via_if2,
+                                 MFIB_ITF_FLAG_FORWARD);
+    mfib_table_entry_path_update(fib_index,
+                                 pfx_s_g,
+                                 MFIB_SOURCE_API,
+                                 &path_via_if3,
+                                 (MFIB_ITF_FLAG_FORWARD |
+                                  MFIB_ITF_FLAG_NEGATE_SIGNAL));
+
+    mfei_s_g = mfib_table_lookup_exact_match(fib_index, pfx_s_g);
+
+    MFIB_TEST(FIB_NODE_INDEX_INVALID != mfei_s_g,
+              "%U present",
+              format_mfib_prefix, pfx_s_g);
+    MFIB_TEST(mfib_test_entry(mfei_s_g,
+                              MFIB_ENTRY_FLAG_NONE,
+                              3,
+                              DPO_ADJACENCY_MCAST, ai_1,
+                              DPO_ADJACENCY_MCAST, ai_2,
+                              DPO_ADJACENCY_MCAST, ai_3),
+              "%U replicate ok",
+              format_mfib_prefix, pfx_s_g);
+    MFIB_TEST_NS(mfib_test_entry_itf(mfei_s_g, tm->hw[0]->sw_if_index,
+                                     MFIB_ITF_FLAG_ACCEPT));
+    MFIB_TEST_NS(mfib_test_entry_itf(mfei_s_g, tm->hw[1]->sw_if_index,
+                                     MFIB_ITF_FLAG_FORWARD));
+    MFIB_TEST_NS(mfib_test_entry_itf(mfei_s_g, tm->hw[2]->sw_if_index,
+                                     MFIB_ITF_FLAG_FORWARD));
+    MFIB_TEST_NS(mfib_test_entry_itf(mfei_s_g, tm->hw[3]->sw_if_index,
+                                     (MFIB_ITF_FLAG_FORWARD |
+                                      MFIB_ITF_FLAG_NEGATE_SIGNAL)));
+
+    /*
+     * A (*,G), which the same G as the (S,G).
+     * different paths. test our LPM.
+     */
+    mfei_g_1 = mfib_table_entry_path_update(fib_index,
+                                            pfx_star_g_1,
+                                            MFIB_SOURCE_API,
+                                            &path_via_if0,
+                                            MFIB_ITF_FLAG_ACCEPT);
+    mfib_table_entry_path_update(fib_index,
+                                 pfx_star_g_1,
+                                 MFIB_SOURCE_API,
+                                 &path_via_if1,
+                                 MFIB_ITF_FLAG_FORWARD);
+
+    /*
+     * test we find the *,G and S,G via LPM and exact matches
+     */
+    mfei = mfib_table_lookup_exact_match(fib_index,
+                                         pfx_star_g_1);
+    MFIB_TEST(mfei == mfei_g_1,
+              "%U found via exact match",
+              format_mfib_prefix, pfx_star_g_1);
+    MFIB_TEST(mfib_test_entry(mfei,
+                              MFIB_ENTRY_FLAG_NONE,
+                              1,
+                              DPO_ADJACENCY_MCAST, ai_1),
+              "%U replicate ok",
+              format_mfib_prefix, pfx_star_g_1);
+
+    mfei = mfib_table_lookup(fib_index,
+                             pfx_star_g_1);
+    MFIB_TEST(mfei == mfei_g_1,
+              "%U found via LP match",
+              format_mfib_prefix, pfx_star_g_1);
+
+    MFIB_TEST(mfib_test_entry(mfei,
+                              MFIB_ENTRY_FLAG_NONE,
+                              1,
+                              DPO_ADJACENCY_MCAST, ai_1),
+              "%U replicate ok",
+              format_mfib_prefix, pfx_star_g_1);
+
+    mfei = mfib_table_lookup_exact_match(fib_index, pfx_s_g);
+    MFIB_TEST(mfei == mfei_s_g,
+              "%U found via exact match",
+              format_mfib_prefix, pfx_s_g);
+
+    MFIB_TEST(mfib_test_entry(mfei,
+                              MFIB_ENTRY_FLAG_NONE,
+                              3,
+                              DPO_ADJACENCY_MCAST, ai_1,
+                              DPO_ADJACENCY_MCAST, ai_2,
+                              DPO_ADJACENCY_MCAST, ai_3),
+              "%U replicate OK",
+              format_mfib_prefix, pfx_s_g);
+    mfei = mfib_table_lookup(fib_index, pfx_s_g);
+    MFIB_TEST(mfei == mfei_s_g,
+              "%U found via LP match",
+              format_mfib_prefix, pfx_s_g);
+
+    MFIB_TEST(mfib_test_entry(mfei,
+                              MFIB_ENTRY_FLAG_NONE,
+                              3,
+                              DPO_ADJACENCY_MCAST, ai_1,
+                              DPO_ADJACENCY_MCAST, ai_2,
+                              DPO_ADJACENCY_MCAST, ai_3),
+              "%U replicate OK",
+              format_mfib_prefix, pfx_s_g);
+
+    /*
+     * A (*,G/m), which the same root G as the (*,G).
+     * different paths. test our LPM.
+     */
+    mfei_g_m = mfib_table_entry_path_update(fib_index,
+                                            pfx_star_g_slash_m,
+                                            MFIB_SOURCE_API,
+                                            &path_via_if2,
+                                            MFIB_ITF_FLAG_ACCEPT);
+    mfib_table_entry_path_update(fib_index,
+                                 pfx_star_g_slash_m,
+                                 MFIB_SOURCE_API,
+                                 &path_via_if3,
+                                 MFIB_ITF_FLAG_FORWARD);
+
+    /*
+     * test we find the (*,G/m), (*,G) and (S,G) via LPM and exact matches
+     */
+    mfei = mfib_table_lookup_exact_match(fib_index, pfx_star_g_1);
+    MFIB_TEST((mfei_g_1 == mfei),
+              "%U found via DP LPM: %d",
+              format_mfib_prefix, pfx_star_g_1, mfei);
+
+    MFIB_TEST(mfib_test_entry(mfei,
+                              MFIB_ENTRY_FLAG_NONE,
+                              1,
+                              DPO_ADJACENCY_MCAST, ai_1),
+              "%U replicate ok",
+              format_mfib_prefix, pfx_star_g_1);
+
+    mfei = mfib_table_lookup(fib_index, pfx_star_g_1);
+
+    MFIB_TEST(mfib_test_entry(mfei,
+                              MFIB_ENTRY_FLAG_NONE,
+                              1,
+                              DPO_ADJACENCY_MCAST, ai_1),
+              "%U replicate ok",
+              format_mfib_prefix, pfx_star_g_1);
+
+    mfei = mfib_table_lookup_exact_match(fib_index, pfx_s_g);
+
+    MFIB_TEST(mfib_test_entry(mfei,
+                              MFIB_ENTRY_FLAG_NONE,
+                              3,
+                              DPO_ADJACENCY_MCAST, ai_1,
+                              DPO_ADJACENCY_MCAST, ai_2,
+                              DPO_ADJACENCY_MCAST, ai_3),
+              "%U replicate OK",
+              format_mfib_prefix, pfx_s_g);
+    mfei = mfib_table_lookup(fib_index, pfx_s_g);
+
+    MFIB_TEST(mfib_test_entry(mfei,
+                              MFIB_ENTRY_FLAG_NONE,
+                              3,
+                              DPO_ADJACENCY_MCAST, ai_1,
+                              DPO_ADJACENCY_MCAST, ai_2,
+                              DPO_ADJACENCY_MCAST, ai_3),
+              "%U replicate OK",
+              format_mfib_prefix, pfx_s_g);
+
+    mfei = mfib_table_lookup_exact_match(fib_index, pfx_star_g_slash_m);
+    MFIB_TEST(mfei = mfei_g_m,
+              "%U Found via exact match",
+              format_mfib_prefix, pfx_star_g_slash_m);
+    MFIB_TEST(mfib_test_entry(mfei,
+                              MFIB_ENTRY_FLAG_NONE,
+                              1,
+                              DPO_ADJACENCY_MCAST, ai_3),
+              "%U replicate OK",
+              format_mfib_prefix, pfx_star_g_slash_m);
+    MFIB_TEST(mfei_g_m == mfib_table_lookup(fib_index, pfx_star_g_slash_m),
+              "%U found via LPM",
+              format_mfib_prefix, pfx_star_g_slash_m);
+
+    /*
+     * Add a for-us path
+     */
+    mfei = mfib_table_entry_path_update(fib_index,
+                                        pfx_s_g,
+                                        MFIB_SOURCE_API,
+                                        &path_for_us,
+                                        MFIB_ITF_FLAG_FORWARD);
+
+    MFIB_TEST(mfib_test_entry(mfei,
+                              MFIB_ENTRY_FLAG_NONE,
+                              4,
+                              DPO_ADJACENCY_MCAST, ai_1,
+                              DPO_ADJACENCY_MCAST, ai_2,
+                              DPO_ADJACENCY_MCAST, ai_3,
+                              DPO_RECEIVE, 0),
+              "%U replicate OK",
+              format_mfib_prefix, pfx_s_g);
+
+    /*
+     * remove a for-us path
+     */
+    mfib_table_entry_path_remove(fib_index,
+                                 pfx_s_g,
+                                 MFIB_SOURCE_API,
+                                 &path_for_us);
+
+    MFIB_TEST(mfib_test_entry(mfei,
+                              MFIB_ENTRY_FLAG_NONE,
+                              3,
+                              DPO_ADJACENCY_MCAST, ai_1,
+                              DPO_ADJACENCY_MCAST, ai_2,
+                              DPO_ADJACENCY_MCAST, ai_3),
+              "%U replicate OK",
+              format_mfib_prefix, pfx_s_g);
+
+    /*
+     * update an existing forwarding path to be only accepting
+     *   - expect it to be removed from the replication set.
+     */
+    mfib_table_entry_path_update(fib_index,
+                                 pfx_s_g,
+                                 MFIB_SOURCE_API,
+                                 &path_via_if3,
+                                 MFIB_ITF_FLAG_ACCEPT);
+
+    MFIB_TEST(mfib_test_entry(mfei,
+                              MFIB_ENTRY_FLAG_NONE,
+                              2,
+                              DPO_ADJACENCY_MCAST, ai_1,
+                              DPO_ADJACENCY_MCAST, ai_2),
+              "%U replicate OK",
+              format_mfib_prefix, pfx_s_g);
+    MFIB_TEST_NS(mfib_test_entry_itf(mfei, tm->hw[0]->sw_if_index,
+                                     MFIB_ITF_FLAG_ACCEPT));
+    MFIB_TEST_NS(mfib_test_entry_itf(mfei, tm->hw[1]->sw_if_index,
+                                     MFIB_ITF_FLAG_FORWARD));
+    MFIB_TEST_NS(mfib_test_entry_itf(mfei, tm->hw[2]->sw_if_index,
+                                     MFIB_ITF_FLAG_FORWARD));
+    MFIB_TEST_NS(mfib_test_entry_itf(mfei, tm->hw[3]->sw_if_index,
+                                     MFIB_ITF_FLAG_ACCEPT));
+    /*
+     * Make the path forwarding again
+     *  - expect it to be added back to the replication set
+     */
+    mfib_table_entry_path_update(fib_index,
+                                 pfx_s_g,
+                                 MFIB_SOURCE_API,
+                                 &path_via_if3,
+                                 (MFIB_ITF_FLAG_FORWARD |
+                                  MFIB_ITF_FLAG_ACCEPT |
+                                  MFIB_ITF_FLAG_NEGATE_SIGNAL));
+
+    mfei = mfib_table_lookup_exact_match(fib_index,
+                                         pfx_s_g);
+
+    MFIB_TEST(mfib_test_entry(mfei,
+                              MFIB_ENTRY_FLAG_NONE,
+                              3,
+                              DPO_ADJACENCY_MCAST, ai_1,
+                              DPO_ADJACENCY_MCAST, ai_2,
+                              DPO_ADJACENCY_MCAST, ai_3),
+              "%U replicate OK",
+              format_mfib_prefix, pfx_s_g);
+    MFIB_TEST_NS(mfib_test_entry_itf(mfei, tm->hw[0]->sw_if_index,
+                                     MFIB_ITF_FLAG_ACCEPT));
+     MFIB_TEST_NS(mfib_test_entry_itf(mfei, tm->hw[1]->sw_if_index,
+                                     MFIB_ITF_FLAG_FORWARD));
+    MFIB_TEST_NS(mfib_test_entry_itf(mfei, tm->hw[2]->sw_if_index,
+                                     MFIB_ITF_FLAG_FORWARD));
+    MFIB_TEST_NS(mfib_test_entry_itf(mfei, tm->hw[3]->sw_if_index,
+                                     (MFIB_ITF_FLAG_FORWARD |
+                                      MFIB_ITF_FLAG_ACCEPT |
+                                      MFIB_ITF_FLAG_NEGATE_SIGNAL)));
+
+    /*
+     * update flags on the entry
+     */
+    mfib_table_entry_update(fib_index,
+                            pfx_s_g,
+                            MFIB_SOURCE_API,
+                            MFIB_ENTRY_FLAG_SIGNAL);
+    MFIB_TEST(mfib_test_entry(mfei,
+                              MFIB_ENTRY_FLAG_SIGNAL,
+                              3,
+                              DPO_ADJACENCY_MCAST, ai_1,
+                              DPO_ADJACENCY_MCAST, ai_2,
+                              DPO_ADJACENCY_MCAST, ai_3),
+              "%U replicate OK",
+              format_mfib_prefix, pfx_s_g);
+
+    /*
+     * remove paths
+     */
+    mfib_table_entry_path_remove(fib_index,
+                                 pfx_s_g,
+                                 MFIB_SOURCE_API,
+                                 &path_via_if3);
+
+    MFIB_TEST(mfib_test_entry(mfei,
+                              MFIB_ENTRY_FLAG_SIGNAL,
+                              2,
+                              DPO_ADJACENCY_MCAST, ai_1,
+                              DPO_ADJACENCY_MCAST, ai_2),
+              "%U replicate OK",
+              format_mfib_prefix, pfx_s_g);
+    MFIB_TEST_NS(mfib_test_entry_itf(mfei, tm->hw[0]->sw_if_index,
+                                     MFIB_ITF_FLAG_ACCEPT));
+    MFIB_TEST_NS(mfib_test_entry_itf(mfei, tm->hw[1]->sw_if_index,
+                                     MFIB_ITF_FLAG_FORWARD));
+    MFIB_TEST_NS(mfib_test_entry_itf(mfei, tm->hw[2]->sw_if_index,
+                                     MFIB_ITF_FLAG_FORWARD));
+    MFIB_TEST_NS(mfib_test_entry_no_itf(mfei, tm->hw[3]->sw_if_index));
+
+    mfib_table_entry_path_remove(fib_index,
+                                 pfx_s_g,
+                                 MFIB_SOURCE_API,
+                                 &path_via_if1);
+
+    MFIB_TEST(mfib_test_entry(mfei,
+                              MFIB_ENTRY_FLAG_SIGNAL,
+                              1,
+                              DPO_ADJACENCY_MCAST, ai_2),
+              "%U replicate OK",
+              format_mfib_prefix, pfx_s_g);
+    MFIB_TEST_NS(mfib_test_entry_itf(mfei, tm->hw[0]->sw_if_index,
+                                     MFIB_ITF_FLAG_ACCEPT));
+    MFIB_TEST_NS(mfib_test_entry_itf(mfei, tm->hw[2]->sw_if_index,
+                                     MFIB_ITF_FLAG_FORWARD));
+    MFIB_TEST_NS(mfib_test_entry_no_itf(mfei, tm->hw[3]->sw_if_index));
+
+    /*
+     * remove the accpeting only interface
+     */
+    mfib_table_entry_path_remove(fib_index,
+                                 pfx_s_g,
+                                 MFIB_SOURCE_API,
+                                 &path_via_if0);
+
+    MFIB_TEST(mfib_test_entry(mfei,
+                              MFIB_ENTRY_FLAG_SIGNAL,
+                              1,
+                              DPO_ADJACENCY_MCAST, ai_2),
+              "%U replicate OK",
+              format_mfib_prefix, pfx_s_g);
+    MFIB_TEST_NS(mfib_test_entry_itf(mfei, tm->hw[2]->sw_if_index,
+                                     MFIB_ITF_FLAG_FORWARD));
+    MFIB_TEST_NS(mfib_test_entry_no_itf(mfei, tm->hw[0]->sw_if_index));
+    MFIB_TEST_NS(mfib_test_entry_no_itf(mfei, tm->hw[1]->sw_if_index));
+    MFIB_TEST_NS(mfib_test_entry_no_itf(mfei, tm->hw[3]->sw_if_index));
+
+    /*
+     * remove the last path, the entry still has flags so it remains
+     */
+    mfib_table_entry_path_remove(fib_index,
+                                 pfx_s_g,
+                                 MFIB_SOURCE_API,
+                                 &path_via_if2);
+
+    MFIB_TEST(mfib_test_entry(mfei,
+                              MFIB_ENTRY_FLAG_SIGNAL,
+                              0),
+              "%U no replications",
+              format_mfib_prefix, pfx_s_g);
+
+    /*
+     * update flags on the entry
+     */
+    mfib_table_entry_update(fib_index,
+                            pfx_s_g,
+                            MFIB_SOURCE_API,
+                            (MFIB_ENTRY_FLAG_SIGNAL |
+                             MFIB_ENTRY_FLAG_CONNECTED));
+    MFIB_TEST(mfib_test_entry(mfei,
+                              (MFIB_ENTRY_FLAG_SIGNAL |
+                               MFIB_ENTRY_FLAG_CONNECTED),
+                              0),
+              "%U no replications",
+              format_mfib_prefix, pfx_s_g);
+
+    /*
+     * An entry with a NS interface
+     */
+    mfei_g_2 = mfib_table_entry_path_update(fib_index,
+                                            pfx_star_g_2,
+                                            MFIB_SOURCE_API,
+                                            &path_via_if0,
+                                            (MFIB_ITF_FLAG_ACCEPT |
+                                             MFIB_ITF_FLAG_NEGATE_SIGNAL));
+    MFIB_TEST(mfib_test_entry(mfei_g_2,
+                              MFIB_ENTRY_FLAG_NONE,
+                              0),
+              "%U No replications",
+              format_mfib_prefix, pfx_star_g_2);
+
+    /*
+     * Simulate a signal from the data-plane
+     */
+    {
+        mfib_entry_t *mfe;
+        mfib_itf_t *mfi;
+
+        mfe = mfib_entry_get(mfei_g_2);
+        mfi = mfib_entry_get_itf(mfe, path_via_if0.frp_sw_if_index);
+
+        mfib_signal_push(mfe, mfi, NULL);
+    }
+
+    /*
+     * An entry with a NS interface
+     */
+    mfei_g_3 = mfib_table_entry_path_update(fib_index,
+                                            pfx_star_g_3,
+                                            MFIB_SOURCE_API,
+                                            &path_via_if0,
+                                            (MFIB_ITF_FLAG_ACCEPT |
+                                             MFIB_ITF_NEGATE_SIGNAL));
+    MFIB_TEST(mfib_test_entry(mfei_g_3,
+                              MFIB_ENTRY_FLAG_NONE,
+                              0),
+              "%U No replications",
+              format_mfib_prefix, pfx_star_g_3);
+
+    /*
+     * Simulate a signal from the data-plane
+     */
+    {
+        mfib_entry_t *mfe;
+        mfib_itf_t *mfi;
+
+        mfe = mfib_entry_get(mfei_g_3);
+        mfi = mfib_entry_get_itf(mfe, path_via_if0.frp_sw_if_index);
+
+        mfib_signal_push(mfe, mfi, NULL);
+    }
+
+    if (FIB_PROTOCOL_IP6 == PROTO)
+    {
+        /*
+         * All the entries are present. let's ensure we can find them all
+         * via exact and longest prefix matches.
+         */
+        /*
+         * A source address we will never match
+         */
+        ip6_address_t src = {
+            .as_u64[0] = clib_host_to_net_u64(0x3001000000000000),
+            .as_u64[1] = clib_host_to_net_u64(0xffffffffffffffff),
+        };
+
+        /*
+         * Find the (*,G/m)
+         */
+        MFIB_TEST((mfei_g_m == ip6_mfib_table_lookup2(
+                                   ip6_mfib_get(fib_index),
+                                   &src,
+                                   &pfx_star_g_slash_m->fp_grp_addr.ip6)),
+                  "%U found via DP LPM grp=%U",
+                  format_mfib_prefix, pfx_star_g_slash_m,
+                  format_ip6_address, &pfx_star_g_slash_m->fp_grp_addr.ip6);
+
+        ip6_address_t tmp = pfx_star_g_slash_m->fp_grp_addr.ip6;
+        tmp.as_u8[15] = 0xff;
+
+        MFIB_TEST((mfei_g_m == ip6_mfib_table_lookup2(
+                                   ip6_mfib_get(fib_index),
+                                   &pfx_s_g->fp_src_addr.ip6,
+                                   &tmp)),
+                  "%U found via DP LPM grp=%U",
+                  format_mfib_prefix, pfx_star_g_slash_m,
+                  format_ip6_address, &tmp);
+
+        /*
+         * Find the (S,G).
+         */
+        mfei = ip6_mfib_table_lookup2(ip6_mfib_get(fib_index),
+                                      &pfx_s_g->fp_src_addr.ip6,
+                                      &pfx_s_g->fp_grp_addr.ip6);
+        MFIB_TEST((mfei_s_g == mfei),
+                  "%U found via DP LPM: %d",
+                  format_mfib_prefix, pfx_s_g, mfei);
+
+        /*
+         * Find the 3 (*,G) s
+         */
+        mfei = ip6_mfib_table_lookup2(ip6_mfib_get(fib_index),
+                                      &src,
+                                      &pfx_star_g_1->fp_grp_addr.ip6);
+        MFIB_TEST((mfei_g_1 == mfei),
+                  "%U found via DP LPM: %d",
+                  format_mfib_prefix, pfx_star_g_1, mfei);
+        mfei = ip6_mfib_table_lookup2(ip6_mfib_get(fib_index),
+                                      &src,
+                                      &pfx_star_g_2->fp_grp_addr.ip6);
+        MFIB_TEST((mfei_g_2 == mfei),
+                  "%U found via DP LPM: %d",
+                  format_mfib_prefix, pfx_star_g_2, mfei);
+        mfei = ip6_mfib_table_lookup2(ip6_mfib_get(fib_index),
+                                      &src,
+                                      &pfx_star_g_3->fp_grp_addr.ip6);
+        MFIB_TEST((mfei_g_3 == mfei),
+                  "%U found via DP LPM: %d",
+                  format_mfib_prefix, pfx_star_g_3, mfei);
+    }
+
+    /*
+     * remove flags on the entry. This is the last of the
+     * state associated with the entry, so now it goes.
+     */
+    mfib_table_entry_update(fib_index,
+                            pfx_s_g,
+                            MFIB_SOURCE_API,
+                            MFIB_ENTRY_FLAG_NONE);
+    mfei = mfib_table_lookup_exact_match(fib_index,
+                                         pfx_s_g);
+    MFIB_TEST(FIB_NODE_INDEX_INVALID == mfei,
+              "%U gone",
+              format_mfib_prefix, pfx_s_g);
+
+    /*
+     * remove the last path on the no forward entry - the last entry
+     */
+    mfib_table_entry_path_remove(fib_index,
+                                 pfx_no_forward,
+                                 MFIB_SOURCE_API,
+                                 &path_via_if0);
+
+    mfei = mfib_table_lookup_exact_match(fib_index, pfx_no_forward);
+    MFIB_TEST(FIB_NODE_INDEX_INVALID == mfei,
+              "%U gone",
+              format_mfib_prefix, pfx_no_forward);
+
+    /*
+     * hard delete the (*,232.1.1.1)
+     */
+    mfib_table_entry_delete(fib_index,
+                            pfx_star_g_1,
+                            MFIB_SOURCE_API);
+
+    mfei = mfib_table_lookup_exact_match(fib_index, pfx_star_g_1);
+    MFIB_TEST(FIB_NODE_INDEX_INVALID == mfei,
+              "%U gone",
+              format_mfib_prefix, pfx_star_g_1);
+    /*
+     * remove the entry whilst the signal is pending
+     */
+    mfib_table_entry_delete(fib_index,
+                            pfx_star_g_2,
+                            MFIB_SOURCE_API);
+
+    mfei = mfib_table_lookup_exact_match(fib_index, pfx_star_g_2);
+    MFIB_TEST(FIB_NODE_INDEX_INVALID == mfei,
+              "%U Gone",
+              format_mfib_prefix, pfx_star_g_2);
+    mfib_table_entry_delete(fib_index,
+                            pfx_star_g_3,
+                            MFIB_SOURCE_API);
+
+    mfei = mfib_table_lookup_exact_match(fib_index, pfx_star_g_3);
+    MFIB_TEST(FIB_NODE_INDEX_INVALID == mfei,
+              "%U Gone",
+              format_mfib_prefix, pfx_star_g_3);
+
+    mfib_table_entry_delete(fib_index,
+                            pfx_star_g_slash_m,
+                            MFIB_SOURCE_API);
+
+    mfei = mfib_table_lookup_exact_match(fib_index, pfx_star_g_slash_m);
+    MFIB_TEST(FIB_NODE_INDEX_INVALID == mfei,
+              "%U Gone",
+              format_mfib_prefix, pfx_star_g_slash_m);
+
+    /*
+     * Unlock the table - it's the last lock so should be gone thereafter
+     */
+    mfib_table_unlock(fib_index, PROTO);
+
+    MFIB_TEST((FIB_NODE_INDEX_INVALID ==
+               mfib_table_find(PROTO, fib_index)),
+              "MFIB table %d gone", fib_index);
+
+    adj_unlock(ai_1);
+    adj_unlock(ai_2);
+    adj_unlock(ai_3);
+
+    /*
+     * test we've leaked no resources
+     */
+    MFIB_TEST(0 == adj_mcast_db_size(), "%d MCAST adjs", adj_mcast_db_size());
+    MFIB_TEST(n_reps == pool_elts(replicate_pool), "%d=%d replicates",
+              n_reps, pool_elts(replicate_pool));
+    MFIB_TEST(n_entries == pool_elts(mfib_entry_pool),
+              " No more entries %d!=%d",
+              n_entries, pool_elts(mfib_entry_pool));
+    MFIB_TEST(n_itfs == pool_elts(mfib_itf_pool),
+              " No more Interfaces %d!=%d",
+              n_itfs, pool_elts(mfib_itf_pool));
+
+    return (0);
+}
+
+static int
+mfib_test_v4 (void)
+{
+    const mfib_prefix_t pfx_224_s_8 = {
+        .fp_len = 8,
+        .fp_proto = FIB_PROTOCOL_IP4,
+        .fp_grp_addr = {
+            .ip4.as_u32 = clib_host_to_net_u32(0xe0000000),
+        }
+    };
+    const mfib_prefix_t pfx_1_1_1_1_c_239_1_1_1 = {
+        .fp_len = 64,
+        .fp_proto = FIB_PROTOCOL_IP4,
+        .fp_grp_addr = {
+            .ip4.as_u32 = clib_host_to_net_u32(0xef010101),
+        },
+        .fp_src_addr = {
+            .ip4.as_u32 = clib_host_to_net_u32(0x01010101),
+        },
+    };
+    const mfib_prefix_t pfx_239_1_1_1 = {
+        .fp_len = 32,
+        .fp_proto = FIB_PROTOCOL_IP4,
+        .fp_grp_addr = {
+            .ip4.as_u32 = clib_host_to_net_u32(0xef010101),
+        },
+        .fp_src_addr = {
+            .ip4.as_u32 = 0,
+        },
+    };
+    const mfib_prefix_t pfx_239_1_1_2 = {
+        .fp_len = 32,
+        .fp_proto = FIB_PROTOCOL_IP4,
+        .fp_grp_addr = {
+            .ip4.as_u32 = clib_host_to_net_u32(0xef010102),
+        },
+        .fp_src_addr = {
+            .ip4.as_u32 = 0,
+        },
+    };
+    const mfib_prefix_t pfx_239_1_1_3 = {
+        .fp_len = 32,
+        .fp_proto = FIB_PROTOCOL_IP4,
+        .fp_grp_addr = {
+            .ip4.as_u32 = clib_host_to_net_u32(0xef010103),
+        },
+        .fp_src_addr = {
+            .ip4.as_u32 = 0,
+        },
+    };
+    const mfib_prefix_t pfx_239 = {
+        .fp_len = 8,
+        .fp_proto = FIB_PROTOCOL_IP4,
+        .fp_grp_addr = {
+            .ip4.as_u32 = clib_host_to_net_u32(0xef000000),
+        },
+        .fp_src_addr = {
+            .ip4.as_u32 = 0,
+        },
+    };
+
+    return (mfib_test_i(FIB_PROTOCOL_IP4,
+                        VNET_LINK_IP4,
+                        &pfx_224_s_8,
+                        &pfx_1_1_1_1_c_239_1_1_1,
+                        &pfx_239_1_1_1,
+                        &pfx_239_1_1_2,
+                        &pfx_239_1_1_3,
+                        &pfx_239));
+}
+
+static int
+mfib_test_v6 (void)
+{
+    const mfib_prefix_t pfx_ffd_s_12 = {
+        .fp_len = 12,
+        .fp_proto = FIB_PROTOCOL_IP6,
+        .fp_grp_addr = {
+            .ip6.as_u64[0] = clib_host_to_net_u64(0xffd0000000000000),
+        }
+    };
+    const mfib_prefix_t pfx_2001_1_c_ff_1 = {
+        .fp_len = 256,
+        .fp_proto = FIB_PROTOCOL_IP6,
+        .fp_grp_addr = {
+            .ip6.as_u64[0] = clib_host_to_net_u64(0xff01000000000000),
+            .ip6.as_u64[1] = clib_host_to_net_u64(0x0000000000000001),
+        },
+        .fp_src_addr = {
+            .ip6.as_u64[0] = clib_host_to_net_u64(0x2001000000000000),
+            .ip6.as_u64[1] = clib_host_to_net_u64(0x0000000000000001),
+        },
+    };
+    const mfib_prefix_t pfx_ff_1 = {
+        .fp_len = 128,
+        .fp_proto = FIB_PROTOCOL_IP6,
+        .fp_grp_addr = {
+            .ip6.as_u64[0] = clib_host_to_net_u64(0xff01000000000000),
+            .ip6.as_u64[1] = clib_host_to_net_u64(0x0000000000000001),
+        },
+    };
+    const mfib_prefix_t pfx_ff_2 = {
+        .fp_len = 128,
+        .fp_proto = FIB_PROTOCOL_IP6,
+        .fp_grp_addr = {
+            .ip6.as_u64[0] = clib_host_to_net_u64(0xff01000000000000),
+            .ip6.as_u64[1] = clib_host_to_net_u64(0x0000000000000002),
+        },
+    };
+    const mfib_prefix_t pfx_ff_3 = {
+        /*
+         * this is the ALL DHCP routers address
+         */
+        .fp_len = 128,
+        .fp_proto = FIB_PROTOCOL_IP6,
+        .fp_grp_addr = {
+            .ip6.as_u64[0] = clib_host_to_net_u64(0xff02000100000000),
+            .ip6.as_u64[1] = clib_host_to_net_u64(0x0000000000000002),
+        },
+    };
+    const mfib_prefix_t pfx_ff = {
+        .fp_len = 16,
+        .fp_proto = FIB_PROTOCOL_IP6,
+        .fp_grp_addr = {
+            .ip6.as_u64[0] = clib_host_to_net_u64(0xff01000000000000),
+            .ip6.as_u64[1] = clib_host_to_net_u64(0x0000000000000000),
+        },
+    };
+
+    return (mfib_test_i(FIB_PROTOCOL_IP6,
+                        VNET_LINK_IP6,
+                        &pfx_ffd_s_12,
+                        &pfx_2001_1_c_ff_1,
+                        &pfx_ff_1,
+                        &pfx_ff_2,
+                        &pfx_ff_3,
+                        &pfx_ff));
+}
+
+static clib_error_t *
+mfib_test (vlib_main_t * vm,
+           unformat_input_t * input,
+           vlib_cli_command_t * cmd_arg)
+{
+    int res = 0;
+
+    res += mfib_test_mk_intf(4);
+    res += mfib_test_v4();
+    res += mfib_test_v6();
+
+    if (res)
+    {
+        return clib_error_return(0, "MFIB Unit Test Failed");
+    }
+    else
+    {
+        return (NULL);
+    }
+}
+
+VLIB_CLI_COMMAND (test_fib_command, static) = {
+    .path = "test mfib",
+    .short_help = "fib unit tests - DO NOT RUN ON A LIVE SYSTEM",
+    .function = mfib_test,
+};
+
+clib_error_t *
+mfib_test_init (vlib_main_t *vm)
+{
+    return 0;
+}
+
+VLIB_INIT_FUNCTION (mfib_test_init);
diff --git a/src/vnet/mfib/mfib_types.c b/src/vnet/mfib/mfib_types.c
new file mode 100644
index 00000000..6d77c3d8
--- /dev/null
+++ b/src/vnet/mfib/mfib_types.c
@@ -0,0 +1,213 @@
+ /*
+ * Copyright (c) 2016 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <vnet/mfib/mfib_types.h>
+
+#include <vnet/ip/ip.h>
+
+/**
+ * String names for each flag
+ */
+static const char *mfib_flag_names[] = MFIB_ENTRY_NAMES_SHORT;
+static const char *mfib_flag_names_long[] = MFIB_ENTRY_NAMES_LONG;
+
+static const char *mfib_itf_flag_long_names[] = MFIB_ITF_NAMES_LONG;
+static const char *mfib_itf_flag_names[] = MFIB_ITF_NAMES_SHORT;
+
+u8 *
+format_mfib_prefix (u8 * s, va_list * args)
+{
+    mfib_prefix_t *fp = va_arg (*args, mfib_prefix_t *);
+
+    /*
+     * protocol specific so it prints ::/0 correctly.
+     */
+    switch (fp->fp_proto)
+    {
+    case FIB_PROTOCOL_IP6:
+    {
+        ip6_address_t p6 = fp->fp_grp_addr.ip6;
+        u32 len = (fp->fp_len > 128 ? 128 : fp->fp_len);
+
+        ip6_address_mask(&p6, &(ip6_main.fib_masks[len]));
+
+        if (ip6_address_is_zero(&fp->fp_src_addr.ip6))
+        {
+            s = format(s, "(*, ");
+        }
+        else
+        {
+            s = format (s, "(%U, ", format_ip6_address, &fp->fp_src_addr.ip6);
+        }
+        s = format (s, "%U", format_ip6_address, &p6);
+        s = format (s, "/%d)", len);
+        break;
+    }
+    case FIB_PROTOCOL_IP4:
+    {
+        ip4_address_t p4 = fp->fp_grp_addr.ip4;
+        u32 len = (fp->fp_len > 32 ? 32 : fp->fp_len);
+
+        p4.as_u32 &= ip4_main.fib_masks[len];
+
+        if (0 == fp->fp_src_addr.ip4.as_u32)
+        {
+            s = format(s, "(*, ");
+        }
+        else
+        {
+            s = format (s, "(%U, ", format_ip4_address, &fp->fp_src_addr.ip4);
+        }
+        s = format (s, "%U", format_ip4_address, &p4);
+        s = format (s, "/%d)", len);
+        break;
+    }
+    case FIB_PROTOCOL_MPLS:
+        break;
+    }
+
+    return (s);
+}
+
+u8 *
+format_mfib_entry_flags (u8 * s, va_list * args)
+{
+    mfib_entry_attribute_t attr;
+    mfib_entry_flags_t flags;
+
+    flags = va_arg (*args, mfib_entry_flags_t);
+
+    if (MFIB_ENTRY_FLAG_NONE != flags) {
+        s = format(s, " flags:");
+        FOR_EACH_MFIB_ATTRIBUTE(attr) {
+            if ((1<<attr) & flags) {
+                s = format (s, "%s,", mfib_flag_names[attr]);
+            }
+        }
+    }
+
+    return (s);
+}
+
+u8 *
+format_mfib_itf_flags (u8 * s, va_list * args)
+{
+    mfib_itf_attribute_t attr;
+    mfib_itf_flags_t flags;
+
+    flags = va_arg (*args, mfib_itf_flags_t);
+
+    FOR_EACH_MFIB_ITF_ATTRIBUTE(attr) {
+        if ((1<<attr) & flags) {
+            s = format (s, "%s,", mfib_itf_flag_long_names[attr]);
+        }
+    }
+
+    return (s);
+}
+
+uword
+unformat_mfib_itf_flags (unformat_input_t * input,
+                         va_list * args)
+{
+    mfib_itf_flags_t old, *iflags = va_arg (*args, mfib_itf_flags_t*);
+    mfib_itf_attribute_t attr;
+
+    old = *iflags;
+    FOR_EACH_MFIB_ITF_ATTRIBUTE(attr) {
+        if (unformat (input, mfib_itf_flag_long_names[attr]))
+            *iflags |= (1 << attr);
+    }
+    FOR_EACH_MFIB_ITF_ATTRIBUTE(attr) {
+        if (unformat (input, mfib_itf_flag_names[attr]))
+            *iflags |= (1 << attr);
+    }
+
+    return (old == *iflags ? 0 : 1);
+}
+
+uword
+unformat_mfib_entry_flags (unformat_input_t * input,
+                           va_list * args)
+{
+    mfib_entry_flags_t old, *eflags = va_arg (*args, mfib_entry_flags_t*);
+    mfib_entry_attribute_t attr;
+
+    old = *eflags;
+    FOR_EACH_MFIB_ATTRIBUTE(attr) {
+        if (unformat (input, mfib_flag_names[attr]))
+            *eflags |= (1 << attr);
+    }
+
+    return (old == *eflags ? 0 : 1);
+}
+
+clib_error_t *
+mfib_show_route_flags (vlib_main_t * vm,
+                       unformat_input_t * main_input,
+                       vlib_cli_command_t * cmd)
+{
+    mfib_entry_attribute_t attr;
+
+    FOR_EACH_MFIB_ATTRIBUTE(attr) {
+        vlib_cli_output(vm, "%s = %s",
+                        mfib_flag_names[attr],
+                        mfib_flag_names_long[attr]);
+    }
+
+    return (NULL);
+}
+
+/*?
+ * This command display the set of support flags applicable to the MFIB route
+ */
+/* *INDENT-OFF* */
+VLIB_CLI_COMMAND (mfib_route_flags_command, static) =
+{
+  .path = "sh mfib route flags",
+  .short_help = "Flags applicable to an MFIB route",
+  .function = mfib_show_route_flags,
+  .is_mp_safe = 1,
+};
+/* *INDENT-ON* */
+
+clib_error_t *
+mfib_show_itf_flags (vlib_main_t * vm,
+                     unformat_input_t * main_input,
+                     vlib_cli_command_t * cmd)
+{
+    mfib_itf_attribute_t attr;
+
+    FOR_EACH_MFIB_ITF_ATTRIBUTE(attr) {
+        vlib_cli_output(vm, "%s = %s",
+                        mfib_itf_flag_names[attr],
+                        mfib_itf_flag_long_names[attr]);
+    }
+
+    return (NULL);
+}
+
+/*?
+ * This command display the set of support flags applicable to the MFIB route
+ */
+/* *INDENT-OFF* */
+VLIB_CLI_COMMAND (mfib_itf_flags_command, static) =
+{
+  .path = "sh mfib itf flags",
+  .short_help = "Flags applicable to an MFIB interfaces",
+  .function = mfib_show_itf_flags,
+  .is_mp_safe = 1,
+};
+/* *INDENT-ON* */
diff --git a/src/vnet/mfib/mfib_types.h b/src/vnet/mfib/mfib_types.h
new file mode 100644
index 00000000..37898a07
--- /dev/null
+++ b/src/vnet/mfib/mfib_types.h
@@ -0,0 +1,185 @@
+ /*
+ * Copyright (c) 2016 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef __MFIB_TYPES_H__
+#define __MFIB_TYPES_H__
+
+#include <vnet/fib/fib_types.h>
+
+/**
+ * Aggregrate type for a prefix
+ */
+typedef struct mfib_prefix_t_ {
+    /**
+     * The mask length
+     */
+    u16 fp_len;
+
+    /**
+     * protocol type
+     */
+    fib_protocol_t fp_proto;
+
+    /**
+     * Pad to keep the address 4 byte aligned
+     */
+    u8 ___fp___pad;
+
+    /**
+     * The address type is not deriveable from the fp_addr member.
+     * If it's v4, then the first 3 u32s of the address will be 0.
+     * v6 addresses (even v4 mapped ones) have at least 2 u32s assigned
+     * to non-zero values. true. but when it's all zero, one cannot decide.
+     */
+    ip46_address_t fp_grp_addr;
+    ip46_address_t fp_src_addr;
+} mfib_prefix_t;
+
+typedef enum mfib_entry_attribute_t_
+{
+    MFIB_ENTRY_ATTRIBUTE_FIRST = 0,
+    /**
+     * The control planes needs packets mathing this entry to generate
+     * a signal.
+     */
+    MFIB_ENTRY_SIGNAL =  MFIB_ENTRY_ATTRIBUTE_FIRST,
+    /**
+     * Drop all traffic to this route
+     */
+    MFIB_ENTRY_DROP,
+    /**
+     * The control plane needs to be informed of coneected sources
+     */
+    MFIB_ENTRY_CONNECTED,
+    /**
+     * Accept packets from any incpoming interface
+     *        Use with extreme caution
+     */
+    MFIB_ENTRY_ACCEPT_ALL_ITF,
+    MFIB_ENTRY_INHERIT_ACCEPT,
+    MFIB_ENTRY_ATTRIBUTE_LAST = MFIB_ENTRY_INHERIT_ACCEPT,
+} mfib_entry_attribute_t;
+
+#define FOR_EACH_MFIB_ATTRIBUTE(_item)			\
+    for (_item = MFIB_ENTRY_ATTRIBUTE_FIRST;		\
+	 _item <= MFIB_ENTRY_ATTRIBUTE_LAST;		\
+	 _item++)
+
+#define MFIB_ENTRY_NAMES_SHORT  {          \
+    [MFIB_ENTRY_SIGNAL]         = "S",     \
+    [MFIB_ENTRY_CONNECTED]      = "C",     \
+    [MFIB_ENTRY_DROP]           = "D",     \
+    [MFIB_ENTRY_ACCEPT_ALL_ITF] = "AA",    \
+    [MFIB_ENTRY_INHERIT_ACCEPT] = "IA",    \
+}
+
+#define MFIB_ENTRY_NAMES_LONG  {                    \
+    [MFIB_ENTRY_SIGNAL]         = "Signal",         \
+    [MFIB_ENTRY_CONNECTED]      = "Connected",      \
+    [MFIB_ENTRY_DROP]           = "Drop",           \
+    [MFIB_ENTRY_ACCEPT_ALL_ITF] = "Accept-all-itf", \
+    [MFIB_ENTRY_INHERIT_ACCEPT] = "Inherit-Accept", \
+}
+
+typedef enum mfib_entry_flags_t_
+{
+    MFIB_ENTRY_FLAG_NONE,
+    MFIB_ENTRY_FLAG_SIGNAL = (1 << MFIB_ENTRY_SIGNAL),
+    MFIB_ENTRY_FLAG_DROP = (1 << MFIB_ENTRY_DROP),
+    MFIB_ENTRY_FLAG_CONNECTED = (1 << MFIB_ENTRY_CONNECTED),
+    MFIB_ENTRY_FLAG_INHERIT_ACCEPT = (1 << MFIB_ENTRY_INHERIT_ACCEPT),
+    MFIB_ENTRY_FLAG_ACCEPT_ALL_ITF = (1 << MFIB_ENTRY_ACCEPT_ALL_ITF),
+} mfib_entry_flags_t;
+
+typedef enum mfib_itf_attribute_t_
+{
+    MFIB_ITF_ATTRIBUTE_FIRST,
+    MFIB_ITF_NEGATE_SIGNAL = MFIB_ITF_ATTRIBUTE_FIRST,
+    MFIB_ITF_ACCEPT,
+    MFIB_ITF_FORWARD,
+    MFIB_ITF_SIGNAL_PRESENT,
+    MFIB_ITF_DONT_PRESERVE,
+    MFIB_ITF_ATTRIBUTE_LAST = MFIB_ITF_DONT_PRESERVE,
+} mfib_itf_attribute_t;
+
+#define FOR_EACH_MFIB_ITF_ATTRIBUTE(_item)	       	\
+    for (_item = MFIB_ITF_ATTRIBUTE_FIRST;       	\
+	 _item <= MFIB_ITF_ATTRIBUTE_LAST;		\
+	 _item++)
+
+#define MFIB_ITF_NAMES_SHORT  {             \
+    [MFIB_ITF_NEGATE_SIGNAL] = "NS",        \
+    [MFIB_ITF_ACCEPT] = "A",                \
+    [MFIB_ITF_FORWARD] = "F",               \
+    [MFIB_ITF_SIGNAL_PRESENT] = "SP",       \
+    [MFIB_ITF_DONT_PRESERVE] = "DP",        \
+}
+
+#define MFIB_ITF_NAMES_LONG  {                    \
+    [MFIB_ITF_NEGATE_SIGNAL] = "Negate-Signal",   \
+    [MFIB_ITF_ACCEPT] = "Accept",                 \
+    [MFIB_ITF_FORWARD] = "Forward",               \
+    [MFIB_ITF_SIGNAL_PRESENT] = "Signal-Present", \
+    [MFIB_ITF_DONT_PRESERVE] = "Don't-Preserve", \
+}
+
+typedef enum mfib_itf_flags_t_
+{
+    MFIB_ITF_FLAG_NONE,
+    MFIB_ITF_FLAG_NEGATE_SIGNAL = (1 << MFIB_ITF_NEGATE_SIGNAL),
+    MFIB_ITF_FLAG_ACCEPT = (1 << MFIB_ITF_ACCEPT),
+    MFIB_ITF_FLAG_FORWARD = (1 << MFIB_ITF_FORWARD),
+    MFIB_ITF_FLAG_SIGNAL_PRESENT = (1 << MFIB_ITF_SIGNAL_PRESENT),
+    MFIB_ITF_FLAG_DONT_PRESERVE = (1 << MFIB_ITF_DONT_PRESERVE),
+} mfib_itf_flags_t;
+
+/**
+ * Possible [control plane] sources of MFIB entries
+ */
+typedef enum mfib_source_t_
+{
+    MFIB_SOURCE_SPECIAL,
+    MFIB_SOURCE_API,
+    MFIB_SOURCE_CLI,
+    MFIB_SOURCE_VXLAN,
+    MFIB_SOURCE_DHCP,
+    MFIB_SOURCE_DEFAULT_ROUTE,
+} mfib_source_t;
+
+#define MFIB_SOURCE_NAMES {                        \
+    [MFIB_SOURCE_SPECIAL] = "Special",             \
+    [MFIB_SOURCE_API] = "API",                     \
+    [MFIB_SOURCE_CLI] = "CLI",                     \
+    [MFIB_SOURCE_DHCP] = "DHCP",                   \
+    [MFIB_SOURCE_VXLAN] = "VXLAN",                 \
+    [MFIB_SOURCE_DEFAULT_ROUTE] = "Default Route", \
+}
+
+/**
+ * \brief Compare two prefixes for equality
+ */
+extern int mfib_prefix_cmp(const mfib_prefix_t *p1,
+                           const mfib_prefix_t *p2);
+
+extern u8 * format_mfib_prefix(u8 * s, va_list * args);
+
+extern u8 *format_mfib_entry_flags(u8 * s, va_list * args);
+extern u8 *format_mfib_itf_flags(u8 * s, va_list * args);
+extern uword unformat_mfib_itf_flags(unformat_input_t * input,
+                                     va_list * args);
+extern uword unformat_mfib_entry_flags(unformat_input_t * input,
+                                       va_list * args);
+
+#endif
diff --git a/src/vnet/misc.c b/src/vnet/misc.c
index 4c8c4cad..9cfe8394 100644
--- a/src/vnet/misc.c
+++ b/src/vnet/misc.c
@@ -83,6 +83,9 @@ vnet_main_init (vlib_main_t * vm)
   if ((error = vlib_call_init_function (vm, fib_module_init)))
     return error;
 
+  if ((error = vlib_call_init_function (vm, mfib_module_init)))
+    return error;
+
   if ((error = vlib_call_init_function (vm, ip_main_init)))
     return error;
 
diff --git a/src/vnet/rewrite.h b/src/vnet/rewrite.h
index 00c1efbd..ce2bce3a 100644
--- a/src/vnet/rewrite.h
+++ b/src/vnet/rewrite.h
@@ -64,6 +64,16 @@ typedef CLIB_PACKED (struct {
      Used for MTU check after packet rewrite. */
   u16 max_l3_packet_bytes;
 
+  /* When dynamically writing a multicast destination L2 addresss
+   * this is the offset within the address to start writing n
+   * bytes of the IP mcast address */
+  u8 dst_mcast_offset;
+
+  /* When dynamically writing a multicast destination L2 addresss
+   * this is the number of bytes of the dest IP address to write into
+   * the MAC rewrite */
+  u8 dst_mcast_n_bytes;
+
   /* Rewrite string starting at end and going backwards. */
   u8 data[0];
 }) vnet_rewrite_header_t;
@@ -261,6 +271,27 @@ _vnet_rewrite_two_headers (vnet_rewrite_header_t * h0,
 			     sizeof ((rw0).rewrite_data),		\
 			     (most_likely_size))
 
+always_inline void
+_vnet_fixup_one_header (vnet_rewrite_header_t * h0,
+			u8 * addr, u32 addr_len,
+			u8 * packet0, int clear_first_bit)
+{
+  /* location to write to in the packet */
+  u8 *p0 = packet0 - h0->dst_mcast_offset;
+  u8 *p1 = p0;
+  /* location to write from in the L3 dest address */
+  u8 *a0 = addr + addr_len - h0->dst_mcast_n_bytes;
+
+  clib_memcpy (p0, a0, h0->dst_mcast_n_bytes);
+  if (clear_first_bit)
+    *p1 &= 0x7f;
+}
+
+#define vnet_fixup_one_header(rw0,addr,p0,clear_first_bit)              \
+  _vnet_fixup_one_header (&((rw0).rewrite_header),                      \
+                          (u8*)(addr), sizeof((*addr)),                 \
+                          (u8*)(p0), (clear_first_bit))
+
 #define VNET_REWRITE_FOR_SW_INTERFACE_ADDRESS_BROADCAST ((void *) 0)
 /** Deprecated */
 void vnet_rewrite_for_sw_interface (struct vnet_main_t *vnm,
diff --git a/src/vnet/sr/sr.c b/src/vnet/sr/sr.c
index 5d0275d9..f30c0da9 100644
--- a/src/vnet/sr/sr.c
+++ b/src/vnet/sr/sr.c
@@ -2161,7 +2161,7 @@ sr_fix_dst_addr (vlib_main_t * vm,
 
 	  adj0 =
 	    ip_get_adjacency (lm, vnet_buffer (b0)->ip.adj_index[VLIB_TX]);
-	  next0 = adj0->mcast_group_index;
+	  next0 = adj0->if_address_index;
 
 	  /* We should be pointing at an Ethernet header... */
 	  eh0 = vlib_buffer_get_current (b0);
@@ -3088,7 +3088,7 @@ set_ip6_sr_rewrite_fn (vlib_main_t * vm,
   adj->rewrite_header.node_index = sr_fix_dst_addr_node.index;
 
   /* $$$$$ hack... steal the mcast group index */
-  adj->mcast_group_index =
+  adj->if_address_index =
     vlib_node_add_next (vm, sr_fix_dst_addr_node.index,
 			hi->output_node_index);
 
diff --git a/src/vnet/util/radix.c b/src/vnet/util/radix.c
new file mode 100644
index 00000000..ff0b0f7b
--- /dev/null
+++ b/src/vnet/util/radix.c
@@ -0,0 +1,1104 @@
+/*	$NetBSD: radix.c,v 1.47 2016/12/12 03:55:57 ozaki-r Exp $	*/
+
+/*
+ * Copyright (c) 1988, 1989, 1993
+ *	The Regents of the University of California.  All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ * 3. Neither the name of the University nor the names of its contributors
+ *    may be used to endorse or promote products derived from this software
+ *    without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ *	@(#)radix.c	8.6 (Berkeley) 10/17/95
+ */
+
+/*
+ * Routines to build and maintain radix trees for routing lookups.
+ */
+
+#include <vnet/util/radix.h>
+
+typedef void (*rn_printer_t)(void *, const char *fmt, ...);
+
+static int max_keylen = 33; // me
+struct radix_mask *rn_mkfreelist;
+struct radix_node_head *mask_rnhead;
+static char *addmask_key;
+static const char normal_chars[] =
+    {0, 0x80, 0xc0, 0xe0, 0xf0, 0xf8, 0xfc, 0xfe, -1};
+static char *rn_zeros, *rn_ones;
+
+#define rn_masktop (mask_rnhead->rnh_treetop)
+
+static int rn_satisfies_leaf(const char *, struct radix_node *, int);
+static int rn_lexobetter(const void *, const void *);
+static struct radix_mask *rn_new_radix_mask(struct radix_node *,
+    struct radix_mask *);
+static struct radix_node *rn_walknext(struct radix_node *, rn_printer_t,
+    void *);
+static struct radix_node *rn_walkfirst(struct radix_node *, rn_printer_t,
+    void *);
+static void rn_nodeprint(struct radix_node *, rn_printer_t, void *,
+    const char *);
+
+#define	SUBTREE_OPEN	"[ "
+#define	SUBTREE_CLOSE	" ]"
+
+#ifdef RN_DEBUG
+static void rn_treeprint(struct radix_node_head *, rn_printer_t, void *);
+#endif /* RN_DEBUG */
+
+#define MIN(x,y) (((x)<(y))?(x):(y))
+
+static struct radix_mask*
+rm_alloc (void)
+{
+    struct radix_mask *rm = clib_mem_alloc(sizeof(struct radix_mask));
+
+    memset(rm, 0, sizeof(*rm));
+
+    return (rm);
+}
+
+static void
+rm_free (struct radix_mask *rm)
+{
+    clib_mem_free(rm);
+}
+
+#define R_Malloc(p, t, n)                               \
+{                                                       \
+    p = (t) clib_mem_alloc((unsigned int)(n));          \
+    memset(p, 0, n);                                    \
+}
+#define Free(p) clib_mem_free((p))
+#define log(a,b, c...)
+#define bool i32
+
+/*
+ * The data structure for the keys is a radix tree with one way
+ * branching removed.  The index rn_b at an internal node n represents a bit
+ * position to be tested.  The tree is arranged so that all descendants
+ * of a node n have keys whose bits all agree up to position rn_b - 1.
+ * (We say the index of n is rn_b.)
+ *
+ * There is at least one descendant which has a one bit at position rn_b,
+ * and at least one with a zero there.
+ *
+ * A route is determined by a pair of key and mask.  We require that the
+ * bit-wise logical and of the key and mask to be the key.
+ * We define the index of a route to associated with the mask to be
+ * the first bit number in the mask where 0 occurs (with bit number 0
+ * representing the highest order bit).
+ *
+ * We say a mask is normal if every bit is 0, past the index of the mask.
+ * If a node n has a descendant (k, m) with index(m) == index(n) == rn_b,
+ * and m is a normal mask, then the route applies to every descendant of n.
+ * If the index(m) < rn_b, this implies the trailing last few bits of k
+ * before bit b are all 0, (and hence consequently true of every descendant
+ * of n), so the route applies to all descendants of the node as well.
+ *
+ * Similar logic shows that a non-normal mask m such that
+ * index(m) <= index(n) could potentially apply to many children of n.
+ * Thus, for each non-host route, we attach its mask to a list at an internal
+ * node as high in the tree as we can go.
+ *
+ * The present version of the code makes use of normal routes in short-
+ * circuiting an explicit mask and compare operation when testing whether
+ * a key satisfies a normal route, and also in remembering the unique leaf
+ * that governs a subtree.
+ */
+
+struct radix_node *
+rn_search(
+	const void *v_arg,
+	struct radix_node *head)
+{
+	const u8 * const v = v_arg;
+	struct radix_node *x;
+
+	for (x = head; x->rn_b >= 0;) {
+		if (x->rn_bmask & v[x->rn_off])
+			x = x->rn_r;
+		else
+			x = x->rn_l;
+	}
+	return x;
+}
+
+struct radix_node *
+rn_search_m(
+	const void *v_arg,
+	struct radix_node *head,
+	const void *m_arg)
+{
+	struct radix_node *x;
+	const u8 * const v = v_arg;
+	const u8 * const m = m_arg;
+
+	for (x = head; x->rn_b >= 0;) {
+		if ((x->rn_bmask & m[x->rn_off]) &&
+		    (x->rn_bmask & v[x->rn_off]))
+			x = x->rn_r;
+		else
+			x = x->rn_l;
+	}
+	return x;
+}
+
+int
+rn_refines(
+	const void *m_arg,
+	const void *n_arg)
+{
+	const char *m = m_arg;
+	const char *n = n_arg;
+	const char *lim = n + *(const u8 *)n;
+	const char *lim2 = lim;
+	int longer = (*(const u8 *)n++) - (int)(*(const u8 *)m++);
+	int masks_are_equal = 1;
+
+	if (longer > 0)
+		lim -= longer;
+	while (n < lim) {
+		if (*n & ~(*m))
+			return 0;
+		if (*n++ != *m++)
+			masks_are_equal = 0;
+	}
+	while (n < lim2)
+		if (*n++)
+			return 0;
+	if (masks_are_equal && (longer < 0))
+		for (lim2 = m - longer; m < lim2; )
+			if (*m++)
+				return 1;
+	return !masks_are_equal;
+}
+
+struct radix_node *
+rn_lookup(
+	const void *v_arg,
+	const void *m_arg,
+	struct radix_node_head *head)
+{
+	struct radix_node *x;
+	const char *netmask = NULL;
+
+	if (m_arg) {
+		if ((x = rn_addmask(m_arg, 1, head->rnh_treetop->rn_off)) == 0)
+			return NULL;
+		netmask = x->rn_key;
+	}
+	x = rn_match(v_arg, head);
+	if (x != NULL && netmask != NULL) {
+		while (x != NULL && x->rn_mask != netmask)
+			x = x->rn_dupedkey;
+	}
+	return x;
+}
+
+static int
+rn_satisfies_leaf(
+	const char *trial,
+	struct radix_node *leaf,
+	int skip)
+{
+	const char *cp = trial;
+	const char *cp2 = leaf->rn_key;
+	const char *cp3 = leaf->rn_mask;
+	const char *cplim;
+	int length = MIN(*(const u8 *)cp, *(const u8 *)cp2);
+
+	if (cp3 == 0)
+		cp3 = rn_ones;
+	else
+		length = MIN(length, *(const u8 *)cp3);
+	cplim = cp + length; cp3 += skip; cp2 += skip;
+	for (cp += skip; cp < cplim; cp++, cp2++, cp3++)
+		if ((*cp ^ *cp2) & *cp3)
+			return 0;
+	return 1;
+}
+
+struct radix_node *
+rn_match(
+	const void *v_arg,
+	struct radix_node_head *head)
+{
+	const char * const v = v_arg;
+	struct radix_node *t = head->rnh_treetop;
+	struct radix_node *top = t;
+	struct radix_node *x;
+	struct radix_node *saved_t;
+	const char *cp = v;
+	const char *cp2;
+	const char *cplim;
+	int off = t->rn_off;
+	int vlen = *(const u8 *)cp;
+	int matched_off;
+	int test, b, rn_b;
+
+	/*
+	 * Open code rn_search(v, top) to avoid overhead of extra
+	 * subroutine call.
+	 */
+	for (; t->rn_b >= 0; ) {
+		if (t->rn_bmask & cp[t->rn_off])
+			t = t->rn_r;
+		else
+			t = t->rn_l;
+	}
+	/*
+	 * See if we match exactly as a host destination
+	 * or at least learn how many bits match, for normal mask finesse.
+	 *
+	 * It doesn't hurt us to limit how many bytes to check
+	 * to the length of the mask, since if it matches we had a genuine
+	 * match and the leaf we have is the most specific one anyway;
+	 * if it didn't match with a shorter length it would fail
+	 * with a long one.  This wins big for class B&C netmasks which
+	 * are probably the most common case...
+	 */
+	if (t->rn_mask)
+		vlen = *(const u8 *)t->rn_mask;
+	cp += off; cp2 = t->rn_key + off; cplim = v + vlen;
+	for (; cp < cplim; cp++, cp2++)
+		if (*cp != *cp2)
+			goto on1;
+	/*
+	 * This extra grot is in case we are explicitly asked
+	 * to look up the default.  Ugh!
+	 */
+	if ((t->rn_flags & RNF_ROOT) && t->rn_dupedkey)
+		t = t->rn_dupedkey;
+	return t;
+on1:
+	test = (*cp ^ *cp2) & 0xff; /* find first bit that differs */
+	for (b = 7; (test >>= 1) > 0;)
+		b--;
+	matched_off = cp - v;
+	b += matched_off << 3;
+	rn_b = -1 - b;
+	/*
+	 * If there is a host route in a duped-key chain, it will be first.
+	 */
+	if ((saved_t = t)->rn_mask == 0)
+		t = t->rn_dupedkey;
+	for (; t; t = t->rn_dupedkey)
+		/*
+		 * Even if we don't match exactly as a host,
+		 * we may match if the leaf we wound up at is
+		 * a route to a net.
+		 */
+		if (t->rn_flags & RNF_NORMAL) {
+			if (rn_b <= t->rn_b)
+				return t;
+		} else if (rn_satisfies_leaf(v, t, matched_off))
+				return t;
+	t = saved_t;
+	/* start searching up the tree */
+	do {
+		struct radix_mask *m;
+		t = t->rn_p;
+		m = t->rn_mklist;
+		if (m) {
+			/*
+			 * If non-contiguous masks ever become important
+			 * we can restore the masking and open coding of
+			 * the search and satisfaction test and put the
+			 * calculation of "off" back before the "do".
+			 */
+			do {
+				if (m->rm_flags & RNF_NORMAL) {
+					if (rn_b <= m->rm_b)
+						return m->rm_leaf;
+				} else {
+					off = MIN(t->rn_off, matched_off);
+					x = rn_search_m(v, t, m->rm_mask);
+					while (x && x->rn_mask != m->rm_mask)
+						x = x->rn_dupedkey;
+					if (x && rn_satisfies_leaf(v, x, off))
+						return x;
+				}
+				m = m->rm_mklist;
+			} while (m);
+		}
+	} while (t != top);
+	return NULL;
+}
+
+static void
+rn_nodeprint(struct radix_node *rn, rn_printer_t printer, void *arg,
+    const char *delim)
+{
+	(*printer)(arg, "%s(%s%p: p<%p> l<%p> r<%p>)",
+	    delim, ((void *)rn == arg) ? "*" : "", rn, rn->rn_p,
+	    rn->rn_l, rn->rn_r);
+}
+
+#ifdef RN_DEBUG
+int	rn_debug =  1;
+
+static void
+rn_dbg_print(void *arg, const char *fmt, ...)
+{
+	va_list ap;
+
+	va_start(ap, fmt);
+	vlog(LOG_DEBUG, fmt, ap);
+	va_end(ap);
+}
+
+static void
+rn_treeprint(struct radix_node_head *h, rn_printer_t printer, void *arg)
+{
+	struct radix_node *dup, *rn;
+	const char *delim;
+
+	if (printer == NULL)
+		return;
+
+	rn = rn_walkfirst(h->rnh_treetop, printer, arg);
+	for (;;) {
+		/* Process leaves */
+		delim = "";
+		for (dup = rn; dup != NULL; dup = dup->rn_dupedkey) {
+			if ((dup->rn_flags & RNF_ROOT) != 0)
+				continue;
+			rn_nodeprint(dup, printer, arg, delim);
+			delim = ", ";
+		}
+		rn = rn_walknext(rn, printer, arg);
+		if (rn->rn_flags & RNF_ROOT)
+			return;
+	}
+	/* NOTREACHED */
+}
+
+#define	traverse(__head, __rn)	rn_treeprint((__head), rn_dbg_print, (__rn))
+#endif /* RN_DEBUG */
+
+struct radix_node *
+rn_newpair(
+	const void *v,
+	int b,
+	struct radix_node nodes[2])
+{
+	struct radix_node *tt = nodes;
+	struct radix_node *t = tt + 1;
+	t->rn_b = b; t->rn_bmask = 0x80 >> (b & 7);
+	t->rn_l = tt; t->rn_off = b >> 3;
+	tt->rn_b = -1; tt->rn_key = v; tt->rn_p = t;
+	tt->rn_flags = t->rn_flags = RNF_ACTIVE;
+	return t;
+}
+
+struct radix_node *
+rn_insert(
+	const void *v_arg,
+	struct radix_node_head *head,
+	int *dupentry,
+	struct radix_node nodes[2])
+{
+	struct radix_node *top = head->rnh_treetop;
+	struct radix_node *t = rn_search(v_arg, top);
+	struct radix_node *tt;
+	const char *v = v_arg;
+	int head_off = top->rn_off;
+	int vlen = *((const u8 *)v);
+	const char *cp = v + head_off;
+	int b;
+    	/*
+	 * Find first bit at which v and t->rn_key differ
+	 */
+    {
+	const char *cp2 = t->rn_key + head_off;
+	const char *cplim = v + vlen;
+	int cmp_res;
+
+	while (cp < cplim)
+		if (*cp2++ != *cp++)
+			goto on1;
+	*dupentry = 1;
+	return t;
+on1:
+	*dupentry = 0;
+	cmp_res = (cp[-1] ^ cp2[-1]) & 0xff;
+	for (b = (cp - v) << 3; cmp_res; b--)
+		cmp_res >>= 1;
+    }
+    {
+	struct radix_node *p, *x = top;
+	cp = v;
+	do {
+		p = x;
+		if (cp[x->rn_off] & x->rn_bmask)
+			x = x->rn_r;
+		else x = x->rn_l;
+	} while (b > (unsigned) x->rn_b); /* x->rn_b < b && x->rn_b >= 0 */
+#ifdef RN_DEBUG
+	if (rn_debug)
+		log(LOG_DEBUG, "%s: Going In:\n", __func__), traverse(head, p);
+#endif
+	t = rn_newpair(v_arg, b, nodes); tt = t->rn_l;
+	if ((cp[p->rn_off] & p->rn_bmask) == 0)
+		p->rn_l = t;
+	else
+		p->rn_r = t;
+	x->rn_p = t; t->rn_p = p; /* frees x, p as temp vars below */
+	if ((cp[t->rn_off] & t->rn_bmask) == 0) {
+		t->rn_r = x;
+	} else {
+		t->rn_r = tt; t->rn_l = x;
+	}
+#ifdef RN_DEBUG
+	if (rn_debug) {
+		log(LOG_DEBUG, "%s: Coming Out:\n", __func__),
+		    traverse(head, p);
+	}
+#endif /* RN_DEBUG */
+    }
+	return tt;
+}
+
+struct radix_node *
+rn_addmask(
+	const void *n_arg,
+	int search,
+	int skip)
+{
+	const char *netmask = n_arg;
+	const char *cp;
+	const char *cplim;
+	struct radix_node *x;
+	struct radix_node *saved_x;
+	int b = 0, mlen, j;
+	int maskduplicated, m0, isnormal;
+	static int last_zeroed = 0;
+
+	if ((mlen = *(const u8 *)netmask) > max_keylen)
+		mlen = max_keylen;
+	if (skip == 0)
+		skip = 1;
+	if (mlen <= skip)
+		return mask_rnhead->rnh_nodes;
+	if (skip > 1)
+		memmove(addmask_key + 1, rn_ones + 1, skip - 1);
+	if ((m0 = mlen) > skip)
+		memmove(addmask_key + skip, netmask + skip, mlen - skip);
+	/*
+	 * Trim trailing zeroes.
+	 */
+	for (cp = addmask_key + mlen; (cp > addmask_key) && cp[-1] == 0;)
+		cp--;
+	mlen = cp - addmask_key;
+	if (mlen <= skip) {
+		if (m0 >= last_zeroed)
+			last_zeroed = mlen;
+		return mask_rnhead->rnh_nodes;
+	}
+	if (m0 < last_zeroed)
+		memset(addmask_key + m0, 0, last_zeroed - m0);
+	*addmask_key = last_zeroed = mlen;
+	x = rn_search(addmask_key, rn_masktop);
+	if (memcmp(addmask_key, x->rn_key, mlen) != 0)
+		x = 0;
+	if (x || search)
+		return x;
+	R_Malloc(x, struct radix_node *, max_keylen + 2 * sizeof (*x));
+	if ((saved_x = x) == NULL)
+		return NULL;
+	memset(x, 0, max_keylen + 2 * sizeof (*x));
+	cp = netmask = (void *)(x + 2);
+	memmove(x + 2, addmask_key, mlen);
+	x = rn_insert(cp, mask_rnhead, &maskduplicated, x);
+	if (maskduplicated) {
+                log(LOG_ERR, "rn_addmask: mask impossibly already in tree\n");
+		Free(saved_x);
+		return x;
+	}
+	/*
+	 * Calculate index of mask, and check for normalcy.
+	 */
+	cplim = netmask + mlen; isnormal = 1;
+	for (cp = netmask + skip; (cp < cplim) && *(const u8 *)cp == 0xff;)
+		cp++;
+	if (cp != cplim) {
+		for (j = 0x80; (j & *cp) != 0; j >>= 1)
+			b++;
+		if (*cp != normal_chars[b] || cp != (cplim - 1))
+			isnormal = 0;
+	}
+	b += (cp - netmask) << 3;
+	x->rn_b = -1 - b;
+	if (isnormal)
+		x->rn_flags |= RNF_NORMAL;
+	return x;
+}
+
+static int	/* XXX: arbitrary ordering for non-contiguous masks */
+rn_lexobetter(
+	const void *m_arg,
+	const void *n_arg)
+{
+	const u8 *mp = m_arg;
+	const u8 *np = n_arg;
+	const u8 *lim;
+
+	if (*mp > *np)
+		return 1;  /* not really, but need to check longer one first */
+	if (*mp == *np)
+		for (lim = mp + *mp; mp < lim;)
+			if (*mp++ > *np++)
+				return 1;
+	return 0;
+}
+
+static struct radix_mask *
+rn_new_radix_mask(
+	struct radix_node *tt,
+	struct radix_mask *next)
+{
+	struct radix_mask *m;
+
+	m = rm_alloc();
+	if (m == NULL) {
+		log(LOG_ERR, "Mask for route not entered\n");
+		return NULL;
+	}
+	memset(m, 0, sizeof(*m));
+	m->rm_b = tt->rn_b;
+	m->rm_flags = tt->rn_flags;
+	if (tt->rn_flags & RNF_NORMAL)
+		m->rm_leaf = tt;
+	else
+		m->rm_mask = tt->rn_mask;
+	m->rm_mklist = next;
+	tt->rn_mklist = m;
+	return m;
+}
+
+struct radix_node *
+rn_addroute(
+	const void *v_arg,
+	const void *n_arg,
+	struct radix_node_head *head,
+	struct radix_node treenodes[2])
+{
+	const char *v = v_arg, *netmask = n_arg;
+	struct radix_node *t, *x = NULL, *tt;
+	struct radix_node *saved_tt, *top = head->rnh_treetop;
+	short b = 0, b_leaf = 0;
+	int keyduplicated;
+	const char *mmask;
+	struct radix_mask *m, **mp;
+
+	/*
+	 * In dealing with non-contiguous masks, there may be
+	 * many different routes which have the same mask.
+	 * We will find it useful to have a unique pointer to
+	 * the mask to speed avoiding duplicate references at
+	 * nodes and possibly save time in calculating indices.
+	 */
+	if (netmask != NULL) {
+		if ((x = rn_addmask(netmask, 0, top->rn_off)) == NULL)
+			return NULL;
+		b_leaf = x->rn_b;
+		b = -1 - x->rn_b;
+		netmask = x->rn_key;
+	}
+	/*
+	 * Deal with duplicated keys: attach node to previous instance
+	 */
+	saved_tt = tt = rn_insert(v, head, &keyduplicated, treenodes);
+	if (keyduplicated) {
+		for (t = tt; tt != NULL; t = tt, tt = tt->rn_dupedkey) {
+			if (tt->rn_mask == netmask)
+				return NULL;
+			if (netmask == NULL ||
+			    (tt->rn_mask != NULL &&
+			     (b_leaf < tt->rn_b || /* index(netmask) > node */
+			       rn_refines(netmask, tt->rn_mask) ||
+			       rn_lexobetter(netmask, tt->rn_mask))))
+				break;
+		}
+		/*
+		 * If the mask is not duplicated, we wouldn't
+		 * find it among possible duplicate key entries
+		 * anyway, so the above test doesn't hurt.
+		 *
+		 * We sort the masks for a duplicated key the same way as
+		 * in a masklist -- most specific to least specific.
+		 * This may require the unfortunate nuisance of relocating
+		 * the head of the list.
+		 *
+		 * We also reverse, or doubly link the list through the
+		 * parent pointer.
+		 */
+		if (tt == saved_tt) {
+			struct	radix_node *xx = x;
+			/* link in at head of list */
+			(tt = treenodes)->rn_dupedkey = t;
+			tt->rn_flags = t->rn_flags;
+			tt->rn_p = x = t->rn_p;
+			t->rn_p = tt;
+			if (x->rn_l == t)
+				x->rn_l = tt;
+			else
+				x->rn_r = tt;
+			saved_tt = tt;
+			x = xx;
+		} else {
+			(tt = treenodes)->rn_dupedkey = t->rn_dupedkey;
+			t->rn_dupedkey = tt;
+			tt->rn_p = t;
+			if (tt->rn_dupedkey)
+				tt->rn_dupedkey->rn_p = tt;
+		}
+		tt->rn_key = v;
+		tt->rn_b = -1;
+		tt->rn_flags = RNF_ACTIVE;
+	}
+	/*
+	 * Put mask in tree.
+	 */
+	if (netmask != NULL) {
+		tt->rn_mask = netmask;
+		tt->rn_b = x->rn_b;
+		tt->rn_flags |= x->rn_flags & RNF_NORMAL;
+	}
+	t = saved_tt->rn_p;
+	if (keyduplicated)
+		goto on2;
+	b_leaf = -1 - t->rn_b;
+	if (t->rn_r == saved_tt)
+		x = t->rn_l;
+	else
+		x = t->rn_r;
+	/* Promote general routes from below */
+	if (x->rn_b < 0) {
+		for (mp = &t->rn_mklist; x != NULL; x = x->rn_dupedkey) {
+			if (x->rn_mask != NULL && x->rn_b >= b_leaf &&
+			    x->rn_mklist == NULL) {
+				*mp = m = rn_new_radix_mask(x, NULL);
+				if (m != NULL)
+					mp = &m->rm_mklist;
+			}
+		}
+	} else if (x->rn_mklist != NULL) {
+		/*
+		 * Skip over masks whose index is > that of new node
+		 */
+		for (mp = &x->rn_mklist; (m = *mp) != NULL; mp = &m->rm_mklist)
+			if (m->rm_b >= b_leaf)
+				break;
+		t->rn_mklist = m;
+		*mp = NULL;
+	}
+on2:
+	/* Add new route to highest possible ancestor's list */
+	if (netmask == NULL || b > t->rn_b)
+		return tt; /* can't lift at all */
+	b_leaf = tt->rn_b;
+	do {
+		x = t;
+		t = t->rn_p;
+	} while (b <= t->rn_b && x != top);
+	/*
+	 * Search through routes associated with node to
+	 * insert new route according to index.
+	 * Need same criteria as when sorting dupedkeys to avoid
+	 * double loop on deletion.
+	 */
+	for (mp = &x->rn_mklist; (m = *mp) != NULL; mp = &m->rm_mklist) {
+		if (m->rm_b < b_leaf)
+			continue;
+		if (m->rm_b > b_leaf)
+			break;
+		if (m->rm_flags & RNF_NORMAL) {
+			mmask = m->rm_leaf->rn_mask;
+			if (tt->rn_flags & RNF_NORMAL) {
+				log(LOG_ERR, "Non-unique normal route,"
+				    " mask not entered\n");
+				return tt;
+			}
+		} else
+			mmask = m->rm_mask;
+		if (mmask == netmask) {
+			m->rm_refs++;
+			tt->rn_mklist = m;
+			return tt;
+		}
+		if (rn_refines(netmask, mmask) || rn_lexobetter(netmask, mmask))
+			break;
+	}
+	*mp = rn_new_radix_mask(tt, *mp);
+	return tt;
+}
+
+struct radix_node *
+rn_delete1(
+	const void *v_arg,
+	const void *netmask_arg,
+	struct radix_node_head *head,
+	struct radix_node *rn)
+{
+	struct radix_node *t, *p, *x, *tt;
+	struct radix_mask *m, *saved_m, **mp;
+	struct radix_node *dupedkey, *saved_tt, *top;
+	const char *v, *netmask;
+	int b, head_off, vlen;
+
+	v = v_arg;
+	netmask = netmask_arg;
+	x = head->rnh_treetop;
+	tt = rn_search(v, x);
+	head_off = x->rn_off;
+	vlen =  *(const u8 *)v;
+	saved_tt = tt;
+	top = x;
+	if (tt == NULL ||
+	    memcmp(v + head_off, tt->rn_key + head_off, vlen - head_off) != 0)
+		return NULL;
+	/*
+	 * Delete our route from mask lists.
+	 */
+	if (netmask != NULL) {
+		if ((x = rn_addmask(netmask, 1, head_off)) == NULL)
+			return NULL;
+		netmask = x->rn_key;
+		while (tt->rn_mask != netmask)
+			if ((tt = tt->rn_dupedkey) == NULL)
+				return NULL;
+	}
+	if (tt->rn_mask == NULL || (saved_m = m = tt->rn_mklist) == NULL)
+		goto on1;
+	if (tt->rn_flags & RNF_NORMAL) {
+		if (m->rm_leaf != tt || m->rm_refs > 0) {
+			log(LOG_ERR, "rn_delete: inconsistent annotation\n");
+			return NULL;  /* dangling ref could cause disaster */
+		}
+	} else {
+		if (m->rm_mask != tt->rn_mask) {
+			log(LOG_ERR, "rn_delete: inconsistent annotation\n");
+			goto on1;
+		}
+		if (--m->rm_refs >= 0)
+			goto on1;
+	}
+	b = -1 - tt->rn_b;
+	t = saved_tt->rn_p;
+	if (b > t->rn_b)
+		goto on1; /* Wasn't lifted at all */
+	do {
+		x = t;
+		t = t->rn_p;
+	} while (b <= t->rn_b && x != top);
+	for (mp = &x->rn_mklist; (m = *mp) != NULL; mp = &m->rm_mklist) {
+		if (m == saved_m) {
+			*mp = m->rm_mklist;
+			rm_free(m);
+			break;
+		}
+	}
+	if (m == NULL) {
+		log(LOG_ERR, "rn_delete: couldn't find our annotation\n");
+		if (tt->rn_flags & RNF_NORMAL)
+			return NULL; /* Dangling ref to us */
+	}
+on1:
+	/*
+	 * Eliminate us from tree
+	 */
+	if (tt->rn_flags & RNF_ROOT)
+		return NULL;
+#ifdef RN_DEBUG
+	if (rn_debug)
+		log(LOG_DEBUG, "%s: Going In:\n", __func__), traverse(head, tt);
+#endif
+	t = tt->rn_p;
+	dupedkey = saved_tt->rn_dupedkey;
+	if (dupedkey != NULL) {
+		/*
+		 * Here, tt is the deletion target, and
+		 * saved_tt is the head of the dupedkey chain.
+		 */
+		if (tt == saved_tt) {
+			x = dupedkey;
+			x->rn_p = t;
+			if (t->rn_l == tt)
+				t->rn_l = x;
+			else
+				t->rn_r = x;
+		} else {
+			/* find node in front of tt on the chain */
+			for (x = p = saved_tt;
+			     p != NULL && p->rn_dupedkey != tt;)
+				p = p->rn_dupedkey;
+			if (p != NULL) {
+				p->rn_dupedkey = tt->rn_dupedkey;
+				if (tt->rn_dupedkey != NULL)
+					tt->rn_dupedkey->rn_p = p;
+			} else
+				log(LOG_ERR, "rn_delete: couldn't find us\n");
+		}
+		t = tt + 1;
+		if  (t->rn_flags & RNF_ACTIVE) {
+			*++x = *t;
+			p = t->rn_p;
+			if (p->rn_l == t)
+				p->rn_l = x;
+			else
+				p->rn_r = x;
+			x->rn_l->rn_p = x;
+			x->rn_r->rn_p = x;
+		}
+		goto out;
+	}
+	if (t->rn_l == tt)
+		x = t->rn_r;
+	else
+		x = t->rn_l;
+	p = t->rn_p;
+	if (p->rn_r == t)
+		p->rn_r = x;
+	else
+		p->rn_l = x;
+	x->rn_p = p;
+	/*
+	 * Demote routes attached to us.
+	 */
+	if (t->rn_mklist == NULL)
+		;
+	else if (x->rn_b >= 0) {
+		for (mp = &x->rn_mklist; (m = *mp) != NULL; mp = &m->rm_mklist)
+			;
+		*mp = t->rn_mklist;
+	} else {
+		/* If there are any key,mask pairs in a sibling
+		   duped-key chain, some subset will appear sorted
+		   in the same order attached to our mklist */
+		for (m = t->rn_mklist;
+		     m != NULL && x != NULL;
+		     x = x->rn_dupedkey) {
+			if (m == x->rn_mklist) {
+				struct radix_mask *mm = m->rm_mklist;
+				x->rn_mklist = NULL;
+				if (--(m->rm_refs) < 0)
+					rm_free(m);
+				m = mm;
+			}
+		}
+		if (m != NULL) {
+			log(LOG_ERR, "rn_delete: Orphaned Mask %p at %p\n",
+			    m, x);
+		}
+	}
+	/*
+	 * We may be holding an active internal node in the tree.
+	 */
+	x = tt + 1;
+	if (t != x) {
+		*t = *x;
+		t->rn_l->rn_p = t;
+		t->rn_r->rn_p = t;
+		p = x->rn_p;
+		if (p->rn_l == x)
+			p->rn_l = t;
+		else
+			p->rn_r = t;
+	}
+out:
+#ifdef RN_DEBUG
+	if (rn_debug) {
+		log(LOG_DEBUG, "%s: Coming Out:\n", __func__),
+		    traverse(head, tt);
+	}
+#endif /* RN_DEBUG */
+	tt->rn_flags &= ~RNF_ACTIVE;
+	tt[1].rn_flags &= ~RNF_ACTIVE;
+	return tt;
+}
+
+struct radix_node *
+rn_delete(
+	const void *v_arg,
+	const void *netmask_arg,
+	struct radix_node_head *head)
+{
+	return rn_delete1(v_arg, netmask_arg, head, NULL);
+}
+
+static struct radix_node *
+rn_walknext(struct radix_node *rn, rn_printer_t printer, void *arg)
+{
+	/* If at right child go back up, otherwise, go right */
+	while (rn->rn_p->rn_r == rn && (rn->rn_flags & RNF_ROOT) == 0) {
+		if (printer != NULL)
+			(*printer)(arg, SUBTREE_CLOSE);
+		rn = rn->rn_p;
+	}
+	if (printer)
+		rn_nodeprint(rn->rn_p, printer, arg, "");
+	/* Find the next *leaf* since next node might vanish, too */
+	for (rn = rn->rn_p->rn_r; rn->rn_b >= 0;) {
+		if (printer != NULL)
+			(*printer)(arg, SUBTREE_OPEN);
+		rn = rn->rn_l;
+	}
+	return rn;
+}
+
+static struct radix_node *
+rn_walkfirst(struct radix_node *rn, rn_printer_t printer, void *arg)
+{
+	/* First time through node, go left */
+	while (rn->rn_b >= 0) {
+		if (printer != NULL)
+			(*printer)(arg, SUBTREE_OPEN);
+		rn = rn->rn_l;
+	}
+	return rn;
+}
+
+int
+rn_walktree(
+	struct radix_node_head *h,
+	int (*f)(struct radix_node *, void *),
+	void *w)
+{
+	int error;
+	struct radix_node *base, *next, *rn;
+	/*
+	 * This gets complicated because we may delete the node
+	 * while applying the function f to it, so we need to calculate
+	 * the successor node in advance.
+	 */
+	rn = rn_walkfirst(h->rnh_treetop, NULL, NULL);
+	for (;;) {
+		base = rn;
+		next = rn_walknext(rn, NULL, NULL);
+		/* Process leaves */
+		while ((rn = base) != NULL) {
+			base = rn->rn_dupedkey;
+			if (!(rn->rn_flags & RNF_ROOT) && (error = (*f)(rn, w)))
+				return error;
+		}
+		rn = next;
+		if (rn->rn_flags & RNF_ROOT)
+			return 0;
+	}
+	/* NOTREACHED */
+}
+
+struct radix_node *
+rn_search_matched(struct radix_node_head *h,
+    int (*matcher)(struct radix_node *, void *), void *w)
+{
+	bool matched;
+	struct radix_node *base, *next, *rn;
+	/*
+	 * This gets complicated because we may delete the node
+	 * while applying the function f to it, so we need to calculate
+	 * the successor node in advance.
+	 */
+	rn = rn_walkfirst(h->rnh_treetop, NULL, NULL);
+	for (;;) {
+		base = rn;
+		next = rn_walknext(rn, NULL, NULL);
+		/* Process leaves */
+		while ((rn = base) != NULL) {
+			base = rn->rn_dupedkey;
+			if (!(rn->rn_flags & RNF_ROOT)) {
+				matched = (*matcher)(rn, w);
+				if (matched)
+					return rn;
+			}
+		}
+		rn = next;
+		if (rn->rn_flags & RNF_ROOT)
+			return NULL;
+	}
+	/* NOTREACHED */
+}
+
+int
+rn_inithead(void **head, int off)
+{
+	struct radix_node_head *rnh;
+
+	if (*head != NULL)
+		return 1;
+	R_Malloc(rnh, struct radix_node_head *, sizeof (*rnh));
+	if (rnh == NULL)
+		return 0;
+	*head = rnh;
+	return rn_inithead0(rnh, off);
+}
+
+int
+rn_inithead0(struct radix_node_head *rnh, int off)
+{
+	struct radix_node *t;
+	struct radix_node *tt;
+	struct radix_node *ttt;
+
+	memset(rnh, 0, sizeof(*rnh));
+	t = rn_newpair(rn_zeros, off, rnh->rnh_nodes);
+	ttt = rnh->rnh_nodes + 2;
+	t->rn_r = ttt;
+	t->rn_p = t;
+	tt = t->rn_l;
+	tt->rn_flags = t->rn_flags = RNF_ROOT | RNF_ACTIVE;
+	tt->rn_b = -1 - off;
+	*ttt = *tt;
+	ttt->rn_key = rn_ones;
+	rnh->rnh_addaddr = rn_addroute;
+	rnh->rnh_deladdr = rn_delete;
+	rnh->rnh_matchaddr = rn_match;
+	rnh->rnh_lookup = rn_lookup;
+	rnh->rnh_treetop = t;
+	return 1;
+}
+
+static clib_error_t *
+rn_module_init (vlib_main_t * vm)
+{
+	char *cp, *cplim;
+
+	R_Malloc(rn_zeros, char *, 3 * max_keylen);
+	if (rn_zeros == NULL)
+            return (clib_error_return (0, "RN Zeros..."));
+
+	memset(rn_zeros, 0, 3 * max_keylen);
+	rn_ones = cp = rn_zeros + max_keylen;
+	addmask_key = cplim = rn_ones + max_keylen;
+	while (cp < cplim)
+		*cp++ = -1;
+	if (rn_inithead((void *)&mask_rnhead, 0) == 0)
+            return (clib_error_return (0, "RN Init 2"));
+
+        return (NULL);
+}
+
+VLIB_INIT_FUNCTION(rn_module_init);
diff --git a/src/vnet/util/radix.h b/src/vnet/util/radix.h
new file mode 100644
index 00000000..d9ba6659
--- /dev/null
+++ b/src/vnet/util/radix.h
@@ -0,0 +1,147 @@
+/*	$NetBSD: radix.h,v 1.23 2016/11/15 01:50:06 ozaki-r Exp $	*/
+
+/*
+ * Copyright (c) 1988, 1989, 1993
+ *	The Regents of the University of California.  All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ * 3. Neither the name of the University nor the names of its contributors
+ *    may be used to endorse or promote products derived from this software
+ *    without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ *	@(#)radix.h	8.2 (Berkeley) 10/31/94
+ */
+
+#ifndef _NET_RADIX_H_
+#define	_NET_RADIX_H_
+
+#include <vlib/vlib.h>
+
+/*
+ * Radix search tree node layout.
+ */
+
+struct radix_node {
+	struct	radix_mask *rn_mklist;	/* list of masks contained in subtree */
+	struct	radix_node *rn_p;	/* parent */
+	i16	rn_b;			/* bit offset; -1-index(netmask) */
+	u8	rn_bmask;		/* node: mask for bit test*/
+	u8	rn_flags;		/* enumerated next */
+#define RNF_NORMAL	1		/* leaf contains normal route */
+#define RNF_ROOT	2		/* leaf is root leaf for tree */
+#define RNF_ACTIVE	4		/* This node is alive (for rtfree) */
+	union {
+		struct {			/* leaf only data: */
+			const char *rn_Key;	/* object of search */
+			const char *rn_Mask;	/* netmask, if present */
+			struct	radix_node *rn_Dupedkey;
+		} rn_leaf;
+		struct {			/* node only data: */
+			int	rn_Off;		/* where to start compare */
+			struct	radix_node *rn_L;/* progeny */
+			struct	radix_node *rn_R;/* progeny */
+		} rn_node;
+	} rn_u;
+#ifdef RN_DEBUG
+	i32 rn_info;
+	struct radix_node *rn_twin;
+	struct radix_node *rn_ybro;
+#endif
+};
+
+#define rn_dupedkey rn_u.rn_leaf.rn_Dupedkey
+#define rn_key rn_u.rn_leaf.rn_Key
+#define rn_mask rn_u.rn_leaf.rn_Mask
+#define rn_off rn_u.rn_node.rn_Off
+#define rn_l rn_u.rn_node.rn_L
+#define rn_r rn_u.rn_node.rn_R
+
+/*
+ * Annotations to tree concerning potential routes applying to subtrees.
+ */
+
+struct radix_mask {
+	i16	rm_b;			/* bit offset; -1-index(netmask) */
+	i8	rm_unused;		/* cf. rn_bmask */
+	u8	rm_flags;		/* cf. rn_flags */
+	struct	radix_mask *rm_mklist;	/* more masks to try */
+	union	{
+		const char *rmu_mask;		/* the mask */
+		struct	radix_node *rmu_leaf;	/* for normal routes */
+	}	rm_rmu;
+	i32	rm_refs;		/* # of references to this struct */
+};
+
+#define rm_mask rm_rmu.rmu_mask
+#define rm_leaf rm_rmu.rmu_leaf		/* extra field would make 32 bytes */
+
+struct radix_node_head {
+	struct	radix_node *rnh_treetop;
+	i32	rnh_addrsize;		/* permit, but not require fixed keys */
+	i32	rnh_pktsize;		/* permit, but not require fixed keys */
+	struct	radix_node *(*rnh_addaddr)	/* add based on sockaddr */
+		(const void *v, const void *mask,
+		     struct radix_node_head *head, struct radix_node nodes[]);
+	struct	radix_node *(*rnh_addpkt)	/* add based on packet hdr */
+		(const void *v, const void *mask,
+		     struct radix_node_head *head, struct radix_node nodes[]);
+	struct	radix_node *(*rnh_deladdr)	/* remove based on sockaddr */
+		(const void *v, const void *mask, struct radix_node_head *head);
+	struct	radix_node *(*rnh_delpkt)	/* remove based on packet hdr */
+		(const void *v, const void *mask, struct radix_node_head *head);
+	struct	radix_node *(*rnh_matchaddr)	/* locate based on sockaddr */
+		(const void *v, struct radix_node_head *head);
+	struct	radix_node *(*rnh_lookup)	/* locate based on sockaddr */
+		(const void *v, const void *mask, struct radix_node_head *head);
+	struct	radix_node *(*rnh_matchpkt)	/* locate based on packet hdr */
+		(const void *v, struct radix_node_head *head);
+	struct	radix_node rnh_nodes[3];	/* empty tree for common case */
+};
+
+void	rn_init(void);
+int	rn_inithead(void **, int);
+void	rn_delayedinit(void **, int);
+int	rn_inithead0(struct radix_node_head *, int);
+int	rn_refines(const void *, const void *);
+int	rn_walktree(struct radix_node_head *,
+	            int (*)(struct radix_node *, void *),
+		    void *);
+struct radix_node *
+	rn_search_matched(struct radix_node_head *,
+	                  int (*)(struct radix_node *, void *),
+		          void *);
+struct radix_node
+	 *rn_addmask(const void *, int, int),
+	 *rn_addroute(const void *, const void *, struct radix_node_head *,
+			struct radix_node [2]),
+	 *rn_delete1(const void *, const void *, struct radix_node_head *,
+			struct radix_node *),
+	 *rn_delete(const void *, const void *, struct radix_node_head *),
+	 *rn_insert(const void *, struct radix_node_head *, int *,
+			struct radix_node [2]),
+	 *rn_lookup(const void *, const void *, struct radix_node_head *),
+	 *rn_match(const void *, struct radix_node_head *),
+	 *rn_newpair(const void *, int, struct radix_node[2]),
+	 *rn_search(const void *, struct radix_node *),
+	 *rn_search_m(const void *, struct radix_node *, const void *);
+
+#endif /* !_NET_RADIX_H_ */
diff --git a/src/vnet/vxlan/vxlan.c b/src/vnet/vxlan/vxlan.c
index f749429f..9ed42875 100644
--- a/src/vnet/vxlan/vxlan.c
+++ b/src/vnet/vxlan/vxlan.c
@@ -16,7 +16,8 @@
 #include <vnet/ip/format.h>
 #include <vnet/fib/fib_entry.h>
 #include <vnet/fib/fib_table.h>
-#include <vnet/dpo/receive_dpo.h>
+#include <vnet/mfib/mfib_table.h>
+#include <vnet/adj/adj_mcast.h>
 #include <vlib/vlib.h>
 
 /**
@@ -337,7 +338,7 @@ vtep_addr_unref(ip46_address_t *ip)
 
 typedef CLIB_PACKED(union {
   struct {
-    fib_node_index_t fib_entry_index;
+    fib_node_index_t mfib_entry_index;
     adj_index_t mcast_adj_index;
   };
   u64 as_u64;
@@ -353,11 +354,28 @@ mcast_shared_get(ip46_address_t * ip)
 }
 
 static inline void
-ip46_multicast_ethernet_address(u8 * ethernet_address, ip46_address_t * ip) {
-          if (ip46_address_is_ip4(ip))
-              ip4_multicast_ethernet_address(ethernet_address, &ip->ip4);
-          else
-              ip6_multicast_ethernet_address(ethernet_address, ip->ip6.as_u32[0]);
+mcast_shared_add(ip46_address_t *dst,
+                 fib_node_index_t mfei,
+                 adj_index_t ai)
+{
+    mcast_shared_t new_ep = {
+        .mcast_adj_index = ai,
+        .mfib_entry_index = mfei,
+    };
+
+    hash_set_key_copy (&vxlan_main.mcast_shared, dst, new_ep.as_u64);
+}
+
+static inline void
+mcast_shared_remove(ip46_address_t *dst)
+{
+    mcast_shared_t ep = mcast_shared_get(dst);
+
+    adj_unlock(ep.mcast_adj_index);
+    mfib_table_entry_delete_index(ep.mfib_entry_index,
+                                  MFIB_SOURCE_VXLAN);
+
+    hash_unset_key_free (&vxlan_main.mcast_shared, dst);
 }
 
 int vnet_vxlan_add_del_tunnel 
@@ -503,28 +521,65 @@ int vnet_vxlan_add_del_tunnel
 	   */
           fib_protocol_t fp = (is_ip6) ? FIB_PROTOCOL_IP6 : FIB_PROTOCOL_IP4;
           dpo_id_t dpo = DPO_INVALID;
-	  dpo_proto_t dproto = fib_proto_to_dpo(fp);
+          mcast_shared_t ep;
 
 	  if (vtep_addr_ref(&t->dst) == 1)
-	    {
-               u8 mcast_mac[6];
-
-               ip46_multicast_ethernet_address(mcast_mac, &t->dst);
-               receive_dpo_add_or_lock(dproto, ~0, NULL, &dpo);
-	       mcast_shared_t new_ep = {
-		 .mcast_adj_index = adj_rewrite_add_and_lock
-                   (fp, fib_proto_to_link(fp), a->mcast_sw_if_index, mcast_mac),
-                 /* Add VRF local mcast adj. */
-                 .fib_entry_index = fib_table_entry_special_dpo_add
-                   (t->encap_fib_index, &tun_dst_pfx,
-                    FIB_SOURCE_SPECIAL, FIB_ENTRY_FLAG_NONE, &dpo)
-		   };
-	       hash_set_key_copy (&vxm->mcast_shared, &t->dst, new_ep.as_u64);
-	       dpo_reset(&dpo);
-	    }
+          {
+              fib_node_index_t mfei;
+              adj_index_t ai;
+              fib_route_path_t path = {
+                  .frp_proto = fp,
+                  .frp_addr = zero_addr,
+                  .frp_sw_if_index = 0xffffffff,
+                  .frp_fib_index = ~0,
+                  .frp_weight = 0,
+                  .frp_flags = FIB_ROUTE_PATH_LOCAL,
+              };
+              const mfib_prefix_t mpfx = {
+                  .fp_proto = fp,
+                  .fp_len = (is_ip6 ? 128 : 32),
+                  .fp_grp_addr = tun_dst_pfx.fp_addr,
+              };
+
+              /*
+               * Setup the (*,G) to receive traffic on the mcast group
+               *  - the forwarding interface is for-us
+               *  - the accepting interface is that from the API
+               */
+              mfib_table_entry_path_update(t->encap_fib_index,
+                                           &mpfx,
+                                           MFIB_SOURCE_VXLAN,
+                                           &path,
+                                           MFIB_ITF_FLAG_FORWARD);
+
+              path.frp_sw_if_index = a->mcast_sw_if_index;
+              path.frp_flags = FIB_ROUTE_PATH_FLAG_NONE;
+              mfei = mfib_table_entry_path_update(t->encap_fib_index,
+                                                  &mpfx,
+                                                  MFIB_SOURCE_VXLAN,
+                                                  &path,
+                                                  MFIB_ITF_FLAG_ACCEPT);
+
+              /*
+               * Create the mcast adjacency to send traffic to the group
+               */
+              ai = adj_mcast_add_or_lock(fp,
+                                         fib_proto_to_link(fp),
+                                         a->mcast_sw_if_index);
+
+              /*
+               * create a new end-point
+               */
+              mcast_shared_add(&t->dst, mfei, ai);
+          }
+
+          ep = mcast_shared_get(&t->dst);
+
           /* Stack shared mcast dst mac addr rewrite on encap */
-	  mcast_shared_t ep = mcast_shared_get(&t->dst);
-          dpo_set (&dpo, DPO_ADJACENCY, dproto, ep.mcast_adj_index);
+          dpo_set (&dpo, DPO_ADJACENCY,
+                   fib_proto_to_dpo(fp),
+                   ep.mcast_adj_index);
+
           dpo_stack_from_node (encap_index, &t->next_dpo, &dpo);
           dpo_reset (&dpo);
 	  flood_class = VNET_FLOOD_CLASS_TUNNEL_MASTER;
@@ -563,10 +618,7 @@ int vnet_vxlan_add_del_tunnel
         }
       else if (vtep_addr_unref(&t->dst) == 0)
         {
-	  mcast_shared_t ep = mcast_shared_get(&t->dst);
-	  adj_unlock(ep.mcast_adj_index);
-	  fib_table_entry_delete_index(ep.fib_entry_index, FIB_SOURCE_SPECIAL);
-          hash_unset_key_free (&vxm->mcast_shared, &t->dst);
+	  mcast_shared_remove(&t->dst);
         }
 
       fib_node_deinit(&t->node);
diff --git a/src/vpp/api/api.c b/src/vpp/api/api.c
index e6227a68..6c8800ec 100644
--- a/src/vpp/api/api.c
+++ b/src/vpp/api/api.c
@@ -964,7 +964,6 @@ ip4_reset_fib_t_handler (vl_api_reset_fib_t * mp)
 
     for (i = 0; i < vec_len (sw_if_indices_to_shut); i++) {
       sw_if_index = sw_if_indices_to_shut[i];
-      // vec_foreach (sw_if_index, sw_if_indices_to_shut) {
 
       u32 flags = vnet_sw_interface_get_flags (vnm, sw_if_index);
       flags &= ~(VNET_SW_INTERFACE_FLAG_ADMIN_UP);
@@ -972,7 +971,6 @@ ip4_reset_fib_t_handler (vl_api_reset_fib_t * mp)
     }
 
     fib_table_flush(fib->index, FIB_PROTOCOL_IP4, FIB_SOURCE_API);
-    fib_table_flush(fib->index, FIB_PROTOCOL_IP4, FIB_SOURCE_INTERFACE);
 
     rv = 0;
     break;
@@ -1013,15 +1011,14 @@ ip6_reset_fib_t_handler (vl_api_reset_fib_t * mp)
 
     /* Shut down interfaces in this FIB / clean out intfc routes */
     pool_foreach (si, im->sw_interfaces,
-                  ({
-                    if (im6->fib_index_by_sw_if_index[si->sw_if_index] ==
-                        fib->index)
-                      vec_add1 (sw_if_indices_to_shut, si->sw_if_index);
-                  }));
+    ({
+      if (im6->fib_index_by_sw_if_index[si->sw_if_index] ==
+          fib->index)
+        vec_add1 (sw_if_indices_to_shut, si->sw_if_index);
+    }));
 
     for (i = 0; i < vec_len (sw_if_indices_to_shut); i++) {
       sw_if_index = sw_if_indices_to_shut[i];
-      // vec_foreach (sw_if_index, sw_if_indices_to_shut) {
 
       u32 flags = vnet_sw_interface_get_flags (vnm, sw_if_index);
       flags &= ~(VNET_SW_INTERFACE_FLAG_ADMIN_UP);
@@ -1029,7 +1026,6 @@ ip6_reset_fib_t_handler (vl_api_reset_fib_t * mp)
     }
 
     fib_table_flush(fib->index, FIB_PROTOCOL_IP6, FIB_SOURCE_API);
-    fib_table_flush(fib->index, FIB_PROTOCOL_IP6, FIB_SOURCE_INTERFACE);
 
     rv = 0;
     break;
diff --git a/src/vppinfra.am b/src/vppinfra.am
index 19485d2d..724f21c9 100644
--- a/src/vppinfra.am
+++ b/src/vppinfra.am
@@ -38,7 +38,7 @@ TESTS  +=  test_bihash_template \
 	   test_time \
 	   test_timing_wheel \
 	   test_vec \
-	   test_zvec 
+	   test_zvec
 endif
 
 noinst_PROGRAMS = $(TESTS)
diff --git a/src/vppinfra/dlist.h b/src/vppinfra/dlist.h
index 7d09b2bb..e445b39f 100644
--- a/src/vppinfra/dlist.h
+++ b/src/vppinfra/dlist.h
@@ -121,7 +121,7 @@ clib_dlist_remove_head (dlist_elt_t * pool, u32 head_index)
 
   ASSERT (head->value == ~0);
 
-  if (head->next == ~0)
+  if (head->next == ~0 || (head->next == head_index))
     return ~0;
 
   rv = head->next;
diff --git a/src/vppinfra/format.c b/src/vppinfra/format.c
index 78e52e9a..999b093c 100644
--- a/src/vppinfra/format.c
+++ b/src/vppinfra/format.c
@@ -150,13 +150,13 @@ justify (u8 * s, format_info_t * fi, uword s_len_orig)
   return s;
 }
 
-static u8 *
-do_percent (u8 ** _s, u8 * fmt, va_list * va)
+static const u8 *
+do_percent (u8 ** _s, const u8 * fmt, va_list * va)
 {
   u8 *s = *_s;
   uword c;
 
-  u8 *f = fmt;
+  const u8 *f = fmt;
 
   format_info_t fi = {
     .justify = '+',
@@ -385,7 +385,7 @@ done:
 u8 *
 va_format (u8 * s, const char *fmt, va_list * va)
 {
-  u8 *f = (u8 *) fmt, *g;
+  const u8 *f = (u8 *) fmt, *g;
   u8 c;
 
   g = f;
diff --git a/src/vppinfra/format.h b/src/vppinfra/format.h
index bc0d6d15..bec1b6b4 100644
--- a/src/vppinfra/format.h
+++ b/src/vppinfra/format.h
@@ -234,7 +234,7 @@ typedef uword (unformat_function_t) (unformat_input_t * input,
 /* External functions. */
 
 /* General unformatting function with programmable input stream. */
-uword unformat (unformat_input_t * i, char *fmt, ...);
+uword unformat (unformat_input_t * i, const char *fmt, ...);
 
 /* Call user defined parse function.
    unformat_user (i, f, ...) is equivalent to unformat (i, "%U", f, ...) */
@@ -242,7 +242,7 @@ uword unformat_user (unformat_input_t * input, unformat_function_t * func,
 		     ...);
 
 /* Alternate version which allows for extensions. */
-uword va_unformat (unformat_input_t * i, char *fmt, va_list * args);
+uword va_unformat (unformat_input_t * i, const char *fmt, va_list * args);
 
 /* Setup for unformat of Unix style command line. */
 void unformat_init_command_line (unformat_input_t * input, char *argv[]);
diff --git a/src/vppinfra/unformat.c b/src/vppinfra/unformat.c
index ac8b7ddc..7c636ccc 100644
--- a/src/vppinfra/unformat.c
+++ b/src/vppinfra/unformat.c
@@ -681,8 +681,8 @@ error:
   return 0;
 }
 
-static char *
-match_input_with_format (unformat_input_t * input, char *f)
+static const char *
+match_input_with_format (unformat_input_t * input, const char *f)
 {
   uword cf, ci;
 
@@ -703,8 +703,8 @@ match_input_with_format (unformat_input_t * input, char *f)
   return f;
 }
 
-static char *
-do_percent (unformat_input_t * input, va_list * va, char *f)
+static const char *
+do_percent (unformat_input_t * input, va_list * va, const char *f)
 {
   uword cf, n, data_bytes = ~0;
 
@@ -824,9 +824,9 @@ unformat_skip_white_space (unformat_input_t * input)
 }
 
 uword
-va_unformat (unformat_input_t * input, char *fmt, va_list * va)
+va_unformat (unformat_input_t * input, const char *fmt, va_list * va)
 {
-  char *f;
+  const char *f;
   uword input_matches_format;
   uword default_skip_input_white_space;
   uword n_input_white_space_skipped;
@@ -937,7 +937,7 @@ va_unformat (unformat_input_t * input, char *fmt, va_list * va)
 
       else
 	{
-	  char *g = match_input_with_format (input, f);
+	  const char *g = match_input_with_format (input, f);
 	  if (!g)
 	    goto parse_fail;
 	  last_non_white_space_match_format = g > f;
@@ -963,7 +963,7 @@ parse_fail:
 }
 
 uword
-unformat (unformat_input_t * input, char *fmt, ...)
+unformat (unformat_input_t * input, const char *fmt, ...)
 {
   va_list va;
   uword result;
diff --git a/test/test_dhcp.py b/test/test_dhcp.py
index bdff679c..04ab2e11 100644
--- a/test/test_dhcp.py
+++ b/test/test_dhcp.py
@@ -4,8 +4,6 @@ import unittest
 import socket
 
 from framework import VppTestCase, VppTestRunner
-from vpp_ip_route import IpRoute, RoutePath
-from vpp_lo_interface import VppLoInterface
 
 from scapy.layers.l2 import Ether, getmacbyip
 from scapy.layers.inet import IP, UDP, ICMP
@@ -482,17 +480,6 @@ class TestDHCP(VppTestCase):
         server_addr_vrf1 = self.pg1.remote_ip6n
         src_addr_vrf1 = self.pg1.local_ip6n
 
-        #
-        # Add the Route to receive the DHCP packets
-        #
-        route_dhcp_vrf0 = IpRoute(self, dhcp_solicit_dst, 128,
-                                  [], is_local=1, is_ip6=1)
-        route_dhcp_vrf0.add_vpp_config()
-        route_dhcp_vrf1 = IpRoute(self, dhcp_solicit_dst, 128,
-                                  [], is_local=1, is_ip6=1,
-                                  table_id=1)
-        route_dhcp_vrf1.add_vpp_config()
-
         dmac = in6_getnsmac(inet_pton(socket.AF_INET6, dhcp_solicit_dst))
         p_solicit_vrf0 = (Ether(dst=dmac, src=self.pg2.remote_mac) /
                           IPv6(src=dhcp_solicit_src_vrf0,
@@ -732,8 +719,5 @@ class TestDHCP(VppTestCase):
                                     is_ipv6=1,
                                     is_add=0)
 
-        route_dhcp_vrf0.remove_vpp_config()
-        route_dhcp_vrf1.remove_vpp_config()
-
 if __name__ == '__main__':
     unittest.main(testRunner=VppTestRunner)
diff --git a/test/test_ip6.py b/test/test_ip6.py
index ea669b70..e188970a 100644
--- a/test/test_ip6.py
+++ b/test/test_ip6.py
@@ -5,6 +5,7 @@ import socket
 
 from framework import VppTestCase, VppTestRunner
 from vpp_sub_interface import VppSubInterface, VppDot1QSubint
+from vpp_pg_interface import is_ipv6_misc
 
 from scapy.packet import Raw
 from scapy.layers.l2 import Ether, Dot1Q
@@ -12,10 +13,9 @@ from scapy.layers.inet6 import IPv6, UDP, ICMPv6ND_NS, ICMPv6ND_RS, \
     ICMPv6ND_RA, ICMPv6NDOptSrcLLAddr, getmacbyip6, ICMPv6MRD_Solicitation
 from util import ppp
 from scapy.utils6 import in6_getnsma, in6_getnsmac, in6_ptop, in6_islladdr, \
-    in6_mactoifaceid
+    in6_mactoifaceid, in6_ismaddr
 from scapy.utils import inet_pton, inet_ntop
 
-
 def mk_ll_addr(mac):
     euid = in6_mactoifaceid(mac)
     addr = "fe80::" + euid
@@ -287,28 +287,39 @@ class TestIPv6(VppTestCase):
         self.send_and_assert_no_replies(self.pg0, pkts,
                                         "No response to NS for unknown target")
 
-    def send_and_expect_ra(self, intf, pkts, remark, src_ip=None):
-        if not src_ip:
-            src_ip = intf.remote_ip6
-        intf.add_stream(pkts)
-        self.pg0.add_stream(pkts)
-        self.pg_enable_capture(self.pg_interfaces)
-        self.pg_start()
-        rx = intf.get_capture(1)
+    def validate_ra(self, intf, rx, dst_ip=None):
+        if not dst_ip:
+            dst_ip = intf.remote_ip6
 
-        self.assertEqual(len(rx), 1)
-        rx = rx[0]
+        # unicasted packets must come to the unicast mac 
+        self.assertEqual(rx[Ether].dst, intf.remote_mac)
+
+        # and from the router's MAC
+        self.assertEqual(rx[Ether].src, intf.local_mac)
 
         # the rx'd RA should be addressed to the sender's source
         self.assertTrue(rx.haslayer(ICMPv6ND_RA))
         self.assertEqual(in6_ptop(rx[IPv6].dst),
-                         in6_ptop(src_ip))
+                         in6_ptop(dst_ip))
 
         # and come from the router's link local
         self.assertTrue(in6_islladdr(rx[IPv6].src))
         self.assertEqual(in6_ptop(rx[IPv6].src),
                          in6_ptop(mk_ll_addr(intf.local_mac)))
 
+
+    def send_and_expect_ra(self, intf, pkts, remark, dst_ip=None,
+                           filter_out_fn=is_ipv6_misc):
+        intf.add_stream(pkts)
+        self.pg0.add_stream(pkts)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+        rx = intf.get_capture(1, filter_out_fn=filter_out_fn)
+
+        self.assertEqual(len(rx), 1)
+        rx = rx[0]
+        self.validate_ra(intf, rx, dst_ip)
+
     def test_rs(self):
         """ IPv6 Router Solicitation Exceptions
 
@@ -319,6 +330,9 @@ class TestIPv6(VppTestCase):
         # Before we begin change the IPv6 RA responses to use the unicast
         # address - that way we will not confuse them with the periodic
         # RAs which go to the mcast address
+        # Sit and wait for the first periodic RA.
+        #
+        # TODO
         #
         self.pg0.ip6_ra_config(send_unicast=1)
 
@@ -365,8 +379,23 @@ class TestIPv6(VppTestCase):
              IPv6(dst=self.pg0.local_ip6, src=ll) /
              ICMPv6ND_RS())
         pkts = [p]
-        self.send_and_expect_ra(
-            self.pg0, pkts, "RS sourced from link-local", src_ip=ll)
+        self.send_and_expect_ra(self.pg0, pkts,
+                                "RS sourced from link-local",
+                                dst_ip=ll)
+
+        #
+        # Send the RS multicast
+        #
+        self.pg0.ip6_ra_config(send_unicast=1)
+        dmac = in6_getnsmac(inet_pton(socket.AF_INET6, "ff02::2"))
+        ll = mk_ll_addr(self.pg0.remote_mac)
+        p = (Ether(dst=dmac, src=self.pg0.remote_mac) /
+             IPv6(dst="ff02::2", src=ll) /
+             ICMPv6ND_RS())
+        pkts = [p]
+        self.send_and_expect_ra(self.pg0, pkts,
+                                "RS sourced from link-local",
+                                dst_ip=ll)
 
         #
         # Source from the unspecified address ::. This happens when the RS
@@ -376,74 +405,20 @@ class TestIPv6(VppTestCase):
         # If we happen to pick up the periodic RA at this point then so be it,
         # it's not an error.
         #
-        self.pg0.ip6_ra_config(send_unicast=1)
-        p = (Ether(dst=self.pg0.local_mac, src=self.pg0.remote_mac) /
-             IPv6(dst=self.pg0.local_ip6, src="::") /
+        self.pg0.ip6_ra_config(send_unicast=1, suppress=1)
+        p = (Ether(dst=dmac, src=self.pg0.remote_mac) /
+             IPv6(dst="ff02::2", src="::") /
              ICMPv6ND_RS())
         pkts = [p]
-
-        self.pg0.add_stream(pkts)
-        self.pg0.add_stream(pkts)
-        self.pg_enable_capture(self.pg_interfaces)
-        self.pg_start()
-        capture = self.pg0.get_capture(1, filter_out_fn=None)
-        found = 0
-        for rx in capture:
-            if (rx.haslayer(ICMPv6ND_RA)):
-                # and come from the router's link local
-                self.assertTrue(in6_islladdr(rx[IPv6].src))
-                self.assertEqual(in6_ptop(rx[IPv6].src),
-                                 in6_ptop(mk_ll_addr(self.pg0.local_mac)))
-                # sent to the all hosts mcast
-                self.assertEqual(in6_ptop(rx[IPv6].dst), "ff02::1")
-
-                found = 1
-        self.assertTrue(found)
-
-    @unittest.skip("Unsupported")
-    def test_mrs(self):
-        """ IPv6 Multicast Router Solicitation Exceptions
-
-        Test scenario:
-        """
-
-        #
-        # An RS from a link source address
-        #  - expect an RA in return
-        #
-        nsma = in6_getnsma(inet_pton(socket.AF_INET6, self.pg0.local_ip6))
-        d = inet_ntop(socket.AF_INET6, nsma)
-
-        p = (Ether(dst=getmacbyip6("ff02::2")) /
-             IPv6(dst=d, src=self.pg0.remote_ip6) /
-             ICMPv6MRD_Solicitation())
-        pkts = [p]
-
-        self.pg0.add_stream(pkts)
-        self.pg_enable_capture(self.pg_interfaces)
-        self.pg_start()
-        self.pg0.assert_nothing_captured(
-            remark="No response to NS source by address not on sub-net")
+        self.send_and_expect_ra(self.pg0, pkts,
+                                "RS sourced from unspecified",
+                                dst_ip="ff02::1",
+                                filter_out_fn=None)
 
         #
-        # An RS from a non link source address
+        # Reset the periodic advertisements back to default values 
         #
-        nsma = in6_getnsma(inet_pton(socket.AF_INET6, self.pg0.local_ip6))
-        d = inet_ntop(socket.AF_INET6, nsma)
-
-        p = (Ether(dst=getmacbyip6("ff02::2")) /
-             IPv6(dst=d, src="2002::2") /
-             ICMPv6MRD_Solicitation())
-        pkts = [p]
-
-        self.send_and_assert_no_replies(self.pg0, pkts,
-                                        "RA rate limited")
-        self.pg0.add_stream(pkts)
-        self.pg_enable_capture(self.pg_interfaces)
-        self.pg_start()
-        self.pg0.assert_nothing_captured(
-            remark="No response to NS source by address not on sub-net")
-
+        self.pg0.ip6_ra_config(no=1, suppress=1, send_unicast=0)
 
 if __name__ == '__main__':
     unittest.main(testRunner=VppTestRunner)
diff --git a/test/test_ip_mcast.py b/test/test_ip_mcast.py
new file mode 100644
index 00000000..028853d2
--- /dev/null
+++ b/test/test_ip_mcast.py
@@ -0,0 +1,612 @@
+#!/usr/bin/env python
+
+import unittest
+
+from framework import VppTestCase, VppTestRunner
+from vpp_sub_interface import VppSubInterface, VppDot1QSubint, VppDot1ADSubint
+from vpp_ip_route import IpMRoute, MRoutePath, MFibSignal
+
+from scapy.packet import Raw
+from scapy.layers.l2 import Ether
+from scapy.layers.inet import IP, UDP, getmacbyip
+from scapy.layers.inet6 import IPv6, getmacbyip6
+from util import ppp
+
+
+class MRouteItfFlags:
+    MFIB_ITF_FLAG_NONE = 0
+    MFIB_ITF_FLAG_NEGATE_SIGNAL = 1
+    MFIB_ITF_FLAG_ACCEPT = 2
+    MFIB_ITF_FLAG_FORWARD = 4
+    MFIB_ITF_FLAG_SIGNAL_PRESENT = 8
+    MFIB_ITF_FLAG_INTERNAL_COPY = 16
+
+
+class MRouteEntryFlags:
+    MFIB_ENTRY_FLAG_NONE = 0
+    MFIB_ENTRY_FLAG_SIGNAL = 1
+    MFIB_ENTRY_FLAG_DROP = 2
+    MFIB_ENTRY_FLAG_CONNECTED = 4
+    MFIB_ENTRY_FLAG_INHERIT_ACCEPT = 8
+
+
+class TestIPMcast(VppTestCase):
+    """ IP Multicast Test Case """
+
+    def setUp(self):
+        super(TestIPMcast, self).setUp()
+
+        # create 4 pg interfaces
+        self.create_pg_interfaces(range(4))
+
+        # setup interfaces
+        for i in self.pg_interfaces:
+            i.admin_up()
+            i.config_ip4()
+            i.config_ip6()
+            i.resolve_arp()
+            i.resolve_ndp()
+
+    def create_stream_ip4(self, src_if, src_ip, dst_ip):
+        pkts = []
+        for i in range(0, 65):
+            info = self.create_packet_info(src_if, src_if)
+            payload = self.info_to_payload(info)
+            p = (Ether(dst=src_if.local_mac, src=src_if.remote_mac) /
+                 IP(src=src_ip, dst=dst_ip) /
+                 UDP(sport=1234, dport=1234) /
+                 Raw(payload))
+            info.data = p.copy()
+            pkts.append(p)
+        return pkts
+
+    def create_stream_ip6(self, src_if, src_ip, dst_ip):
+        pkts = []
+        for i in range(0, 65):
+            info = self.create_packet_info(src_if, src_if)
+            payload = self.info_to_payload(info)
+            p = (Ether(dst=src_if.local_mac, src=src_if.remote_mac) /
+                 IPv6(src=src_ip, dst=dst_ip) /
+                 UDP(sport=1234, dport=1234) /
+                 Raw(payload))
+            info.data = p.copy()
+            pkts.append(p)
+        return pkts
+
+    def verify_filter(self, capture, sent):
+        if not len(capture) == len(sent):
+            # filter out any IPv6 RAs from the captur
+            for p in capture:
+                if (p.haslayer(IPv6)):
+                    capture.remove(p)
+        return capture
+
+    def verify_capture_ip4(self, src_if, sent):
+        rxd = self.pg1.get_capture(65)
+
+        try:
+            capture = self.verify_filter(rxd, sent)
+
+            self.assertEqual(len(capture), len(sent))
+
+            for i in range(len(capture)):
+                tx = sent[i]
+                rx = capture[i]
+
+                # the rx'd packet has the MPLS label popped
+                eth = rx[Ether]
+                self.assertEqual(eth.type, 0x800)
+
+                tx_ip = tx[IP]
+                rx_ip = rx[IP]
+
+                # check the MAC address on the RX'd packet is correctly formed
+                self.assertEqual(eth.dst, getmacbyip(rx_ip.dst))
+
+                self.assertEqual(rx_ip.src, tx_ip.src)
+                self.assertEqual(rx_ip.dst, tx_ip.dst)
+                # IP processing post pop has decremented the TTL
+                self.assertEqual(rx_ip.ttl + 1, tx_ip.ttl)
+
+        except:
+            raise
+
+    def verify_capture_ip6(self, src_if, sent):
+        capture = self.pg1.get_capture(65)
+
+        self.assertEqual(len(capture), len(sent))
+
+        for i in range(len(capture)):
+            tx = sent[i]
+            rx = capture[i]
+
+            # the rx'd packet has the MPLS label popped
+            eth = rx[Ether]
+            self.assertEqual(eth.type, 0x86DD)
+
+            tx_ip = tx[IPv6]
+            rx_ip = rx[IPv6]
+
+            # check the MAC address on the RX'd packet is correctly formed
+            self.assertEqual(eth.dst, getmacbyip6(rx_ip.dst))
+
+            self.assertEqual(rx_ip.src, tx_ip.src)
+            self.assertEqual(rx_ip.dst, tx_ip.dst)
+            # IP processing post pop has decremented the TTL
+            self.assertEqual(rx_ip.hlim + 1, tx_ip.hlim)
+
+    def test_ip_mcast(self):
+        """ IP Multicast Replication """
+
+        #
+        # a stream that matches the default route. gets dropped.
+        #
+        self.vapi.cli("clear trace")
+        tx = self.create_stream_ip4(self.pg0, "1.1.1.1", "232.1.1.1")
+        self.pg0.add_stream(tx)
+
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        self.pg0.assert_nothing_captured(
+            remark="IP multicast packets forwarded on default route")
+
+        #
+        # A (*,G).
+        # one accepting interface, pg0, 3 forwarding interfaces
+        #
+        route_232_1_1_1 = IpMRoute(
+            self,
+            "0.0.0.0",
+            "232.1.1.1", 32,
+            MRouteEntryFlags.MFIB_ENTRY_FLAG_NONE,
+            [MRoutePath(self.pg0.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_ACCEPT),
+             MRoutePath(self.pg1.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_FORWARD),
+             MRoutePath(self.pg2.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_FORWARD),
+             MRoutePath(self.pg3.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_FORWARD)])
+        route_232_1_1_1.add_vpp_config()
+
+        #
+        # An (S,G).
+        # one accepting interface, pg0, 2 forwarding interfaces
+        #
+        route_1_1_1_1_232_1_1_1 = IpMRoute(
+            self,
+            "1.1.1.1",
+            "232.1.1.1", 64,
+            MRouteEntryFlags.MFIB_ENTRY_FLAG_NONE,
+            [MRoutePath(self.pg0.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_ACCEPT),
+             MRoutePath(self.pg1.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_FORWARD),
+             MRoutePath(self.pg2.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_FORWARD)])
+        route_1_1_1_1_232_1_1_1.add_vpp_config()
+
+        #
+        # An (*,G/m).
+        # one accepting interface, pg0, 1 forwarding interfaces
+        #
+        route_232 = IpMRoute(
+            self,
+            "0.0.0.0",
+            "232.0.0.0", 8,
+            MRouteEntryFlags.MFIB_ENTRY_FLAG_NONE,
+            [MRoutePath(self.pg0.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_ACCEPT),
+             MRoutePath(self.pg1.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_FORWARD)])
+        route_232.add_vpp_config()
+
+        #
+        # a stream that matches the route for (1.1.1.1,232.1.1.1)
+        #
+        self.vapi.cli("clear trace")
+        tx = self.create_stream_ip4(self.pg0, "1.1.1.1", "232.1.1.1")
+        self.pg0.add_stream(tx)
+
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        # We expect replications on Pg1, 2,
+        self.verify_capture_ip4(self.pg1, tx)
+        self.verify_capture_ip4(self.pg2, tx)
+
+        # no replications on Pg0
+        self.pg0.assert_nothing_captured(
+            remark="IP multicast packets forwarded on PG0")
+        self.pg3.assert_nothing_captured(
+            remark="IP multicast packets forwarded on PG3")
+
+        #
+        # a stream that matches the route for (*,232.0.0.0/8)
+        # Send packets with the 9th bit set so we test the correct clearing
+        # of that bit in the mac rewrite
+        #
+        self.vapi.cli("clear trace")
+        tx = self.create_stream_ip4(self.pg0, "1.1.1.1", "232.255.255.255")
+        self.pg0.add_stream(tx)
+
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        # We expect replications on Pg1 only
+        self.verify_capture_ip4(self.pg1, tx)
+
+        # no replications on Pg0, Pg2 not Pg3
+        self.pg0.assert_nothing_captured(
+            remark="IP multicast packets forwarded on PG0")
+        self.pg2.assert_nothing_captured(
+            remark="IP multicast packets forwarded on PG2")
+        self.pg3.assert_nothing_captured(
+            remark="IP multicast packets forwarded on PG3")
+
+        #
+        # a stream that matches the route for (*,232.1.1.1)
+        #
+        self.vapi.cli("clear trace")
+        tx = self.create_stream_ip4(self.pg0, "1.1.1.2", "232.1.1.1")
+        self.pg0.add_stream(tx)
+
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        # We expect replications on Pg1, 2, 3.
+        self.verify_capture_ip4(self.pg1, tx)
+        self.verify_capture_ip4(self.pg2, tx)
+        self.verify_capture_ip4(self.pg3, tx)
+
+        # no replications on Pg0
+        self.pg0.assert_nothing_captured(
+            remark="IP multicast packets forwarded on PG0")
+
+        route_232_1_1_1.remove_vpp_config()
+        route_1_1_1_1_232_1_1_1.remove_vpp_config()
+        route_232.remove_vpp_config()
+
+    def test_ip6_mcast(self):
+        """ IPv6 Multicast Replication """
+
+        #
+        # a stream that matches the default route. gets dropped.
+        #
+        self.vapi.cli("clear trace")
+        tx = self.create_stream_ip6(self.pg0, "2001::1", "ff01::1")
+        self.pg0.add_stream(tx)
+
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        self.pg0.assert_nothing_captured(
+            remark="IPv6 multicast packets forwarded on default route")
+
+        #
+        # A (*,G).
+        # one accepting interface, pg0, 3 forwarding interfaces
+        #
+        route_ff01_1 = IpMRoute(
+            self,
+            "::",
+            "ff01::1", 128,
+            MRouteEntryFlags.MFIB_ENTRY_FLAG_NONE,
+            [MRoutePath(self.pg0.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_ACCEPT),
+             MRoutePath(self.pg1.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_FORWARD),
+             MRoutePath(self.pg2.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_FORWARD),
+             MRoutePath(self.pg3.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_FORWARD)],
+            is_ip6=1)
+        route_ff01_1.add_vpp_config()
+
+        #
+        # An (S,G).
+        # one accepting interface, pg0, 2 forwarding interfaces
+        #
+        route_2001_ff01_1 = IpMRoute(
+            self,
+            "2001::1",
+            "ff01::1", 256,
+            MRouteEntryFlags.MFIB_ENTRY_FLAG_NONE,
+            [MRoutePath(self.pg0.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_ACCEPT),
+             MRoutePath(self.pg1.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_FORWARD),
+             MRoutePath(self.pg2.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_FORWARD)],
+            is_ip6=1)
+        route_2001_ff01_1.add_vpp_config()
+
+        #
+        # An (*,G/m).
+        # one accepting interface, pg0, 1 forwarding interface
+        #
+        route_ff01 = IpMRoute(
+            self,
+            "::",
+            "ff01::", 16,
+            MRouteEntryFlags.MFIB_ENTRY_FLAG_NONE,
+            [MRoutePath(self.pg0.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_ACCEPT),
+             MRoutePath(self.pg1.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_FORWARD)],
+            is_ip6=1)
+        route_ff01.add_vpp_config()
+
+        #
+        # a stream that matches the route for (*, ff01::/16)
+        #
+        self.vapi.cli("clear trace")
+        tx = self.create_stream_ip6(self.pg0, "2002::1", "ff01:2::255")
+        self.pg0.add_stream(tx)
+
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        # We expect replications on Pg1
+        self.verify_capture_ip6(self.pg1, tx)
+
+        # no replications on Pg0, Pg3
+        self.pg0.assert_nothing_captured(
+            remark="IP multicast packets forwarded on PG0")
+        self.pg2.assert_nothing_captured(
+            remark="IP multicast packets forwarded on PG2")
+        self.pg3.assert_nothing_captured(
+            remark="IP multicast packets forwarded on PG3")
+
+        #
+        # a stream that matches the route for (*,ff01::1)
+        #
+        self.vapi.cli("clear trace")
+        tx = self.create_stream_ip6(self.pg0, "2002::2", "ff01::1")
+        self.pg0.add_stream(tx)
+
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        # We expect replications on Pg1, 2, 3.
+        self.verify_capture_ip6(self.pg1, tx)
+        self.verify_capture_ip6(self.pg2, tx)
+        self.verify_capture_ip6(self.pg3, tx)
+
+        # no replications on Pg0
+        self.pg0.assert_nothing_captured(
+            remark="IPv6 multicast packets forwarded on PG0")
+
+        #
+        # a stream that matches the route for (2001::1, ff00::1)
+        #
+        self.vapi.cli("clear trace")
+        tx = self.create_stream_ip6(self.pg0, "2001::1", "ff01::1")
+        self.pg0.add_stream(tx)
+
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        # We expect replications on Pg1, 2,
+        self.verify_capture_ip6(self.pg1, tx)
+        self.verify_capture_ip6(self.pg2, tx)
+
+        # no replications on Pg0, Pg3
+        self.pg0.assert_nothing_captured(
+            remark="IP multicast packets forwarded on PG0")
+        self.pg3.assert_nothing_captured(
+            remark="IP multicast packets forwarded on PG3")
+
+        route_ff01.remove_vpp_config()
+        route_ff01_1.remove_vpp_config()
+        route_2001_ff01_1.remove_vpp_config()
+
+    def _mcast_connected_send_stream(self, dst_ip):
+        self.vapi.cli("clear trace")
+        tx = self.create_stream_ip4(self.pg0,
+                                    self.pg0.remote_ip4,
+                                    dst_ip)
+        self.pg0.add_stream(tx)
+
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        # We expect replications on Pg1.
+        self.verify_capture_ip4(self.pg1, tx)
+
+        return tx
+
+    def test_ip_mcast_connected(self):
+        """ IP Multicast Connected Source check """
+
+        #
+        # A (*,G).
+        # one accepting interface, pg0, 1 forwarding interfaces
+        #
+        route_232_1_1_1 = IpMRoute(
+            self,
+            "0.0.0.0",
+            "232.1.1.1", 32,
+            MRouteEntryFlags.MFIB_ENTRY_FLAG_NONE,
+            [MRoutePath(self.pg0.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_ACCEPT),
+             MRoutePath(self.pg1.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_FORWARD)])
+
+        route_232_1_1_1.add_vpp_config()
+        route_232_1_1_1.update_entry_flags(
+            MRouteEntryFlags.MFIB_ENTRY_FLAG_CONNECTED)
+
+        #
+        # Now the (*,G) is present, send from connected source
+        #
+        tx = self._mcast_connected_send_stream("232.1.1.1")
+
+        #
+        # Constrct a representation of the signal we expect on pg0
+        #
+        signal_232_1_1_1_itf_0 = MFibSignal(self,
+                                            route_232_1_1_1,
+                                            self.pg0.sw_if_index,
+                                            tx[0])
+
+        #
+        # read the only expected signal
+        #
+        signals = self.vapi.mfib_signal_dump()
+
+        self.assertEqual(1, len(signals))
+
+        signal_232_1_1_1_itf_0.compare(signals[0])
+
+        #
+        # reading the signal allows for the generation of another
+        # so send more packets and expect the next signal
+        #
+        tx = self._mcast_connected_send_stream("232.1.1.1")
+
+        signals = self.vapi.mfib_signal_dump()
+        self.assertEqual(1, len(signals))
+        signal_232_1_1_1_itf_0.compare(signals[0])
+
+        #
+        # A Second entry with connected check
+        # one accepting interface, pg0, 1 forwarding interfaces
+        #
+        route_232_1_1_2 = IpMRoute(
+            self,
+            "0.0.0.0",
+            "232.1.1.2", 32,
+            MRouteEntryFlags.MFIB_ENTRY_FLAG_NONE,
+            [MRoutePath(self.pg0.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_ACCEPT),
+             MRoutePath(self.pg1.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_FORWARD)])
+
+        route_232_1_1_2.add_vpp_config()
+        route_232_1_1_2.update_entry_flags(
+            MRouteEntryFlags.MFIB_ENTRY_FLAG_CONNECTED)
+
+        #
+        # Send traffic to both entries. One read should net us two signals
+        #
+        signal_232_1_1_2_itf_0 = MFibSignal(self,
+                                            route_232_1_1_2,
+                                            self.pg0.sw_if_index,
+                                            tx[0])
+        tx = self._mcast_connected_send_stream("232.1.1.1")
+        tx2 = self._mcast_connected_send_stream("232.1.1.2")
+
+        #
+        # read the only expected signal
+        #
+        signals = self.vapi.mfib_signal_dump()
+
+        self.assertEqual(2, len(signals))
+
+        signal_232_1_1_1_itf_0.compare(signals[1])
+        signal_232_1_1_2_itf_0.compare(signals[0])
+
+        route_232_1_1_1.remove_vpp_config()
+        route_232_1_1_2.remove_vpp_config()
+
+    def test_ip_mcast_signal(self):
+        """ IP Multicast Signal """
+
+        #
+        # A (*,G).
+        # one accepting interface, pg0, 1 forwarding interfaces
+        #
+        route_232_1_1_1 = IpMRoute(
+            self,
+            "0.0.0.0",
+            "232.1.1.1", 32,
+            MRouteEntryFlags.MFIB_ENTRY_FLAG_NONE,
+            [MRoutePath(self.pg0.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_ACCEPT),
+             MRoutePath(self.pg1.sw_if_index,
+                        MRouteItfFlags.MFIB_ITF_FLAG_FORWARD)])
+
+        route_232_1_1_1.add_vpp_config()
+        route_232_1_1_1.update_entry_flags(
+            MRouteEntryFlags.MFIB_ENTRY_FLAG_SIGNAL)
+
+        #
+        # Now the (*,G) is present, send from connected source
+        #
+        tx = self._mcast_connected_send_stream("232.1.1.1")
+
+        #
+        # Constrct a representation of the signal we expect on pg0
+        #
+        signal_232_1_1_1_itf_0 = MFibSignal(self,
+                                            route_232_1_1_1,
+                                            self.pg0.sw_if_index,
+                                            tx[0])
+
+        #
+        # read the only expected signal
+        #
+        signals = self.vapi.mfib_signal_dump()
+
+        self.assertEqual(1, len(signals))
+
+        signal_232_1_1_1_itf_0.compare(signals[0])
+
+        #
+        # reading the signal allows for the generation of another
+        # so send more packets and expect the next signal
+        #
+        tx = self._mcast_connected_send_stream("232.1.1.1")
+
+        signals = self.vapi.mfib_signal_dump()
+        self.assertEqual(1, len(signals))
+        signal_232_1_1_1_itf_0.compare(signals[0])
+
+        #
+        # Set the negate-signal on the accepting interval - the signals
+        # should stop
+        #
+        route_232_1_1_1.update_path_flags(
+            self.pg0.sw_if_index,
+            (MRouteItfFlags.MFIB_ITF_FLAG_ACCEPT |
+             MRouteItfFlags.MFIB_ITF_FLAG_NEGATE_SIGNAL))
+
+        tx = self._mcast_connected_send_stream("232.1.1.1")
+
+        signals = self.vapi.mfib_signal_dump()
+        self.assertEqual(0, len(signals))
+
+        #
+        # Clear the SIGNAL flag on the entry and the signals should
+        # come back since the interface is still NEGATE-SIGNAL
+        #
+        route_232_1_1_1.update_entry_flags(
+            MRouteEntryFlags.MFIB_ENTRY_FLAG_NONE)
+
+        tx = self._mcast_connected_send_stream("232.1.1.1")
+
+        signals = self.vapi.mfib_signal_dump()
+        self.assertEqual(1, len(signals))
+        signal_232_1_1_1_itf_0.compare(signals[0])
+
+        #
+        # Lastly remove the NEGATE-SIGNAL from the interface and the
+        # signals should stop
+        #
+        route_232_1_1_1.update_path_flags(self.pg0.sw_if_index,
+                                          MRouteItfFlags.MFIB_ITF_FLAG_ACCEPT)
+
+        tx = self._mcast_connected_send_stream("232.1.1.1")
+        signals = self.vapi.mfib_signal_dump()
+        self.assertEqual(0, len(signals))
+
+        #
+        # Cleanup
+        #
+        route_232_1_1_1.remove_vpp_config()
+
+
+if __name__ == '__main__':
+    unittest.main(testRunner=VppTestRunner)
diff --git a/test/test_mfib.py b/test/test_mfib.py
new file mode 100644
index 00000000..4d0d2a3f
--- /dev/null
+++ b/test/test_mfib.py
@@ -0,0 +1,23 @@
+#!/usr/bin/env python
+
+import unittest
+
+from framework import VppTestCase, VppTestRunner
+
+
+class TestMFIB(VppTestCase):
+    """ MFIB Test Case """
+
+    def setUp(self):
+        super(TestMFIB, self).setUp()
+
+    def test_mfib(self):
+        """ MFIB Unit Tests """
+        error = self.vapi.cli("test mfib")
+
+        if error:
+            self.logger.critical(error)
+        self.assertEqual(error.find("Failed"), -1)
+
+if __name__ == '__main__':
+    unittest.main(testRunner=VppTestRunner)
diff --git a/test/vpp_interface.py b/test/vpp_interface.py
index ee4a9ef6..29edb70e 100644
--- a/test/vpp_interface.py
+++ b/test/vpp_interface.py
@@ -260,9 +260,10 @@ class VppInterface(object):
         """Configure IPv6 RA suppress on the VPP interface."""
         self.test.vapi.sw_interface_ra_suppress(self.sw_if_index)
 
-    def ip6_ra_config(self, suppress=0, send_unicast=0):
+    def ip6_ra_config(self, no=0, suppress=0, send_unicast=0):
         """Configure IPv6 RA suppress on the VPP interface."""
         self.test.vapi.ip6_sw_interface_ra_config(self.sw_if_index,
+                                                  no,
                                                   suppress,
                                                   send_unicast)
 
diff --git a/test/vpp_ip_route.py b/test/vpp_ip_route.py
index fc9133fb..1804fbbd 100644
--- a/test/vpp_ip_route.py
+++ b/test/vpp_ip_route.py
@@ -11,7 +11,7 @@ MPLS_IETF_MAX_LABEL = 0xfffff
 MPLS_LABEL_INVALID = MPLS_IETF_MAX_LABEL + 1
 
 
-class RoutePath:
+class RoutePath(object):
 
     def __init__(
             self,
@@ -31,6 +31,14 @@ class RoutePath:
             self.nh_addr = socket.inet_pton(socket.AF_INET, nh_addr)
 
 
+class MRoutePath(RoutePath):
+
+    def __init__(self, nh_sw_if_index, flags):
+        super(MRoutePath, self).__init__("0.0.0.0",
+                                         nh_sw_if_index)
+        self.nh_i_flags = flags
+
+
 class IpRoute:
     """
     IP Route
@@ -94,6 +102,97 @@ class IpRoute:
                                                  is_add=0)
 
 
+class IpMRoute:
+    """
+    IP Multicast Route
+    """
+
+    def __init__(self, test, src_addr, grp_addr,
+                 grp_addr_len, e_flags, paths, table_id=0, is_ip6=0):
+        self._test = test
+        self.paths = paths
+        self.grp_addr_len = grp_addr_len
+        self.table_id = table_id
+        self.e_flags = e_flags
+        self.is_ip6 = is_ip6
+
+        if is_ip6:
+            self.grp_addr = socket.inet_pton(socket.AF_INET6, grp_addr)
+            self.src_addr = socket.inet_pton(socket.AF_INET6, src_addr)
+        else:
+            self.grp_addr = socket.inet_pton(socket.AF_INET, grp_addr)
+            self.src_addr = socket.inet_pton(socket.AF_INET, src_addr)
+
+    def add_vpp_config(self):
+        for path in self.paths:
+            self._test.vapi.ip_mroute_add_del(self.src_addr,
+                                              self.grp_addr,
+                                              self.grp_addr_len,
+                                              self.e_flags,
+                                              path.nh_itf,
+                                              path.nh_i_flags,
+                                              table_id=self.table_id,
+                                              is_ipv6=self.is_ip6)
+
+    def remove_vpp_config(self):
+        for path in self.paths:
+            self._test.vapi.ip_mroute_add_del(self.src_addr,
+                                              self.grp_addr,
+                                              self.grp_addr_len,
+                                              self.e_flags,
+                                              path.nh_itf,
+                                              path.nh_i_flags,
+                                              table_id=self.table_id,
+                                              is_add=0,
+                                              is_ipv6=self.is_ip6)
+
+    def update_entry_flags(self, flags):
+        self.e_flags = flags
+        self._test.vapi.ip_mroute_add_del(self.src_addr,
+                                          self.grp_addr,
+                                          self.grp_addr_len,
+                                          self.e_flags,
+                                          0xffffffff,
+                                          0,
+                                          table_id=self.table_id,
+                                          is_ipv6=self.is_ip6)
+
+    def update_path_flags(self, itf, flags):
+        for path in self.paths:
+            if path.nh_itf == itf:
+                path.nh_i_flags = flags
+                break
+        self._test.vapi.ip_mroute_add_del(self.src_addr,
+                                          self.grp_addr,
+                                          self.grp_addr_len,
+                                          self.e_flags,
+                                          path.nh_itf,
+                                          path.nh_i_flags,
+                                          table_id=self.table_id,
+                                          is_ipv6=self.is_ip6)
+
+
+class MFibSignal:
+    def __init__(self, test, route, interface, packet):
+        self.route = route
+        self.interface = interface
+        self.packet = packet
+        self.test = test
+
+    def compare(self, signal):
+        self.test.assertEqual(self.interface, signal.sw_if_index)
+        self.test.assertEqual(self.route.table_id, signal.table_id)
+        self.test.assertEqual(self.route.grp_addr_len,
+                              signal.grp_address_len)
+        for i in range(self.route.grp_addr_len / 8):
+            self.test.assertEqual(self.route.grp_addr[i],
+                                  signal.grp_address[i])
+        if (self.route.grp_addr_len > 32):
+            for i in range(4):
+                self.test.assertEqual(self.route.src_addr[i],
+                                      signal.src_address[i])
+
+
 class MplsIpBind:
     """
     MPLS to IP Binding
diff --git a/test/vpp_papi_provider.py b/test/vpp_papi_provider.py
index 1b2895e9..90c954dc 100644
--- a/test/vpp_papi_provider.py
+++ b/test/vpp_papi_provider.py
@@ -252,10 +252,12 @@ class VppPapiProvider(object):
                         {'sw_if_index': sw_if_index})
 
     def ip6_sw_interface_ra_config(self, sw_if_index,
+                                   no,
                                    suppress,
-                                   send_unicast,):
+                                   send_unicast):
         return self.api(self.papi.sw_interface_ip6nd_ra_config,
                         {'sw_if_index': sw_if_index,
+                         'is_no': no,
                          'suppress': suppress,
                          'send_unicast': send_unicast})
 
@@ -1178,3 +1180,33 @@ class VppPapiProvider(object):
                 'is_add': is_add,
                 'oui': oui,
             })
+
+    def ip_mroute_add_del(self,
+                          src_address,
+                          grp_address,
+                          grp_address_length,
+                          e_flags,
+                          next_hop_sw_if_index,
+                          i_flags,
+                          table_id=0,
+                          create_vrf_if_needed=0,
+                          is_add=1,
+                          is_ipv6=0,
+                          is_local=0):
+        """
+        """
+        return self.api(
+            self.papi.ip_mroute_add_del,
+            {'next_hop_sw_if_index': next_hop_sw_if_index,
+             'entry_flags': e_flags,
+             'itf_flags': i_flags,
+             'create_vrf_if_needed': create_vrf_if_needed,
+             'is_add': is_add,
+             'is_ipv6': is_ipv6,
+             'is_local': is_local,
+             'grp_address_length': grp_address_length,
+             'grp_address': grp_address,
+             'src_address': src_address})
+
+    def mfib_signal_dump(self):
+        return self.api(self.papi.mfib_signal_dump, {})
-- 
cgit 1.2.3-korg


From 20a175a18414c67e38b5ce0709b33fb1df8069c9 Mon Sep 17 00:00:00 2001
From: Neale Ranns <nranns@cisco.com>
Date: Tue, 14 Feb 2017 07:28:41 -0800
Subject: dhcp: multiple additions

DHCP additions:
1) DHCPv4 will only relay a message back to the client, if the Option82 information is present. So make this the default.
2) It is no longer possible to select via the API to "insert circuit ID" - since this is now default
3) Remove the version 2 API since it's now the same as version 1.
4) Adding the VSS option is now conditional only on the presence of VSS config (not the 'insert' option in the set API)
5) DHCP proxy dump via API

Change-Id: Ia7271ba8c1d4dbf34a02c401d268ccfbb1b74f17
Signed-off-by: Neale Ranns <nranns@cisco.com>
---
 src/scripts/vnet/dhcp/proxy   |  21 ++
 src/vat/api_format.c          | 173 ++++++-----
 src/vnet/dhcp/client.c        |   1 +
 src/vnet/dhcp/client.h        |   3 +
 src/vnet/dhcp/dhcp.api        |  67 ++---
 src/vnet/dhcp/dhcp_api.c      | 112 ++++---
 src/vnet/dhcp/proxy.h         |  33 ++-
 src/vnet/dhcp/proxy_error.def |   3 +-
 src/vnet/dhcp/proxy_node.c    | 676 +++++++++++++++++++++++-------------------
 src/vnet/dhcpv6/proxy.h       |  19 +-
 src/vnet/dhcpv6/proxy_node.c  | 513 ++++++++++++++++++--------------
 src/vnet/dpo/receive_dpo.c    |   5 +
 src/vpp/api/custom_dump.c     |  34 ---
 test/test_dhcp.py             | 166 ++++++-----
 test/vpp_papi_provider.py     |   6 +-
 15 files changed, 978 insertions(+), 854 deletions(-)
 create mode 100644 src/scripts/vnet/dhcp/proxy

(limited to 'test/test_dhcp.py')

diff --git a/src/scripts/vnet/dhcp/proxy b/src/scripts/vnet/dhcp/proxy
new file mode 100644
index 00000000..c709d87d
--- /dev/null
+++ b/src/scripts/vnet/dhcp/proxy
@@ -0,0 +1,21 @@
+loop create
+loop create
+
+set int state loop0 up
+set int state loop1 up
+
+set int ip  table loop1 1
+set int ip6 table loop1 1
+
+set int ip addr loop0 10.0.0.1/24
+set int ip addr loop0 10.0.1.1/24
+
+set int ip addr loop0 2001::1/64
+set int ip addr loop0 2001:1::1/64
+
+set dhcp proxy server 10.255.0.1 src-address 10.0.0.1 server-fib-id 0 rx-fib-id 0
+set dhcp proxy server 10.255.0.2 src-address 10.0.1.1 server-fib-id 1 rx-fib-id 1
+
+set dhcpv6 proxy server 3001::1 src-address 2001::1 server-fib-id 0 rx-fib-id 0
+set dhcpv6 proxy server 3002::1 src-address 2001:1::1 server-fib-id 1 rx-fib-id 1
+
diff --git a/src/vat/api_format.c b/src/vat/api_format.c
index 11e68214..78c5e279 100644
--- a/src/vat/api_format.c
+++ b/src/vat/api_format.c
@@ -3819,7 +3819,6 @@ _(reset_vrf_reply)                                      \
 _(oam_add_del_reply)                                    \
 _(reset_fib_reply)                                      \
 _(dhcp_proxy_config_reply)                              \
-_(dhcp_proxy_config_2_reply)                            \
 _(dhcp_proxy_set_vss_reply)                             \
 _(dhcp_client_config_reply)                             \
 _(set_ip_flow_hash_reply)                               \
@@ -4033,8 +4032,8 @@ _(CREATE_SUBIF_REPLY, create_subif_reply)                     		\
 _(OAM_ADD_DEL_REPLY, oam_add_del_reply)                                 \
 _(RESET_FIB_REPLY, reset_fib_reply)                                     \
 _(DHCP_PROXY_CONFIG_REPLY, dhcp_proxy_config_reply)                     \
-_(DHCP_PROXY_CONFIG_2_REPLY, dhcp_proxy_config_2_reply)                 \
 _(DHCP_PROXY_SET_VSS_REPLY, dhcp_proxy_set_vss_reply)                   \
+_(DHCP_PROXY_DETAILS, dhcp_proxy_details)                               \
 _(DHCP_CLIENT_CONFIG_REPLY, dhcp_client_config_reply)                   \
 _(SET_IP_FLOW_HASH_REPLY, set_ip_flow_hash_reply)                       \
 _(SW_INTERFACE_IP6_ENABLE_DISABLE_REPLY,                                \
@@ -7635,9 +7634,9 @@ api_dhcp_proxy_config (vat_main_t * vam)
 {
   unformat_input_t *i = vam->input;
   vl_api_dhcp_proxy_config_t *mp;
-  u32 vrf_id = 0;
+  u32 rx_vrf_id = 0;
+  u32 server_vrf_id = 0;
   u8 is_add = 1;
-  u8 insert_cid = 1;
   u8 v4_address_set = 0;
   u8 v6_address_set = 0;
   ip4_address_t v4address;
@@ -7653,9 +7652,9 @@ api_dhcp_proxy_config (vat_main_t * vam)
     {
       if (unformat (i, "del"))
 	is_add = 0;
-      else if (unformat (i, "vrf %d", &vrf_id))
+      else if (unformat (i, "rx_vrf_id %d", &rx_vrf_id))
 	;
-      else if (unformat (i, "insert-cid %d", &insert_cid))
+      else if (unformat (i, "server_vrf_id %d", &server_vrf_id))
 	;
       else if (unformat (i, "svr %U", unformat_ip4_address, &v4address))
 	v4_address_set = 1;
@@ -7701,9 +7700,9 @@ api_dhcp_proxy_config (vat_main_t * vam)
   /* Construct the API message */
   M (DHCP_PROXY_CONFIG, mp);
 
-  mp->insert_circuit_id = insert_cid;
   mp->is_add = is_add;
-  mp->vrf_id = ntohl (vrf_id);
+  mp->rx_vrf_id = ntohl (rx_vrf_id);
+  mp->server_vrf_id = ntohl (server_vrf_id);
   if (v6_address_set)
     {
       mp->is_ipv6 = 1;
@@ -7724,100 +7723,98 @@ api_dhcp_proxy_config (vat_main_t * vam)
   return ret;
 }
 
-static int
-api_dhcp_proxy_config_2 (vat_main_t * vam)
+#define vl_api_dhcp_proxy_details_t_endian vl_noop_handler
+#define vl_api_dhcp_proxy_details_t_print vl_noop_handler
+
+static void
+vl_api_dhcp_proxy_details_t_handler (vl_api_dhcp_proxy_details_t * mp)
 {
-  unformat_input_t *i = vam->input;
-  vl_api_dhcp_proxy_config_2_t *mp;
-  u32 rx_vrf_id = 0;
-  u32 server_vrf_id = 0;
-  u8 is_add = 1;
-  u8 insert_cid = 1;
-  u8 v4_address_set = 0;
-  u8 v6_address_set = 0;
-  ip4_address_t v4address;
-  ip6_address_t v6address;
-  u8 v4_src_address_set = 0;
-  u8 v6_src_address_set = 0;
-  ip4_address_t v4srcaddress;
-  ip6_address_t v6srcaddress;
-  int ret;
+  vat_main_t *vam = &vat_main;
 
-  /* Parse args required to build the message */
-  while (unformat_check_input (i) != UNFORMAT_END_OF_INPUT)
-    {
-      if (unformat (i, "del"))
-	is_add = 0;
-      else if (unformat (i, "rx_vrf_id %d", &rx_vrf_id))
-	;
-      else if (unformat (i, "server_vrf_id %d", &server_vrf_id))
-	;
-      else if (unformat (i, "insert-cid %d", &insert_cid))
-	;
-      else if (unformat (i, "svr %U", unformat_ip4_address, &v4address))
-	v4_address_set = 1;
-      else if (unformat (i, "svr %U", unformat_ip6_address, &v6address))
-	v6_address_set = 1;
-      else if (unformat (i, "src %U", unformat_ip4_address, &v4srcaddress))
-	v4_src_address_set = 1;
-      else if (unformat (i, "src %U", unformat_ip6_address, &v6srcaddress))
-	v6_src_address_set = 1;
-      else
-	break;
-    }
+  if (mp->is_ipv6)
+    print (vam->ofp,
+	   "RX Table-ID %d, Server Table-ID %d, Server Address %U, Source Address %U, VSS FIB-ID %d, VSS OUI %d",
+	   ntohl (mp->rx_vrf_id),
+	   ntohl (mp->server_vrf_id),
+	   format_ip6_address, mp->dhcp_server,
+	   format_ip6_address, mp->dhcp_src_address,
+	   ntohl (mp->vss_oui), ntohl (mp->vss_fib_id));
+  else
+    print (vam->ofp,
+	   "RX Table-ID %d, Server Table-ID %d, Server Address %U, Source Address %U, VSS FIB-ID %d, VSS OUI %d",
+	   ntohl (mp->rx_vrf_id),
+	   ntohl (mp->server_vrf_id),
+	   format_ip4_address, mp->dhcp_server,
+	   format_ip4_address, mp->dhcp_src_address,
+	   ntohl (mp->vss_oui), ntohl (mp->vss_fib_id));
+}
 
-  if (v4_address_set && v6_address_set)
-    {
-      errmsg ("both v4 and v6 server addresses set");
-      return -99;
-    }
-  if (!v4_address_set && !v6_address_set)
-    {
-      errmsg ("no server addresses set");
-      return -99;
-    }
+static void vl_api_dhcp_proxy_details_t_handler_json
+  (vl_api_dhcp_proxy_details_t * mp)
+{
+  vat_main_t *vam = &vat_main;
+  vat_json_node_t *node = NULL;
+  struct in_addr ip4;
+  struct in6_addr ip6;
 
-  if (v4_src_address_set && v6_src_address_set)
+  if (VAT_JSON_ARRAY != vam->json_tree.type)
     {
-      errmsg ("both v4 and v6  src addresses set");
-      return -99;
+      ASSERT (VAT_JSON_NONE == vam->json_tree.type);
+      vat_json_init_array (&vam->json_tree);
     }
-  if (!v4_src_address_set && !v6_src_address_set)
+  node = vat_json_array_add (&vam->json_tree);
+
+  vat_json_init_object (node);
+  vat_json_object_add_uint (node, "rx-table-id", ntohl (mp->rx_vrf_id));
+  vat_json_object_add_uint (node, "server-table-id",
+			    ntohl (mp->server_vrf_id));
+  if (mp->is_ipv6)
     {
-      errmsg ("no src addresses set");
-      return -99;
+      clib_memcpy (&ip6, &mp->dhcp_server, sizeof (ip6));
+      vat_json_object_add_ip6 (node, "server_address", ip6);
+      clib_memcpy (&ip6, &mp->dhcp_src_address, sizeof (ip6));
+      vat_json_object_add_ip6 (node, "src_address", ip6);
     }
-
-  if (!(v4_src_address_set && v4_address_set) &&
-      !(v6_src_address_set && v6_address_set))
+  else
     {
-      errmsg ("no matching server and src addresses set");
-      return -99;
+      clib_memcpy (&ip4, &mp->dhcp_server, sizeof (ip4));
+      vat_json_object_add_ip4 (node, "server_address", ip4);
+      clib_memcpy (&ip4, &mp->dhcp_src_address, sizeof (ip4));
+      vat_json_object_add_ip4 (node, "src_address", ip4);
     }
+  vat_json_object_add_uint (node, "vss-fib-id", ntohl (mp->vss_fib_id));
+  vat_json_object_add_uint (node, "vss-oui", ntohl (mp->vss_oui));
+}
 
-  /* Construct the API message */
-  M (DHCP_PROXY_CONFIG_2, mp);
+static int
+api_dhcp_proxy_dump (vat_main_t * vam)
+{
+  unformat_input_t *i = vam->input;
+  vl_api_control_ping_t *mp_ping;
+  vl_api_dhcp_proxy_dump_t *mp;
+  u8 is_ipv6 = 0;
+  int ret;
 
-  mp->insert_circuit_id = insert_cid;
-  mp->is_add = is_add;
-  mp->rx_vrf_id = ntohl (rx_vrf_id);
-  mp->server_vrf_id = ntohl (server_vrf_id);
-  if (v6_address_set)
-    {
-      mp->is_ipv6 = 1;
-      clib_memcpy (mp->dhcp_server, &v6address, sizeof (v6address));
-      clib_memcpy (mp->dhcp_src_address, &v6srcaddress, sizeof (v6address));
-    }
-  else
+  while (unformat_check_input (i) != UNFORMAT_END_OF_INPUT)
     {
-      clib_memcpy (mp->dhcp_server, &v4address, sizeof (v4address));
-      clib_memcpy (mp->dhcp_src_address, &v4srcaddress, sizeof (v4address));
+      if (unformat (i, "ipv6"))
+	is_ipv6 = 1;
+      else
+	{
+	  clib_warning ("parse error '%U'", format_unformat_error, i);
+	  return -99;
+	}
     }
 
-  /* send it... */
+  M (DHCP_PROXY_DUMP, mp);
+
+  mp->is_ip6 = is_ipv6;
   S (mp);
 
-  /* Wait for a reply, return good/bad news  */
+  /* Use a control ping for synchronization */
+  M (CONTROL_PING, mp_ping);
+  S (mp_ping);
+
   W (ret);
   return ret;
 }
@@ -18187,12 +18184,10 @@ _(oam_add_del, "src <ip4-address> dst <ip4-address> [vrf <n>] [del]")   \
 _(reset_fib, "vrf <n> [ipv6]")                                          \
 _(dhcp_proxy_config,                                                    \
   "svr <v46-address> src <v46-address>\n"                               \
-   "insert-cid <n> [del]")                                              \
-_(dhcp_proxy_config_2,                                                  \
-  "svr <v46-address> src <v46-address>\n"                               \
-   "rx_vrf_id <nn> server_vrf_id <nn> insert-cid <n> [del]")            \
+   "rx_vrf_id <nn> server_vrf_id <nn>  [del]")                          \
 _(dhcp_proxy_set_vss,                                                   \
   "tbl_id <n> fib_id <n> oui <n> [ipv6] [del]")                         \
+_(dhcp_proxy_dump, "ip6")                                               \
 _(dhcp_client_config,                                                   \
   "<intfc> | sw_if_index <id> [hostname <name>] [disable_event] [del]") \
 _(set_ip_flow_hash,                                                     \
diff --git a/src/vnet/dhcp/client.c b/src/vnet/dhcp/client.c
index c352e310..8a1a43b3 100644
--- a/src/vnet/dhcp/client.c
+++ b/src/vnet/dhcp/client.c
@@ -13,6 +13,7 @@
  * limitations under the License.
  */
 #include <vlib/vlib.h>
+#include <vnet/dhcp/client.h>
 #include <vnet/dhcp/proxy.h>
 #include <vnet/fib/fib_table.h>
 
diff --git a/src/vnet/dhcp/client.h b/src/vnet/dhcp/client.h
index d15e686b..a74368cb 100644
--- a/src/vnet/dhcp/client.h
+++ b/src/vnet/dhcp/client.h
@@ -19,6 +19,9 @@
 #ifndef included_dhcp_client_h
 #define included_dhcp_client_h
 
+#include <vnet/ip/ip.h>
+#include <vnet/dhcp/packet.h>
+
 #define foreach_dhcp_client_state               \
 _(DHCP_DISCOVER)                                \
 _(DHCP_REQUEST)                                 \
diff --git a/src/vnet/dhcp/dhcp.api b/src/vnet/dhcp/dhcp.api
index c228cd04..8daadd8c 100644
--- a/src/vnet/dhcp/dhcp.api
+++ b/src/vnet/dhcp/dhcp.api
@@ -16,7 +16,8 @@
 /** \brief DHCP Proxy config add / del request
     @param client_index - opaque cookie to identify the sender
     @param context - sender context, to match reply w/ request
-    @param vrf_id - vrf id
+    @param rx_vrf_id - Rx/interface vrf id
+    @param server_vrf_id - server vrf id
     @param if_ipv6 - ipv6 if non-zero, else ipv4
     @param is_add - add the config if non-zero, else delete
     @param insert_circuit_id - option82 suboption 1 fib number
@@ -27,10 +28,10 @@ define dhcp_proxy_config
 {
   u32 client_index;
   u32 context;
-  u32 vrf_id;
+  u32 rx_vrf_id;
+  u32 server_vrf_id;
   u8 is_ipv6;
   u8 is_add;
-  u8 insert_circuit_id;
   u8 dhcp_server[16];
   u8 dhcp_src_address[16];
 };
@@ -45,40 +46,6 @@ define dhcp_proxy_config_reply
   i32 retval;
 };
 
-/** \brief DHCP Proxy config 2 add / del request
-    @param client_index - opaque cookie to identify the sender
-    @param context - sender context, to match reply w/ request
-    @param rx_vrf_id - receive vrf id
-    @param server_vrf_id - server vrf id
-    @param if_ipv6 - ipv6 if non-zero, else ipv4
-    @param is_add - add the config if non-zero, else delete
-    @param insert_circuit_id - option82 suboption 1 fib number
-    @param dhcp_server[] - server address
-    @param dhcp_src_address[] - <fix this, need details>
-*/
-define dhcp_proxy_config_2
-{
-  u32 client_index;
-  u32 context;
-  u32 rx_vrf_id;
-  u32 server_vrf_id;
-  u8 is_ipv6;
-  u8 is_add;
-  u8 insert_circuit_id;
-  u8 dhcp_server[16];
-  u8 dhcp_src_address[16];
-};
-
-/** \brief DHCP Proxy config 2 add / del response
-    @param context - sender context, to match reply w/ request
-    @param retval - return code for request
-*/
-define dhcp_proxy_config_2_reply
-{
-  u32 context;
-  i32 retval;
-};
-
 /** \brief DHCP Proxy set / unset vss request
     @param client_index - opaque cookie to identify the sender
     @param context - sender context, to match reply w/ request
@@ -159,6 +126,32 @@ define dhcp_compl_event
   u8 host_mac[6];
 };
 
+/** \brief Dump DHCP proxy table
+    @param client_index - opaque cookie to identify the sender
+    @param True for IPv6 proxy table
+*/
+define dhcp_proxy_dump
+{
+  u32 client_index;
+  u32 context;
+  u8  is_ip6;
+};
+
+/** \brief Tell client about a DHCP completion event
+    @param client_index - opaque cookie to identify the sender
+*/
+define dhcp_proxy_details
+{
+  u32 context;
+  u32 rx_vrf_id;
+  u32 server_vrf_id;
+  u32 vss_oui;
+  u32 vss_fib_id;
+  u8 is_ipv6;
+  u8 dhcp_server[16];
+  u8 dhcp_src_address[16];
+};
+
 /*
  * Local Variables:
  * eval: (c-set-style "gnu")
diff --git a/src/vnet/dhcp/dhcp_api.c b/src/vnet/dhcp/dhcp_api.c
index 88b32b24..ce9039b7 100644
--- a/src/vnet/dhcp/dhcp_api.c
+++ b/src/vnet/dhcp/dhcp_api.c
@@ -46,7 +46,8 @@
 
 #define foreach_vpe_api_msg                       \
 _(DHCP_PROXY_CONFIG,dhcp_proxy_config)            \
-_(DHCP_PROXY_CONFIG_2,dhcp_proxy_config_2)        \
+_(DHCP_PROXY_DUMP,dhcp_proxy_dump)                \
+_(DHCP_PROXY_DETAILS,dhcp_proxy_details)          \
 _(DHCP_PROXY_SET_VSS,dhcp_proxy_set_vss)          \
 _(DHCP_CLIENT_CONFIG, dhcp_client_config)
 
@@ -58,8 +59,8 @@ dhcpv4_proxy_config (vl_api_dhcp_proxy_config_t * mp)
 
   rv = dhcp_proxy_set_server ((ip4_address_t *) (&mp->dhcp_server),
 			      (ip4_address_t *) (&mp->dhcp_src_address),
-			      (u32) ntohl (mp->vrf_id),
-			      (int) mp->insert_circuit_id,
+			      (u32) ntohl (mp->rx_vrf_id),
+			      (u32) ntohl (mp->server_vrf_id),
 			      (int) (mp->is_add == 0));
 
   REPLY_MACRO (VL_API_DHCP_PROXY_CONFIG_REPLY);
@@ -74,44 +75,11 @@ dhcpv6_proxy_config (vl_api_dhcp_proxy_config_t * mp)
 
   rv = dhcpv6_proxy_set_server ((ip6_address_t *) (&mp->dhcp_server),
 				(ip6_address_t *) (&mp->dhcp_src_address),
-				(u32) ntohl (mp->vrf_id),
-				(int) mp->insert_circuit_id,
-				(int) (mp->is_add == 0));
-
-  REPLY_MACRO (VL_API_DHCP_PROXY_CONFIG_REPLY);
-}
-
-static void
-dhcpv4_proxy_config_2 (vl_api_dhcp_proxy_config_2_t * mp)
-{
-  vl_api_dhcp_proxy_config_reply_t *rmp;
-  int rv;
-
-  rv = dhcp_proxy_set_server_2 ((ip4_address_t *) (&mp->dhcp_server),
-				(ip4_address_t *) (&mp->dhcp_src_address),
 				(u32) ntohl (mp->rx_vrf_id),
 				(u32) ntohl (mp->server_vrf_id),
-				(int) mp->insert_circuit_id,
 				(int) (mp->is_add == 0));
 
-  REPLY_MACRO (VL_API_DHCP_PROXY_CONFIG_2_REPLY);
-}
-
-
-static void
-dhcpv6_proxy_config_2 (vl_api_dhcp_proxy_config_2_t * mp)
-{
-  vl_api_dhcp_proxy_config_reply_t *rmp;
-  int rv = -1;
-
-  rv = dhcpv6_proxy_set_server_2 ((ip6_address_t *) (&mp->dhcp_server),
-				  (ip6_address_t *) (&mp->dhcp_src_address),
-				  (u32) ntohl (mp->rx_vrf_id),
-				  (u32) ntohl (mp->server_vrf_id),
-				  (int) mp->insert_circuit_id,
-				  (int) (mp->is_add == 0));
-
-  REPLY_MACRO (VL_API_DHCP_PROXY_CONFIG_2_REPLY);
+  REPLY_MACRO (VL_API_DHCP_PROXY_CONFIG_REPLY);
 }
 
 
@@ -143,6 +111,67 @@ static void vl_api_dhcp_proxy_config_t_handler
     dhcpv6_proxy_config (mp);
 }
 
+static void
+vl_api_dhcp_proxy_dump_t_handler (vl_api_dhcp_proxy_dump_t * mp)
+{
+  unix_shared_memory_queue_t *q;
+
+  q = vl_api_client_index_to_input_queue (mp->client_index);
+  if (q == 0)
+    return;
+
+  if (mp->is_ip6 == 0)
+    dhcp_proxy_dump (q, mp->context);
+  else
+    dhcpv6_proxy_dump (q, mp->context);
+}
+
+void
+dhcp_send_details (void *opaque,
+		   u32 context,
+		   const ip46_address_t * server,
+		   const ip46_address_t * src,
+		   u32 server_fib_id,
+		   u32 rx_fib_id, u32 vss_fib_id, u32 vss_oui)
+{
+  vl_api_dhcp_proxy_details_t *mp;
+  unix_shared_memory_queue_t *q = opaque;
+
+  mp = vl_msg_api_alloc (sizeof (*mp));
+  if (!mp)
+    return;
+  memset (mp, 0, sizeof (*mp));
+  mp->_vl_msg_id = ntohs (VL_API_DHCP_PROXY_DETAILS);
+  mp->context = context;
+
+  mp->rx_vrf_id = htonl (rx_fib_id);
+  mp->server_vrf_id = htonl (server_fib_id);
+  mp->vss_oui = htonl (vss_oui);
+  mp->vss_fib_id = htonl (vss_fib_id);
+
+  mp->is_ipv6 = !ip46_address_is_ip4 (server);
+
+  if (mp->is_ipv6)
+    {
+      memcpy (mp->dhcp_server, server, 16);
+      memcpy (mp->dhcp_src_address, src, 16);
+    }
+  else
+    {
+      /* put the address in the first bytes */
+      memcpy (mp->dhcp_server, &server->ip4, 4);
+      memcpy (mp->dhcp_src_address, &src->ip4, 4);
+    }
+  vl_msg_api_send_shmem (q, (u8 *) & mp);
+}
+
+
+static void
+vl_api_dhcp_proxy_details_t_handler (vl_api_dhcp_proxy_details_t * mp)
+{
+  clib_warning ("BUG");
+}
+
 void
 dhcp_compl_event_callback (u32 client_index, u32 pid, u8 * hostname,
 			   u8 is_ipv6, u8 * host_address, u8 * router_address,
@@ -172,15 +201,6 @@ dhcp_compl_event_callback (u32 client_index, u32 pid, u8 * hostname,
   vl_msg_api_send_shmem (q, (u8 *) & mp);
 }
 
-static void vl_api_dhcp_proxy_config_2_t_handler
-  (vl_api_dhcp_proxy_config_2_t * mp)
-{
-  if (mp->is_ipv6 == 0)
-    dhcpv4_proxy_config_2 (mp);
-  else
-    dhcpv6_proxy_config_2 (mp);
-}
-
 static void vl_api_dhcp_client_config_t_handler
   (vl_api_dhcp_client_config_t * mp)
 {
diff --git a/src/vnet/dhcp/proxy.h b/src/vnet/dhcp/proxy.h
index e12c0d00..4b115c74 100644
--- a/src/vnet/dhcp/proxy.h
+++ b/src/vnet/dhcp/proxy.h
@@ -27,7 +27,6 @@
 #include <vnet/pg/pg.h>
 #include <vnet/ip/format.h>
 #include <vnet/ip/udp.h>
-#include <vnet/dhcp/client.h>
 
 typedef enum {
 #define dhcp_proxy_error(n,s) DHCP_PROXY_ERROR_##n,
@@ -49,9 +48,7 @@ typedef union {
 typedef struct {
   ip4_address_t dhcp_server;
   ip4_address_t dhcp_src_address;
-  u32 insert_option_82;
   u32 server_fib_index;
-  u32 valid;
 } dhcp_server_t;
 
 typedef struct {
@@ -64,29 +61,39 @@ typedef struct {
   /* to drop pkts in server-to-client direction */
   u32 error_drop_node_index;
 
-  vss_info *opt82vss;
+  vss_info *vss;
 
   /* hash lookup specific vrf_id -> option 82 vss suboption  */
-  uword * opt82vss_index_by_vrf_id;
+  u32 *vss_index_by_rx_fib_index;
 
   /* convenience */
-  dhcp_client_main_t * dhcp_client_main;
   vlib_main_t * vlib_main;
   vnet_main_t * vnet_main;
 } dhcp_proxy_main_t;
 
-dhcp_proxy_main_t dhcp_proxy_main;
+extern dhcp_proxy_main_t dhcp_proxy_main;
 
-int dhcp_proxy_set_server (ip4_address_t *addr, ip4_address_t *src_address,
-                           u32 fib_id, int insert_option_82, int is_del);
+void dhcp_send_details (void *opaque,
+                        u32 context,
+                        const ip46_address_t *server,
+                        const ip46_address_t *src,
+                        u32 server_fib_id,
+                        u32 rx_fib_id,
+                        u32 vss_fib_id,
+                        u32 vss_oui);
 
-int dhcp_proxy_set_server_2 (ip4_address_t *addr, ip4_address_t *src_address,
-                             u32 rx_fib_id,
-                             u32 server_fib_id, 
-                             int insert_option_82, int is_del);
+int dhcp_proxy_set_server (ip4_address_t *addr,
+                           ip4_address_t *src_address,
+                           u32 fib_id,
+                           u32 server_fib_id, 
+                           int is_del);
 
 int dhcp_proxy_set_option82_vss(u32 vrf_id,
                                 u32 oui,
                                 u32 fib_id, 
                                 int is_del);
+
+void dhcp_proxy_dump(void *opaque,
+                     u32 context);
+
 #endif /* included_dhcp_proxy_h */
diff --git a/src/vnet/dhcp/proxy_error.def b/src/vnet/dhcp/proxy_error.def
index 6aa06eb5..6d790d73 100644
--- a/src/vnet/dhcp/proxy_error.def
+++ b/src/vnet/dhcp/proxy_error.def
@@ -21,7 +21,8 @@ dhcp_proxy_error (RELAY_TO_SERVER, "DHCP packets relayed to the server")
 dhcp_proxy_error (RELAY_TO_CLIENT, "DHCP packets relayed to clients")
 dhcp_proxy_error (OPTION_82_ERROR, "DHCP failed to insert option 82")
 dhcp_proxy_error (NO_OPTION_82, "DHCP option 82 missing")
-dhcp_proxy_error (BAD_OPTION_82, "Bad DHCP option 82 value")
+dhcp_proxy_error (BAD_OPTION_82_ITF, "Bad DHCP option 82 interface value")
+dhcp_proxy_error (BAD_OPTION_82_ADDR, "Bad DHCP option 82 address value")
 dhcp_proxy_error (BAD_FIB_ID, "DHCP option 82 fib-id to fib-index map failure")
 dhcp_proxy_error (NO_INTERFACE_ADDRESS, "DHCP no interface address")
 dhcp_proxy_error (OPTION_82_VSS_NOT_PROCESSED, "DHCP VSS not processed by DHCP server")
diff --git a/src/vnet/dhcp/proxy_node.c b/src/vnet/dhcp/proxy_node.c
index 6a58fcdb..ab6819fe 100644
--- a/src/vnet/dhcp/proxy_node.c
+++ b/src/vnet/dhcp/proxy_node.c
@@ -18,6 +18,7 @@
 #include <vlib/vlib.h>
 #include <vnet/pg/pg.h>
 #include <vnet/dhcp/proxy.h>
+#include <vnet/dhcp/client.h>
 #include <vnet/fib/ip4_fib.h>
 
 static char * dhcp_proxy_error_strings[] = {
@@ -57,6 +58,8 @@ typedef struct {
 vlib_node_registration_t dhcp_proxy_to_server_node;
 vlib_node_registration_t dhcp_proxy_to_client_node;
 
+dhcp_proxy_main_t dhcp_proxy_main;
+
 u8 * format_dhcp_proxy_trace (u8 * s, va_list * args)
 {
   CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *);
@@ -94,6 +97,42 @@ u8 * format_dhcp_proxy_header_with_length (u8 * s, va_list * args)
   return s;
 }
 
+static inline vss_info *
+dhcp_get_vss_info (dhcp_proxy_main_t *dm,
+                   u32 rx_fib_index)
+{
+  vss_info *v;
+
+  if (vec_len(dm->vss_index_by_rx_fib_index) <= rx_fib_index ||
+      dm->vss_index_by_rx_fib_index[rx_fib_index] == ~0)
+  {
+      v = NULL;
+  }
+  else
+  {
+      v = pool_elt_at_index (dm->vss,
+                             dm->vss_index_by_rx_fib_index[rx_fib_index]);
+  }
+
+  return (v);
+}
+
+static inline dhcp_server_t *
+dhcp_get_server (dhcp_proxy_main_t *dm,
+                 u32 rx_fib_index)
+{
+  dhcp_server_t *s = NULL;
+
+  if (vec_len(dm->dhcp_server_index_by_rx_fib_index) > rx_fib_index &&
+      dm->dhcp_server_index_by_rx_fib_index[rx_fib_index] != ~0)
+  {
+      s = pool_elt_at_index (dm->dhcp_servers,
+                             dm->dhcp_server_index_by_rx_fib_index[rx_fib_index]);
+  }
+
+  return (s);
+}
+
 static uword
 dhcp_proxy_to_server_input (vlib_main_t * vm,
                             vlib_node_runtime_t * node,
@@ -131,9 +170,12 @@ dhcp_proxy_to_server_input (vlib_main_t * vm,
           u32 sw_if_index = 0;
           u32 original_sw_if_index = 0;
           u8  *end = NULL;
-          u32 fib_index, server_index;
+          u32 fib_index;
           dhcp_server_t * server;
           u32 rx_sw_if_index;
+          dhcp_option_t *o;
+          u32 len = 0;
+          vlib_buffer_free_list_t *fl;
 
 	  bi0 = from[0];
 	  to_next[0] = bi0;
@@ -166,26 +208,16 @@ dhcp_proxy_to_server_input (vlib_main_t * vm,
           rx_sw_if_index = vnet_buffer(b0)->sw_if_index[VLIB_RX];
 
           fib_index = im->fib_index_by_sw_if_index [rx_sw_if_index];
-
-          if (fib_index < vec_len(dpm->dhcp_server_index_by_rx_fib_index))
-            server_index = dpm->dhcp_server_index_by_rx_fib_index[fib_index];
-          else
-            server_index = 0;
+          server = dhcp_get_server(dpm, fib_index);
           
-          if (PREDICT_FALSE (pool_is_free_index (dpm->dhcp_servers, 
-                                                 server_index)))
+          if (PREDICT_FALSE (NULL == server))
             {
-            no_server:
               error0 = DHCP_PROXY_ERROR_NO_SERVER;
               next0 = DHCP_PROXY_TO_SERVER_INPUT_NEXT_DROP;
               pkts_no_server++;
               goto do_trace;
             }
           
-          server = pool_elt_at_index (dpm->dhcp_servers, server_index);
-          if (server->valid == 0)
-            goto no_server;
-
           vlib_buffer_advance (b0, -(sizeof(*ip0)));
           ip0 = vlib_buffer_get_current (b0);
 
@@ -216,142 +248,131 @@ dhcp_proxy_to_server_input (vlib_main_t * vm,
           h0->gateway_ip_address.as_u32 = server->dhcp_src_address.as_u32;
           pkts_to_server++;
 
-          if (server->insert_option_82) 
-            {
-              u32 fib_index, fib_id, opt82_fib_id=0, opt82_oui=0;
-	      ip4_fib_t * fib;
-              dhcp_option_t *o = (dhcp_option_t *) h0->options;
-              u32 len = 0;
-              vlib_buffer_free_list_t *fl;
+          o = (dhcp_option_t *) h0->options;
               
-              fib_index = im->fib_index_by_sw_if_index 
-                [vnet_buffer(b0)->sw_if_index[VLIB_RX]];
-	      fib = ip4_fib_get (fib_index);
-	      fib_id = fib->table_id;
-
-              end = b0->data + b0->current_data + b0->current_length;
-              /* TLVs are not performance-friendly... */
-              while  (o->option != 0xFF /* end of options */ && (u8 *)o < end) 
-                  o = (dhcp_option_t *) (((uword) o) + (o->length + 2));
-
-              fl = vlib_buffer_get_free_list (vm, b0->free_list_index);
-              // start write at (option*)o, some packets have padding
-              if (((u8 *)o - (u8 *)b0->data + VPP_DHCP_OPTION82_SIZE) > fl->n_data_bytes)
-                {
-                  next0 = DHCP_PROXY_TO_SERVER_INPUT_NEXT_DROP;
-                  pkts_too_big++;
-                  goto do_trace;
-                }
+          fib_index = im->fib_index_by_sw_if_index 
+              [vnet_buffer(b0)->sw_if_index[VLIB_RX]];
+
+          end = b0->data + b0->current_data + b0->current_length;
+          /* TLVs are not performance-friendly... */
+          while  (o->option != 0xFF /* end of options */ && (u8 *)o < end) 
+              o = (dhcp_option_t *) (((uword) o) + (o->length + 2));
 
-              if ((o->option == 0xFF)  && ((u8 *)o <= end))
-                {  
-                  vnet_main_t *vnm = vnet_get_main();   
-                  u16 old_l0, new_l0;
-                  ip4_address_t _ia0, * ia0 = &_ia0;
-                  uword  *p_vss;
-                  vss_info *vss;
-                  vnet_sw_interface_t *swif;
-                  sw_if_index = 0;
-                  original_sw_if_index = 0;
+          fl = vlib_buffer_get_free_list (vm, b0->free_list_index);
+          // start write at (option*)o, some packets have padding
+          if (((u8 *)o - (u8 *)b0->data + VPP_DHCP_OPTION82_SIZE) > fl->n_data_bytes)
+          {
+              next0 = DHCP_PROXY_TO_SERVER_INPUT_NEXT_DROP;
+              pkts_too_big++;
+              goto do_trace;
+          }
+
+          if ((o->option == 0xFF)  && ((u8 *)o <= end))
+          {  
+              vnet_main_t *vnm = vnet_get_main();   
+              u16 old_l0, new_l0;
+              ip4_address_t _ia0, * ia0 = &_ia0;
+              vss_info *vss;
+              vnet_sw_interface_t *swif;
+              sw_if_index = 0;
+              original_sw_if_index = 0;
                   
-                  original_sw_if_index = sw_if_index = 
-                      vnet_buffer(b0)->sw_if_index[VLIB_RX];
-                  swif = vnet_get_sw_interface (vnm, sw_if_index);
-                  if (swif->flags & VNET_SW_INTERFACE_FLAG_UNNUMBERED)
-                      sw_if_index = swif->unnumbered_sw_if_index;
+              original_sw_if_index = sw_if_index = 
+                  vnet_buffer(b0)->sw_if_index[VLIB_RX];
+              swif = vnet_get_sw_interface (vnm, sw_if_index);
+              if (swif->flags & VNET_SW_INTERFACE_FLAG_UNNUMBERED)
+                  sw_if_index = swif->unnumbered_sw_if_index;
                   
-                  p_vss = hash_get (dpm->opt82vss_index_by_vrf_id,
-                                    fib_id);
-                  if (p_vss) 
-                    {
-                      vss = pool_elt_at_index (dpm->opt82vss, p_vss[0]);
-                      opt82_oui =  vss->vpn_id.oui;
-                      opt82_fib_id =  vss->vpn_id.fib_id;
-                    }
-                  /* 
-                   * Get the first ip4 address on the [client-side] 
-                   * RX interface, if not unnumbered. otherwise use
-                   * the loopback interface's ip address.
-                   */
-                  ia0 = ip4_interface_first_address(&ip4_main, sw_if_index, 0);
+              /* 
+               * Get the first ip4 address on the [client-side] 
+               * RX interface, if not unnumbered. otherwise use
+               * the loopback interface's ip address.
+               */
+              ia0 = ip4_interface_first_address(&ip4_main, sw_if_index, 0);
                   
-                  if (ia0 == 0)
-                    {
-                      error0 = DHCP_PROXY_ERROR_NO_INTERFACE_ADDRESS;
-                      next0 = DHCP_PROXY_TO_SERVER_INPUT_NEXT_DROP;
-                      pkts_no_interface_address++;
-                      goto do_trace;
-                    }
-
-                  /* Add option 82 */
-                  o->option = 82;   /* option 82 */
-                  o->length = 12;   /* 12 octets to follow */
-                  o->data[0] = 1;   /* suboption 1, circuit ID (=FIB id) */
-                  o->data[1] = 4;   /* length of suboption */
-                  o->data[2] = (original_sw_if_index >> 24) & 0xFF;
-                  o->data[3] = (original_sw_if_index >> 16) & 0xFF;
-                  o->data[4] = (original_sw_if_index >> 8)  & 0xFF;
-                  o->data[5] = (original_sw_if_index >> 0)  & 0xFF;
-		  o->data[6] = 5; /* suboption 5 (client RX intfc address) */
-		  o->data[7] = 4; /* length 4 */
-		  o->data[8] = ia0->as_u8[0];
-		  o->data[9] = ia0->as_u8[1];
-		  o->data[10] = ia0->as_u8[2];
-		  o->data[11] = ia0->as_u8[3];
-                  o->data[12] = 0xFF;
-                  if (opt82_oui !=0 || opt82_fib_id != 0)
-                    {
-                      o->data[12] = 151; /* vss suboption */
-                      if (255 == opt82_fib_id) {
-                          o->data[13] = 1;   /* length */
-                          o->data[14] = 255;   /* vss option type */
-                          o->data[15] = 152; /* vss control suboption */
-                          o->data[16] = 0;   /* length */
-                          /* and a new "end-of-options" option (0xff) */
-                          o->data[17] = 0xFF;
-                          o->length += 5;
-                      } else {
-                          o->data[13] = 8;   /* length */
-                          o->data[14] = 1;   /* vss option type */
-                          o->data[15] = (opt82_oui >> 16) & 0xff;
-                          o->data[16] = (opt82_oui >> 8) & 0xff;
-                          o->data[17] = (opt82_oui ) & 0xff;
-                          o->data[18] = (opt82_fib_id >> 24) & 0xff;
-                          o->data[19] = (opt82_fib_id >> 16) & 0xff;
-                          o->data[20] = (opt82_fib_id >> 8) & 0xff;
-                          o->data[21] = (opt82_fib_id) & 0xff;
-                          o->data[22] = 152; /* vss control suboption */
-                          o->data[23] = 0;   /* length */
+              if (ia0 == 0)
+              {
+                  error0 = DHCP_PROXY_ERROR_NO_INTERFACE_ADDRESS;
+                  next0 = DHCP_PROXY_TO_SERVER_INPUT_NEXT_DROP;
+                  pkts_no_interface_address++;
+                  goto do_trace;
+              }
+
+              /* Add option 82 */
+              o->option = 82;   /* option 82 */
+              o->length = 12;   /* 12 octets to follow */
+              o->data[0] = 1;   /* suboption 1, circuit ID (=FIB id) */
+              o->data[1] = 4;   /* length of suboption */
+              o->data[2] = (original_sw_if_index >> 24) & 0xFF;
+              o->data[3] = (original_sw_if_index >> 16) & 0xFF;
+              o->data[4] = (original_sw_if_index >> 8)  & 0xFF;
+              o->data[5] = (original_sw_if_index >> 0)  & 0xFF;
+              o->data[6] = 5; /* suboption 5 (client RX intfc address) */
+              o->data[7] = 4; /* length 4 */
+              o->data[8] = ia0->as_u8[0];
+              o->data[9] = ia0->as_u8[1];
+              o->data[10] = ia0->as_u8[2];
+              o->data[11] = ia0->as_u8[3];
+              o->data[12] = 0xFF;
+
+              vss = dhcp_get_vss_info (dpm, fib_index);
+              if (NULL != vss)
+              {
+                  u32 opt82_fib_id=0, opt82_oui=0;
+
+                  opt82_oui =  vss->vpn_id.oui;
+                  opt82_fib_id =  vss->vpn_id.fib_id;
+
+                  o->data[12] = 151; /* vss suboption */
+                  if (255 == opt82_fib_id) {
+                      o->data[13] = 1;   /* length */
+                      o->data[14] = 255;   /* vss option type */
+                      o->data[15] = 152; /* vss control suboption */
+                      o->data[16] = 0;   /* length */
+                      /* and a new "end-of-options" option (0xff) */
+                      o->data[17] = 0xFF;
+                      o->length += 5;
+                  } else {
+                      o->data[13] = 8;   /* length */
+                      o->data[14] = 1;   /* vss option type */
+                      o->data[15] = (opt82_oui >> 16) & 0xff;
+                      o->data[16] = (opt82_oui >> 8) & 0xff;
+                      o->data[17] = (opt82_oui ) & 0xff;
+                      o->data[18] = (opt82_fib_id >> 24) & 0xff;
+                      o->data[19] = (opt82_fib_id >> 16) & 0xff;
+                      o->data[20] = (opt82_fib_id >> 8) & 0xff;
+                      o->data[21] = (opt82_fib_id) & 0xff;
+                      o->data[22] = 152; /* vss control suboption */
+                      o->data[23] = 0;   /* length */
                           
-                          /* and a new "end-of-options" option (0xff) */
-                          o->data[24] = 0xFF;
-                          o->length += 12;
-                      }
+                      /* and a new "end-of-options" option (0xff) */
+                      o->data[24] = 0xFF;
+                      o->length += 12;
                   }
-
-                  len = o->length + 3;
-                  b0->current_length += len;
-                  /* Fix IP header length and checksum */
-                  old_l0 = ip0->length;
-                  new_l0 = clib_net_to_host_u16 (old_l0);
-                  new_l0 += len;
-                  new_l0 = clib_host_to_net_u16 (new_l0);
-                  ip0->length = new_l0;
-                  sum0 = ip0->checksum;
-                  sum0 = ip_csum_update (sum0, old_l0, new_l0, ip4_header_t,
-                                         length /* changed member */);
-                  ip0->checksum = ip_csum_fold (sum0);
-
-                  /* Fix UDP length */
-                  new_l0 = clib_net_to_host_u16 (u0->length);
-                  new_l0 += len;
-                  u0->length = clib_host_to_net_u16 (new_l0);
-                } else {
-                  vlib_node_increment_counter 
-                      (vm, dhcp_proxy_to_server_node.index,
-                       DHCP_PROXY_ERROR_OPTION_82_ERROR, 1);
-                }
-            }
+              }
+
+              len = o->length + 3;
+              b0->current_length += len;
+              /* Fix IP header length and checksum */
+              old_l0 = ip0->length;
+              new_l0 = clib_net_to_host_u16 (old_l0);
+              new_l0 += len;
+              new_l0 = clib_host_to_net_u16 (new_l0);
+              ip0->length = new_l0;
+              sum0 = ip0->checksum;
+              sum0 = ip_csum_update (sum0, old_l0, new_l0, ip4_header_t,
+                                     length /* changed member */);
+              ip0->checksum = ip_csum_fold (sum0);
+
+              /* Fix UDP length */
+              new_l0 = clib_net_to_host_u16 (u0->length);
+              new_l0 += len;
+              u0->length = clib_host_to_net_u16 (new_l0);
+          } else {
+              vlib_node_increment_counter 
+                  (vm, dhcp_proxy_to_server_node.index,
+                   DHCP_PROXY_ERROR_OPTION_82_ERROR, 1);
+          }
           
           next0 = DHCP_PROXY_TO_SERVER_INPUT_NEXT_LOOKUP;
 
@@ -451,11 +472,13 @@ dhcp_proxy_to_client_input (vlib_main_t * vm,
       vnet_sw_interface_t *si0;
       u32 error0 = (u32)~0;
       vnet_sw_interface_t *swif;
-      u32 server_index;
       u32 fib_index;
       dhcp_server_t * server;
       u32 original_sw_if_index = (u32) ~0;
-          
+      ip4_address_t relay_addr = {
+          .as_u32 = 0,
+      };
+
       bi0 = from[0];
       from += 1;
       n_left_from -= 1;
@@ -501,13 +524,21 @@ dhcp_proxy_to_client_input (vlib_main_t * vm,
                            and the sw_if_index */
                         if (sub->option == 1 && sub->length == 4)
                           {
-                            sw_if_index = (o->data[2] << 24)
-                                | (o->data[3] << 16)
-                                | (o->data[4] << 8)
-                                | (o->data[5]);
-                          } else if (sub->option == 151 &&
-                                     sub->length == 7 &&
-                                     sub->data[0] == 1)
+                            sw_if_index = ((sub->data[0] << 24) |
+                                           (sub->data[1] << 16) |
+                                           (sub->data[2] << 8)  |
+                                           (sub->data[3]));
+                          }
+                        else if (sub->option == 5 && sub->length == 4)
+                          {
+                              relay_addr.as_u8[0] = sub->data[0];
+                              relay_addr.as_u8[1] = sub->data[1];
+                              relay_addr.as_u8[2] = sub->data[2];
+                              relay_addr.as_u8[3] = sub->data[3];
+                          }
+                        else if (sub->option == 151 &&
+                                 sub->length == 7 &&
+                                 sub->data[0] == 1)
                             vss_exist = 1;
                         else if (sub->option == 152 && sub->length == 0)
                             vss_ctrl = 1;
@@ -539,34 +570,27 @@ dhcp_proxy_to_client_input (vlib_main_t * vm,
           goto do_trace;
         }
       
+      if (relay_addr.as_u32 == 0)
+        {
+          error0 = DHCP_PROXY_ERROR_BAD_OPTION_82_ADDR;
+          goto drop_packet;
+        }
 
       if (sw_if_index >= vec_len (im->fib_index_by_sw_if_index))
         {
-          error0 = DHCP_PROXY_ERROR_BAD_OPTION_82;
+          error0 = DHCP_PROXY_ERROR_BAD_OPTION_82_ITF;
           goto drop_packet;
         }
 
       fib_index = im->fib_index_by_sw_if_index [sw_if_index];
+      server = dhcp_get_server(dpm, fib_index);
 
-      if (fib_index < vec_len(dpm->dhcp_server_index_by_rx_fib_index))
-        server_index = dpm->dhcp_server_index_by_rx_fib_index[fib_index];
-      else
-        server_index = 0;
-
-      if (PREDICT_FALSE (pool_is_free_index (dpm->dhcp_servers, 
-                                             server_index)))
-        {
-          error0 = DHCP_PROXY_ERROR_BAD_OPTION_82;
-          goto drop_packet;
-        }
-      
-      server = pool_elt_at_index (dpm->dhcp_servers, server_index);
-      if (server->valid == 0)
+      if (PREDICT_FALSE (NULL == server))
         {
           error0 = DHCP_PROXY_ERROR_NO_SERVER;
           goto drop_packet;
         }
-
+      
       if (ip0->src_address.as_u32 != server->dhcp_server.as_u32)
         {             
           error0 = DHCP_PROXY_ERROR_BAD_SVR_FIB_OR_ADDRESS;
@@ -587,6 +611,12 @@ dhcp_proxy_to_client_input (vlib_main_t * vm,
           goto drop_packet;
         }
 
+      if (relay_addr.as_u32 != ia0->as_u32)
+        {             
+          error0 = DHCP_PROXY_ERROR_BAD_YIADDR;
+          goto drop_packet;
+        }
+
       u0->checksum = 0;
       u0->dst_port = clib_net_to_host_u16 (UDP_DST_PORT_dhcp_to_client);
       sum0 = ip0->checksum;
@@ -677,7 +707,7 @@ clib_error_t * dhcp_proxy_init (vlib_main_t * vm)
   error_drop_node = vlib_get_node_by_name (vm, (u8 *) "error-drop");
   dm->error_drop_node_index = error_drop_node->index;
 
-  dm->opt82vss_index_by_vrf_id = hash_create (0, sizeof (uword));
+  dm->vss_index_by_rx_fib_index = NULL;
 
   udp_register_dst_port (vm, UDP_DST_PORT_dhcp_to_client, 
                          dhcp_proxy_to_client_node.index, 1 /* is_ip4 */);
@@ -694,15 +724,17 @@ clib_error_t * dhcp_proxy_init (vlib_main_t * vm)
 
 VLIB_INIT_FUNCTION (dhcp_proxy_init);
 
-int dhcp_proxy_set_server_2 (ip4_address_t *addr, ip4_address_t *src_address,
-                             u32 rx_fib_id,
-                             u32 server_fib_id, 
-                             int insert_option_82, int is_del)
+int dhcp_proxy_set_server (ip4_address_t *addr,
+                           ip4_address_t *src_address,
+                           u32 rx_fib_id,
+                           u32 server_fib_id, 
+                           int is_del)
 {
   dhcp_proxy_main_t * dpm = &dhcp_proxy_main;
   dhcp_server_t * server = 0;
   u32 server_index = 0;
   u32 rx_fib_index = 0;
+
   const fib_prefix_t all_1s =
   {
       .fp_len = 32,
@@ -719,97 +751,68 @@ int dhcp_proxy_set_server_2 (ip4_address_t *addr, ip4_address_t *src_address,
   rx_fib_index = fib_table_find_or_create_and_lock(FIB_PROTOCOL_IP4,
                                                    rx_fib_id);
 
-  if (rx_fib_id == 0)
-    {
-      server = pool_elt_at_index (dpm->dhcp_servers, 0);
-      
-      if (is_del)
-        {
-          memset (server, 0, sizeof (*server));
-          fib_table_entry_special_remove(rx_fib_index,
-                                         &all_1s,
-                                         FIB_SOURCE_DHCP);
-         return 0;
-        }
-      if (!server->valid)
-          fib_table_entry_special_add(rx_fib_index,
-                                      &all_1s,
-                                      FIB_SOURCE_DHCP,
-                                      FIB_ENTRY_FLAG_LOCAL,
-                                      ADJ_INDEX_INVALID);
-
-      goto initialize_it;
-    }
-
   if (is_del)
     {
       if (rx_fib_index >= vec_len(dpm->dhcp_server_index_by_rx_fib_index))
         return VNET_API_ERROR_NO_SUCH_ENTRY;
       
       server_index = dpm->dhcp_server_index_by_rx_fib_index[rx_fib_index];
-      ASSERT(server_index > 0);
+
+      if (server_index == ~0)
+        return VNET_API_ERROR_NO_SUCH_ENTRY;
 
       /* Use the default server again.  */
-      dpm->dhcp_server_index_by_rx_fib_index[rx_fib_index] = 0;
+      dpm->dhcp_server_index_by_rx_fib_index[rx_fib_index] = ~0;
       server = pool_elt_at_index (dpm->dhcp_servers, server_index);
-      memset (server, 0, sizeof (*server));
-      pool_put (dpm->dhcp_servers, server);
 
       fib_table_entry_special_remove(rx_fib_index,
                                      &all_1s,
                                      FIB_SOURCE_DHCP);
+      fib_table_unlock (rx_fib_index,
+                        FIB_PROTOCOL_IP4);
+      fib_table_unlock (server->server_fib_index,
+                        FIB_PROTOCOL_IP4);
 
+      memset (server, 0, sizeof (*server));
+      pool_put (dpm->dhcp_servers, server);
       return 0;
     }
-
-  if (rx_fib_index < vec_len(dpm->dhcp_server_index_by_rx_fib_index))
-    {
-      server_index = dpm->dhcp_server_index_by_rx_fib_index[rx_fib_index];
-      if (server_index != 0)
-        {
-          server = pool_elt_at_index (dpm->dhcp_servers, server_index);
-          goto initialize_it;
-        }
-    }
-
-  pool_get (dpm->dhcp_servers, server);
-
-  fib_table_entry_special_add(rx_fib_index,
-                              &all_1s,
-                              FIB_SOURCE_DHCP,
-                              FIB_ENTRY_FLAG_LOCAL,
-                              ADJ_INDEX_INVALID);
-  
- initialize_it:
-
-
-  server->dhcp_server.as_u32 = addr->as_u32;
-  server->server_fib_index = 
-      fib_table_find_or_create_and_lock(FIB_PROTOCOL_IP4,
-	  				server_fib_id);
-  server->dhcp_src_address.as_u32 = src_address->as_u32;
-  server->insert_option_82 = insert_option_82;
-  server->valid = 1;
-  if (rx_fib_index)
-    {
-      vec_validate (dpm->dhcp_server_index_by_rx_fib_index, rx_fib_index);
+  else
+  {
+      vec_validate_init_empty(dpm->dhcp_server_index_by_rx_fib_index,
+                              rx_fib_index,
+                              ~0);
+
+      pool_get (dpm->dhcp_servers, server);
+
+      server->dhcp_server.as_u32 = addr->as_u32;
+      server->dhcp_src_address.as_u32 = src_address->as_u32;
+
+      fib_table_entry_special_add(rx_fib_index,
+                                  &all_1s,
+                                  FIB_SOURCE_DHCP,
+                                  FIB_ENTRY_FLAG_LOCAL,
+                                  ADJ_INDEX_INVALID);
+      fib_table_lock (rx_fib_index,
+                      FIB_PROTOCOL_IP4);
+
+      server->server_fib_index = 
+          fib_table_find_or_create_and_lock(FIB_PROTOCOL_IP4,
+                                            server_fib_id);
+
+      vec_validate_init_empty (dpm->dhcp_server_index_by_rx_fib_index,
+                               rx_fib_index,
+                               ~0);
       dpm->dhcp_server_index_by_rx_fib_index[rx_fib_index] = 
-        server - dpm->dhcp_servers;
-    }
+          server - dpm->dhcp_servers;
+  }
 
-  return 0;
-}
+  fib_table_unlock (rx_fib_index,
+                    FIB_PROTOCOL_IP4);
 
-/* Old API, manipulates the default server (only) */
-int dhcp_proxy_set_server (ip4_address_t *addr, ip4_address_t *src_address,
-                           u32 fib_id, int insert_option_82, int is_del)
-{
-  return dhcp_proxy_set_server_2 (addr, src_address, 0 /* rx_fib_id */,
-                                  fib_id /* server_fib_id */, 
-                                  insert_option_82, is_del);
+  return 0;
 }
 
-
 static clib_error_t *
 dhcp_proxy_set_command_fn (vlib_main_t * vm,
                            unformat_input_t * input,
@@ -818,7 +821,6 @@ dhcp_proxy_set_command_fn (vlib_main_t * vm,
   ip4_address_t server_addr, src_addr;
   u32 server_fib_id = 0, rx_fib_id = 0;
   int is_del = 0;
-  int add_option_82 = 0;
   int set_src = 0, set_server = 0;
   
   while (unformat_check_input(input) != UNFORMAT_END_OF_INPUT) 
@@ -833,9 +835,6 @@ dhcp_proxy_set_command_fn (vlib_main_t * vm,
       else if (unformat(input, "src-address %U", 
 			unformat_ip4_address, &src_addr))
         set_src = 1;
-      else if (unformat (input, "add-option-82")
-               || unformat (input, "insert-option-82"))
-        add_option_82 = 1;
       else if (unformat (input, "delete") ||
                unformat (input, "del"))
         is_del = 1;
@@ -847,8 +846,8 @@ dhcp_proxy_set_command_fn (vlib_main_t * vm,
     {
       int rv;
 
-      rv = dhcp_proxy_set_server_2 (&server_addr, &src_addr, rx_fib_id, 
-                                    server_fib_id, add_option_82, is_del);
+      rv = dhcp_proxy_set_server (&server_addr, &src_addr, rx_fib_id, 
+                                  server_fib_id, is_del);
       switch (rv)
         {
         case 0:
@@ -882,7 +881,7 @@ dhcp_proxy_set_command_fn (vlib_main_t * vm,
 
 VLIB_CLI_COMMAND (dhcp_proxy_set_command, static) = {
   .path = "set dhcp proxy",
-  .short_help = "set dhcp proxy [del] server <ip-addr> src-address <ip-addr> [add-option-82] [server-fib-id <n>] [rx-fib-id <n>]",
+  .short_help = "set dhcp proxy [del] server <ip-addr> src-address <ip-addr> [server-fib-id <n>] [rx-fib-id <n>]",
   .function = dhcp_proxy_set_command_fn,
 };
 
@@ -896,8 +895,8 @@ u8 * format_dhcp_proxy_server (u8 * s, va_list * args)
 
   if (dm == 0)
     {
-      s = format (s, "%=16s%=16s%=14s%=14s%=20s", "Server", "Src Address", 
-                  "Server FIB", "RX FIB", "Insert Option 82");
+      s = format (s, "%=16s%=16s%=14s%=14s", "Server", "Src Address", 
+                  "Server FIB", "RX FIB");
       return s;
     }
 
@@ -911,11 +910,10 @@ u8 * format_dhcp_proxy_server (u8 * s, va_list * args)
   if (rx_fib)
     rx_fib_id = rx_fib->table_id;
 
-  s = format (s, "%=16U%=16U%=14u%=14u%=20s",
+  s = format (s, "%=16U%=16U%=14u%=14u",
               format_ip4_address, &server->dhcp_server,
               format_ip4_address, &server->dhcp_src_address,
-              server_fib_id, rx_fib_id,
-              server->insert_option_82 ? "yes" : "no");
+              server_fib_id, rx_fib_id);
   return s;
 }
 
@@ -925,24 +923,22 @@ dhcp_proxy_show_command_fn (vlib_main_t * vm,
                             vlib_cli_command_t * cmd)
 {
   dhcp_proxy_main_t * dpm = &dhcp_proxy_main;
-  ip4_main_t * im = &ip4_main;
   dhcp_server_t * server;
-  u32 server_index;
-  int i;
+  u32 server_index, i;
 
   vlib_cli_output (vm, "%U", format_dhcp_proxy_server, 0 /* header line */,
                    0, 0);
 
-  for (i = 0; i < vec_len (im->fibs); i++)
-    {
-      if (i < vec_len(dpm->dhcp_server_index_by_rx_fib_index))
-        server_index = dpm->dhcp_server_index_by_rx_fib_index[i];
-      else
-        server_index = 0;
+  vec_foreach_index (i, dpm->dhcp_server_index_by_rx_fib_index)
+  {
+      server_index = dpm->dhcp_server_index_by_rx_fib_index[i];
+      if (~0 == server_index)
+          continue;
+
       server = pool_elt_at_index (dpm->dhcp_servers, server_index);
-      if (server->valid)
-        vlib_cli_output (vm, "%U", format_dhcp_proxy_server, dpm, 
-                         server, i);
+
+      vlib_cli_output (vm, "%U", format_dhcp_proxy_server, dpm, 
+                       server, i);
     }
 
   return 0;
@@ -954,50 +950,104 @@ VLIB_CLI_COMMAND (dhcp_proxy_show_command, static) = {
   .function = dhcp_proxy_show_command_fn,
 };
 
+void
+dhcp_proxy_dump (void *opaque,
+                 u32 context)
+{
+  dhcp_proxy_main_t * dpm = &dhcp_proxy_main;
+  ip4_fib_t *s_fib, *r_fib;
+  dhcp_server_t * server;
+  u32 server_index, i;
+  vss_info *v;
+
+  vec_foreach_index (i, dpm->dhcp_server_index_by_rx_fib_index)
+  {
+      server_index = dpm->dhcp_server_index_by_rx_fib_index[i];
+      if (~0 == server_index)
+          continue;
+
+      server = pool_elt_at_index (dpm->dhcp_servers, server_index);
+      v = dhcp_get_vss_info(dpm, i);
+
+      ip46_address_t src_addr = {
+          .ip4 = server->dhcp_src_address,
+      };
+      ip46_address_t server_addr = {
+          .ip4 = server->dhcp_server,
+      };
+
+      s_fib = ip4_fib_get(server->server_fib_index);
+      r_fib = ip4_fib_get(i);
+
+      dhcp_send_details(opaque,
+                        context,
+                        &server_addr,
+                        &src_addr,
+                        s_fib->table_id,
+                        r_fib->table_id,
+                        (v ? v->vpn_id.fib_id : 0),
+                        (v ? v->vpn_id.oui : 0));
+  }
+}
 
-int dhcp_proxy_set_option82_vss(  u32 vrf_id,
-                                  u32 oui,
-                                  u32 fib_id, 
-                                  int is_del)
+int dhcp_proxy_set_option82_vss(u32 tbl_id,
+                                u32 oui,
+                                u32 fib_id, 
+                                int is_del)
 {
   dhcp_proxy_main_t *dm = &dhcp_proxy_main;
-  uword *p;
-  vss_info *a;
-  u32 old_oui=0, old_fib_id=0;
+  vss_info *v = NULL;
+  u32  rx_fib_index;
+  int rc = 0;
   
-  p = hash_get (dm->opt82vss_index_by_vrf_id, vrf_id);
+  rx_fib_index = ip4_fib_table_find_or_create_and_lock(tbl_id);
+  v = dhcp_get_vss_info(dm, rx_fib_index);
 
-  if (p) 
-    {
-      a = pool_elt_at_index (dm->opt82vss, p[0]);
-      if (!a) 
-        return VNET_API_ERROR_NO_SUCH_FIB;
-      old_oui = a->vpn_id.oui;
-      old_fib_id = a->vpn_id.fib_id;
-          
+  if (NULL != v)
+  {
       if (is_del)
-        {
-          if (old_oui == oui &&
-              old_fib_id == fib_id)
-            {
-              pool_put(dm->opt82vss, a);
-              hash_unset (dm->opt82vss_index_by_vrf_id, vrf_id);
-              return 0;
-            }
-          else
-            return VNET_API_ERROR_NO_SUCH_ENTRY;
-        }
-      pool_put(dm->opt82vss, a);
-      hash_unset (dm->opt82vss_index_by_vrf_id, vrf_id);
-  } else if (is_del)
-    return VNET_API_ERROR_NO_SUCH_ENTRY;
-  pool_get (dm->opt82vss, a);
-  memset (a, ~0, sizeof (a[0]));
-  a->vpn_id.oui = oui;
-  a->vpn_id.fib_id = fib_id;
-  hash_set (dm->opt82vss_index_by_vrf_id, vrf_id, a - dm->opt82vss);
+      {
+          /* release the lock held on the table when the VSS
+           * info was created */
+          fib_table_unlock (rx_fib_index,
+                            FIB_PROTOCOL_IP4);
+
+          pool_put (dm->vss, v);
+          dm->vss_index_by_rx_fib_index[rx_fib_index] = ~0;
+      }
+      else
+      {
+          /* this is a modify */
+          v->vpn_id.fib_id = fib_id;
+          v->vpn_id.oui = oui;
+      }
+  }
+  else
+  {
+      if (is_del)
+          rc = VNET_API_ERROR_NO_SUCH_ENTRY;
+      else
+      {
+          /* create a new entry */
+          vec_validate_init_empty(dm->vss_index_by_rx_fib_index,
+                                  rx_fib_index, ~0);
+
+          /* hold a lock on the table whilst the VSS info exist */
+          fib_table_lock (rx_fib_index,
+                          FIB_PROTOCOL_IP4);
+
+          pool_get (dm->vss, v);
+          v->vpn_id.fib_id = fib_id;
+          v->vpn_id.oui = oui;
+          dm->vss_index_by_rx_fib_index[rx_fib_index] = v - dm->vss;
+      }
+  }
+
+  /* Release the lock taken during the create_or_lock at the start */
+  fib_table_unlock (rx_fib_index,
+                    FIB_PROTOCOL_IP4);
   
-  return 0;
+  return (rc);
 }
 
 static clib_error_t *
@@ -1065,20 +1115,20 @@ dhcp_vss_show_command_fn (vlib_main_t * vm,
   
 {
   dhcp_proxy_main_t * dm = &dhcp_proxy_main;
+  ip4_fib_t *fib;
+  u32 *fib_index;
   vss_info *v;
-  u32 oui;
-  u32 fib_id;
-  u32 tbl_id;
-  uword index;
   
   vlib_cli_output (vm, "%=9s%=11s%=12s","Table", "OUI", "VPN-ID");
-  hash_foreach (tbl_id, index, dm->opt82vss_index_by_vrf_id,
+  pool_foreach (fib_index, dm->vss_index_by_rx_fib_index,
   ({
-     v = pool_elt_at_index (dm->opt82vss, index);
-     oui = v->vpn_id.oui;
-     fib_id = v->vpn_id.fib_id;
-     vlib_cli_output (vm, "%=9d 0x%08x%=12d",
-                      tbl_id, oui, fib_id);
+      fib = ip4_fib_get (*fib_index);
+      v = pool_elt_at_index (dm->vss, *fib_index);
+
+      vlib_cli_output (vm, "%=6d%=6d%=12d",
+                       fib->table_id,
+                       v->vpn_id.oui,
+                       v->vpn_id.fib_id);
   }));
   
   return 0;
diff --git a/src/vnet/dhcpv6/proxy.h b/src/vnet/dhcpv6/proxy.h
index 9e18913a..77ced361 100644
--- a/src/vnet/dhcpv6/proxy.h
+++ b/src/vnet/dhcpv6/proxy.h
@@ -48,9 +48,7 @@ typedef union {
 typedef struct {
   ip6_address_t dhcp6_server;
   ip6_address_t dhcp6_src_address;
-  u32 insert_vss;
   u32 server_fib6_index;
-  u32 valid;
 } dhcpv6_server_t;
 
 typedef struct {
@@ -70,7 +68,7 @@ typedef struct {
   dhcpv6_vss_info *vss;
 
   /* hash lookup specific vrf_id -> VSS vector index*/
-  uword  *vss_index_by_vrf_id;
+  u32 *vss_index_by_rx_fib_index;
    
   /* convenience */
   vlib_main_t * vlib_main;
@@ -79,17 +77,18 @@ typedef struct {
 
 dhcpv6_proxy_main_t dhcpv6_proxy_main;
 
-int dhcpv6_proxy_set_server (ip6_address_t *addr, ip6_address_t *src_address,
-                             u32 fib_id, int insert_vss, int is_del);
-
 int dhcpv6_proxy_set_vss(u32 tbl_id,
                          u32 oui,
                          u32 fib_id, 
                          int is_del);
 
-int dhcpv6_proxy_set_server_2 (ip6_address_t *addr, ip6_address_t *src_address,
-                             u32 rx_fib_id,
-                             u32 server_fib_id,
-                             int insert_vss, int is_del);
+int dhcpv6_proxy_set_server(ip6_address_t *addr,
+                            ip6_address_t *src_address,
+                            u32 rx_fib_id,
+                            u32 server_fib_id,
+                            int is_del);
+
+void dhcpv6_proxy_dump(void *opaque,
+                       u32 context);
 
 #endif /* included_dhcpv6_proxy_h */
diff --git a/src/vnet/dhcpv6/proxy_node.c b/src/vnet/dhcpv6/proxy_node.c
index 4137624c..f40798e6 100644
--- a/src/vnet/dhcpv6/proxy_node.c
+++ b/src/vnet/dhcpv6/proxy_node.c
@@ -18,6 +18,7 @@
 #include <vlib/vlib.h>
 #include <vnet/pg/pg.h>
 #include <vnet/dhcpv6/proxy.h>
+#include <vnet/dhcp/proxy.h>
 #include <vnet/fib/ip6_fib.h>
 #include <vnet/mfib/mfib_table.h>
 #include <vnet/mfib/ip6_mfib.h>
@@ -117,6 +118,42 @@ static inline void copy_ip6_address (ip6_address_t *dst, ip6_address_t *src)
   dst->as_u64[1] = src->as_u64[1];
 } 
 
+static inline dhcpv6_vss_info *
+dhcpv6_get_vss_info (dhcpv6_proxy_main_t *dm,
+                     u32 rx_fib_index)
+{
+  dhcpv6_vss_info *v;
+
+  if (vec_len(dm->vss_index_by_rx_fib_index) <= rx_fib_index ||
+      dm->vss_index_by_rx_fib_index[rx_fib_index] == ~0)
+  {
+      v = NULL;
+  }
+  else
+  {
+      v = pool_elt_at_index (dm->vss,
+                             dm->vss_index_by_rx_fib_index[rx_fib_index]);
+  }
+
+  return (v);
+}
+
+static inline dhcpv6_server_t *
+dhcpv6_get_server (dhcpv6_proxy_main_t *dm,
+                   u32 rx_fib_index)
+{
+  dhcpv6_server_t *s = NULL;
+
+  if (vec_len(dm->dhcp6_server_index_by_rx_fib_index) > rx_fib_index &&
+      dm->dhcp6_server_index_by_rx_fib_index[rx_fib_index] != ~0)
+  {
+      s = pool_elt_at_index (dm->dhcp6_servers,
+                             dm->dhcp6_server_index_by_rx_fib_index[rx_fib_index]);
+  }
+
+  return (s);
+}
+
 static uword
 dhcpv6_proxy_to_server_input (vlib_main_t * vm,
                             vlib_node_runtime_t * node,
@@ -132,13 +169,10 @@ dhcpv6_proxy_to_server_input (vlib_main_t * vm,
   u32 pkts_wrong_msg_type=0;
   u32 pkts_too_big=0;
   ip6_main_t * im = &ip6_main;
-  ip6_fib_t * fib;
   ip6_address_t * src;
   int bogus_length;
   dhcpv6_server_t * server;
   u32  rx_fib_idx = 0, server_fib_idx = 0;
-  u32 server_idx;
-  u32 fib_id1 = 0;
 
   next_index = node->cached_next_index;
 
@@ -172,12 +206,8 @@ dhcpv6_proxy_to_server_input (vlib_main_t * vm,
           ethernet_header_t * e_h0;
           u8 client_src_mac[6];
           vlib_buffer_free_list_t *fl;
-
-          uword *p_vss;
-          u32  oui1=0;
           dhcpv6_vss_info *vss;
 
-
 	  bi0 = from[0];
 	  to_next[0] = bi0;
 	  from += 1;
@@ -228,25 +258,15 @@ dhcpv6_proxy_to_server_input (vlib_main_t * vm,
           /* Send to DHCPV6 server via the configured FIB */
           rx_sw_if_index = sw_if_index =  vnet_buffer(b0)->sw_if_index[VLIB_RX];
           rx_fib_idx = im->fib_index_by_sw_if_index [rx_sw_if_index];
+          server = dhcpv6_get_server(dpm, rx_fib_idx);
 
-	  if (vec_len(dpm->dhcp6_server_index_by_rx_fib_index) <= rx_fib_idx)
-	    goto no_server;
-
-	  server_idx = dpm->dhcp6_server_index_by_rx_fib_index[rx_fib_idx];
-
-          if (PREDICT_FALSE (pool_is_free_index (dpm->dhcp6_servers,
-                                                          server_idx)))
-                     {
-                     no_server:
-                       error0 = DHCPV6_PROXY_ERROR_NO_SERVER;
-                       next0 = DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_DROP;
-                       pkts_no_server++;
-                       goto do_trace;
-                     }
-
-          server = pool_elt_at_index(dpm->dhcp6_servers, server_idx);
-          if (server->valid == 0)
-            goto no_server;
+          if (PREDICT_FALSE (NULL == server))
+          {
+              error0 = DHCPV6_PROXY_ERROR_NO_SERVER;
+              next0 = DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_DROP;
+              pkts_no_server++;
+              goto do_trace;
+          }
 
           server_fib_idx = server->server_fib6_index;
           vnet_buffer(b0)->sw_if_index[VLIB_TX] = server_fib_idx;
@@ -331,19 +351,6 @@ dhcpv6_proxy_to_server_input (vlib_main_t * vm,
           id1 = (dhcpv6_int_id_t *) (((uword) ip1) + b0->current_length);
           b0->current_length += (sizeof (*id1));
 
-
-          fib = ip6_fib_get (rx_fib_idx);
-
-          //TODO: Revisit if hash makes sense here
-          p_vss = hash_get (dpm->vss_index_by_vrf_id,
-                            fib->table_id);
-          if (p_vss)
-            {
-              vss = pool_elt_at_index (dpm->vss, p_vss[0]);
-              oui1 =  vss->vpn_id.oui;
-              fib_id1 =  vss->vpn_id.fib_id;
-            }
-
           id1->opt.option = clib_host_to_net_u16(DHCPV6_OPTION_INTERFACE_ID);
           id1->opt.length = clib_host_to_net_u16(sizeof(rx_sw_if_index));
           id1->int_idx = clib_host_to_net_u32(rx_sw_if_index);
@@ -360,20 +367,24 @@ dhcpv6_proxy_to_server_input (vlib_main_t * vm,
                clib_memcpy(cmac->data, client_src_mac, 6);
                u1->length += sizeof(*cmac);
             }
-          if (server->insert_vss !=0 ) {
+
+          //TODO: Revisit if hash makes sense here
+          vss = dhcpv6_get_vss_info(dpm, rx_fib_idx);
+
+          if (NULL != vss) {
               vss1 = (dhcpv6_vss_t *) (((uword) ip1) + b0->current_length);
               b0->current_length += (sizeof (*vss1));
               vss1->opt.length =clib_host_to_net_u16(sizeof(*vss1) -
 						     sizeof(vss1->opt));
               vss1->opt.option = clib_host_to_net_u16(DHCPV6_OPTION_VSS);
               vss1->data[0] = 1;   // type
-              vss1->data[1] = oui1>>16 & 0xff;
-              vss1->data[2] = oui1>>8  & 0xff;
-              vss1->data[3] = oui1 & 0xff;
-              vss1->data[4] = fib_id1>>24 & 0xff;
-              vss1->data[5] = fib_id1>>16 & 0xff;
-              vss1->data[6] = fib_id1>>8 & 0xff;
-              vss1->data[7] = fib_id1 & 0xff;
+              vss1->data[1] = vss->vpn_id.oui >>16 & 0xff;
+              vss1->data[2] = vss->vpn_id.oui >>8  & 0xff;
+              vss1->data[3] = vss->vpn_id.oui & 0xff;
+              vss1->data[4] = vss->vpn_id.fib_id >> 24 & 0xff;
+              vss1->data[5] = vss->vpn_id.fib_id >> 16 & 0xff;
+              vss1->data[6] = vss->vpn_id.fib_id >> 8 & 0xff;
+              vss1->data[7] = vss->vpn_id.fib_id & 0xff;
               u1->length += sizeof(*vss1);
           }
 
@@ -524,9 +535,8 @@ dhcpv6_proxy_to_client_input (vlib_main_t * vm,
       u16 len = 0;
       u8 interface_opt_flag = 0;
       u8 relay_msg_opt_flag = 0;
-      ip6_fib_t * svr_fib;
       ip6_main_t * im = &ip6_main;
-      u32 server_fib_idx, svr_fib_id, client_fib_idx, server_idx;
+      u32 server_fib_idx, client_fib_idx;
 
       bi0 = from[0];
       from += 1;
@@ -608,31 +618,18 @@ dhcpv6_proxy_to_client_input (vlib_main_t * vm,
       vlib_buffer_advance (b0, sizeof(*r0));
 
       client_fib_idx = im->fib_index_by_sw_if_index[sw_if_index];
-      if (client_fib_idx < vec_len(dm->dhcp6_server_index_by_rx_fib_index))
-    	  server_idx = dm->dhcp6_server_index_by_rx_fib_index[client_fib_idx];
-      else
-    	  server_idx = 0;
-
-      if (PREDICT_FALSE (pool_is_free_index (dm->dhcp6_servers, server_idx)))
-        {
-          error0 = DHCPV6_PROXY_ERROR_WRONG_INTERFACE_ID_OPTION;
-          goto drop_packet;
-        }
+      server = dhcpv6_get_server(dm, client_fib_idx);
 
-      server = pool_elt_at_index (dm->dhcp6_servers, server_idx);
-      if (server->valid == 0)
+      if (NULL == server)
       {
     	  error0 = DHCPV6_PROXY_ERROR_NO_SERVER;
           goto drop_packet;
       }
 
-
       server_fib_idx = im->fib_index_by_sw_if_index
           [vnet_buffer(b0)->sw_if_index[VLIB_RX]];
-      svr_fib = ip6_fib_get (server_fib_idx);
-      svr_fib_id = svr_fib->table_id;
 
-      if (svr_fib_id != server->server_fib6_index ||
+      if (server_fib_idx != server->server_fib6_index ||
           ip0->src_address.as_u64[0] != server->dhcp6_server.as_u64[0] ||
           ip0->src_address.as_u64[1] != server->dhcp6_server.as_u64[1])
         {
@@ -760,7 +757,7 @@ clib_error_t * dhcpv6_proxy_init (vlib_main_t * vm)
   error_drop_node = vlib_get_node_by_name (vm, (u8 *) "error-drop");
   dm->error_drop_node_index = error_drop_node->index;
 
-  dm->vss_index_by_vrf_id = hash_create (0, sizeof (uword));
+  dm->vss_index_by_rx_fib_index = NULL;
 
   /* RFC says this is the dhcpv6 server address  */
   dm->all_dhcpv6_server_address.as_u64[0] = clib_host_to_net_u64 (0xFF05000000000000);
@@ -785,121 +782,138 @@ clib_error_t * dhcpv6_proxy_init (vlib_main_t * vm)
 
 VLIB_INIT_FUNCTION (dhcpv6_proxy_init);
 
-/* Old API, manipulates a single server (only) shared by all Rx VRFs */
-int dhcpv6_proxy_set_server (ip6_address_t *addr, ip6_address_t *src_address,
-                             u32 fib_id, int insert_vss, int is_del)
-{
-	return dhcpv6_proxy_set_server_2 (addr, src_address,
-			0, fib_id,
-			insert_vss, is_del);
-}
-
-int dhcpv6_proxy_set_server_2 (ip6_address_t *addr, ip6_address_t *src_address,
-                               u32 rx_fib_id, u32 server_fib_id,
-                               int insert_vss, int is_del)
+int dhcpv6_proxy_set_server (ip6_address_t *addr,
+                             ip6_address_t *src_address,
+                             u32 rx_fib_id,
+                             u32 server_fib_id,
+                             int is_del)
 {
   dhcpv6_proxy_main_t * dm = &dhcpv6_proxy_main;
   dhcpv6_server_t * server = 0;
-  u32 server_fib_index = 0;
   u32 rx_fib_index = 0;
+  int rc = 0;
 
   rx_fib_index = ip6_mfib_table_find_or_create_and_lock(rx_fib_id);
-  server_fib_index = ip6_fib_table_find_or_create_and_lock(server_fib_id);
-
-  if (is_del)
-      {
-
-	  if (rx_fib_index >= vec_len(dm->dhcp6_server_index_by_rx_fib_index))
-		  return VNET_API_ERROR_NO_SUCH_ENTRY;
 
-	  server_fib_index = dm->dhcp6_server_index_by_rx_fib_index[rx_fib_index];
-
-	  dm->dhcp6_server_index_by_rx_fib_index[rx_fib_index] = 0;
-	  server = pool_elt_at_index (dm->dhcp6_servers, server_fib_index);
-	  memset (server, 0, sizeof (*server));
-	  pool_put (dm->dhcp6_servers, server);
-	  return 0;
+  const mfib_prefix_t all_dhcp_servers = {
+      .fp_len = 128,
+      .fp_proto = FIB_PROTOCOL_IP6,
+      .fp_grp_addr = {
+          .ip6 = dm->all_dhcpv6_server_relay_agent_address,
       }
+  };
 
-  if (addr->as_u64[0] == 0 &&
-        addr->as_u64[1] == 0 )
-      return VNET_API_ERROR_INVALID_DST_ADDRESS;
-
-    if (src_address->as_u64[0] == 0 &&
-        src_address->as_u64[1] == 0)
-      return VNET_API_ERROR_INVALID_SRC_ADDRESS;
-
-  if (rx_fib_id == 0)
+  if (is_del)
     {
-      server = pool_elt_at_index (dm->dhcp6_servers, 0);
-      if (server->valid)
-          goto reconfigure_it;
-      else
-          goto initialize_it;
-    }
+      server = dhcpv6_get_server(dm, rx_fib_index);
 
-  if (rx_fib_index < vec_len(dm->dhcp6_server_index_by_rx_fib_index))
-    {
-      server_fib_index = dm->dhcp6_server_index_by_rx_fib_index[rx_fib_index];
-      if (server_fib_index != 0)
+      if (NULL == server)
         {
-          server = pool_elt_at_index (dm->dhcp6_servers, server_fib_index);
-          goto initialize_it;
+          rc = VNET_API_ERROR_NO_SUCH_ENTRY;
+          goto out;
         }
-    }
 
-  /*Allocate a new server*/
-  pool_get (dm->dhcp6_servers, server);
-
-  initialize_it:
-  {
-      const mfib_prefix_t all_dhcp_servers = {
-          .fp_len = 128,
-          .fp_proto = FIB_PROTOCOL_IP6,
-          .fp_grp_addr = {
-              .ip6 = dm->all_dhcpv6_server_relay_agent_address,
-          }
-      };
-      const fib_route_path_t path_for_us = {
-          .frp_proto = FIB_PROTOCOL_IP6,
-          .frp_addr = zero_addr,
-          .frp_sw_if_index = 0xffffffff,
-          .frp_fib_index = ~0,
-          .frp_weight = 0,
-          .frp_flags = FIB_ROUTE_PATH_LOCAL,
-      };
-      mfib_table_entry_path_update(rx_fib_index,
-                                   &all_dhcp_servers,
-                                   MFIB_SOURCE_DHCP,
-                                   &path_for_us,
-                                   MFIB_ITF_FLAG_FORWARD);
       /*
-       * Each interface that is enabled in this table, needs to be added
-       * as an accepting interface, but this is not easily doable in VPP.
-       * So we cheat. Add a flag to the entry that indicates accept form
-       * any interface.
-       * We will still only accept on v6 enabled interfaces, since the input
-       * feature ensures this.
+       * release the locks held on the server fib and rx mfib
        */
-      mfib_table_entry_update(rx_fib_index,
+      mfib_table_entry_delete(rx_fib_index,
                               &all_dhcp_servers,
-                              MFIB_SOURCE_DHCP,
-                              MFIB_ENTRY_FLAG_ACCEPT_ALL_ITF);
-  }
+                              MFIB_SOURCE_DHCP);
+      mfib_table_unlock(rx_fib_index, FIB_PROTOCOL_IP6);
+      fib_table_unlock(server->server_fib6_index, FIB_PROTOCOL_IP6);
 
-reconfigure_it:
+      dm->dhcp6_server_index_by_rx_fib_index[rx_fib_index] = ~0;
 
-  copy_ip6_address(&server->dhcp6_server, addr);
-  copy_ip6_address(&server->dhcp6_src_address, src_address);
-  server->server_fib6_index = server_fib_index;
-  server->valid = 1;
-  server->insert_vss = insert_vss;
+      memset (server, 0, sizeof (*server));
+      pool_put (dm->dhcp6_servers, server);
+    }
+  else
+    {
+      if (addr->as_u64[0] == 0 &&
+          addr->as_u64[1] == 0 )
+      {
+          rc = VNET_API_ERROR_INVALID_DST_ADDRESS;
+          goto out;
+      }
+      if (src_address->as_u64[0] == 0 &&
+          src_address->as_u64[1] == 0)
+      {
+          rc = VNET_API_ERROR_INVALID_SRC_ADDRESS;
+          goto out;
+      }
 
-  vec_validate (dm->dhcp6_server_index_by_rx_fib_index, rx_fib_index);
-  dm->dhcp6_server_index_by_rx_fib_index[rx_fib_index] =
-		  server - dm->dhcp6_servers;
+      server = dhcpv6_get_server(dm, rx_fib_index);
 
-  return 0;
+      if (NULL != server)
+        {
+          /* modify of an existing entry */
+          ip6_fib_t *fib;
+
+          fib = ip6_fib_get(server->server_fib6_index);
+
+          if (fib->table_id != server_fib_id)
+            {
+              /* swap tables */
+              fib_table_unlock(server->server_fib6_index, FIB_PROTOCOL_IP6);
+              server->server_fib6_index =
+                  ip6_fib_table_find_or_create_and_lock(server_fib_id);
+            }
+        }
+      else
+        {
+          /* Allocate a new server */
+          pool_get (dm->dhcp6_servers, server);
+
+          vec_validate_init_empty (dm->dhcp6_server_index_by_rx_fib_index,
+                                   rx_fib_index, ~0);
+          dm->dhcp6_server_index_by_rx_fib_index[rx_fib_index] =
+              server - dm->dhcp6_servers;
+
+          server->server_fib6_index =
+              ip6_fib_table_find_or_create_and_lock(server_fib_id);
+          mfib_table_lock(rx_fib_index, FIB_PROTOCOL_IP6);
+
+          const mfib_prefix_t all_dhcp_servers = {
+              .fp_len = 128,
+              .fp_proto = FIB_PROTOCOL_IP6,
+              .fp_grp_addr = {
+                  .ip6 = dm->all_dhcpv6_server_relay_agent_address,
+              }
+            };
+          const fib_route_path_t path_for_us = {
+              .frp_proto = FIB_PROTOCOL_IP6,
+              .frp_addr = zero_addr,
+              .frp_sw_if_index = 0xffffffff,
+              .frp_fib_index = ~0,
+              .frp_weight = 0,
+              .frp_flags = FIB_ROUTE_PATH_LOCAL,
+          };
+          mfib_table_entry_path_update(rx_fib_index,
+                                       &all_dhcp_servers,
+                                       MFIB_SOURCE_DHCP,
+                                       &path_for_us,
+                                       MFIB_ITF_FLAG_FORWARD);
+          /*
+           * Each interface that is enabled in this table, needs to be added
+           * as an accepting interface, but this is not easily doable in VPP.
+           * So we cheat. Add a flag to the entry that indicates accept form
+           * any interface.
+           * We will still only accept on v6 enabled interfaces, since the
+           * input feature ensures this.
+           */
+          mfib_table_entry_update(rx_fib_index,
+                                  &all_dhcp_servers,
+                                  MFIB_SOURCE_DHCP,
+                                  MFIB_ENTRY_FLAG_ACCEPT_ALL_ITF);
+        }
+      copy_ip6_address(&server->dhcp6_server, addr);
+      copy_ip6_address(&server->dhcp6_src_address, src_address);
+  }
+
+out:
+  mfib_table_unlock(rx_fib_index, FIB_PROTOCOL_IP6);
+
+  return (rc);
 }
 
 static clib_error_t *
@@ -910,7 +924,7 @@ dhcpv6_proxy_set_command_fn (vlib_main_t * vm,
   ip6_address_t addr, src_addr;
   int set_server = 0, set_src_address = 0;
   u32 rx_fib_id = 0, server_fib_id = 0;
-  int is_del = 0, add_vss = 0;
+  int is_del = 0;
 
   while (unformat_check_input(input) != UNFORMAT_END_OF_INPUT)
     {
@@ -924,9 +938,6 @@ dhcpv6_proxy_set_command_fn (vlib_main_t * vm,
         ;
        else if (unformat (input, "rx-fib-id %d", &rx_fib_id))
          ;
-       else if (unformat (input, "add-vss-option")
-               || unformat (input, "insert-option"))
-          add_vss = 1;
        else if (unformat (input, "delete") ||
                 unformat (input, "del"))
            is_del = 1;
@@ -938,8 +949,8 @@ dhcpv6_proxy_set_command_fn (vlib_main_t * vm,
   {
       int rv;
 
-      rv = dhcpv6_proxy_set_server_2 (&addr, &src_addr, rx_fib_id,
-    		  server_fib_id, add_vss, is_del);
+      rv = dhcpv6_proxy_set_server (&addr, &src_addr, rx_fib_id,
+                                    server_fib_id, is_del);
 
       //TODO: Complete the errors
       switch (rv)
@@ -962,7 +973,7 @@ dhcpv6_proxy_set_command_fn (vlib_main_t * vm,
 VLIB_CLI_COMMAND (dhcpv6_proxy_set_command, static) = {
   .path = "set dhcpv6 proxy",
   .short_help = "set dhcpv6 proxy [del] server <ipv6-addr> src-address <ipv6-addr> "
-		  "[add-vss-option] [server-fib-id <fib-id>] [rx-fib-id <fib-id>] ",
+		  "[server-fib-id <fib-id>] [rx-fib-id <fib-id>] ",
   .function = dhcpv6_proxy_set_command_fn,
 };
 
@@ -976,8 +987,8 @@ u8 * format_dhcpv6_proxy_server (u8 * s, va_list * args)
 
   if (dm == 0)
     {
-      s = format (s, "%=40s%=40s%=14s%=14s%=20s", "Server Address", "Source Address",
-                  "Server FIB", "RX FIB", "Insert VSS Option");
+      s = format (s, "%=40s%=40s%=14s%=14s", "Server Address", "Source Address",
+                  "Server FIB", "RX FIB");
       return s;
     }
 
@@ -990,11 +1001,10 @@ u8 * format_dhcpv6_proxy_server (u8 * s, va_list * args)
   if (rx_fib)
 	  rx_fib_id = rx_fib->table_id;
 
-  s = format (s, "%=40U%=40U%=14u%=14u%=20s",
+  s = format (s, "%=40U%=40U%=14u%=14u",
               format_ip6_address, &server->dhcp6_server,
               format_ip6_address, &server->dhcp6_src_address,
-			  server_fib_id, rx_fib_id,
-			                server->insert_vss ? "yes" : "no");
+			  server_fib_id, rx_fib_id);
   return s;
 }
 
@@ -1003,25 +1013,25 @@ dhcpv6_proxy_show_command_fn (vlib_main_t * vm,
                             unformat_input_t * input,
                             vlib_cli_command_t * cmd)
 {
-  dhcpv6_proxy_main_t * dm = &dhcpv6_proxy_main;
-  ip6_main_t * im = &ip6_main;
+  dhcpv6_proxy_main_t * dpm = &dhcpv6_proxy_main;
   int i;
   u32 server_index;
   dhcpv6_server_t * server;
 
   vlib_cli_output (vm, "%U", format_dhcpv6_proxy_server, 0 /* header line */,
 		  0, 0);
-  for (i = 0; i < vec_len (im->fibs); i++)
-      {
-        if (i < vec_len(dm->dhcp6_server_index_by_rx_fib_index))
-          server_index = dm->dhcp6_server_index_by_rx_fib_index[i];
-        else
-          server_index = 0;
-        server = pool_elt_at_index (dm->dhcp6_servers, server_index);
-        if (server->valid)
-          vlib_cli_output (vm, "%U", format_dhcpv6_proxy_server, dm,
-		  server, i);
-      }
+  vec_foreach_index (i, dpm->dhcp6_server_index_by_rx_fib_index)
+  {
+      server_index = dpm->dhcp6_server_index_by_rx_fib_index[i];
+      if (~0 == server_index)
+          continue;
+
+      server = pool_elt_at_index (dpm->dhcp6_servers, server_index);
+
+      vlib_cli_output (vm, "%U", format_dhcpv6_proxy_server, dpm, 
+                       server, i);
+    }
+
   return 0;
 }
 
@@ -1031,51 +1041,104 @@ VLIB_CLI_COMMAND (dhcpv6_proxy_show_command, static) = {
   .function = dhcpv6_proxy_show_command_fn,
 };
 
+void
+dhcpv6_proxy_dump (void *opaque,
+                   u32 context)
+{
+  dhcpv6_proxy_main_t * dpm = &dhcpv6_proxy_main;
+  ip6_fib_t *s_fib, *r_fib;
+  dhcpv6_server_t * server;
+  u32 server_index, i;
+  dhcpv6_vss_info *v;
+
+  vec_foreach_index (i, dpm->dhcp6_server_index_by_rx_fib_index)
+  {
+      server_index = dpm->dhcp6_server_index_by_rx_fib_index[i];
+      if (~0 == server_index)
+          continue;
+
+      server = pool_elt_at_index (dpm->dhcp6_servers, server_index);
+      v = dhcpv6_get_vss_info(dpm, i);
+
+      ip46_address_t src_addr = {
+          .ip6 = server->dhcp6_src_address,
+      };
+      ip46_address_t server_addr = {
+          .ip6 = server->dhcp6_server,
+      };
+
+      s_fib = ip6_fib_get(server->server_fib6_index);
+      r_fib = ip6_fib_get(i);
+
+      dhcp_send_details(opaque,
+                        context,
+                        &server_addr,
+                        &src_addr,
+                        s_fib->table_id,
+                        r_fib->table_id,
+                        (v ? v->vpn_id.fib_id : 0),
+                        (v ? v->vpn_id.oui : 0));
+  }
+}
+
 int dhcpv6_proxy_set_vss(u32 tbl_id,
                          u32 oui,
                          u32 fib_id,
                          int is_del)
 {
   dhcpv6_proxy_main_t *dm = &dhcpv6_proxy_main;
-  u32 old_oui, old_fib_id;
-  uword *p;
-  dhcpv6_vss_info *v;
+  dhcpv6_vss_info *v = NULL;
+  u32  rx_fib_index;
+  int rc = 0;
 
-  p = hash_get (dm->vss_index_by_vrf_id, tbl_id);
+  rx_fib_index = ip6_fib_table_find_or_create_and_lock(tbl_id);
+  v = dhcpv6_get_vss_info(dm, rx_fib_index);
 
-  if (p) {
-      v = pool_elt_at_index (dm->vss, p[0]);
-      if (!v)
-        return VNET_API_ERROR_NO_SUCH_FIB;
-
-      old_oui = v->vpn_id.oui;
-      old_fib_id = v->vpn_id.fib_id;
+  if (NULL != v)
+  {
+      if (is_del)
+      {
+          /* release the lock held on the table when the VSS
+           * info was created */
+          fib_table_unlock (rx_fib_index,
+                            FIB_PROTOCOL_IP6);
 
+          pool_put (dm->vss, v);
+          dm->vss_index_by_rx_fib_index[rx_fib_index] = ~0;
+      }
+      else
+      {
+          /* this is a modify */
+          v->vpn_id.fib_id = fib_id;
+          v->vpn_id.oui = oui;
+      }
+  }
+  else
+  {
       if (is_del)
+          rc = VNET_API_ERROR_NO_SUCH_ENTRY;
+      else
       {
-          if (old_oui == oui &&
-              old_fib_id == fib_id )
-          {
-              pool_put(dm->vss, v);
-              hash_unset (dm->vss_index_by_vrf_id, tbl_id);
-              return 0;
-          }
-          else
-            return VNET_API_ERROR_NO_SUCH_ENTRY;
+          /* create a new entry */
+          vec_validate_init_empty(dm->vss_index_by_rx_fib_index,
+                                  rx_fib_index, ~0);
+
+          /* hold a lock on the table whilst the VSS info exist */
+          fib_table_lock (rx_fib_index,
+                          FIB_PROTOCOL_IP6);
+
+          pool_get (dm->vss, v);
+          v->vpn_id.fib_id = fib_id;
+          v->vpn_id.oui = oui;
+          dm->vss_index_by_rx_fib_index[rx_fib_index] = v - dm->vss;
       }
+  }
 
-      pool_put(dm->vss, v);
-      hash_unset (dm->vss_index_by_vrf_id, tbl_id);
-  } else if (is_del)
-    return VNET_API_ERROR_NO_SUCH_ENTRY;
-
-  pool_get (dm->vss, v);
-  memset (v, ~0, sizeof (*v));
-  v->vpn_id.fib_id = fib_id;
-  v->vpn_id.oui = oui;
-  hash_set (dm->vss_index_by_vrf_id, tbl_id, v - dm->vss);
+  /* Release the lock taken during the create_or_lock at the start */
+  fib_table_unlock (rx_fib_index,
+                    FIB_PROTOCOL_IP6);
 
-  return 0;
+  return (rc);
 }
 
 
@@ -1147,19 +1210,19 @@ dhcpv6_vss_show_command_fn (vlib_main_t * vm,
 {
   dhcpv6_proxy_main_t * dm = &dhcpv6_proxy_main;
   dhcpv6_vss_info *v;
-  u32 oui;
-  u32 fib_id;
-  u32 tbl_id;
-  uword index;
+  ip6_fib_t *fib;
+  u32 *fib_index;
 
   vlib_cli_output (vm, "%=6s%=6s%=12s","Table", "OUI", "VPN ID");
-  hash_foreach (tbl_id, index, dm->vss_index_by_vrf_id,
+  pool_foreach (fib_index, dm->vss_index_by_rx_fib_index,
   ({
-     v = pool_elt_at_index (dm->vss, index);
-     oui = v->vpn_id.oui;
-     fib_id = v->vpn_id.fib_id;
-     vlib_cli_output (vm, "%=6d%=6d%=12d",
-                      tbl_id, oui, fib_id);
+      fib = ip6_fib_get (*fib_index);
+      v = pool_elt_at_index (dm->vss, *fib_index);
+
+      vlib_cli_output (vm, "%=6d%=6d%=12d",
+                       fib->table_id,
+                       v->vpn_id.oui,
+                       v->vpn_id.fib_id);
   }));
 
   return 0;
diff --git a/src/vnet/dpo/receive_dpo.c b/src/vnet/dpo/receive_dpo.c
index 2b2571c6..83e33ed8 100644
--- a/src/vnet/dpo/receive_dpo.c
+++ b/src/vnet/dpo/receive_dpo.c
@@ -102,6 +102,11 @@ format_receive_dpo (u8 *s, va_list *ap)
     vnet_main_t * vnm = vnet_get_main();
     receive_dpo_t *rd;
 
+    if (pool_is_free_index(receive_dpo_pool, index))
+    {
+        return (format(s, "dpo-receive DELETED"));
+    }
+
     rd = receive_dpo_get(index);
 
     if (~0 != rd->rd_sw_if_index)
diff --git a/src/vpp/api/custom_dump.c b/src/vpp/api/custom_dump.c
index a7dca989..70b4e4c9 100644
--- a/src/vpp/api/custom_dump.c
+++ b/src/vpp/api/custom_dump.c
@@ -772,37 +772,6 @@ static void *vl_api_dhcp_proxy_config_t_print
 {
   u8 *s;
 
-  s = format (0, "SCRIPT: dhcp_proxy_config ");
-
-  s = format (s, "vrf_id %d ", ntohl (mp->vrf_id));
-
-  if (mp->is_ipv6)
-    {
-      s = format (s, "svr %U ", format_ip6_address,
-		  (ip6_address_t *) mp->dhcp_server);
-      s = format (s, "src %U ", format_ip6_address,
-		  (ip6_address_t *) mp->dhcp_src_address);
-    }
-  else
-    {
-      s = format (s, "svr %U ", format_ip4_address,
-		  (ip4_address_t *) mp->dhcp_server);
-      s = format (s, "src %U ", format_ip4_address,
-		  (ip4_address_t *) mp->dhcp_src_address);
-    }
-  if (mp->is_add == 0)
-    s = format (s, "del ");
-
-  s = format (s, "insert-cid %d ", mp->insert_circuit_id);
-
-  FINISH;
-}
-
-static void *vl_api_dhcp_proxy_config_2_t_print
-  (vl_api_dhcp_proxy_config_2_t * mp, void *handle)
-{
-  u8 *s;
-
   s = format (0, "SCRIPT: dhcp_proxy_config_2 ");
 
   s = format (s, "rx_vrf_id %d ", ntohl (mp->rx_vrf_id));
@@ -825,8 +794,6 @@ static void *vl_api_dhcp_proxy_config_2_t_print
   if (mp->is_add == 0)
     s = format (s, "del ");
 
-  s = format (s, "insert-cid %d ", mp->insert_circuit_id);
-
   FINISH;
 }
 
@@ -2954,7 +2921,6 @@ _(BRIDGE_DOMAIN_DUMP, bridge_domain_dump)                               \
 _(CLASSIFY_SET_INTERFACE_IP_TABLE, classify_set_interface_ip_table)	\
 _(CLASSIFY_SET_INTERFACE_L2_TABLES, classify_set_interface_l2_tables)	\
 _(ADD_NODE_NEXT, add_node_next)						\
-_(DHCP_PROXY_CONFIG_2, dhcp_proxy_config_2)	                        \
 _(DHCP_CLIENT_CONFIG, dhcp_client_config)	                        \
 _(L2TPV3_CREATE_TUNNEL, l2tpv3_create_tunnel)                           \
 _(L2TPV3_SET_TUNNEL_COOKIES, l2tpv3_set_tunnel_cookies)                 \
diff --git a/test/test_dhcp.py b/test/test_dhcp.py
index 04ab2e11..fbfb8a0c 100644
--- a/test/test_dhcp.py
+++ b/test/test_dhcp.py
@@ -65,7 +65,7 @@ class TestDHCP(VppTestCase):
         for i in self.pg_interfaces:
             i.assert_nothing_captured(remark=remark)
 
-    def validate_option_82(self, pkt, intf, ip_addr):
+    def validate_relay_options(self, pkt, intf, ip_addr, fib_id, oui):
         dhcp = pkt[DHCP]
         found = 0
         data = []
@@ -77,7 +77,10 @@ class TestDHCP(VppTestCase):
                     # There are two sb-options present - each of length 6.
                     #
                     data = i[1]
-                    self.assertEqual(len(data), 12)
+                    if oui != 0:
+                        self.assertEqual(len(data), 24)
+                    else:
+                        self.assertEqual(len(data), 12)
 
                     #
                     # First sub-option is ID 1, len 4, then encoded
@@ -107,12 +110,30 @@ class TestDHCP(VppTestCase):
                     self.assertEqual(data[10], claddr[2])
                     self.assertEqual(data[11], claddr[3])
 
+                    if oui != 0:
+                        # sub-option 151 encodes the 3 byte oui
+                        # and the 4 byte fib_id
+                        self.assertEqual(ord(data[12]), 151)
+                        self.assertEqual(ord(data[13]), 8)
+                        self.assertEqual(ord(data[14]), 1)
+                        self.assertEqual(ord(data[15]), 0)
+                        self.assertEqual(ord(data[16]), 0)
+                        self.assertEqual(ord(data[17]), oui)
+                        self.assertEqual(ord(data[18]), 0)
+                        self.assertEqual(ord(data[19]), 0)
+                        self.assertEqual(ord(data[20]), 0)
+                        self.assertEqual(ord(data[21]), fib_id)
+
+                        # VSS control sub-option
+                        self.assertEqual(ord(data[22]), 152)
+                        self.assertEqual(ord(data[23]), 0)
+
                     found = 1
         self.assertTrue(found)
 
         return data
 
-    def verify_dhcp_offer(self, pkt, intf, check_option_82=True):
+    def verify_dhcp_offer(self, pkt, intf):
         ether = pkt[Ether]
         self.assertEqual(ether.dst, "ff:ff:ff:ff:ff:ff")
         self.assertEqual(ether.src, intf.local_mac)
@@ -134,11 +155,9 @@ class TestDHCP(VppTestCase):
                     is_offer = True
         self.assertTrue(is_offer)
 
-        if check_option_82:
-            data = self.validate_option_82(pkt, intf, intf.local_ip4)
+        data = self.validate_relay_options(pkt, intf, intf.local_ip4, 0, 0)
 
-    def verify_dhcp_discover(self, pkt, intf, src_intf=None,
-                             option_82_present=True):
+    def verify_dhcp_discover(self, pkt, intf, src_intf=None, fib_id=0, oui=0):
         ether = pkt[Ether]
         self.assertEqual(ether.dst, intf.remote_mac)
         self.assertEqual(ether.src, intf.local_mac)
@@ -161,13 +180,10 @@ class TestDHCP(VppTestCase):
                     is_discover = True
         self.assertTrue(is_discover)
 
-        if option_82_present:
-            data = self.validate_option_82(pkt, src_intf, src_intf.local_ip4)
-            return data
-        else:
-            for i in dhcp.options:
-                if type(i) is tuple:
-                    self.assertNotEqual(i[0], "relay_agent_Information")
+        data = self.validate_relay_options(pkt, src_intf,
+                                           src_intf.local_ip4,
+                                           fib_id, oui)
+        return data
 
     def verify_dhcp6_solicit(self, pkt, intf,
                              peer_ip, peer_mac,
@@ -193,18 +209,19 @@ class TestDHCP(VppTestCase):
         self.assertEqual(cll.lltype, 1)
         self.assertEqual(cll.clladdr, peer_mac)
 
-        vss = pkt[DHCP6OptVSS]
-        self.assertEqual(vss.optlen, 8)
-        self.assertEqual(vss.type, 1)
-        # the OUI and FIB-id are really 3 and 4 bytes resp.
-        # but the tested range is small
-        self.assertEqual(ord(vss.data[0]), 0)
-        self.assertEqual(ord(vss.data[1]), 0)
-        self.assertEqual(ord(vss.data[2]), oui)
-        self.assertEqual(ord(vss.data[3]), 0)
-        self.assertEqual(ord(vss.data[4]), 0)
-        self.assertEqual(ord(vss.data[5]), 0)
-        self.assertEqual(ord(vss.data[6]), fib_id)
+        if fib_id != 0:
+            vss = pkt[DHCP6OptVSS]
+            self.assertEqual(vss.optlen, 8)
+            self.assertEqual(vss.type, 1)
+            # the OUI and FIB-id are really 3 and 4 bytes resp.
+            # but the tested range is small
+            self.assertEqual(ord(vss.data[0]), 0)
+            self.assertEqual(ord(vss.data[1]), 0)
+            self.assertEqual(ord(vss.data[2]), oui)
+            self.assertEqual(ord(vss.data[3]), 0)
+            self.assertEqual(ord(vss.data[4]), 0)
+            self.assertEqual(ord(vss.data[5]), 0)
+            self.assertEqual(ord(vss.data[6]), fib_id)
 
         # the relay message should be an encoded Solicit
         msg = pkt[DHCP6OptRelayMsg]
@@ -267,29 +284,16 @@ class TestDHCP(VppTestCase):
                                     rx_table_id=0)
 
         #
-        # Now a DHCP request on pg2, which is in the same VRF
-        # as the DHCP config, will result in a relayed DHCP
-        # message to the [fake] server
-        #
-        self.pg2.add_stream(pkts_disc_vrf0)
-        self.pg_enable_capture(self.pg_interfaces)
-        self.pg_start()
-
-        rx = self.pg0.get_capture(1)
-        rx = rx[0]
-
-        #
-        # Rx'd packet should be to the server address and from the configured
-        # source address
-        # UDP source ports are unchanged
-        # we've no option 82 config so that should be absent
+        # Discover packets from the client are dropped because there is no
+        # IP address configured on the client facing interface
         #
-        self.verify_dhcp_discover(rx, self.pg0, option_82_present=False)
+        self.send_and_assert_no_replies(self.pg2, pkts_disc_vrf0,
+                                        "Discover DHCP no relay address")
 
         #
         # Inject a response from the server
-        #  VPP will only relay the offer if option 82 is present.
-        #  so this one is dropped
+        #  dropped, because there is no IP addrees on the
+        #  clinet interfce to fill in the option.
         #
         p = (Ether(dst=self.pg0.local_mac, src=self.pg0.remote_mac) /
              IP(src=self.pg0.remote_ip4, dst=self.pg0.local_ip4) /
@@ -298,24 +302,8 @@ class TestDHCP(VppTestCase):
              DHCP(options=[('message-type', 'offer'), ('end')]))
         pkts = [p]
 
-        self.send_and_assert_no_replies(self.pg0, pkts,
-                                        "DHCP offer no option 82")
-
-        #
-        # Configure sending option 82 in relayed messages
-        #
-        self.vapi.dhcp_proxy_config(server_addr,
-                                    src_addr,
-                                    rx_table_id=0,
-                                    insert_circuit_id=1)
-
-        #
-        # Send a request:
-        #  again dropped, but ths time because there is no IP addrees on the
-        #  clinet interfce to fill in the option.
-        #
-        self.send_and_assert_no_replies(self.pg2, pkts_disc_vrf0,
-                                        "DHCP no relay address")
+        self.send_and_assert_no_replies(self.pg2, pkts,
+                                        "Offer DHCP no relay address")
 
         #
         # configure an IP address on the client facing interface
@@ -376,15 +364,8 @@ class TestDHCP(VppTestCase):
                            ('relay_agent_Information', bad_ip),
                            ('end')]))
         pkts = [p]
-
-        self.pg0.add_stream(pkts)
-        self.pg_enable_capture(self.pg_interfaces)
-        self.pg_start()
-        rx = self.pg2.get_capture(1)
-        rx = rx[0]
-
-        self.verify_dhcp_offer(rx, self.pg2, check_option_82=False)
-        self.pg0.assert_nothing_captured(remark="")
+        self.send_and_assert_no_replies(self.pg0, pkts,
+                                        "DHCP offer option 82 bad address")
 
         # 2. Not a sw_if_index VPP knows
         bad_if_index = option_82[0:2] + chr(33) + option_82[3:]
@@ -413,8 +394,7 @@ class TestDHCP(VppTestCase):
         self.vapi.dhcp_proxy_config(server_addr,
                                     src_addr,
                                     rx_table_id=0,
-                                    is_add=0,
-                                    insert_circuit_id=1)
+                                    is_add=0)
 
         self.send_and_assert_no_replies(self.pg2, pkts_disc_vrf0,
                                         "DHCP config removed VRF 0")
@@ -429,8 +409,7 @@ class TestDHCP(VppTestCase):
         self.vapi.dhcp_proxy_config(server_addr,
                                     src_addr,
                                     rx_table_id=1,
-                                    server_table_id=1,
-                                    insert_circuit_id=1)
+                                    server_table_id=1)
 
         #
         # Confim DHCP requests ok in VRF 1.
@@ -452,14 +431,41 @@ class TestDHCP(VppTestCase):
         rx = rx[0]
         self.verify_dhcp_discover(rx, self.pg1, src_intf=self.pg3)
 
+        #
+        # Add VSS config
+        #  table=1, fib=id=1, oui=4
+        self.vapi.dhcp_proxy_set_vss(1, 1, 4)
+
+        self.pg3.add_stream(pkts_disc_vrf1)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg1.get_capture(1)
+        rx = rx[0]
+        self.verify_dhcp_discover(rx, self.pg1, src_intf=self.pg3,
+                                  fib_id=1, oui=4)
+
+        #
+        # Remove the VSS config
+        #  relayed DHCP has default vlaues in the option.
+        #
+        self.vapi.dhcp_proxy_set_vss(1, 1, 4, is_add=0)
+
+        self.pg3.add_stream(pkts_disc_vrf1)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg1.get_capture(1)
+        rx = rx[0]
+        self.verify_dhcp_discover(rx, self.pg1, src_intf=self.pg3)
+
         #
         # remove DHCP config to cleanup
         #
         self.vapi.dhcp_proxy_config(server_addr,
                                     src_addr,
                                     rx_table_id=1,
-                                    server_table_id=1,
-                                    insert_circuit_id=1,
+                                    server_table_id=11,
                                     is_add=0)
 
         self.send_and_assert_no_replies(self.pg2, pkts_disc_vrf0,
@@ -510,7 +516,6 @@ class TestDHCP(VppTestCase):
                                     src_addr_vrf0,
                                     rx_table_id=0,
                                     server_table_id=0,
-                                    insert_circuit_id=1,
                                     is_ipv6=1)
 
         self.send_and_assert_no_replies(self.pg2, pkts_solicit_vrf0,
@@ -630,7 +635,6 @@ class TestDHCP(VppTestCase):
                                     src_addr_vrf1,
                                     rx_table_id=1,
                                     server_table_id=1,
-                                    insert_circuit_id=1,
                                     is_ipv6=1)
         self.pg3.config_ip6()
 
@@ -708,14 +712,12 @@ class TestDHCP(VppTestCase):
                                     src_addr_vrf1,
                                     rx_table_id=1,
                                     server_table_id=1,
-                                    insert_circuit_id=1,
                                     is_ipv6=1,
                                     is_add=0)
         self.vapi.dhcp_proxy_config(server_addr_vrf1,
                                     src_addr_vrf1,
                                     rx_table_id=0,
                                     server_table_id=0,
-                                    insert_circuit_id=1,
                                     is_ipv6=1,
                                     is_add=0)
 
diff --git a/test/vpp_papi_provider.py b/test/vpp_papi_provider.py
index 32680424..59e58ad0 100644
--- a/test/vpp_papi_provider.py
+++ b/test/vpp_papi_provider.py
@@ -1240,16 +1240,14 @@ class VppPapiProvider(object):
                           rx_table_id=0,
                           server_table_id=0,
                           is_add=1,
-                          is_ipv6=0,
-                          insert_circuit_id=0):
+                          is_ipv6=0):
         return self.api(
-            self.papi.dhcp_proxy_config_2,
+            self.papi.dhcp_proxy_config,
             {
                 'rx_vrf_id': rx_table_id,
                 'server_vrf_id': server_table_id,
                 'is_ipv6': is_ipv6,
                 'is_add': is_add,
-                'insert_circuit_id': insert_circuit_id,
                 'dhcp_server': dhcp_server,
                 'dhcp_src_address': dhcp_src_address,
             })
-- 
cgit 1.2.3-korg


From 2dd6852d8109e39d15a5c60f7ba58f1abcf9e455 Mon Sep 17 00:00:00 2001
From: Neale Ranns <nranns@cisco.com>
Date: Thu, 16 Feb 2017 03:38:59 -0800
Subject: Consolidate DHCP v4 and V6 implementation. No functional change
 intended

The DHCP proxy and VSS information maintained by VPP is the same for v4 and v6, so we can manage this state using the same code.
Packet handling is cleary different, so this is kept separate.

Change-Id: I10f10cc1f7f19debcd4c4b099c6de64e56bb0c69
Signed-off-by: Neale Ranns <nranns@cisco.com>
---
 src/vnet.am                         |   26 +-
 src/vnet/dhcp/client.c              |    2 +-
 src/vnet/dhcp/client.h              |    2 +-
 src/vnet/dhcp/dhcp4_packet.h        |   61 ++
 src/vnet/dhcp/dhcp4_proxy_error.def |   32 +
 src/vnet/dhcp/dhcp4_proxy_node.c    |  983 +++++++++++++++++++++++++++
 src/vnet/dhcp/dhcp6_packet.h        |  183 +++++
 src/vnet/dhcp/dhcp6_proxy_error.def |   29 +
 src/vnet/dhcp/dhcp6_proxy_node.c    | 1065 +++++++++++++++++++++++++++++
 src/vnet/dhcp/dhcp_api.c            |   95 ++-
 src/vnet/dhcp/dhcp_proxy.c          |  275 ++++++++
 src/vnet/dhcp/dhcp_proxy.h          |  248 +++++++
 src/vnet/dhcp/packet.h              |   61 --
 src/vnet/dhcp/proxy.h               |   99 ---
 src/vnet/dhcp/proxy_error.def       |   31 -
 src/vnet/dhcp/proxy_node.c          | 1192 --------------------------------
 src/vnet/dhcpv6/packet.h            |  183 -----
 src/vnet/dhcpv6/proxy.h             |   94 ---
 src/vnet/dhcpv6/proxy_error.def     |   29 -
 src/vnet/dhcpv6/proxy_node.c        | 1280 -----------------------------------
 src/vpp/api/custom_dump.c           |    3 +-
 test/test_dhcp.py                   |    2 +-
 22 files changed, 2933 insertions(+), 3042 deletions(-)
 create mode 100644 src/vnet/dhcp/dhcp4_packet.h
 create mode 100644 src/vnet/dhcp/dhcp4_proxy_error.def
 create mode 100644 src/vnet/dhcp/dhcp4_proxy_node.c
 create mode 100644 src/vnet/dhcp/dhcp6_packet.h
 create mode 100644 src/vnet/dhcp/dhcp6_proxy_error.def
 create mode 100644 src/vnet/dhcp/dhcp6_proxy_node.c
 create mode 100644 src/vnet/dhcp/dhcp_proxy.c
 create mode 100644 src/vnet/dhcp/dhcp_proxy.h
 delete mode 100644 src/vnet/dhcp/packet.h
 delete mode 100644 src/vnet/dhcp/proxy.h
 delete mode 100644 src/vnet/dhcp/proxy_error.def
 delete mode 100644 src/vnet/dhcp/proxy_node.c
 delete mode 100644 src/vnet/dhcpv6/packet.h
 delete mode 100644 src/vnet/dhcpv6/proxy.h
 delete mode 100644 src/vnet/dhcpv6/proxy_error.def
 delete mode 100644 src/vnet/dhcpv6/proxy_node.c

(limited to 'test/test_dhcp.py')

diff --git a/src/vnet.am b/src/vnet.am
index 70f1e7e9..64484e18 100644
--- a/src/vnet.am
+++ b/src/vnet.am
@@ -674,7 +674,7 @@ libvnet_la_SOURCES +=				\
  vnet/dhcp/dhcp_api.c
 
 nobase_include_HEADERS +=			\
- vnet/dhcp/client.h					\
+ vnet/dhcp/client.h				\
  vnet/dhcp/dhcp.api.h
 
 API_FILES += vnet/dhcp/dhcp.api
@@ -683,13 +683,16 @@ API_FILES += vnet/dhcp/dhcp.api
 # DHCP proxy
 ########################################
 libvnet_la_SOURCES +=				\
- vnet/dhcp/proxy_node.c				\
- vnet/dhcp/proxy.h
+ vnet/dhcp/dhcp6_proxy_node.c                   \
+ vnet/dhcp/dhcp4_proxy_node.c			\
+ vnet/dhcp/dhcp_proxy.c
 
 nobase_include_HEADERS +=			\
- vnet/dhcp/packet.h				\
- vnet/dhcp/proxy.h				\
- vnet/dhcp/proxy_error.def
+ vnet/dhcp/dhcp4_packet.h		        \
+ vnet/dhcp/dhcp6_packet.h		        \
+ vnet/dhcp/dhcp_proxy.h				\
+ vnet/dhcp/dhcp6_proxy_error.def                \
+ vnet/dhcp/dhcp4_proxy_error.def
 
 ########################################
 # ipv6 segment routing
@@ -709,17 +712,6 @@ nobase_include_HEADERS +=			\
 
 API_FILES += vnet/sr/sr.api
 
-########################################
-# DHCPv6 proxy
-########################################
-libvnet_la_SOURCES +=				\
- vnet/dhcpv6/proxy_node.c
-
-nobase_include_HEADERS +=			\
- vnet/dhcpv6/packet.h				\
- vnet/dhcpv6/proxy.h				\
- vnet/dhcpv6/proxy_error.def
-
 ########################################
 # IPFIX / netflow v10
 ########################################
diff --git a/src/vnet/dhcp/client.c b/src/vnet/dhcp/client.c
index 8a1a43b3..d34c5a64 100644
--- a/src/vnet/dhcp/client.c
+++ b/src/vnet/dhcp/client.c
@@ -14,7 +14,7 @@
  */
 #include <vlib/vlib.h>
 #include <vnet/dhcp/client.h>
-#include <vnet/dhcp/proxy.h>
+#include <vnet/dhcp/dhcp_proxy.h>
 #include <vnet/fib/fib_table.h>
 
 dhcp_client_main_t dhcp_client_main;
diff --git a/src/vnet/dhcp/client.h b/src/vnet/dhcp/client.h
index a74368cb..1f85d7ce 100644
--- a/src/vnet/dhcp/client.h
+++ b/src/vnet/dhcp/client.h
@@ -20,7 +20,7 @@
 #define included_dhcp_client_h
 
 #include <vnet/ip/ip.h>
-#include <vnet/dhcp/packet.h>
+#include <vnet/dhcp/dhcp4_packet.h>
 
 #define foreach_dhcp_client_state               \
 _(DHCP_DISCOVER)                                \
diff --git a/src/vnet/dhcp/dhcp4_packet.h b/src/vnet/dhcp/dhcp4_packet.h
new file mode 100644
index 00000000..28c4b156
--- /dev/null
+++ b/src/vnet/dhcp/dhcp4_packet.h
@@ -0,0 +1,61 @@
+#ifndef included_vnet_dhcp4_packet_h
+#define included_vnet_dhcp4_packet_h
+
+/*
+ * DHCP packet format
+ *
+ * Copyright (c) 2013 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+#include <vnet/ip/ip4_packet.h>
+
+typedef struct {
+  u8 opcode;                    /* 1 = request, 2 = reply */
+  u8 hardware_type;             /* 1 = ethernet */
+  u8 hardware_address_length;
+  u8 hops;
+  u32 transaction_identifier;
+  u16 seconds;
+  u16 flags;
+#define DHCP_FLAG_BROADCAST (1<<15)
+  ip4_address_t client_ip_address;
+  ip4_address_t your_ip_address; /* use this one */
+  ip4_address_t server_ip_address;
+  ip4_address_t gateway_ip_address; /* use option 3, not this one */
+  u8 client_hardware_address[16];
+  u8 server_name[64];
+  u8 boot_filename[128];
+  ip4_address_t magic_cookie;
+  u8 options[0];
+} dhcp_header_t;
+
+typedef struct {
+  u8 option;
+  u8 length;
+  union {
+    u8 data[0];
+    u32 data_as_u32[0];
+  };
+} __attribute__((packed)) dhcp_option_t;
+
+typedef enum {
+  DHCP_PACKET_DISCOVER=1,
+  DHCP_PACKET_OFFER,
+  DHCP_PACKET_REQUEST,
+  DHCP_PACKET_ACK=5,
+} dhcp_packet_type_t;
+
+/* charming antique: 99.130.83.99 is the dhcp magic cookie */
+#define DHCP_MAGIC (clib_host_to_net_u32(0x63825363))
+
+#endif /* included_vnet_dhcp4_packet_h */
diff --git a/src/vnet/dhcp/dhcp4_proxy_error.def b/src/vnet/dhcp/dhcp4_proxy_error.def
new file mode 100644
index 00000000..adf04808
--- /dev/null
+++ b/src/vnet/dhcp/dhcp4_proxy_error.def
@@ -0,0 +1,32 @@
+/*
+ * dhcp_proxy_error.def: dhcp proxy errors
+ *
+ * Copyright (c) 2013 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+dhcp_proxy_error (NONE, "no error")
+dhcp_proxy_error (NO_SERVER, "no dhcp server configured")
+dhcp_proxy_error (RELAY_TO_SERVER, "DHCP packets relayed to the server")
+dhcp_proxy_error (RELAY_TO_CLIENT, "DHCP packets relayed to clients")
+dhcp_proxy_error (OPTION_82_ERROR, "DHCP failed to insert option 82")
+dhcp_proxy_error (NO_OPTION_82, "DHCP option 82 missing")
+dhcp_proxy_error (BAD_OPTION_82_ITF, "Bad DHCP option 82 interface value")
+dhcp_proxy_error (BAD_OPTION_82_ADDR, "Bad DHCP option 82 address value")
+dhcp_proxy_error (BAD_FIB_ID, "DHCP option 82 fib-id to fib-index map failure")
+dhcp_proxy_error (NO_INTERFACE_ADDRESS, "DHCP no interface address")
+dhcp_proxy_error (OPTION_82_VSS_NOT_PROCESSED, "DHCP VSS not processed by DHCP server")
+dhcp_proxy_error (BAD_YIADDR, "DHCP packets with bad your_ip_address fields")
+dhcp_proxy_error (BAD_SVR_FIB_OR_ADDRESS, "DHCP packets not from DHCP server or server FIB.")
+dhcp_proxy_error (PKT_TOO_BIG, "DHCP packets which are too big.")
+
diff --git a/src/vnet/dhcp/dhcp4_proxy_node.c b/src/vnet/dhcp/dhcp4_proxy_node.c
new file mode 100644
index 00000000..88a99249
--- /dev/null
+++ b/src/vnet/dhcp/dhcp4_proxy_node.c
@@ -0,0 +1,983 @@
+/*
+ * proxy_node.c: dhcp proxy node processing
+ *
+ * Copyright (c) 2013 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <vlib/vlib.h>
+#include <vnet/pg/pg.h>
+#include <vnet/dhcp/dhcp_proxy.h>
+#include <vnet/dhcp/client.h>
+#include <vnet/fib/ip4_fib.h>
+
+static char * dhcp_proxy_error_strings[] = {
+#define dhcp_proxy_error(n,s) s,
+#include <vnet/dhcp/dhcp4_proxy_error.def>
+#undef dhcp_proxy_error
+};
+
+#define foreach_dhcp_proxy_to_server_input_next \
+  _ (DROP, "error-drop")			\
+  _ (LOOKUP, "ip4-lookup")			\
+  _ (SEND_TO_CLIENT, "dhcp-proxy-to-client")
+
+typedef enum {
+#define _(s,n) DHCP_PROXY_TO_SERVER_INPUT_NEXT_##s,
+  foreach_dhcp_proxy_to_server_input_next
+#undef _
+  DHCP_PROXY_TO_SERVER_INPUT_N_NEXT,
+} dhcp_proxy_to_server_input_next_t;
+
+typedef struct {
+  /* 0 => to server, 1 => to client */
+  int which; 
+  ip4_address_t trace_ip4_address;
+  u32 error;
+  u32 sw_if_index;
+  u32 original_sw_if_index;
+} dhcp_proxy_trace_t;
+
+#define VPP_DHCP_OPTION82_SUB1_SIZE   6
+#define VPP_DHCP_OPTION82_SUB5_SIZE   6
+#define VPP_DHCP_OPTION82_VSS_SIZE    12
+#define VPP_DHCP_OPTION82_SIZE (VPP_DHCP_OPTION82_SUB1_SIZE + \
+                                VPP_DHCP_OPTION82_SUB5_SIZE + \
+                                VPP_DHCP_OPTION82_VSS_SIZE +3)
+
+static vlib_node_registration_t dhcp_proxy_to_server_node;
+static vlib_node_registration_t dhcp_proxy_to_client_node;
+
+static u8 *
+format_dhcp_proxy_trace (u8 * s, va_list * args)
+{
+  CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *);
+  CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *);
+  dhcp_proxy_trace_t * t = va_arg (*args, dhcp_proxy_trace_t *);
+    
+  if (t->which == 0)
+    s = format (s, "DHCP proxy: sent to server %U\n",
+                format_ip4_address, &t->trace_ip4_address, t->error);
+  else
+    s = format (s, "DHCP proxy: broadcast to client from %U\n",
+                format_ip4_address, &t->trace_ip4_address);
+      
+  if (t->error != (u32)~0)
+    s = format (s, "  error: %s\n", dhcp_proxy_error_strings[t->error]);
+
+  s = format (s, "  original_sw_if_index: %d, sw_if_index: %d\n",
+              t->original_sw_if_index, t->sw_if_index);
+  
+  return s;
+}
+
+static u8 *
+format_dhcp_proxy_header_with_length (u8 * s, va_list * args)
+{
+  dhcp_header_t * h = va_arg (*args, dhcp_header_t *);
+  u32 max_header_bytes = va_arg (*args, u32);
+  u32 header_bytes;
+
+  header_bytes = sizeof (h[0]);
+  if (max_header_bytes != 0 && header_bytes > max_header_bytes)
+    return format (s, "dhcp header truncated");
+
+  s = format (s, "DHCP Proxy");
+
+  return s;
+}
+
+static uword
+dhcp_proxy_to_server_input (vlib_main_t * vm,
+                            vlib_node_runtime_t * node,
+                            vlib_frame_t * from_frame)
+{
+  u32 n_left_from, next_index, * from, * to_next;
+  dhcp_proxy_main_t * dpm = &dhcp_proxy_main;
+  from = vlib_frame_vector_args (from_frame);
+  n_left_from = from_frame->n_vectors;
+  u32 pkts_to_server=0, pkts_to_client=0, pkts_no_server=0;
+  u32 pkts_no_interface_address=0;
+  u32 pkts_too_big=0;
+  ip4_main_t * im = &ip4_main;
+
+  next_index = node->cached_next_index;
+
+  while (n_left_from > 0)
+    {
+      u32 n_left_to_next;
+
+      vlib_get_next_frame (vm, node, next_index,
+			   to_next, n_left_to_next);
+
+      while (n_left_from > 0 && n_left_to_next > 0)
+	{
+	  u32 bi0;
+	  vlib_buffer_t * b0;
+          udp_header_t * u0;
+	  dhcp_header_t * h0;
+          ip4_header_t * ip0;
+	  u32 next0;
+          u32 old0, new0;
+          ip_csum_t sum0;
+          u32 error0 = (u32) ~0;
+          u32 sw_if_index = 0;
+          u32 original_sw_if_index = 0;
+          u8  *end = NULL;
+          u32 fib_index;
+          dhcp_server_t * server;
+          u32 rx_sw_if_index;
+          dhcp_option_t *o;
+          u32 len = 0;
+          vlib_buffer_free_list_t *fl;
+
+	  bi0 = from[0];
+	  to_next[0] = bi0;
+	  from += 1;
+	  to_next += 1;
+	  n_left_from -= 1;
+	  n_left_to_next -= 1;
+
+	  b0 = vlib_get_buffer (vm, bi0);
+
+          h0 = vlib_buffer_get_current (b0);
+
+          /* 
+           * udp_local hands us the DHCP header, need udp hdr, 
+           * ip hdr to relay to server
+           */
+          vlib_buffer_advance (b0, -(sizeof(*u0)));
+	  u0 = vlib_buffer_get_current (b0);
+
+          /* This blows. Return traffic has src_port = 67, dst_port = 67 */
+          if (u0->src_port == clib_net_to_host_u16(UDP_DST_PORT_dhcp_to_server))
+            {
+              vlib_buffer_advance (b0, sizeof(*u0));
+              next0 = DHCP_PROXY_TO_SERVER_INPUT_NEXT_SEND_TO_CLIENT;
+              error0 = 0;
+              pkts_to_client++;
+              goto do_enqueue;
+            }
+
+          rx_sw_if_index = vnet_buffer(b0)->sw_if_index[VLIB_RX];
+
+          fib_index = im->fib_index_by_sw_if_index [rx_sw_if_index];
+          server = dhcp_get_server(dpm, fib_index, FIB_PROTOCOL_IP4);
+          
+          if (PREDICT_FALSE (NULL == server))
+            {
+              error0 = DHCP_PROXY_ERROR_NO_SERVER;
+              next0 = DHCP_PROXY_TO_SERVER_INPUT_NEXT_DROP;
+              pkts_no_server++;
+              goto do_trace;
+            }
+          
+          vlib_buffer_advance (b0, -(sizeof(*ip0)));
+          ip0 = vlib_buffer_get_current (b0);
+
+          /* disable UDP checksum */
+          u0->checksum = 0;
+          sum0 = ip0->checksum;
+          old0 = ip0->dst_address.as_u32;
+          new0 = server->dhcp_server.ip4.as_u32;
+          ip0->dst_address.as_u32 = server->dhcp_server.ip4.as_u32;
+          sum0 = ip_csum_update (sum0, old0, new0, 
+                                ip4_header_t /* structure */, 
+                                dst_address /* changed member */);
+          ip0->checksum = ip_csum_fold (sum0);
+
+          sum0 = ip0->checksum;
+          old0 = ip0->src_address.as_u32;
+          new0 = server->dhcp_src_address.ip4.as_u32;
+          ip0->src_address.as_u32 = new0;
+          sum0 = ip_csum_update (sum0, old0, new0, 
+                                ip4_header_t /* structure */, 
+                                src_address /* changed member */);
+          ip0->checksum = ip_csum_fold (sum0);
+
+          /* Send to DHCP server via the configured FIB */
+          vnet_buffer(b0)->sw_if_index[VLIB_TX] =
+            server->server_fib_index;
+
+          h0->gateway_ip_address.as_u32 = server->dhcp_src_address.ip4.as_u32;
+          pkts_to_server++;
+
+          o = (dhcp_option_t *) h0->options;
+              
+          fib_index = im->fib_index_by_sw_if_index 
+              [vnet_buffer(b0)->sw_if_index[VLIB_RX]];
+
+          end = b0->data + b0->current_data + b0->current_length;
+          /* TLVs are not performance-friendly... */
+          while  (o->option != 0xFF /* end of options */ && (u8 *)o < end) 
+              o = (dhcp_option_t *) (((uword) o) + (o->length + 2));
+
+          fl = vlib_buffer_get_free_list (vm, b0->free_list_index);
+          // start write at (option*)o, some packets have padding
+          if (((u8 *)o - (u8 *)b0->data + VPP_DHCP_OPTION82_SIZE) > fl->n_data_bytes)
+          {
+              next0 = DHCP_PROXY_TO_SERVER_INPUT_NEXT_DROP;
+              pkts_too_big++;
+              goto do_trace;
+          }
+
+          if ((o->option == 0xFF)  && ((u8 *)o <= end))
+          {  
+              vnet_main_t *vnm = vnet_get_main();   
+              u16 old_l0, new_l0;
+              ip4_address_t _ia0, * ia0 = &_ia0;
+              dhcp_vss_t *vss;
+              vnet_sw_interface_t *swif;
+              sw_if_index = 0;
+              original_sw_if_index = 0;
+                  
+              original_sw_if_index = sw_if_index = 
+                  vnet_buffer(b0)->sw_if_index[VLIB_RX];
+              swif = vnet_get_sw_interface (vnm, sw_if_index);
+              if (swif->flags & VNET_SW_INTERFACE_FLAG_UNNUMBERED)
+                  sw_if_index = swif->unnumbered_sw_if_index;
+                  
+              /* 
+               * Get the first ip4 address on the [client-side] 
+               * RX interface, if not unnumbered. otherwise use
+               * the loopback interface's ip address.
+               */
+              ia0 = ip4_interface_first_address(&ip4_main, sw_if_index, 0);
+                  
+              if (ia0 == 0)
+              {
+                  error0 = DHCP_PROXY_ERROR_NO_INTERFACE_ADDRESS;
+                  next0 = DHCP_PROXY_TO_SERVER_INPUT_NEXT_DROP;
+                  pkts_no_interface_address++;
+                  goto do_trace;
+              }
+
+              /* Add option 82 */
+              o->option = 82;   /* option 82 */
+              o->length = 12;   /* 12 octets to follow */
+              o->data[0] = 1;   /* suboption 1, circuit ID (=FIB id) */
+              o->data[1] = 4;   /* length of suboption */
+              o->data[2] = (original_sw_if_index >> 24) & 0xFF;
+              o->data[3] = (original_sw_if_index >> 16) & 0xFF;
+              o->data[4] = (original_sw_if_index >> 8)  & 0xFF;
+              o->data[5] = (original_sw_if_index >> 0)  & 0xFF;
+              o->data[6] = 5; /* suboption 5 (client RX intfc address) */
+              o->data[7] = 4; /* length 4 */
+              o->data[8] = ia0->as_u8[0];
+              o->data[9] = ia0->as_u8[1];
+              o->data[10] = ia0->as_u8[2];
+              o->data[11] = ia0->as_u8[3];
+              o->data[12] = 0xFF;
+
+              vss = dhcp_get_vss_info (dpm, fib_index, FIB_PROTOCOL_IP4);
+              if (NULL != vss)
+              {
+                  u32 opt82_fib_id=0, opt82_oui=0;
+
+                  opt82_oui =  vss->oui;
+                  opt82_fib_id =  vss->fib_id;
+
+                  o->data[12] = 151; /* vss suboption */
+                  if (255 == opt82_fib_id) {
+                      o->data[13] = 1;   /* length */
+                      o->data[14] = 255;   /* vss option type */
+                      o->data[15] = 152; /* vss control suboption */
+                      o->data[16] = 0;   /* length */
+                      /* and a new "end-of-options" option (0xff) */
+                      o->data[17] = 0xFF;
+                      o->length += 5;
+                  } else {
+                      o->data[13] = 8;   /* length */
+                      o->data[14] = 1;   /* vss option type */
+                      o->data[15] = (opt82_oui >> 16) & 0xff;
+                      o->data[16] = (opt82_oui >> 8) & 0xff;
+                      o->data[17] = (opt82_oui ) & 0xff;
+                      o->data[18] = (opt82_fib_id >> 24) & 0xff;
+                      o->data[19] = (opt82_fib_id >> 16) & 0xff;
+                      o->data[20] = (opt82_fib_id >> 8) & 0xff;
+                      o->data[21] = (opt82_fib_id) & 0xff;
+                      o->data[22] = 152; /* vss control suboption */
+                      o->data[23] = 0;   /* length */
+                          
+                      /* and a new "end-of-options" option (0xff) */
+                      o->data[24] = 0xFF;
+                      o->length += 12;
+                  }
+              }
+
+              len = o->length + 3;
+              b0->current_length += len;
+              /* Fix IP header length and checksum */
+              old_l0 = ip0->length;
+              new_l0 = clib_net_to_host_u16 (old_l0);
+              new_l0 += len;
+              new_l0 = clib_host_to_net_u16 (new_l0);
+              ip0->length = new_l0;
+              sum0 = ip0->checksum;
+              sum0 = ip_csum_update (sum0, old_l0, new_l0, ip4_header_t,
+                                     length /* changed member */);
+              ip0->checksum = ip_csum_fold (sum0);
+
+              /* Fix UDP length */
+              new_l0 = clib_net_to_host_u16 (u0->length);
+              new_l0 += len;
+              u0->length = clib_host_to_net_u16 (new_l0);
+          } else {
+              vlib_node_increment_counter 
+                  (vm, dhcp_proxy_to_server_node.index,
+                   DHCP_PROXY_ERROR_OPTION_82_ERROR, 1);
+          }
+          
+          next0 = DHCP_PROXY_TO_SERVER_INPUT_NEXT_LOOKUP;
+
+        do_trace:
+          if (PREDICT_FALSE(b0->flags & VLIB_BUFFER_IS_TRACED)) 
+            {
+               dhcp_proxy_trace_t *tr = vlib_add_trace (vm, node, 
+                                                        b0, sizeof (*tr));
+               tr->which = 0; /* to server */
+               tr->error = error0;
+               tr->original_sw_if_index = original_sw_if_index;
+               tr->sw_if_index = sw_if_index;
+               if (next0 == DHCP_PROXY_TO_SERVER_INPUT_NEXT_LOOKUP)
+                 tr->trace_ip4_address.as_u32 = server->dhcp_server.ip4.as_u32;
+            }
+
+        do_enqueue:
+	  vlib_validate_buffer_enqueue_x1 (vm, node, next_index,
+					   to_next, n_left_to_next,
+					   bi0, next0);
+	}
+
+      vlib_put_next_frame (vm, node, next_index, n_left_to_next);
+    }
+
+  vlib_node_increment_counter (vm, dhcp_proxy_to_server_node.index,
+                               DHCP_PROXY_ERROR_RELAY_TO_CLIENT,
+                               pkts_to_client);
+  vlib_node_increment_counter (vm, dhcp_proxy_to_server_node.index,
+                               DHCP_PROXY_ERROR_RELAY_TO_SERVER,
+                               pkts_to_server);
+  vlib_node_increment_counter (vm, dhcp_proxy_to_server_node.index,
+                               DHCP_PROXY_ERROR_NO_SERVER,
+                               pkts_no_server);
+  vlib_node_increment_counter (vm, dhcp_proxy_to_server_node.index,
+                               DHCP_PROXY_ERROR_NO_INTERFACE_ADDRESS,
+                               pkts_no_interface_address);
+ vlib_node_increment_counter (vm, dhcp_proxy_to_server_node.index,
+                              DHCP_PROXY_ERROR_PKT_TOO_BIG,
+                              pkts_too_big);
+  return from_frame->n_vectors;
+}
+
+VLIB_REGISTER_NODE (dhcp_proxy_to_server_node, static) = {
+  .function = dhcp_proxy_to_server_input,
+  .name = "dhcp-proxy-to-server",
+  /* Takes a vector of packets. */
+  .vector_size = sizeof (u32),
+
+  .n_errors = DHCP_PROXY_N_ERROR,
+  .error_strings = dhcp_proxy_error_strings,
+
+  .n_next_nodes = DHCP_PROXY_TO_SERVER_INPUT_N_NEXT,
+  .next_nodes = {
+#define _(s,n) [DHCP_PROXY_TO_SERVER_INPUT_NEXT_##s] = n,
+    foreach_dhcp_proxy_to_server_input_next
+#undef _
+  },
+
+  .format_buffer = format_dhcp_proxy_header_with_length,
+  .format_trace = format_dhcp_proxy_trace,
+#if 0
+  .unformat_buffer = unformat_dhcp_proxy_header,
+#endif
+};
+
+static uword
+dhcp_proxy_to_client_input (vlib_main_t * vm,
+                            vlib_node_runtime_t * node,
+                            vlib_frame_t * from_frame)
+{
+  u32 n_left_from, * from;
+  ethernet_main_t *em = ethernet_get_main (vm);
+  dhcp_proxy_main_t * dpm = &dhcp_proxy_main;
+  vnet_main_t * vnm = vnet_get_main();
+  ip4_main_t * im = &ip4_main;
+
+  from = vlib_frame_vector_args (from_frame);
+  n_left_from = from_frame->n_vectors;
+
+  while (n_left_from > 0)
+    {
+      u32 bi0;
+      vlib_buffer_t * b0;
+      udp_header_t * u0;
+      dhcp_header_t * h0;
+      ip4_header_t * ip0 = 0;
+      ip4_address_t * ia0 = 0;
+      u32 old0, new0;
+      ip_csum_t sum0;
+      ethernet_interface_t *ei0;
+      ethernet_header_t *mac0;
+      vnet_hw_interface_t *hi0;
+      vlib_frame_t *f0;
+      u32 * to_next0;
+      u32 sw_if_index = ~0;
+      vnet_sw_interface_t *si0;
+      u32 error0 = (u32)~0;
+      vnet_sw_interface_t *swif;
+      u32 fib_index;
+      dhcp_server_t * server;
+      u32 original_sw_if_index = (u32) ~0;
+      ip4_address_t relay_addr = {
+          .as_u32 = 0,
+      };
+
+      bi0 = from[0];
+      from += 1;
+      n_left_from -= 1;
+
+      b0 = vlib_get_buffer (vm, bi0);
+      h0 = vlib_buffer_get_current (b0);
+
+      /* 
+       * udp_local hands us the DHCP header, need udp hdr, 
+       * ip hdr to relay to client
+       */
+      vlib_buffer_advance (b0, -(sizeof(*u0)));
+      u0 = vlib_buffer_get_current (b0);
+
+      vlib_buffer_advance (b0, -(sizeof(*ip0)));
+      ip0 = vlib_buffer_get_current (b0);
+
+      /* Consumed by dhcp client code? */
+      if (dhcp_client_for_us (bi0, b0, ip0, u0, h0))
+          continue;
+
+      if (1 /* dpm->insert_option_82 */)
+        {
+          dhcp_option_t *o = (dhcp_option_t *) h0->options;
+          dhcp_option_t *sub;
+              
+          /* Parse through TLVs looking for option 82.
+             The circuit-ID is the FIB number we need
+             to track down the client-facing interface */
+
+          while (o->option != 0xFF /* end of options */ &&
+                 (u8 *) o < (b0->data + b0->current_data + b0->current_length))
+            {
+              if (o->option == 82)
+                {
+                    u32 vss_exist = 0;
+                    u32 vss_ctrl = 0;
+                    sub = (dhcp_option_t *) &o->data[0];
+                    while (sub->option != 0xFF /* end of options */ &&
+                           (u8 *) sub < (u8 *)(o + o->length)) {
+                        /* If this is one of ours, it will have
+                           total length 12, circuit-id suboption type,
+                           and the sw_if_index */
+                        if (sub->option == 1 && sub->length == 4)
+                          {
+                            sw_if_index = ((sub->data[0] << 24) |
+                                           (sub->data[1] << 16) |
+                                           (sub->data[2] << 8)  |
+                                           (sub->data[3]));
+                          }
+                        else if (sub->option == 5 && sub->length == 4)
+                          {
+                              relay_addr.as_u8[0] = sub->data[0];
+                              relay_addr.as_u8[1] = sub->data[1];
+                              relay_addr.as_u8[2] = sub->data[2];
+                              relay_addr.as_u8[3] = sub->data[3];
+                          }
+                        else if (sub->option == 151 &&
+                                 sub->length == 7 &&
+                                 sub->data[0] == 1)
+                            vss_exist = 1;
+                        else if (sub->option == 152 && sub->length == 0)
+                            vss_ctrl = 1;
+                        sub = (dhcp_option_t *) 
+                          (((uword) sub) + (sub->length + 2));
+                    }
+                    if (vss_ctrl && vss_exist)
+                      vlib_node_increment_counter 
+                        (vm, dhcp_proxy_to_client_node.index,
+                         DHCP_PROXY_ERROR_OPTION_82_VSS_NOT_PROCESSED, 1);
+
+                }
+              o = (dhcp_option_t *) (((uword) o) + (o->length + 2));
+            }
+        }
+
+      if (sw_if_index == (u32)~0)
+        {
+          error0 = DHCP_PROXY_ERROR_NO_OPTION_82;
+          
+        drop_packet:
+          vlib_node_increment_counter (vm, dhcp_proxy_to_client_node.index,
+                                       error0, 1);
+          f0 = vlib_get_frame_to_node (vm, dpm->error_drop_node_index);
+          to_next0 = vlib_frame_vector_args (f0);
+          to_next0[0] = bi0;
+          f0->n_vectors = 1;
+          vlib_put_frame_to_node (vm, dpm->error_drop_node_index, f0);
+          goto do_trace;
+        }
+      
+      if (relay_addr.as_u32 == 0)
+        {
+          error0 = DHCP_PROXY_ERROR_BAD_OPTION_82_ADDR;
+          goto drop_packet;
+        }
+
+      if (sw_if_index >= vec_len (im->fib_index_by_sw_if_index))
+        {
+          error0 = DHCP_PROXY_ERROR_BAD_OPTION_82_ITF;
+          goto drop_packet;
+        }
+
+      fib_index = im->fib_index_by_sw_if_index [sw_if_index];
+      server = dhcp_get_server(dpm, fib_index, FIB_PROTOCOL_IP4);
+
+      if (PREDICT_FALSE (NULL == server))
+        {
+          error0 = DHCP_PROXY_ERROR_NO_SERVER;
+          goto drop_packet;
+        }
+      
+      if (ip0->src_address.as_u32 != server->dhcp_server.ip4.as_u32)
+        {             
+          error0 = DHCP_PROXY_ERROR_BAD_SVR_FIB_OR_ADDRESS;
+          goto drop_packet;
+        }
+
+      vnet_buffer (b0)->sw_if_index[VLIB_TX] = sw_if_index;
+
+      swif = vnet_get_sw_interface (vnm, sw_if_index);
+      original_sw_if_index = sw_if_index;
+      if (swif->flags & VNET_SW_INTERFACE_FLAG_UNNUMBERED)
+          sw_if_index = swif->unnumbered_sw_if_index;
+
+      ia0 = ip4_interface_first_address (&ip4_main, sw_if_index, 0);
+      if (ia0 == 0)
+        {
+          error0 = DHCP_PROXY_ERROR_NO_INTERFACE_ADDRESS;
+          goto drop_packet;
+        }
+
+      if (relay_addr.as_u32 != ia0->as_u32)
+        {             
+          error0 = DHCP_PROXY_ERROR_BAD_YIADDR;
+          goto drop_packet;
+        }
+
+      u0->checksum = 0;
+      u0->dst_port = clib_net_to_host_u16 (UDP_DST_PORT_dhcp_to_client);
+      sum0 = ip0->checksum;
+      old0 = ip0->dst_address.as_u32;
+      new0 = 0xFFFFFFFF;
+      ip0->dst_address.as_u32 = new0;
+      sum0 = ip_csum_update (sum0, old0, new0, 
+                            ip4_header_t /* structure */, 
+                            dst_address /* offset of changed member */);
+      ip0->checksum = ip_csum_fold (sum0);
+
+      sum0 = ip0->checksum;
+      old0 = ip0->src_address.as_u32;
+      new0 = ia0->as_u32;
+      ip0->src_address.as_u32 = new0;
+      sum0 = ip_csum_update (sum0, old0, new0, 
+                            ip4_header_t /* structure */, 
+                            src_address /* offset of changed member */);
+      ip0->checksum = ip_csum_fold (sum0);
+
+      vlib_buffer_advance (b0, -(sizeof(ethernet_header_t)));
+      si0 = vnet_get_sw_interface (vnm, original_sw_if_index);
+      if (si0->type == VNET_SW_INTERFACE_TYPE_SUB)
+	  vlib_buffer_advance (b0, -4 /* space for VLAN tag */);
+
+      mac0 = vlib_buffer_get_current (b0);
+
+      hi0 = vnet_get_sup_hw_interface (vnm, original_sw_if_index);
+      ei0 = pool_elt_at_index (em->interfaces, hi0->hw_instance);
+      clib_memcpy (mac0->src_address, ei0->address, sizeof (ei0->address));
+      memset (mac0->dst_address, 0xff, sizeof (mac0->dst_address));
+      mac0->type = (si0->type == VNET_SW_INTERFACE_TYPE_SUB) ?
+	clib_net_to_host_u16(0x8100) : clib_net_to_host_u16 (0x0800);
+
+      if (si0->type == VNET_SW_INTERFACE_TYPE_SUB)
+	{
+	  u32 * vlan_tag = (u32 *)(mac0+1);
+	  u32 tmp;
+	  tmp = (si0->sub.id << 16) | 0x0800;
+	  *vlan_tag = clib_host_to_net_u32 (tmp);
+	}
+
+      /* $$$ This needs to be rewritten, for sure */
+      f0 = vlib_get_frame_to_node (vm, hi0->output_node_index);
+      to_next0 = vlib_frame_vector_args (f0);
+      to_next0[0] = bi0;
+      f0->n_vectors = 1;
+      vlib_put_frame_to_node (vm, hi0->output_node_index, f0);
+
+    do_trace:
+      if (PREDICT_FALSE(b0->flags & VLIB_BUFFER_IS_TRACED)) 
+        {
+          dhcp_proxy_trace_t *tr = vlib_add_trace (vm, node, 
+                                                   b0, sizeof (*tr));
+          tr->which = 1; /* to client */
+          tr->trace_ip4_address.as_u32 = ia0 ? ia0->as_u32 : 0;
+          tr->error = error0;
+          tr->original_sw_if_index = original_sw_if_index;
+          tr->sw_if_index = sw_if_index;
+        }
+    }
+  return from_frame->n_vectors;
+}
+
+VLIB_REGISTER_NODE (dhcp_proxy_to_client_node, static) = {
+  .function = dhcp_proxy_to_client_input,
+  .name = "dhcp-proxy-to-client",
+  /* Takes a vector of packets. */
+  .vector_size = sizeof (u32),
+
+  .n_errors = DHCP_PROXY_N_ERROR,
+  .error_strings = dhcp_proxy_error_strings,
+  .format_buffer = format_dhcp_proxy_header_with_length,
+  .format_trace = format_dhcp_proxy_trace,
+#if 0
+  .unformat_buffer = unformat_dhcp_proxy_header,
+#endif
+};
+
+static clib_error_t *
+dhcp4_proxy_init (vlib_main_t * vm)
+{
+  dhcp_proxy_main_t * dm = &dhcp_proxy_main;
+  vlib_node_t * error_drop_node;
+
+  error_drop_node = vlib_get_node_by_name (vm, (u8 *) "error-drop");
+  dm->error_drop_node_index = error_drop_node->index;
+
+  udp_register_dst_port (vm, UDP_DST_PORT_dhcp_to_client, 
+                         dhcp_proxy_to_client_node.index, 1 /* is_ip4 */);
+
+  udp_register_dst_port (vm, UDP_DST_PORT_dhcp_to_server, 
+                         dhcp_proxy_to_server_node.index, 1 /* is_ip4 */);
+
+  return 0;
+}
+
+
+VLIB_INIT_FUNCTION (dhcp4_proxy_init);
+
+int
+dhcp4_proxy_set_server (ip46_address_t *addr,
+                        ip46_address_t *src_addr,
+                        u32 rx_table_id,
+                        u32 server_table_id, 
+                        int is_del)
+{
+  u32 rx_fib_index = 0;
+  int rc = 0;
+
+  const fib_prefix_t all_1s =
+  {
+      .fp_len = 32,
+      .fp_addr.ip4.as_u32 = 0xffffffff,
+      .fp_proto = FIB_PROTOCOL_IP4,
+  };
+
+  if (ip46_address_is_zero(addr))
+    return VNET_API_ERROR_INVALID_DST_ADDRESS;
+  
+  if (ip46_address_is_zero(src_addr))
+    return VNET_API_ERROR_INVALID_SRC_ADDRESS;
+
+  rx_fib_index = fib_table_find_or_create_and_lock(FIB_PROTOCOL_IP4,
+                                                   rx_table_id);
+
+  if (is_del)
+    {
+      rc = dhcp_proxy_server_del (FIB_PROTOCOL_IP4, rx_fib_index);
+
+      if (0 == rc)
+      {
+          fib_table_entry_special_remove(rx_fib_index,
+                                         &all_1s,
+                                         FIB_SOURCE_DHCP);
+          fib_table_unlock (rx_fib_index, FIB_PROTOCOL_IP4);
+      }
+    }
+  else
+  {
+      if (dhcp_proxy_server_add (FIB_PROTOCOL_IP4,
+                                 addr, src_addr,
+                                 rx_fib_index, server_table_id))
+      {
+          fib_table_entry_special_add(rx_fib_index,
+                                      &all_1s,
+                                      FIB_SOURCE_DHCP,
+                                      FIB_ENTRY_FLAG_LOCAL,
+                                      ADJ_INDEX_INVALID);
+          fib_table_lock (rx_fib_index, FIB_PROTOCOL_IP4);
+      }
+  }
+  fib_table_unlock (rx_fib_index, FIB_PROTOCOL_IP4);
+
+  return (rc);
+}
+
+static clib_error_t *
+dhcp4_proxy_set_command_fn (vlib_main_t * vm,
+                            unformat_input_t * input,
+                            vlib_cli_command_t * cmd)
+{
+  ip46_address_t server_addr, src_addr;
+  u32 server_table_id = 0, rx_table_id = 0;
+  int is_del = 0;
+  int set_src = 0, set_server = 0;
+
+  memset(&server_addr, 0, sizeof(server_addr));
+  memset(&src_addr, 0, sizeof(src_addr));
+
+  while (unformat_check_input(input) != UNFORMAT_END_OF_INPUT) 
+    {
+      if (unformat (input, "server %U", 
+                    unformat_ip4_address, &server_addr.ip4)) 
+        set_server = 1;
+      else if (unformat (input, "server-fib-id %d", &server_table_id))
+        ;
+      else if (unformat (input, "rx-fib-id %d", &rx_table_id))
+        ;
+      else if (unformat(input, "src-address %U", 
+			unformat_ip4_address, &src_addr.ip4))
+        set_src = 1;
+      else if (unformat (input, "delete") ||
+               unformat (input, "del"))
+        is_del = 1;
+      else
+        break;
+    }
+
+  if (is_del || (set_server && set_src))
+    {
+      int rv;
+
+      rv = dhcp4_proxy_set_server (&server_addr, &src_addr, rx_table_id, 
+                                   server_table_id, is_del);
+      switch (rv)
+        {
+        case 0:
+          return 0;
+
+        case VNET_API_ERROR_INVALID_DST_ADDRESS:
+          return clib_error_return (0, "Invalid server address");
+          
+        case VNET_API_ERROR_INVALID_SRC_ADDRESS:
+          return clib_error_return (0, "Invalid src address");
+
+        case VNET_API_ERROR_NO_SUCH_ENTRY:
+          return clib_error_return 
+            (0, "Fib id %d: no per-fib DHCP server configured", rx_table_id);
+
+        default:
+          return clib_error_return (0, "BUG: rv %d", rv);
+        }
+    }
+  else
+    return clib_error_return (0, "parse error`%U'",
+                              format_unformat_error, input);
+}
+
+VLIB_CLI_COMMAND (dhcp_proxy_set_command, static) = {
+  .path = "set dhcp proxy",
+  .short_help = "set dhcp proxy [del] server <ip-addr> src-address <ip-addr> [server-fib-id <n>] [rx-fib-id <n>]",
+  .function = dhcp4_proxy_set_command_fn,
+};
+
+static u8 *
+format_dhcp4_proxy_server (u8 * s, va_list * args)
+{
+  dhcp_server_t * server = va_arg (*args, dhcp_server_t *);
+  ip4_fib_t * rx_fib, * server_fib;
+
+  if (server == 0)
+    {
+      s = format (s, "%=16s%=16s%=14s%=14s", "Server", "Src Address", 
+                  "Server FIB", "RX FIB");
+      return s;
+    }
+
+  server_fib = ip4_fib_get(server->server_fib_index);
+  rx_fib = ip4_fib_get(server->rx_fib_index);
+
+  s = format (s, "%=16U%=16U%=14u%=14u",
+              format_ip46_address, &server->dhcp_server, IP46_TYPE_ANY,
+              format_ip46_address, &server->dhcp_src_address, IP46_TYPE_ANY,
+              server_fib->table_id,
+              rx_fib->table_id);
+  return s;
+}
+
+static int
+dhcp4_proxy_show_walk (dhcp_server_t *server,
+                       void *ctx)
+{
+    vlib_main_t * vm = ctx;
+
+    vlib_cli_output (vm, "%U", format_dhcp4_proxy_server, server);
+
+    return (1);
+}
+
+static clib_error_t *
+dhcp4_proxy_show_command_fn (vlib_main_t * vm,
+                             unformat_input_t * input,
+                             vlib_cli_command_t * cmd)
+{
+  vlib_cli_output (vm, "%U", format_dhcp4_proxy_server, NULL /* header line */);
+
+  dhcp_proxy_walk(FIB_PROTOCOL_IP4, dhcp4_proxy_show_walk, vm);
+
+  return (NULL);
+}
+
+VLIB_CLI_COMMAND (dhcp_proxy_show_command, static) = {
+  .path = "show dhcp proxy",
+  .short_help = "Display dhcp proxy server info",
+  .function = dhcp4_proxy_show_command_fn,
+};
+
+static clib_error_t *
+dhcp_option_82_vss_fn (vlib_main_t * vm,
+                        unformat_input_t * input,
+                        vlib_cli_command_t * cmd)
+{
+  int is_del = 0, got_new_vpn_id=0;
+  u32 oui=0, fib_id=0, tbl_id=~0;
+
+  while (unformat_check_input(input) != UNFORMAT_END_OF_INPUT) 
+    {
+
+      if (unformat(input, "delete") || unformat(input, "del"))
+          is_del = 1;    
+      else if (unformat (input, "oui %d", &oui))
+          got_new_vpn_id = 1;
+      else if (unformat (input, "vpn-id %d", &fib_id))
+          got_new_vpn_id = 1;
+      else if (unformat (input, "table %d", &tbl_id))
+          got_new_vpn_id = 1;
+      else
+          break;
+  }
+  if (tbl_id == ~0)
+      return clib_error_return (0, "no table ID specified.");
+  
+  if (is_del || got_new_vpn_id)
+    {
+      int rv;
+      rv = dhcp_proxy_set_vss(FIB_PROTOCOL_IP4, tbl_id, oui, fib_id, is_del);
+      switch (rv)
+        {
+        case 0:
+            return 0;
+            
+        case VNET_API_ERROR_NO_SUCH_FIB:
+            return clib_error_return (0, "option 82 vss(oui:%d, vpn-id:%d) not found in table %d",
+                                      oui, fib_id, tbl_id);
+            
+        case VNET_API_ERROR_NO_SUCH_ENTRY:
+            return clib_error_return (0, "option 82 vss for table %d not found in in pool.",
+                                      tbl_id);
+        default:
+          return clib_error_return (0, "BUG: rv %d", rv);
+        }
+    }
+  else
+      return clib_error_return (0, "parse error`%U'",
+                                format_unformat_error, input);
+}
+
+VLIB_CLI_COMMAND (dhcp_proxy_vss_command,static) = {
+  .path = "set dhcp option-82 vss",
+  .short_help = "set dhcp option-82 vss [del] table <table id> oui <oui> vpn-id <vpn-id>",
+  .function = dhcp_option_82_vss_fn,
+};
+
+static clib_error_t *
+dhcp_vss_show_command_fn (vlib_main_t * vm,
+                          unformat_input_t * input,
+                          vlib_cli_command_t * cmd)
+  
+{
+  dhcp_vss_walk(FIB_PROTOCOL_IP4, dhcp_vss_show_walk, vm);
+
+  return (NULL);
+}
+
+VLIB_CLI_COMMAND (dhcp_proxy_vss_show_command, static) = {
+  .path = "show dhcp vss",
+  .short_help = "show dhcp VSS",
+  .function = dhcp_vss_show_command_fn,
+};
+
+static clib_error_t *
+dhcp_option_82_address_show_command_fn (vlib_main_t * vm,
+                                unformat_input_t * input,
+                                vlib_cli_command_t * cmd)
+  
+{
+  vnet_main_t *vnm = vnet_get_main();                                     
+  u32 sw_if_index0=0, sw_if_index;
+  vnet_sw_interface_t *swif;
+  ip4_address_t *ia0;
+  
+  while (unformat_check_input(input) != UNFORMAT_END_OF_INPUT) 
+    {
+      
+      if (unformat(input, "%U",
+                   unformat_vnet_sw_interface, vnm, &sw_if_index0))
+        {
+          swif = vnet_get_sw_interface (vnm, sw_if_index0);
+          sw_if_index = (swif->flags & VNET_SW_INTERFACE_FLAG_UNNUMBERED) ?
+            swif->unnumbered_sw_if_index : sw_if_index0;
+          ia0 = ip4_interface_first_address(&ip4_main, sw_if_index, 0);
+          if (ia0)
+            {
+              vlib_cli_output (vm, "%=20s%=20s", "interface", 
+                               "source IP address");
+              
+              vlib_cli_output (vm, "%=20U%=20U",
+                               format_vnet_sw_if_index_name, 
+                               vnm, sw_if_index0,
+                               format_ip4_address, ia0);
+            }
+          else
+            vlib_cli_output (vm, "%=34s %=20U", 
+                             "No IPv4 address configured on",
+                             format_vnet_sw_if_index_name, 
+                             vnm, sw_if_index);
+        }
+      else
+        break;
+    }
+  
+  return 0;
+}
+
+VLIB_CLI_COMMAND (dhcp_proxy_address_show_command,static) = {
+  .path = "show dhcp option-82-address interface",
+  .short_help = "show dhcp option-82-address interface <interface>",
+  .function = dhcp_option_82_address_show_command_fn,
+};
diff --git a/src/vnet/dhcp/dhcp6_packet.h b/src/vnet/dhcp/dhcp6_packet.h
new file mode 100644
index 00000000..ddcde7a0
--- /dev/null
+++ b/src/vnet/dhcp/dhcp6_packet.h
@@ -0,0 +1,183 @@
+#ifndef included_vnet_dhcp6_packet_h
+#define included_vnet_dhcp6_packet_h
+
+/*
+ * DHCP packet format
+ *
+ * Copyright (c) 2013 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+#include <vnet/ip/ip6_packet.h>
+
+// #define DHCP_VRF_NAME_MAX_LEN L3VM_MAX_NAME_STR_LEN
+// #define DHCPV6_MAX_VRF_NAME_LEN L3VM_MAX_NAME_STR_LEN
+#define DHCP_MAX_RELAY_ADDR    16
+#define PROTO_UDP 17
+#define DHCPV6_CLIENT_PORT 546
+#define DHCPV6_SERVER_PORT 547
+#define HOP_COUNT_LIMIT  32
+#define DHCPV6_CISCO_ENT_NUM 9
+
+/*
+ * DHCPv6 message types
+ */
+typedef enum dhcpv6_msg_type_{
+    DHCPV6_MSG_SOLICIT              = 1,
+    DHCPV6_MSG_ADVERTISE            = 2,
+    DHCPV6_MSG_REQUEST              = 3,
+    DHCPV6_MSG_CONFIRM              = 4,
+    DHCPV6_MSG_RENEW                = 5,
+    DHCPV6_MSG_REBIND               = 6,
+    DHCPV6_MSG_REPLY                = 7,
+    DHCPV6_MSG_RELEASE              = 8,
+    DHCPV6_MSG_DECLINE              = 9,
+    DHCPV6_MSG_RECONFIGURE          = 10,
+    DHCPV6_MSG_INFORMATION_REQUEST  = 11,
+    DHCPV6_MSG_RELAY_FORW           = 12,
+    DHCPV6_MSG_RELAY_REPL           = 13,
+} dhcpv6_msg_type_t;
+
+/*
+ * DHCPv6 options types
+ */
+enum {
+    DHCPV6_OPTION_CLIENTID      = 1,
+    DHCPV6_OPTION_SERVERID      = 2,
+    DHCPV6_OPTION_IA_NA         = 3,
+    DHCPV6_OPTION_IA_TA         = 4,
+    DHCPV6_OPTION_IAADDR        = 5,
+    DHCPV6_OPTION_ORO           = 6,
+    DHCPV6_OPTION_PREFERENCE    = 7,
+    DHCPV6_OPTION_ELAPSED_TIME  = 8,
+    DHCPV6_OPTION_RELAY_MSG     = 9,
+    DHCPV6_OPTION_AUTH          = 11,
+    DHCPV6_OPTION_UNICAST       = 12,
+    DHCPV6_OPTION_STATUS_CODE   = 13,
+    DHCPV6_OPTION_RAPID_COMMIT  = 14,
+    DHCPV6_OPTION_USER_CLASS    = 15,
+    DHCPV6_OPTION_VENDOR_CLASS  = 16,
+    DHCPV6_OPTION_VENDOR_OPTS   = 17,
+    DHCPV6_OPTION_INTERFACE_ID  = 18, // relay agent fills this
+    DHCPV6_OPTION_RECONF_MSG    = 19,
+    DHCPV6_OPTION_RECONF_ACCEPT = 20,
+    DHCPV6_OPTION_REMOTEID      = 37, // relay agent fills this
+    DHCPV6_OPTION_VSS      = 68, // relay agent fills this
+    DHCPV6_OPTION_CLIENT_LINK_LAYER_ADDRESS = 79, 
+    DHCPV6_OPTION_MAX
+};
+
+/*
+* DHCPv6 status codes
+ */
+enum {
+    DHCPV6_STATUS_SUCCESS       = 0,
+    DHCPV6_STATUS_UNSPEC_FAIL   = 1,
+    DHCPV6_STATUS_NOADDRS_AVAIL = 2,
+    DHCPV6_STATUS_NO_BINDING    = 3,
+    DHCPV6_STATUS_NOT_ONLINK    = 4,
+    DHCPV6_STATUS_USE_MULTICAST = 5,
+};
+
+/*
+ * DHCPv6 DUID types
+ */
+enum {
+    DHCPV6_DUID_LLT =   1,  /* DUID Based on Link-layer Address Plus Time */
+    DHCPV6_DUID_EN =    2,  /* DUID Based on Enterprise Number */
+    DHCPV6_DUID_LL =    3,  /* DUID Based on Link-layer Address */
+};
+
+//Structure for DHCPv6 payload from client
+typedef struct dhcpv6_hdr_ {
+    union {
+        u8 msg_type; //DHCP msg type
+        u32 xid;     // transaction id
+    }u;
+    u8  data[0];
+} dhcpv6_header_t;
+
+
+
+typedef CLIB_PACKED (struct dhcpv6_relay_ctx_ {
+    dhcpv6_header_t *pkt;
+    u32  pkt_len;
+    u32  dhcpv6_len; //DHCPv6 payload load
+//    if_ordinal iod;
+    u32 if_index;
+    u32 ctx_id;
+    char ctx_name[32+1];
+    u8 dhcp_msg_type;
+}) dhcpv6_relay_ctx_t;
+
+//Structure for DHCPv6 RELAY-FORWARD and DHCPv6 RELAY-REPLY pkts
+typedef CLIB_PACKED (struct dhcpv6_relay_hdr_ {
+    u8           msg_type;
+    u8           hop_count;
+    ip6_address_t    link_addr;
+    ip6_address_t    peer_addr;
+    u8           data[0];
+}) dhcpv6_relay_hdr_t;
+
+typedef enum dhcp_stats_action_type_ {
+    DHCP_STATS_ACTION_FORWARDED=1,
+    DHCP_STATS_ACTION_RECEIVED,
+    DHCP_STATS_ACTION_DROPPED
+} dhcp_stats_action_type_t;
+//Generic counters for a packet
+typedef  struct dhcp_stats_counters_ {
+    u64  rx_pkts; //counter for received pkts
+    u64  tx_pkts; //counter for forwarded pkts
+    u64  drops;   //counter for dropped pkts
+} dhcp_stats_counters_t;
+
+
+typedef enum dhcpv6_stats_drop_reason_ {
+    DHCPV6_RELAY_PKT_DROP_RELAYDISABLE = 1,
+    DHCPV6_RELAY_PKT_DROP_MAX_HOPS,
+    DHCPV6_RELAY_PKT_DROP_VALIDATION_FAIL,
+    DHCPV6_RELAY_PKT_DROP_UNKNOWN_OP_INTF,
+    DHCPV6_RELAY_PKT_DROP_BAD_CONTEXT,
+    DHCPV6_RELAY_PKT_DROP_OPT_INSERT_FAIL,
+    DHCPV6_RELAY_PKT_DROP_REPLY_FROM_CLIENT,
+} dhcpv6_stats_drop_reason_t;
+
+typedef CLIB_PACKED (struct {
+  u16 option;
+  u16 length;
+  u8 data[0];
+}) dhcpv6_option_t;
+
+typedef CLIB_PACKED (struct {
+  dhcpv6_option_t opt;
+  u32 int_idx;
+}) dhcpv6_int_id_t;
+
+typedef CLIB_PACKED (struct {
+  dhcpv6_option_t opt;
+  u8 data[8];  // data[0]:type, data[1..7]: VPN ID
+}) dhcpv6_vss_t;
+
+typedef CLIB_PACKED (struct {
+  dhcpv6_option_t opt;
+  u32 ent_num;
+  u32 rmt_id;
+}) dhcpv6_rmt_id_t;
+
+typedef CLIB_PACKED (struct {
+  dhcpv6_option_t opt;
+  u16 link_type;
+  u8 data[6];  // data[0]:data[5]: MAC address
+}) dhcpv6_client_mac_t;
+
+
+#endif /* included_vnet_dhcp6_packet_h */
diff --git a/src/vnet/dhcp/dhcp6_proxy_error.def b/src/vnet/dhcp/dhcp6_proxy_error.def
new file mode 100644
index 00000000..55fa7317
--- /dev/null
+++ b/src/vnet/dhcp/dhcp6_proxy_error.def
@@ -0,0 +1,29 @@
+/*
+ * dhcp_proxy_error.def: dhcp proxy errors
+ *
+ * Copyright (c) 2013 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+dhcpv6_proxy_error (NONE, "no error")
+dhcpv6_proxy_error (NO_SERVER, "no dhcpv6 server configured")
+dhcpv6_proxy_error (RELAY_TO_SERVER, "DHCPV6 packets relayed to the server")
+dhcpv6_proxy_error (RELAY_TO_CLIENT, "DHCPV6 packets relayed to clients")
+dhcpv6_proxy_error (NO_INTERFACE_ADDRESS, "DHCPV6 no interface address")
+dhcpv6_proxy_error (WRONG_MESSAGE_TYPE, "DHCPV6 wrong message type.")
+dhcpv6_proxy_error (NO_SRC_ADDRESS, "DHCPV6 no srouce IPv6 address configured.")
+dhcpv6_proxy_error (NO_CIRCUIT_ID_OPTION, "DHCPv6 reply packets without circuit ID option")
+dhcpv6_proxy_error (NO_RELAY_MESSAGE_OPTION, "DHCPv6 reply packets without relay message option")
+dhcpv6_proxy_error (BAD_SVR_FIB_OR_ADDRESS, "DHCPv6 packets not from DHCPv6 server or server FIB.")
+dhcpv6_proxy_error (PKT_TOO_BIG, "DHCPv6 packets which are too big.")
+dhcpv6_proxy_error (WRONG_INTERFACE_ID_OPTION, "DHCPv6 reply to invalid interface.")
diff --git a/src/vnet/dhcp/dhcp6_proxy_node.c b/src/vnet/dhcp/dhcp6_proxy_node.c
new file mode 100644
index 00000000..ed44977d
--- /dev/null
+++ b/src/vnet/dhcp/dhcp6_proxy_node.c
@@ -0,0 +1,1065 @@
+/*
+ * dhcp6_proxy_node.c: dhcpv6 proxy node processing
+ *
+ * Copyright (c) 2013 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <vlib/vlib.h>
+#include <vnet/pg/pg.h>
+#include <vnet/dhcp/dhcp_proxy.h>
+#include <vnet/dhcp/dhcp6_packet.h>
+#include <vnet/fib/ip6_fib.h>
+#include <vnet/mfib/mfib_table.h>
+#include <vnet/mfib/ip6_mfib.h>
+
+static char * dhcpv6_proxy_error_strings[] = {
+#define dhcpv6_proxy_error(n,s) s,
+#include <vnet/dhcp/dhcp6_proxy_error.def>
+#undef dhcpv6_proxy_error
+};
+
+#define foreach_dhcpv6_proxy_to_server_input_next \
+  _ (DROP, "error-drop")			\
+  _ (LOOKUP, "ip6-lookup")                      \
+  _ (SEND_TO_CLIENT, "dhcpv6-proxy-to-client")
+
+
+typedef enum {
+#define _(s,n) DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_##s,
+  foreach_dhcpv6_proxy_to_server_input_next
+#undef _
+  DHCPV6_PROXY_TO_SERVER_INPUT_N_NEXT,
+} dhcpv6_proxy_to_server_input_next_t;
+
+typedef struct {
+  /* 0 => to server, 1 => to client */
+  int which;
+  u8 packet_data[64];
+  u32 error;
+  u32 sw_if_index;
+  u32 original_sw_if_index;
+} dhcpv6_proxy_trace_t;
+
+static vlib_node_registration_t dhcpv6_proxy_to_server_node;
+static vlib_node_registration_t dhcpv6_proxy_to_client_node;
+
+/* all DHCP servers address */
+static ip6_address_t all_dhcpv6_server_address;
+static ip6_address_t all_dhcpv6_server_relay_agent_address;
+
+static u8 *
+format_dhcpv6_proxy_trace (u8 * s, va_list * args)
+{
+  CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *);
+  CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *);
+  dhcpv6_proxy_trace_t * t = va_arg (*args, dhcpv6_proxy_trace_t *);
+
+  if (t->which == 0)
+    s = format (s, "DHCPV6 proxy: sent to server %U",
+                format_ip6_address, &t->packet_data, sizeof (ip6_address_t));
+  else
+    s = format (s, "DHCPV6 proxy: sent to client from %U",
+                format_ip6_address, &t->packet_data, sizeof (ip6_address_t));
+  if (t->error != (u32)~0)
+    s = format (s, " error: %s\n", dhcpv6_proxy_error_strings[t->error]);
+
+  s = format (s, "  original_sw_if_index: %d, sw_if_index: %d\n",
+              t->original_sw_if_index, t->sw_if_index);
+
+  return s;
+}
+
+static u8 *
+format_dhcpv6_proxy_header_with_length (u8 * s, va_list * args)
+{
+  dhcpv6_header_t * h = va_arg (*args, dhcpv6_header_t *);
+  u32 max_header_bytes = va_arg (*args, u32);
+  u32 header_bytes;
+
+  header_bytes = sizeof (h[0]);
+  if (max_header_bytes != 0 && header_bytes > max_header_bytes)
+    return format (s, "dhcpv6 header truncated");
+
+  s = format (s, "DHCPV6 Proxy");
+
+  return s;
+}
+/* get first interface address */
+static ip6_address_t *
+ip6_interface_first_global_or_site_address (ip6_main_t * im, u32 sw_if_index)
+{
+  ip_lookup_main_t * lm = &im->lookup_main;
+  ip_interface_address_t * ia = 0;
+  ip6_address_t * result = 0;
+
+  foreach_ip_interface_address (lm, ia, sw_if_index,
+                                1 /* honor unnumbered */,
+  ({
+    ip6_address_t * a = ip_interface_address_get_address (lm, ia);
+    if ((a->as_u8[0] & 0xe0) == 0x20 ||
+        (a->as_u8[0] & 0xfe) == 0xfc)  {
+        result = a;
+        break;
+    }
+  }));
+  return result;
+}
+
+static inline void copy_ip6_address (ip6_address_t *dst,
+                                     ip6_address_t *src) 
+{
+  dst->as_u64[0] = src->as_u64[0];
+  dst->as_u64[1] = src->as_u64[1];
+} 
+
+static uword
+dhcpv6_proxy_to_server_input (vlib_main_t * vm,
+                              vlib_node_runtime_t * node,
+                              vlib_frame_t * from_frame)
+{
+  u32 n_left_from, next_index, * from, * to_next;
+  dhcp_proxy_main_t * dpm = &dhcp_proxy_main;
+  from = vlib_frame_vector_args (from_frame);
+  n_left_from = from_frame->n_vectors;
+  u32 pkts_to_server=0, pkts_to_client=0, pkts_no_server=0;
+  u32 pkts_no_interface_address=0, pkts_no_exceeding_max_hop=0;
+  u32 pkts_no_src_address=0;
+  u32 pkts_wrong_msg_type=0;
+  u32 pkts_too_big=0;
+  ip6_main_t * im = &ip6_main;
+  ip6_address_t * src;
+  int bogus_length;
+  dhcp_server_t * server;
+  u32  rx_fib_idx = 0, server_fib_idx = 0;
+
+  next_index = node->cached_next_index;
+
+  while (n_left_from > 0)
+    {
+      u32 n_left_to_next;
+
+      vlib_get_next_frame (vm, node, next_index,
+			   to_next, n_left_to_next);
+
+      while (n_left_from > 0 && n_left_to_next > 0)
+	{
+          vnet_main_t *vnm = vnet_get_main();
+          u32 sw_if_index = 0;
+          u32 rx_sw_if_index = 0;
+          vnet_sw_interface_t *swif;
+ 	  u32 bi0;
+	  vlib_buffer_t * b0;
+          udp_header_t * u0, *u1;
+	  dhcpv6_header_t * h0;  // client msg hdr
+          ip6_header_t * ip0, *ip1;
+          ip6_address_t _ia0, *ia0=&_ia0;
+	  u32 next0;
+          u32 error0 = (u32) ~0;
+          dhcpv6_option_t *fwd_opt;
+          dhcpv6_relay_hdr_t *r1;
+          u16 len;
+          dhcpv6_int_id_t *id1;
+          dhcpv6_vss_t *vss1;
+          dhcpv6_client_mac_t *cmac; // client mac
+          ethernet_header_t * e_h0;
+          u8 client_src_mac[6];
+          vlib_buffer_free_list_t *fl;
+          dhcp_vss_t *vss;
+
+	  bi0 = from[0];
+	  to_next[0] = bi0;
+	  from += 1;
+	  to_next += 1;
+	  n_left_from -= 1;
+	  n_left_to_next -= 1;
+
+	  b0 = vlib_get_buffer (vm, bi0);
+
+          h0 = vlib_buffer_get_current (b0);
+
+          /*
+           * udp_local hands us the DHCPV6 header.
+           */
+          u0 = (void *)h0 -(sizeof(*u0));
+          ip0 = (void *)u0 -(sizeof(*ip0));
+          e_h0 = (void *)ip0 - ethernet_buffer_header_size(b0);
+
+          clib_memcpy(client_src_mac, e_h0->src_address, 6);
+
+          switch (h0->u.msg_type) {
+            case DHCPV6_MSG_SOLICIT:
+            case DHCPV6_MSG_REQUEST:
+            case DHCPV6_MSG_CONFIRM:
+            case DHCPV6_MSG_RENEW:
+            case DHCPV6_MSG_REBIND:
+            case DHCPV6_MSG_RELEASE:
+            case DHCPV6_MSG_DECLINE:
+            case DHCPV6_MSG_INFORMATION_REQUEST:
+            case DHCPV6_MSG_RELAY_FORW:
+                /* send to server */
+                break;
+            case DHCPV6_MSG_RELAY_REPL:
+                /* send to client */
+                next0 = DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_SEND_TO_CLIENT;
+                error0 = 0;
+                pkts_to_client++;
+                goto do_enqueue;
+            default:
+                /* drop the packet */
+                pkts_wrong_msg_type++;
+                error0 = DHCPV6_PROXY_ERROR_WRONG_MESSAGE_TYPE;
+                next0 = DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_DROP;
+                goto do_trace;
+
+          }
+
+          /* Send to DHCPV6 server via the configured FIB */
+          rx_sw_if_index = sw_if_index =  vnet_buffer(b0)->sw_if_index[VLIB_RX];
+          rx_fib_idx = im->fib_index_by_sw_if_index [rx_sw_if_index];
+          server = dhcp_get_server(dpm, rx_fib_idx, FIB_PROTOCOL_IP6);
+
+          if (PREDICT_FALSE (NULL == server))
+          {
+              error0 = DHCPV6_PROXY_ERROR_NO_SERVER;
+              next0 = DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_DROP;
+              pkts_no_server++;
+              goto do_trace;
+          }
+
+          server_fib_idx = server->server_fib_index;
+          vnet_buffer(b0)->sw_if_index[VLIB_TX] = server_fib_idx;
+
+
+          /* relay-option header pointer */
+          vlib_buffer_advance(b0, -(sizeof(*fwd_opt)));
+          fwd_opt = vlib_buffer_get_current(b0);
+          /* relay message header pointer */
+          vlib_buffer_advance(b0, -(sizeof(*r1)));
+          r1 = vlib_buffer_get_current(b0);
+
+          vlib_buffer_advance(b0, -(sizeof(*u1)));
+          u1 = vlib_buffer_get_current(b0);
+
+          vlib_buffer_advance(b0, -(sizeof(*ip1)));
+          ip1 = vlib_buffer_get_current(b0);
+
+          /* fill in all that rubbish... */
+          len = clib_net_to_host_u16(u0->length) - sizeof(udp_header_t);
+          copy_ip6_address(&r1->peer_addr, &ip0->src_address);
+
+          r1->msg_type = DHCPV6_MSG_RELAY_FORW;
+          fwd_opt->length = clib_host_to_net_u16(len);
+          fwd_opt->option = clib_host_to_net_u16(DHCPV6_OPTION_RELAY_MSG);
+
+          r1->hop_count++;
+          r1->hop_count = (h0->u.msg_type != DHCPV6_MSG_RELAY_FORW) ? 0 : r1->hop_count;
+
+          if (PREDICT_FALSE(r1->hop_count >= HOP_COUNT_LIMIT))
+            {
+              error0 =  DHCPV6_RELAY_PKT_DROP_MAX_HOPS;
+              next0 = DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_DROP;
+              pkts_no_exceeding_max_hop++;
+              goto do_trace;
+            }
+
+
+          /* If relay-fwd and src address is site or global unicast address  */
+          if (h0->u.msg_type == DHCPV6_MSG_RELAY_FORW &&
+              ((ip0->src_address.as_u8[0] & 0xe0) == 0x20 ||
+               (ip0->src_address.as_u8[0] & 0xfe) == 0xfc))
+            {
+              /* Set link address to zero */
+              r1->link_addr.as_u64[0] = 0;
+              r1->link_addr.as_u64[1] = 0;
+              goto link_address_set;
+            }
+
+          /* if receiving interface is unnumbered, use receiving interface
+           * IP address as link address, otherwise use the loopback interface
+           * IP address as link address.
+           */
+
+          swif = vnet_get_sw_interface (vnm, rx_sw_if_index);
+          if (swif->flags & VNET_SW_INTERFACE_FLAG_UNNUMBERED)
+              sw_if_index = swif->unnumbered_sw_if_index;
+
+          ia0 = ip6_interface_first_global_or_site_address(&ip6_main, sw_if_index);
+          if (ia0 == 0)
+            {
+              error0 = DHCPV6_PROXY_ERROR_NO_INTERFACE_ADDRESS;
+              next0 = DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_DROP;
+              pkts_no_interface_address++;
+              goto do_trace;
+            }
+
+          copy_ip6_address(&r1->link_addr, ia0);
+
+        link_address_set:
+          fl = vlib_buffer_get_free_list (vm, b0->free_list_index);
+
+          if ((b0->current_length+sizeof(*id1)+sizeof(*vss1)+sizeof(*cmac))
+              > fl->n_data_bytes)
+            {
+              error0 = DHCPV6_PROXY_ERROR_PKT_TOO_BIG;
+              next0 = DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_DROP;
+              pkts_too_big++;
+              goto do_trace;
+            }
+
+          id1 = (dhcpv6_int_id_t *) (((uword) ip1) + b0->current_length);
+          b0->current_length += (sizeof (*id1));
+
+          id1->opt.option = clib_host_to_net_u16(DHCPV6_OPTION_INTERFACE_ID);
+          id1->opt.length = clib_host_to_net_u16(sizeof(rx_sw_if_index));
+          id1->int_idx = clib_host_to_net_u32(rx_sw_if_index);
+
+          u1->length =0;
+          if (h0->u.msg_type != DHCPV6_MSG_RELAY_FORW)
+            {
+               cmac = (dhcpv6_client_mac_t *) (((uword) ip1) + b0->current_length);
+               b0->current_length += (sizeof (*cmac));
+               cmac->opt.length =clib_host_to_net_u16(sizeof(*cmac) -
+                                                      sizeof(cmac->opt));
+               cmac->opt.option = clib_host_to_net_u16(DHCPV6_OPTION_CLIENT_LINK_LAYER_ADDRESS);
+               cmac->link_type = clib_host_to_net_u16(1); // ethernet
+               clib_memcpy(cmac->data, client_src_mac, 6);
+               u1->length += sizeof(*cmac);
+            }
+
+          vss = dhcp_get_vss_info(dpm, rx_fib_idx, FIB_PROTOCOL_IP6);
+
+          if (NULL != vss) {
+              vss1 = (dhcpv6_vss_t *) (((uword) ip1) + b0->current_length);
+              b0->current_length += (sizeof (*vss1));
+              vss1->opt.length =clib_host_to_net_u16(sizeof(*vss1) -
+						     sizeof(vss1->opt));
+              vss1->opt.option = clib_host_to_net_u16(DHCPV6_OPTION_VSS);
+              vss1->data[0] = 1;   // type
+              vss1->data[1] = vss->oui >>16 & 0xff;
+              vss1->data[2] = vss->oui >>8  & 0xff;
+              vss1->data[3] = vss->oui & 0xff;
+              vss1->data[4] = vss->fib_id >> 24 & 0xff;
+              vss1->data[5] = vss->fib_id >> 16 & 0xff;
+              vss1->data[6] = vss->fib_id >> 8 & 0xff;
+              vss1->data[7] = vss->fib_id & 0xff;
+              u1->length += sizeof(*vss1);
+          }
+
+          pkts_to_server++;
+          u1->checksum = 0;
+          u1->src_port = clib_host_to_net_u16(UDP_DST_PORT_dhcpv6_to_client);
+          u1->dst_port = clib_host_to_net_u16(UDP_DST_PORT_dhcpv6_to_server);
+
+          u1->length =
+              clib_host_to_net_u16( clib_net_to_host_u16(fwd_opt->length) +
+                                    sizeof(*r1) + sizeof(*fwd_opt) +
+                                    sizeof(*u1) + sizeof(*id1) + u1->length);
+
+          memset(ip1, 0, sizeof(*ip1));
+          ip1->ip_version_traffic_class_and_flow_label = 0x60;
+          ip1->payload_length =  u1->length;
+          ip1->protocol = PROTO_UDP;
+          ip1->hop_limit = HOP_COUNT_LIMIT;
+	      src = (server->dhcp_server.ip6.as_u64[0] ||
+                     server->dhcp_server.ip6.as_u64[1]) ?
+	        &server->dhcp_server.ip6 : &all_dhcpv6_server_address;
+          copy_ip6_address(&ip1->dst_address, src);
+
+
+          ia0 = ip6_interface_first_global_or_site_address
+              (&ip6_main, vnet_buffer(b0)->sw_if_index[VLIB_RX]);
+
+	      src = (server->dhcp_src_address.ip6.as_u64[0] ||
+                     server->dhcp_src_address.ip6.as_u64[1]) ?
+	        &server->dhcp_src_address.ip6 : ia0;
+          if (ia0 == 0)
+            {
+              error0 = DHCPV6_PROXY_ERROR_NO_SRC_ADDRESS;
+              next0 = DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_DROP;
+              pkts_no_src_address++;
+              goto do_trace;
+            }
+
+	  copy_ip6_address (&ip1->src_address, src);
+
+
+          u1->checksum = ip6_tcp_udp_icmp_compute_checksum(vm, b0, ip1,
+                                                           &bogus_length);
+          ASSERT(bogus_length == 0);
+
+          next0 = DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_LOOKUP;
+
+        do_trace:
+          if (PREDICT_FALSE(b0->flags & VLIB_BUFFER_IS_TRACED))
+            {
+               dhcpv6_proxy_trace_t *tr = vlib_add_trace (vm, node,
+                                                          b0, sizeof (*tr));
+               tr->which = 0; /* to server */
+               tr->error = error0;
+               tr->original_sw_if_index = rx_sw_if_index;
+               tr->sw_if_index = sw_if_index;
+               if (DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_LOOKUP == next0)
+                 copy_ip6_address((ip6_address_t *)&tr->packet_data[0], &server->dhcp_server.ip6);
+            }
+
+        do_enqueue:
+	  vlib_validate_buffer_enqueue_x1 (vm, node, next_index,
+					   to_next, n_left_to_next,
+					   bi0, next0);
+	}
+
+      vlib_put_next_frame (vm, node, next_index, n_left_to_next);
+    }
+
+  vlib_node_increment_counter (vm, dhcpv6_proxy_to_server_node.index,
+                               DHCPV6_PROXY_ERROR_RELAY_TO_CLIENT,
+                               pkts_to_client);
+  vlib_node_increment_counter (vm, dhcpv6_proxy_to_server_node.index,
+                               DHCPV6_PROXY_ERROR_RELAY_TO_SERVER,
+                               pkts_to_server);
+  vlib_node_increment_counter (vm, dhcpv6_proxy_to_server_node.index,
+                               DHCPV6_PROXY_ERROR_NO_INTERFACE_ADDRESS,
+                               pkts_no_interface_address);
+  vlib_node_increment_counter (vm, dhcpv6_proxy_to_server_node.index,
+                               DHCPV6_PROXY_ERROR_WRONG_MESSAGE_TYPE,
+                               pkts_wrong_msg_type);
+  vlib_node_increment_counter (vm, dhcpv6_proxy_to_server_node.index,
+                               DHCPV6_PROXY_ERROR_NO_SRC_ADDRESS,
+                               pkts_no_src_address);
+  vlib_node_increment_counter (vm, dhcpv6_proxy_to_server_node.index,
+                               DHCPV6_PROXY_ERROR_PKT_TOO_BIG,
+                               pkts_too_big);
+  return from_frame->n_vectors;
+}
+
+VLIB_REGISTER_NODE (dhcpv6_proxy_to_server_node, static) = {
+  .function = dhcpv6_proxy_to_server_input,
+  .name = "dhcpv6-proxy-to-server",
+  /* Takes a vector of packets. */
+  .vector_size = sizeof (u32),
+
+  .n_errors = DHCPV6_PROXY_N_ERROR,
+  .error_strings = dhcpv6_proxy_error_strings,
+
+  .n_next_nodes = DHCPV6_PROXY_TO_SERVER_INPUT_N_NEXT,
+  .next_nodes = {
+#define _(s,n) [DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_##s] = n,
+    foreach_dhcpv6_proxy_to_server_input_next
+#undef _
+  },
+
+  .format_buffer = format_dhcpv6_proxy_header_with_length,
+  .format_trace = format_dhcpv6_proxy_trace,
+#if 0
+  .unformat_buffer = unformat_dhcpv6_proxy_header,
+#endif
+};
+
+static uword
+dhcpv6_proxy_to_client_input (vlib_main_t * vm,
+                            vlib_node_runtime_t * node,
+                            vlib_frame_t * from_frame)
+{
+
+  u32 n_left_from, * from;
+  ethernet_main_t *em = ethernet_get_main (vm);
+  dhcp_proxy_main_t * dm = &dhcp_proxy_main;
+  dhcp_server_t * server;
+  vnet_main_t * vnm = vnet_get_main();
+  int bogus_length;
+
+  from = vlib_frame_vector_args (from_frame);
+  n_left_from = from_frame->n_vectors;
+
+  while (n_left_from > 0)
+    {
+      u32 bi0;
+      vlib_buffer_t * b0;
+      udp_header_t * u0, *u1=0;
+      dhcpv6_relay_hdr_t * h0;
+      ip6_header_t * ip1 = 0, *ip0;
+      ip6_address_t _ia0, * ia0 = &_ia0;
+      ip6_address_t client_address;
+      ethernet_interface_t *ei0;
+      ethernet_header_t *mac0;
+      vnet_hw_interface_t *hi0;
+      vlib_frame_t *f0;
+      u32 * to_next0;
+      u32 sw_if_index = ~0;
+      u32 original_sw_if_index = ~0;
+      vnet_sw_interface_t *si0;
+      u32 error0 = (u32)~0;
+      vnet_sw_interface_t *swif;
+      dhcpv6_option_t *r0 = 0, *o;
+      u16 len = 0;
+      u8 interface_opt_flag = 0;
+      u8 relay_msg_opt_flag = 0;
+      ip6_main_t * im = &ip6_main;
+      u32 server_fib_idx, client_fib_idx;
+
+      bi0 = from[0];
+      from += 1;
+      n_left_from -= 1;
+
+      b0 = vlib_get_buffer (vm, bi0);
+      h0 = vlib_buffer_get_current (b0);
+
+      if (DHCPV6_MSG_RELAY_REPL != h0->msg_type)
+        {
+          error0 =  DHCPV6_PROXY_ERROR_WRONG_MESSAGE_TYPE;
+
+        drop_packet:
+          vlib_node_increment_counter (vm, dhcpv6_proxy_to_client_node.index,
+                                       error0, 1);
+
+          f0 = vlib_get_frame_to_node (vm, dm->error_drop_node_index);
+          to_next0 = vlib_frame_vector_args (f0);
+          to_next0[0] = bi0;
+          f0->n_vectors = 1;
+          vlib_put_frame_to_node (vm, dm->error_drop_node_index, f0);
+          goto do_trace;
+        }
+      /* hop count seems not need to be checked */
+      if (HOP_COUNT_LIMIT < h0->hop_count)
+        {
+          error0 =  DHCPV6_RELAY_PKT_DROP_MAX_HOPS;
+          goto drop_packet;
+        }
+      u0 = (void *)h0 -(sizeof(*u0));
+      ip0 = (void *)u0 -(sizeof(*ip0));
+
+      vlib_buffer_advance (b0, sizeof(*h0));
+      o = vlib_buffer_get_current (b0);
+
+      /* Parse through TLVs looking for option 18 (DHCPV6_OPTION_INTERFACE_ID)
+         _and_ option 9 (DHCPV6_OPTION_RELAY_MSG) option which must be there.
+         Currently assuming no other options need to be processed
+         The interface-ID is the FIB number we need
+         to track down the client-facing interface */
+
+      while ((u8 *) o < (b0->data + b0->current_data + b0->current_length))
+        {
+           if (DHCPV6_OPTION_INTERFACE_ID == clib_net_to_host_u16(o->option))
+             {
+                interface_opt_flag = 1;
+                if (clib_net_to_host_u16(o->length) == sizeof(sw_if_index))
+                    sw_if_index = clib_net_to_host_u32(((dhcpv6_int_id_t*)o)->int_idx);
+                if (sw_if_index >= vec_len (im->fib_index_by_sw_if_index))
+                  {
+                    error0 = DHCPV6_PROXY_ERROR_WRONG_INTERFACE_ID_OPTION;
+                    goto drop_packet;
+                  }
+             }
+           if (DHCPV6_OPTION_RELAY_MSG == clib_net_to_host_u16(o->option))
+             {
+                relay_msg_opt_flag = 1;
+                r0 = vlib_buffer_get_current (b0);
+             }
+           if ((relay_msg_opt_flag == 1) && (interface_opt_flag == 1))
+             break;
+           vlib_buffer_advance (b0, sizeof(*o) + clib_net_to_host_u16(o->length));
+           o = (dhcpv6_option_t *) (((uword) o) + clib_net_to_host_u16(o->length) + sizeof(*o));
+        }
+
+      if ((relay_msg_opt_flag == 0) || (r0 == 0))
+        {
+    	  error0 = DHCPV6_PROXY_ERROR_NO_RELAY_MESSAGE_OPTION;
+    	  goto drop_packet;
+        }
+
+      if ((u32)~0 == sw_if_index)
+        {
+          error0 = DHCPV6_PROXY_ERROR_NO_CIRCUIT_ID_OPTION;
+          goto drop_packet;
+        }
+
+      //Advance buffer to start of encapsulated DHCPv6 message
+      vlib_buffer_advance (b0, sizeof(*r0));
+
+      client_fib_idx = im->fib_index_by_sw_if_index[sw_if_index];
+      server = dhcp_get_server(dm, client_fib_idx, FIB_PROTOCOL_IP6);
+
+      if (NULL == server)
+      {
+    	  error0 = DHCPV6_PROXY_ERROR_NO_SERVER;
+          goto drop_packet;
+      }
+
+      server_fib_idx = im->fib_index_by_sw_if_index
+          [vnet_buffer(b0)->sw_if_index[VLIB_RX]];
+
+      if (server_fib_idx != server->server_fib_index ||
+          ip0->src_address.as_u64[0] != server->dhcp_server.ip6.as_u64[0] ||
+          ip0->src_address.as_u64[1] != server->dhcp_server.ip6.as_u64[1])
+        {
+          //drop packet if not from server with configured address or FIB
+          error0 = DHCPV6_PROXY_ERROR_BAD_SVR_FIB_OR_ADDRESS;
+          goto drop_packet;
+        }
+
+      vnet_buffer (b0)->sw_if_index[VLIB_TX] = original_sw_if_index
+          = sw_if_index;
+
+      swif = vnet_get_sw_interface (vnm, original_sw_if_index);
+      if (swif->flags & VNET_SW_INTERFACE_FLAG_UNNUMBERED)
+          sw_if_index = swif->unnumbered_sw_if_index;
+
+
+      /*
+       * udp_local hands us the DHCPV6 header, need udp hdr,
+       * ip hdr to relay to client
+       */
+      vlib_buffer_advance (b0, -(sizeof(*u1)));
+      u1 = vlib_buffer_get_current (b0);
+
+      vlib_buffer_advance (b0, -(sizeof(*ip1)));
+      ip1 = vlib_buffer_get_current (b0);
+
+      copy_ip6_address(&client_address, &h0->peer_addr);
+
+      ia0 = ip6_interface_first_address (&ip6_main, sw_if_index);
+      if (ia0 == 0)
+        {
+          error0 = DHCPV6_PROXY_ERROR_NO_INTERFACE_ADDRESS;
+          goto drop_packet;
+        }
+
+      len =  clib_net_to_host_u16(r0->length);
+      memset(ip1, 0, sizeof(*ip1));
+      copy_ip6_address(&ip1->dst_address, &client_address);
+      u1->checksum = 0;
+      u1->src_port = clib_net_to_host_u16 (UDP_DST_PORT_dhcpv6_to_server);
+      u1->dst_port = clib_net_to_host_u16 (UDP_DST_PORT_dhcpv6_to_client);
+      u1->length = clib_host_to_net_u16 (len + sizeof(udp_header_t));
+
+      ip1->ip_version_traffic_class_and_flow_label =
+          ip0->ip_version_traffic_class_and_flow_label &
+          0x00000fff;
+      ip1->payload_length =  u1->length;
+      ip1->protocol = PROTO_UDP;
+      ip1->hop_limit = HOP_COUNT_LIMIT;
+      copy_ip6_address(&ip1->src_address, ia0);
+
+      u1->checksum = ip6_tcp_udp_icmp_compute_checksum(vm, b0, ip1,
+                                                       &bogus_length);
+      ASSERT(bogus_length == 0);
+
+      vlib_buffer_advance (b0, -(sizeof(ethernet_header_t)));
+      si0 = vnet_get_sw_interface (vnm, original_sw_if_index);
+      if (si0->type == VNET_SW_INTERFACE_TYPE_SUB)
+	  vlib_buffer_advance (b0, -4 /* space for VLAN tag */);
+
+      mac0 = vlib_buffer_get_current (b0);
+
+      hi0 = vnet_get_sup_hw_interface (vnm, original_sw_if_index);
+      ei0 = pool_elt_at_index (em->interfaces, hi0->hw_instance);
+      clib_memcpy (mac0->src_address, ei0->address, sizeof (ei0->address));
+      memset (&mac0->dst_address, 0xff, sizeof (mac0->dst_address));
+      mac0->type = (si0->type == VNET_SW_INTERFACE_TYPE_SUB) ?
+	clib_net_to_host_u16(0x8100) : clib_net_to_host_u16 (0x86dd);
+
+      if (si0->type == VNET_SW_INTERFACE_TYPE_SUB)
+	{
+	  u32 * vlan_tag = (u32 *)(mac0+1);
+	  u32 tmp;
+	  tmp = (si0->sub.id << 16) | 0x0800;
+	  *vlan_tag = clib_host_to_net_u32 (tmp);
+	}
+
+      /* $$$ consider adding a dynamic next to the graph node, for performance */
+      f0 = vlib_get_frame_to_node (vm, hi0->output_node_index);
+      to_next0 = vlib_frame_vector_args (f0);
+      to_next0[0] = bi0;
+      f0->n_vectors = 1;
+      vlib_put_frame_to_node (vm, hi0->output_node_index, f0);
+
+    do_trace:
+      if (PREDICT_FALSE(b0->flags & VLIB_BUFFER_IS_TRACED))
+        {
+          dhcpv6_proxy_trace_t *tr = vlib_add_trace (vm, node,
+                                                     b0, sizeof (*tr));
+          tr->which = 1; /* to client */
+          if (ia0)
+              copy_ip6_address((ip6_address_t*)tr->packet_data, ia0);
+          tr->error = error0;
+          tr->original_sw_if_index = original_sw_if_index;
+          tr->sw_if_index = sw_if_index;
+        }
+    }
+  return from_frame->n_vectors;
+
+}
+
+VLIB_REGISTER_NODE (dhcpv6_proxy_to_client_node, static) = {
+  .function = dhcpv6_proxy_to_client_input,
+  .name = "dhcpv6-proxy-to-client",
+  /* Takes a vector of packets. */
+  .vector_size = sizeof (u32),
+
+  .n_errors = DHCPV6_PROXY_N_ERROR,
+  .error_strings = dhcpv6_proxy_error_strings,
+  .format_buffer = format_dhcpv6_proxy_header_with_length,
+  .format_trace = format_dhcpv6_proxy_trace,
+#if 0
+  .unformat_buffer = unformat_dhcpv6_proxy_header,
+#endif
+};
+
+static clib_error_t *
+dhcp6_proxy_init (vlib_main_t * vm)
+{
+  dhcp_proxy_main_t * dm = &dhcp_proxy_main;
+  vlib_node_t * error_drop_node;
+
+  error_drop_node = vlib_get_node_by_name (vm, (u8 *) "error-drop");
+  dm->error_drop_node_index = error_drop_node->index;
+
+  /* RFC says this is the dhcpv6 server address  */
+  all_dhcpv6_server_address.as_u64[0] = clib_host_to_net_u64 (0xFF05000000000000);
+  all_dhcpv6_server_address.as_u64[1] = clib_host_to_net_u64 (0x00010003);
+
+  /* RFC says this is the server and agent address */
+  all_dhcpv6_server_relay_agent_address.as_u64[0] = clib_host_to_net_u64 (0xFF02000000000000);
+  all_dhcpv6_server_relay_agent_address.as_u64[1] = clib_host_to_net_u64 (0x00010002);
+
+  udp_register_dst_port (vm, UDP_DST_PORT_dhcpv6_to_client,
+                         dhcpv6_proxy_to_client_node.index, 0 /* is_ip6 */);
+
+  udp_register_dst_port (vm, UDP_DST_PORT_dhcpv6_to_server,
+                         dhcpv6_proxy_to_server_node.index, 0 /* is_ip6 */);
+
+  return 0;
+}
+
+VLIB_INIT_FUNCTION (dhcp6_proxy_init);
+
+int
+dhcp6_proxy_set_server (ip46_address_t *addr,
+                        ip46_address_t *src_addr,
+                        u32 rx_table_id,
+                        u32 server_table_id, 
+                        int is_del)
+{
+  u32 rx_fib_index = 0;
+  int rc = 0;
+
+  const mfib_prefix_t all_dhcp_servers = {
+      .fp_len = 128,
+      .fp_proto = FIB_PROTOCOL_IP6,
+      .fp_grp_addr = {
+          .ip6 = all_dhcpv6_server_relay_agent_address,
+      }
+  };
+
+  if (ip46_address_is_zero(addr))
+    return VNET_API_ERROR_INVALID_DST_ADDRESS;
+  
+  if (ip46_address_is_zero(src_addr))
+    return VNET_API_ERROR_INVALID_SRC_ADDRESS;
+
+  rx_fib_index = mfib_table_find_or_create_and_lock(FIB_PROTOCOL_IP6,
+                                                    rx_table_id);
+
+  if (is_del)
+    {
+      rc = dhcp_proxy_server_del (FIB_PROTOCOL_IP6, rx_fib_index);
+
+      if (0 == rc)
+      {
+          mfib_table_entry_delete(rx_fib_index,
+                                  &all_dhcp_servers,
+                                  MFIB_SOURCE_DHCP);
+          mfib_table_unlock(rx_fib_index, FIB_PROTOCOL_IP6);
+      }
+    }
+  else
+    {
+     const fib_route_path_t path_for_us = {
+          .frp_proto = FIB_PROTOCOL_IP6,
+          .frp_addr = zero_addr,
+          .frp_sw_if_index = 0xffffffff,
+          .frp_fib_index = ~0,
+          .frp_weight = 0,
+          .frp_flags = FIB_ROUTE_PATH_LOCAL,
+      };
+     if (dhcp_proxy_server_add (FIB_PROTOCOL_IP6, addr, src_addr,
+                                rx_fib_index, server_table_id))
+     {
+         mfib_table_entry_path_update(rx_fib_index,
+                                      &all_dhcp_servers,
+                                      MFIB_SOURCE_DHCP,
+                                      &path_for_us,
+                                      MFIB_ITF_FLAG_FORWARD);
+         /*
+          * Each interface that is enabled in this table, needs to be added
+          * as an accepting interface, but this is not easily doable in VPP.
+          * So we cheat. Add a flag to the entry that indicates accept form
+          * any interface.
+          * We will still only accept on v6 enabled interfaces, since the
+          * input feature ensures this.
+          */
+         mfib_table_entry_update(rx_fib_index,
+                                 &all_dhcp_servers,
+                                 MFIB_SOURCE_DHCP,
+                                 MFIB_ENTRY_FLAG_ACCEPT_ALL_ITF);
+         mfib_table_lock(rx_fib_index, FIB_PROTOCOL_IP6);
+     }
+    }
+
+  mfib_table_unlock(rx_fib_index, FIB_PROTOCOL_IP6);
+
+  return (rc);
+}
+
+static clib_error_t *
+dhcpv6_proxy_set_command_fn (vlib_main_t * vm,
+                           unformat_input_t * input,
+                           vlib_cli_command_t * cmd)
+{
+  ip46_address_t addr, src_addr;
+  int set_server = 0, set_src_address = 0;
+  u32 rx_table_id = 0, server_table_id = 0;
+  int is_del = 0;
+
+  while (unformat_check_input(input) != UNFORMAT_END_OF_INPUT)
+    {
+      if (unformat (input, "server %U",
+                    unformat_ip6_address, &addr.ip6))
+	 set_server = 1;
+      else if (unformat(input, "src-address %U",
+			unformat_ip6_address, &src_addr.ip6))
+	  set_src_address =1;
+       else if (unformat (input, "server-fib-id %d", &server_table_id))
+        ;
+       else if (unformat (input, "rx-fib-id %d", &rx_table_id))
+         ;
+       else if (unformat (input, "delete") ||
+                unformat (input, "del"))
+           is_del = 1;
+      else
+        break;
+    }
+
+  if (is_del || (set_server && set_src_address))
+  {
+      int rv;
+
+      rv = dhcp6_proxy_set_server (&addr, &src_addr, rx_table_id,
+                                   server_table_id, is_del);
+
+      //TODO: Complete the errors
+      switch (rv)
+        {
+        case 0:
+          return 0;
+
+        case VNET_API_ERROR_INVALID_DST_ADDRESS:
+          return clib_error_return (0, "Invalid server address");
+
+        case VNET_API_ERROR_INVALID_SRC_ADDRESS:
+          return clib_error_return (0, "Invalid src address");
+
+        case VNET_API_ERROR_NO_SUCH_ENTRY:
+          return clib_error_return
+            (0, "Fib id %d: no per-fib DHCP server configured", rx_table_id);
+
+        default:
+          return clib_error_return (0, "BUG: rv %d", rv);
+        }
+  }
+  else
+    return clib_error_return (0, "parse error`%U'",
+                              format_unformat_error, input);
+}
+
+VLIB_CLI_COMMAND (dhcpv6_proxy_set_command, static) = {
+  .path = "set dhcpv6 proxy",
+  .short_help = "set dhcpv6 proxy [del] server <ipv6-addr> src-address <ipv6-addr> "
+		  "[server-fib-id <fib-id>] [rx-fib-id <fib-id>] ",
+  .function = dhcpv6_proxy_set_command_fn,
+};
+
+static u8 *
+format_dhcp6_proxy_server (u8 * s, va_list * args)
+{
+  dhcp_server_t * server = va_arg (*args, dhcp_server_t *);
+  ip6_fib_t * rx_fib, * server_fib;
+
+  if (NULL == server)
+    {
+      s = format (s, "%=40s%=40s%=14s%=14s", "Server Address", "Source Address",
+                  "Server FIB", "RX FIB");
+      return s;
+    }
+
+  server_fib = ip6_fib_get(server->server_fib_index);
+  rx_fib = ip6_fib_get(server->rx_fib_index);
+
+
+  s = format (s, "%=40U%=40U%=14u%=14u",
+              format_ip46_address, &server->dhcp_server, IP46_TYPE_ANY,
+              format_ip46_address, &server->dhcp_src_address, IP46_TYPE_ANY,
+              server_fib->table_id, rx_fib->table_id);
+  return s;
+}
+
+static int
+dhcp6_proxy_show_walk (dhcp_server_t *server,
+                       void *ctx)
+{
+    vlib_main_t * vm = ctx;
+
+    vlib_cli_output (vm, "%U", format_dhcp6_proxy_server, server);
+
+    return (1);
+}
+
+static clib_error_t *
+dhcpv6_proxy_show_command_fn (vlib_main_t * vm,
+                            unformat_input_t * input,
+                            vlib_cli_command_t * cmd)
+{
+  vlib_cli_output (vm, "%U", format_dhcp6_proxy_server, NULL /* header line */);
+
+  dhcp_proxy_walk(FIB_PROTOCOL_IP6, dhcp6_proxy_show_walk, vm);
+
+  return (NULL);
+}
+
+VLIB_CLI_COMMAND (dhcpv6_proxy_show_command, static) = {
+  .path = "show dhcpv6 proxy",
+  .short_help = "Display dhcpv6 proxy info",
+  .function = dhcpv6_proxy_show_command_fn,
+};
+
+static clib_error_t *
+dhcpv6_vss_command_fn (vlib_main_t * vm,
+                       unformat_input_t * input,
+                       vlib_cli_command_t * cmd)
+{
+  int is_del = 0, got_new_vss=0;
+  u32 oui=0;
+  u32 fib_id=0, tbl_id=~0;
+
+  while (unformat_check_input(input) != UNFORMAT_END_OF_INPUT)
+    {
+      if (unformat (input, "oui %d", &oui))
+          got_new_vss = 1;
+      else if (unformat (input, "vpn-id %d", &fib_id))
+          got_new_vss = 1;
+      else if (unformat (input, "table %d", &tbl_id))
+          got_new_vss = 1;
+      else if (unformat(input, "delete") || unformat(input, "del"))
+          is_del = 1;
+      else
+          break;
+    }
+
+  if (tbl_id ==~0)
+      return clib_error_return (0, "no table ID specified.");
+
+  if (is_del || got_new_vss)
+    {
+      int rv;
+
+      rv = dhcp_proxy_set_vss(FIB_PROTOCOL_IP6, tbl_id, oui, fib_id, is_del);
+      switch (rv)
+        {
+        case 0:
+          return 0;
+
+        case VNET_API_ERROR_NO_SUCH_FIB:
+            return clib_error_return (0, "vss info (oui:%d, vpn-id:%d)  not found in table %d.",
+                                      oui, fib_id, tbl_id);
+
+        case VNET_API_ERROR_NO_SUCH_ENTRY:
+            return clib_error_return (0, "vss for table %d not found in pool.",
+                                      tbl_id);
+
+        default:
+          return clib_error_return (0, "BUG: rv %d", rv);
+        }
+    }
+  else
+      return clib_error_return (0, "parse error`%U'",
+                                format_unformat_error, input);
+
+}
+
+VLIB_CLI_COMMAND (dhcpv6_proxy_vss_command, static) = {
+  .path = "set dhcpv6 vss",
+  .short_help = "set dhcpv6 vss table <table-id> oui <oui> vpn-idx <vpn-idx>",
+  .function = dhcpv6_vss_command_fn,
+};
+
+static clib_error_t *
+dhcpv6_vss_show_command_fn (vlib_main_t * vm,
+                            unformat_input_t * input,
+                            vlib_cli_command_t * cmd)
+
+{
+  dhcp_vss_walk(FIB_PROTOCOL_IP6, dhcp_vss_show_walk, vm);
+
+  return (NULL);
+}
+
+VLIB_CLI_COMMAND (dhcpv6_proxy_vss_show_command, static) = {
+  .path = "show dhcpv6 vss",
+  .short_help = "show dhcpv6 VSS",
+  .function = dhcpv6_vss_show_command_fn,
+};
+
+static clib_error_t *
+dhcpv6_link_address_show_command_fn (vlib_main_t * vm,
+                                unformat_input_t * input,
+                                vlib_cli_command_t * cmd)
+
+{
+  vnet_main_t *vnm = vnet_get_main();
+  u32 sw_if_index0=0, sw_if_index;
+  vnet_sw_interface_t *swif;
+  ip6_address_t *ia0;
+
+  while (unformat_check_input(input) != UNFORMAT_END_OF_INPUT)
+    {
+
+      if (unformat(input, "%U",
+                   unformat_vnet_sw_interface, vnm, &sw_if_index0))
+        {
+            swif = vnet_get_sw_interface (vnm, sw_if_index0);
+            sw_if_index = (swif->flags & VNET_SW_INTERFACE_FLAG_UNNUMBERED) ?
+                swif->unnumbered_sw_if_index : sw_if_index0;
+            ia0 = ip6_interface_first_address(&ip6_main, sw_if_index);
+            if (ia0)
+              {
+                  vlib_cli_output (vm, "%=20s%=48s", "interface", "link-address");
+
+                  vlib_cli_output (vm, "%=20U%=48U",
+                                   format_vnet_sw_if_index_name, vnm, sw_if_index0,
+                                   format_ip6_address, ia0);
+              } else
+                vlib_cli_output (vm, "%=34s%=20U", "No IPv6 address configured on",
+                                 format_vnet_sw_if_index_name, vnm, sw_if_index);
+        } else
+          break;
+    }
+
+  return 0;
+}
+
+VLIB_CLI_COMMAND (dhcpv6_proxy_address_show_command, static) = {
+  .path = "show dhcpv6 link-address interface",
+  .short_help = "show dhcpv6 link-address interface <interface>",
+  .function = dhcpv6_link_address_show_command_fn,
+};
diff --git a/src/vnet/dhcp/dhcp_api.c b/src/vnet/dhcp/dhcp_api.c
index ce9039b7..bdf02cae 100644
--- a/src/vnet/dhcp/dhcp_api.c
+++ b/src/vnet/dhcp/dhcp_api.c
@@ -22,9 +22,8 @@
 
 #include <vnet/interface.h>
 #include <vnet/api_errno.h>
-#include <vnet/dhcp/proxy.h>
+#include <vnet/dhcp/dhcp_proxy.h>
 #include <vnet/dhcp/client.h>
-#include <vnet/dhcpv6/proxy.h>
 
 #include <vnet/vnet_msg_enum.h>
 
@@ -51,52 +50,19 @@ _(DHCP_PROXY_DETAILS,dhcp_proxy_details)          \
 _(DHCP_PROXY_SET_VSS,dhcp_proxy_set_vss)          \
 _(DHCP_CLIENT_CONFIG, dhcp_client_config)
 
-static void
-dhcpv4_proxy_config (vl_api_dhcp_proxy_config_t * mp)
-{
-  vl_api_dhcp_proxy_config_reply_t *rmp;
-  int rv;
-
-  rv = dhcp_proxy_set_server ((ip4_address_t *) (&mp->dhcp_server),
-			      (ip4_address_t *) (&mp->dhcp_src_address),
-			      (u32) ntohl (mp->rx_vrf_id),
-			      (u32) ntohl (mp->server_vrf_id),
-			      (int) (mp->is_add == 0));
-
-  REPLY_MACRO (VL_API_DHCP_PROXY_CONFIG_REPLY);
-}
-
-
-static void
-dhcpv6_proxy_config (vl_api_dhcp_proxy_config_t * mp)
-{
-  vl_api_dhcp_proxy_config_reply_t *rmp;
-  int rv = -1;
-
-  rv = dhcpv6_proxy_set_server ((ip6_address_t *) (&mp->dhcp_server),
-				(ip6_address_t *) (&mp->dhcp_src_address),
-				(u32) ntohl (mp->rx_vrf_id),
-				(u32) ntohl (mp->server_vrf_id),
-				(int) (mp->is_add == 0));
-
-  REPLY_MACRO (VL_API_DHCP_PROXY_CONFIG_REPLY);
-}
-
 
 static void
 vl_api_dhcp_proxy_set_vss_t_handler (vl_api_dhcp_proxy_set_vss_t * mp)
 {
   vl_api_dhcp_proxy_set_vss_reply_t *rmp;
   int rv;
-  if (!mp->is_ipv6)
-    rv = dhcp_proxy_set_option82_vss (ntohl (mp->tbl_id),
-				      ntohl (mp->oui),
-				      ntohl (mp->fib_id),
-				      (int) mp->is_add == 0);
-  else
-    rv = dhcpv6_proxy_set_vss (ntohl (mp->tbl_id),
-			       ntohl (mp->oui),
-			       ntohl (mp->fib_id), (int) mp->is_add == 0);
+
+  rv = dhcp_proxy_set_vss ((mp->is_ipv6 ?
+			    FIB_PROTOCOL_IP6 :
+			    FIB_PROTOCOL_IP4),
+			   ntohl (mp->tbl_id),
+			   ntohl (mp->oui),
+			   ntohl (mp->fib_id), (int) mp->is_add == 0);
 
   REPLY_MACRO (VL_API_DHCP_PROXY_SET_VSS_REPLY);
 }
@@ -105,10 +71,38 @@ vl_api_dhcp_proxy_set_vss_t_handler (vl_api_dhcp_proxy_set_vss_t * mp)
 static void vl_api_dhcp_proxy_config_t_handler
   (vl_api_dhcp_proxy_config_t * mp)
 {
-  if (mp->is_ipv6 == 0)
-    dhcpv4_proxy_config (mp);
+  vl_api_dhcp_proxy_set_vss_reply_t *rmp;
+  ip46_address_t src, server;
+  int rv = -1;
+
+  if (mp->is_ipv6)
+    {
+      clib_memcpy (&src.ip6, mp->dhcp_src_address, sizeof (src.ip6));
+      clib_memcpy (&server.ip6, mp->dhcp_server, sizeof (server.ip6));
+
+      rv = dhcp6_proxy_set_server (&server,
+				   &src,
+				   (u32) ntohl (mp->rx_vrf_id),
+				   (u32) ntohl (mp->server_vrf_id),
+				   (int) (mp->is_add == 0));
+    }
   else
-    dhcpv6_proxy_config (mp);
+    {
+      ip46_address_reset (&src);
+      ip46_address_reset (&server);
+
+      clib_memcpy (&src.ip4, mp->dhcp_src_address, sizeof (src.ip4));
+      clib_memcpy (&server.ip4, mp->dhcp_server, sizeof (server.ip4));
+
+      rv = dhcp4_proxy_set_server (&server,
+				   &src,
+				   (u32) ntohl (mp->rx_vrf_id),
+				   (u32) ntohl (mp->server_vrf_id),
+				   (int) (mp->is_add == 0));
+    }
+
+
+  REPLY_MACRO (VL_API_DHCP_PROXY_CONFIG_REPLY);
 }
 
 static void
@@ -120,14 +114,13 @@ vl_api_dhcp_proxy_dump_t_handler (vl_api_dhcp_proxy_dump_t * mp)
   if (q == 0)
     return;
 
-  if (mp->is_ip6 == 0)
-    dhcp_proxy_dump (q, mp->context);
-  else
-    dhcpv6_proxy_dump (q, mp->context);
+  dhcp_proxy_dump ((mp->is_ip6 == 0 ?
+		    FIB_PROTOCOL_IP6 : FIB_PROTOCOL_IP4), q, mp->context);
 }
 
 void
-dhcp_send_details (void *opaque,
+dhcp_send_details (fib_protocol_t proto,
+		   void *opaque,
 		   u32 context,
 		   const ip46_address_t * server,
 		   const ip46_address_t * src,
@@ -149,7 +142,7 @@ dhcp_send_details (void *opaque,
   mp->vss_oui = htonl (vss_oui);
   mp->vss_fib_id = htonl (vss_fib_id);
 
-  mp->is_ipv6 = !ip46_address_is_ip4 (server);
+  mp->is_ipv6 = (proto == FIB_PROTOCOL_IP6);
 
   if (mp->is_ipv6)
     {
diff --git a/src/vnet/dhcp/dhcp_proxy.c b/src/vnet/dhcp/dhcp_proxy.c
new file mode 100644
index 00000000..da2deea6
--- /dev/null
+++ b/src/vnet/dhcp/dhcp_proxy.c
@@ -0,0 +1,275 @@
+/*
+ * proxy_node.c: common dhcp v4 and v6 proxy node processing
+ *
+ * Copyright (c) 2013 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <vnet/dhcp/dhcp_proxy.h>
+#include <vnet/fib/fib_table.h>
+
+/**
+ * @brief Shard 4/6 instance of DHCP main
+ */
+dhcp_proxy_main_t dhcp_proxy_main;
+
+void
+dhcp_proxy_walk (fib_protocol_t proto,
+                 dhcp_proxy_walk_fn_t fn,
+                 void *ctx)
+{
+  dhcp_proxy_main_t * dpm = &dhcp_proxy_main;
+  dhcp_server_t * server;
+  u32 server_index, i;
+
+  vec_foreach_index (i, dpm->dhcp_server_index_by_rx_fib_index[proto])
+  {
+      server_index = dpm->dhcp_server_index_by_rx_fib_index[proto][i];
+      if (~0 == server_index)
+          continue;
+
+      server = pool_elt_at_index (dpm->dhcp_servers[proto], server_index);
+
+      if (!fn(server, ctx))
+          break;
+    }
+}
+
+void
+dhcp_vss_walk (fib_protocol_t proto,
+               dhcp_vss_walk_fn_t fn,
+               void *ctx)
+{
+  dhcp_proxy_main_t * dpm = &dhcp_proxy_main;
+  dhcp_vss_t * vss;
+  u32 vss_index, i;
+  fib_table_t *fib;
+
+
+  vec_foreach_index (i, dpm->vss_index_by_rx_fib_index[proto])
+  {
+      vss_index = dpm->vss_index_by_rx_fib_index[proto][i];
+      if (~0 == vss_index)
+          continue;
+
+      vss = pool_elt_at_index (dpm->vss[proto], vss_index);
+
+      fib = fib_table_get(i, proto);
+
+      if (!fn(vss, fib->ft_table_id, ctx))
+          break;
+    }
+}
+
+int
+dhcp_proxy_server_del (fib_protocol_t proto,
+                       u32 rx_fib_index)
+{
+  dhcp_proxy_main_t * dpm = &dhcp_proxy_main;
+  dhcp_server_t * server = 0;
+  int rc = 0;
+
+  server = dhcp_get_server(dpm, rx_fib_index, proto);
+
+  if (NULL == server)
+  {
+      rc = VNET_API_ERROR_NO_SUCH_ENTRY;
+  }
+  else
+  {
+      /* Use the default server again.  */
+      dpm->dhcp_server_index_by_rx_fib_index[proto][rx_fib_index] = ~0;
+
+      fib_table_unlock (server->server_fib_index, proto);
+
+      pool_put (dpm->dhcp_servers[proto], server);
+  }
+
+  return (rc);
+}
+
+int
+dhcp_proxy_server_add (fib_protocol_t proto,
+                       ip46_address_t *addr,
+                       ip46_address_t *src_address,
+                       u32 rx_fib_index,
+                       u32 server_table_id)
+{
+  dhcp_proxy_main_t * dpm = &dhcp_proxy_main;
+  dhcp_server_t * server = 0;
+  int new = 0;
+
+  server = dhcp_get_server(dpm, rx_fib_index, proto);
+
+  if (NULL == server)
+  {
+      vec_validate_init_empty(dpm->dhcp_server_index_by_rx_fib_index[proto],
+                              rx_fib_index,
+                              ~0);
+
+      pool_get (dpm->dhcp_servers[proto], server);
+      memset (server, 0, sizeof (*server));
+      new = 1;
+
+      dpm->dhcp_server_index_by_rx_fib_index[proto][rx_fib_index] =
+          server - dpm->dhcp_servers[proto];
+
+      server->rx_fib_index = rx_fib_index;
+      server->server_fib_index = 
+          fib_table_find_or_create_and_lock(proto, server_table_id);
+  }
+  else
+  {
+      /* modify, may need to swap server FIBs */
+      u32 tmp_index;
+
+      tmp_index = fib_table_find(proto, server_table_id);
+
+      if (tmp_index != server->server_fib_index)
+      {
+          tmp_index = server->server_fib_index;
+
+          /* certainly swapping if the fib doesn't exist */
+          server->server_fib_index = 
+              fib_table_find_or_create_and_lock(proto, server_table_id);
+          fib_table_unlock (tmp_index, proto);
+      }
+  }
+
+  server->dhcp_server = *addr;
+  server->dhcp_src_address = *src_address;
+
+  return (new);
+}
+
+typedef struct dhcp4_proxy_dump_walk_ctx_t_
+{
+    fib_protocol_t proto;
+    void *opaque;
+    u32 context;
+} dhcp_proxy_dump_walk_cxt_t;
+
+static int
+dhcp_proxy_dump_walk (dhcp_server_t *server,
+                      void *arg)
+{
+  dhcp_proxy_dump_walk_cxt_t *ctx = arg;
+  fib_table_t *s_fib, *r_fib;
+  dhcp_vss_t *v;
+
+  v = dhcp_get_vss_info(&dhcp_proxy_main,
+                        server->rx_fib_index,
+                        ctx->proto);
+
+  s_fib = fib_table_get(server->server_fib_index, ctx->proto);
+  r_fib = fib_table_get(server->rx_fib_index, ctx->proto);
+
+  dhcp_send_details(ctx->proto,
+                    ctx->opaque,
+                    ctx->context,
+                    &server->dhcp_server,
+                    &server->dhcp_src_address,
+                    s_fib->ft_table_id,
+                    r_fib->ft_table_id,
+                    (v ? v->fib_id : 0),
+                    (v ? v->oui : 0));
+
+  return (1);
+}
+
+void
+dhcp_proxy_dump (fib_protocol_t proto,
+                 void *opaque,
+                 u32 context)
+{
+    dhcp_proxy_dump_walk_cxt_t ctx =  {
+        .proto = proto,
+        .opaque = opaque,
+        .context = context,
+    };
+    dhcp_proxy_walk(proto, dhcp_proxy_dump_walk, &ctx);
+}
+
+int
+dhcp_vss_show_walk (dhcp_vss_t *vss,
+                    u32 rx_table_id,
+                    void *ctx)
+{
+    vlib_main_t * vm = ctx;
+
+    vlib_cli_output (vm, "%=6d%=6d%=12d",
+                     rx_table_id,
+                     vss->oui,
+                     vss->fib_id);
+
+    return (1);
+}
+
+int dhcp_proxy_set_vss (fib_protocol_t proto,
+                        u32 tbl_id,
+                        u32 oui,
+                        u32 fib_id, 
+                        int is_del)
+{
+  dhcp_proxy_main_t *dm = &dhcp_proxy_main;
+  dhcp_vss_t *v = NULL;
+  u32  rx_fib_index;
+  int rc = 0;
+  
+  rx_fib_index = fib_table_find_or_create_and_lock(proto, tbl_id);
+  v = dhcp_get_vss_info(dm, rx_fib_index, proto);
+
+  if (NULL != v)
+  {
+      if (is_del)
+      {
+          /* release the lock held on the table when the VSS
+           * info was created */
+          fib_table_unlock (rx_fib_index, proto);
+
+          pool_put (dm->vss[proto], v);
+          dm->vss_index_by_rx_fib_index[proto][rx_fib_index] = ~0;
+      }
+      else
+      {
+          /* this is a modify */
+          v->fib_id = fib_id;
+          v->oui = oui;
+      }
+  }
+  else
+  {
+      if (is_del)
+          rc = VNET_API_ERROR_NO_SUCH_ENTRY;
+      else
+      {
+          /* create a new entry */
+          vec_validate_init_empty(dm->vss_index_by_rx_fib_index[proto],
+                                  rx_fib_index, ~0);
+
+          /* hold a lock on the table whilst the VSS info exist */
+          fib_table_lock (rx_fib_index, proto);
+
+          pool_get (dm->vss[proto], v);
+          v->fib_id = fib_id;
+          v->oui = oui;
+          dm->vss_index_by_rx_fib_index[proto][rx_fib_index] =
+              v - dm->vss[proto];
+      }
+  }
+
+  /* Release the lock taken during the create_or_lock at the start */
+  fib_table_unlock (rx_fib_index, proto);
+  
+  return (rc);
+}
diff --git a/src/vnet/dhcp/dhcp_proxy.h b/src/vnet/dhcp/dhcp_proxy.h
new file mode 100644
index 00000000..c0d79c41
--- /dev/null
+++ b/src/vnet/dhcp/dhcp_proxy.h
@@ -0,0 +1,248 @@
+/*
+ * dhcp_proxy.h: DHCP v4 & v6 proxy common functions/types
+ *
+ * Copyright (c) 2013 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef included_dhcp_proxy_h
+#define included_dhcp_proxy_h
+
+#include <vnet/vnet.h>
+#include <vnet/dhcp/dhcp4_packet.h>
+#include <vnet/ethernet/ethernet.h>
+#include <vnet/ip/ip.h>
+#include <vnet/ip/ip4.h>
+#include <vnet/ip/ip4_packet.h>
+#include <vnet/pg/pg.h>
+#include <vnet/ip/format.h>
+#include <vnet/ip/udp.h>
+
+typedef enum {
+#define dhcp_proxy_error(n,s) DHCP_PROXY_ERROR_##n,
+#include <vnet/dhcp/dhcp4_proxy_error.def>
+#undef dhcp_proxy_error
+  DHCP_PROXY_N_ERROR,
+} dhcp_proxy_error_t;
+
+typedef enum {
+#define dhcpv6_proxy_error(n,s) DHCPV6_PROXY_ERROR_##n,
+#include <vnet/dhcp/dhcp6_proxy_error.def>
+#undef dhcpv6_proxy_error
+  DHCPV6_PROXY_N_ERROR,
+} dhcpv6_proxy_error_t;
+
+
+/**
+ * @brief The Virtual Sub-net Selection information for a given RX FIB
+ */
+typedef struct dhcp_vss_t_ {
+    /**
+     * @brief ?? RFC doesn't say
+     */
+    u32 oui;
+    /**
+     * @brief VPN-ID
+     */
+    u32 fib_id;
+} dhcp_vss_t;
+
+/**
+ * @brief A DHCP proxy server represenation
+ */
+typedef struct dhcp_server_t_ {
+    /**
+     * @brief The address of the DHCP server to which to relay the client's
+     *        messages
+     */
+    ip46_address_t dhcp_server;
+
+    /**
+     * @brief The source address to use in relayed messaes
+     */
+    ip46_address_t dhcp_src_address;
+
+    /**
+     * @brief The FIB index (not the external Table-ID) in which the server
+     *        is reachable.
+     */
+    u32 server_fib_index;
+
+    /**
+     * @brief The FIB index (not the external Table-ID) in which the client
+     *        is resides.
+     */
+    u32 rx_fib_index;
+} dhcp_server_t;
+
+#define DHCP_N_PROTOS (FIB_PROTOCOL_IP6 + 1)
+
+/**
+ * @brief Collection of global DHCP proxy data
+ */
+typedef struct {
+  /* Pool of DHCP servers */
+  dhcp_server_t *dhcp_servers[DHCP_N_PROTOS];
+
+  /* Pool of selected DHCP server. Zero is the default server */
+  u32 * dhcp_server_index_by_rx_fib_index[DHCP_N_PROTOS];
+
+  /* to drop pkts in server-to-client direction */
+  u32 error_drop_node_index;
+
+  dhcp_vss_t *vss[DHCP_N_PROTOS];
+
+  /* hash lookup specific vrf_id -> option 82 vss suboption  */
+  u32 *vss_index_by_rx_fib_index[DHCP_N_PROTOS];
+
+} dhcp_proxy_main_t;
+
+extern dhcp_proxy_main_t dhcp_proxy_main;
+
+/**
+ * @brief Send the details of a proxy session to the API client during a dump
+ */
+void dhcp_send_details (fib_protocol_t proto,
+                        void *opaque,
+                        u32 context,
+                        const ip46_address_t *server,
+                        const ip46_address_t *src,
+                        u32 server_fib_id,
+                        u32 rx_fib_id,
+                        u32 vss_fib_id,
+                        u32 vss_oui);
+
+/**
+ * @brief Show (on CLI) a VSS config during a show walk
+ */
+int dhcp_vss_show_walk (dhcp_vss_t *vss,
+                        u32 rx_table_id,
+                        void *ctx);
+
+/**
+ * @brief Configure/set a new VSS info
+ */
+int dhcp_proxy_set_vss(fib_protocol_t proto,
+                       u32 vrf_id,
+                       u32 oui,
+                       u32 fib_id,
+                       int is_del);
+
+/**
+ * @brief Dump the proxy configs to the API
+ */
+void dhcp_proxy_dump(fib_protocol_t proto,
+                     void *opaque,
+                     u32 context);
+
+/**
+ * @brief Add a new DHCP proxy server configuration.
+ * @return 1 is the config is new,
+ *         0 otherwise (implying a modify of an existing)
+ */
+int dhcp_proxy_server_add(fib_protocol_t proto,
+                          ip46_address_t *addr,
+                          ip46_address_t *src_address,
+                          u32 rx_fib_iindex,
+                          u32 server_table_id);
+
+/**
+ * @brief Delete a DHCP proxy config
+ * @return 0 is deleted, otherwise an error code
+ */
+int dhcp_proxy_server_del(fib_protocol_t proto,
+                          u32 rx_fib_index);
+
+/**
+ * @brief Callback function invoked for each DHCP proxy entry
+ *  return 0 to break the walk, non-zero otherwise.
+ */
+typedef int (*dhcp_proxy_walk_fn_t)(dhcp_server_t *server,
+                                    void *ctx);
+
+/**
+ * @brief Walk/Visit each DHCP proxy server
+ */
+void dhcp_proxy_walk(fib_protocol_t proto,
+                     dhcp_proxy_walk_fn_t fn,
+                     void *ctx);
+
+/**
+ * @brief Callback function invoked for each DHCP VSS entry
+ *  return 0 to break the walk, non-zero otherwise.
+ */
+typedef int (*dhcp_vss_walk_fn_t)(dhcp_vss_t *server,
+                                  u32 rx_table_id,
+                                  void *ctx);
+
+/**
+ * @brief Walk/Visit each DHCP proxy VSS
+ */
+void dhcp_vss_walk(fib_protocol_t proto,
+                   dhcp_vss_walk_fn_t fn,
+                   void *ctx);
+
+/**
+ * @brief Get the VSS data for the FIB index
+ */
+static inline dhcp_vss_t *
+dhcp_get_vss_info (dhcp_proxy_main_t *dm,
+                   u32 rx_fib_index,
+                   fib_protocol_t proto)
+{
+  dhcp_vss_t *v = NULL;
+
+  if (vec_len(dm->vss_index_by_rx_fib_index[proto]) > rx_fib_index &&
+      dm->vss_index_by_rx_fib_index[proto][rx_fib_index] != ~0)
+  {
+      v = pool_elt_at_index (
+              dm->vss[proto],
+              dm->vss_index_by_rx_fib_index[proto][rx_fib_index]);
+  }
+
+  return (v);
+}
+
+/**
+ * @brief Get the DHCP proxy server data for the FIB index
+ */
+static inline dhcp_server_t *
+dhcp_get_server (dhcp_proxy_main_t *dm,
+                 u32 rx_fib_index,
+                 fib_protocol_t proto)
+{
+  dhcp_server_t *s = NULL;
+
+  if (vec_len(dm->dhcp_server_index_by_rx_fib_index[proto]) > rx_fib_index &&
+      dm->dhcp_server_index_by_rx_fib_index[proto][rx_fib_index] != ~0)
+  {
+      s = pool_elt_at_index (
+              dm->dhcp_servers[proto],
+              dm->dhcp_server_index_by_rx_fib_index[proto][rx_fib_index]);
+  }
+
+  return (s);
+}
+
+int dhcp6_proxy_set_server (ip46_address_t *addr,
+                            ip46_address_t *src_addr,
+                            u32 rx_table_id,
+                            u32 server_table_id,
+                            int is_del);
+int dhcp4_proxy_set_server (ip46_address_t *addr,
+                            ip46_address_t *src_addr,
+                            u32 rx_table_id,
+                            u32 server_table_id,
+                            int is_del);
+
+#endif /* included_dhcp_proxy_h */
diff --git a/src/vnet/dhcp/packet.h b/src/vnet/dhcp/packet.h
deleted file mode 100644
index 267a8eaf..00000000
--- a/src/vnet/dhcp/packet.h
+++ /dev/null
@@ -1,61 +0,0 @@
-#ifndef included_vnet_dhcp_packet_h
-#define included_vnet_dhcp_packet_h
-
-/*
- * DHCP packet format
- *
- * Copyright (c) 2013 Cisco and/or its affiliates.
- * Licensed under the Apache License, Version 2.0 (the "License");
- * you may not use this file except in compliance with the License.
- * You may obtain a copy of the License at:
- *
- *     http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-#include <vnet/ip/ip4_packet.h>
-
-typedef struct {
-  u8 opcode;                    /* 1 = request, 2 = reply */
-  u8 hardware_type;             /* 1 = ethernet */
-  u8 hardware_address_length;
-  u8 hops;
-  u32 transaction_identifier;
-  u16 seconds;
-  u16 flags;
-#define DHCP_FLAG_BROADCAST (1<<15)
-  ip4_address_t client_ip_address;
-  ip4_address_t your_ip_address; /* use this one */
-  ip4_address_t server_ip_address;
-  ip4_address_t gateway_ip_address; /* use option 3, not this one */
-  u8 client_hardware_address[16];
-  u8 server_name[64];
-  u8 boot_filename[128];
-  ip4_address_t magic_cookie;
-  u8 options[0];
-} dhcp_header_t;
-
-typedef struct {
-  u8 option;
-  u8 length;
-  union {
-    u8 data[0];
-    u32 data_as_u32[0];
-  };
-} __attribute__((packed)) dhcp_option_t;
-
-typedef enum {
-  DHCP_PACKET_DISCOVER=1,
-  DHCP_PACKET_OFFER,
-  DHCP_PACKET_REQUEST,
-  DHCP_PACKET_ACK=5,
-} dhcp_packet_type_t;
-
-/* charming antique: 99.130.83.99 is the dhcp magic cookie */
-#define DHCP_MAGIC (clib_host_to_net_u32(0x63825363))
-
-#endif /* included_vnet_dhcp_packet_h */
diff --git a/src/vnet/dhcp/proxy.h b/src/vnet/dhcp/proxy.h
deleted file mode 100644
index 4b115c74..00000000
--- a/src/vnet/dhcp/proxy.h
+++ /dev/null
@@ -1,99 +0,0 @@
-/*
- * proxy.h: dhcp proxy 
- *
- * Copyright (c) 2013 Cisco and/or its affiliates.
- * Licensed under the Apache License, Version 2.0 (the "License");
- * you may not use this file except in compliance with the License.
- * You may obtain a copy of the License at:
- *
- *     http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-
-#ifndef included_dhcp_proxy_h
-#define included_dhcp_proxy_h
-
-#include <vnet/vnet.h>
-#include <vnet/dhcp/packet.h>
-#include <vnet/ethernet/ethernet.h>
-#include <vnet/ip/ip.h>
-#include <vnet/ip/ip4.h>
-#include <vnet/ip/ip4_packet.h>
-#include <vnet/pg/pg.h>
-#include <vnet/ip/format.h>
-#include <vnet/ip/udp.h>
-
-typedef enum {
-#define dhcp_proxy_error(n,s) DHCP_PROXY_ERROR_##n,
-#include <vnet/dhcp/proxy_error.def>
-#undef dhcp_proxy_error
-  DHCP_PROXY_N_ERROR,
-} dhcp_proxy_error_t;
-
-typedef struct {
-  u32 oui;
-  u32 fib_id;
-} vss_id;
-
-typedef union {
-  u8 as_u8[8];
-  vss_id vpn_id;
-} vss_info;
-
-typedef struct {
-  ip4_address_t dhcp_server;
-  ip4_address_t dhcp_src_address;
-  u32 server_fib_index;
-} dhcp_server_t;
-
-typedef struct {
-  /* Pool of DHCP servers */
-  dhcp_server_t * dhcp_servers;
-
-  /* Pool of selected DHCP server. Zero is the default server */
-  u32 * dhcp_server_index_by_rx_fib_index;
-
-  /* to drop pkts in server-to-client direction */
-  u32 error_drop_node_index;
-
-  vss_info *vss;
-
-  /* hash lookup specific vrf_id -> option 82 vss suboption  */
-  u32 *vss_index_by_rx_fib_index;
-
-  /* convenience */
-  vlib_main_t * vlib_main;
-  vnet_main_t * vnet_main;
-} dhcp_proxy_main_t;
-
-extern dhcp_proxy_main_t dhcp_proxy_main;
-
-void dhcp_send_details (void *opaque,
-                        u32 context,
-                        const ip46_address_t *server,
-                        const ip46_address_t *src,
-                        u32 server_fib_id,
-                        u32 rx_fib_id,
-                        u32 vss_fib_id,
-                        u32 vss_oui);
-
-int dhcp_proxy_set_server (ip4_address_t *addr,
-                           ip4_address_t *src_address,
-                           u32 fib_id,
-                           u32 server_fib_id, 
-                           int is_del);
-
-int dhcp_proxy_set_option82_vss(u32 vrf_id,
-                                u32 oui,
-                                u32 fib_id, 
-                                int is_del);
-
-void dhcp_proxy_dump(void *opaque,
-                     u32 context);
-
-#endif /* included_dhcp_proxy_h */
diff --git a/src/vnet/dhcp/proxy_error.def b/src/vnet/dhcp/proxy_error.def
deleted file mode 100644
index 6d790d73..00000000
--- a/src/vnet/dhcp/proxy_error.def
+++ /dev/null
@@ -1,31 +0,0 @@
-/*
- * dhcp_proxy_error.def: dhcp proxy errors
- *
- * Copyright (c) 2013 Cisco and/or its affiliates.
- * Licensed under the Apache License, Version 2.0 (the "License");
- * you may not use this file except in compliance with the License.
- * You may obtain a copy of the License at:
- *
- *     http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-
-dhcp_proxy_error (NONE, "no error")
-dhcp_proxy_error (NO_SERVER, "no dhcp server configured")
-dhcp_proxy_error (RELAY_TO_SERVER, "DHCP packets relayed to the server")
-dhcp_proxy_error (RELAY_TO_CLIENT, "DHCP packets relayed to clients")
-dhcp_proxy_error (OPTION_82_ERROR, "DHCP failed to insert option 82")
-dhcp_proxy_error (NO_OPTION_82, "DHCP option 82 missing")
-dhcp_proxy_error (BAD_OPTION_82_ITF, "Bad DHCP option 82 interface value")
-dhcp_proxy_error (BAD_OPTION_82_ADDR, "Bad DHCP option 82 address value")
-dhcp_proxy_error (BAD_FIB_ID, "DHCP option 82 fib-id to fib-index map failure")
-dhcp_proxy_error (NO_INTERFACE_ADDRESS, "DHCP no interface address")
-dhcp_proxy_error (OPTION_82_VSS_NOT_PROCESSED, "DHCP VSS not processed by DHCP server")
-dhcp_proxy_error (BAD_YIADDR, "DHCP packets with bad your_ip_address fields")
-dhcp_proxy_error (BAD_SVR_FIB_OR_ADDRESS, "DHCP packets not from DHCP server or server FIB.")
-dhcp_proxy_error (PKT_TOO_BIG, "DHCP packets which are too big.")
diff --git a/src/vnet/dhcp/proxy_node.c b/src/vnet/dhcp/proxy_node.c
deleted file mode 100644
index ab6819fe..00000000
--- a/src/vnet/dhcp/proxy_node.c
+++ /dev/null
@@ -1,1192 +0,0 @@
-/*
- * proxy_node.c: dhcp proxy node processing
- *
- * Copyright (c) 2013 Cisco and/or its affiliates.
- * Licensed under the Apache License, Version 2.0 (the "License");
- * you may not use this file except in compliance with the License.
- * You may obtain a copy of the License at:
- *
- *     http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-
-#include <vlib/vlib.h>
-#include <vnet/pg/pg.h>
-#include <vnet/dhcp/proxy.h>
-#include <vnet/dhcp/client.h>
-#include <vnet/fib/ip4_fib.h>
-
-static char * dhcp_proxy_error_strings[] = {
-#define dhcp_proxy_error(n,s) s,
-#include "proxy_error.def"
-#undef dhcp_proxy_error
-};
-
-#define foreach_dhcp_proxy_to_server_input_next \
-  _ (DROP, "error-drop")			\
-  _ (LOOKUP, "ip4-lookup")			\
-  _ (SEND_TO_CLIENT, "dhcp-proxy-to-client")
-
-typedef enum {
-#define _(s,n) DHCP_PROXY_TO_SERVER_INPUT_NEXT_##s,
-  foreach_dhcp_proxy_to_server_input_next
-#undef _
-  DHCP_PROXY_TO_SERVER_INPUT_N_NEXT,
-} dhcp_proxy_to_server_input_next_t;
-
-typedef struct {
-  /* 0 => to server, 1 => to client */
-  int which; 
-  ip4_address_t trace_ip4_address;
-  u32 error;
-  u32 sw_if_index;
-  u32 original_sw_if_index;
-} dhcp_proxy_trace_t;
-
-#define VPP_DHCP_OPTION82_SUB1_SIZE   6
-#define VPP_DHCP_OPTION82_SUB5_SIZE   6
-#define VPP_DHCP_OPTION82_VSS_SIZE    12
-#define VPP_DHCP_OPTION82_SIZE (VPP_DHCP_OPTION82_SUB1_SIZE + \
-                                VPP_DHCP_OPTION82_SUB5_SIZE + \
-                                VPP_DHCP_OPTION82_VSS_SIZE +3)
-
-vlib_node_registration_t dhcp_proxy_to_server_node;
-vlib_node_registration_t dhcp_proxy_to_client_node;
-
-dhcp_proxy_main_t dhcp_proxy_main;
-
-u8 * format_dhcp_proxy_trace (u8 * s, va_list * args)
-{
-  CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *);
-  CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *);
-  dhcp_proxy_trace_t * t = va_arg (*args, dhcp_proxy_trace_t *);
-    
-  if (t->which == 0)
-    s = format (s, "DHCP proxy: sent to server %U\n",
-                format_ip4_address, &t->trace_ip4_address, t->error);
-  else
-    s = format (s, "DHCP proxy: broadcast to client from %U\n",
-                format_ip4_address, &t->trace_ip4_address);
-      
-  if (t->error != (u32)~0)
-    s = format (s, "  error: %s\n", dhcp_proxy_error_strings[t->error]);
-
-  s = format (s, "  original_sw_if_index: %d, sw_if_index: %d\n",
-              t->original_sw_if_index, t->sw_if_index);
-  
-  return s;
-}
-
-u8 * format_dhcp_proxy_header_with_length (u8 * s, va_list * args)
-{
-  dhcp_header_t * h = va_arg (*args, dhcp_header_t *);
-  u32 max_header_bytes = va_arg (*args, u32);
-  u32 header_bytes;
-
-  header_bytes = sizeof (h[0]);
-  if (max_header_bytes != 0 && header_bytes > max_header_bytes)
-    return format (s, "dhcp header truncated");
-
-  s = format (s, "DHCP Proxy");
-
-  return s;
-}
-
-static inline vss_info *
-dhcp_get_vss_info (dhcp_proxy_main_t *dm,
-                   u32 rx_fib_index)
-{
-  vss_info *v;
-
-  if (vec_len(dm->vss_index_by_rx_fib_index) <= rx_fib_index ||
-      dm->vss_index_by_rx_fib_index[rx_fib_index] == ~0)
-  {
-      v = NULL;
-  }
-  else
-  {
-      v = pool_elt_at_index (dm->vss,
-                             dm->vss_index_by_rx_fib_index[rx_fib_index]);
-  }
-
-  return (v);
-}
-
-static inline dhcp_server_t *
-dhcp_get_server (dhcp_proxy_main_t *dm,
-                 u32 rx_fib_index)
-{
-  dhcp_server_t *s = NULL;
-
-  if (vec_len(dm->dhcp_server_index_by_rx_fib_index) > rx_fib_index &&
-      dm->dhcp_server_index_by_rx_fib_index[rx_fib_index] != ~0)
-  {
-      s = pool_elt_at_index (dm->dhcp_servers,
-                             dm->dhcp_server_index_by_rx_fib_index[rx_fib_index]);
-  }
-
-  return (s);
-}
-
-static uword
-dhcp_proxy_to_server_input (vlib_main_t * vm,
-                            vlib_node_runtime_t * node,
-                            vlib_frame_t * from_frame)
-{
-  u32 n_left_from, next_index, * from, * to_next;
-  dhcp_proxy_main_t * dpm = &dhcp_proxy_main;
-  from = vlib_frame_vector_args (from_frame);
-  n_left_from = from_frame->n_vectors;
-  u32 pkts_to_server=0, pkts_to_client=0, pkts_no_server=0;
-  u32 pkts_no_interface_address=0;
-  u32 pkts_too_big=0;
-  ip4_main_t * im = &ip4_main;
-
-  next_index = node->cached_next_index;
-
-  while (n_left_from > 0)
-    {
-      u32 n_left_to_next;
-
-      vlib_get_next_frame (vm, node, next_index,
-			   to_next, n_left_to_next);
-
-      while (n_left_from > 0 && n_left_to_next > 0)
-	{
-	  u32 bi0;
-	  vlib_buffer_t * b0;
-          udp_header_t * u0;
-	  dhcp_header_t * h0;
-          ip4_header_t * ip0;
-	  u32 next0;
-          u32 old0, new0;
-          ip_csum_t sum0;
-          u32 error0 = (u32) ~0;
-          u32 sw_if_index = 0;
-          u32 original_sw_if_index = 0;
-          u8  *end = NULL;
-          u32 fib_index;
-          dhcp_server_t * server;
-          u32 rx_sw_if_index;
-          dhcp_option_t *o;
-          u32 len = 0;
-          vlib_buffer_free_list_t *fl;
-
-	  bi0 = from[0];
-	  to_next[0] = bi0;
-	  from += 1;
-	  to_next += 1;
-	  n_left_from -= 1;
-	  n_left_to_next -= 1;
-
-	  b0 = vlib_get_buffer (vm, bi0);
-
-          h0 = vlib_buffer_get_current (b0);
-
-          /* 
-           * udp_local hands us the DHCP header, need udp hdr, 
-           * ip hdr to relay to server
-           */
-          vlib_buffer_advance (b0, -(sizeof(*u0)));
-	  u0 = vlib_buffer_get_current (b0);
-
-          /* This blows. Return traffic has src_port = 67, dst_port = 67 */
-          if (u0->src_port == clib_net_to_host_u16(UDP_DST_PORT_dhcp_to_server))
-            {
-              vlib_buffer_advance (b0, sizeof(*u0));
-              next0 = DHCP_PROXY_TO_SERVER_INPUT_NEXT_SEND_TO_CLIENT;
-              error0 = 0;
-              pkts_to_client++;
-              goto do_enqueue;
-            }
-
-          rx_sw_if_index = vnet_buffer(b0)->sw_if_index[VLIB_RX];
-
-          fib_index = im->fib_index_by_sw_if_index [rx_sw_if_index];
-          server = dhcp_get_server(dpm, fib_index);
-          
-          if (PREDICT_FALSE (NULL == server))
-            {
-              error0 = DHCP_PROXY_ERROR_NO_SERVER;
-              next0 = DHCP_PROXY_TO_SERVER_INPUT_NEXT_DROP;
-              pkts_no_server++;
-              goto do_trace;
-            }
-          
-          vlib_buffer_advance (b0, -(sizeof(*ip0)));
-          ip0 = vlib_buffer_get_current (b0);
-
-          /* disable UDP checksum */
-          u0->checksum = 0;
-          sum0 = ip0->checksum;
-          old0 = ip0->dst_address.as_u32;
-          new0 = server->dhcp_server.as_u32;
-          ip0->dst_address.as_u32 = server->dhcp_server.as_u32;
-          sum0 = ip_csum_update (sum0, old0, new0, 
-                                ip4_header_t /* structure */, 
-                                dst_address /* changed member */);
-          ip0->checksum = ip_csum_fold (sum0);
-
-          sum0 = ip0->checksum;
-          old0 = ip0->src_address.as_u32;
-          new0 = server->dhcp_src_address.as_u32;
-          ip0->src_address.as_u32 = new0;
-          sum0 = ip_csum_update (sum0, old0, new0, 
-                                ip4_header_t /* structure */, 
-                                src_address /* changed member */);
-          ip0->checksum = ip_csum_fold (sum0);
-
-          /* Send to DHCP server via the configured FIB */
-          vnet_buffer(b0)->sw_if_index[VLIB_TX] =
-            server->server_fib_index;
-
-          h0->gateway_ip_address.as_u32 = server->dhcp_src_address.as_u32;
-          pkts_to_server++;
-
-          o = (dhcp_option_t *) h0->options;
-              
-          fib_index = im->fib_index_by_sw_if_index 
-              [vnet_buffer(b0)->sw_if_index[VLIB_RX]];
-
-          end = b0->data + b0->current_data + b0->current_length;
-          /* TLVs are not performance-friendly... */
-          while  (o->option != 0xFF /* end of options */ && (u8 *)o < end) 
-              o = (dhcp_option_t *) (((uword) o) + (o->length + 2));
-
-          fl = vlib_buffer_get_free_list (vm, b0->free_list_index);
-          // start write at (option*)o, some packets have padding
-          if (((u8 *)o - (u8 *)b0->data + VPP_DHCP_OPTION82_SIZE) > fl->n_data_bytes)
-          {
-              next0 = DHCP_PROXY_TO_SERVER_INPUT_NEXT_DROP;
-              pkts_too_big++;
-              goto do_trace;
-          }
-
-          if ((o->option == 0xFF)  && ((u8 *)o <= end))
-          {  
-              vnet_main_t *vnm = vnet_get_main();   
-              u16 old_l0, new_l0;
-              ip4_address_t _ia0, * ia0 = &_ia0;
-              vss_info *vss;
-              vnet_sw_interface_t *swif;
-              sw_if_index = 0;
-              original_sw_if_index = 0;
-                  
-              original_sw_if_index = sw_if_index = 
-                  vnet_buffer(b0)->sw_if_index[VLIB_RX];
-              swif = vnet_get_sw_interface (vnm, sw_if_index);
-              if (swif->flags & VNET_SW_INTERFACE_FLAG_UNNUMBERED)
-                  sw_if_index = swif->unnumbered_sw_if_index;
-                  
-              /* 
-               * Get the first ip4 address on the [client-side] 
-               * RX interface, if not unnumbered. otherwise use
-               * the loopback interface's ip address.
-               */
-              ia0 = ip4_interface_first_address(&ip4_main, sw_if_index, 0);
-                  
-              if (ia0 == 0)
-              {
-                  error0 = DHCP_PROXY_ERROR_NO_INTERFACE_ADDRESS;
-                  next0 = DHCP_PROXY_TO_SERVER_INPUT_NEXT_DROP;
-                  pkts_no_interface_address++;
-                  goto do_trace;
-              }
-
-              /* Add option 82 */
-              o->option = 82;   /* option 82 */
-              o->length = 12;   /* 12 octets to follow */
-              o->data[0] = 1;   /* suboption 1, circuit ID (=FIB id) */
-              o->data[1] = 4;   /* length of suboption */
-              o->data[2] = (original_sw_if_index >> 24) & 0xFF;
-              o->data[3] = (original_sw_if_index >> 16) & 0xFF;
-              o->data[4] = (original_sw_if_index >> 8)  & 0xFF;
-              o->data[5] = (original_sw_if_index >> 0)  & 0xFF;
-              o->data[6] = 5; /* suboption 5 (client RX intfc address) */
-              o->data[7] = 4; /* length 4 */
-              o->data[8] = ia0->as_u8[0];
-              o->data[9] = ia0->as_u8[1];
-              o->data[10] = ia0->as_u8[2];
-              o->data[11] = ia0->as_u8[3];
-              o->data[12] = 0xFF;
-
-              vss = dhcp_get_vss_info (dpm, fib_index);
-              if (NULL != vss)
-              {
-                  u32 opt82_fib_id=0, opt82_oui=0;
-
-                  opt82_oui =  vss->vpn_id.oui;
-                  opt82_fib_id =  vss->vpn_id.fib_id;
-
-                  o->data[12] = 151; /* vss suboption */
-                  if (255 == opt82_fib_id) {
-                      o->data[13] = 1;   /* length */
-                      o->data[14] = 255;   /* vss option type */
-                      o->data[15] = 152; /* vss control suboption */
-                      o->data[16] = 0;   /* length */
-                      /* and a new "end-of-options" option (0xff) */
-                      o->data[17] = 0xFF;
-                      o->length += 5;
-                  } else {
-                      o->data[13] = 8;   /* length */
-                      o->data[14] = 1;   /* vss option type */
-                      o->data[15] = (opt82_oui >> 16) & 0xff;
-                      o->data[16] = (opt82_oui >> 8) & 0xff;
-                      o->data[17] = (opt82_oui ) & 0xff;
-                      o->data[18] = (opt82_fib_id >> 24) & 0xff;
-                      o->data[19] = (opt82_fib_id >> 16) & 0xff;
-                      o->data[20] = (opt82_fib_id >> 8) & 0xff;
-                      o->data[21] = (opt82_fib_id) & 0xff;
-                      o->data[22] = 152; /* vss control suboption */
-                      o->data[23] = 0;   /* length */
-                          
-                      /* and a new "end-of-options" option (0xff) */
-                      o->data[24] = 0xFF;
-                      o->length += 12;
-                  }
-              }
-
-              len = o->length + 3;
-              b0->current_length += len;
-              /* Fix IP header length and checksum */
-              old_l0 = ip0->length;
-              new_l0 = clib_net_to_host_u16 (old_l0);
-              new_l0 += len;
-              new_l0 = clib_host_to_net_u16 (new_l0);
-              ip0->length = new_l0;
-              sum0 = ip0->checksum;
-              sum0 = ip_csum_update (sum0, old_l0, new_l0, ip4_header_t,
-                                     length /* changed member */);
-              ip0->checksum = ip_csum_fold (sum0);
-
-              /* Fix UDP length */
-              new_l0 = clib_net_to_host_u16 (u0->length);
-              new_l0 += len;
-              u0->length = clib_host_to_net_u16 (new_l0);
-          } else {
-              vlib_node_increment_counter 
-                  (vm, dhcp_proxy_to_server_node.index,
-                   DHCP_PROXY_ERROR_OPTION_82_ERROR, 1);
-          }
-          
-          next0 = DHCP_PROXY_TO_SERVER_INPUT_NEXT_LOOKUP;
-
-        do_trace:
-          if (PREDICT_FALSE(b0->flags & VLIB_BUFFER_IS_TRACED)) 
-            {
-               dhcp_proxy_trace_t *tr = vlib_add_trace (vm, node, 
-                                                        b0, sizeof (*tr));
-               tr->which = 0; /* to server */
-               tr->error = error0;
-               tr->original_sw_if_index = original_sw_if_index;
-               tr->sw_if_index = sw_if_index;
-               if (next0 == DHCP_PROXY_TO_SERVER_INPUT_NEXT_LOOKUP)
-                 tr->trace_ip4_address.as_u32 = server->dhcp_server.as_u32;
-            }
-
-        do_enqueue:
-	  vlib_validate_buffer_enqueue_x1 (vm, node, next_index,
-					   to_next, n_left_to_next,
-					   bi0, next0);
-	}
-
-      vlib_put_next_frame (vm, node, next_index, n_left_to_next);
-    }
-
-  vlib_node_increment_counter (vm, dhcp_proxy_to_server_node.index,
-                               DHCP_PROXY_ERROR_RELAY_TO_CLIENT,
-                               pkts_to_client);
-  vlib_node_increment_counter (vm, dhcp_proxy_to_server_node.index,
-                               DHCP_PROXY_ERROR_RELAY_TO_SERVER,
-                               pkts_to_server);
-  vlib_node_increment_counter (vm, dhcp_proxy_to_server_node.index,
-                               DHCP_PROXY_ERROR_NO_SERVER,
-                               pkts_no_server);
-  vlib_node_increment_counter (vm, dhcp_proxy_to_server_node.index,
-                               DHCP_PROXY_ERROR_NO_INTERFACE_ADDRESS,
-                               pkts_no_interface_address);
- vlib_node_increment_counter (vm, dhcp_proxy_to_server_node.index,
-                              DHCP_PROXY_ERROR_PKT_TOO_BIG,
-                              pkts_too_big);
-  return from_frame->n_vectors;
-}
-
-VLIB_REGISTER_NODE (dhcp_proxy_to_server_node) = {
-  .function = dhcp_proxy_to_server_input,
-  .name = "dhcp-proxy-to-server",
-  /* Takes a vector of packets. */
-  .vector_size = sizeof (u32),
-
-  .n_errors = DHCP_PROXY_N_ERROR,
-  .error_strings = dhcp_proxy_error_strings,
-
-  .n_next_nodes = DHCP_PROXY_TO_SERVER_INPUT_N_NEXT,
-  .next_nodes = {
-#define _(s,n) [DHCP_PROXY_TO_SERVER_INPUT_NEXT_##s] = n,
-    foreach_dhcp_proxy_to_server_input_next
-#undef _
-  },
-
-  .format_buffer = format_dhcp_proxy_header_with_length,
-  .format_trace = format_dhcp_proxy_trace,
-#if 0
-  .unformat_buffer = unformat_dhcp_proxy_header,
-#endif
-};
-
-static uword
-dhcp_proxy_to_client_input (vlib_main_t * vm,
-                            vlib_node_runtime_t * node,
-                            vlib_frame_t * from_frame)
-{
-  u32 n_left_from, * from;
-  ethernet_main_t *em = ethernet_get_main (vm);
-  dhcp_proxy_main_t * dpm = &dhcp_proxy_main;
-  vnet_main_t * vnm = vnet_get_main();
-  ip4_main_t * im = &ip4_main;
-
-  from = vlib_frame_vector_args (from_frame);
-  n_left_from = from_frame->n_vectors;
-
-  while (n_left_from > 0)
-    {
-      u32 bi0;
-      vlib_buffer_t * b0;
-      udp_header_t * u0;
-      dhcp_header_t * h0;
-      ip4_header_t * ip0 = 0;
-      ip4_address_t * ia0 = 0;
-      u32 old0, new0;
-      ip_csum_t sum0;
-      ethernet_interface_t *ei0;
-      ethernet_header_t *mac0;
-      vnet_hw_interface_t *hi0;
-      vlib_frame_t *f0;
-      u32 * to_next0;
-      u32 sw_if_index = ~0;
-      vnet_sw_interface_t *si0;
-      u32 error0 = (u32)~0;
-      vnet_sw_interface_t *swif;
-      u32 fib_index;
-      dhcp_server_t * server;
-      u32 original_sw_if_index = (u32) ~0;
-      ip4_address_t relay_addr = {
-          .as_u32 = 0,
-      };
-
-      bi0 = from[0];
-      from += 1;
-      n_left_from -= 1;
-
-      b0 = vlib_get_buffer (vm, bi0);
-      h0 = vlib_buffer_get_current (b0);
-
-      /* 
-       * udp_local hands us the DHCP header, need udp hdr, 
-       * ip hdr to relay to client
-       */
-      vlib_buffer_advance (b0, -(sizeof(*u0)));
-      u0 = vlib_buffer_get_current (b0);
-
-      vlib_buffer_advance (b0, -(sizeof(*ip0)));
-      ip0 = vlib_buffer_get_current (b0);
-
-      /* Consumed by dhcp client code? */
-      if (dhcp_client_for_us (bi0, b0, ip0, u0, h0))
-          continue;
-
-      if (1 /* dpm->insert_option_82 */)
-        {
-          dhcp_option_t *o = (dhcp_option_t *) h0->options;
-          dhcp_option_t *sub;
-              
-          /* Parse through TLVs looking for option 82.
-             The circuit-ID is the FIB number we need
-             to track down the client-facing interface */
-
-          while (o->option != 0xFF /* end of options */ &&
-                 (u8 *) o < (b0->data + b0->current_data + b0->current_length))
-            {
-              if (o->option == 82)
-                {
-                    u32 vss_exist = 0;
-                    u32 vss_ctrl = 0;
-                    sub = (dhcp_option_t *) &o->data[0];
-                    while (sub->option != 0xFF /* end of options */ &&
-                           (u8 *) sub < (u8 *)(o + o->length)) {
-                        /* If this is one of ours, it will have
-                           total length 12, circuit-id suboption type,
-                           and the sw_if_index */
-                        if (sub->option == 1 && sub->length == 4)
-                          {
-                            sw_if_index = ((sub->data[0] << 24) |
-                                           (sub->data[1] << 16) |
-                                           (sub->data[2] << 8)  |
-                                           (sub->data[3]));
-                          }
-                        else if (sub->option == 5 && sub->length == 4)
-                          {
-                              relay_addr.as_u8[0] = sub->data[0];
-                              relay_addr.as_u8[1] = sub->data[1];
-                              relay_addr.as_u8[2] = sub->data[2];
-                              relay_addr.as_u8[3] = sub->data[3];
-                          }
-                        else if (sub->option == 151 &&
-                                 sub->length == 7 &&
-                                 sub->data[0] == 1)
-                            vss_exist = 1;
-                        else if (sub->option == 152 && sub->length == 0)
-                            vss_ctrl = 1;
-                        sub = (dhcp_option_t *) 
-                          (((uword) sub) + (sub->length + 2));
-                    }
-                    if (vss_ctrl && vss_exist)
-                      vlib_node_increment_counter 
-                        (vm, dhcp_proxy_to_client_node.index,
-                         DHCP_PROXY_ERROR_OPTION_82_VSS_NOT_PROCESSED, 1);
-
-                }
-              o = (dhcp_option_t *) (((uword) o) + (o->length + 2));
-            }
-        }
-
-      if (sw_if_index == (u32)~0)
-        {
-          error0 = DHCP_PROXY_ERROR_NO_OPTION_82;
-          
-        drop_packet:
-          vlib_node_increment_counter (vm, dhcp_proxy_to_client_node.index,
-                                       error0, 1);
-          f0 = vlib_get_frame_to_node (vm, dpm->error_drop_node_index);
-          to_next0 = vlib_frame_vector_args (f0);
-          to_next0[0] = bi0;
-          f0->n_vectors = 1;
-          vlib_put_frame_to_node (vm, dpm->error_drop_node_index, f0);
-          goto do_trace;
-        }
-      
-      if (relay_addr.as_u32 == 0)
-        {
-          error0 = DHCP_PROXY_ERROR_BAD_OPTION_82_ADDR;
-          goto drop_packet;
-        }
-
-      if (sw_if_index >= vec_len (im->fib_index_by_sw_if_index))
-        {
-          error0 = DHCP_PROXY_ERROR_BAD_OPTION_82_ITF;
-          goto drop_packet;
-        }
-
-      fib_index = im->fib_index_by_sw_if_index [sw_if_index];
-      server = dhcp_get_server(dpm, fib_index);
-
-      if (PREDICT_FALSE (NULL == server))
-        {
-          error0 = DHCP_PROXY_ERROR_NO_SERVER;
-          goto drop_packet;
-        }
-      
-      if (ip0->src_address.as_u32 != server->dhcp_server.as_u32)
-        {             
-          error0 = DHCP_PROXY_ERROR_BAD_SVR_FIB_OR_ADDRESS;
-          goto drop_packet;
-        }
-
-      vnet_buffer (b0)->sw_if_index[VLIB_TX] = sw_if_index;
-
-      swif = vnet_get_sw_interface (vnm, sw_if_index);
-      original_sw_if_index = sw_if_index;
-      if (swif->flags & VNET_SW_INTERFACE_FLAG_UNNUMBERED)
-          sw_if_index = swif->unnumbered_sw_if_index;
-
-      ia0 = ip4_interface_first_address (&ip4_main, sw_if_index, 0);
-      if (ia0 == 0)
-        {
-          error0 = DHCP_PROXY_ERROR_NO_INTERFACE_ADDRESS;
-          goto drop_packet;
-        }
-
-      if (relay_addr.as_u32 != ia0->as_u32)
-        {             
-          error0 = DHCP_PROXY_ERROR_BAD_YIADDR;
-          goto drop_packet;
-        }
-
-      u0->checksum = 0;
-      u0->dst_port = clib_net_to_host_u16 (UDP_DST_PORT_dhcp_to_client);
-      sum0 = ip0->checksum;
-      old0 = ip0->dst_address.as_u32;
-      new0 = 0xFFFFFFFF;
-      ip0->dst_address.as_u32 = new0;
-      sum0 = ip_csum_update (sum0, old0, new0, 
-                            ip4_header_t /* structure */, 
-                            dst_address /* offset of changed member */);
-      ip0->checksum = ip_csum_fold (sum0);
-
-      sum0 = ip0->checksum;
-      old0 = ip0->src_address.as_u32;
-      new0 = ia0->as_u32;
-      ip0->src_address.as_u32 = new0;
-      sum0 = ip_csum_update (sum0, old0, new0, 
-                            ip4_header_t /* structure */, 
-                            src_address /* offset of changed member */);
-      ip0->checksum = ip_csum_fold (sum0);
-
-      vlib_buffer_advance (b0, -(sizeof(ethernet_header_t)));
-      si0 = vnet_get_sw_interface (vnm, original_sw_if_index);
-      if (si0->type == VNET_SW_INTERFACE_TYPE_SUB)
-	  vlib_buffer_advance (b0, -4 /* space for VLAN tag */);
-
-      mac0 = vlib_buffer_get_current (b0);
-
-      hi0 = vnet_get_sup_hw_interface (vnm, original_sw_if_index);
-      ei0 = pool_elt_at_index (em->interfaces, hi0->hw_instance);
-      clib_memcpy (mac0->src_address, ei0->address, sizeof (ei0->address));
-      memset (mac0->dst_address, 0xff, sizeof (mac0->dst_address));
-      mac0->type = (si0->type == VNET_SW_INTERFACE_TYPE_SUB) ?
-	clib_net_to_host_u16(0x8100) : clib_net_to_host_u16 (0x0800);
-
-      if (si0->type == VNET_SW_INTERFACE_TYPE_SUB)
-	{
-	  u32 * vlan_tag = (u32 *)(mac0+1);
-	  u32 tmp;
-	  tmp = (si0->sub.id << 16) | 0x0800;
-	  *vlan_tag = clib_host_to_net_u32 (tmp);
-	}
-
-      /* $$$ This needs to be rewritten, for sure */
-      f0 = vlib_get_frame_to_node (vm, hi0->output_node_index);
-      to_next0 = vlib_frame_vector_args (f0);
-      to_next0[0] = bi0;
-      f0->n_vectors = 1;
-      vlib_put_frame_to_node (vm, hi0->output_node_index, f0);
-
-    do_trace:
-      if (PREDICT_FALSE(b0->flags & VLIB_BUFFER_IS_TRACED)) 
-        {
-          dhcp_proxy_trace_t *tr = vlib_add_trace (vm, node, 
-                                                   b0, sizeof (*tr));
-          tr->which = 1; /* to client */
-          tr->trace_ip4_address.as_u32 = ia0 ? ia0->as_u32 : 0;
-          tr->error = error0;
-          tr->original_sw_if_index = original_sw_if_index;
-          tr->sw_if_index = sw_if_index;
-        }
-    }
-  return from_frame->n_vectors;
-}
-
-VLIB_REGISTER_NODE (dhcp_proxy_to_client_node) = {
-  .function = dhcp_proxy_to_client_input,
-  .name = "dhcp-proxy-to-client",
-  /* Takes a vector of packets. */
-  .vector_size = sizeof (u32),
-
-  .n_errors = DHCP_PROXY_N_ERROR,
-  .error_strings = dhcp_proxy_error_strings,
-  .format_buffer = format_dhcp_proxy_header_with_length,
-  .format_trace = format_dhcp_proxy_trace,
-#if 0
-  .unformat_buffer = unformat_dhcp_proxy_header,
-#endif
-};
-
-clib_error_t * dhcp_proxy_init (vlib_main_t * vm)
-{
-  dhcp_proxy_main_t * dm = &dhcp_proxy_main;
-  vlib_node_t * error_drop_node;
-  dhcp_server_t * server;
-
-  dm->vlib_main = vm;
-  dm->vnet_main = vnet_get_main();
-  error_drop_node = vlib_get_node_by_name (vm, (u8 *) "error-drop");
-  dm->error_drop_node_index = error_drop_node->index;
-
-  dm->vss_index_by_rx_fib_index = NULL;
-
-  udp_register_dst_port (vm, UDP_DST_PORT_dhcp_to_client, 
-                         dhcp_proxy_to_client_node.index, 1 /* is_ip4 */);
-
-  udp_register_dst_port (vm, UDP_DST_PORT_dhcp_to_server, 
-                         dhcp_proxy_to_server_node.index, 1 /* is_ip4 */);
-
-  /* Create the default server, don't mark it valid */
-  pool_get (dm->dhcp_servers, server);
-  memset (server, 0, sizeof (*server));
-
-  return 0;
-}
-
-VLIB_INIT_FUNCTION (dhcp_proxy_init);
-
-int dhcp_proxy_set_server (ip4_address_t *addr,
-                           ip4_address_t *src_address,
-                           u32 rx_fib_id,
-                           u32 server_fib_id, 
-                           int is_del)
-{
-  dhcp_proxy_main_t * dpm = &dhcp_proxy_main;
-  dhcp_server_t * server = 0;
-  u32 server_index = 0;
-  u32 rx_fib_index = 0;
-
-  const fib_prefix_t all_1s =
-  {
-      .fp_len = 32,
-      .fp_addr.ip4.as_u32 = 0xffffffff,
-      .fp_proto = FIB_PROTOCOL_IP4,
-  };
-
-  if (addr->as_u32 == 0)
-    return VNET_API_ERROR_INVALID_DST_ADDRESS;
-  
-  if (src_address->as_u32 == 0)
-    return VNET_API_ERROR_INVALID_SRC_ADDRESS;
-
-  rx_fib_index = fib_table_find_or_create_and_lock(FIB_PROTOCOL_IP4,
-                                                   rx_fib_id);
-
-  if (is_del)
-    {
-      if (rx_fib_index >= vec_len(dpm->dhcp_server_index_by_rx_fib_index))
-        return VNET_API_ERROR_NO_SUCH_ENTRY;
-      
-      server_index = dpm->dhcp_server_index_by_rx_fib_index[rx_fib_index];
-
-      if (server_index == ~0)
-        return VNET_API_ERROR_NO_SUCH_ENTRY;
-
-      /* Use the default server again.  */
-      dpm->dhcp_server_index_by_rx_fib_index[rx_fib_index] = ~0;
-      server = pool_elt_at_index (dpm->dhcp_servers, server_index);
-
-      fib_table_entry_special_remove(rx_fib_index,
-                                     &all_1s,
-                                     FIB_SOURCE_DHCP);
-      fib_table_unlock (rx_fib_index,
-                        FIB_PROTOCOL_IP4);
-      fib_table_unlock (server->server_fib_index,
-                        FIB_PROTOCOL_IP4);
-
-      memset (server, 0, sizeof (*server));
-      pool_put (dpm->dhcp_servers, server);
-      return 0;
-    }
-  else
-  {
-      vec_validate_init_empty(dpm->dhcp_server_index_by_rx_fib_index,
-                              rx_fib_index,
-                              ~0);
-
-      pool_get (dpm->dhcp_servers, server);
-
-      server->dhcp_server.as_u32 = addr->as_u32;
-      server->dhcp_src_address.as_u32 = src_address->as_u32;
-
-      fib_table_entry_special_add(rx_fib_index,
-                                  &all_1s,
-                                  FIB_SOURCE_DHCP,
-                                  FIB_ENTRY_FLAG_LOCAL,
-                                  ADJ_INDEX_INVALID);
-      fib_table_lock (rx_fib_index,
-                      FIB_PROTOCOL_IP4);
-
-      server->server_fib_index = 
-          fib_table_find_or_create_and_lock(FIB_PROTOCOL_IP4,
-                                            server_fib_id);
-
-      vec_validate_init_empty (dpm->dhcp_server_index_by_rx_fib_index,
-                               rx_fib_index,
-                               ~0);
-      dpm->dhcp_server_index_by_rx_fib_index[rx_fib_index] = 
-          server - dpm->dhcp_servers;
-  }
-
-  fib_table_unlock (rx_fib_index,
-                    FIB_PROTOCOL_IP4);
-
-  return 0;
-}
-
-static clib_error_t *
-dhcp_proxy_set_command_fn (vlib_main_t * vm,
-                           unformat_input_t * input,
-                           vlib_cli_command_t * cmd)
-{
-  ip4_address_t server_addr, src_addr;
-  u32 server_fib_id = 0, rx_fib_id = 0;
-  int is_del = 0;
-  int set_src = 0, set_server = 0;
-  
-  while (unformat_check_input(input) != UNFORMAT_END_OF_INPUT) 
-    {
-      if (unformat (input, "server %U", 
-                    unformat_ip4_address, &server_addr)) 
-        set_server = 1;
-      else if (unformat (input, "server-fib-id %d", &server_fib_id))
-        ;
-      else if (unformat (input, "rx-fib-id %d", &rx_fib_id))
-        ;
-      else if (unformat(input, "src-address %U", 
-			unformat_ip4_address, &src_addr))
-        set_src = 1;
-      else if (unformat (input, "delete") ||
-               unformat (input, "del"))
-        is_del = 1;
-      else
-        break;
-    }
-
-  if (is_del || (set_server && set_src))
-    {
-      int rv;
-
-      rv = dhcp_proxy_set_server (&server_addr, &src_addr, rx_fib_id, 
-                                  server_fib_id, is_del);
-      switch (rv)
-        {
-        case 0:
-          return 0;
-
-        case VNET_API_ERROR_INVALID_DST_ADDRESS:
-          return clib_error_return (0, "Invalid server address");
-          
-        case VNET_API_ERROR_INVALID_SRC_ADDRESS:
-          return clib_error_return (0, "Invalid src address");
-          
-        case VNET_API_ERROR_NO_SUCH_INNER_FIB:
-          return clib_error_return (0, "No such rx fib id %d", rx_fib_id);
-          
-        case VNET_API_ERROR_NO_SUCH_FIB:
-          return clib_error_return (0, "No such server fib id %d", 
-                                    server_fib_id);
-
-        case VNET_API_ERROR_NO_SUCH_ENTRY:
-          return clib_error_return 
-            (0, "Fib id %d: no per-fib DHCP server configured", rx_fib_id);
-
-        default:
-          return clib_error_return (0, "BUG: rv %d", rv);
-        }
-    }
-  else
-    return clib_error_return (0, "parse error`%U'",
-                              format_unformat_error, input);
-}
-
-VLIB_CLI_COMMAND (dhcp_proxy_set_command, static) = {
-  .path = "set dhcp proxy",
-  .short_help = "set dhcp proxy [del] server <ip-addr> src-address <ip-addr> [server-fib-id <n>] [rx-fib-id <n>]",
-  .function = dhcp_proxy_set_command_fn,
-};
-
-u8 * format_dhcp_proxy_server (u8 * s, va_list * args)
-{
-  dhcp_proxy_main_t * dm = va_arg (*args, dhcp_proxy_main_t *);
-  dhcp_server_t * server = va_arg (*args, dhcp_server_t *);
-  u32 rx_fib_index = va_arg (*args, u32);
-  ip4_fib_t * rx_fib, * server_fib;
-  u32 server_fib_id = ~0, rx_fib_id = ~0;
-
-  if (dm == 0)
-    {
-      s = format (s, "%=16s%=16s%=14s%=14s", "Server", "Src Address", 
-                  "Server FIB", "RX FIB");
-      return s;
-    }
-
-  server_fib = ip4_fib_get(server->server_fib_index);
-
-  if (server_fib)
-    server_fib_id = server_fib->table_id;
-
-  rx_fib = ip4_fib_get(rx_fib_index);
-
-  if (rx_fib)
-    rx_fib_id = rx_fib->table_id;
-
-  s = format (s, "%=16U%=16U%=14u%=14u",
-              format_ip4_address, &server->dhcp_server,
-              format_ip4_address, &server->dhcp_src_address,
-              server_fib_id, rx_fib_id);
-  return s;
-}
-
-static clib_error_t *
-dhcp_proxy_show_command_fn (vlib_main_t * vm,
-                            unformat_input_t * input,
-                            vlib_cli_command_t * cmd)
-{
-  dhcp_proxy_main_t * dpm = &dhcp_proxy_main;
-  dhcp_server_t * server;
-  u32 server_index, i;
-
-  vlib_cli_output (vm, "%U", format_dhcp_proxy_server, 0 /* header line */,
-                   0, 0);
-
-  vec_foreach_index (i, dpm->dhcp_server_index_by_rx_fib_index)
-  {
-      server_index = dpm->dhcp_server_index_by_rx_fib_index[i];
-      if (~0 == server_index)
-          continue;
-
-      server = pool_elt_at_index (dpm->dhcp_servers, server_index);
-
-      vlib_cli_output (vm, "%U", format_dhcp_proxy_server, dpm, 
-                       server, i);
-    }
-
-  return 0;
-}
-
-VLIB_CLI_COMMAND (dhcp_proxy_show_command, static) = {
-  .path = "show dhcp proxy",
-  .short_help = "Display dhcp proxy server info",
-  .function = dhcp_proxy_show_command_fn,
-};
-
-void
-dhcp_proxy_dump (void *opaque,
-                 u32 context)
-{
-  dhcp_proxy_main_t * dpm = &dhcp_proxy_main;
-  ip4_fib_t *s_fib, *r_fib;
-  dhcp_server_t * server;
-  u32 server_index, i;
-  vss_info *v;
-
-  vec_foreach_index (i, dpm->dhcp_server_index_by_rx_fib_index)
-  {
-      server_index = dpm->dhcp_server_index_by_rx_fib_index[i];
-      if (~0 == server_index)
-          continue;
-
-      server = pool_elt_at_index (dpm->dhcp_servers, server_index);
-      v = dhcp_get_vss_info(dpm, i);
-
-      ip46_address_t src_addr = {
-          .ip4 = server->dhcp_src_address,
-      };
-      ip46_address_t server_addr = {
-          .ip4 = server->dhcp_server,
-      };
-
-      s_fib = ip4_fib_get(server->server_fib_index);
-      r_fib = ip4_fib_get(i);
-
-      dhcp_send_details(opaque,
-                        context,
-                        &server_addr,
-                        &src_addr,
-                        s_fib->table_id,
-                        r_fib->table_id,
-                        (v ? v->vpn_id.fib_id : 0),
-                        (v ? v->vpn_id.oui : 0));
-  }
-}
-
-int dhcp_proxy_set_option82_vss(u32 tbl_id,
-                                u32 oui,
-                                u32 fib_id, 
-                                int is_del)
-{
-  dhcp_proxy_main_t *dm = &dhcp_proxy_main;
-  vss_info *v = NULL;
-  u32  rx_fib_index;
-  int rc = 0;
-  
-  rx_fib_index = ip4_fib_table_find_or_create_and_lock(tbl_id);
-  v = dhcp_get_vss_info(dm, rx_fib_index);
-
-  if (NULL != v)
-  {
-      if (is_del)
-      {
-          /* release the lock held on the table when the VSS
-           * info was created */
-          fib_table_unlock (rx_fib_index,
-                            FIB_PROTOCOL_IP4);
-
-          pool_put (dm->vss, v);
-          dm->vss_index_by_rx_fib_index[rx_fib_index] = ~0;
-      }
-      else
-      {
-          /* this is a modify */
-          v->vpn_id.fib_id = fib_id;
-          v->vpn_id.oui = oui;
-      }
-  }
-  else
-  {
-      if (is_del)
-          rc = VNET_API_ERROR_NO_SUCH_ENTRY;
-      else
-      {
-          /* create a new entry */
-          vec_validate_init_empty(dm->vss_index_by_rx_fib_index,
-                                  rx_fib_index, ~0);
-
-          /* hold a lock on the table whilst the VSS info exist */
-          fib_table_lock (rx_fib_index,
-                          FIB_PROTOCOL_IP4);
-
-          pool_get (dm->vss, v);
-          v->vpn_id.fib_id = fib_id;
-          v->vpn_id.oui = oui;
-          dm->vss_index_by_rx_fib_index[rx_fib_index] = v - dm->vss;
-      }
-  }
-
-  /* Release the lock taken during the create_or_lock at the start */
-  fib_table_unlock (rx_fib_index,
-                    FIB_PROTOCOL_IP4);
-  
-  return (rc);
-}
-
-static clib_error_t *
-dhcp_option_82_vss_fn (vlib_main_t * vm,
-                        unformat_input_t * input,
-                        vlib_cli_command_t * cmd)
-{
-  int is_del = 0, got_new_vpn_id=0;
-  u32 oui=0, fib_id=0, tbl_id=~0;
- 
-
-  while (unformat_check_input(input) != UNFORMAT_END_OF_INPUT) 
-    {
-
-      if (unformat(input, "delete") || unformat(input, "del"))
-          is_del = 1;    
-      else if (unformat (input, "oui %d", &oui))
-          got_new_vpn_id = 1;
-      else if (unformat (input, "vpn-id %d", &fib_id))
-          got_new_vpn_id = 1;
-      else if (unformat (input, "table %d", &tbl_id))
-          got_new_vpn_id = 1;
-      else
-          break;
-  }
-  if (tbl_id == ~0)
-      return clib_error_return (0, "no table ID specified.");
-  
-  if (is_del || got_new_vpn_id)
-    {
-      int rv;
-      rv = dhcp_proxy_set_option82_vss(tbl_id, oui, fib_id, is_del);
-      switch (rv)
-        {
-        case 0:
-            return 0;
-            
-        case VNET_API_ERROR_NO_SUCH_FIB:
-            return clib_error_return (0, "option 82 vss(oui:%d, vpn-id:%d) not found in table %d",
-                                      oui, fib_id, tbl_id);
-            
-        case VNET_API_ERROR_NO_SUCH_ENTRY:
-            return clib_error_return (0, "option 82 vss for table %d not found in in pool.",
-                                      tbl_id);
-        default:
-          return clib_error_return (0, "BUG: rv %d", rv);
-        }
-    }
-  else
-      return clib_error_return (0, "parse error`%U'",
-                                format_unformat_error, input);
-}
-
-VLIB_CLI_COMMAND (dhcp_proxy_vss_command,static) = {
-  .path = "set dhcp option-82 vss",
-  .short_help = "set dhcp option-82 vss [del] table <table id> oui <oui> vpn-id <vpn-id>",
-  .function = dhcp_option_82_vss_fn,
-};
-
-
-static clib_error_t *
-dhcp_vss_show_command_fn (vlib_main_t * vm,
-                          unformat_input_t * input,
-                          vlib_cli_command_t * cmd)
-  
-{
-  dhcp_proxy_main_t * dm = &dhcp_proxy_main;
-  ip4_fib_t *fib;
-  u32 *fib_index;
-  vss_info *v;
-  
-  vlib_cli_output (vm, "%=9s%=11s%=12s","Table", "OUI", "VPN-ID");
-  pool_foreach (fib_index, dm->vss_index_by_rx_fib_index,
-  ({
-      fib = ip4_fib_get (*fib_index);
-      v = pool_elt_at_index (dm->vss, *fib_index);
-
-      vlib_cli_output (vm, "%=6d%=6d%=12d",
-                       fib->table_id,
-                       v->vpn_id.oui,
-                       v->vpn_id.fib_id);
-  }));
-  
-  return 0;
-}
-
-VLIB_CLI_COMMAND (dhcp_proxy_vss_show_command, static) = {
-  .path = "show dhcp vss",
-  .short_help = "show dhcp VSS",
-  .function = dhcp_vss_show_command_fn,
-};
-
-static clib_error_t *
-dhcp_option_82_address_show_command_fn (vlib_main_t * vm,
-                                unformat_input_t * input,
-                                vlib_cli_command_t * cmd)
-  
-{
-  dhcp_proxy_main_t *dm = &dhcp_proxy_main;
-  vnet_main_t *vnm = vnet_get_main();                                     
-  u32 sw_if_index0=0, sw_if_index;
-  ip4_address_t *ia0;
-  vnet_sw_interface_t *swif;
-  
-  while (unformat_check_input(input) != UNFORMAT_END_OF_INPUT) 
-    {
-      
-      if (unformat(input, "%U",
-                   unformat_vnet_sw_interface, dm->vnet_main, &sw_if_index0))
-        {
-          swif = vnet_get_sw_interface (vnm, sw_if_index0);
-          sw_if_index = (swif->flags & VNET_SW_INTERFACE_FLAG_UNNUMBERED) ?
-            swif->unnumbered_sw_if_index : sw_if_index0;
-          ia0 = ip4_interface_first_address(&ip4_main, sw_if_index, 0);
-          if (ia0)
-            {
-              vlib_cli_output (vm, "%=20s%=20s", "interface", 
-                               "source IP address");
-              
-              vlib_cli_output (vm, "%=20U%=20U",
-                               format_vnet_sw_if_index_name, 
-                               dm->vnet_main, sw_if_index0,
-                               format_ip4_address, ia0);
-            }
-          else
-            vlib_cli_output (vm, "%=34s %=20U", 
-                             "No IPv4 address configured on",
-                             format_vnet_sw_if_index_name, 
-                             dm->vnet_main, sw_if_index);
-        }
-      else
-        break;
-    }
-  
-  return 0;
-}
-
-VLIB_CLI_COMMAND (dhcp_proxy_address_show_command,static) = {
-  .path = "show dhcp option-82-address interface",
-  .short_help = "show dhcp option-82-address interface <interface>",
-  .function = dhcp_option_82_address_show_command_fn,
-};
diff --git a/src/vnet/dhcpv6/packet.h b/src/vnet/dhcpv6/packet.h
deleted file mode 100644
index 8634b5d8..00000000
--- a/src/vnet/dhcpv6/packet.h
+++ /dev/null
@@ -1,183 +0,0 @@
-#ifndef included_vnet_dhcp_packet_h
-#define included_vnet_dhcp_packet_h
-
-/*
- * DHCP packet format
- *
- * Copyright (c) 2013 Cisco and/or its affiliates.
- * Licensed under the Apache License, Version 2.0 (the "License");
- * you may not use this file except in compliance with the License.
- * You may obtain a copy of the License at:
- *
- *     http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-#include <vnet/ip/ip6_packet.h>
-
-// #define DHCP_VRF_NAME_MAX_LEN L3VM_MAX_NAME_STR_LEN
-// #define DHCPV6_MAX_VRF_NAME_LEN L3VM_MAX_NAME_STR_LEN
-#define DHCP_MAX_RELAY_ADDR    16
-#define PROTO_UDP 17
-#define DHCPV6_CLIENT_PORT 546
-#define DHCPV6_SERVER_PORT 547
-#define HOP_COUNT_LIMIT  32
-#define DHCPV6_CISCO_ENT_NUM 9
-
-/*
- * DHCPv6 message types
- */
-typedef enum dhcpv6_msg_type_{
-    DHCPV6_MSG_SOLICIT              = 1,
-    DHCPV6_MSG_ADVERTISE            = 2,
-    DHCPV6_MSG_REQUEST              = 3,
-    DHCPV6_MSG_CONFIRM              = 4,
-    DHCPV6_MSG_RENEW                = 5,
-    DHCPV6_MSG_REBIND               = 6,
-    DHCPV6_MSG_REPLY                = 7,
-    DHCPV6_MSG_RELEASE              = 8,
-    DHCPV6_MSG_DECLINE              = 9,
-    DHCPV6_MSG_RECONFIGURE          = 10,
-    DHCPV6_MSG_INFORMATION_REQUEST  = 11,
-    DHCPV6_MSG_RELAY_FORW           = 12,
-    DHCPV6_MSG_RELAY_REPL           = 13,
-} dhcpv6_msg_type_t;
-
-/*
- * DHCPv6 options types
- */
-enum {
-    DHCPV6_OPTION_CLIENTID      = 1,
-    DHCPV6_OPTION_SERVERID      = 2,
-    DHCPV6_OPTION_IA_NA         = 3,
-    DHCPV6_OPTION_IA_TA         = 4,
-    DHCPV6_OPTION_IAADDR        = 5,
-    DHCPV6_OPTION_ORO           = 6,
-    DHCPV6_OPTION_PREFERENCE    = 7,
-    DHCPV6_OPTION_ELAPSED_TIME  = 8,
-    DHCPV6_OPTION_RELAY_MSG     = 9,
-    DHCPV6_OPTION_AUTH          = 11,
-    DHCPV6_OPTION_UNICAST       = 12,
-    DHCPV6_OPTION_STATUS_CODE   = 13,
-    DHCPV6_OPTION_RAPID_COMMIT  = 14,
-    DHCPV6_OPTION_USER_CLASS    = 15,
-    DHCPV6_OPTION_VENDOR_CLASS  = 16,
-    DHCPV6_OPTION_VENDOR_OPTS   = 17,
-    DHCPV6_OPTION_INTERFACE_ID  = 18, // relay agent fills this
-    DHCPV6_OPTION_RECONF_MSG    = 19,
-    DHCPV6_OPTION_RECONF_ACCEPT = 20,
-    DHCPV6_OPTION_REMOTEID      = 37, // relay agent fills this
-    DHCPV6_OPTION_VSS      = 68, // relay agent fills this
-    DHCPV6_OPTION_CLIENT_LINK_LAYER_ADDRESS = 79, 
-    DHCPV6_OPTION_MAX
-};
-
-/*
-* DHCPv6 status codes
- */
-enum {
-    DHCPV6_STATUS_SUCCESS       = 0,
-    DHCPV6_STATUS_UNSPEC_FAIL   = 1,
-    DHCPV6_STATUS_NOADDRS_AVAIL = 2,
-    DHCPV6_STATUS_NO_BINDING    = 3,
-    DHCPV6_STATUS_NOT_ONLINK    = 4,
-    DHCPV6_STATUS_USE_MULTICAST = 5,
-};
-
-/*
- * DHCPv6 DUID types
- */
-enum {
-    DHCPV6_DUID_LLT =   1,  /* DUID Based on Link-layer Address Plus Time */
-    DHCPV6_DUID_EN =    2,  /* DUID Based on Enterprise Number */
-    DHCPV6_DUID_LL =    3,  /* DUID Based on Link-layer Address */
-};
-
-//Structure for DHCPv6 payload from client
-typedef struct dhcpv6_hdr_ {
-    union {
-        u8 msg_type; //DHCP msg type
-        u32 xid;     // transaction id
-    }u;
-    u8  data[0];
-} dhcpv6_header_t;
-
-
-
-typedef CLIB_PACKED (struct dhcpv6_relay_ctx_ {
-    dhcpv6_header_t *pkt;
-    u32  pkt_len;
-    u32  dhcpv6_len; //DHCPv6 payload load
-//    if_ordinal iod;
-    u32 if_index;
-    u32 ctx_id;
-    char ctx_name[32+1];
-    u8 dhcp_msg_type;
-}) dhcpv6_relay_ctx_t;
-
-//Structure for DHCPv6 RELAY-FORWARD and DHCPv6 RELAY-REPLY pkts
-typedef CLIB_PACKED (struct dhcpv6_relay_hdr_ {
-    u8           msg_type;
-    u8           hop_count;
-    ip6_address_t    link_addr;
-    ip6_address_t    peer_addr;
-    u8           data[0];
-}) dhcpv6_relay_hdr_t;
-
-typedef enum dhcp_stats_action_type_ {
-    DHCP_STATS_ACTION_FORWARDED=1,
-    DHCP_STATS_ACTION_RECEIVED,
-    DHCP_STATS_ACTION_DROPPED
-} dhcp_stats_action_type_t;
-//Generic counters for a packet
-typedef  struct dhcp_stats_counters_ {
-    u64  rx_pkts; //counter for received pkts
-    u64  tx_pkts; //counter for forwarded pkts
-    u64  drops;   //counter for dropped pkts
-} dhcp_stats_counters_t;
-
-
-typedef enum dhcpv6_stats_drop_reason_ {
-    DHCPV6_RELAY_PKT_DROP_RELAYDISABLE = 1,
-    DHCPV6_RELAY_PKT_DROP_MAX_HOPS,
-    DHCPV6_RELAY_PKT_DROP_VALIDATION_FAIL,
-    DHCPV6_RELAY_PKT_DROP_UNKNOWN_OP_INTF,
-    DHCPV6_RELAY_PKT_DROP_BAD_CONTEXT,
-    DHCPV6_RELAY_PKT_DROP_OPT_INSERT_FAIL,
-    DHCPV6_RELAY_PKT_DROP_REPLY_FROM_CLIENT,
-} dhcpv6_stats_drop_reason_t;
-
-typedef CLIB_PACKED (struct {
-  u16 option;
-  u16 length;
-  u8 data[0];
-}) dhcpv6_option_t;
-
-typedef CLIB_PACKED (struct {
-  dhcpv6_option_t opt;
-  u32 int_idx;
-}) dhcpv6_int_id_t;
-
-typedef CLIB_PACKED (struct {
-  dhcpv6_option_t opt;
-  u8 data[8];  // data[0]:type, data[1..7]: VPN ID
-}) dhcpv6_vss_t;
-
-typedef CLIB_PACKED (struct {
-  dhcpv6_option_t opt;
-  u32 ent_num;
-  u32 rmt_id;
-}) dhcpv6_rmt_id_t;
-
-typedef CLIB_PACKED (struct {
-  dhcpv6_option_t opt;
-  u16 link_type;
-  u8 data[6];  // data[0]:data[5]: MAC address
-}) dhcpv6_client_mac_t;
-
-
-#endif /* included_vnet_dhcp_packet_h */
diff --git a/src/vnet/dhcpv6/proxy.h b/src/vnet/dhcpv6/proxy.h
deleted file mode 100644
index 77ced361..00000000
--- a/src/vnet/dhcpv6/proxy.h
+++ /dev/null
@@ -1,94 +0,0 @@
-/*
- * proxy.h: dhcp proxy 
- *
- * Copyright (c) 2013 Cisco and/or its affiliates.
- * Licensed under the Apache License, Version 2.0 (the "License");
- * you may not use this file except in compliance with the License.
- * You may obtain a copy of the License at:
- *
- *     http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-
-#ifndef included_dhcpv6_proxy_h
-#define included_dhcpv6_proxy_h
-
-#include <vnet/vnet.h>
-#include <vnet/ethernet/ethernet.h>
-#include <vnet/ip/ip.h>
-#include <vnet/ip/ip4.h>
-#include <vnet/ip/ip6_packet.h>
-#include <vnet/pg/pg.h>
-#include <vnet/ip/format.h>
-#include <vnet/ip/udp.h>
-#include <vnet/dhcpv6/packet.h>
-
-typedef enum {
-#define dhcpv6_proxy_error(n,s) DHCPV6_PROXY_ERROR_##n,
-#include <vnet/dhcpv6/proxy_error.def>
-#undef dhcpv6_proxy_error
-  DHCPV6_PROXY_N_ERROR,
-} dhcpv6_proxy_error_t;
-
-typedef struct {
-  u32 oui;
-  u32 fib_id;
-} dhcpv6_vss_id;
-
-typedef union {
-  u8 as_u8[8];
-  dhcpv6_vss_id vpn_id;
-} dhcpv6_vss_info;
-
-typedef struct {
-  ip6_address_t dhcp6_server;
-  ip6_address_t dhcp6_src_address;
-  u32 server_fib6_index;
-} dhcpv6_server_t;
-
-typedef struct {
-  /* Pool of DHCP servers */
-  dhcpv6_server_t * dhcp6_servers;
-
-  /* Pool of selected DHCP server. Zero is the default server */
-   u32 * dhcp6_server_index_by_rx_fib_index;
-
-  /* all DHCP servers address */
-  ip6_address_t all_dhcpv6_server_address;
-  ip6_address_t all_dhcpv6_server_relay_agent_address;
-
-  /* to drop pkts in server-to-client direction */
-  u32 error_drop_node_index;
-
-  dhcpv6_vss_info *vss;
-
-  /* hash lookup specific vrf_id -> VSS vector index*/
-  u32 *vss_index_by_rx_fib_index;
-   
-  /* convenience */
-  vlib_main_t * vlib_main;
-  vnet_main_t * vnet_main;
-} dhcpv6_proxy_main_t;
-
-dhcpv6_proxy_main_t dhcpv6_proxy_main;
-
-int dhcpv6_proxy_set_vss(u32 tbl_id,
-                         u32 oui,
-                         u32 fib_id, 
-                         int is_del);
-
-int dhcpv6_proxy_set_server(ip6_address_t *addr,
-                            ip6_address_t *src_address,
-                            u32 rx_fib_id,
-                            u32 server_fib_id,
-                            int is_del);
-
-void dhcpv6_proxy_dump(void *opaque,
-                       u32 context);
-
-#endif /* included_dhcpv6_proxy_h */
diff --git a/src/vnet/dhcpv6/proxy_error.def b/src/vnet/dhcpv6/proxy_error.def
deleted file mode 100644
index 55fa7317..00000000
--- a/src/vnet/dhcpv6/proxy_error.def
+++ /dev/null
@@ -1,29 +0,0 @@
-/*
- * dhcp_proxy_error.def: dhcp proxy errors
- *
- * Copyright (c) 2013 Cisco and/or its affiliates.
- * Licensed under the Apache License, Version 2.0 (the "License");
- * you may not use this file except in compliance with the License.
- * You may obtain a copy of the License at:
- *
- *     http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-
-dhcpv6_proxy_error (NONE, "no error")
-dhcpv6_proxy_error (NO_SERVER, "no dhcpv6 server configured")
-dhcpv6_proxy_error (RELAY_TO_SERVER, "DHCPV6 packets relayed to the server")
-dhcpv6_proxy_error (RELAY_TO_CLIENT, "DHCPV6 packets relayed to clients")
-dhcpv6_proxy_error (NO_INTERFACE_ADDRESS, "DHCPV6 no interface address")
-dhcpv6_proxy_error (WRONG_MESSAGE_TYPE, "DHCPV6 wrong message type.")
-dhcpv6_proxy_error (NO_SRC_ADDRESS, "DHCPV6 no srouce IPv6 address configured.")
-dhcpv6_proxy_error (NO_CIRCUIT_ID_OPTION, "DHCPv6 reply packets without circuit ID option")
-dhcpv6_proxy_error (NO_RELAY_MESSAGE_OPTION, "DHCPv6 reply packets without relay message option")
-dhcpv6_proxy_error (BAD_SVR_FIB_OR_ADDRESS, "DHCPv6 packets not from DHCPv6 server or server FIB.")
-dhcpv6_proxy_error (PKT_TOO_BIG, "DHCPv6 packets which are too big.")
-dhcpv6_proxy_error (WRONG_INTERFACE_ID_OPTION, "DHCPv6 reply to invalid interface.")
diff --git a/src/vnet/dhcpv6/proxy_node.c b/src/vnet/dhcpv6/proxy_node.c
deleted file mode 100644
index f40798e6..00000000
--- a/src/vnet/dhcpv6/proxy_node.c
+++ /dev/null
@@ -1,1280 +0,0 @@
-/*
- * proxy_node.c: dhcpv6 proxy node processing
- *
- * Copyright (c) 2013 Cisco and/or its affiliates.
- * Licensed under the Apache License, Version 2.0 (the "License");
- * you may not use this file except in compliance with the License.
- * You may obtain a copy of the License at:
- *
- *     http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-
-#include <vlib/vlib.h>
-#include <vnet/pg/pg.h>
-#include <vnet/dhcpv6/proxy.h>
-#include <vnet/dhcp/proxy.h>
-#include <vnet/fib/ip6_fib.h>
-#include <vnet/mfib/mfib_table.h>
-#include <vnet/mfib/ip6_mfib.h>
-
-static char * dhcpv6_proxy_error_strings[] = {
-#define dhcpv6_proxy_error(n,s) s,
-#include "proxy_error.def"
-#undef dhcpv6_proxy_error
-};
-
-#define foreach_dhcpv6_proxy_to_server_input_next \
-  _ (DROP, "error-drop")			\
-  _ (LOOKUP, "ip6-lookup")                      \
-  _ (SEND_TO_CLIENT, "dhcpv6-proxy-to-client")
-
-
-typedef enum {
-#define _(s,n) DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_##s,
-  foreach_dhcpv6_proxy_to_server_input_next
-#undef _
-  DHCPV6_PROXY_TO_SERVER_INPUT_N_NEXT,
-} dhcpv6_proxy_to_server_input_next_t;
-
-typedef struct {
-  /* 0 => to server, 1 => to client */
-  int which;
-  u8 packet_data[64];
-  u32 error;
-  u32 sw_if_index;
-  u32 original_sw_if_index;
-} dhcpv6_proxy_trace_t;
-
-vlib_node_registration_t dhcpv6_proxy_to_server_node;
-vlib_node_registration_t dhcpv6_proxy_to_client_node;
-
-
-u8 * format_dhcpv6_proxy_trace (u8 * s, va_list * args)
-{
-  CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *);
-  CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *);
-  dhcpv6_proxy_trace_t * t = va_arg (*args, dhcpv6_proxy_trace_t *);
-
-  if (t->which == 0)
-    s = format (s, "DHCPV6 proxy: sent to server %U",
-                format_ip6_address, &t->packet_data, sizeof (ip6_address_t));
-  else
-    s = format (s, "DHCPV6 proxy: sent to client from %U",
-                format_ip6_address, &t->packet_data, sizeof (ip6_address_t));
-  if (t->error != (u32)~0)
-    s = format (s, " error: %s\n", dhcpv6_proxy_error_strings[t->error]);
-
-  s = format (s, "  original_sw_if_index: %d, sw_if_index: %d\n",
-              t->original_sw_if_index, t->sw_if_index);
-
-  return s;
-}
-
-u8 * format_dhcpv6_proxy_header_with_length (u8 * s, va_list * args)
-{
-  dhcpv6_header_t * h = va_arg (*args, dhcpv6_header_t *);
-  u32 max_header_bytes = va_arg (*args, u32);
-  u32 header_bytes;
-
-  header_bytes = sizeof (h[0]);
-  if (max_header_bytes != 0 && header_bytes > max_header_bytes)
-    return format (s, "dhcpv6 header truncated");
-
-  s = format (s, "DHCPV6 Proxy");
-
-  return s;
-}
-/* get first interface address */
-static ip6_address_t *
-ip6_interface_first_global_or_site_address (ip6_main_t * im, u32 sw_if_index)
-{
-  ip_lookup_main_t * lm = &im->lookup_main;
-  ip_interface_address_t * ia = 0;
-  ip6_address_t * result = 0;
-
-  foreach_ip_interface_address (lm, ia, sw_if_index,
-                                1 /* honor unnumbered */,
-  ({
-    ip6_address_t * a = ip_interface_address_get_address (lm, ia);
-    if ((a->as_u8[0] & 0xe0) == 0x20 ||
-        (a->as_u8[0] & 0xfe) == 0xfc)  {
-        result = a;
-        break;
-    }
-  }));
-  return result;
-}
-
-static inline void copy_ip6_address (ip6_address_t *dst, ip6_address_t *src) 
-{
-
-  dst->as_u64[0] = src->as_u64[0];
-  dst->as_u64[1] = src->as_u64[1];
-} 
-
-static inline dhcpv6_vss_info *
-dhcpv6_get_vss_info (dhcpv6_proxy_main_t *dm,
-                     u32 rx_fib_index)
-{
-  dhcpv6_vss_info *v;
-
-  if (vec_len(dm->vss_index_by_rx_fib_index) <= rx_fib_index ||
-      dm->vss_index_by_rx_fib_index[rx_fib_index] == ~0)
-  {
-      v = NULL;
-  }
-  else
-  {
-      v = pool_elt_at_index (dm->vss,
-                             dm->vss_index_by_rx_fib_index[rx_fib_index]);
-  }
-
-  return (v);
-}
-
-static inline dhcpv6_server_t *
-dhcpv6_get_server (dhcpv6_proxy_main_t *dm,
-                   u32 rx_fib_index)
-{
-  dhcpv6_server_t *s = NULL;
-
-  if (vec_len(dm->dhcp6_server_index_by_rx_fib_index) > rx_fib_index &&
-      dm->dhcp6_server_index_by_rx_fib_index[rx_fib_index] != ~0)
-  {
-      s = pool_elt_at_index (dm->dhcp6_servers,
-                             dm->dhcp6_server_index_by_rx_fib_index[rx_fib_index]);
-  }
-
-  return (s);
-}
-
-static uword
-dhcpv6_proxy_to_server_input (vlib_main_t * vm,
-                            vlib_node_runtime_t * node,
-                            vlib_frame_t * from_frame)
-{
-  u32 n_left_from, next_index, * from, * to_next;
-  dhcpv6_proxy_main_t * dpm = &dhcpv6_proxy_main;
-  from = vlib_frame_vector_args (from_frame);
-  n_left_from = from_frame->n_vectors;
-  u32 pkts_to_server=0, pkts_to_client=0, pkts_no_server=0;
-  u32 pkts_no_interface_address=0, pkts_no_exceeding_max_hop=0;
-  u32 pkts_no_src_address=0;
-  u32 pkts_wrong_msg_type=0;
-  u32 pkts_too_big=0;
-  ip6_main_t * im = &ip6_main;
-  ip6_address_t * src;
-  int bogus_length;
-  dhcpv6_server_t * server;
-  u32  rx_fib_idx = 0, server_fib_idx = 0;
-
-  next_index = node->cached_next_index;
-
-  while (n_left_from > 0)
-    {
-      u32 n_left_to_next;
-
-      vlib_get_next_frame (vm, node, next_index,
-			   to_next, n_left_to_next);
-
-      while (n_left_from > 0 && n_left_to_next > 0)
-	{
-          vnet_main_t *vnm = vnet_get_main();
-          u32 sw_if_index = 0;
-          u32 rx_sw_if_index = 0;
-          vnet_sw_interface_t *swif;
- 	  u32 bi0;
-	  vlib_buffer_t * b0;
-          udp_header_t * u0, *u1;
-	  dhcpv6_header_t * h0;  // client msg hdr
-          ip6_header_t * ip0, *ip1;
-          ip6_address_t _ia0, *ia0=&_ia0;
-	  u32 next0;
-          u32 error0 = (u32) ~0;
-          dhcpv6_option_t *fwd_opt;
-          dhcpv6_relay_hdr_t *r1;
-          u16 len;
-          dhcpv6_int_id_t *id1;
-          dhcpv6_vss_t *vss1;
-          dhcpv6_client_mac_t *cmac; // client mac
-          ethernet_header_t * e_h0;
-          u8 client_src_mac[6];
-          vlib_buffer_free_list_t *fl;
-          dhcpv6_vss_info *vss;
-
-	  bi0 = from[0];
-	  to_next[0] = bi0;
-	  from += 1;
-	  to_next += 1;
-	  n_left_from -= 1;
-	  n_left_to_next -= 1;
-
-	  b0 = vlib_get_buffer (vm, bi0);
-
-          h0 = vlib_buffer_get_current (b0);
-
-          /*
-           * udp_local hands us the DHCPV6 header.
-           */
-          u0 = (void *)h0 -(sizeof(*u0));
-          ip0 = (void *)u0 -(sizeof(*ip0));
-          e_h0 = (void *)ip0 - ethernet_buffer_header_size(b0);
-
-          clib_memcpy(client_src_mac, e_h0->src_address, 6);
-
-          switch (h0->u.msg_type) {
-            case DHCPV6_MSG_SOLICIT:
-            case DHCPV6_MSG_REQUEST:
-            case DHCPV6_MSG_CONFIRM:
-            case DHCPV6_MSG_RENEW:
-            case DHCPV6_MSG_REBIND:
-            case DHCPV6_MSG_RELEASE:
-            case DHCPV6_MSG_DECLINE:
-            case DHCPV6_MSG_INFORMATION_REQUEST:
-            case DHCPV6_MSG_RELAY_FORW:
-                /* send to server */
-                break;
-            case DHCPV6_MSG_RELAY_REPL:
-                /* send to client */
-                next0 = DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_SEND_TO_CLIENT;
-                error0 = 0;
-                pkts_to_client++;
-                goto do_enqueue;
-            default:
-                /* drop the packet */
-                pkts_wrong_msg_type++;
-                error0 = DHCPV6_PROXY_ERROR_WRONG_MESSAGE_TYPE;
-                next0 = DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_DROP;
-                goto do_trace;
-
-          }
-
-          /* Send to DHCPV6 server via the configured FIB */
-          rx_sw_if_index = sw_if_index =  vnet_buffer(b0)->sw_if_index[VLIB_RX];
-          rx_fib_idx = im->fib_index_by_sw_if_index [rx_sw_if_index];
-          server = dhcpv6_get_server(dpm, rx_fib_idx);
-
-          if (PREDICT_FALSE (NULL == server))
-          {
-              error0 = DHCPV6_PROXY_ERROR_NO_SERVER;
-              next0 = DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_DROP;
-              pkts_no_server++;
-              goto do_trace;
-          }
-
-          server_fib_idx = server->server_fib6_index;
-          vnet_buffer(b0)->sw_if_index[VLIB_TX] = server_fib_idx;
-
-
-          /* relay-option header pointer */
-          vlib_buffer_advance(b0, -(sizeof(*fwd_opt)));
-          fwd_opt = vlib_buffer_get_current(b0);
-          /* relay message header pointer */
-          vlib_buffer_advance(b0, -(sizeof(*r1)));
-          r1 = vlib_buffer_get_current(b0);
-
-          vlib_buffer_advance(b0, -(sizeof(*u1)));
-          u1 = vlib_buffer_get_current(b0);
-
-          vlib_buffer_advance(b0, -(sizeof(*ip1)));
-          ip1 = vlib_buffer_get_current(b0);
-
-          /* fill in all that rubbish... */
-          len = clib_net_to_host_u16(u0->length) - sizeof(udp_header_t);
-          copy_ip6_address(&r1->peer_addr, &ip0->src_address);
-
-          r1->msg_type = DHCPV6_MSG_RELAY_FORW;
-          fwd_opt->length = clib_host_to_net_u16(len);
-          fwd_opt->option = clib_host_to_net_u16(DHCPV6_OPTION_RELAY_MSG);
-
-          r1->hop_count++;
-          r1->hop_count = (h0->u.msg_type != DHCPV6_MSG_RELAY_FORW) ? 0 : r1->hop_count;
-
-          if (PREDICT_FALSE(r1->hop_count >= HOP_COUNT_LIMIT))
-            {
-              error0 =  DHCPV6_RELAY_PKT_DROP_MAX_HOPS;
-              next0 = DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_DROP;
-              pkts_no_exceeding_max_hop++;
-              goto do_trace;
-            }
-
-
-          /* If relay-fwd and src address is site or global unicast address  */
-          if (h0->u.msg_type == DHCPV6_MSG_RELAY_FORW &&
-              ((ip0->src_address.as_u8[0] & 0xe0) == 0x20 ||
-               (ip0->src_address.as_u8[0] & 0xfe) == 0xfc))
-            {
-              /* Set link address to zero */
-              r1->link_addr.as_u64[0] = 0;
-              r1->link_addr.as_u64[1] = 0;
-              goto link_address_set;
-            }
-
-          /* if receiving interface is unnumbered, use receiving interface
-           * IP address as link address, otherwise use the loopback interface
-           * IP address as link address.
-           */
-
-          swif = vnet_get_sw_interface (vnm, rx_sw_if_index);
-          if (swif->flags & VNET_SW_INTERFACE_FLAG_UNNUMBERED)
-              sw_if_index = swif->unnumbered_sw_if_index;
-
-          ia0 = ip6_interface_first_global_or_site_address(&ip6_main, sw_if_index);
-          if (ia0 == 0)
-            {
-              error0 = DHCPV6_PROXY_ERROR_NO_INTERFACE_ADDRESS;
-              next0 = DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_DROP;
-              pkts_no_interface_address++;
-              goto do_trace;
-            }
-
-          copy_ip6_address(&r1->link_addr, ia0);
-
-        link_address_set:
-          fl = vlib_buffer_get_free_list (vm, b0->free_list_index);
-
-          if ((b0->current_length+sizeof(*id1)+sizeof(*vss1)+sizeof(*cmac))
-              > fl->n_data_bytes)
-            {
-              error0 = DHCPV6_PROXY_ERROR_PKT_TOO_BIG;
-              next0 = DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_DROP;
-              pkts_too_big++;
-              goto do_trace;
-            }
-
-          id1 = (dhcpv6_int_id_t *) (((uword) ip1) + b0->current_length);
-          b0->current_length += (sizeof (*id1));
-
-          id1->opt.option = clib_host_to_net_u16(DHCPV6_OPTION_INTERFACE_ID);
-          id1->opt.length = clib_host_to_net_u16(sizeof(rx_sw_if_index));
-          id1->int_idx = clib_host_to_net_u32(rx_sw_if_index);
-
-          u1->length =0;
-          if (h0->u.msg_type != DHCPV6_MSG_RELAY_FORW)
-            {
-               cmac = (dhcpv6_client_mac_t *) (((uword) ip1) + b0->current_length);
-               b0->current_length += (sizeof (*cmac));
-               cmac->opt.length =clib_host_to_net_u16(sizeof(*cmac) -
-                                                      sizeof(cmac->opt));
-               cmac->opt.option = clib_host_to_net_u16(DHCPV6_OPTION_CLIENT_LINK_LAYER_ADDRESS);
-               cmac->link_type = clib_host_to_net_u16(1); // ethernet
-               clib_memcpy(cmac->data, client_src_mac, 6);
-               u1->length += sizeof(*cmac);
-            }
-
-          //TODO: Revisit if hash makes sense here
-          vss = dhcpv6_get_vss_info(dpm, rx_fib_idx);
-
-          if (NULL != vss) {
-              vss1 = (dhcpv6_vss_t *) (((uword) ip1) + b0->current_length);
-              b0->current_length += (sizeof (*vss1));
-              vss1->opt.length =clib_host_to_net_u16(sizeof(*vss1) -
-						     sizeof(vss1->opt));
-              vss1->opt.option = clib_host_to_net_u16(DHCPV6_OPTION_VSS);
-              vss1->data[0] = 1;   // type
-              vss1->data[1] = vss->vpn_id.oui >>16 & 0xff;
-              vss1->data[2] = vss->vpn_id.oui >>8  & 0xff;
-              vss1->data[3] = vss->vpn_id.oui & 0xff;
-              vss1->data[4] = vss->vpn_id.fib_id >> 24 & 0xff;
-              vss1->data[5] = vss->vpn_id.fib_id >> 16 & 0xff;
-              vss1->data[6] = vss->vpn_id.fib_id >> 8 & 0xff;
-              vss1->data[7] = vss->vpn_id.fib_id & 0xff;
-              u1->length += sizeof(*vss1);
-          }
-
-          pkts_to_server++;
-          u1->checksum = 0;
-          u1->src_port = clib_host_to_net_u16(UDP_DST_PORT_dhcpv6_to_client);
-          u1->dst_port = clib_host_to_net_u16(UDP_DST_PORT_dhcpv6_to_server);
-
-          u1->length =
-              clib_host_to_net_u16( clib_net_to_host_u16(fwd_opt->length) +
-                                    sizeof(*r1) + sizeof(*fwd_opt) +
-                                    sizeof(*u1) + sizeof(*id1) + u1->length);
-
-          memset(ip1, 0, sizeof(*ip1));
-          ip1->ip_version_traffic_class_and_flow_label = 0x60;
-          ip1->payload_length =  u1->length;
-          ip1->protocol = PROTO_UDP;
-          ip1->hop_limit = HOP_COUNT_LIMIT;
-	      src = (server->dhcp6_server.as_u64[0] || server->dhcp6_server.as_u64[1]) ?
-	        &server->dhcp6_server : &dpm->all_dhcpv6_server_address;
-          copy_ip6_address(&ip1->dst_address, src);
-
-
-          ia0 = ip6_interface_first_global_or_site_address
-              (&ip6_main, vnet_buffer(b0)->sw_if_index[VLIB_RX]);
-
-	      src = (server->dhcp6_src_address.as_u64[0] || server->dhcp6_src_address.as_u64[1]) ?
-	        &server->dhcp6_src_address : ia0;
-          if (ia0 == 0)
-            {
-              error0 = DHCPV6_PROXY_ERROR_NO_SRC_ADDRESS;
-              next0 = DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_DROP;
-              pkts_no_src_address++;
-              goto do_trace;
-            }
-
-	  copy_ip6_address (&ip1->src_address, src);
-
-
-          u1->checksum = ip6_tcp_udp_icmp_compute_checksum(vm, b0, ip1,
-                                                           &bogus_length);
-          ASSERT(bogus_length == 0);
-
-          next0 = DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_LOOKUP;
-
-        do_trace:
-          if (PREDICT_FALSE(b0->flags & VLIB_BUFFER_IS_TRACED))
-            {
-               dhcpv6_proxy_trace_t *tr = vlib_add_trace (vm, node,
-                                                          b0, sizeof (*tr));
-               tr->which = 0; /* to server */
-               tr->error = error0;
-               tr->original_sw_if_index = rx_sw_if_index;
-               tr->sw_if_index = sw_if_index;
-               if (DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_LOOKUP == next0)
-                 copy_ip6_address((ip6_address_t *)&tr->packet_data[0], &server->dhcp6_server);
-            }
-
-        do_enqueue:
-	  vlib_validate_buffer_enqueue_x1 (vm, node, next_index,
-					   to_next, n_left_to_next,
-					   bi0, next0);
-	}
-
-      vlib_put_next_frame (vm, node, next_index, n_left_to_next);
-    }
-
-  vlib_node_increment_counter (vm, dhcpv6_proxy_to_server_node.index,
-                               DHCPV6_PROXY_ERROR_RELAY_TO_CLIENT,
-                               pkts_to_client);
-  vlib_node_increment_counter (vm, dhcpv6_proxy_to_server_node.index,
-                               DHCPV6_PROXY_ERROR_RELAY_TO_SERVER,
-                               pkts_to_server);
-  vlib_node_increment_counter (vm, dhcpv6_proxy_to_server_node.index,
-                               DHCPV6_PROXY_ERROR_NO_INTERFACE_ADDRESS,
-                               pkts_no_interface_address);
-  vlib_node_increment_counter (vm, dhcpv6_proxy_to_server_node.index,
-                               DHCPV6_PROXY_ERROR_WRONG_MESSAGE_TYPE,
-                               pkts_wrong_msg_type);
-  vlib_node_increment_counter (vm, dhcpv6_proxy_to_server_node.index,
-                               DHCPV6_PROXY_ERROR_NO_SRC_ADDRESS,
-                               pkts_no_src_address);
-  vlib_node_increment_counter (vm, dhcpv6_proxy_to_server_node.index,
-                               DHCPV6_PROXY_ERROR_PKT_TOO_BIG,
-                               pkts_too_big);
-  return from_frame->n_vectors;
-}
-
-VLIB_REGISTER_NODE (dhcpv6_proxy_to_server_node) = {
-  .function = dhcpv6_proxy_to_server_input,
-  .name = "dhcpv6-proxy-to-server",
-  /* Takes a vector of packets. */
-  .vector_size = sizeof (u32),
-
-  .n_errors = DHCPV6_PROXY_N_ERROR,
-  .error_strings = dhcpv6_proxy_error_strings,
-
-  .n_next_nodes = DHCPV6_PROXY_TO_SERVER_INPUT_N_NEXT,
-  .next_nodes = {
-#define _(s,n) [DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_##s] = n,
-    foreach_dhcpv6_proxy_to_server_input_next
-#undef _
-  },
-
-  .format_buffer = format_dhcpv6_proxy_header_with_length,
-  .format_trace = format_dhcpv6_proxy_trace,
-#if 0
-  .unformat_buffer = unformat_dhcpv6_proxy_header,
-#endif
-};
-
-static uword
-dhcpv6_proxy_to_client_input (vlib_main_t * vm,
-                            vlib_node_runtime_t * node,
-                            vlib_frame_t * from_frame)
-{
-
-  u32 n_left_from, * from;
-  ethernet_main_t *em = ethernet_get_main (vm);
-  dhcpv6_proxy_main_t * dm = &dhcpv6_proxy_main;
-  dhcpv6_server_t * server;
-  vnet_main_t * vnm = vnet_get_main();
-  int bogus_length;
-
-  from = vlib_frame_vector_args (from_frame);
-  n_left_from = from_frame->n_vectors;
-
-  while (n_left_from > 0)
-    {
-      u32 bi0;
-      vlib_buffer_t * b0;
-      udp_header_t * u0, *u1=0;
-      dhcpv6_relay_hdr_t * h0;
-      ip6_header_t * ip1 = 0, *ip0;
-      ip6_address_t _ia0, * ia0 = &_ia0;
-      ip6_address_t client_address;
-      ethernet_interface_t *ei0;
-      ethernet_header_t *mac0;
-      vnet_hw_interface_t *hi0;
-      vlib_frame_t *f0;
-      u32 * to_next0;
-      u32 sw_if_index = ~0;
-      u32 original_sw_if_index = ~0;
-      vnet_sw_interface_t *si0;
-      u32 error0 = (u32)~0;
-      vnet_sw_interface_t *swif;
-      dhcpv6_option_t *r0 = 0, *o;
-      u16 len = 0;
-      u8 interface_opt_flag = 0;
-      u8 relay_msg_opt_flag = 0;
-      ip6_main_t * im = &ip6_main;
-      u32 server_fib_idx, client_fib_idx;
-
-      bi0 = from[0];
-      from += 1;
-      n_left_from -= 1;
-
-      b0 = vlib_get_buffer (vm, bi0);
-      h0 = vlib_buffer_get_current (b0);
-
-      if (DHCPV6_MSG_RELAY_REPL != h0->msg_type)
-        {
-          error0 =  DHCPV6_PROXY_ERROR_WRONG_MESSAGE_TYPE;
-
-        drop_packet:
-          vlib_node_increment_counter (vm, dhcpv6_proxy_to_client_node.index,
-                                       error0, 1);
-
-          f0 = vlib_get_frame_to_node (vm, dm->error_drop_node_index);
-          to_next0 = vlib_frame_vector_args (f0);
-          to_next0[0] = bi0;
-          f0->n_vectors = 1;
-          vlib_put_frame_to_node (vm, dm->error_drop_node_index, f0);
-          goto do_trace;
-        }
-      /* hop count seems not need to be checked */
-      if (HOP_COUNT_LIMIT < h0->hop_count)
-        {
-          error0 =  DHCPV6_RELAY_PKT_DROP_MAX_HOPS;
-          goto drop_packet;
-        }
-      u0 = (void *)h0 -(sizeof(*u0));
-      ip0 = (void *)u0 -(sizeof(*ip0));
-
-      vlib_buffer_advance (b0, sizeof(*h0));
-      o = vlib_buffer_get_current (b0);
-
-      /* Parse through TLVs looking for option 18 (DHCPV6_OPTION_INTERFACE_ID)
-         _and_ option 9 (DHCPV6_OPTION_RELAY_MSG) option which must be there.
-         Currently assuming no other options need to be processed
-         The interface-ID is the FIB number we need
-         to track down the client-facing interface */
-
-      while ((u8 *) o < (b0->data + b0->current_data + b0->current_length))
-        {
-           if (DHCPV6_OPTION_INTERFACE_ID == clib_net_to_host_u16(o->option))
-             {
-                interface_opt_flag = 1;
-                if (clib_net_to_host_u16(o->length) == sizeof(sw_if_index))
-                    sw_if_index = clib_net_to_host_u32(((dhcpv6_int_id_t*)o)->int_idx);
-                if (sw_if_index >= vec_len (im->fib_index_by_sw_if_index))
-                  {
-                    error0 = DHCPV6_PROXY_ERROR_WRONG_INTERFACE_ID_OPTION;
-                    goto drop_packet;
-                  }
-             }
-           if (DHCPV6_OPTION_RELAY_MSG == clib_net_to_host_u16(o->option))
-             {
-                relay_msg_opt_flag = 1;
-                r0 = vlib_buffer_get_current (b0);
-             }
-           if ((relay_msg_opt_flag == 1) && (interface_opt_flag == 1))
-             break;
-           vlib_buffer_advance (b0, sizeof(*o) + clib_net_to_host_u16(o->length));
-           o = (dhcpv6_option_t *) (((uword) o) + clib_net_to_host_u16(o->length) + sizeof(*o));
-        }
-
-      if ((relay_msg_opt_flag == 0) || (r0 == 0))
-        {
-    	  error0 = DHCPV6_PROXY_ERROR_NO_RELAY_MESSAGE_OPTION;
-    	  goto drop_packet;
-        }
-
-      if ((u32)~0 == sw_if_index)
-        {
-          error0 = DHCPV6_PROXY_ERROR_NO_CIRCUIT_ID_OPTION;
-          goto drop_packet;
-        }
-
-      //Advance buffer to start of encapsulated DHCPv6 message
-      vlib_buffer_advance (b0, sizeof(*r0));
-
-      client_fib_idx = im->fib_index_by_sw_if_index[sw_if_index];
-      server = dhcpv6_get_server(dm, client_fib_idx);
-
-      if (NULL == server)
-      {
-    	  error0 = DHCPV6_PROXY_ERROR_NO_SERVER;
-          goto drop_packet;
-      }
-
-      server_fib_idx = im->fib_index_by_sw_if_index
-          [vnet_buffer(b0)->sw_if_index[VLIB_RX]];
-
-      if (server_fib_idx != server->server_fib6_index ||
-          ip0->src_address.as_u64[0] != server->dhcp6_server.as_u64[0] ||
-          ip0->src_address.as_u64[1] != server->dhcp6_server.as_u64[1])
-        {
-          //drop packet if not from server with configured address or FIB
-          error0 = DHCPV6_PROXY_ERROR_BAD_SVR_FIB_OR_ADDRESS;
-          goto drop_packet;
-        }
-
-      vnet_buffer (b0)->sw_if_index[VLIB_TX] = original_sw_if_index
-          = sw_if_index;
-
-      swif = vnet_get_sw_interface (vnm, original_sw_if_index);
-      if (swif->flags & VNET_SW_INTERFACE_FLAG_UNNUMBERED)
-          sw_if_index = swif->unnumbered_sw_if_index;
-
-
-      /*
-       * udp_local hands us the DHCPV6 header, need udp hdr,
-       * ip hdr to relay to client
-       */
-      vlib_buffer_advance (b0, -(sizeof(*u1)));
-      u1 = vlib_buffer_get_current (b0);
-
-      vlib_buffer_advance (b0, -(sizeof(*ip1)));
-      ip1 = vlib_buffer_get_current (b0);
-
-      copy_ip6_address(&client_address, &h0->peer_addr);
-
-      ia0 = ip6_interface_first_address (&ip6_main, sw_if_index);
-      if (ia0 == 0)
-        {
-          error0 = DHCPV6_PROXY_ERROR_NO_INTERFACE_ADDRESS;
-          goto drop_packet;
-        }
-
-      len =  clib_net_to_host_u16(r0->length);
-      memset(ip1, 0, sizeof(*ip1));
-      copy_ip6_address(&ip1->dst_address, &client_address);
-      u1->checksum = 0;
-      u1->src_port = clib_net_to_host_u16 (UDP_DST_PORT_dhcpv6_to_server);
-      u1->dst_port = clib_net_to_host_u16 (UDP_DST_PORT_dhcpv6_to_client);
-      u1->length = clib_host_to_net_u16 (len + sizeof(udp_header_t));
-
-      ip1->ip_version_traffic_class_and_flow_label =
-          ip0->ip_version_traffic_class_and_flow_label &
-          0x00000fff;
-      ip1->payload_length =  u1->length;
-      ip1->protocol = PROTO_UDP;
-      ip1->hop_limit = HOP_COUNT_LIMIT;
-      copy_ip6_address(&ip1->src_address, ia0);
-
-      u1->checksum = ip6_tcp_udp_icmp_compute_checksum(vm, b0, ip1,
-                                                       &bogus_length);
-      ASSERT(bogus_length == 0);
-
-      vlib_buffer_advance (b0, -(sizeof(ethernet_header_t)));
-      si0 = vnet_get_sw_interface (vnm, original_sw_if_index);
-      if (si0->type == VNET_SW_INTERFACE_TYPE_SUB)
-	  vlib_buffer_advance (b0, -4 /* space for VLAN tag */);
-
-      mac0 = vlib_buffer_get_current (b0);
-
-      hi0 = vnet_get_sup_hw_interface (vnm, original_sw_if_index);
-      ei0 = pool_elt_at_index (em->interfaces, hi0->hw_instance);
-      clib_memcpy (mac0->src_address, ei0->address, sizeof (ei0->address));
-      memset (&mac0->dst_address, 0xff, sizeof (mac0->dst_address));
-      mac0->type = (si0->type == VNET_SW_INTERFACE_TYPE_SUB) ?
-	clib_net_to_host_u16(0x8100) : clib_net_to_host_u16 (0x86dd);
-
-      if (si0->type == VNET_SW_INTERFACE_TYPE_SUB)
-	{
-	  u32 * vlan_tag = (u32 *)(mac0+1);
-	  u32 tmp;
-	  tmp = (si0->sub.id << 16) | 0x0800;
-	  *vlan_tag = clib_host_to_net_u32 (tmp);
-	}
-
-      /* $$$ consider adding a dynamic next to the graph node, for performance */
-      f0 = vlib_get_frame_to_node (vm, hi0->output_node_index);
-      to_next0 = vlib_frame_vector_args (f0);
-      to_next0[0] = bi0;
-      f0->n_vectors = 1;
-      vlib_put_frame_to_node (vm, hi0->output_node_index, f0);
-
-    do_trace:
-      if (PREDICT_FALSE(b0->flags & VLIB_BUFFER_IS_TRACED))
-        {
-          dhcpv6_proxy_trace_t *tr = vlib_add_trace (vm, node,
-                                                     b0, sizeof (*tr));
-          tr->which = 1; /* to client */
-          if (ia0)
-              copy_ip6_address((ip6_address_t*)tr->packet_data, ia0);
-          tr->error = error0;
-          tr->original_sw_if_index = original_sw_if_index;
-          tr->sw_if_index = sw_if_index;
-        }
-    }
-  return from_frame->n_vectors;
-
-}
-
-VLIB_REGISTER_NODE (dhcpv6_proxy_to_client_node) = {
-  .function = dhcpv6_proxy_to_client_input,
-  .name = "dhcpv6-proxy-to-client",
-  /* Takes a vector of packets. */
-  .vector_size = sizeof (u32),
-
-  .n_errors = DHCPV6_PROXY_N_ERROR,
-  .error_strings = dhcpv6_proxy_error_strings,
-  .format_buffer = format_dhcpv6_proxy_header_with_length,
-  .format_trace = format_dhcpv6_proxy_trace,
-#if 0
-  .unformat_buffer = unformat_dhcpv6_proxy_header,
-#endif
-};
-
-clib_error_t * dhcpv6_proxy_init (vlib_main_t * vm)
-{
-  dhcpv6_proxy_main_t * dm = &dhcpv6_proxy_main;
-  vlib_node_t * error_drop_node;
-  dhcpv6_server_t * server;
-
-  dm->vlib_main = vm;
-  dm->vnet_main = vnet_get_main();
-  error_drop_node = vlib_get_node_by_name (vm, (u8 *) "error-drop");
-  dm->error_drop_node_index = error_drop_node->index;
-
-  dm->vss_index_by_rx_fib_index = NULL;
-
-  /* RFC says this is the dhcpv6 server address  */
-  dm->all_dhcpv6_server_address.as_u64[0] = clib_host_to_net_u64 (0xFF05000000000000);
-  dm->all_dhcpv6_server_address.as_u64[1] = clib_host_to_net_u64 (0x00010003);
-
-  /* RFC says this is the server and agent address */
-  dm->all_dhcpv6_server_relay_agent_address.as_u64[0] = clib_host_to_net_u64 (0xFF02000000000000);
-  dm->all_dhcpv6_server_relay_agent_address.as_u64[1] = clib_host_to_net_u64 (0x00010002);
-
-  udp_register_dst_port (vm, UDP_DST_PORT_dhcpv6_to_client,
-                         dhcpv6_proxy_to_client_node.index, 0 /* is_ip6 */);
-
-  udp_register_dst_port (vm, UDP_DST_PORT_dhcpv6_to_server,
-                         dhcpv6_proxy_to_server_node.index, 0 /* is_ip6 */);
-
-  /* Create the default server, don't mark it valid */
-  pool_get (dm->dhcp6_servers, server);
-  memset (server, 0, sizeof (*server));
-
-  return 0;
-}
-
-VLIB_INIT_FUNCTION (dhcpv6_proxy_init);
-
-int dhcpv6_proxy_set_server (ip6_address_t *addr,
-                             ip6_address_t *src_address,
-                             u32 rx_fib_id,
-                             u32 server_fib_id,
-                             int is_del)
-{
-  dhcpv6_proxy_main_t * dm = &dhcpv6_proxy_main;
-  dhcpv6_server_t * server = 0;
-  u32 rx_fib_index = 0;
-  int rc = 0;
-
-  rx_fib_index = ip6_mfib_table_find_or_create_and_lock(rx_fib_id);
-
-  const mfib_prefix_t all_dhcp_servers = {
-      .fp_len = 128,
-      .fp_proto = FIB_PROTOCOL_IP6,
-      .fp_grp_addr = {
-          .ip6 = dm->all_dhcpv6_server_relay_agent_address,
-      }
-  };
-
-  if (is_del)
-    {
-      server = dhcpv6_get_server(dm, rx_fib_index);
-
-      if (NULL == server)
-        {
-          rc = VNET_API_ERROR_NO_SUCH_ENTRY;
-          goto out;
-        }
-
-      /*
-       * release the locks held on the server fib and rx mfib
-       */
-      mfib_table_entry_delete(rx_fib_index,
-                              &all_dhcp_servers,
-                              MFIB_SOURCE_DHCP);
-      mfib_table_unlock(rx_fib_index, FIB_PROTOCOL_IP6);
-      fib_table_unlock(server->server_fib6_index, FIB_PROTOCOL_IP6);
-
-      dm->dhcp6_server_index_by_rx_fib_index[rx_fib_index] = ~0;
-
-      memset (server, 0, sizeof (*server));
-      pool_put (dm->dhcp6_servers, server);
-    }
-  else
-    {
-      if (addr->as_u64[0] == 0 &&
-          addr->as_u64[1] == 0 )
-      {
-          rc = VNET_API_ERROR_INVALID_DST_ADDRESS;
-          goto out;
-      }
-      if (src_address->as_u64[0] == 0 &&
-          src_address->as_u64[1] == 0)
-      {
-          rc = VNET_API_ERROR_INVALID_SRC_ADDRESS;
-          goto out;
-      }
-
-      server = dhcpv6_get_server(dm, rx_fib_index);
-
-      if (NULL != server)
-        {
-          /* modify of an existing entry */
-          ip6_fib_t *fib;
-
-          fib = ip6_fib_get(server->server_fib6_index);
-
-          if (fib->table_id != server_fib_id)
-            {
-              /* swap tables */
-              fib_table_unlock(server->server_fib6_index, FIB_PROTOCOL_IP6);
-              server->server_fib6_index =
-                  ip6_fib_table_find_or_create_and_lock(server_fib_id);
-            }
-        }
-      else
-        {
-          /* Allocate a new server */
-          pool_get (dm->dhcp6_servers, server);
-
-          vec_validate_init_empty (dm->dhcp6_server_index_by_rx_fib_index,
-                                   rx_fib_index, ~0);
-          dm->dhcp6_server_index_by_rx_fib_index[rx_fib_index] =
-              server - dm->dhcp6_servers;
-
-          server->server_fib6_index =
-              ip6_fib_table_find_or_create_and_lock(server_fib_id);
-          mfib_table_lock(rx_fib_index, FIB_PROTOCOL_IP6);
-
-          const mfib_prefix_t all_dhcp_servers = {
-              .fp_len = 128,
-              .fp_proto = FIB_PROTOCOL_IP6,
-              .fp_grp_addr = {
-                  .ip6 = dm->all_dhcpv6_server_relay_agent_address,
-              }
-            };
-          const fib_route_path_t path_for_us = {
-              .frp_proto = FIB_PROTOCOL_IP6,
-              .frp_addr = zero_addr,
-              .frp_sw_if_index = 0xffffffff,
-              .frp_fib_index = ~0,
-              .frp_weight = 0,
-              .frp_flags = FIB_ROUTE_PATH_LOCAL,
-          };
-          mfib_table_entry_path_update(rx_fib_index,
-                                       &all_dhcp_servers,
-                                       MFIB_SOURCE_DHCP,
-                                       &path_for_us,
-                                       MFIB_ITF_FLAG_FORWARD);
-          /*
-           * Each interface that is enabled in this table, needs to be added
-           * as an accepting interface, but this is not easily doable in VPP.
-           * So we cheat. Add a flag to the entry that indicates accept form
-           * any interface.
-           * We will still only accept on v6 enabled interfaces, since the
-           * input feature ensures this.
-           */
-          mfib_table_entry_update(rx_fib_index,
-                                  &all_dhcp_servers,
-                                  MFIB_SOURCE_DHCP,
-                                  MFIB_ENTRY_FLAG_ACCEPT_ALL_ITF);
-        }
-      copy_ip6_address(&server->dhcp6_server, addr);
-      copy_ip6_address(&server->dhcp6_src_address, src_address);
-  }
-
-out:
-  mfib_table_unlock(rx_fib_index, FIB_PROTOCOL_IP6);
-
-  return (rc);
-}
-
-static clib_error_t *
-dhcpv6_proxy_set_command_fn (vlib_main_t * vm,
-                           unformat_input_t * input,
-                           vlib_cli_command_t * cmd)
-{
-  ip6_address_t addr, src_addr;
-  int set_server = 0, set_src_address = 0;
-  u32 rx_fib_id = 0, server_fib_id = 0;
-  int is_del = 0;
-
-  while (unformat_check_input(input) != UNFORMAT_END_OF_INPUT)
-    {
-      if (unformat (input, "server %U",
-                    unformat_ip6_address, &addr))
-	 set_server = 1;
-      else if (unformat(input, "src-address %U",
-			unformat_ip6_address, &src_addr))
-	  set_src_address =1;
-       else if (unformat (input, "server-fib-id %d", &server_fib_id))
-        ;
-       else if (unformat (input, "rx-fib-id %d", &rx_fib_id))
-         ;
-       else if (unformat (input, "delete") ||
-                unformat (input, "del"))
-           is_del = 1;
-      else
-        break;
-    }
-
-  if (is_del || (set_server && set_src_address))
-  {
-      int rv;
-
-      rv = dhcpv6_proxy_set_server (&addr, &src_addr, rx_fib_id,
-                                    server_fib_id, is_del);
-
-      //TODO: Complete the errors
-      switch (rv)
-        {
-        case 0:
-          return 0;
-
-        case -1:
-          return clib_error_return (0, "FIB id %d does not exist", server_fib_id);
-
-        default:
-          return clib_error_return (0, "BUG: rv %d", rv);
-        }
-  }
-  else
-    return clib_error_return (0, "parse error`%U'",
-                              format_unformat_error, input);
-}
-
-VLIB_CLI_COMMAND (dhcpv6_proxy_set_command, static) = {
-  .path = "set dhcpv6 proxy",
-  .short_help = "set dhcpv6 proxy [del] server <ipv6-addr> src-address <ipv6-addr> "
-		  "[server-fib-id <fib-id>] [rx-fib-id <fib-id>] ",
-  .function = dhcpv6_proxy_set_command_fn,
-};
-
-u8 * format_dhcpv6_proxy_server (u8 * s, va_list * args)
-{
-  dhcpv6_proxy_main_t * dm = va_arg (*args, dhcpv6_proxy_main_t *);
-  dhcpv6_server_t * server = va_arg (*args, dhcpv6_server_t *);
-  u32 rx_fib_index = va_arg (*args, u32);
-  ip6_fib_t * rx_fib, * server_fib;
-  u32 server_fib_id = (u32)~0, rx_fib_id = ~0;
-
-  if (dm == 0)
-    {
-      s = format (s, "%=40s%=40s%=14s%=14s", "Server Address", "Source Address",
-                  "Server FIB", "RX FIB");
-      return s;
-    }
-
-  server_fib = ip6_fib_get(server->server_fib6_index);
-  if (server_fib)
-	  server_fib_id= server_fib->table_id;
-
-  rx_fib= ip6_fib_get(rx_fib_index);
-
-  if (rx_fib)
-	  rx_fib_id = rx_fib->table_id;
-
-  s = format (s, "%=40U%=40U%=14u%=14u",
-              format_ip6_address, &server->dhcp6_server,
-              format_ip6_address, &server->dhcp6_src_address,
-			  server_fib_id, rx_fib_id);
-  return s;
-}
-
-static clib_error_t *
-dhcpv6_proxy_show_command_fn (vlib_main_t * vm,
-                            unformat_input_t * input,
-                            vlib_cli_command_t * cmd)
-{
-  dhcpv6_proxy_main_t * dpm = &dhcpv6_proxy_main;
-  int i;
-  u32 server_index;
-  dhcpv6_server_t * server;
-
-  vlib_cli_output (vm, "%U", format_dhcpv6_proxy_server, 0 /* header line */,
-		  0, 0);
-  vec_foreach_index (i, dpm->dhcp6_server_index_by_rx_fib_index)
-  {
-      server_index = dpm->dhcp6_server_index_by_rx_fib_index[i];
-      if (~0 == server_index)
-          continue;
-
-      server = pool_elt_at_index (dpm->dhcp6_servers, server_index);
-
-      vlib_cli_output (vm, "%U", format_dhcpv6_proxy_server, dpm, 
-                       server, i);
-    }
-
-  return 0;
-}
-
-VLIB_CLI_COMMAND (dhcpv6_proxy_show_command, static) = {
-  .path = "show dhcpv6 proxy",
-  .short_help = "Display dhcpv6 proxy info",
-  .function = dhcpv6_proxy_show_command_fn,
-};
-
-void
-dhcpv6_proxy_dump (void *opaque,
-                   u32 context)
-{
-  dhcpv6_proxy_main_t * dpm = &dhcpv6_proxy_main;
-  ip6_fib_t *s_fib, *r_fib;
-  dhcpv6_server_t * server;
-  u32 server_index, i;
-  dhcpv6_vss_info *v;
-
-  vec_foreach_index (i, dpm->dhcp6_server_index_by_rx_fib_index)
-  {
-      server_index = dpm->dhcp6_server_index_by_rx_fib_index[i];
-      if (~0 == server_index)
-          continue;
-
-      server = pool_elt_at_index (dpm->dhcp6_servers, server_index);
-      v = dhcpv6_get_vss_info(dpm, i);
-
-      ip46_address_t src_addr = {
-          .ip6 = server->dhcp6_src_address,
-      };
-      ip46_address_t server_addr = {
-          .ip6 = server->dhcp6_server,
-      };
-
-      s_fib = ip6_fib_get(server->server_fib6_index);
-      r_fib = ip6_fib_get(i);
-
-      dhcp_send_details(opaque,
-                        context,
-                        &server_addr,
-                        &src_addr,
-                        s_fib->table_id,
-                        r_fib->table_id,
-                        (v ? v->vpn_id.fib_id : 0),
-                        (v ? v->vpn_id.oui : 0));
-  }
-}
-
-int dhcpv6_proxy_set_vss(u32 tbl_id,
-                         u32 oui,
-                         u32 fib_id,
-                         int is_del)
-{
-  dhcpv6_proxy_main_t *dm = &dhcpv6_proxy_main;
-  dhcpv6_vss_info *v = NULL;
-  u32  rx_fib_index;
-  int rc = 0;
-
-  rx_fib_index = ip6_fib_table_find_or_create_and_lock(tbl_id);
-  v = dhcpv6_get_vss_info(dm, rx_fib_index);
-
-  if (NULL != v)
-  {
-      if (is_del)
-      {
-          /* release the lock held on the table when the VSS
-           * info was created */
-          fib_table_unlock (rx_fib_index,
-                            FIB_PROTOCOL_IP6);
-
-          pool_put (dm->vss, v);
-          dm->vss_index_by_rx_fib_index[rx_fib_index] = ~0;
-      }
-      else
-      {
-          /* this is a modify */
-          v->vpn_id.fib_id = fib_id;
-          v->vpn_id.oui = oui;
-      }
-  }
-  else
-  {
-      if (is_del)
-          rc = VNET_API_ERROR_NO_SUCH_ENTRY;
-      else
-      {
-          /* create a new entry */
-          vec_validate_init_empty(dm->vss_index_by_rx_fib_index,
-                                  rx_fib_index, ~0);
-
-          /* hold a lock on the table whilst the VSS info exist */
-          fib_table_lock (rx_fib_index,
-                          FIB_PROTOCOL_IP6);
-
-          pool_get (dm->vss, v);
-          v->vpn_id.fib_id = fib_id;
-          v->vpn_id.oui = oui;
-          dm->vss_index_by_rx_fib_index[rx_fib_index] = v - dm->vss;
-      }
-  }
-
-  /* Release the lock taken during the create_or_lock at the start */
-  fib_table_unlock (rx_fib_index,
-                    FIB_PROTOCOL_IP6);
-
-  return (rc);
-}
-
-
-static clib_error_t *
-dhcpv6_vss_command_fn (vlib_main_t * vm,
-                           unformat_input_t * input,
-                           vlib_cli_command_t * cmd)
-{
-  int is_del = 0, got_new_vss=0;
-  u32 oui=0;
-  u32 fib_id=0, tbl_id=~0;
-
-  while (unformat_check_input(input) != UNFORMAT_END_OF_INPUT)
-    {
-      if (unformat (input, "oui %d", &oui))
-          got_new_vss = 1;
-      else if (unformat (input, "vpn-id %d", &fib_id))
-          got_new_vss = 1;
-      else if (unformat (input, "table %d", &tbl_id))
-          got_new_vss = 1;
-      else if (unformat(input, "delete") || unformat(input, "del"))
-          is_del = 1;
-      else
-          break;
-    }
-
-  if (tbl_id ==~0)
-      return clib_error_return (0, "no table ID specified.");
-
-  if (is_del || got_new_vss)
-    {
-      int rv;
-
-      rv = dhcpv6_proxy_set_vss(tbl_id, oui, fib_id, is_del);
-      switch (rv)
-        {
-        case 0:
-          return 0;
-
-        case VNET_API_ERROR_NO_SUCH_FIB:
-            return clib_error_return (0, "vss info (oui:%d, vpn-id:%d)  not found in table %d.",
-                                      oui, fib_id, tbl_id);
-
-        case VNET_API_ERROR_NO_SUCH_ENTRY:
-            return clib_error_return (0, "vss for table %d not found in pool.",
-                                      tbl_id);
-
-        default:
-          return clib_error_return (0, "BUG: rv %d", rv);
-        }
-    }
-  else
-      return clib_error_return (0, "parse error`%U'",
-                                format_unformat_error, input);
-
-}
-
-VLIB_CLI_COMMAND (dhcpv6_proxy_vss_command, static) = {
-  .path = "set dhcpv6 vss",
-  .short_help = "set dhcpv6 vss table <table-id> oui <oui> vpn-idx <vpn-idx>",
-  .function = dhcpv6_vss_command_fn,
-};
-
-static clib_error_t *
-dhcpv6_vss_show_command_fn (vlib_main_t * vm,
-                            unformat_input_t * input,
-                            vlib_cli_command_t * cmd)
-
-{
-  dhcpv6_proxy_main_t * dm = &dhcpv6_proxy_main;
-  dhcpv6_vss_info *v;
-  ip6_fib_t *fib;
-  u32 *fib_index;
-
-  vlib_cli_output (vm, "%=6s%=6s%=12s","Table", "OUI", "VPN ID");
-  pool_foreach (fib_index, dm->vss_index_by_rx_fib_index,
-  ({
-      fib = ip6_fib_get (*fib_index);
-      v = pool_elt_at_index (dm->vss, *fib_index);
-
-      vlib_cli_output (vm, "%=6d%=6d%=12d",
-                       fib->table_id,
-                       v->vpn_id.oui,
-                       v->vpn_id.fib_id);
-  }));
-
-  return 0;
-}
-
-VLIB_CLI_COMMAND (dhcpv6_proxy_vss_show_command, static) = {
-  .path = "show dhcpv6 vss",
-  .short_help = "show dhcpv6 VSS",
-  .function = dhcpv6_vss_show_command_fn,
-};
-
-static clib_error_t *
-dhcpv6_link_address_show_command_fn (vlib_main_t * vm,
-                                unformat_input_t * input,
-                                vlib_cli_command_t * cmd)
-
-{
-  dhcpv6_proxy_main_t *dm = &dhcpv6_proxy_main;
-  vnet_main_t *vnm = vnet_get_main();
-  u32 sw_if_index0=0, sw_if_index;
-  ip6_address_t *ia0;
-  vnet_sw_interface_t *swif;
-
-  while (unformat_check_input(input) != UNFORMAT_END_OF_INPUT)
-    {
-
-      if (unformat(input, "%U",
-                   unformat_vnet_sw_interface, dm->vnet_main, &sw_if_index0))
-        {
-            swif = vnet_get_sw_interface (vnm, sw_if_index0);
-            sw_if_index = (swif->flags & VNET_SW_INTERFACE_FLAG_UNNUMBERED) ?
-                swif->unnumbered_sw_if_index : sw_if_index0;
-            ia0 = ip6_interface_first_address(&ip6_main, sw_if_index);
-            if (ia0)
-              {
-                  vlib_cli_output (vm, "%=20s%=48s", "interface", "link-address");
-
-                  vlib_cli_output (vm, "%=20U%=48U",
-                                   format_vnet_sw_if_index_name, dm->vnet_main, sw_if_index0,
-                                   format_ip6_address, ia0);
-              } else
-                vlib_cli_output (vm, "%=34s%=20U", "No IPv6 address configured on",
-                                 format_vnet_sw_if_index_name, dm->vnet_main, sw_if_index);
-        } else
-          break;
-    }
-
-  return 0;
-}
-
-VLIB_CLI_COMMAND (dhcpv6_proxy_address_show_command, static) = {
-  .path = "show dhcpv6 link-address interface",
-  .short_help = "show dhcpv6 link-address interface <interface>",
-  .function = dhcpv6_link_address_show_command_fn,
-};
diff --git a/src/vpp/api/custom_dump.c b/src/vpp/api/custom_dump.c
index 70b4e4c9..4cc6aa73 100644
--- a/src/vpp/api/custom_dump.c
+++ b/src/vpp/api/custom_dump.c
@@ -21,8 +21,7 @@
 #include <vnet/ip/ip.h>
 #include <vnet/unix/tuntap.h>
 #include <vnet/mpls/mpls.h>
-#include <vnet/dhcp/proxy.h>
-#include <vnet/dhcpv6/proxy.h>
+#include <vnet/dhcp/dhcp_proxy.h>
 #include <vnet/l2tp/l2tp.h>
 #include <vnet/l2/l2_input.h>
 #include <vnet/sr/sr_packet.h>
diff --git a/test/test_dhcp.py b/test/test_dhcp.py
index fbfb8a0c..6299975b 100644
--- a/test/test_dhcp.py
+++ b/test/test_dhcp.py
@@ -293,7 +293,7 @@ class TestDHCP(VppTestCase):
         #
         # Inject a response from the server
         #  dropped, because there is no IP addrees on the
-        #  clinet interfce to fill in the option.
+        #  client interfce to fill in the option.
         #
         p = (Ether(dst=self.pg0.local_mac, src=self.pg0.remote_mac) /
              IP(src=self.pg0.remote_ip4, dst=self.pg0.local_ip4) /
-- 
cgit 1.2.3-korg


From 4008ac998f43265451281cb6e759cd6184e50bed Mon Sep 17 00:00:00 2001
From: Neale Ranns <nranns@cisco.com>
Date: Mon, 13 Feb 2017 23:20:04 -0800
Subject: Changing the IP table for an interface is an error if the interface
 already has an address configured (VPP-601)

Change-Id: I311fc264f73dd3b2b3ce9d7d1c33cd0515b36c4a
Signed-off-by: Neale Ranns <nranns@cisco.com>
---
 src/vnet/adj/adj.c                  | 22 ------------
 src/vnet/api_errno.h                |  3 +-
 src/vnet/interface_api.c            | 32 +++++++++++++++++
 src/vnet/ip/ip4_forward.c           | 71 ++++++++++++++++++++++---------------
 src/vnet/ip/ip6_forward.c           | 39 ++++++++++++++------
 src/vnet/ip/ip6_neighbor.c          |  4 +++
 src/vnet/ip/lookup.h                |  5 ---
 test/test_dhcp.py                   |  7 ++++
 test/test_gre.py                    |  4 +++
 test/test_ip4_vrf_multi_instance.py |  1 +
 test/test_ip6_vrf_multi_instance.py |  1 +
 test/test_mpls.py                   |  9 ++---
 test/test_neighbor.py               |  7 ++++
 13 files changed, 135 insertions(+), 70 deletions(-)

(limited to 'test/test_dhcp.py')

diff --git a/src/vnet/adj/adj.c b/src/vnet/adj/adj.c
index a99f173f..11f1680d 100644
--- a/src/vnet/adj/adj.c
+++ b/src/vnet/adj/adj.c
@@ -441,25 +441,3 @@ VLIB_CLI_COMMAND (adj_show_command, static) = {
     .short_help = "show adj [<adj_index>] [interface]",
     .function = adj_show,
 };
-
-/* 
- * DEPRECATED: DO NOT USE
- */
-ip_adjacency_t *
-ip_add_adjacency (ip_lookup_main_t * lm,
-		  ip_adjacency_t * copy_adj,
-		  u32 n_adj,
-		  u32 * adj_index_return)
-{
-  ip_adjacency_t * adj;
-
-  ASSERT(1==n_adj);
-
-  adj = adj_alloc(FIB_PROTOCOL_IP4);
-
-  if (copy_adj)
-      *adj = *copy_adj;
-
-  *adj_index_return = adj_get_index(adj);
-  return adj;
-}
diff --git a/src/vnet/api_errno.h b/src/vnet/api_errno.h
index a5bcb377..5e65ac7b 100644
--- a/src/vnet/api_errno.h
+++ b/src/vnet/api_errno.h
@@ -102,7 +102,8 @@ _(URI_FIFO_CREATE_FAILED, -109, "URI FIFO segment create failed")       \
 _(LISP_RLOC_LOCAL, -110, "RLOC address is local")                       \
 _(BFD_EAGAIN, -111, "BFD object cannot be manipulated at this time")	\
 _(INVALID_GPE_MODE, -112, "Invalid GPE mode")                           \
-_(LISP_GPE_ENTRIES_PRESENT, -113, "LISP GPE entries are present")
+_(LISP_GPE_ENTRIES_PRESENT, -113, "LISP GPE entries are present")       \
+_(ADDRESS_FOUND_FOR_INTERFACE, -114, "Address found for interface")
 
 typedef enum
 {
diff --git a/src/vnet/interface_api.c b/src/vnet/interface_api.c
index 60cd6d40..f94928b6 100644
--- a/src/vnet/interface_api.c
+++ b/src/vnet/interface_api.c
@@ -25,6 +25,7 @@
 #include <vnet/ethernet/ethernet.h>
 #include <vnet/ip/ip.h>
 #include <vnet/fib/fib_table.h>
+#include <vnet/mfib/mfib_table.h>
 #include <vnet/l2/l2_vtr.h>
 #include <vnet/vnet_msg_enum.h>
 #include <vnet/fib/fib_api.h>
@@ -318,6 +319,7 @@ vl_api_sw_interface_set_table_t_handler (vl_api_sw_interface_set_table_t * mp)
   u32 table_id = ntohl (mp->vrf_id);
   u32 sw_if_index = ntohl (mp->sw_if_index);
   vl_api_sw_interface_set_table_reply_t *rmp;
+  CLIB_UNUSED (ip_interface_address_t * ia);
   u32 fib_index;
 
   VALIDATE_SW_IF_INDEX (mp);
@@ -326,21 +328,51 @@ vl_api_sw_interface_set_table_t_handler (vl_api_sw_interface_set_table_t * mp)
 
   if (mp->is_ipv6)
     {
+      /* *INDENT-OFF* */
+      foreach_ip_interface_address (&ip6_main.lookup_main,
+				    ia, sw_if_index,
+				    1 /* honor unnumbered */ ,
+      ({
+        rv = VNET_API_ERROR_ADDRESS_FOUND_FOR_INTERFACE;
+        goto done;
+      }));
+      /* *INDENT-ON* */
+
       fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6,
 						     table_id);
 
       vec_validate (ip6_main.fib_index_by_sw_if_index, sw_if_index);
       ip6_main.fib_index_by_sw_if_index[sw_if_index] = fib_index;
+      fib_index = mfib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6,
+						      table_id);
+      vec_validate (ip6_main.mfib_index_by_sw_if_index, sw_if_index);
+      ip6_main.mfib_index_by_sw_if_index[sw_if_index] = fib_index;
     }
   else
     {
+      /* *INDENT-OFF* */
+      foreach_ip_interface_address (&ip4_main.lookup_main,
+				    ia, sw_if_index,
+				    1 /* honor unnumbered */ ,
+      ({
+        rv = VNET_API_ERROR_ADDRESS_FOUND_FOR_INTERFACE;
+        goto done;
+      }));
+      /* *INDENT-ON* */
 
       fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4,
 						     table_id);
 
       vec_validate (ip4_main.fib_index_by_sw_if_index, sw_if_index);
       ip4_main.fib_index_by_sw_if_index[sw_if_index] = fib_index;
+
+      fib_index = mfib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4,
+						      table_id);
+      vec_validate (ip4_main.mfib_index_by_sw_if_index, sw_if_index);
+      ip4_main.mfib_index_by_sw_if_index[sw_if_index] = fib_index;
     }
+
+done:
   stats_dsunlock ();
 
   BAD_SW_IF_INDEX_LABEL;
diff --git a/src/vnet/ip/ip4_forward.c b/src/vnet/ip/ip4_forward.c
index 66d91ab6..fe4d6767 100644
--- a/src/vnet/ip/ip4_forward.c
+++ b/src/vnet/ip/ip4_forward.c
@@ -2781,6 +2781,7 @@ add_del_interface_table (vlib_main_t * vm,
 			 unformat_input_t * input, vlib_cli_command_t * cmd)
 {
   vnet_main_t *vnm = vnet_get_main ();
+  ip_interface_address_t *ia;
   clib_error_t *error = 0;
   u32 sw_if_index, table_id;
 
@@ -2802,29 +2803,44 @@ add_del_interface_table (vlib_main_t * vm,
       goto done;
     }
 
-  {
-    ip4_main_t *im = &ip4_main;
-    u32 fib_index;
-
-    fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4,
-						   table_id);
-
-    //
-    // FIXME-LATER
-    //  changing an interface's table has consequences for any connecteds
-    //  and adj-fibs already installed.
-    //
-    vec_validate (im->fib_index_by_sw_if_index, sw_if_index);
-    im->fib_index_by_sw_if_index[sw_if_index] = fib_index;
-
-    fib_index = mfib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4,
-                                                   table_id);
-    vec_validate (im->mfib_index_by_sw_if_index, sw_if_index);
-    im->mfib_index_by_sw_if_index[sw_if_index] = fib_index;
-  }
+  /*
+   * If the interface already has in IP address, then a change int
+   * VRF is not allowed. The IP address applied must first be removed.
+   * We do not do that automatically here, since VPP has no knowledge
+   * of whether thoses subnets are valid in the destination VRF.
+   */
+  /* *INDENT-OFF* */
+  foreach_ip_interface_address (&ip4_main.lookup_main,
+                                ia, sw_if_index,
+                                1 /* honor unnumbered */,
+  ({
+      ip4_address_t * a;
+
+      a = ip_interface_address_get_address (&ip4_main.lookup_main, ia);
+      error = clib_error_return (0, "interface %U has address %U",
+                                 format_vnet_sw_if_index_name, vnm,
+                                 sw_if_index,
+                                 format_ip4_address, a);
+      goto done;
+   }));
+   /* *INDENT-ON* */
+
+{
+  ip4_main_t *im = &ip4_main;
+  u32 fib_index;
+
+  fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, table_id);
+
+  vec_validate (im->fib_index_by_sw_if_index, sw_if_index);
+  im->fib_index_by_sw_if_index[sw_if_index] = fib_index;
+
+  fib_index = mfib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, table_id);
+  vec_validate (im->mfib_index_by_sw_if_index, sw_if_index);
+  im->mfib_index_by_sw_if_index[sw_if_index] = fib_index;
+}
 
 done:
-  return error;
+return error;
 }
 
 /*?
@@ -2835,13 +2851,12 @@ done:
  * an IP Address is assigned to an interface in the table (which adds a route
  * automatically).
  *
- * @note IP addresses added after setting the interface IP table end up in
- * the indicated FIB table. If the IP address is added prior to adding the
- * interface to the FIB table, it will NOT be part of the FIB table. Predictable
- * but potentially counter-intuitive results occur if you provision interface
- * addresses in multiple FIBs. Upon RX, packets will be processed in the last
- * IP table ID provisioned. It might be marginally useful to evade source RPF
- * drops to put an interface address into multiple FIBs.
+ * @note IP addresses added after setting the interface IP table are added to
+ * the indicated FIB table. If an IP address is added prior to changing the
+ * table then this is an error. The control plane must remove these addresses
+ * first and then change the table. VPP will not automatically move the
+ * addresses from the old to the new table as it does not know the validity
+ * of such a change.
  *
  * @cliexpar
  * Example of how to add an interface to an IPv4 FIB table (where 2 is the table-id):
diff --git a/src/vnet/ip/ip6_forward.c b/src/vnet/ip/ip6_forward.c
index 5dd22b99..6f77c6dd 100644
--- a/src/vnet/ip/ip6_forward.c
+++ b/src/vnet/ip/ip6_forward.c
@@ -415,9 +415,6 @@ ip6_sw_interface_enable_disable (u32 sw_if_index, u32 is_enable)
 	return;
     }
 
-  if (sw_if_index != 0)
-    ip6_mfib_interface_enable_disable (sw_if_index, is_enable);
-
   vnet_feature_enable_disable ("ip6-unicast", "ip6-lookup", sw_if_index,
 			       is_enable, 0, 0);
 
@@ -2972,6 +2969,7 @@ add_del_ip6_interface_table (vlib_main_t * vm,
 			     vlib_cli_command_t * cmd)
 {
   vnet_main_t *vnm = vnet_get_main ();
+  ip_interface_address_t *ia;
   clib_error_t *error = 0;
   u32 sw_if_index, table_id;
 
@@ -2993,6 +2991,28 @@ add_del_ip6_interface_table (vlib_main_t * vm,
       goto done;
     }
 
+  /*
+   * If the interface already has in IP address, then a change int
+   * VRF is not allowed. The IP address applied must first be removed.
+   * We do not do that automatically here, since VPP has no knowledge
+   * of whether thoses subnets are valid in the destination VRF.
+   */
+  /* *INDENT-OFF* */
+  foreach_ip_interface_address (&ip6_main.lookup_main,
+                                ia, sw_if_index,
+                                1 /* honor unnumbered */,
+  ({
+      ip4_address_t * a;
+
+      a = ip_interface_address_get_address (&ip6_main.lookup_main, ia);
+      error = clib_error_return (0, "interface %U has address %U",
+                                 format_vnet_sw_if_index_name, vnm,
+                                 sw_if_index,
+                                 format_ip6_address, a);
+      goto done;
+  }));
+  /* *INDENT-ON* */
+
   {
     u32 fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6,
 						       table_id);
@@ -3020,13 +3040,12 @@ done:
  * an IP Address is assigned to an interface in the table (which adds a route
  * automatically).
  *
- * @note IP addresses added after setting the interface IP table end up in
- * the indicated FIB table. If the IP address is added prior to adding the
- * interface to the FIB table, it will NOT be part of the FIB table. Predictable
- * but potentially counter-intuitive results occur if you provision interface
- * addresses in multiple FIBs. Upon RX, packets will be processed in the last
- * IP table ID provisioned. It might be marginally useful to evade source RPF
- * drops to put an interface address into multiple FIBs.
+ * @note IP addresses added after setting the interface IP table are added to
+ * the indicated FIB table. If an IP address is added prior to changing the
+ * table then this is an error. The control plane must remove these addresses
+ * first and then change the table. VPP will not automatically move the
+ * addresses from the old to the new table as it does not know the validity
+ * of such a change.
  *
  * @cliexpar
  * Example of how to add an interface to an IPv6 FIB table (where 2 is the table-id):
diff --git a/src/vnet/ip/ip6_neighbor.c b/src/vnet/ip/ip6_neighbor.c
index 6b53137f..91ff224c 100644
--- a/src/vnet/ip/ip6_neighbor.c
+++ b/src/vnet/ip/ip6_neighbor.c
@@ -24,6 +24,7 @@
 #include <vnet/adj/adj_mcast.h>
 #include <vnet/fib/fib_table.h>
 #include <vnet/fib/ip6_fib.h>
+#include <vnet/mfib/ip6_mfib.h>
 
 /**
  * @file
@@ -3283,6 +3284,7 @@ disable_ip6_interface (vlib_main_t * vm, u32 sw_if_index)
 
 	  ip6_neighbor_sw_interface_add_del (vnm, sw_if_index,
 					     0 /* is_add */ );
+	  ip6_mfib_interface_enable_disable (sw_if_index, 0);
 	}
     }
   return error;
@@ -3369,6 +3371,8 @@ enable_ip6_interface (vlib_main_t * vm, u32 sw_if_index)
 		      link_local_address.as_u8[8] &= 0xfd;
 		    }
 
+		  ip6_mfib_interface_enable_disable (sw_if_index, 1);
+
 		  /* essentially "enables" ipv6 on this interface */
 		  error = ip6_add_del_interface_address (vm, sw_if_index,
 							 &link_local_address,
diff --git a/src/vnet/ip/lookup.h b/src/vnet/ip/lookup.h
index 27c70943..48360b5b 100644
--- a/src/vnet/ip/lookup.h
+++ b/src/vnet/ip/lookup.h
@@ -389,11 +389,6 @@ do {								\
   CLIB_PREFETCH (_adj, sizeof (_adj[0]), type);			\
 } while (0)
 
-/* Create new block of given number of contiguous adjacencies. */
-ip_adjacency_t *ip_add_adjacency (ip_lookup_main_t * lm,
-				  ip_adjacency_t * adj,
-				  u32 n_adj, u32 * adj_index_result);
-
 clib_error_t *ip_interface_address_add_del (ip_lookup_main_t * lm,
 					    u32 sw_if_index,
 					    void *address,
diff --git a/test/test_dhcp.py b/test/test_dhcp.py
index 6299975b..a09c9bdb 100644
--- a/test/test_dhcp.py
+++ b/test/test_dhcp.py
@@ -58,6 +58,13 @@ class TestDHCP(VppTestCase):
             i.set_table_ip6(table_id)
             table_id += 1
 
+    def tearDown(self):
+        super(TestDHCP, self).tearDown()
+        for i in self.pg_interfaces:
+            i.unconfig_ip4()
+            i.unconfig_ip6()
+            i.admin_down()
+
     def send_and_assert_no_replies(self, intf, pkts, remark):
         intf.add_stream(pkts)
         self.pg_enable_capture(self.pg_interfaces)
diff --git a/test/test_gre.py b/test/test_gre.py
index 89f39e4e..f2a5e0b0 100644
--- a/test/test_gre.py
+++ b/test/test_gre.py
@@ -39,6 +39,10 @@ class TestGRE(VppTestCase):
 
     def tearDown(self):
         super(TestGRE, self).tearDown()
+        for i in self.pg_interfaces:
+            i.unconfig_ip4()
+            i.unconfig_ip6()
+            i.admin_down()
 
     def create_stream_ip4(self, src_if, src_ip, dst_ip):
         pkts = []
diff --git a/test/test_ip4_vrf_multi_instance.py b/test/test_ip4_vrf_multi_instance.py
index ddf8f593..b73ac948 100644
--- a/test/test_ip4_vrf_multi_instance.py
+++ b/test/test_ip4_vrf_multi_instance.py
@@ -208,6 +208,7 @@ class TestIp4VrfMultiInst(VppTestCase):
             self.vrf_deleted_list.append(vrf_id)
         for j in range(self.pg_ifs_per_vrf):
             pg_if = self.pg_if_by_vrf_id[vrf_id][j]
+            pg_if.unconfig_ip4()
             if pg_if in self.pg_in_vrf:
                 self.pg_in_vrf.remove(pg_if)
             if pg_if not in self.pg_not_in_vrf:
diff --git a/test/test_ip6_vrf_multi_instance.py b/test/test_ip6_vrf_multi_instance.py
index 7bd4d89c..af80b5ba 100644
--- a/test/test_ip6_vrf_multi_instance.py
+++ b/test/test_ip6_vrf_multi_instance.py
@@ -224,6 +224,7 @@ class TestIP6VrfMultiInst(VppTestCase):
             self.vrf_reset_list.append(vrf_id)
         for j in range(self.pg_ifs_per_vrf):
             pg_if = self.pg_if_by_vrf_id[vrf_id][j]
+            pg_if.unconfig_ip6()
             if pg_if in self.pg_in_vrf:
                 self.pg_in_vrf.remove(pg_if)
             if pg_if not in self.pg_not_in_vrf:
diff --git a/test/test_mpls.py b/test/test_mpls.py
index 41d9426b..9082637b 100644
--- a/test/test_mpls.py
+++ b/test/test_mpls.py
@@ -17,10 +17,6 @@ from scapy.contrib.mpls import MPLS
 class TestMPLS(VppTestCase):
     """ MPLS Test Case """
 
-    @classmethod
-    def setUpClass(cls):
-        super(TestMPLS, cls).setUpClass()
-
     def setUp(self):
         super(TestMPLS, self).setUp()
 
@@ -44,6 +40,11 @@ class TestMPLS(VppTestCase):
 
     def tearDown(self):
         super(TestMPLS, self).tearDown()
+        for i in self.pg_interfaces:
+            i.unconfig_ip4()
+            i.unconfig_ip6()
+            i.ip6_disable()
+            i.admin_down()
 
     # the default of 64 matches the IP packet TTL default
     def create_stream_labelled_ip4(
diff --git a/test/test_neighbor.py b/test/test_neighbor.py
index 6a608091..885bf5a4 100644
--- a/test/test_neighbor.py
+++ b/test/test_neighbor.py
@@ -40,6 +40,13 @@ class ARPTestCase(VppTestCase):
         self.pg3.set_table_ip4(1)
         self.pg3.config_ip4()
 
+    def tearDown(self):
+        super(ARPTestCase, self).tearDown()
+        for i in self.pg_interfaces:
+            i.unconfig_ip4()
+            i.unconfig_ip6()
+            i.admin_down()
+
     def verify_arp_req(self, rx, smac, sip, dip):
         ether = rx[Ether]
         self.assertEqual(ether.dst, "ff:ff:ff:ff:ff:ff")
-- 
cgit 1.2.3-korg


From 3466c30261950823828d1dad0d2fb170ee2f9aaf Mon Sep 17 00:00:00 2001
From: Neale Ranns <nranns@cisco.com>
Date: Thu, 16 Feb 2017 07:45:03 -0800
Subject: DHCP Multiple Servers (VPP-602, VPP-605)

Multiple DHCP (4 and/or 6) servers can be added and removed through multiple calls to the 'set dhcp server' API.
All 4/6/ discover/solicit messages will then be replicated to all servers in the list. The expectation is that the servers/system is configured in such a way that this is viable.
If VSS information is providied for the clinet VRF which also has multiple servers configured, then the same VSS information is sent to each server. Likewise the source address of packets sent to from VPP to each server is the same.

Change-Id: I3287cb084c84b3f612b78bc69cfcb5b9c1f8934d
Signed-off-by: Neale Ranns <nranns@cisco.com>
---
 src/scripts/vnet/dhcp/proxy      |   3 +-
 src/vat/api_format.c             |  56 ++++--
 src/vnet/dhcp/client.c           |   2 +-
 src/vnet/dhcp/dhcp.api           |  12 +-
 src/vnet/dhcp/dhcp4_packet.h     |   5 +
 src/vnet/dhcp/dhcp4_proxy_node.c | 151 +++++++++++++----
 src/vnet/dhcp/dhcp6_proxy_node.c | 158 ++++++++++++-----
 src/vnet/dhcp/dhcp_api.c         |  66 +++++---
 src/vnet/dhcp/dhcp_proxy.c       | 129 +++++++-------
 src/vnet/dhcp/dhcp_proxy.h       |  85 +++++++---
 src/vnet/ip/ip6_packet.h         |   2 +
 src/vnet/pg/input.c              |   6 +
 test/test_dhcp.py                | 356 ++++++++++++++++++++++++++++++++++-----
 13 files changed, 796 insertions(+), 235 deletions(-)

(limited to 'test/test_dhcp.py')

diff --git a/src/scripts/vnet/dhcp/proxy b/src/scripts/vnet/dhcp/proxy
index c709d87d..42dff2a0 100644
--- a/src/scripts/vnet/dhcp/proxy
+++ b/src/scripts/vnet/dhcp/proxy
@@ -14,7 +14,8 @@ set int ip addr loop0 2001::1/64
 set int ip addr loop0 2001:1::1/64
 
 set dhcp proxy server 10.255.0.1 src-address 10.0.0.1 server-fib-id 0 rx-fib-id 0
-set dhcp proxy server 10.255.0.2 src-address 10.0.1.1 server-fib-id 1 rx-fib-id 1
+set dhcp proxy server 10.255.0.2 src-address 10.0.0.1 server-fib-id 0 rx-fib-id 0
+set dhcp proxy server 10.255.1.2 src-address 10.0.1.1 server-fib-id 1 rx-fib-id 1
 
 set dhcpv6 proxy server 3001::1 src-address 2001::1 server-fib-id 0 rx-fib-id 0
 set dhcpv6 proxy server 3002::1 src-address 2001:1::1 server-fib-id 1 rx-fib-id 1
diff --git a/src/vat/api_format.c b/src/vat/api_format.c
index 0b60b910..b5943f03 100644
--- a/src/vat/api_format.c
+++ b/src/vat/api_format.c
@@ -7573,23 +7573,35 @@ static void
 vl_api_dhcp_proxy_details_t_handler (vl_api_dhcp_proxy_details_t * mp)
 {
   vat_main_t *vam = &vat_main;
+  u32 i, count = mp->count;
+  vl_api_dhcp_server_t *s;
 
   if (mp->is_ipv6)
     print (vam->ofp,
-	   "RX Table-ID %d, Server Table-ID %d, Server Address %U, Source Address %U, VSS FIB-ID %d, VSS OUI %d",
+	   "RX Table-ID %d, Source Address %U, VSS FIB-ID %d, VSS OUI %d",
 	   ntohl (mp->rx_vrf_id),
-	   ntohl (mp->server_vrf_id),
-	   format_ip6_address, mp->dhcp_server,
 	   format_ip6_address, mp->dhcp_src_address,
 	   ntohl (mp->vss_oui), ntohl (mp->vss_fib_id));
   else
     print (vam->ofp,
-	   "RX Table-ID %d, Server Table-ID %d, Server Address %U, Source Address %U, VSS FIB-ID %d, VSS OUI %d",
+	   "RX Table-ID %d, Source Address %U, VSS FIB-ID %d, VSS OUI %d",
 	   ntohl (mp->rx_vrf_id),
-	   ntohl (mp->server_vrf_id),
-	   format_ip4_address, mp->dhcp_server,
 	   format_ip4_address, mp->dhcp_src_address,
 	   ntohl (mp->vss_oui), ntohl (mp->vss_fib_id));
+
+  for (i = 0; i < count; i++)
+    {
+      s = &mp->servers[i];
+
+      if (mp->is_ipv6)
+	print (vam->ofp,
+	       " Server Table-ID %d, Server Address %U",
+	       ntohl (s->server_vrf_id), format_ip6_address, s->dhcp_server);
+      else
+	print (vam->ofp,
+	       " Server Table-ID %d, Server Address %U",
+	       ntohl (s->server_vrf_id), format_ip4_address, s->dhcp_server);
+    }
 }
 
 static void vl_api_dhcp_proxy_details_t_handler_json
@@ -7597,8 +7609,10 @@ static void vl_api_dhcp_proxy_details_t_handler_json
 {
   vat_main_t *vam = &vat_main;
   vat_json_node_t *node = NULL;
+  u32 i, count = mp->count;
   struct in_addr ip4;
   struct in6_addr ip6;
+  vl_api_dhcp_server_t *s;
 
   if (VAT_JSON_ARRAY != vam->json_tree.type)
     {
@@ -7609,24 +7623,38 @@ static void vl_api_dhcp_proxy_details_t_handler_json
 
   vat_json_init_object (node);
   vat_json_object_add_uint (node, "rx-table-id", ntohl (mp->rx_vrf_id));
-  vat_json_object_add_uint (node, "server-table-id",
-			    ntohl (mp->server_vrf_id));
+  vat_json_object_add_uint (node, "vss-fib-id", ntohl (mp->vss_fib_id));
+  vat_json_object_add_uint (node, "vss-oui", ntohl (mp->vss_oui));
+
   if (mp->is_ipv6)
     {
-      clib_memcpy (&ip6, &mp->dhcp_server, sizeof (ip6));
-      vat_json_object_add_ip6 (node, "server_address", ip6);
       clib_memcpy (&ip6, &mp->dhcp_src_address, sizeof (ip6));
       vat_json_object_add_ip6 (node, "src_address", ip6);
     }
   else
     {
-      clib_memcpy (&ip4, &mp->dhcp_server, sizeof (ip4));
-      vat_json_object_add_ip4 (node, "server_address", ip4);
       clib_memcpy (&ip4, &mp->dhcp_src_address, sizeof (ip4));
       vat_json_object_add_ip4 (node, "src_address", ip4);
     }
-  vat_json_object_add_uint (node, "vss-fib-id", ntohl (mp->vss_fib_id));
-  vat_json_object_add_uint (node, "vss-oui", ntohl (mp->vss_oui));
+
+  for (i = 0; i < count; i++)
+    {
+      s = &mp->servers[i];
+
+      vat_json_object_add_uint (node, "server-table-id",
+				ntohl (s->server_vrf_id));
+
+      if (mp->is_ipv6)
+	{
+	  clib_memcpy (&ip4, &s->dhcp_server, sizeof (ip4));
+	  vat_json_object_add_ip4 (node, "src_address", ip4);
+	}
+      else
+	{
+	  clib_memcpy (&ip6, &s->dhcp_server, sizeof (ip6));
+	  vat_json_object_add_ip6 (node, "server_address", ip6);
+	}
+    }
 }
 
 static int
diff --git a/src/vnet/dhcp/client.c b/src/vnet/dhcp/client.c
index d34c5a64..29749a33 100644
--- a/src/vnet/dhcp/client.c
+++ b/src/vnet/dhcp/client.c
@@ -366,7 +366,7 @@ send_dhcp_pkt (dhcp_client_main_t * dcm, dhcp_client_t * c,
   o = (dhcp_option_t * )dhcp->options;
 
   /* Send option 53, the DHCP message type */
-  o->option = 53;
+  o->option = DHCP_PACKET_OPTION_MSG_TYPE;
   o->length = 1;
   o->data[0] = type;
   o = (dhcp_option_t *) (((uword) o) + (o->length + 2));
diff --git a/src/vnet/dhcp/dhcp.api b/src/vnet/dhcp/dhcp.api
index 8daadd8c..2db85a79 100644
--- a/src/vnet/dhcp/dhcp.api
+++ b/src/vnet/dhcp/dhcp.api
@@ -137,19 +137,25 @@ define dhcp_proxy_dump
   u8  is_ip6;
 };
 
+typeonly manual_print manual_endian define dhcp_server
+{
+  u32 server_vrf_id;
+  u8 dhcp_server[16];
+};
+
 /** \brief Tell client about a DHCP completion event
     @param client_index - opaque cookie to identify the sender
 */
-define dhcp_proxy_details
+manual_endian manual_print define dhcp_proxy_details
 {
   u32 context;
   u32 rx_vrf_id;
-  u32 server_vrf_id;
   u32 vss_oui;
   u32 vss_fib_id;
   u8 is_ipv6;
-  u8 dhcp_server[16];
   u8 dhcp_src_address[16];
+  u8 count;
+  vl_api_dhcp_server_t servers[count];
 };
 
 /*
diff --git a/src/vnet/dhcp/dhcp4_packet.h b/src/vnet/dhcp/dhcp4_packet.h
index 28c4b156..07829f48 100644
--- a/src/vnet/dhcp/dhcp4_packet.h
+++ b/src/vnet/dhcp/dhcp4_packet.h
@@ -55,6 +55,11 @@ typedef enum {
   DHCP_PACKET_ACK=5,
 } dhcp_packet_type_t;
 
+typedef enum dhcp_packet_option_t_
+{
+    DHCP_PACKET_OPTION_MSG_TYPE = 53,
+} dhcp_packet_option_t;
+
 /* charming antique: 99.130.83.99 is the dhcp magic cookie */
 #define DHCP_MAGIC (clib_host_to_net_u32(0x63825363))
 
diff --git a/src/vnet/dhcp/dhcp4_proxy_node.c b/src/vnet/dhcp/dhcp4_proxy_node.c
index 88a99249..1c84881a 100644
--- a/src/vnet/dhcp/dhcp4_proxy_node.c
+++ b/src/vnet/dhcp/dhcp4_proxy_node.c
@@ -135,18 +135,17 @@ dhcp_proxy_to_server_input (vlib_main_t * vm,
           u32 original_sw_if_index = 0;
           u8  *end = NULL;
           u32 fib_index;
-          dhcp_server_t * server;
+          dhcp_proxy_t *proxy;
+          dhcp_server_t *server;
           u32 rx_sw_if_index;
           dhcp_option_t *o;
           u32 len = 0;
           vlib_buffer_free_list_t *fl;
+          u8 is_discover = 0;
 
 	  bi0 = from[0];
-	  to_next[0] = bi0;
 	  from += 1;
-	  to_next += 1;
 	  n_left_from -= 1;
-	  n_left_to_next -= 1;
 
 	  b0 = vlib_get_buffer (vm, bi0);
 
@@ -172,16 +171,17 @@ dhcp_proxy_to_server_input (vlib_main_t * vm,
           rx_sw_if_index = vnet_buffer(b0)->sw_if_index[VLIB_RX];
 
           fib_index = im->fib_index_by_sw_if_index [rx_sw_if_index];
-          server = dhcp_get_server(dpm, fib_index, FIB_PROTOCOL_IP4);
-          
-          if (PREDICT_FALSE (NULL == server))
+          proxy = dhcp_get_proxy(dpm, fib_index, FIB_PROTOCOL_IP4);
+
+          if (PREDICT_FALSE (NULL == proxy))
             {
               error0 = DHCP_PROXY_ERROR_NO_SERVER;
               next0 = DHCP_PROXY_TO_SERVER_INPUT_NEXT_DROP;
               pkts_no_server++;
               goto do_trace;
             }
-          
+
+          server = &proxy->dhcp_servers[0];
           vlib_buffer_advance (b0, -(sizeof(*ip0)));
           ip0 = vlib_buffer_get_current (b0);
 
@@ -198,7 +198,7 @@ dhcp_proxy_to_server_input (vlib_main_t * vm,
 
           sum0 = ip0->checksum;
           old0 = ip0->src_address.as_u32;
-          new0 = server->dhcp_src_address.ip4.as_u32;
+          new0 = proxy->dhcp_src_address.ip4.as_u32;
           ip0->src_address.as_u32 = new0;
           sum0 = ip_csum_update (sum0, old0, new0, 
                                 ip4_header_t /* structure */, 
@@ -209,7 +209,7 @@ dhcp_proxy_to_server_input (vlib_main_t * vm,
           vnet_buffer(b0)->sw_if_index[VLIB_TX] =
             server->server_fib_index;
 
-          h0->gateway_ip_address.as_u32 = server->dhcp_src_address.ip4.as_u32;
+          h0->gateway_ip_address.as_u32 = proxy->dhcp_src_address.ip4.as_u32;
           pkts_to_server++;
 
           o = (dhcp_option_t *) h0->options;
@@ -220,7 +220,16 @@ dhcp_proxy_to_server_input (vlib_main_t * vm,
           end = b0->data + b0->current_data + b0->current_length;
           /* TLVs are not performance-friendly... */
           while  (o->option != 0xFF /* end of options */ && (u8 *)o < end) 
+            {
+              if (DHCP_PACKET_OPTION_MSG_TYPE == o->option)
+                {
+                  if (DHCP_PACKET_DISCOVER == o->data[0])
+                    {
+                      is_discover = 1;
+                    }
+                }
               o = (dhcp_option_t *) (((uword) o) + (o->length + 2));
+          }
 
           fl = vlib_buffer_get_free_list (vm, b0->free_list_index);
           // start write at (option*)o, some packets have padding
@@ -340,6 +349,65 @@ dhcp_proxy_to_server_input (vlib_main_t * vm,
           
           next0 = DHCP_PROXY_TO_SERVER_INPUT_NEXT_LOOKUP;
 
+          /*
+           * If we have multiple servers configured and this is the
+           * client's discover message, then send copies to each of
+           * those servers
+           */
+          if (is_discover && vec_len(proxy->dhcp_servers) > 1)
+          {
+              u32 ii;
+
+              for (ii = 1; ii < vec_len(proxy->dhcp_servers); ii++)
+              {
+                  vlib_buffer_t *c0;
+                  u32 ci0;
+              
+                  c0 = vlib_buffer_copy(vm, b0);
+                  ci0 = vlib_get_buffer_index(vm, c0);
+                  server = &proxy->dhcp_servers[ii];
+
+                  ip0 = vlib_buffer_get_current (c0);
+
+                  sum0 = ip0->checksum;
+                  old0 = ip0->dst_address.as_u32;
+                  new0 = server->dhcp_server.ip4.as_u32;
+                  ip0->dst_address.as_u32 = server->dhcp_server.ip4.as_u32;
+                  sum0 = ip_csum_update (sum0, old0, new0, 
+                                         ip4_header_t /* structure */, 
+                                         dst_address /* changed member */);
+                  ip0->checksum = ip_csum_fold (sum0);
+
+                  to_next[0] = ci0;
+                  to_next += 1;
+                  n_left_to_next -= 1;
+
+                  vlib_validate_buffer_enqueue_x1 (vm, node, next_index,
+                                                   to_next, n_left_to_next,
+                                                   ci0, next0);
+
+                  if (PREDICT_FALSE(b0->flags & VLIB_BUFFER_IS_TRACED)) 
+                  {
+                      dhcp_proxy_trace_t *tr;
+
+                      tr = vlib_add_trace (vm, node, c0, sizeof (*tr));
+                      tr->which = 0; /* to server */
+                      tr->error = error0;
+                      tr->original_sw_if_index = original_sw_if_index;
+                      tr->sw_if_index = sw_if_index;
+                      if (next0 == DHCP_PROXY_TO_SERVER_INPUT_NEXT_LOOKUP)
+                          tr->trace_ip4_address.as_u32 = server->dhcp_server.ip4.as_u32;
+                  }
+
+                  if (PREDICT_FALSE(0 == n_left_to_next))
+                  {
+                      vlib_put_next_frame (vm, node, next_index,
+                                           n_left_to_next);
+                      vlib_get_next_frame (vm, node, next_index,
+                                           to_next, n_left_to_next);
+                  }
+              }
+          }
         do_trace:
           if (PREDICT_FALSE(b0->flags & VLIB_BUFFER_IS_TRACED)) 
             {
@@ -350,10 +418,15 @@ dhcp_proxy_to_server_input (vlib_main_t * vm,
                tr->original_sw_if_index = original_sw_if_index;
                tr->sw_if_index = sw_if_index;
                if (next0 == DHCP_PROXY_TO_SERVER_INPUT_NEXT_LOOKUP)
-                 tr->trace_ip4_address.as_u32 = server->dhcp_server.ip4.as_u32;
+                 tr->trace_ip4_address.as_u32 =
+                     proxy->dhcp_servers[0].dhcp_server.ip4.as_u32;
             }
 
         do_enqueue:
+	  to_next[0] = bi0;
+	  to_next += 1;
+	  n_left_to_next -= 1;
+
 	  vlib_validate_buffer_enqueue_x1 (vm, node, next_index,
 					   to_next, n_left_to_next,
 					   bi0, next0);
@@ -437,7 +510,8 @@ dhcp_proxy_to_client_input (vlib_main_t * vm,
       u32 error0 = (u32)~0;
       vnet_sw_interface_t *swif;
       u32 fib_index;
-      dhcp_server_t * server;
+      dhcp_proxy_t *proxy;
+      dhcp_server_t *server;
       u32 original_sw_if_index = (u32) ~0;
       ip4_address_t relay_addr = {
           .as_u32 = 0,
@@ -547,20 +621,26 @@ dhcp_proxy_to_client_input (vlib_main_t * vm,
         }
 
       fib_index = im->fib_index_by_sw_if_index [sw_if_index];
-      server = dhcp_get_server(dpm, fib_index, FIB_PROTOCOL_IP4);
+      proxy = dhcp_get_proxy(dpm, fib_index, FIB_PROTOCOL_IP4);
 
-      if (PREDICT_FALSE (NULL == server))
+      if (PREDICT_FALSE (NULL == proxy))
         {
           error0 = DHCP_PROXY_ERROR_NO_SERVER;
           goto drop_packet;
         }
       
-      if (ip0->src_address.as_u32 != server->dhcp_server.ip4.as_u32)
-        {             
-          error0 = DHCP_PROXY_ERROR_BAD_SVR_FIB_OR_ADDRESS;
-          goto drop_packet;
+      vec_foreach(server, proxy->dhcp_servers)
+        {
+          if (ip0->src_address.as_u32 == server->dhcp_server.ip4.as_u32)
+            {
+              goto server_found;
+            }
         }
 
+      error0 = DHCP_PROXY_ERROR_BAD_SVR_FIB_OR_ADDRESS;
+      goto drop_packet;
+
+    server_found:
       vnet_buffer (b0)->sw_if_index[VLIB_TX] = sw_if_index;
 
       swif = vnet_get_sw_interface (vnm, sw_if_index);
@@ -709,9 +789,8 @@ dhcp4_proxy_set_server (ip46_address_t *addr,
 
   if (is_del)
     {
-      rc = dhcp_proxy_server_del (FIB_PROTOCOL_IP4, rx_fib_index);
-
-      if (0 == rc)
+      if (dhcp_proxy_server_del (FIB_PROTOCOL_IP4, rx_fib_index,
+                                 addr, server_table_id))
       {
           fib_table_entry_special_remove(rx_fib_index,
                                          &all_1s,
@@ -809,29 +888,35 @@ VLIB_CLI_COMMAND (dhcp_proxy_set_command, static) = {
 static u8 *
 format_dhcp4_proxy_server (u8 * s, va_list * args)
 {
-  dhcp_server_t * server = va_arg (*args, dhcp_server_t *);
+  dhcp_proxy_t *proxy = va_arg (*args, dhcp_proxy_t *);
   ip4_fib_t * rx_fib, * server_fib;
+  dhcp_server_t *server;
 
-  if (server == 0)
+  if (proxy == 0)
     {
-      s = format (s, "%=16s%=16s%=14s%=14s", "Server", "Src Address", 
-                  "Server FIB", "RX FIB");
+        s = format (s, "%=14s%=16s%s", "RX FIB", "Src Address", 
+                    "Servers FIB,Address");
       return s;
     }
 
-  server_fib = ip4_fib_get(server->server_fib_index);
-  rx_fib = ip4_fib_get(server->rx_fib_index);
+  rx_fib = ip4_fib_get(proxy->rx_fib_index);
+
+  s = format (s, "%=14u%=16U",
+              rx_fib->table_id,
+              format_ip46_address, &proxy->dhcp_src_address, IP46_TYPE_ANY);
 
-  s = format (s, "%=16U%=16U%=14u%=14u",
-              format_ip46_address, &server->dhcp_server, IP46_TYPE_ANY,
-              format_ip46_address, &server->dhcp_src_address, IP46_TYPE_ANY,
-              server_fib->table_id,
-              rx_fib->table_id);
+  vec_foreach(server, proxy->dhcp_servers)
+  {
+      server_fib = ip4_fib_get(server->server_fib_index);
+      s = format (s, "%u,%U  ",
+                  server_fib->table_id,
+                  format_ip46_address, &server->dhcp_server, IP46_TYPE_ANY);
+  }
   return s;
 }
 
 static int
-dhcp4_proxy_show_walk (dhcp_server_t *server,
+dhcp4_proxy_show_walk (dhcp_proxy_t *server,
                        void *ctx)
 {
     vlib_main_t * vm = ctx;
diff --git a/src/vnet/dhcp/dhcp6_proxy_node.c b/src/vnet/dhcp/dhcp6_proxy_node.c
index 58674209..524cb095 100644
--- a/src/vnet/dhcp/dhcp6_proxy_node.c
+++ b/src/vnet/dhcp/dhcp6_proxy_node.c
@@ -140,7 +140,8 @@ dhcpv6_proxy_to_server_input (vlib_main_t * vm,
   ip6_main_t * im = &ip6_main;
   ip6_address_t * src;
   int bogus_length;
-  dhcp_server_t * server;
+  dhcp_proxy_t *proxy;
+  dhcp_server_t *server;
   u32  rx_fib_idx = 0, server_fib_idx = 0;
 
   next_index = node->cached_next_index;
@@ -176,13 +177,11 @@ dhcpv6_proxy_to_server_input (vlib_main_t * vm,
           u8 client_src_mac[6];
           vlib_buffer_free_list_t *fl;
           dhcp_vss_t *vss;
+          u8 is_solicit = 0;
 
 	  bi0 = from[0];
-	  to_next[0] = bi0;
 	  from += 1;
-	  to_next += 1;
 	  n_left_from -= 1;
-	  n_left_to_next -= 1;
 
 	  b0 = vlib_get_buffer (vm, bi0);
 
@@ -227,9 +226,9 @@ dhcpv6_proxy_to_server_input (vlib_main_t * vm,
           /* Send to DHCPV6 server via the configured FIB */
           rx_sw_if_index = sw_if_index =  vnet_buffer(b0)->sw_if_index[VLIB_RX];
           rx_fib_idx = im->mfib_index_by_sw_if_index [rx_sw_if_index];
-          server = dhcp_get_server(dpm, rx_fib_idx, FIB_PROTOCOL_IP6);
+          proxy = dhcp_get_proxy(dpm, rx_fib_idx, FIB_PROTOCOL_IP6);
 
-          if (PREDICT_FALSE (NULL == server))
+          if (PREDICT_FALSE (NULL == proxy))
           {
               error0 = DHCPV6_PROXY_ERROR_NO_SERVER;
               next0 = DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_DROP;
@@ -237,6 +236,7 @@ dhcpv6_proxy_to_server_input (vlib_main_t * vm,
               goto do_trace;
           }
 
+          server = &proxy->dhcp_servers[0];
           server_fib_idx = server->server_fib_index;
           vnet_buffer(b0)->sw_if_index[VLIB_TX] = server_fib_idx;
 
@@ -371,18 +371,19 @@ dhcpv6_proxy_to_server_input (vlib_main_t * vm,
           ip1->payload_length =  u1->length;
           ip1->protocol = PROTO_UDP;
           ip1->hop_limit = HOP_COUNT_LIMIT;
-	      src = (server->dhcp_server.ip6.as_u64[0] ||
-                     server->dhcp_server.ip6.as_u64[1]) ?
-	        &server->dhcp_server.ip6 : &all_dhcpv6_server_address;
+          src = ((server->dhcp_server.ip6.as_u64[0] ||
+                  server->dhcp_server.ip6.as_u64[1]) ?
+                 &server->dhcp_server.ip6 :
+                 &all_dhcpv6_server_address);
           copy_ip6_address(&ip1->dst_address, src);
 
 
           ia0 = ip6_interface_first_global_or_site_address
               (&ip6_main, vnet_buffer(b0)->sw_if_index[VLIB_RX]);
 
-	      src = (server->dhcp_src_address.ip6.as_u64[0] ||
-                     server->dhcp_src_address.ip6.as_u64[1]) ?
-	        &server->dhcp_src_address.ip6 : ia0;
+	      src = (proxy->dhcp_src_address.ip6.as_u64[0] ||
+                     proxy->dhcp_src_address.ip6.as_u64[1]) ?
+	        &proxy->dhcp_src_address.ip6 : ia0;
           if (ia0 == 0)
             {
               error0 = DHCPV6_PROXY_ERROR_NO_SRC_ADDRESS;
@@ -400,6 +401,66 @@ dhcpv6_proxy_to_server_input (vlib_main_t * vm,
 
           next0 = DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_LOOKUP;
 
+          is_solicit = (DHCPV6_MSG_SOLICIT == h0->u.msg_type);
+
+          /*
+           * If we have multiple servers configured and this is the
+           * client's discover message, then send copies to each of
+           * those servers
+           */
+          if (is_solicit && vec_len(proxy->dhcp_servers) > 1)
+          {
+              u32 ii;
+
+              for (ii = 1; ii < vec_len(proxy->dhcp_servers); ii++)
+              {
+                  vlib_buffer_t *c0;
+                  u32 ci0;
+              
+                  c0 = vlib_buffer_copy(vm, b0);
+                  ci0 = vlib_get_buffer_index(vm, c0);
+                  server = &proxy->dhcp_servers[ii];
+
+                  ip0 = vlib_buffer_get_current (c0);
+
+                  src = ((server->dhcp_server.ip6.as_u64[0] ||
+                          server->dhcp_server.ip6.as_u64[1]) ?
+                         &server->dhcp_server.ip6 :
+                         &all_dhcpv6_server_address);
+                  copy_ip6_address(&ip1->dst_address, src);
+
+                  to_next[0] = ci0;
+                  to_next += 1;
+                  n_left_to_next -= 1;
+
+                  vlib_validate_buffer_enqueue_x1 (vm, node, next_index,
+                                                   to_next, n_left_to_next,
+                                                   ci0, next0);
+
+                  if (PREDICT_FALSE(b0->flags & VLIB_BUFFER_IS_TRACED)) 
+                  {
+                      dhcpv6_proxy_trace_t *tr;
+
+                      tr = vlib_add_trace (vm, node, c0, sizeof (*tr));
+                      tr->which = 0; /* to server */
+                      tr->error = error0;
+                      tr->original_sw_if_index = rx_sw_if_index;
+                      tr->sw_if_index = sw_if_index;
+                      if (next0 == DHCPV6_PROXY_TO_SERVER_INPUT_NEXT_LOOKUP)
+                          copy_ip6_address((ip6_address_t *)&tr->packet_data[0],
+                                           &server->dhcp_server.ip6);
+                  }
+
+                  if (PREDICT_FALSE(0 == n_left_to_next))
+                  {
+                      vlib_put_next_frame (vm, node, next_index,
+                                           n_left_to_next);
+                      vlib_get_next_frame (vm, node, next_index,
+                                           to_next, n_left_to_next);
+                  }
+              }
+          }
+
         do_trace:
           if (PREDICT_FALSE(b0->flags & VLIB_BUFFER_IS_TRACED))
             {
@@ -414,6 +475,10 @@ dhcpv6_proxy_to_server_input (vlib_main_t * vm,
             }
 
         do_enqueue:
+	  to_next[0] = bi0;
+	  to_next += 1;
+	  n_left_to_next -= 1;
+
 	  vlib_validate_buffer_enqueue_x1 (vm, node, next_index,
 					   to_next, n_left_to_next,
 					   bi0, next0);
@@ -475,7 +540,8 @@ dhcpv6_proxy_to_client_input (vlib_main_t * vm,
   u32 n_left_from, * from;
   ethernet_main_t *em = ethernet_get_main (vm);
   dhcp_proxy_main_t * dm = &dhcp_proxy_main;
-  dhcp_server_t * server;
+  dhcp_proxy_t *proxy;
+  dhcp_server_t *server;
   vnet_main_t * vnm = vnet_get_main();
   int bogus_length;
 
@@ -588,9 +654,9 @@ dhcpv6_proxy_to_client_input (vlib_main_t * vm,
       vlib_buffer_advance (b0, sizeof(*r0));
 
       client_fib_idx = im->mfib_index_by_sw_if_index[sw_if_index];
-      server = dhcp_get_server(dm, client_fib_idx, FIB_PROTOCOL_IP6);
+      proxy = dhcp_get_proxy(dm, client_fib_idx, FIB_PROTOCOL_IP6);
 
-      if (NULL == server)
+      if (NULL == proxy)
       {
     	  error0 = DHCPV6_PROXY_ERROR_NO_SERVER;
           goto drop_packet;
@@ -599,15 +665,21 @@ dhcpv6_proxy_to_client_input (vlib_main_t * vm,
       server_fib_idx = im->fib_index_by_sw_if_index
           [vnet_buffer(b0)->sw_if_index[VLIB_RX]];
 
-      if (server_fib_idx != server->server_fib_index ||
-          ip0->src_address.as_u64[0] != server->dhcp_server.ip6.as_u64[0] ||
-          ip0->src_address.as_u64[1] != server->dhcp_server.ip6.as_u64[1])
-        {
-          //drop packet if not from server with configured address or FIB
-          error0 = DHCPV6_PROXY_ERROR_BAD_SVR_FIB_OR_ADDRESS;
-          goto drop_packet;
-        }
+      vec_foreach(server, proxy->dhcp_servers)
+      {
+          if (server_fib_idx == server->server_fib_index &&
+              ip0->src_address.as_u64[0] == server->dhcp_server.ip6.as_u64[0] &&
+              ip0->src_address.as_u64[1] == server->dhcp_server.ip6.as_u64[1])
+          {
+              goto server_found;
+          }
+      }
+
+      //drop packet if not from server with configured address or FIB
+      error0 = DHCPV6_PROXY_ERROR_BAD_SVR_FIB_OR_ADDRESS;
+      goto drop_packet;
 
+    server_found:
       vnet_buffer (b0)->sw_if_index[VLIB_TX] = original_sw_if_index
           = sw_if_index;
 
@@ -773,9 +845,8 @@ dhcp6_proxy_set_server (ip46_address_t *addr,
 
   if (is_del)
     {
-      rc = dhcp_proxy_server_del (FIB_PROTOCOL_IP6, rx_fib_index);
-
-      if (0 == rc)
+      if (dhcp_proxy_server_del (FIB_PROTOCOL_IP6, rx_fib_index,
+                                 addr, server_table_id))
       {
           mfib_table_entry_delete(rx_fib_index,
                                   &all_dhcp_servers,
@@ -893,43 +964,50 @@ VLIB_CLI_COMMAND (dhcpv6_proxy_set_command, static) = {
 static u8 *
 format_dhcp6_proxy_server (u8 * s, va_list * args)
 {
-  dhcp_server_t * server = va_arg (*args, dhcp_server_t *);
+  dhcp_proxy_t * proxy = va_arg (*args, dhcp_proxy_t *);
   ip6_fib_t *server_fib;
+  dhcp_server_t *server;
   ip6_mfib_t *rx_fib;
 
-  if (NULL == server)
+  if (proxy == 0)
     {
-      s = format (s, "%=40s%=40s%=14s%=14s", "Server Address", "Source Address",
-                  "Server FIB", "RX FIB");
+        s = format (s, "%=14s%=16s%s", "RX FIB", "Src Address", 
+                    "Servers FIB,Address");
       return s;
     }
 
-  server_fib = ip6_fib_get(server->server_fib_index);
-  rx_fib = ip6_mfib_get(server->rx_fib_index);
+  rx_fib = ip6_mfib_get(proxy->rx_fib_index);
+
+  s = format (s, "%=14u%=16U",
+              rx_fib->table_id,
+              format_ip46_address, &proxy->dhcp_src_address, IP46_TYPE_ANY);
 
+  vec_foreach(server, proxy->dhcp_servers)
+  {
+      server_fib = ip6_fib_get(server->server_fib_index);
+      s = format (s, "%u,%U  ",
+                  server_fib->table_id,
+                  format_ip46_address, &server->dhcp_server, IP46_TYPE_ANY);
+  }
 
-  s = format (s, "%=40U%=40U%=14u%=14u",
-              format_ip46_address, &server->dhcp_server, IP46_TYPE_ANY,
-              format_ip46_address, &server->dhcp_src_address, IP46_TYPE_ANY,
-              server_fib->table_id, rx_fib->table_id);
   return s;
 }
 
 static int
-dhcp6_proxy_show_walk (dhcp_server_t *server,
+dhcp6_proxy_show_walk (dhcp_proxy_t *proxy,
                        void *ctx)
 {
     vlib_main_t * vm = ctx;
 
-    vlib_cli_output (vm, "%U", format_dhcp6_proxy_server, server);
+    vlib_cli_output (vm, "%U", format_dhcp6_proxy_server, proxy);
 
     return (1);
 }
 
 static clib_error_t *
 dhcpv6_proxy_show_command_fn (vlib_main_t * vm,
-                            unformat_input_t * input,
-                            vlib_cli_command_t * cmd)
+                              unformat_input_t * input,
+                              vlib_cli_command_t * cmd)
 {
   vlib_cli_output (vm, "%U", format_dhcp6_proxy_server, NULL /* header line */);
 
diff --git a/src/vnet/dhcp/dhcp_api.c b/src/vnet/dhcp/dhcp_api.c
index ce34f6a4..e9c757e8 100644
--- a/src/vnet/dhcp/dhcp_api.c
+++ b/src/vnet/dhcp/dhcp_api.c
@@ -24,6 +24,7 @@
 #include <vnet/api_errno.h>
 #include <vnet/dhcp/dhcp_proxy.h>
 #include <vnet/dhcp/client.h>
+#include <vnet/fib/fib_table.h>
 
 #include <vnet/vnet_msg_enum.h>
 
@@ -113,46 +114,73 @@ vl_api_dhcp_proxy_dump_t_handler (vl_api_dhcp_proxy_dump_t * mp)
   if (q == 0)
     return;
 
-  dhcp_proxy_dump ((mp->is_ip6 == 0 ?
+  dhcp_proxy_dump ((mp->is_ip6 == 1 ?
 		    FIB_PROTOCOL_IP6 : FIB_PROTOCOL_IP4), q, mp->context);
 }
 
 void
 dhcp_send_details (fib_protocol_t proto,
-		   void *opaque,
-		   u32 context,
-		   const ip46_address_t * server,
-		   const ip46_address_t * src,
-		   u32 server_fib_id,
-		   u32 rx_fib_id, u32 vss_fib_id, u32 vss_oui)
+		   void *opaque, u32 context, dhcp_proxy_t * proxy)
 {
   vl_api_dhcp_proxy_details_t *mp;
   unix_shared_memory_queue_t *q = opaque;
-
-  mp = vl_msg_api_alloc (sizeof (*mp));
+  vl_api_dhcp_server_t *v_server;
+  dhcp_server_t *server;
+  fib_table_t *s_fib;
+  dhcp_vss_t *vss;
+  u32 count;
+  size_t n;
+
+  count = vec_len (proxy->dhcp_servers);
+  n = sizeof (*mp) + (count * sizeof (vl_api_dhcp_server_t));
+  mp = vl_msg_api_alloc (n);
   if (!mp)
     return;
-  memset (mp, 0, sizeof (*mp));
+  memset (mp, 0, n);
   mp->_vl_msg_id = ntohs (VL_API_DHCP_PROXY_DETAILS);
   mp->context = context;
-
-  mp->rx_vrf_id = htonl (rx_fib_id);
-  mp->server_vrf_id = htonl (server_fib_id);
-  mp->vss_oui = htonl (vss_oui);
-  mp->vss_fib_id = htonl (vss_fib_id);
+  mp->count = count;
 
   mp->is_ipv6 = (proto == FIB_PROTOCOL_IP6);
+  mp->rx_vrf_id =
+    htonl (dhcp_proxy_rx_table_get_table_id (proto, proxy->rx_fib_index));
+
+  vss = dhcp_get_vss_info (&dhcp_proxy_main, proxy->rx_fib_index, proto);
+
+  if (NULL != vss)
+    {
+      mp->vss_oui = htonl (vss->oui);
+      mp->vss_fib_id = htonl (vss->fib_id);
+    }
+
+  vec_foreach_index (count, proxy->dhcp_servers)
+  {
+    server = &proxy->dhcp_servers[count];
+    v_server = &mp->servers[count];
+
+    s_fib = fib_table_get (server->server_fib_index, proto);
+
+    v_server->server_vrf_id = htonl (s_fib->ft_table_id);
+
+    if (mp->is_ipv6)
+      {
+	memcpy (v_server->dhcp_server, &server->dhcp_server.ip6, 16);
+      }
+    else
+      {
+	/* put the address in the first bytes */
+	memcpy (v_server->dhcp_server, &server->dhcp_server.ip4, 4);
+      }
+  }
 
   if (mp->is_ipv6)
     {
-      memcpy (mp->dhcp_server, server, 16);
-      memcpy (mp->dhcp_src_address, src, 16);
+      memcpy (mp->dhcp_src_address, &proxy->dhcp_src_address.ip6, 16);
     }
   else
     {
       /* put the address in the first bytes */
-      memcpy (mp->dhcp_server, &server->ip4, 4);
-      memcpy (mp->dhcp_src_address, &src->ip4, 4);
+      memcpy (mp->dhcp_src_address, &proxy->dhcp_src_address.ip4, 4);
     }
   vl_msg_api_send_shmem (q, (u8 *) & mp);
 }
diff --git a/src/vnet/dhcp/dhcp_proxy.c b/src/vnet/dhcp/dhcp_proxy.c
index 8e31c3db..ba7f354e 100644
--- a/src/vnet/dhcp/dhcp_proxy.c
+++ b/src/vnet/dhcp/dhcp_proxy.c
@@ -44,7 +44,7 @@ dhcp_proxy_rx_table_unlock (fib_protocol_t proto,
         mfib_table_unlock(fib_index, proto);
 }
 
-static u32
+ u32
 dhcp_proxy_rx_table_get_table_id (fib_protocol_t proto,
                                   u32 fib_index)
 {
@@ -72,7 +72,7 @@ dhcp_proxy_walk (fib_protocol_t proto,
                  void *ctx)
 {
   dhcp_proxy_main_t * dpm = &dhcp_proxy_main;
-  dhcp_server_t * server;
+  dhcp_proxy_t * server;
   u32 server_index, i;
 
   vec_foreach_index (i, dpm->dhcp_server_index_by_rx_fib_index[proto])
@@ -124,31 +124,68 @@ dhcp_vss_walk (fib_protocol_t proto,
     }
 }
 
+static u32
+dhcp_proxy_server_find (dhcp_proxy_t *proxy,
+                        fib_protocol_t proto,
+                        ip46_address_t *addr,
+                        u32 server_table_id)
+{
+    dhcp_server_t *server;
+    u32 ii, fib_index;
+
+    vec_foreach_index(ii, proxy->dhcp_servers)
+    {
+        server = &proxy->dhcp_servers[ii];
+        fib_index = fib_table_find(proto, server_table_id);
+
+        if (ip46_address_is_equal(&server->dhcp_server,
+                                  addr) &&
+            (server->server_fib_index == fib_index))
+        {
+            return (ii);
+        }
+    }
+    return (~0);
+}
+
 int
 dhcp_proxy_server_del (fib_protocol_t proto,
-                       u32 rx_fib_index)
+                       u32 rx_fib_index,
+                       ip46_address_t *addr,
+                       u32 server_table_id)
 {
   dhcp_proxy_main_t * dpm = &dhcp_proxy_main;
-  dhcp_server_t * server = 0;
-  int rc = 0;
+  dhcp_proxy_t *proxy = 0;
 
-  server = dhcp_get_server(dpm, rx_fib_index, proto);
+  proxy = dhcp_get_proxy(dpm, rx_fib_index, proto);
 
-  if (NULL == server)
-  {
-      rc = VNET_API_ERROR_NO_SUCH_ENTRY;
-  }
-  else
+  if (NULL != proxy)
   {
-      /* Use the default server again.  */
-      dpm->dhcp_server_index_by_rx_fib_index[proto][rx_fib_index] = ~0;
+      dhcp_server_t *server;
+      u32 index;
 
-      fib_table_unlock (server->server_fib_index, proto);
+      index = dhcp_proxy_server_find(proxy, proto, addr, server_table_id);
 
-      pool_put (dpm->dhcp_servers[proto], server);
+      if (~0 != index)
+      {
+          server = &proxy->dhcp_servers[index];
+          fib_table_unlock (server->server_fib_index, proto);
+
+          vec_del1(proxy->dhcp_servers, index);
+
+          if (0 == vec_len(proxy->dhcp_servers))
+          {
+              /* no servers left, delete the proxy config */
+              dpm->dhcp_server_index_by_rx_fib_index[proto][rx_fib_index] = ~0;
+              vec_free(proxy->dhcp_servers);
+              pool_put (dpm->dhcp_servers[proto], proxy);
+              return (1);
+          }
+      }
   }
 
-  return (rc);
+  /* the proxy still exists */
+  return (0);
 }
 
 int
@@ -159,48 +196,42 @@ dhcp_proxy_server_add (fib_protocol_t proto,
                        u32 server_table_id)
 {
   dhcp_proxy_main_t * dpm = &dhcp_proxy_main;
-  dhcp_server_t * server = 0;
+  dhcp_proxy_t * proxy = 0;
   int new = 0;
 
-  server = dhcp_get_server(dpm, rx_fib_index, proto);
+  proxy = dhcp_get_proxy(dpm, rx_fib_index, proto);
 
-  if (NULL == server)
+  if (NULL == proxy)
   {
       vec_validate_init_empty(dpm->dhcp_server_index_by_rx_fib_index[proto],
                               rx_fib_index,
                               ~0);
 
-      pool_get (dpm->dhcp_servers[proto], server);
-      memset (server, 0, sizeof (*server));
+      pool_get (dpm->dhcp_servers[proto], proxy);
+      memset (proxy, 0, sizeof (*proxy));
       new = 1;
 
       dpm->dhcp_server_index_by_rx_fib_index[proto][rx_fib_index] =
-          server - dpm->dhcp_servers[proto];
+          proxy - dpm->dhcp_servers[proto];
 
-      server->rx_fib_index = rx_fib_index;
-      server->server_fib_index = 
-          fib_table_find_or_create_and_lock(proto, server_table_id);
+      proxy->dhcp_src_address = *src_address;
+      proxy->rx_fib_index = rx_fib_index;
   }
   else
   {
-      /* modify, may need to swap server FIBs */
-      u32 tmp_index;
-
-      tmp_index = fib_table_find(proto, server_table_id);
-
-      if (tmp_index != server->server_fib_index)
+      if (~0 != dhcp_proxy_server_find(proxy, proto, addr, server_table_id))
       {
-          tmp_index = server->server_fib_index;
-
-          /* certainly swapping if the fib doesn't exist */
-          server->server_fib_index = 
-              fib_table_find_or_create_and_lock(proto, server_table_id);
-          fib_table_unlock (tmp_index, proto);
+          return (new);
       }
   }
 
-  server->dhcp_server = *addr;
-  server->dhcp_src_address = *src_address;
+  dhcp_server_t server = {
+      .dhcp_server = *addr,
+      .server_fib_index = fib_table_find_or_create_and_lock(proto,
+                                                            server_table_id),
+  };
+
+  vec_add1(proxy->dhcp_servers, server);
 
   return (new);
 }
@@ -213,31 +244,15 @@ typedef struct dhcp4_proxy_dump_walk_ctx_t_
 } dhcp_proxy_dump_walk_cxt_t;
 
 static int
-dhcp_proxy_dump_walk (dhcp_server_t *server,
+dhcp_proxy_dump_walk (dhcp_proxy_t *proxy,
                       void *arg)
 {
   dhcp_proxy_dump_walk_cxt_t *ctx = arg;
-  fib_table_t *s_fib;
-  u32 rx_table_id;
-  dhcp_vss_t *v;
-
-  v = dhcp_get_vss_info(&dhcp_proxy_main,
-                        server->rx_fib_index,
-                        ctx->proto);
-
-  s_fib = fib_table_get(server->server_fib_index, ctx->proto);
-  rx_table_id = dhcp_proxy_rx_table_get_table_id(server->rx_fib_index,
-                                                 ctx->proto);
 
   dhcp_send_details(ctx->proto,
                     ctx->opaque,
                     ctx->context,
-                    &server->dhcp_server,
-                    &server->dhcp_src_address,
-                    s_fib->ft_table_id,
-                    rx_table_id,
-                    (v ? v->fib_id : 0),
-                    (v ? v->oui : 0));
+                    proxy);
 
   return (1);
 }
diff --git a/src/vnet/dhcp/dhcp_proxy.h b/src/vnet/dhcp/dhcp_proxy.h
index 708e92f3..ef2bc0a1 100644
--- a/src/vnet/dhcp/dhcp_proxy.h
+++ b/src/vnet/dhcp/dhcp_proxy.h
@@ -58,32 +58,58 @@ typedef struct dhcp_vss_t_ {
 } dhcp_vss_t;
 
 /**
- * @brief A DHCP proxy server represenation
+ * @brief A representation of a single DHCP Server within a given VRF config
  */
-typedef struct dhcp_server_t_ {
+typedef struct dhcp_server_t_
+{
     /**
      * @brief The address of the DHCP server to which to relay the client's
      *        messages
      */
     ip46_address_t dhcp_server;
 
-    /**
-     * @brief The source address to use in relayed messaes
-     */
-    ip46_address_t dhcp_src_address;
-
     /**
      * @brief The FIB index (not the external Table-ID) in which the server
      *        is reachable.
      */
     u32 server_fib_index;
+} dhcp_server_t;
+
+/**
+ * @brief A DHCP proxy represenation fpr per-client VRF config
+ */
+typedef struct dhcp_proxy_t_ {
+    /**
+     * @brief The set of DHCP servers to which messages are relayed.
+     *  If multiple servers are configured then discover/solict messages
+     * are relayed to each. A cookie is maintained for the relay, and only
+     * one message is replayed to the client, based on the presence of the
+     * cookie.
+     * The expectation is there are only 1 or 2 servers, hence no fancy DB.
+     */
+    dhcp_server_t *dhcp_servers;
+
+    /**
+     * @brief Hash table of pending requets key'd on the clients MAC address
+     */
+    uword *dhcp_pending;
+
+    /**
+     * @brief A lock for the pending request DB.
+     */
+    int lock;
+
+    /**
+     * @brief The source address to use in relayed messaes
+     */
+    ip46_address_t dhcp_src_address;
 
     /**
      * @brief The FIB index (not the external Table-ID) in which the client
      *        is resides.
      */
     u32 rx_fib_index;
-} dhcp_server_t;
+} dhcp_proxy_t;
 
 #define DHCP_N_PROTOS (FIB_PROTOCOL_IP6 + 1)
 
@@ -92,7 +118,7 @@ typedef struct dhcp_server_t_ {
  */
 typedef struct {
   /* Pool of DHCP servers */
-  dhcp_server_t *dhcp_servers[DHCP_N_PROTOS];
+  dhcp_proxy_t *dhcp_servers[DHCP_N_PROTOS];
 
   /* Pool of selected DHCP server. Zero is the default server */
   u32 * dhcp_server_index_by_rx_fib_index[DHCP_N_PROTOS];
@@ -114,12 +140,7 @@ extern dhcp_proxy_main_t dhcp_proxy_main;
 void dhcp_send_details (fib_protocol_t proto,
                         void *opaque,
                         u32 context,
-                        const ip46_address_t *server,
-                        const ip46_address_t *src,
-                        u32 server_fib_id,
-                        u32 rx_fib_id,
-                        u32 vss_fib_id,
-                        u32 vss_oui);
+                        dhcp_proxy_t *proxy);
 
 /**
  * @brief Show (on CLI) a VSS config during a show walk
@@ -157,16 +178,22 @@ int dhcp_proxy_server_add(fib_protocol_t proto,
 
 /**
  * @brief Delete a DHCP proxy config
- * @return 0 is deleted, otherwise an error code
+ * @return 1 if the proxy is deleted, 0 otherwise
  */
 int dhcp_proxy_server_del(fib_protocol_t proto,
-                          u32 rx_fib_index);
+                          u32 rx_fib_index,
+                          ip46_address_t *addr,
+                          u32 server_table_id);
+
+u32
+dhcp_proxy_rx_table_get_table_id (fib_protocol_t proto,
+                                  u32 fib_index);
 
 /**
  * @brief Callback function invoked for each DHCP proxy entry
  *  return 0 to break the walk, non-zero otherwise.
  */
-typedef int (*dhcp_proxy_walk_fn_t)(dhcp_server_t *server,
+typedef int (*dhcp_proxy_walk_fn_t)(dhcp_proxy_t *server,
                                     void *ctx);
 
 /**
@@ -191,6 +218,18 @@ void dhcp_vss_walk(fib_protocol_t proto,
                    dhcp_vss_walk_fn_t fn,
                    void *ctx);
 
+/**
+ * @brief Lock a proxy object to prevent simultaneous access of its
+ *  pending store
+ */
+void dhcp_proxy_lock (dhcp_proxy_t *server);
+
+/**
+ * @brief Lock a proxy object to prevent simultaneous access of its
+ *  pending store
+ */
+void dhcp_proxy_unlock (dhcp_proxy_t *server);
+
 /**
  * @brief Get the VSS data for the FIB index
  */
@@ -215,12 +254,12 @@ dhcp_get_vss_info (dhcp_proxy_main_t *dm,
 /**
  * @brief Get the DHCP proxy server data for the FIB index
  */
-static inline dhcp_server_t *
-dhcp_get_server (dhcp_proxy_main_t *dm,
-                 u32 rx_fib_index,
-                 fib_protocol_t proto)
+static inline dhcp_proxy_t *
+dhcp_get_proxy (dhcp_proxy_main_t *dm,
+                u32 rx_fib_index,
+                fib_protocol_t proto)
 {
-  dhcp_server_t *s = NULL;
+  dhcp_proxy_t *s = NULL;
 
   if (vec_len(dm->dhcp_server_index_by_rx_fib_index[proto]) > rx_fib_index &&
       dm->dhcp_server_index_by_rx_fib_index[proto][rx_fib_index] != ~0)
diff --git a/src/vnet/ip/ip6_packet.h b/src/vnet/ip/ip6_packet.h
index 6eabeef1..9bf19edb 100644
--- a/src/vnet/ip/ip6_packet.h
+++ b/src/vnet/ip/ip6_packet.h
@@ -79,6 +79,8 @@ typedef CLIB_PACKED (union {
 #define ip46_address_reset(ip46)	((ip46)->as_u64[0] = (ip46)->as_u64[1] = 0)
 #define ip46_address_cmp(ip46_1, ip46_2) (memcmp(ip46_1, ip46_2, sizeof(*ip46_1)))
 #define ip46_address_is_zero(ip46)	(((ip46)->as_u64[0] == 0) && ((ip46)->as_u64[1] == 0))
+#define ip46_address_is_equal(a1, a2)	(((a1)->as_u64[0] == (a2)->as_u64[0]) \
+                                         && ((a1)->as_u64[1] == (a2)->as_u64[1]))
 
 always_inline void
 ip46_from_addr_buf (u32 is_ipv6, u8 * buf, ip46_address_t * ip)
diff --git a/src/vnet/pg/input.c b/src/vnet/pg/input.c
index 4a65b024..2649798b 100644
--- a/src/vnet/pg/input.c
+++ b/src/vnet/pg/input.c
@@ -1373,6 +1373,7 @@ typedef struct
   u32 stream_index;
 
   u32 packet_length;
+  u32 sw_if_index;
 
   /* Use pre data for packet data. */
   vlib_buffer_t buffer;
@@ -1399,6 +1400,7 @@ format_pg_input_trace (u8 * s, va_list * va)
     s = format (s, "stream %d", t->stream_index);
 
   s = format (s, ", %d bytes", t->packet_length);
+  s = format (s, ", %d sw_if_index", t->sw_if_index);
 
   s = format (s, "\n%U%U",
 	      format_white_space, indent, format_vlib_buffer, &t->buffer);
@@ -1458,6 +1460,9 @@ pg_input_trace (pg_main_t * pg,
       t0->packet_length = vlib_buffer_length_in_chain (vm, b0);
       t1->packet_length = vlib_buffer_length_in_chain (vm, b1);
 
+      t0->sw_if_index = vnet_buffer (b0)->sw_if_index[VLIB_RX];
+      t1->sw_if_index = vnet_buffer (b1)->sw_if_index[VLIB_RX];
+
       clib_memcpy (&t0->buffer, b0, sizeof (b0[0]) - sizeof (b0->pre_data));
       clib_memcpy (&t1->buffer, b1, sizeof (b1[0]) - sizeof (b1->pre_data));
 
@@ -1484,6 +1489,7 @@ pg_input_trace (pg_main_t * pg,
 
       t0->stream_index = stream_index;
       t0->packet_length = vlib_buffer_length_in_chain (vm, b0);
+      t0->sw_if_index = vnet_buffer (b0)->sw_if_index[VLIB_RX];
       clib_memcpy (&t0->buffer, b0, sizeof (b0[0]) - sizeof (b0->pre_data));
       clib_memcpy (t0->buffer.pre_data, b0->data,
 		   sizeof (t0->buffer.pre_data));
diff --git a/test/test_dhcp.py b/test/test_dhcp.py
index a09c9bdb..89667d3d 100644
--- a/test/test_dhcp.py
+++ b/test/test_dhcp.py
@@ -2,8 +2,10 @@
 
 import unittest
 import socket
+import struct
 
 from framework import VppTestCase, VppTestRunner
+from vpp_neighbor import VppNeighbor
 
 from scapy.layers.l2 import Ether, getmacbyip
 from scapy.layers.inet import IP, UDP, ICMP
@@ -11,7 +13,7 @@ from scapy.layers.inet6 import IPv6, in6_getnsmac, in6_mactoifaceid
 from scapy.layers.dhcp import DHCP, BOOTP, DHCPTypes
 from scapy.layers.dhcp6 import DHCP6, DHCP6_Solicit, DHCP6_RelayForward, \
     DHCP6_RelayReply, DHCP6_Advertise, DHCP6OptRelayMsg, DHCP6OptIfaceId, \
-    DHCP6OptStatusCode, DHCP6OptVSS, DHCP6OptClientLinkLayerAddr
+    DHCP6OptStatusCode, DHCP6OptVSS, DHCP6OptClientLinkLayerAddr, DHCP6_Request
 from socket import AF_INET, AF_INET6
 from scapy.utils import inet_pton, inet_ntop
 from scapy.utils6 import in6_ptop
@@ -140,7 +142,7 @@ class TestDHCP(VppTestCase):
 
         return data
 
-    def verify_dhcp_offer(self, pkt, intf):
+    def verify_dhcp_offer(self, pkt, intf, fib_id=0, oui=0):
         ether = pkt[Ether]
         self.assertEqual(ether.dst, "ff:ff:ff:ff:ff:ff")
         self.assertEqual(ether.src, intf.local_mac)
@@ -162,15 +164,22 @@ class TestDHCP(VppTestCase):
                     is_offer = True
         self.assertTrue(is_offer)
 
-        data = self.validate_relay_options(pkt, intf, intf.local_ip4, 0, 0)
+        data = self.validate_relay_options(pkt, intf, intf.local_ip4,
+                                           fib_id, oui)
+
+    def verify_dhcp_discover(self, pkt, intf, src_intf=None, fib_id=0, oui=0,
+                             dst_mac=None, dst_ip=None):
+        if not dst_mac:
+            dst_mac = intf.remote_mac
+        if not dst_ip:
+            dst_ip = intf.remote_ip4
 
-    def verify_dhcp_discover(self, pkt, intf, src_intf=None, fib_id=0, oui=0):
         ether = pkt[Ether]
-        self.assertEqual(ether.dst, intf.remote_mac)
+        self.assertEqual(ether.dst, dst_mac)
         self.assertEqual(ether.src, intf.local_mac)
 
         ip = pkt[IP]
-        self.assertEqual(ip.dst, intf.remote_ip4)
+        self.assertEqual(ip.dst, dst_ip)
         self.assertEqual(ip.src, intf.local_ip4)
 
         udp = pkt[UDP]
@@ -195,13 +204,20 @@ class TestDHCP(VppTestCase):
     def verify_dhcp6_solicit(self, pkt, intf,
                              peer_ip, peer_mac,
                              fib_id=0,
-                             oui=0):
+                             oui=0,
+                             dst_mac=None,
+                             dst_ip=None):
+        if not dst_mac:
+            dst_mac = intf.remote_mac
+        if not dst_ip:
+            dst_ip = in6_ptop(intf.remote_ip6)
+
         ether = pkt[Ether]
-        self.assertEqual(ether.dst, intf.remote_mac)
+        self.assertEqual(ether.dst, dst_mac)
         self.assertEqual(ether.src, intf.local_mac)
 
         ip = pkt[IPv6]
-        self.assertEqual(in6_ptop(ip.dst), in6_ptop(intf.remote_ip6))
+        self.assertEqual(in6_ptop(ip.dst), dst_ip)
         self.assertEqual(in6_ptop(ip.src), in6_ptop(intf.local_ip6))
 
         udp = pkt[UDP]
@@ -447,6 +463,128 @@ class TestDHCP(VppTestCase):
         self.pg_enable_capture(self.pg_interfaces)
         self.pg_start()
 
+        rx = self.pg1.get_capture(1)
+        rx = rx[0]
+        self.verify_dhcp_discover(rx, self.pg1, src_intf=self.pg3,
+                                  fib_id=1, oui=4)
+
+        #
+        # Add a second DHCP server in VRF 1
+        #  expect clients messages to be relay to both configured servers
+        #
+        self.pg1.generate_remote_hosts(2)
+        server_addr2 = socket.inet_pton(AF_INET, self.pg1.remote_hosts[1].ip4)
+
+        self.vapi.dhcp_proxy_config(server_addr2,
+                                    src_addr,
+                                    rx_table_id=1,
+                                    server_table_id=1,
+                                    is_add=1)
+
+        #
+        # We'll need an ARP entry for the server to send it packets
+        #
+        arp_entry = VppNeighbor(self,
+                                self.pg1.sw_if_index,
+                                self.pg1.remote_hosts[1].mac,
+                                self.pg1.remote_hosts[1].ip4)
+        arp_entry.add_vpp_config()
+
+        #
+        # Send a discover from the client. expect two relayed messages
+        # The frist packet is sent to the second server
+        # We're not enforcing that here, it's just the way it is.
+        #
+        self.pg3.add_stream(pkts_disc_vrf1)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg1.get_capture(2)
+
+        option_82 = self.verify_dhcp_discover(
+            rx[0], self.pg1,
+            src_intf=self.pg3,
+            dst_mac=self.pg1.remote_hosts[1].mac,
+            dst_ip=self.pg1.remote_hosts[1].ip4,
+            fib_id=1, oui=4)
+        self.verify_dhcp_discover(rx[1], self.pg1, src_intf=self.pg3,
+                                  fib_id=1, oui=4)
+
+        #
+        # Send both packets back. Client gets both.
+        #
+        p1 = (Ether(dst=self.pg1.local_mac, src=self.pg1.remote_mac) /
+              IP(src=self.pg1.remote_ip4, dst=self.pg1.local_ip4) /
+              UDP(sport=DHCP4_SERVER_PORT, dport=DHCP4_SERVER_PORT) /
+              BOOTP(op=1) /
+              DHCP(options=[('message-type', 'offer'),
+                            ('relay_agent_Information', option_82),
+                            ('end')]))
+        p2 = (Ether(dst=self.pg1.local_mac, src=self.pg1.remote_mac) /
+              IP(src=self.pg1.remote_hosts[1].ip4, dst=self.pg1.local_ip4) /
+              UDP(sport=DHCP4_SERVER_PORT, dport=DHCP4_SERVER_PORT) /
+              BOOTP(op=1) /
+              DHCP(options=[('message-type', 'offer'),
+                            ('relay_agent_Information', option_82),
+                            ('end')]))
+        pkts = [p1, p2]
+
+        self.pg1.add_stream(pkts)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg3.get_capture(2)
+
+        self.verify_dhcp_offer(rx[0], self.pg3, fib_id=1, oui=4)
+        self.verify_dhcp_offer(rx[1], self.pg3, fib_id=1, oui=4)
+
+        #
+        # Ensure offers from non-servers are dropeed
+        #
+        p2 = (Ether(dst=self.pg1.local_mac, src=self.pg1.remote_mac) /
+              IP(src="8.8.8.8", dst=self.pg1.local_ip4) /
+              UDP(sport=DHCP4_SERVER_PORT, dport=DHCP4_SERVER_PORT) /
+              BOOTP(op=1) /
+              DHCP(options=[('message-type', 'offer'),
+                            ('relay_agent_Information', option_82),
+                            ('end')]))
+        self.send_and_assert_no_replies(self.pg1, p2,
+                                        "DHCP offer from non-server")
+
+        #
+        # Ensure only the discover is sent to multiple servers
+        #
+        p_req_vrf1 = (Ether(dst="ff:ff:ff:ff:ff:ff",
+                            src=self.pg3.remote_mac) /
+                      IP(src="0.0.0.0", dst="255.255.255.255") /
+                      UDP(sport=DHCP4_CLIENT_PORT,
+                          dport=DHCP4_SERVER_PORT) /
+                      BOOTP(op=1) /
+                      DHCP(options=[('message-type', 'request'),
+                                    ('end')]))
+
+        self.pg3.add_stream(p_req_vrf1)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg1.get_capture(1)
+
+        #
+        # Remove the second DHCP server
+        #
+        self.vapi.dhcp_proxy_config(server_addr2,
+                                    src_addr,
+                                    rx_table_id=1,
+                                    server_table_id=1,
+                                    is_add=0)
+
+        #
+        # Test we can still relay with the first
+        #
+        self.pg3.add_stream(pkts_disc_vrf1)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
         rx = self.pg1.get_capture(1)
         rx = rx[0]
         self.verify_dhcp_discover(rx, self.pg1, src_intf=self.pg3,
@@ -472,7 +610,7 @@ class TestDHCP(VppTestCase):
         self.vapi.dhcp_proxy_config(server_addr,
                                     src_addr,
                                     rx_table_id=1,
-                                    server_table_id=11,
+                                    server_table_id=1,
                                     is_add=0)
 
         self.send_and_assert_no_replies(self.pg2, pkts_disc_vrf0,
@@ -500,18 +638,16 @@ class TestDHCP(VppTestCase):
                           UDP(sport=DHCP6_SERVER_PORT,
                               dport=DHCP6_CLIENT_PORT) /
                           DHCP6_Solicit())
-        pkts_solicit_vrf0 = [p_solicit_vrf0]
         p_solicit_vrf1 = (Ether(dst=dmac, src=self.pg3.remote_mac) /
                           IPv6(src=dhcp_solicit_src_vrf1,
                                dst=dhcp_solicit_dst) /
                           UDP(sport=DHCP6_SERVER_PORT,
                               dport=DHCP6_CLIENT_PORT) /
                           DHCP6_Solicit())
-        pkts_solicit_vrf1 = [p_solicit_vrf1]
 
-        self.send_and_assert_no_replies(self.pg2, pkts_solicit_vrf0,
+        self.send_and_assert_no_replies(self.pg2, p_solicit_vrf0,
                                         "DHCP with no configuration")
-        self.send_and_assert_no_replies(self.pg3, pkts_solicit_vrf1,
+        self.send_and_assert_no_replies(self.pg3, p_solicit_vrf1,
                                         "DHCP with no configuration")
 
         #
@@ -525,9 +661,9 @@ class TestDHCP(VppTestCase):
                                     server_table_id=0,
                                     is_ipv6=1)
 
-        self.send_and_assert_no_replies(self.pg2, pkts_solicit_vrf0,
+        self.send_and_assert_no_replies(self.pg2, p_solicit_vrf0,
                                         "DHCP with no configuration")
-        self.send_and_assert_no_replies(self.pg3, pkts_solicit_vrf1,
+        self.send_and_assert_no_replies(self.pg3, p_solicit_vrf1,
                                         "DHCP with no configuration")
 
         #
@@ -538,13 +674,13 @@ class TestDHCP(VppTestCase):
         #
         # Now the DHCP requests are relayed to the server
         #
-        self.pg2.add_stream(pkts_solicit_vrf0)
+        self.pg2.add_stream(p_solicit_vrf0)
         self.pg_enable_capture(self.pg_interfaces)
         self.pg_start()
 
         rx = self.pg0.get_capture(1)
-        rx = rx[0]
-        self.verify_dhcp6_solicit(rx, self.pg0,
+
+        self.verify_dhcp6_solicit(rx[0], self.pg0,
                                   dhcp_solicit_src_vrf0,
                                   self.pg2.remote_mac)
 
@@ -557,8 +693,7 @@ class TestDHCP(VppTestCase):
                       IPv6(dst=self.pg0.local_ip6, src=self.pg0.remote_ip6) /
                       UDP(sport=DHCP6_SERVER_PORT, dport=DHCP6_SERVER_PORT) /
                       DHCP6_Advertise())
-        pkts_adv_vrf0 = [p_adv_vrf0]
-        self.send_and_assert_no_replies(self.pg2, pkts_adv_vrf0,
+        self.send_and_assert_no_replies(self.pg2, p_adv_vrf0,
                                         "DHCP6 not a relay reply")
 
         # 2 - no relay message option
@@ -567,8 +702,7 @@ class TestDHCP(VppTestCase):
                       UDP(sport=DHCP6_SERVER_PORT, dport=DHCP6_SERVER_PORT) /
                       DHCP6_RelayReply() /
                       DHCP6_Advertise())
-        pkts_adv_vrf0 = [p_adv_vrf0]
-        self.send_and_assert_no_replies(self.pg2, pkts_adv_vrf0,
+        self.send_and_assert_no_replies(self.pg2, p_adv_vrf0,
                                         "DHCP not a relay message")
 
         # 3 - no circuit ID
@@ -578,8 +712,7 @@ class TestDHCP(VppTestCase):
                       DHCP6_RelayReply() /
                       DHCP6OptRelayMsg(optlen=0) /
                       DHCP6_Advertise())
-        pkts_adv_vrf0 = [p_adv_vrf0]
-        self.send_and_assert_no_replies(self.pg2, pkts_adv_vrf0,
+        self.send_and_assert_no_replies(self.pg2, p_adv_vrf0,
                                         "DHCP6 no circuit ID")
         # 4 - wrong circuit ID
         p_adv_vrf0 = (Ether(dst=self.pg0.local_mac, src=self.pg0.remote_mac) /
@@ -589,8 +722,7 @@ class TestDHCP(VppTestCase):
                       DHCP6OptIfaceId(optlen=4, ifaceid='\x00\x00\x00\x05') /
                       DHCP6OptRelayMsg(optlen=0) /
                       DHCP6_Advertise())
-        pkts_adv_vrf0 = [p_adv_vrf0]
-        self.send_and_assert_no_replies(self.pg2, pkts_adv_vrf0,
+        self.send_and_assert_no_replies(self.pg2, p_adv_vrf0,
                                         "DHCP6 wrong circuit ID")
 
         #
@@ -611,8 +743,8 @@ class TestDHCP(VppTestCase):
         self.pg_start()
 
         rx = self.pg2.get_capture(1)
-        rx = rx[0]
-        self.verify_dhcp6_advert(rx, self.pg2, "::")
+
+        self.verify_dhcp6_advert(rx[0], self.pg2, "::")
 
         #
         # Send the relay response (the advertisement)
@@ -632,8 +764,8 @@ class TestDHCP(VppTestCase):
         self.pg_start()
 
         rx = self.pg2.get_capture(1)
-        rx = rx[0]
-        self.verify_dhcp6_advert(rx, self.pg2, dhcp_solicit_src_vrf0)
+
+        self.verify_dhcp6_advert(rx[0], self.pg2, dhcp_solicit_src_vrf0)
 
         #
         # Add all the config for VRF 1
@@ -648,13 +780,13 @@ class TestDHCP(VppTestCase):
         #
         # VRF 1 solicit
         #
-        self.pg3.add_stream(pkts_solicit_vrf1)
+        self.pg3.add_stream(p_solicit_vrf1)
         self.pg_enable_capture(self.pg_interfaces)
         self.pg_start()
 
         rx = self.pg1.get_capture(1)
-        rx = rx[0]
-        self.verify_dhcp6_solicit(rx, self.pg1,
+
+        self.verify_dhcp6_solicit(rx[0], self.pg1,
                                   dhcp_solicit_src_vrf1,
                                   self.pg3.remote_mac)
 
@@ -676,21 +808,21 @@ class TestDHCP(VppTestCase):
         self.pg_start()
 
         rx = self.pg3.get_capture(1)
-        rx = rx[0]
-        self.verify_dhcp6_advert(rx, self.pg3, dhcp_solicit_src_vrf1)
+
+        self.verify_dhcp6_advert(rx[0], self.pg3, dhcp_solicit_src_vrf1)
 
         #
         # Add VSS config
         #  table=1, fib=id=1, oui=4
         self.vapi.dhcp_proxy_set_vss(1, 1, 4, is_ip6=1)
 
-        self.pg3.add_stream(pkts_solicit_vrf1)
+        self.pg3.add_stream(p_solicit_vrf1)
         self.pg_enable_capture(self.pg_interfaces)
         self.pg_start()
 
         rx = self.pg1.get_capture(1)
-        rx = rx[0]
-        self.verify_dhcp6_solicit(rx, self.pg1,
+
+        self.verify_dhcp6_solicit(rx[0], self.pg1,
                                   dhcp_solicit_src_vrf1,
                                   self.pg3.remote_mac,
                                   fib_id=1,
@@ -702,27 +834,163 @@ class TestDHCP(VppTestCase):
         #
         self.vapi.dhcp_proxy_set_vss(1, 1, 4, is_ip6=1, is_add=0)
 
-        self.pg3.add_stream(pkts_solicit_vrf1)
+        self.pg3.add_stream(p_solicit_vrf1)
         self.pg_enable_capture(self.pg_interfaces)
         self.pg_start()
 
         rx = self.pg1.get_capture(1)
-        rx = rx[0]
-        self.verify_dhcp6_solicit(rx, self.pg1,
+
+        self.verify_dhcp6_solicit(rx[0], self.pg1,
                                   dhcp_solicit_src_vrf1,
                                   self.pg3.remote_mac)
 
         #
-        # Cleanup
+        # Add a second DHCP server in VRF 1
+        #  expect clients messages to be relay to both configured servers
         #
-        self.vapi.dhcp_proxy_config(server_addr_vrf1,
+        self.pg1.generate_remote_hosts(2)
+        server_addr2 = socket.inet_pton(AF_INET6, self.pg1.remote_hosts[1].ip6)
+
+        self.vapi.dhcp_proxy_config(server_addr2,
+                                    src_addr_vrf1,
+                                    rx_table_id=1,
+                                    server_table_id=1,
+                                    is_ipv6=1)
+
+        #
+        # We'll need an ND entry for the server to send it packets
+        #
+        nd_entry = VppNeighbor(self,
+                               self.pg1.sw_if_index,
+                               self.pg1.remote_hosts[1].mac,
+                               self.pg1.remote_hosts[1].ip6,
+                               af=AF_INET6)
+        nd_entry.add_vpp_config()
+
+        #
+        # Send a discover from the client. expect two relayed messages
+        # The frist packet is sent to the second server
+        # We're not enforcing that here, it's just the way it is.
+        #
+        self.pg3.add_stream(p_solicit_vrf1)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg1.get_capture(2)
+
+        self.verify_dhcp6_solicit(rx[0], self.pg1,
+                                  dhcp_solicit_src_vrf1,
+                                  self.pg3.remote_mac)
+        self.verify_dhcp6_solicit(rx[1], self.pg1,
+                                  dhcp_solicit_src_vrf1,
+                                  self.pg3.remote_mac,
+                                  dst_mac=self.pg1.remote_hosts[1].mac,
+                                  dst_ip=self.pg1.remote_hosts[1].ip6)
+
+        #
+        # Send both packets back. Client gets both.
+        #
+        p1 = (Ether(dst=self.pg1.local_mac, src=self.pg1.remote_mac) /
+              IPv6(dst=self.pg1.local_ip6, src=self.pg1.remote_ip6) /
+              UDP(sport=DHCP6_SERVER_PORT, dport=DHCP6_SERVER_PORT) /
+              DHCP6_RelayReply(peeraddr=dhcp_solicit_src_vrf1) /
+              DHCP6OptIfaceId(optlen=4, ifaceid='\x00\x00\x00\x04') /
+              DHCP6OptRelayMsg(optlen=0) /
+              DHCP6_Advertise(trid=1) /
+              DHCP6OptStatusCode(statuscode=0))
+        p2 = (Ether(dst=self.pg1.local_mac, src=self.pg1.remote_hosts[1].mac) /
+              IPv6(dst=self.pg1.local_ip6, src=self.pg1._remote_hosts[1].ip6) /
+              UDP(sport=DHCP6_SERVER_PORT, dport=DHCP6_SERVER_PORT) /
+              DHCP6_RelayReply(peeraddr=dhcp_solicit_src_vrf1) /
+              DHCP6OptIfaceId(optlen=4, ifaceid='\x00\x00\x00\x04') /
+              DHCP6OptRelayMsg(optlen=0) /
+              DHCP6_Advertise(trid=1) /
+              DHCP6OptStatusCode(statuscode=0))
+
+        pkts = [p1, p2]
+
+        self.pg1.add_stream(pkts)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg3.get_capture(2)
+
+        self.verify_dhcp6_advert(rx[0], self.pg3, dhcp_solicit_src_vrf1)
+        self.verify_dhcp6_advert(rx[1], self.pg3, dhcp_solicit_src_vrf1)
+
+        #
+        # Ensure only solicit messages are duplicated
+        #
+        p_request_vrf1 = (Ether(dst=dmac, src=self.pg3.remote_mac) /
+                          IPv6(src=dhcp_solicit_src_vrf1,
+                               dst=dhcp_solicit_dst) /
+                          UDP(sport=DHCP6_SERVER_PORT,
+                              dport=DHCP6_CLIENT_PORT) /
+                          DHCP6_Request())
+
+        self.pg3.add_stream(p_request_vrf1)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg1.get_capture(1)
+
+        #
+        # Test we drop DHCP packets from addresses that are not configured as
+        # DHCP servers
+        #
+        p2 = (Ether(dst=self.pg1.local_mac, src=self.pg1.remote_hosts[1].mac) /
+              IPv6(dst=self.pg1.local_ip6, src="3001::1") /
+              UDP(sport=DHCP6_SERVER_PORT, dport=DHCP6_SERVER_PORT) /
+              DHCP6_RelayReply(peeraddr=dhcp_solicit_src_vrf1) /
+              DHCP6OptIfaceId(optlen=4, ifaceid='\x00\x00\x00\x04') /
+              DHCP6OptRelayMsg(optlen=0) /
+              DHCP6_Advertise(trid=1) /
+              DHCP6OptStatusCode(statuscode=0))
+        self.send_and_assert_no_replies(self.pg1, p2,
+                                        "DHCP6 not from server")
+
+        #
+        # Remove the second DHCP server
+        #
+        self.vapi.dhcp_proxy_config(server_addr2,
                                     src_addr_vrf1,
                                     rx_table_id=1,
                                     server_table_id=1,
                                     is_ipv6=1,
                                     is_add=0)
+
+        #
+        # Test we can still relay with the first
+        #
+        self.pg3.add_stream(p_solicit_vrf1)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg1.get_capture(1)
+
+        self.verify_dhcp6_solicit(rx[0], self.pg1,
+                                  dhcp_solicit_src_vrf1,
+                                  self.pg3.remote_mac)
+
+        #
+        # Cleanup
+        #
         self.vapi.dhcp_proxy_config(server_addr_vrf1,
                                     src_addr_vrf1,
+                                    rx_table_id=1,
+                                    server_table_id=1,
+                                    is_ipv6=1,
+                                    is_add=0)
+        self.vapi.dhcp_proxy_config(server_addr_vrf0,
+                                    src_addr_vrf0,
+                                    rx_table_id=0,
+                                    server_table_id=0,
+                                    is_ipv6=1,
+                                    is_add=0)
+
+        # duplicate delete
+        self.vapi.dhcp_proxy_config(server_addr_vrf0,
+                                    src_addr_vrf0,
                                     rx_table_id=0,
                                     server_table_id=0,
                                     is_ipv6=1,
-- 
cgit 1.2.3-korg


From 2a3ea49d5cc224ffb2cf02bacaf0c02ddae12b86 Mon Sep 17 00:00:00 2001
From: Neale Ranns <nranns@cisco.com>
Date: Wed, 19 Apr 2017 05:24:40 -0700
Subject: Learn IP6 link-local ND entries from NSs sourced from link-local
 address

Change-Id: I4c3ce4d58df7977490fc94991291422ea1e31ee3
Signed-off-by: Neale Ranns <nranns@cisco.com>
---
 src/vnet/ethernet/arp.c    |   3 ++
 src/vnet/ip/ip6_neighbor.c |   7 ++-
 test/test_dhcp.py          |   8 +---
 test/test_ip6.py           | 106 ++++++++++++++++++++++++++++++++++-----------
 test/util.py               |  21 ++++++++-
 test/vpp_interface.py      |  18 +++++++-
 test/vpp_neighbor.py       |   2 +-
 7 files changed, 126 insertions(+), 39 deletions(-)

(limited to 'test/test_dhcp.py')

diff --git a/src/vnet/ethernet/arp.c b/src/vnet/ethernet/arp.c
index 3e292e4d..31af4fba 100644
--- a/src/vnet/ethernet/arp.c
+++ b/src/vnet/ethernet/arp.c
@@ -584,6 +584,9 @@ vnet_arp_set_ip4_over_ethernet_internal (vnet_main_t * vnm,
 					     FIB_PROTOCOL_IP4, &pfx.fp_addr,
 					     e->sw_if_index, ~0, 1, NULL,
 					     FIB_ROUTE_PATH_FLAG_NONE);
+	}
+      else
+	{
 	  e->flags |= ETHERNET_ARP_IP4_ENTRY_FLAG_NO_FIB_ENTRY;
 	}
     }
diff --git a/src/vnet/ip/ip6_neighbor.c b/src/vnet/ip/ip6_neighbor.c
index 42edb79a..31182770 100644
--- a/src/vnet/ip/ip6_neighbor.c
+++ b/src/vnet/ip/ip6_neighbor.c
@@ -634,6 +634,9 @@ vnet_set_ip6_ethernet_neighbor (vlib_main_t * vm,
 					     FIB_PROTOCOL_IP6, &pfx.fp_addr,
 					     n->key.sw_if_index, ~0, 1, NULL,
 					     FIB_ROUTE_PATH_FLAG_NONE);
+	}
+      else
+	{
 	  n->flags |= IP6_NEIGHBOR_FLAG_NO_FIB_ENTRY;
 	}
     }
@@ -1027,7 +1030,7 @@ icmp6_neighbor_solicitation_or_advertisement (vlib_main_t * vm,
 
 	  /* If src address unspecified or link local, donot learn neighbor MAC */
 	  if (PREDICT_TRUE (error0 == ICMP6_ERROR_NONE && o0 != 0 &&
-			    !ip6_sadd_unspecified && !ip6_sadd_link_local))
+			    !ip6_sadd_unspecified))
 	    {
 	      ip6_neighbor_main_t *nm = &ip6_neighbor_main;
 	      if (nm->limit_neighbor_cache_size &&
@@ -1040,7 +1043,7 @@ icmp6_neighbor_solicitation_or_advertisement (vlib_main_t * vm,
 					      &h0->target_address,
 					      o0->ethernet_address,
 					      sizeof (o0->ethernet_address),
-					      0, 0);
+					      0, ip6_sadd_link_local);
 	    }
 
 	  if (is_solicitation && error0 == ICMP6_ERROR_NONE)
diff --git a/test/test_dhcp.py b/test/test_dhcp.py
index 89667d3d..03c749d3 100644
--- a/test/test_dhcp.py
+++ b/test/test_dhcp.py
@@ -6,6 +6,7 @@ import struct
 
 from framework import VppTestCase, VppTestRunner
 from vpp_neighbor import VppNeighbor
+from util import mk_ll_addr
 
 from scapy.layers.l2 import Ether, getmacbyip
 from scapy.layers.inet import IP, UDP, ICMP
@@ -24,13 +25,6 @@ DHCP6_CLIENT_PORT = 547
 DHCP6_SERVER_PORT = 546
 
 
-def mk_ll_addr(mac):
-
-    euid = in6_mactoifaceid(mac)
-    addr = "fe80::" + euid
-    return addr
-
-
 class TestDHCP(VppTestCase):
     """ DHCP Test Case """
 
diff --git a/test/test_ip6.py b/test/test_ip6.py
index a8e8d4de..3ba09230 100644
--- a/test/test_ip6.py
+++ b/test/test_ip6.py
@@ -77,7 +77,6 @@ class TestIPv6ND(VppTestCase):
     def send_and_expect_ra(self, intf, pkts, remark, dst_ip=None,
                            filter_out_fn=is_ipv6_misc):
         intf.add_stream(pkts)
-        self.pg0.add_stream(pkts)
         self.pg_enable_capture(self.pg_interfaces)
         self.pg_start()
         rx = intf.get_capture(1, filter_out_fn=filter_out_fn)
@@ -86,11 +85,25 @@ class TestIPv6ND(VppTestCase):
         rx = rx[0]
         self.validate_ra(intf, rx, dst_ip)
 
+    def send_and_expect_na(self, intf, pkts, remark, dst_ip=None,
+                           tgt_ip=None,
+                           filter_out_fn=is_ipv6_misc):
+        intf.add_stream(pkts)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+        rx = intf.get_capture(1, filter_out_fn=filter_out_fn)
+
+        self.assertEqual(len(rx), 1)
+        rx = rx[0]
+        self.validate_na(intf, rx, dst_ip, tgt_ip)
+
     def send_and_assert_no_replies(self, intf, pkts, remark):
         intf.add_stream(pkts)
         self.pg_enable_capture(self.pg_interfaces)
         self.pg_start()
-        intf.assert_nothing_captured(remark=remark)
+        for i in self.pg_interfaces:
+            i.get_capture(0)
+            i.assert_nothing_captured(remark=remark)
 
 
 class TestIPv6(TestIPv6ND):
@@ -376,6 +389,59 @@ class TestIPv6(TestIPv6ND):
                                     128,
                                     inet=AF_INET6))
 
+        #
+        # send an NS from a link local address to the interface's global
+        # address
+        #
+        p = (Ether(dst=in6_getnsmac(nsma), src=self.pg0.remote_mac) /
+             IPv6(dst=d, src=self.pg0._remote_hosts[2].ip6_ll) /
+             ICMPv6ND_NS(tgt=self.pg0.local_ip6) /
+             ICMPv6NDOptSrcLLAddr(lladdr=self.pg0.remote_mac))
+
+        self.send_and_expect_na(self.pg0, p,
+                                "NS from link-local",
+                                dst_ip=self.pg0._remote_hosts[2].ip6_ll,
+                                tgt_ip=self.pg0.local_ip6)
+
+        #
+        # we should have learned an ND entry for the peer's link-local
+        # but not inserted a route to it in the FIB
+        #
+        self.assertTrue(find_nbr(self,
+                                 self.pg0.sw_if_index,
+                                 self.pg0._remote_hosts[2].ip6_ll,
+                                 inet=AF_INET6))
+        self.assertFalse(find_route(self,
+                                    self.pg0._remote_hosts[2].ip6_ll,
+                                    128,
+                                    inet=AF_INET6))
+
+        #
+        # An NS to the router's own Link-local
+        #
+        p = (Ether(dst=in6_getnsmac(nsma), src=self.pg0.remote_mac) /
+             IPv6(dst=d, src=self.pg0._remote_hosts[3].ip6_ll) /
+             ICMPv6ND_NS(tgt=self.pg0.local_ip6_ll) /
+             ICMPv6NDOptSrcLLAddr(lladdr=self.pg0.remote_mac))
+
+        self.send_and_expect_na(self.pg0, p,
+                                "NS to/from link-local",
+                                dst_ip=self.pg0._remote_hosts[3].ip6_ll,
+                                tgt_ip=self.pg0.local_ip6_ll)
+
+        #
+        # we should have learned an ND entry for the peer's link-local
+        # but not inserted a route to it in the FIB
+        #
+        self.assertTrue(find_nbr(self,
+                                 self.pg0.sw_if_index,
+                                 self.pg0._remote_hosts[3].ip6_ll,
+                                 inet=AF_INET6))
+        self.assertFalse(find_route(self,
+                                    self.pg0._remote_hosts[3].ip6_ll,
+                                    128,
+                                    inet=AF_INET6))
+
     def validate_ra(self, intf, rx, dst_ip=None, mtu=9000, pi_opt=None):
         if not dst_ip:
             dst_ip = intf.remote_ip6
@@ -770,14 +836,10 @@ class IPv6NDProxyTest(TestIPv6ND):
         #
         # try that NS again. this time we expect an NA back
         #
-        self.pg1.add_stream(ns_pg1)
-        self.pg_enable_capture(self.pg_interfaces)
-        self.pg_start()
-        rx = self.pg1.get_capture(1)
-
-        self.validate_na(self.pg1, rx[0],
-                         dst_ip=self.pg0._remote_hosts[2].ip6,
-                         tgt_ip=self.pg0.local_ip6)
+        self.send_and_expect_na(self.pg1, ns_pg1,
+                                "NS to proxy entry",
+                                dst_ip=self.pg0._remote_hosts[2].ip6,
+                                tgt_ip=self.pg0.local_ip6)
 
         #
         # ... and that we have an entry in the ND cache
@@ -816,14 +878,10 @@ class IPv6NDProxyTest(TestIPv6ND):
                   ICMPv6ND_NS(tgt=self.pg0._remote_hosts[2].ip6) /
                   ICMPv6NDOptSrcLLAddr(lladdr=self.pg0.remote_mac))
 
-        self.pg0.add_stream(ns_pg0)
-        self.pg_enable_capture(self.pg_interfaces)
-        self.pg_start()
-        rx = self.pg0.get_capture(1)
-
-        self.validate_na(self.pg0, rx[0],
-                         tgt_ip=self.pg0._remote_hosts[2].ip6,
-                         dst_ip=self.pg0.remote_ip6)
+        self.send_and_expect_na(self.pg0, ns_pg0,
+                                "NS to proxy entry on main",
+                                tgt_ip=self.pg0._remote_hosts[2].ip6,
+                                dst_ip=self.pg0.remote_ip6)
 
         #
         # Setup and resolve proxy for another host on another interface
@@ -837,14 +895,10 @@ class IPv6NDProxyTest(TestIPv6ND):
             inet_pton(AF_INET6, self.pg0._remote_hosts[3].ip6),
             self.pg2.sw_if_index)
 
-        self.pg2.add_stream(ns_pg2)
-        self.pg_enable_capture(self.pg_interfaces)
-        self.pg_start()
-        rx = self.pg2.get_capture(1)
-
-        self.validate_na(self.pg2, rx[0],
-                         dst_ip=self.pg0._remote_hosts[3].ip6,
-                         tgt_ip=self.pg0.local_ip6)
+        self.send_and_expect_na(self.pg2, ns_pg2,
+                                "NS to proxy entry other interface",
+                                dst_ip=self.pg0._remote_hosts[3].ip6,
+                                tgt_ip=self.pg0.local_ip6)
 
         self.assertTrue(find_nbr(self,
                                  self.pg2.sw_if_index,
diff --git a/test/util.py b/test/util.py
index d6b77f9d..aeba2ab4 100644
--- a/test/util.py
+++ b/test/util.py
@@ -4,6 +4,7 @@ import socket
 import sys
 from abc import abstractmethod, ABCMeta
 from cStringIO import StringIO
+from scapy.layers.inet6 import in6_mactoifaceid
 
 
 def ppp(headline, packet):
@@ -52,6 +53,12 @@ def mactobinary(mac):
     return mac.replace(':', '').decode('hex')
 
 
+def mk_ll_addr(mac):
+    euid = in6_mactoifaceid(mac)
+    addr = "fe80::" + euid
+    return addr
+
+
 class NumericConstant(object):
     __metaclass__ = ABCMeta
 
@@ -101,10 +108,22 @@ class Host(object):
         """ IPv6 address of remote host - raw, suitable as API parameter."""
         return socket.inet_pton(socket.AF_INET6, self._ip6)
 
-    def __init__(self, mac=None, ip4=None, ip6=None):
+    @property
+    def ip6_ll(self):
+        """ IPv6 link-local address - string """
+        return self._ip6_ll
+
+    @property
+    def ip6n_ll(self):
+        """ IPv6 link-local address of remote host -
+        raw, suitable as API parameter."""
+        return socket.inet_pton(socket.AF_INET6, self._ip6_ll)
+
+    def __init__(self, mac=None, ip4=None, ip6=None, ip6_ll=None):
         self._mac = mac
         self._ip4 = ip4
         self._ip6 = ip6
+        self._ip6_ll = ip6_ll
 
 
 class ForeignAddressFactory(object):
diff --git a/test/vpp_interface.py b/test/vpp_interface.py
index 5dba0978..662015ea 100644
--- a/test/vpp_interface.py
+++ b/test/vpp_interface.py
@@ -1,7 +1,7 @@
 from abc import abstractmethod, ABCMeta
 import socket
 
-from util import Host
+from util import Host, mk_ll_addr
 from vpp_neighbor import VppNeighbor
 
 
@@ -54,6 +54,16 @@ class VppInterface(object):
         """Local IPv6 address - raw, suitable as API parameter."""
         return socket.inet_pton(socket.AF_INET6, self.local_ip6)
 
+    @property
+    def local_ip6_ll(self):
+        """Local IPv6 linnk-local address on VPP interface (string)."""
+        return self._local_ip6_ll
+
+    @property
+    def local_ip6n_ll(self):
+        """Local IPv6 link-local address - raw, suitable as API parameter."""
+        return self.local_ip6n_ll
+
     @property
     def remote_ip6(self):
         """IPv6 address of remote peer "connected" to this interface."""
@@ -133,7 +143,8 @@ class VppInterface(object):
             mac = "02:%02x:00:00:ff:%02x" % (self.sw_if_index, i)
             ip4 = "172.16.%u.%u" % (self.sw_if_index, i)
             ip6 = "fd01:%x::%x" % (self.sw_if_index, i)
-            host = Host(mac, ip4, ip6)
+            ip6_ll = mk_ll_addr(mac)
+            host = Host(mac, ip4, ip6, ip6_ll)
             self._remote_hosts.append(host)
             self._hosts_by_mac[mac] = host
             self._hosts_by_ip4[ip4] = host
@@ -176,6 +187,9 @@ class VppInterface(object):
                 "Could not find interface with sw_if_index %d "
                 "in interface dump %s" %
                 (self.sw_if_index, repr(r)))
+        self._local_ip6_ll = mk_ll_addr(self.local_mac)
+        self._local_ip6n_ll = socket.inet_pton(socket.AF_INET6,
+                                               self.local_ip6_ll)
 
     def config_ip4(self):
         """Configure IPv4 address on the VPP interface."""
diff --git a/test/vpp_neighbor.py b/test/vpp_neighbor.py
index 6968b5f6..5919cf8e 100644
--- a/test/vpp_neighbor.py
+++ b/test/vpp_neighbor.py
@@ -31,7 +31,7 @@ class VppNeighbor(VppObject):
     """
 
     def __init__(self, test, sw_if_index, mac_addr, nbr_addr,
-                 af=AF_INET, is_static=False, is_no_fib_entry=False):
+                 af=AF_INET, is_static=False, is_no_fib_entry=0):
         self._test = test
         self.sw_if_index = sw_if_index
         self.mac_addr = mactobinary(mac_addr)
-- 
cgit 1.2.3-korg


From a2fbf6ba0e2553687f56d9a9fb63e2972bdb26c6 Mon Sep 17 00:00:00 2001
From: Neale Ranns <nranns@cisco.com>
Date: Tue, 18 Jul 2017 08:23:32 -0700
Subject: DHCP client - remove interface address when DHCP de-configured

Change-Id: I63c59e3c13859b51999d283774f7783ef0a6a5ed
Signed-off-by: Neale Ranns <nranns@cisco.com>
---
 src/vnet/dhcp/client.c    |   1 +
 test/test_dhcp.py         | 158 ++++++++++++++++++++++++++++++++++++++++------
 test/vpp_papi_provider.py |  15 +++++
 3 files changed, 153 insertions(+), 21 deletions(-)

(limited to 'test/test_dhcp.py')

diff --git a/src/vnet/dhcp/client.c b/src/vnet/dhcp/client.c
index 014f17a1..8f033d25 100644
--- a/src/vnet/dhcp/client.c
+++ b/src/vnet/dhcp/client.c
@@ -821,6 +821,7 @@ int dhcp_client_add_del (dhcp_client_add_del_args_t * a)
 				      1,
 				      FIB_ROUTE_PATH_FLAG_NONE);
       }
+      dhcp_client_release_address (dcm, c);
       ip4_sw_interface_enable_disable (c->sw_if_index, 0);
 
       vec_free (c->option_55_data);
diff --git a/test/test_dhcp.py b/test/test_dhcp.py
index 03c749d3..1700f6ba 100644
--- a/test/test_dhcp.py
+++ b/test/test_dhcp.py
@@ -6,6 +6,7 @@ import struct
 
 from framework import VppTestCase, VppTestRunner
 from vpp_neighbor import VppNeighbor
+from vpp_ip_route import find_route
 from util import mk_ll_addr
 
 from scapy.layers.l2 import Ether, getmacbyip
@@ -68,6 +69,18 @@ class TestDHCP(VppTestCase):
         for i in self.pg_interfaces:
             i.assert_nothing_captured(remark=remark)
 
+    def verify_dhcp_has_option(self, pkt, option, value):
+        dhcp = pkt[DHCP]
+        found = False
+
+        for i in dhcp.options:
+            if type(i) is tuple:
+                if i[0] == option:
+                    self.assertEqual(i[1], value)
+                    found = True
+
+        self.assertTrue(found)
+
     def validate_relay_options(self, pkt, intf, ip_addr, fib_id, oui):
         dhcp = pkt[DHCP]
         found = 0
@@ -136,6 +149,16 @@ class TestDHCP(VppTestCase):
 
         return data
 
+    def verify_dhcp_msg_type(self, pkt, name):
+        dhcp = pkt[DHCP]
+        found = False
+        for o in dhcp.options:
+            if type(o) is tuple:
+                if o[0] == "message-type" \
+                   and DHCPTypes[o[1]] == name:
+                    found = True
+        self.assertTrue(found)
+
     def verify_dhcp_offer(self, pkt, intf, fib_id=0, oui=0):
         ether = pkt[Ether]
         self.assertEqual(ether.dst, "ff:ff:ff:ff:ff:ff")
@@ -149,20 +172,39 @@ class TestDHCP(VppTestCase):
         self.assertEqual(udp.dport, DHCP4_CLIENT_PORT)
         self.assertEqual(udp.sport, DHCP4_SERVER_PORT)
 
-        dhcp = pkt[DHCP]
-        is_offer = False
-        for o in dhcp.options:
-            if type(o) is tuple:
-                if o[0] == "message-type" \
-                   and DHCPTypes[o[1]] == "offer":
-                    is_offer = True
-        self.assertTrue(is_offer)
-
+        self.verify_dhcp_msg_type(pkt, "offer")
         data = self.validate_relay_options(pkt, intf, intf.local_ip4,
                                            fib_id, oui)
 
-    def verify_dhcp_discover(self, pkt, intf, src_intf=None, fib_id=0, oui=0,
-                             dst_mac=None, dst_ip=None):
+    def verify_orig_dhcp_pkt(self, pkt, intf):
+        ether = pkt[Ether]
+        self.assertEqual(ether.dst, "ff:ff:ff:ff:ff:ff")
+        self.assertEqual(ether.src, intf.local_mac)
+
+        ip = pkt[IP]
+        self.assertEqual(ip.dst, "255.255.255.255")
+        self.assertEqual(ip.src, "0.0.0.0")
+
+        udp = pkt[UDP]
+        self.assertEqual(udp.dport, DHCP4_SERVER_PORT)
+        self.assertEqual(udp.sport, DHCP4_CLIENT_PORT)
+
+    def verify_orig_dhcp_discover(self, pkt, intf, hostname):
+        self.verify_orig_dhcp_pkt(pkt, intf)
+
+        self.verify_dhcp_msg_type(pkt, "discover")
+        self.verify_dhcp_has_option(pkt, "hostname", hostname)
+
+    def verify_orig_dhcp_request(self, pkt, intf, hostname, ip):
+        self.verify_orig_dhcp_pkt(pkt, intf)
+
+        self.verify_dhcp_msg_type(pkt, "request")
+        self.verify_dhcp_has_option(pkt, "hostname", hostname)
+        self.verify_dhcp_has_option(pkt, "requested_addr", ip)
+
+    def verify_relayed_dhcp_discover(self, pkt, intf, src_intf=None,
+                                     fib_id=0, oui=0,
+                                     dst_mac=None, dst_ip=None):
         if not dst_mac:
             dst_mac = intf.remote_mac
         if not dst_ip:
@@ -341,7 +383,8 @@ class TestDHCP(VppTestCase):
         rx = self.pg0.get_capture(1)
         rx = rx[0]
 
-        option_82 = self.verify_dhcp_discover(rx, self.pg0, src_intf=self.pg2)
+        option_82 = self.verify_relayed_dhcp_discover(rx, self.pg0,
+                                                      src_intf=self.pg2)
 
         #
         # Create an DHCP offer reply from the server with a correctly formatted
@@ -446,7 +489,7 @@ class TestDHCP(VppTestCase):
 
         rx = self.pg1.get_capture(1)
         rx = rx[0]
-        self.verify_dhcp_discover(rx, self.pg1, src_intf=self.pg3)
+        self.verify_relayed_dhcp_discover(rx, self.pg1, src_intf=self.pg3)
 
         #
         # Add VSS config
@@ -459,8 +502,9 @@ class TestDHCP(VppTestCase):
 
         rx = self.pg1.get_capture(1)
         rx = rx[0]
-        self.verify_dhcp_discover(rx, self.pg1, src_intf=self.pg3,
-                                  fib_id=1, oui=4)
+        self.verify_relayed_dhcp_discover(rx, self.pg1,
+                                          src_intf=self.pg3,
+                                          fib_id=1, oui=4)
 
         #
         # Add a second DHCP server in VRF 1
@@ -495,14 +539,15 @@ class TestDHCP(VppTestCase):
 
         rx = self.pg1.get_capture(2)
 
-        option_82 = self.verify_dhcp_discover(
+        option_82 = self.verify_relayed_dhcp_discover(
             rx[0], self.pg1,
             src_intf=self.pg3,
             dst_mac=self.pg1.remote_hosts[1].mac,
             dst_ip=self.pg1.remote_hosts[1].ip4,
             fib_id=1, oui=4)
-        self.verify_dhcp_discover(rx[1], self.pg1, src_intf=self.pg3,
-                                  fib_id=1, oui=4)
+        self.verify_relayed_dhcp_discover(rx[1], self.pg1,
+                                          src_intf=self.pg3,
+                                          fib_id=1, oui=4)
 
         #
         # Send both packets back. Client gets both.
@@ -581,8 +626,9 @@ class TestDHCP(VppTestCase):
 
         rx = self.pg1.get_capture(1)
         rx = rx[0]
-        self.verify_dhcp_discover(rx, self.pg1, src_intf=self.pg3,
-                                  fib_id=1, oui=4)
+        self.verify_relayed_dhcp_discover(rx, self.pg1,
+                                          src_intf=self.pg3,
+                                          fib_id=1, oui=4)
 
         #
         # Remove the VSS config
@@ -596,7 +642,7 @@ class TestDHCP(VppTestCase):
 
         rx = self.pg1.get_capture(1)
         rx = rx[0]
-        self.verify_dhcp_discover(rx, self.pg1, src_intf=self.pg3)
+        self.verify_relayed_dhcp_discover(rx, self.pg1, src_intf=self.pg3)
 
         #
         # remove DHCP config to cleanup
@@ -990,5 +1036,75 @@ class TestDHCP(VppTestCase):
                                     is_ipv6=1,
                                     is_add=0)
 
+    def test_dhcp_client(self):
+        """ DHCP Client"""
+
+        hostname = 'universal-dp'
+
+        self.pg_enable_capture(self.pg_interfaces)
+
+        #
+        # Configure DHCP client on PG2 and capture the discover sent
+        #
+        self.vapi.dhcp_client(self.pg2.sw_if_index, hostname)
+
+        rx = self.pg2.get_capture(1)
+
+        self.verify_orig_dhcp_discover(rx[0], self.pg2, hostname)
+
+        #
+        # Sned back on offer, expect the request
+        #
+        p = (Ether(dst=self.pg2.local_mac, src=self.pg2.remote_mac) /
+             IP(src=self.pg2.remote_ip4, dst="255.255.255.255") /
+             UDP(sport=DHCP4_SERVER_PORT, dport=DHCP4_CLIENT_PORT) /
+             BOOTP(op=1,
+                   yiaddr=self.pg2.local_ip4) /
+             DHCP(options=[('message-type', 'offer'), ('end')]))
+
+        self.pg2.add_stream(p)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg2.get_capture(1)
+        self.verify_orig_dhcp_request(rx[0], self.pg2, hostname,
+                                      self.pg2.local_ip4)
+
+        #
+        # Send an acknowloedgement
+        #
+        p = (Ether(dst=self.pg2.local_mac, src=self.pg2.remote_mac) /
+             IP(src=self.pg2.remote_ip4, dst="255.255.255.255") /
+             UDP(sport=DHCP4_SERVER_PORT, dport=DHCP4_CLIENT_PORT) /
+             BOOTP(op=1,
+                   yiaddr=self.pg2.local_ip4) /
+             DHCP(options=[('message-type', 'ack'),
+                           ('subnet_mask', "255.255.255.0"),
+                           ('router', self.pg2.remote_ip4),
+                           ('server_id', self.pg2.remote_ip4),
+                           ('lease_time', 43200),
+                           ('end')]))
+
+        self.pg2.add_stream(p)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        #
+        # At the end of this procedure there should be a connected route
+        # in the FIB
+        #
+        self.assertTrue(find_route(self, self.pg2.local_ip4, 32))
+
+        #
+        # remove the DHCP config
+        #
+        self.vapi.dhcp_client(self.pg2.sw_if_index, hostname, is_add=0)
+
+        #
+        # and now the route should be gone
+        #
+        self.assertFalse(find_route(self, self.pg2.local_ip4, 32))
+
+
 if __name__ == '__main__':
     unittest.main(testRunner=VppTestRunner)
diff --git a/test/vpp_papi_provider.py b/test/vpp_papi_provider.py
index 51c359e8..31eadad8 100644
--- a/test/vpp_papi_provider.py
+++ b/test/vpp_papi_provider.py
@@ -1755,6 +1755,21 @@ class VppPapiProvider(object):
                 'oui': oui,
             })
 
+    def dhcp_client(self,
+                    sw_if_index,
+                    hostname,
+                    is_add=1,
+                    want_dhcp_events=0):
+        return self.api(
+            self.papi.dhcp_client_config,
+            {
+                'sw_if_index': sw_if_index,
+                'hostname': hostname,
+                'is_add': is_add,
+                'want_dhcp_event': want_dhcp_events,
+                'pid': os.getpid(),
+            })
+
     def ip_mroute_add_del(self,
                           src_address,
                           grp_address,
-- 
cgit 1.2.3-korg


From 51822bf07a3f0fe72834ea94659faf6e262475ba Mon Sep 17 00:00:00 2001
From: Neale Ranns <nranns@cisco.com>
Date: Tue, 18 Jul 2017 09:26:53 -0700
Subject: DHCP client option 61 "client_id"

the existing seeting of client_id to a VPP version number was unused and so overridden

Change-Id: If9ebea936336f1fcca8d07e67186c95f8f8f0ccd
Signed-off-by: Neale Ranns <nranns@cisco.com>
---
 src/vnet/dhcp/client.c    | 15 ++++++++++++++-
 src/vnet/dhcp/client.h    |  1 +
 src/vnet/dhcp/dhcp.api    |  2 ++
 src/vnet/dhcp/dhcp_api.c  |  3 ++-
 test/test_dhcp.py         | 34 ++++++++++++++++++++++++++++++++--
 test/vpp_papi_provider.py |  2 ++
 6 files changed, 53 insertions(+), 4 deletions(-)

(limited to 'test/test_dhcp.py')

diff --git a/src/vnet/dhcp/client.c b/src/vnet/dhcp/client.c
index 8f033d25..25ab3176 100644
--- a/src/vnet/dhcp/client.c
+++ b/src/vnet/dhcp/client.c
@@ -414,6 +414,16 @@ send_dhcp_pkt (dhcp_client_main_t * dcm, dhcp_client_t * c,
       o = (dhcp_option_t *) (((uword) o) + (o->length + 2));
     }
 
+  /* send option 61, client_id */
+  if (vec_len (c->client_identifier))
+    {
+      o->option = 61;
+      o->length = vec_len (c->client_identifier);
+      clib_memcpy (o->data, c->client_identifier,
+                   vec_len (c->client_identifier));
+      o = (dhcp_option_t *) (((uword) o) + (o->length + 2));
+    }
+
   /* $$ maybe send the client s/w version if anyone cares */
 
   /* 
@@ -838,6 +848,7 @@ int
 dhcp_client_config (vlib_main_t * vm,
                     u32 sw_if_index,
                     u8 * hostname,
+                    u8 * client_id,
                     u32 is_add,
                     u32 client_index,
                     void * event_callback,
@@ -854,7 +865,9 @@ dhcp_client_config (vlib_main_t * vm,
   a->event_callback = event_callback;
   vec_validate(a->hostname, strlen((char *)hostname) - 1);
   strncpy((char *)a->hostname, (char *)hostname, vec_len(a->hostname));
-  a->client_identifier = format (0, "vpe 1.0%c", 0);
+  vec_validate(a->client_identifier, strlen((char *)client_id) - 1);
+  strncpy((char *)a->client_identifier, (char *)client_id, vec_len(a->client_identifier));
+
   /* 
    * Option 55 request list. These data precisely match
    * the Ubuntu dhcp client. YMMV.
diff --git a/src/vnet/dhcp/client.h b/src/vnet/dhcp/client.h
index 1f85d7ce..509d5d4c 100644
--- a/src/vnet/dhcp/client.h
+++ b/src/vnet/dhcp/client.h
@@ -113,6 +113,7 @@ int dhcp_client_for_us (u32 bi0,
 int dhcp_client_config (vlib_main_t * vm,
                         u32 sw_if_index,
                         u8 * hostname,
+                        u8 * client_id,
                         u32 is_add,
                         u32 client_index,
                         void *event_callback,
diff --git a/src/vnet/dhcp/dhcp.api b/src/vnet/dhcp/dhcp.api
index a2803728..c632c087 100644
--- a/src/vnet/dhcp/dhcp.api
+++ b/src/vnet/dhcp/dhcp.api
@@ -61,6 +61,7 @@ autoreply define dhcp_proxy_set_vss
     @param context - sender context, to match reply w/ request
     @param sw_if_index - index of the interface for DHCP client
     @param hostname - hostname
+    @param client_id - Client ID - option 61
     @param is_add - add the config if non-zero, else delete
     @param want_dhcp_event - DHCP event sent to the sender
            via dhcp_compl_event API message if non-zero
@@ -72,6 +73,7 @@ autoreply define dhcp_client_config
   u32 context;
   u32 sw_if_index;
   u8 hostname[64];
+  u8 client_id[64];
   u8 is_add;
   u8 want_dhcp_event;
   u32 pid;
diff --git a/src/vnet/dhcp/dhcp_api.c b/src/vnet/dhcp/dhcp_api.c
index 5ea93660..d6984f2d 100644
--- a/src/vnet/dhcp/dhcp_api.c
+++ b/src/vnet/dhcp/dhcp_api.c
@@ -227,7 +227,8 @@ static void vl_api_dhcp_client_config_t_handler
   VALIDATE_SW_IF_INDEX (mp);
 
   rv = dhcp_client_config (vm, ntohl (mp->sw_if_index),
-			   mp->hostname, mp->is_add, mp->client_index,
+			   mp->hostname, mp->client_id,
+			   mp->is_add, mp->client_index,
 			   mp->want_dhcp_event ? dhcp_compl_event_callback :
 			   NULL, mp->pid);
 
diff --git a/test/test_dhcp.py b/test/test_dhcp.py
index 1700f6ba..4e8ed4ce 100644
--- a/test/test_dhcp.py
+++ b/test/test_dhcp.py
@@ -9,7 +9,7 @@ from vpp_neighbor import VppNeighbor
 from vpp_ip_route import find_route
 from util import mk_ll_addr
 
-from scapy.layers.l2 import Ether, getmacbyip
+from scapy.layers.l2 import Ether, getmacbyip, ARP
 from scapy.layers.inet import IP, UDP, ICMP
 from scapy.layers.inet6 import IPv6, in6_getnsmac, in6_mactoifaceid
 from scapy.layers.dhcp import DHCP, BOOTP, DHCPTypes
@@ -189,11 +189,13 @@ class TestDHCP(VppTestCase):
         self.assertEqual(udp.dport, DHCP4_SERVER_PORT)
         self.assertEqual(udp.sport, DHCP4_CLIENT_PORT)
 
-    def verify_orig_dhcp_discover(self, pkt, intf, hostname):
+    def verify_orig_dhcp_discover(self, pkt, intf, hostname, client_id=None):
         self.verify_orig_dhcp_pkt(pkt, intf)
 
         self.verify_dhcp_msg_type(pkt, "discover")
         self.verify_dhcp_has_option(pkt, "hostname", hostname)
+        if client_id:
+            self.verify_dhcp_has_option(pkt, "client_id", client_id)
 
     def verify_orig_dhcp_request(self, pkt, intf, hostname, ip):
         self.verify_orig_dhcp_pkt(pkt, intf)
@@ -1089,12 +1091,25 @@ class TestDHCP(VppTestCase):
         self.pg_enable_capture(self.pg_interfaces)
         self.pg_start()
 
+        #
+        # We'll get an ARP request for the router address
+        #
+        rx = self.pg2.get_capture(1)
+
+        self.assertEqual(rx[0][ARP].pdst, self.pg2.remote_ip4)
+        self.pg_enable_capture(self.pg_interfaces)
+
         #
         # At the end of this procedure there should be a connected route
         # in the FIB
         #
         self.assertTrue(find_route(self, self.pg2.local_ip4, 32))
 
+        # remove the left over ARP entry
+        self.vapi.ip_neighbor_add_del(self.pg2.sw_if_index,
+                                      self.pg2.remote_mac,
+                                      self.pg2.remote_ip4,
+                                      is_add=0)
         #
         # remove the DHCP config
         #
@@ -1105,6 +1120,21 @@ class TestDHCP(VppTestCase):
         #
         self.assertFalse(find_route(self, self.pg2.local_ip4, 32))
 
+        #
+        # Start the procedure again. this time have VPP send the clientiid
+        #
+        self.vapi.dhcp_client(self.pg2.sw_if_index, hostname,
+                              client_id=self.pg2.local_mac)
+
+        rx = self.pg2.get_capture(1)
+
+        self.verify_orig_dhcp_discover(rx[0], self.pg2, hostname,
+                                       self.pg2.local_mac)
+
+        #
+        # remove the DHCP config
+        #
+        self.vapi.dhcp_client(self.pg2.sw_if_index, hostname, is_add=0)
 
 if __name__ == '__main__':
     unittest.main(testRunner=VppTestRunner)
diff --git a/test/vpp_papi_provider.py b/test/vpp_papi_provider.py
index 31eadad8..2814ef97 100644
--- a/test/vpp_papi_provider.py
+++ b/test/vpp_papi_provider.py
@@ -1758,6 +1758,7 @@ class VppPapiProvider(object):
     def dhcp_client(self,
                     sw_if_index,
                     hostname,
+                    client_id='',
                     is_add=1,
                     want_dhcp_events=0):
         return self.api(
@@ -1765,6 +1766,7 @@ class VppPapiProvider(object):
             {
                 'sw_if_index': sw_if_index,
                 'hostname': hostname,
+                'client_id': client_id,
                 'is_add': is_add,
                 'want_dhcp_event': want_dhcp_events,
                 'pid': os.getpid(),
-- 
cgit 1.2.3-korg


From 808c5b21c2759564689933d004223052b7895a42 Mon Sep 17 00:00:00 2001
From: Neale Ranns <nranns@cisco.com>
Date: Wed, 2 Aug 2017 05:15:07 -0700
Subject: DHCP Client: receive unicast ACKs

despite VPP DHCP client setting neither ciaddr nor giaddr and setting the broadcast bit (see RFC 2131 section 4.1) some DHCP servers will still send a unicast DCHPACK. So as not to drop this VPP must have both 1) a receive FIB entry for the OFFERED IP adress and 2) a 'don't drop me because of uRPF' FIB entry for the DHCP server's address.

Change-Id: I167d858deb45629318cbdccf5bf67d971730a42f
Signed-off-by: Neale Ranns <nranns@cisco.com>
---
 src/vnet/dhcp/client.c   | 79 ++++++++++++++++++++++++++++++++++++++++-
 src/vnet/fib/fib_entry.h |  9 ++---
 test/test_dhcp.py        | 92 +++++++++++++++++++++++++++++++++++++-----------
 3 files changed, 151 insertions(+), 29 deletions(-)

(limited to 'test/test_dhcp.py')

diff --git a/src/vnet/dhcp/client.c b/src/vnet/dhcp/client.c
index 25ab3176..cfe62a6f 100644
--- a/src/vnet/dhcp/client.c
+++ b/src/vnet/dhcp/client.c
@@ -22,6 +22,70 @@ static u8 * format_dhcp_client_state (u8 * s, va_list * va);
 static vlib_node_registration_t dhcp_client_process_node;
 
 static void 
+dhcp_client_add_rx_address (dhcp_client_main_t * dcm, dhcp_client_t * c)
+{
+  /* Install a local entry for the offered address */
+  fib_prefix_t rx =
+    {
+      .fp_len = 32,
+      .fp_addr.ip4 = c->leased_address,
+      .fp_proto = FIB_PROTOCOL_IP4,
+    };
+
+  fib_table_entry_special_add(fib_table_get_index_for_sw_if_index(
+                                  FIB_PROTOCOL_IP4,
+                                  c->sw_if_index),
+                              &rx,
+                              FIB_SOURCE_DHCP,
+                              (FIB_ENTRY_FLAG_LOCAL));
+
+  /* And add the server's address as uRPF exempt so we can accept
+   * local packets from it */
+  fib_prefix_t server =
+    {
+      .fp_len = 32,
+      .fp_addr.ip4 = c->dhcp_server,
+      .fp_proto = FIB_PROTOCOL_IP4,
+    };
+
+  fib_table_entry_special_add(fib_table_get_index_for_sw_if_index(
+                                  FIB_PROTOCOL_IP4,
+                                  c->sw_if_index),
+                              &server,
+                              FIB_SOURCE_URPF_EXEMPT,
+                              (FIB_ENTRY_FLAG_DROP));
+}
+
+static void
+dhcp_client_remove_rx_address (dhcp_client_main_t * dcm, dhcp_client_t * c)
+{
+  fib_prefix_t rx =
+    {
+      .fp_len = 32,
+      .fp_addr.ip4 = c->leased_address,
+      .fp_proto = FIB_PROTOCOL_IP4,
+    };
+
+  fib_table_entry_special_remove(fib_table_get_index_for_sw_if_index(
+                                     FIB_PROTOCOL_IP4,
+                                     c->sw_if_index),
+                                 &rx,
+                                 FIB_SOURCE_DHCP);
+  fib_prefix_t server =
+    {
+      .fp_len = 32,
+      .fp_addr.ip4 = c->dhcp_server,
+      .fp_proto = FIB_PROTOCOL_IP4,
+    };
+
+  fib_table_entry_special_remove(fib_table_get_index_for_sw_if_index(
+                                     FIB_PROTOCOL_IP4,
+                                     c->sw_if_index),
+                                 &server,
+                                 FIB_SOURCE_URPF_EXEMPT);
+}
+
+static void
 dhcp_client_acquire_address (dhcp_client_main_t * dcm, dhcp_client_t * c)
 {
   /* 
@@ -95,7 +159,9 @@ int dhcp_client_for_us (u32 bi, vlib_buffer_t * b,
   /* parse through the packet, learn what we can */
   if (dhcp->your_ip_address.as_u32)
     c->leased_address.as_u32 = dhcp->your_ip_address.as_u32;
-      
+
+  c->dhcp_server.as_u32 = dhcp->server_ip_address.as_u32;
+
   o = (dhcp_option_t *) dhcp->options;
   
   while (o->option != 0xFF /* end of options */ &&
@@ -172,6 +238,14 @@ int dhcp_client_for_us (u32 bi, vlib_buffer_t * b,
           c->next_transmit = now + 5.0;
           break;
         }
+      /*
+       * in order to accept unicasted ACKs we need to configure the offered
+       * address on the interface. However, at this point we may not know the
+       * subnet-mask (an OFFER may not contain it). So add a temporary receice
+       * and uRPF excempt entry
+       */
+      dhcp_client_add_rx_address (dcm, c);
+
       /* Received an offer, go send a request */
       c->state = DHCP_REQUEST;
       c->retry_count = 0;
@@ -196,6 +270,8 @@ int dhcp_client_for_us (u32 bi, vlib_buffer_t * b,
         {
           void (*fp)(u32, u32, u8 *, u8, u8, u8 *, u8 *, u8 *) = c->event_callback;
 
+          /* replace the temporary RX address with the correct subnet */
+          dhcp_client_remove_rx_address (dcm, c);
           dhcp_client_acquire_address (dcm, c);
 
           /*
@@ -831,6 +907,7 @@ int dhcp_client_add_del (dhcp_client_add_del_args_t * a)
 				      1,
 				      FIB_ROUTE_PATH_FLAG_NONE);
       }
+      dhcp_client_remove_rx_address (dcm, c);
       dhcp_client_release_address (dcm, c);
       ip4_sw_interface_enable_disable (c->sw_if_index, 0);
 
diff --git a/src/vnet/fib/fib_entry.h b/src/vnet/fib/fib_entry.h
index 5f6ff312..93b8016d 100644
--- a/src/vnet/fib/fib_entry.h
+++ b/src/vnet/fib/fib_entry.h
@@ -205,14 +205,9 @@ typedef enum fib_entry_attribute_t_ {
     /**
      * Marker. add new entries before this one.
      */
-    FIB_ENTRY_ATTRIBUTE_LAST = FIB_ENTRY_ATTRIBUTE_MULTICAST,
+    FIB_ENTRY_ATTRIBUTE_LAST = FIB_ENTRY_ATTRIBUTE_URPF_EXEMPT,
 } fib_entry_attribute_t;
 
-/**
- * The maximum number of sources
- */
-#define FIB_ENTRY_ATTRIBUTE_MAX (FIB_ENTRY_ATTRIBUTE_LAST+1)
-
 #define FIB_ENTRY_ATTRIBUTES {		       		\
     [FIB_ENTRY_ATTRIBUTE_CONNECTED] = "connected",	\
     [FIB_ENTRY_ATTRIBUTE_ATTACHED]  = "attached",	\
@@ -226,7 +221,7 @@ typedef enum fib_entry_attribute_t_ {
 
 #define FOR_EACH_FIB_ATTRIBUTE(_item)			\
     for (_item = FIB_ENTRY_ATTRIBUTE_FIRST;		\
-	 _item < FIB_ENTRY_ATTRIBUTE_MAX;		\
+	 _item <= FIB_ENTRY_ATTRIBUTE_LAST;		\
 	 _item++)
 
 typedef enum fib_entry_flag_t_ {
diff --git a/test/test_dhcp.py b/test/test_dhcp.py
index 4e8ed4ce..6fc29182 100644
--- a/test/test_dhcp.py
+++ b/test/test_dhcp.py
@@ -196,6 +196,10 @@ class TestDHCP(VppTestCase):
         self.verify_dhcp_has_option(pkt, "hostname", hostname)
         if client_id:
             self.verify_dhcp_has_option(pkt, "client_id", client_id)
+        bootp = pkt[BOOTP]
+        self.assertEqual(bootp.ciaddr, "0.0.0.0")
+        self.assertEqual(bootp.giaddr, "0.0.0.0")
+        self.assertEqual(bootp.flags, 0x8000)
 
     def verify_orig_dhcp_request(self, pkt, intf, hostname, ip):
         self.verify_orig_dhcp_pkt(pkt, intf)
@@ -203,6 +207,10 @@ class TestDHCP(VppTestCase):
         self.verify_dhcp_msg_type(pkt, "request")
         self.verify_dhcp_has_option(pkt, "hostname", hostname)
         self.verify_dhcp_has_option(pkt, "requested_addr", ip)
+        bootp = pkt[BOOTP]
+        self.assertEqual(bootp.ciaddr, "0.0.0.0")
+        self.assertEqual(bootp.giaddr, "0.0.0.0")
+        self.assertEqual(bootp.flags, 0x8000)
 
     def verify_relayed_dhcp_discover(self, pkt, intf, src_intf=None,
                                      fib_id=0, oui=0,
@@ -1057,14 +1065,15 @@ class TestDHCP(VppTestCase):
         #
         # Sned back on offer, expect the request
         #
-        p = (Ether(dst=self.pg2.local_mac, src=self.pg2.remote_mac) /
-             IP(src=self.pg2.remote_ip4, dst="255.255.255.255") /
-             UDP(sport=DHCP4_SERVER_PORT, dport=DHCP4_CLIENT_PORT) /
-             BOOTP(op=1,
-                   yiaddr=self.pg2.local_ip4) /
-             DHCP(options=[('message-type', 'offer'), ('end')]))
+        p_offer = (Ether(dst=self.pg2.local_mac, src=self.pg2.remote_mac) /
+                   IP(src=self.pg2.remote_ip4, dst="255.255.255.255") /
+                   UDP(sport=DHCP4_SERVER_PORT, dport=DHCP4_CLIENT_PORT) /
+                   BOOTP(op=1, yiaddr=self.pg2.local_ip4) /
+                   DHCP(options=[('message-type', 'offer'),
+                                 ('server_id', self.pg2.remote_ip4),
+                                 ('end')]))
 
-        self.pg2.add_stream(p)
+        self.pg2.add_stream(p_offer)
         self.pg_enable_capture(self.pg_interfaces)
         self.pg_start()
 
@@ -1075,19 +1084,18 @@ class TestDHCP(VppTestCase):
         #
         # Send an acknowloedgement
         #
-        p = (Ether(dst=self.pg2.local_mac, src=self.pg2.remote_mac) /
-             IP(src=self.pg2.remote_ip4, dst="255.255.255.255") /
-             UDP(sport=DHCP4_SERVER_PORT, dport=DHCP4_CLIENT_PORT) /
-             BOOTP(op=1,
-                   yiaddr=self.pg2.local_ip4) /
-             DHCP(options=[('message-type', 'ack'),
-                           ('subnet_mask', "255.255.255.0"),
-                           ('router', self.pg2.remote_ip4),
-                           ('server_id', self.pg2.remote_ip4),
-                           ('lease_time', 43200),
-                           ('end')]))
-
-        self.pg2.add_stream(p)
+        p_ack = (Ether(dst=self.pg2.local_mac, src=self.pg2.remote_mac) /
+                 IP(src=self.pg2.remote_ip4, dst="255.255.255.255") /
+                 UDP(sport=DHCP4_SERVER_PORT, dport=DHCP4_CLIENT_PORT) /
+                 BOOTP(op=1, yiaddr=self.pg2.local_ip4) /
+                 DHCP(options=[('message-type', 'ack'),
+                               ('subnet_mask', "255.255.255.0"),
+                               ('router', self.pg2.remote_ip4),
+                               ('server_id', self.pg2.remote_ip4),
+                               ('lease_time', 43200),
+                               ('end')]))
+
+        self.pg2.add_stream(p_ack)
         self.pg_enable_capture(self.pg_interfaces)
         self.pg_start()
 
@@ -1103,6 +1111,7 @@ class TestDHCP(VppTestCase):
         # At the end of this procedure there should be a connected route
         # in the FIB
         #
+        self.assertTrue(find_route(self, self.pg2.local_ip4, 24))
         self.assertTrue(find_route(self, self.pg2.local_ip4, 32))
 
         # remove the left over ARP entry
@@ -1119,10 +1128,14 @@ class TestDHCP(VppTestCase):
         # and now the route should be gone
         #
         self.assertFalse(find_route(self, self.pg2.local_ip4, 32))
+        self.assertFalse(find_route(self, self.pg2.local_ip4, 24))
 
         #
-        # Start the procedure again. this time have VPP send the clientiid
+        # Start the procedure again. this time have VPP send the client-ID
         #
+        self.pg2.admin_down()
+        self.sleep(1)
+        self.pg2.admin_up()
         self.vapi.dhcp_client(self.pg2.sw_if_index, hostname,
                               client_id=self.pg2.local_mac)
 
@@ -1131,10 +1144,47 @@ class TestDHCP(VppTestCase):
         self.verify_orig_dhcp_discover(rx[0], self.pg2, hostname,
                                        self.pg2.local_mac)
 
+        self.pg2.add_stream(p_offer)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        rx = self.pg2.get_capture(1)
+        self.verify_orig_dhcp_request(rx[0], self.pg2, hostname,
+                                      self.pg2.local_ip4)
+
+        #
+        # unicast the ack to the offered address
+        #
+        p_ack = (Ether(dst=self.pg2.local_mac, src=self.pg2.remote_mac) /
+                 IP(src=self.pg2.remote_ip4, dst=self.pg2.local_ip4) /
+                 UDP(sport=DHCP4_SERVER_PORT, dport=DHCP4_CLIENT_PORT) /
+                 BOOTP(op=1, yiaddr=self.pg2.local_ip4) /
+                 DHCP(options=[('message-type', 'ack'),
+                               ('subnet_mask', "255.255.255.0"),
+                               ('router', self.pg2.remote_ip4),
+                               ('server_id', self.pg2.remote_ip4),
+                               ('lease_time', 43200),
+                               ('end')]))
+
+        self.pg2.add_stream(p_ack)
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        #
+        # At the end of this procedure there should be a connected route
+        # in the FIB
+        #
+        self.assertTrue(find_route(self, self.pg2.local_ip4, 32))
+        self.assertTrue(find_route(self, self.pg2.local_ip4, 24))
+
         #
         # remove the DHCP config
         #
         self.vapi.dhcp_client(self.pg2.sw_if_index, hostname, is_add=0)
 
+        self.assertFalse(find_route(self, self.pg2.local_ip4, 32))
+        self.assertFalse(find_route(self, self.pg2.local_ip4, 24))
+
+
 if __name__ == '__main__':
     unittest.main(testRunner=VppTestRunner)
-- 
cgit 1.2.3-korg


From 1500254bee11355bbd69cc1dd9705be4f002f2bd Mon Sep 17 00:00:00 2001
From: Neale Ranns <nranns@cisco.com>
Date: Sun, 10 Sep 2017 04:39:11 -0700
Subject: FIB table add/delete API

part 2;
  - this adds the code to create an IP and MPLS table via the API.
  - but the enforcement that the table must be created before it is used is still missing, this is so that CSIT can pass.

Change-Id: Id124d884ade6cb7da947225200e3bb193454c555
Signed-off-by: Neale Ranns <nranns@cisco.com>
---
 src/plugins/nat/nat.c                         |  17 +-
 src/plugins/nat/nat64.c                       |  13 +-
 src/vnet/classify/vnet_classify.c             |  16 +-
 src/vnet/dhcp/dhcp4_proxy_node.c              |   9 +-
 src/vnet/dhcp/dhcp6_proxy_node.c              |   9 +-
 src/vnet/dhcp/dhcp_proxy.c                    |  19 ++-
 src/vnet/dpo/lookup_dpo.c                     |  20 ++-
 src/vnet/dpo/mpls_label_dpo.c                 |  12 +-
 src/vnet/ethernet/arp.c                       | 127 +++++++++++----
 src/vnet/fib/fib_api.h                        |   1 -
 src/vnet/fib/fib_entry.c                      |  15 +-
 src/vnet/fib/fib_entry.h                      |   1 +
 src/vnet/fib/fib_entry_src_mpls.c             |   7 +-
 src/vnet/fib/fib_table.c                      |  43 +++--
 src/vnet/fib/fib_table.h                      |  32 +++-
 src/vnet/fib/fib_test.c                       |  27 ++--
 src/vnet/fib/ip4_fib.c                        |  41 +++--
 src/vnet/fib/ip4_fib.h                        |   5 +-
 src/vnet/fib/ip6_fib.c                        |  41 +++--
 src/vnet/fib/ip6_fib.h                        |   5 +-
 src/vnet/fib/mpls_fib.c                       |  16 +-
 src/vnet/fib/mpls_fib.h                       |   5 +-
 src/vnet/interface_api.c                      | 177 ++++++++++++++++----
 src/vnet/ip/ip.h                              |   7 +
 src/vnet/ip/ip4.h                             |  13 ++
 src/vnet/ip/ip4_forward.c                     | 101 +-----------
 src/vnet/ip/ip4_source_and_port_range_check.c |  11 +-
 src/vnet/ip/ip6.h                             |  13 ++
 src/vnet/ip/ip6_forward.c                     | 103 +-----------
 src/vnet/ip/ip6_neighbor.c                    | 108 +++++++++----
 src/vnet/ip/ip_api.c                          | 122 +++++++++++---
 src/vnet/ip/lookup.c                          | 225 ++++++++++++++++++++++++++
 src/vnet/lisp-gpe/interface.c                 |  11 +-
 src/vnet/lisp-gpe/lisp_gpe_fwd_entry.c        |   9 +-
 src/vnet/lisp-gpe/lisp_gpe_sub_interface.c    |  11 +-
 src/vnet/mfib/ip4_mfib.c                      |  12 +-
 src/vnet/mfib/ip4_mfib.h                      |   5 +-
 src/vnet/mfib/ip6_mfib.c                      |  12 +-
 src/vnet/mfib/ip6_mfib.h                      |   5 +-
 src/vnet/mfib/mfib_entry.c                    |  11 ++
 src/vnet/mfib/mfib_entry.h                    |   2 +
 src/vnet/mfib/mfib_table.c                    |  88 ++++++++--
 src/vnet/mfib/mfib_table.h                    |  29 +++-
 src/vnet/mfib/mfib_test.c                     |  11 +-
 src/vnet/mfib/mfib_types.h                    |   8 +-
 src/vnet/mpls/interface.c                     |  26 ++-
 src/vnet/mpls/mpls.c                          |  76 ++++++++-
 src/vnet/mpls/mpls.h                          |  16 +-
 src/vnet/mpls/mpls_api.c                      |  66 ++++++--
 src/vnet/srv6/sr_policy_rewrite.c             |   6 +-
 src/vnet/srv6/sr_steering.c                   |   6 +-
 src/vpp/api/api.c                             |   5 +-
 src/vpp/api/custom_dump.c                     |   3 -
 test/test_dhcp.py                             |  24 ++-
 test/test_gre.py                              |   8 +-
 test/test_ip4.py                              |  11 +-
 test/test_ip4_vrf_multi_instance.py           |   4 +-
 test/test_ip6.py                              |   7 +-
 test/test_ip6_vrf_multi_instance.py           |   4 +-
 test/test_ip_mcast.py                         |  98 ++++++++++-
 test/test_mpls.py                             |  48 +++++-
 test/test_nat.py                              |  13 ++
 test/test_neighbor.py                         |  66 +++++++-
 test/vpp_ip_route.py                          |  73 +++++++++
 test/vpp_papi_provider.py                     |  46 ++++--
 65 files changed, 1643 insertions(+), 538 deletions(-)

(limited to 'test/test_dhcp.py')

diff --git a/src/plugins/nat/nat.c b/src/plugins/nat/nat.c
index aa7ef10a..8aecac6d 100644
--- a/src/plugins/nat/nat.c
+++ b/src/plugins/nat/nat.c
@@ -167,7 +167,8 @@ void snat_add_address (snat_main_t *sm, ip4_address_t *addr, u32 vrf_id)
   ap->addr = *addr;
   if (vrf_id != ~0)
     ap->fib_index =
-      fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, vrf_id);
+      fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, vrf_id,
+                                         FIB_SOURCE_PLUGIN_HI);
   else
     ap->fib_index = ~0;
 #define _(N, i, n, s) \
@@ -625,7 +626,8 @@ int nat44_add_del_lb_static_mapping (ip4_address_t e_addr, u16 e_port,
         return VNET_API_ERROR_INVALID_VALUE;
 
       fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4,
-                                                     vrf_id);
+                                                     vrf_id,
+                                                     FIB_SOURCE_PLUGIN_HI);
 
       /* Find external address in allocated addresses and reserve port for
          address and port pair mapping when dynamic translations enabled */
@@ -754,7 +756,7 @@ int nat44_add_del_lb_static_mapping (ip4_address_t e_addr, u16 e_port,
       if (!m)
         return VNET_API_ERROR_NO_SUCH_ENTRY;
 
-      fib_table_unlock (m->fib_index, FIB_PROTOCOL_IP4);
+      fib_table_unlock (m->fib_index, FIB_PROTOCOL_IP4, FIB_SOURCE_PLUGIN_HI);
 
       /* Free external address port */
       if (!sm->static_mapping_only)
@@ -874,7 +876,8 @@ int snat_del_address (snat_main_t *sm, ip4_address_t addr, u8 delete_sm)
     }
 
   if (a->fib_index != ~0)
-    fib_table_unlock(a->fib_index, FIB_PROTOCOL_IP4);
+    fib_table_unlock(a->fib_index, FIB_PROTOCOL_IP4,
+                     FIB_SOURCE_PLUGIN_HI);
 
   /* Delete sessions using address */
   if (a->busy_tcp_ports || a->busy_udp_ports || a->busy_icmp_ports)
@@ -2151,10 +2154,12 @@ snat_config (vlib_main_t * vm, unformat_input_t * input)
   sm->max_translations_per_user = max_translations_per_user;
   sm->outside_vrf_id = outside_vrf_id;
   sm->outside_fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4,
-                                                             outside_vrf_id);
+                                                             outside_vrf_id,
+                                                             FIB_SOURCE_PLUGIN_HI);
   sm->inside_vrf_id = inside_vrf_id;
   sm->inside_fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4,
-                                                            inside_vrf_id);
+                                                            inside_vrf_id,
+                                                            FIB_SOURCE_PLUGIN_HI);
   sm->static_mapping_only = static_mapping_only;
   sm->static_mapping_connection_tracking = static_mapping_connection_tracking;
 
diff --git a/src/plugins/nat/nat64.c b/src/plugins/nat/nat64.c
index b04901fa..bfcfa9b3 100644
--- a/src/plugins/nat/nat64.c
+++ b/src/plugins/nat/nat64.c
@@ -107,7 +107,8 @@ nat64_add_del_pool_addr (ip4_address_t * addr, u32 vrf_id, u8 is_add)
       a->fib_index = 0;
       if (vrf_id != ~0)
 	a->fib_index =
-	  fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6, vrf_id);
+	  fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6, vrf_id,
+					     FIB_SOURCE_PLUGIN_HI);
 #define _(N, i, n, s) \
       clib_bitmap_alloc (a->busy_##n##_port_bitmap, 65535);
       foreach_snat_protocol
@@ -119,7 +120,8 @@ nat64_add_del_pool_addr (ip4_address_t * addr, u32 vrf_id, u8 is_add)
 	return VNET_API_ERROR_NO_SUCH_ENTRY;
 
       if (a->fib_index)
-	fib_table_unlock (a->fib_index, FIB_PROTOCOL_IP6);
+	fib_table_unlock (a->fib_index, FIB_PROTOCOL_IP6,
+			  FIB_SOURCE_PLUGIN_HI);
 
 #define _(N, id, n, s) \
       clib_bitmap_free (a->busy_##n##_port_bitmap);
@@ -353,8 +355,8 @@ nat64_add_del_static_bib_entry (ip6_address_t * in_addr,
 {
   nat64_main_t *nm = &nat64_main;
   nat64_db_bib_entry_t *bibe;
-  u32 fib_index =
-    fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6, vrf_id);
+  u32 fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6, vrf_id,
+						     FIB_SOURCE_PLUGIN_HI);
   snat_protocol_t p = ip_proto_to_snat_proto (proto);
   ip46_address_t addr;
   int i;
@@ -644,7 +646,8 @@ nat64_add_del_prefix (ip6_address_t * prefix, u8 plen, u32 vrf_id, u8 is_add)
 	{
 	  vec_add2 (nm->pref64, p, 1);
 	  p->fib_index =
-	    fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6, vrf_id);
+	    fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6, vrf_id,
+					       FIB_SOURCE_PLUGIN_HI);
 	  p->vrf_id = vrf_id;
 	}
 
diff --git a/src/vnet/classify/vnet_classify.c b/src/vnet/classify/vnet_classify.c
index 879fba3c..57d86748 100644
--- a/src/vnet/classify/vnet_classify.c
+++ b/src/vnet/classify/vnet_classify.c
@@ -368,10 +368,10 @@ vnet_classify_entry_claim_resource (vnet_classify_entry_t *e)
     switch (e->action)
     {
     case CLASSIFY_ACTION_SET_IP4_FIB_INDEX:
-        fib_table_lock (e->metadata, FIB_PROTOCOL_IP4);
+        fib_table_lock (e->metadata, FIB_PROTOCOL_IP4, FIB_SOURCE_CLASSIFY);
         break;
     case CLASSIFY_ACTION_SET_IP6_FIB_INDEX:
-        fib_table_lock (e->metadata, FIB_PROTOCOL_IP6);
+        fib_table_lock (e->metadata, FIB_PROTOCOL_IP6, FIB_SOURCE_CLASSIFY);
         break;
     }
 }
@@ -382,10 +382,10 @@ vnet_classify_entry_release_resource (vnet_classify_entry_t *e)
     switch (e->action)
     {
     case CLASSIFY_ACTION_SET_IP4_FIB_INDEX:
-        fib_table_unlock (e->metadata, FIB_PROTOCOL_IP4);
+        fib_table_unlock (e->metadata, FIB_PROTOCOL_IP4, FIB_SOURCE_CLASSIFY);
         break;
     case CLASSIFY_ACTION_SET_IP6_FIB_INDEX:
-        fib_table_unlock (e->metadata, FIB_PROTOCOL_IP6);
+        fib_table_unlock (e->metadata, FIB_PROTOCOL_IP6, FIB_SOURCE_CLASSIFY);
         break;
     }
 }
@@ -2096,9 +2096,13 @@ int vnet_classify_add_del_session (vnet_classify_main_t * cm,
   e->flags = 0;
   e->action = action;
   if (e->action == CLASSIFY_ACTION_SET_IP4_FIB_INDEX)
-    e->metadata = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, metadata);
+    e->metadata = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4,
+                                                     metadata,
+                                                     FIB_SOURCE_CLASSIFY);
   else if (e->action == CLASSIFY_ACTION_SET_IP6_FIB_INDEX)
-    e->metadata = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6, metadata);
+    e->metadata = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6,
+                                                     metadata,
+                                                     FIB_SOURCE_CLASSIFY);
   else
     e->metadata = 0;
 
diff --git a/src/vnet/dhcp/dhcp4_proxy_node.c b/src/vnet/dhcp/dhcp4_proxy_node.c
index 1b59cdea..339a7885 100644
--- a/src/vnet/dhcp/dhcp4_proxy_node.c
+++ b/src/vnet/dhcp/dhcp4_proxy_node.c
@@ -785,7 +785,8 @@ dhcp4_proxy_set_server (ip46_address_t *addr,
     return VNET_API_ERROR_INVALID_SRC_ADDRESS;
 
   rx_fib_index = fib_table_find_or_create_and_lock(FIB_PROTOCOL_IP4,
-                                                   rx_table_id);
+                                                   rx_table_id,
+                                                   FIB_SOURCE_DHCP);
 
   if (is_del)
     {
@@ -795,7 +796,7 @@ dhcp4_proxy_set_server (ip46_address_t *addr,
           fib_table_entry_special_remove(rx_fib_index,
                                          &all_1s,
                                          FIB_SOURCE_DHCP);
-          fib_table_unlock (rx_fib_index, FIB_PROTOCOL_IP4);
+          fib_table_unlock (rx_fib_index, FIB_PROTOCOL_IP4, FIB_SOURCE_DHCP);
       }
     }
   else
@@ -808,10 +809,10 @@ dhcp4_proxy_set_server (ip46_address_t *addr,
                                       &all_1s,
                                       FIB_SOURCE_DHCP,
                                       FIB_ENTRY_FLAG_LOCAL);
-          fib_table_lock (rx_fib_index, FIB_PROTOCOL_IP4);
+          fib_table_lock (rx_fib_index, FIB_PROTOCOL_IP4, FIB_SOURCE_DHCP);
       }
   }
-  fib_table_unlock (rx_fib_index, FIB_PROTOCOL_IP4);
+  fib_table_unlock (rx_fib_index, FIB_PROTOCOL_IP4, FIB_SOURCE_DHCP);
 
   return (rc);
 }
diff --git a/src/vnet/dhcp/dhcp6_proxy_node.c b/src/vnet/dhcp/dhcp6_proxy_node.c
index 9c2f5220..ce7a8fca 100644
--- a/src/vnet/dhcp/dhcp6_proxy_node.c
+++ b/src/vnet/dhcp/dhcp6_proxy_node.c
@@ -841,7 +841,8 @@ dhcp6_proxy_set_server (ip46_address_t *addr,
     return VNET_API_ERROR_INVALID_SRC_ADDRESS;
 
   rx_fib_index = mfib_table_find_or_create_and_lock(FIB_PROTOCOL_IP6,
-                                                    rx_table_id);
+                                                    rx_table_id,
+                                                    MFIB_SOURCE_DHCP);
 
   if (is_del)
     {
@@ -851,7 +852,7 @@ dhcp6_proxy_set_server (ip46_address_t *addr,
           mfib_table_entry_delete(rx_fib_index,
                                   &all_dhcp_servers,
                                   MFIB_SOURCE_DHCP);
-          mfib_table_unlock(rx_fib_index, FIB_PROTOCOL_IP6);
+          mfib_table_unlock(rx_fib_index, FIB_PROTOCOL_IP6, MFIB_SOURCE_DHCP);
       }
     }
   else
@@ -885,11 +886,11 @@ dhcp6_proxy_set_server (ip46_address_t *addr,
                                  MFIB_SOURCE_DHCP,
                                  MFIB_RPF_ID_NONE,
                                  MFIB_ENTRY_FLAG_ACCEPT_ALL_ITF);
-         mfib_table_lock(rx_fib_index, FIB_PROTOCOL_IP6);
+         mfib_table_lock(rx_fib_index, FIB_PROTOCOL_IP6, MFIB_SOURCE_DHCP);
      }
     }
 
-  mfib_table_unlock(rx_fib_index, FIB_PROTOCOL_IP6);
+  mfib_table_unlock(rx_fib_index, FIB_PROTOCOL_IP6, MFIB_SOURCE_DHCP);
 
   return (rc);
 }
diff --git a/src/vnet/dhcp/dhcp_proxy.c b/src/vnet/dhcp/dhcp_proxy.c
index ba7f354e..1784906b 100644
--- a/src/vnet/dhcp/dhcp_proxy.c
+++ b/src/vnet/dhcp/dhcp_proxy.c
@@ -29,9 +29,9 @@ dhcp_proxy_rx_table_lock (fib_protocol_t proto,
                           u32 fib_index)
 {
     if (FIB_PROTOCOL_IP4 == proto)
-        fib_table_lock(fib_index, proto);
+        fib_table_lock(fib_index, proto, FIB_SOURCE_DHCP);
     else
-        mfib_table_lock(fib_index, proto);
+        mfib_table_lock(fib_index, proto, MFIB_SOURCE_DHCP);
 }
 
 static void
@@ -39,9 +39,9 @@ dhcp_proxy_rx_table_unlock (fib_protocol_t proto,
                             u32 fib_index)
 {
     if (FIB_PROTOCOL_IP4 == proto)
-        fib_table_unlock(fib_index, proto);
+        fib_table_unlock(fib_index, proto, FIB_SOURCE_DHCP);
     else
-        mfib_table_unlock(fib_index, proto);
+        mfib_table_unlock(fib_index, proto, MFIB_SOURCE_DHCP);
 }
 
  u32
@@ -169,7 +169,7 @@ dhcp_proxy_server_del (fib_protocol_t proto,
       if (~0 != index)
       {
           server = &proxy->dhcp_servers[index];
-          fib_table_unlock (server->server_fib_index, proto);
+          fib_table_unlock (server->server_fib_index, proto, FIB_SOURCE_DHCP);
 
           vec_del1(proxy->dhcp_servers, index);
 
@@ -228,7 +228,8 @@ dhcp_proxy_server_add (fib_protocol_t proto,
   dhcp_server_t server = {
       .dhcp_server = *addr,
       .server_fib_index = fib_table_find_or_create_and_lock(proto,
-                                                            server_table_id),
+                                                            server_table_id,
+                                                            FIB_SOURCE_DHCP),
   };
 
   vec_add1(proxy->dhcp_servers, server);
@@ -297,9 +298,11 @@ int dhcp_proxy_set_vss (fib_protocol_t proto,
   int rc = 0;
   
   if (proto == FIB_PROTOCOL_IP4)
-      rx_fib_index = fib_table_find_or_create_and_lock(proto, tbl_id);
+      rx_fib_index = fib_table_find_or_create_and_lock(proto, tbl_id,
+                                                       FIB_SOURCE_DHCP);
   else
-      rx_fib_index = mfib_table_find_or_create_and_lock(proto, tbl_id);
+      rx_fib_index = mfib_table_find_or_create_and_lock(proto, tbl_id,
+                                                        MFIB_SOURCE_DHCP);
   v = dhcp_get_vss_info(dm, rx_fib_index, proto);
 
   if (NULL != v)
diff --git a/src/vnet/dpo/lookup_dpo.c b/src/vnet/dpo/lookup_dpo.c
index 26363a2f..af189eda 100644
--- a/src/vnet/dpo/lookup_dpo.c
+++ b/src/vnet/dpo/lookup_dpo.c
@@ -135,11 +135,15 @@ lookup_dpo_add_or_lock_w_fib_index (fib_node_index_t fib_index,
     {
         if (LOOKUP_UNICAST == cast)
         {
-            fib_table_lock(fib_index, dpo_proto_to_fib(proto));
+            fib_table_lock(fib_index,
+                           dpo_proto_to_fib(proto),
+                           FIB_SOURCE_RR);
         }
         else
         {
-            mfib_table_lock(fib_index, dpo_proto_to_fib(proto));
+            mfib_table_lock(fib_index,
+                            dpo_proto_to_fib(proto),
+                            MFIB_SOURCE_RR);
         }
     }
     lookup_dpo_add_or_lock_i(fib_index, proto, cast, input, table_config, dpo);
@@ -161,13 +165,15 @@ lookup_dpo_add_or_lock_w_table_id (u32 table_id,
         {
             fib_index =
                 fib_table_find_or_create_and_lock(dpo_proto_to_fib(proto),
-                                                  table_id);
+                                                  table_id,
+                                                  FIB_SOURCE_RR);
         }
         else
         {
             fib_index =
                 mfib_table_find_or_create_and_lock(dpo_proto_to_fib(proto),
-                                                   table_id);
+                                                   table_id,
+                                                   MFIB_SOURCE_RR);
         }
     }
 
@@ -238,12 +244,14 @@ lookup_dpo_unlock (dpo_id_t *dpo)
             if (LOOKUP_UNICAST == lkd->lkd_cast)
             {
                 fib_table_unlock(lkd->lkd_fib_index,
-                                 dpo_proto_to_fib(lkd->lkd_proto));
+                                 dpo_proto_to_fib(lkd->lkd_proto),
+                                 FIB_SOURCE_RR);
             }
             else
             {
                 mfib_table_unlock(lkd->lkd_fib_index,
-                                  dpo_proto_to_fib(lkd->lkd_proto));
+                                  dpo_proto_to_fib(lkd->lkd_proto),
+                                  MFIB_SOURCE_RR);
             }
         }
         pool_put(lookup_dpo_pool, lkd);
diff --git a/src/vnet/dpo/mpls_label_dpo.c b/src/vnet/dpo/mpls_label_dpo.c
index b178a902..2a6e7dd5 100644
--- a/src/vnet/dpo/mpls_label_dpo.c
+++ b/src/vnet/dpo/mpls_label_dpo.c
@@ -105,10 +105,18 @@ format_mpls_label_dpo (u8 *s, va_list *args)
     mpls_label_dpo_t *mld;
     u32 ii;
 
-    mld = mpls_label_dpo_get(index);
-
     s = format(s, "mpls-label:[%d]:", index);
 
+    if (pool_is_free_index(mpls_label_dpo_pool, index))
+    {
+        /*
+         * the packet trace can be printed after the DPO has been deleted
+         */
+        return (s);
+    }
+
+    mld = mpls_label_dpo_get(index);
+
     for (ii = 0; ii < mld->mld_n_labels; ii++)
     {
 	hdr.label_exp_s_ttl =
diff --git a/src/vnet/ethernet/arp.c b/src/vnet/ethernet/arp.c
index c84ff47b..08e91373 100644
--- a/src/vnet/ethernet/arp.c
+++ b/src/vnet/ethernet/arp.c
@@ -522,6 +522,24 @@ arp_update_adjacency (vnet_main_t * vnm, u32 sw_if_index, u32 ai)
     }
 }
 
+static void
+arp_adj_fib_add (ethernet_arp_ip4_entry_t * e, uint32_t fib_index)
+{
+  fib_prefix_t pfx = {
+    .fp_len = 32,
+    .fp_proto = FIB_PROTOCOL_IP4,
+    .fp_addr.ip4 = e->ip4_address,
+  };
+
+  e->fib_entry_index =
+    fib_table_entry_path_add (fib_index, &pfx, FIB_SOURCE_ADJ,
+			      FIB_ENTRY_FLAG_ATTACHED,
+			      DPO_PROTO_IP4, &pfx.fp_addr,
+			      e->sw_if_index, ~0, 1, NULL,
+			      FIB_ROUTE_PATH_FLAG_NONE);
+  fib_table_lock (fib_index, FIB_PROTOCOL_IP4, FIB_SOURCE_ADJ);
+}
+
 int
 vnet_arp_set_ip4_over_ethernet_internal (vnet_main_t * vnm,
 					 vnet_arp_set_ip4_over_ethernet_rpc_args_t
@@ -576,21 +594,9 @@ vnet_arp_set_ip4_over_ethernet_internal (vnet_main_t * vnm,
 
       if (!is_no_fib_entry)
 	{
-	  fib_prefix_t pfx = {
-	    .fp_len = 32,
-	    .fp_proto = FIB_PROTOCOL_IP4,
-	    .fp_addr.ip4 = a->ip4,
-	  };
-	  u32 fib_index;
-
-	  fib_index =
-	    ip4_fib_table_get_index_for_sw_if_index (e->sw_if_index);
-	  e->fib_entry_index =
-	    fib_table_entry_path_add (fib_index, &pfx, FIB_SOURCE_ADJ,
-				      FIB_ENTRY_FLAG_ATTACHED,
-				      DPO_PROTO_IP4, &pfx.fp_addr,
-				      e->sw_if_index, ~0, 1, NULL,
-				      FIB_ROUTE_PATH_FLAG_NONE);
+	  arp_adj_fib_add (e,
+			   ip4_fib_table_get_index_for_sw_if_index
+			   (e->sw_if_index));
 	}
       else
 	{
@@ -1561,6 +1567,65 @@ arp_add_del_interface_address (ip4_main_t * im,
     }
 }
 
+void
+arp_adj_fib_remove (ethernet_arp_ip4_entry_t * e, uint32_t fib_index)
+{
+  if (FIB_NODE_INDEX_INVALID != e->fib_entry_index)
+    {
+      fib_prefix_t pfx = {
+	.fp_len = 32,
+	.fp_proto = FIB_PROTOCOL_IP4,
+	.fp_addr.ip4 = e->ip4_address,
+      };
+      u32 fib_index;
+
+      fib_index = ip4_fib_table_get_index_for_sw_if_index (e->sw_if_index);
+
+      fib_table_entry_path_remove (fib_index, &pfx,
+				   FIB_SOURCE_ADJ,
+				   DPO_PROTO_IP4,
+				   &pfx.fp_addr,
+				   e->sw_if_index, ~0, 1,
+				   FIB_ROUTE_PATH_FLAG_NONE);
+      fib_table_unlock (fib_index, FIB_PROTOCOL_IP4, FIB_SOURCE_ADJ);
+    }
+}
+
+static void
+arp_table_bind (ip4_main_t * im,
+		uword opaque,
+		u32 sw_if_index, u32 new_fib_index, u32 old_fib_index)
+{
+  ethernet_arp_main_t *am = &ethernet_arp_main;
+  ethernet_arp_interface_t *eai;
+  ethernet_arp_ip4_entry_t *e;
+  hash_pair_t *pair;
+
+  /*
+   * the IP table that the interface is bound to has changed.
+   * reinstall all the adj fibs.
+   */
+
+  if (vec_len (am->ethernet_arp_by_sw_if_index) <= sw_if_index)
+    return;
+
+  eai = &am->ethernet_arp_by_sw_if_index[sw_if_index];
+
+  /* *INDENT-OFF* */
+  hash_foreach_pair (pair, eai->arp_entries,
+  ({
+    e = pool_elt_at_index(am->ip4_entry_pool,
+                          pair->value[0]);
+    /*
+     * remove the adj-fib from the old table and add to the new
+     */
+    arp_adj_fib_remove(e, old_fib_index);
+    arp_adj_fib_add(e, new_fib_index);
+  }));
+  /* *INDENT-ON* */
+
+}
+
 static clib_error_t *
 ethernet_arp_init (vlib_main_t * vm)
 {
@@ -1606,6 +1671,11 @@ ethernet_arp_init (vlib_main_t * vm)
   cb.function_opaque = 0;
   vec_add1 (im->add_del_interface_address_callbacks, cb);
 
+  ip4_table_bind_callback_t cbt;
+  cbt.function = arp_table_bind;
+  cbt.function_opaque = 0;
+  vec_add1 (im->table_bind_callbacks, cbt);
+
   return 0;
 }
 
@@ -1616,24 +1686,9 @@ arp_entry_free (ethernet_arp_interface_t * eai, ethernet_arp_ip4_entry_t * e)
 {
   ethernet_arp_main_t *am = &ethernet_arp_main;
 
-  if (FIB_NODE_INDEX_INVALID != e->fib_entry_index)
-    {
-      fib_prefix_t pfx = {
-	.fp_len = 32,
-	.fp_proto = FIB_PROTOCOL_IP4,
-	.fp_addr.ip4 = e->ip4_address,
-      };
-      u32 fib_index;
-
-      fib_index = ip4_fib_table_get_index_for_sw_if_index (e->sw_if_index);
-
-      fib_table_entry_path_remove (fib_index, &pfx,
-				   FIB_SOURCE_ADJ,
-				   DPO_PROTO_IP4,
-				   &pfx.fp_addr,
-				   e->sw_if_index, ~0, 1,
-				   FIB_ROUTE_PATH_FLAG_NONE);
-    }
+  arp_adj_fib_remove (e,
+		      ip4_fib_table_get_index_for_sw_if_index
+		      (e->sw_if_index));
   hash_unset (eai->arp_entries, e->ip4_address.as_u32);
   pool_put (am->ip4_entry_pool, e);
 }
@@ -1693,7 +1748,11 @@ vnet_arp_flush_ip4_over_ethernet_internal (vnet_main_t * vnm,
        * does in response to interface events. unset is only done
        * by the control plane.
        */
-      if (e->flags & ETHERNET_ARP_IP4_ENTRY_FLAG_DYNAMIC)
+      if (e->flags & ETHERNET_ARP_IP4_ENTRY_FLAG_STATIC)
+	{
+	  e->flags &= ETHERNET_ARP_IP4_ENTRY_FLAG_DYNAMIC;
+	}
+      else if (e->flags & ETHERNET_ARP_IP4_ENTRY_FLAG_DYNAMIC)
 	{
 	  arp_entry_free (eai, e);
 	}
diff --git a/src/vnet/fib/fib_api.h b/src/vnet/fib/fib_api.h
index d07d6cae..f5a107ca 100644
--- a/src/vnet/fib/fib_api.h
+++ b/src/vnet/fib/fib_api.h
@@ -23,7 +23,6 @@ add_del_route_check (fib_protocol_t table_proto,
 		     u32 next_hop_sw_if_index,
 		     dpo_proto_t next_hop_table_proto,
 		     u32 next_hop_table_id,
-		     u8 create_missing_tables,
                      u8 is_rpf_id,
 		     u32 * fib_index, u32 * next_hop_fib_index);
 
diff --git a/src/vnet/fib/fib_entry.c b/src/vnet/fib/fib_entry.c
index 2027f2be..4cb6cf60 100644
--- a/src/vnet/fib/fib_entry.c
+++ b/src/vnet/fib/fib_entry.c
@@ -89,6 +89,17 @@ fib_entry_get_default_chain_type (const fib_entry_t *fib_entry)
     return (FIB_FORW_CHAIN_TYPE_UNICAST_IP4);
 }
 
+u8 *
+format_fib_source (u8 * s, va_list * args)
+{
+    fib_source_t source = va_arg (*args, int);
+
+    s = format (s, "\n  src:%s ",
+                fib_source_names[source]);
+
+    return (s);
+}
+
 u8 *
 format_fib_entry (u8 * s, va_list * args)
 {
@@ -114,8 +125,8 @@ format_fib_entry (u8 * s, va_list * args)
 
 	FOR_EACH_SRC_ADDED(fib_entry, src, source,
         ({
-	    s = format (s, "\n  src:%s ",
-			fib_source_names[source]);
+	    s = format (s, "\n  src:%U ",
+			format_fib_source, source);
 	    s = fib_entry_src_format(fib_entry, source, s);
 	    s = format (s, " refs:%d ", src->fes_ref_count);
 	    if (FIB_ENTRY_FLAG_NONE != src->fes_entry_flags) {
diff --git a/src/vnet/fib/fib_entry.h b/src/vnet/fib/fib_entry.h
index 93b8016d..2f6e37fe 100644
--- a/src/vnet/fib/fib_entry.h
+++ b/src/vnet/fib/fib_entry.h
@@ -431,6 +431,7 @@ typedef struct fib_entry_t_ {
 #define FIB_ENTRY_FORMAT_DETAIL2 (0x2)
 
 extern u8 *format_fib_entry (u8 * s, va_list * args);
+extern u8 *format_fib_source (u8 * s, va_list * args);
 
 extern fib_node_index_t fib_entry_create_special(u32 fib_index,
 						 const fib_prefix_t *prefix,
diff --git a/src/vnet/fib/fib_entry_src_mpls.c b/src/vnet/fib/fib_entry_src_mpls.c
index a616458f..6fdd5c0a 100644
--- a/src/vnet/fib/fib_entry_src_mpls.c
+++ b/src/vnet/fib/fib_entry_src_mpls.c
@@ -94,7 +94,9 @@ fib_entry_src_mpls_set_data (fib_entry_src_t *src,
 	    fib_table_entry_delete_index(src->mpls.fesm_lfes[eos],
 					 FIB_SOURCE_SPECIAL);
         }
-        fib_table_unlock(MPLS_FIB_DEFAULT_TABLE_ID, FIB_PROTOCOL_MPLS);
+        fib_table_unlock(MPLS_FIB_DEFAULT_TABLE_ID,
+                         FIB_PROTOCOL_MPLS,
+                         FIB_SOURCE_MPLS);
         src->mpls.fesm_label = label;
     }
     else
@@ -113,7 +115,8 @@ fib_entry_src_mpls_set_data (fib_entry_src_t *src,
         {
             fib_index =
 		fib_table_find_or_create_and_lock(FIB_PROTOCOL_MPLS,
-						  MPLS_FIB_DEFAULT_TABLE_ID);
+						  MPLS_FIB_DEFAULT_TABLE_ID,
+                                                  FIB_SOURCE_MPLS);
         }
 	else
 	{
diff --git a/src/vnet/fib/fib_table.c b/src/vnet/fib/fib_table.c
index 6b6cc5cb..75d15628 100644
--- a/src/vnet/fib/fib_table.c
+++ b/src/vnet/fib/fib_table.c
@@ -1039,7 +1039,8 @@ fib_table_find (fib_protocol_t proto,
 
 u32
 fib_table_find_or_create_and_lock (fib_protocol_t proto,
-				   u32 table_id)
+				   u32 table_id,
+                                   fib_source_t src)
 {
     fib_table_t *fib_table;
     fib_node_index_t fi;
@@ -1047,13 +1048,13 @@ fib_table_find_or_create_and_lock (fib_protocol_t proto,
     switch (proto)
     {
     case FIB_PROTOCOL_IP4:
-	fi = ip4_fib_table_find_or_create_and_lock(table_id);
+	fi = ip4_fib_table_find_or_create_and_lock(table_id, src);
         break;
     case FIB_PROTOCOL_IP6:
-	fi = ip6_fib_table_find_or_create_and_lock(table_id);
+	fi = ip6_fib_table_find_or_create_and_lock(table_id, src);
         break;
     case FIB_PROTOCOL_MPLS:
-	fi = mpls_fib_table_find_or_create_and_lock(table_id);
+	fi = mpls_fib_table_find_or_create_and_lock(table_id, src);
         break;
     default:
         return (~0);        
@@ -1070,6 +1071,7 @@ fib_table_find_or_create_and_lock (fib_protocol_t proto,
 
 u32
 fib_table_create_and_lock (fib_protocol_t proto,
+                           fib_source_t src,
                            const char *const fmt,
                            ...)
 {
@@ -1082,13 +1084,13 @@ fib_table_create_and_lock (fib_protocol_t proto,
     switch (proto)
     {
     case FIB_PROTOCOL_IP4:
-	fi = ip4_fib_table_create_and_lock();
+	fi = ip4_fib_table_create_and_lock(src);
         break;
     case FIB_PROTOCOL_IP6:
-	fi = ip6_fib_table_create_and_lock();
+	fi = ip6_fib_table_create_and_lock(src);
         break;
      case FIB_PROTOCOL_MPLS:
-	fi = mpls_fib_table_create_and_lock();
+	fi = mpls_fib_table_create_and_lock(src);
         break;
    default:
         return (~0);        
@@ -1143,26 +1145,43 @@ fib_table_walk (u32 fib_index,
 
 void
 fib_table_unlock (u32 fib_index,
-		  fib_protocol_t proto)
+		  fib_protocol_t proto,
+                  fib_source_t source)
 {
     fib_table_t *fib_table;
 
     fib_table = fib_table_get(fib_index, proto);
-    fib_table->ft_locks--;
+    fib_table->ft_locks[source]--;
+    fib_table->ft_locks[FIB_TABLE_TOTAL_LOCKS]--;
 
-    if (0 == fib_table->ft_locks)
+    if (0 == fib_table->ft_locks[source])
     {
+        /*
+         * The source no longer needs the table. flush any routes
+         * from it just in case
+         */
+        fib_table_flush(fib_index, proto, source);
+    }
+
+    if (0 == fib_table->ft_locks[FIB_TABLE_TOTAL_LOCKS])
+    {
+        /*
+         * no more locak from any source - kill it
+         */
 	fib_table_destroy(fib_table);
     }
 }
+
 void
 fib_table_lock (u32 fib_index,
-		fib_protocol_t proto)
+		fib_protocol_t proto,
+                fib_source_t source)
 {
     fib_table_t *fib_table;
 
     fib_table = fib_table_get(fib_index, proto);
-    fib_table->ft_locks++;
+    fib_table->ft_locks[source]++;
+    fib_table->ft_locks[FIB_TABLE_TOTAL_LOCKS]++;
 }
 
 u32
diff --git a/src/vnet/fib/fib_table.h b/src/vnet/fib/fib_table.h
index 579740e9..6b7011b3 100644
--- a/src/vnet/fib/fib_table.h
+++ b/src/vnet/fib/fib_table.h
@@ -22,6 +22,12 @@
 #include <vnet/mpls/mpls.h>
 #include <vnet/mpls/packet.h>
 
+/**
+ * Keep a lock per-source and a total
+ */
+#define FIB_TABLE_N_LOCKS (FIB_SOURCE_MAX+1)
+#define FIB_TABLE_TOTAL_LOCKS FIB_SOURCE_MAX
+
 /**
  * @brief 
  *   A protocol Independent FIB table
@@ -34,9 +40,9 @@ typedef struct fib_table_t_
     fib_protocol_t ft_proto;
 
     /**
-     * number of locks on the table
+     * per-source number of locks on the table
      */
-    u16 ft_locks;
+    u16 ft_locks[FIB_TABLE_N_LOCKS];
 
     /**
      * Table ID (hash key) for this FIB.
@@ -628,9 +634,13 @@ extern u32 fib_table_find(fib_protocol_t proto, u32 table_id);
  *
  * @return fib_index
  *  The index of the FIB
+ *
+ * @param source
+ *  The ID of the client/source.
  */
 extern u32 fib_table_find_or_create_and_lock(fib_protocol_t proto,
-					     u32 table_id);
+					     u32 table_id,
+                                             fib_source_t source);
 
 /**
  * @brief
@@ -643,10 +653,14 @@ extern u32 fib_table_find_or_create_and_lock(fib_protocol_t proto,
  * @param fmt
  *  A string to describe the table
  *
+ * @param source
+ *  The ID of the client/source.
+ *
  * @return fib_index
  *  The index of the FIB
  */
 extern u32 fib_table_create_and_lock(fib_protocol_t proto,
+                                     fib_source_t source,
                                      const char *const fmt,
                                      ...);
 
@@ -704,9 +718,13 @@ extern void fib_table_set_flow_hash_config(u32 fib_index,
  *
  * @paran proto
  *  The protocol of the FIB (and thus the entries therein)
+ *
+ * @param source
+ *  The ID of the client/source.
  */ 
 extern void fib_table_unlock(u32 fib_index,
-			     fib_protocol_t proto);
+			     fib_protocol_t proto,
+                             fib_source_t source);
 
 /**
  * @brief
@@ -718,9 +736,13 @@ extern void fib_table_unlock(u32 fib_index,
  *
  * @paran proto
  *  The protocol of the FIB (and thus the entries therein)
+ *
+ * @param source
+ *  The ID of the client/source.
  */ 
 extern void fib_table_lock(u32 fib_index,
-			   fib_protocol_t proto);
+			   fib_protocol_t proto,
+                           fib_source_t source);
 
 /**
  * @brief
diff --git a/src/vnet/fib/fib_test.c b/src/vnet/fib/fib_test.c
index 6867cca8..572d7f0d 100644
--- a/src/vnet/fib/fib_test.c
+++ b/src/vnet/fib/fib_test.c
@@ -739,7 +739,8 @@ fib_test_v4 (void)
     lb_count = pool_elts(load_balance_pool);
 
     /* Find or create FIB table 11 */
-    fib_index = fib_table_find_or_create_and_lock(FIB_PROTOCOL_IP4, 11);
+    fib_index = fib_table_find_or_create_and_lock(FIB_PROTOCOL_IP4, 11,
+                                                  FIB_SOURCE_API);
 
     for (ii = 0; ii < 4; ii++)
     {
@@ -4150,7 +4151,7 @@ fib_test_v4 (void)
                                              FIB_SOURCE_INTERFACE)),
              "NO INterface Source'd prefixes");
 
-    fib_table_unlock(fib_index, FIB_PROTOCOL_IP4);
+    fib_table_unlock(fib_index, FIB_PROTOCOL_IP4, FIB_SOURCE_API);
 
     FIB_TEST((0  == fib_path_list_db_size()), "path list DB population:%d",
     	     fib_path_list_db_size());
@@ -4201,7 +4202,8 @@ fib_test_v6 (void)
     dpo_drop = drop_dpo_get(DPO_PROTO_IP6);
 
     /* Find or create FIB table 11 */
-    fib_index = fib_table_find_or_create_and_lock(FIB_PROTOCOL_IP6, 11);
+    fib_index = fib_table_find_or_create_and_lock(FIB_PROTOCOL_IP6, 11,
+                                                  FIB_SOURCE_API);
 
     for (ii = 0; ii < 4; ii++)
     {
@@ -5025,7 +5027,7 @@ fib_test_v6 (void)
     /*
      * now remove the VRF
      */
-    fib_table_unlock(fib_index, FIB_PROTOCOL_IP6);
+    fib_table_unlock(fib_index, FIB_PROTOCOL_IP6, FIB_SOURCE_API);
 
     FIB_TEST((0 == fib_path_list_db_size()),   "path list DB population:%d",
 	     fib_path_list_db_size());
@@ -5157,7 +5159,9 @@ fib_test_ae (void)
      */
     u32 import_fib_index1;
 
-    import_fib_index1 = fib_table_find_or_create_and_lock(FIB_PROTOCOL_IP4, 11);
+    import_fib_index1 = fib_table_find_or_create_and_lock(FIB_PROTOCOL_IP4,
+                                                          11,
+                                                          FIB_SOURCE_CLI);
 
     /*
      * Add an attached route in the import FIB
@@ -5233,7 +5237,8 @@ fib_test_ae (void)
      */
     u32 import_fib_index2;
 
-    import_fib_index2 = fib_table_find_or_create_and_lock(FIB_PROTOCOL_IP4, 12);
+    import_fib_index2 = fib_table_find_or_create_and_lock(FIB_PROTOCOL_IP4, 12,
+                                                          FIB_SOURCE_CLI);
 
     /*
      * Add an attached route in the import FIB
@@ -5595,8 +5600,8 @@ fib_test_ae (void)
 			   &local_pfx,
 			   FIB_SOURCE_API);
 
-    fib_table_unlock(import_fib_index1, FIB_PROTOCOL_IP4);
-    fib_table_unlock(import_fib_index2, FIB_PROTOCOL_IP4);
+    fib_table_unlock(import_fib_index1, FIB_PROTOCOL_IP4, FIB_SOURCE_CLI);
+    fib_table_unlock(import_fib_index2, FIB_PROTOCOL_IP4, FIB_SOURCE_CLI);
 
     FIB_TEST((0 == adj_nbr_db_size()), "ADJ DB size is %d",
 	     adj_nbr_db_size());
@@ -8168,9 +8173,10 @@ lfib_test (void)
     /*
      * MPLS enable an interface so we get the MPLS table created
      */
+    mpls_table_create(MPLS_FIB_DEFAULT_TABLE_ID, FIB_SOURCE_API);
     mpls_sw_interface_enable_disable(&mpls_main,
                                      tm->hw[0]->sw_if_index,
-                                     1);
+                                     1, 1);
 
     ip46_address_t nh_10_10_10_1 = {
 	.ip4.as_u32 = clib_host_to_net_u32(0x0a0a0a01),
@@ -8662,7 +8668,8 @@ lfib_test (void)
      */
     mpls_sw_interface_enable_disable(&mpls_main,
                                      tm->hw[0]->sw_if_index,
-                                     0);
+                                     0, 1);
+    mpls_table_delete(MPLS_FIB_DEFAULT_TABLE_ID, FIB_SOURCE_API);
 
     FIB_TEST(lb_count == pool_elts(load_balance_pool),
 	     "Load-balance resources freed %d of %d",
diff --git a/src/vnet/fib/ip4_fib.c b/src/vnet/fib/ip4_fib.c
index d563bafd..865e2dd5 100644
--- a/src/vnet/fib/ip4_fib.c
+++ b/src/vnet/fib/ip4_fib.c
@@ -101,7 +101,8 @@ static const ip4_fib_table_special_prefix_t ip4_specials[] = {
 
 
 static u32
-ip4_create_fib_with_table_id (u32 table_id)
+ip4_create_fib_with_table_id (u32 table_id,
+                              fib_source_t src)
 {
     fib_table_t *fib_table;
     ip4_fib_t *v4_fib;
@@ -128,7 +129,7 @@ ip4_create_fib_with_table_id (u32 table_id)
     v4_fib->fwd_classify_table_index = ~0;
     v4_fib->rev_classify_table_index = ~0;
     
-    fib_table_lock(fib_table->ft_index, FIB_PROTOCOL_IP4);
+    fib_table_lock(fib_table->ft_index, FIB_PROTOCOL_IP4, src);
 
     ip4_mtrie_init(&v4_fib->mtrie);
 
@@ -198,23 +199,24 @@ ip4_fib_table_destroy (u32 fib_index)
 
 
 u32
-ip4_fib_table_find_or_create_and_lock (u32 table_id)
+ip4_fib_table_find_or_create_and_lock (u32 table_id,
+                                       fib_source_t src)
 {
     u32 index;
 
     index = ip4_fib_index_from_table_id(table_id);
     if (~0 == index)
-	return ip4_create_fib_with_table_id(table_id);
+	return ip4_create_fib_with_table_id(table_id, src);
 
-    fib_table_lock(index, FIB_PROTOCOL_IP4);
+    fib_table_lock(index, FIB_PROTOCOL_IP4, src);
 
     return (index);
 }
 
 u32
-ip4_fib_table_create_and_lock (void)
+ip4_fib_table_create_and_lock (fib_source_t src)
 {
-    return (ip4_create_fib_with_table_id(~0));
+    return (ip4_create_fib_with_table_id(~0, src));
 }
 
 u32
@@ -525,17 +527,32 @@ ip4_show_fib (vlib_main_t * vm,
     pool_foreach (fib_table, im4->fibs,
     ({
 	ip4_fib_t *fib = pool_elt_at_index(im4->v4_fibs, fib_table->ft_index);
+        fib_source_t source;
+        u8 *s = NULL;
 
 	if (table_id >= 0 && table_id != (int)fib->table_id)
 	    continue;
 	if (fib_index != ~0 && fib_index != (int)fib->index)
 	    continue;
 
-	vlib_cli_output (vm, "%U, fib_index:%d, flow hash:[%U] locks:%d", 
-			 format_fib_table_name, fib->index, FIB_PROTOCOL_IP4,
-			 fib->index,
-			 format_ip_flow_hash_config, fib_table->ft_flow_hash_config,
-                         fib_table->ft_locks);
+	s = format(s, "%U, fib_index:%d, flow hash:[%U] locks:[",
+                   format_fib_table_name, fib->index,
+                   FIB_PROTOCOL_IP4,
+                   fib->index,
+                   format_ip_flow_hash_config,
+                   fib_table->ft_flow_hash_config);
+	FOR_EACH_FIB_SOURCE(source)
+        {
+            if (0 != fib_table->ft_locks[source])
+            {
+                s = format(s, "%U:%d, ",
+                           format_fib_source, source,
+                           fib_table->ft_locks[source]);
+            }
+        }
+        s = format (s, "]");
+        vlib_cli_output (vm, "%V", s);
+        vec_free(s);
 
 	/* Show summary? */
 	if (! verbose)
diff --git a/src/vnet/fib/ip4_fib.h b/src/vnet/fib/ip4_fib.h
index 006163b4..495b45cc 100644
--- a/src/vnet/fib/ip4_fib.h
+++ b/src/vnet/fib/ip4_fib.h
@@ -127,8 +127,9 @@ ip4_fib_lookup (ip4_main_t * im, u32 sw_if_index, ip4_address_t * dst)
  * @returns A pointer to the retrieved or created fib.
  *
  */
-extern u32 ip4_fib_table_find_or_create_and_lock(u32 table_id);
-extern u32 ip4_fib_table_create_and_lock(void);
+extern u32 ip4_fib_table_find_or_create_and_lock(u32 table_id,
+                                                 fib_source_t src);
+extern u32 ip4_fib_table_create_and_lock(fib_source_t src);
 
 
 static inline 
diff --git a/src/vnet/fib/ip6_fib.c b/src/vnet/fib/ip6_fib.c
index 8fde6f9f..3ddb8453 100644
--- a/src/vnet/fib/ip6_fib.c
+++ b/src/vnet/fib/ip6_fib.c
@@ -50,7 +50,8 @@ vnet_ip6_fib_init (u32 fib_index)
 }
 
 static u32
-create_fib_with_table_id (u32 table_id)
+create_fib_with_table_id (u32 table_id,
+                          fib_source_t src)
 {
     fib_table_t *fib_table;
     ip6_fib_t *v6_fib;
@@ -77,29 +78,30 @@ create_fib_with_table_id (u32 table_id)
     fib_table->ft_flow_hash_config = IP_FLOW_HASH_DEFAULT;
 
     vnet_ip6_fib_init(fib_table->ft_index);
-    fib_table_lock(fib_table->ft_index, FIB_PROTOCOL_IP6);
+    fib_table_lock(fib_table->ft_index, FIB_PROTOCOL_IP6, src);
 
     return (fib_table->ft_index);
 }
 
 u32
-ip6_fib_table_find_or_create_and_lock (u32 table_id)
+ip6_fib_table_find_or_create_and_lock (u32 table_id,
+                                       fib_source_t src)
 {
     uword * p;
 
     p = hash_get (ip6_main.fib_index_by_table_id, table_id);
     if (NULL == p)
-	return create_fib_with_table_id(table_id);
+	return create_fib_with_table_id(table_id, src);
     
-    fib_table_lock(p[0], FIB_PROTOCOL_IP6);
+    fib_table_lock(p[0], FIB_PROTOCOL_IP6, src);
 
     return (p[0]);
 }
 
 u32
-ip6_fib_table_create_and_lock (void)
+ip6_fib_table_create_and_lock (fib_source_t src)
 {
-    return (create_fib_with_table_id(~0));
+    return (create_fib_with_table_id(~0, src));
 }
 
 void
@@ -588,16 +590,33 @@ ip6_show_fib (vlib_main_t * vm,
 
     pool_foreach (fib_table, im6->fibs,
     ({
+        fib_source_t source;
+        u8 *s = NULL;
+
 	fib = pool_elt_at_index(im6->v6_fibs, fib_table->ft_index);
 	if (table_id >= 0 && table_id != (int)fib->table_id)
 	    continue;
 	if (fib_index != ~0 && fib_index != (int)fib->index)
 	    continue;
 
-	vlib_cli_output (vm, "%s, fib_index:%d, flow hash:[%U] locks:%d", 
-			 fib_table->ft_desc, fib->index,
-			 format_ip_flow_hash_config, fib_table->ft_flow_hash_config,
-                         fib_table->ft_locks);
+	s = format(s, "%U, fib_index:%d, flow hash:[%U] locks:[",
+                   format_fib_table_name, fib->index,
+                   FIB_PROTOCOL_IP6,
+                   fib->index,
+                   format_ip_flow_hash_config,
+                   fib_table->ft_flow_hash_config);
+	FOR_EACH_FIB_SOURCE(source)
+        {
+            if (0 != fib_table->ft_locks[source])
+            {
+                s = format(s, "%U:%d, ",
+                           format_fib_source, source,
+                           fib_table->ft_locks[source]);
+            }
+        }
+        s = format (s, "]");
+        vlib_cli_output (vm, "%V", s);
+        vec_free(s);
 
 	/* Show summary? */
 	if (! verbose)
diff --git a/src/vnet/fib/ip6_fib.h b/src/vnet/fib/ip6_fib.h
index aad8305c..9728eecc 100644
--- a/src/vnet/fib/ip6_fib.h
+++ b/src/vnet/fib/ip6_fib.h
@@ -144,8 +144,9 @@ ip6_src_lookup_for_packet (ip6_main_t * im,
  * \returns A pointer to the retrieved or created fib.
  *
  */
-extern u32 ip6_fib_table_find_or_create_and_lock(u32 table_id);
-extern u32 ip6_fib_table_create_and_lock(void);
+extern u32 ip6_fib_table_find_or_create_and_lock(u32 table_id,
+                                                 fib_source_t src);
+extern u32 ip6_fib_table_create_and_lock(fib_source_t src);
 
 static inline ip6_fib_t *
 ip6_fib_get (fib_node_index_t index)
diff --git a/src/vnet/fib/mpls_fib.c b/src/vnet/fib/mpls_fib.c
index ca6271fe..4eeef7ab 100644
--- a/src/vnet/fib/mpls_fib.c
+++ b/src/vnet/fib/mpls_fib.c
@@ -83,7 +83,8 @@ mpls_fib_index_from_table_id (u32 table_id)
 }
 
 static u32
-mpls_fib_create_with_table_id (u32 table_id)
+mpls_fib_create_with_table_id (u32 table_id,
+                               fib_source_t src)
 {
     dpo_id_t dpo = DPO_INVALID;
     fib_table_t *fib_table;
@@ -107,7 +108,7 @@ mpls_fib_create_with_table_id (u32 table_id)
     fib_table->ft_table_id = table_id;
     fib_table->ft_flow_hash_config = MPLS_FLOW_HASH_DEFAULT;
     
-    fib_table_lock(fib_table->ft_index, FIB_PROTOCOL_MPLS);
+    fib_table_lock(fib_table->ft_index, FIB_PROTOCOL_MPLS, src);
 
     if (INDEX_INVALID == mpls_fib_drop_dpo_index)
     {
@@ -220,22 +221,23 @@ mpls_fib_create_with_table_id (u32 table_id)
 }
 
 u32
-mpls_fib_table_find_or_create_and_lock (u32 table_id)
+mpls_fib_table_find_or_create_and_lock (u32 table_id,
+                                        fib_source_t src)
 {
     u32 index;
 
     index = mpls_fib_index_from_table_id(table_id);
     if (~0 == index)
-	return mpls_fib_create_with_table_id(table_id);
+	return mpls_fib_create_with_table_id(table_id, src);
 
-    fib_table_lock(index, FIB_PROTOCOL_MPLS);
+    fib_table_lock(index, FIB_PROTOCOL_MPLS, src);
 
     return (index);
 }
 u32
-mpls_fib_table_create_and_lock (void)
+mpls_fib_table_create_and_lock (fib_source_t src)
 {
-    return (mpls_fib_create_with_table_id(~0));
+    return (mpls_fib_create_with_table_id(~0, src));
 }
 
 void
diff --git a/src/vnet/fib/mpls_fib.h b/src/vnet/fib/mpls_fib.h
index dfb8b7fc..29cd1d20 100644
--- a/src/vnet/fib/mpls_fib.h
+++ b/src/vnet/fib/mpls_fib.h
@@ -59,8 +59,9 @@ mpls_fib_get (fib_node_index_t index)
     return (pool_elt_at_index(mpls_main.mpls_fibs, index));
 }
 
-extern u32 mpls_fib_table_find_or_create_and_lock(u32 table_id);
-extern u32 mpls_fib_table_create_and_lock(void);
+extern u32 mpls_fib_table_find_or_create_and_lock(u32 table_id,
+                                                  fib_source_t src);
+extern u32 mpls_fib_table_create_and_lock(fib_source_t src);
 // extern mpls_fib_t * mpls_fib_find(u32 table_id);
 extern u32 mpls_fib_index_from_table_id(u32 table_id);
 
diff --git a/src/vnet/interface_api.c b/src/vnet/interface_api.c
index 113728cd..419fef94 100644
--- a/src/vnet/interface_api.c
+++ b/src/vnet/interface_api.c
@@ -320,68 +320,189 @@ stats_dsunlock (void)
 static void
 vl_api_sw_interface_set_table_t_handler (vl_api_sw_interface_set_table_t * mp)
 {
-  int rv = 0;
-  u32 table_id = ntohl (mp->vrf_id);
-  u32 sw_if_index = ntohl (mp->sw_if_index);
   vl_api_sw_interface_set_table_reply_t *rmp;
-  CLIB_UNUSED (ip_interface_address_t * ia);
-  u32 fib_index;
+  u32 sw_if_index = ntohl (mp->sw_if_index);
+  u32 table_id = ntohl (mp->vrf_id);
+  int rv = 0;
 
   VALIDATE_SW_IF_INDEX (mp);
 
   stats_dslock_with_hint (1 /* release hint */ , 4 /* tag */ );
 
   if (mp->is_ipv6)
+    rv = ip_table_bind (FIB_PROTOCOL_IP6, sw_if_index, table_id, 1);
+  else
+    rv = ip_table_bind (FIB_PROTOCOL_IP4, sw_if_index, table_id, 1);
+
+  stats_dsunlock ();
+
+  BAD_SW_IF_INDEX_LABEL;
+
+  REPLY_MACRO (VL_API_SW_INTERFACE_SET_TABLE_REPLY);
+}
+
+int
+ip_table_bind (fib_protocol_t fproto,
+	       uint32_t sw_if_index, uint32_t table_id, u8 is_api)
+{
+  CLIB_UNUSED (ip_interface_address_t * ia);
+  u32 fib_index, mfib_index;
+  fib_source_t src;
+  mfib_source_t msrc;
+
+  if (is_api)
+    {
+      src = FIB_SOURCE_API;
+      msrc = MFIB_SOURCE_API;
+    }
+  else
+    {
+      src = FIB_SOURCE_CLI;
+      msrc = MFIB_SOURCE_CLI;
+    }
+
+  /*
+   * This is temporary whilst I do the song and dance with the CSIT version
+   */
+  if (0 != table_id)
     {
+      fib_index = fib_table_find_or_create_and_lock (fproto, table_id, src);
+      mfib_index =
+	mfib_table_find_or_create_and_lock (fproto, table_id, msrc);
+    }
+  else
+    {
+      fib_index = 0;
+      mfib_index = 0;
+    }
+
+  /*
+   * This if table does not exist = error is what we want in the end.
+   */
+  /* fib_index = fib_table_find (fproto, table_id); */
+  /* mfib_index = mfib_table_find (fproto, table_id); */
+
+  /* if (~0 == fib_index || ~0 == mfib_index) */
+  /*   { */
+  /*     return (VNET_API_ERROR_NO_SUCH_FIB); */
+  /*   } */
+
+  if (FIB_PROTOCOL_IP6 == fproto)
+    {
+      /*
+       * If the interface already has in IP address, then a change int
+       * VRF is not allowed. The IP address applied must first be removed.
+       * We do not do that automatically here, since VPP has no knowledge
+       * of whether thoses subnets are valid in the destination VRF.
+       */
       /* *INDENT-OFF* */
       foreach_ip_interface_address (&ip6_main.lookup_main,
 				    ia, sw_if_index,
 				    1 /* honor unnumbered */ ,
       ({
-        rv = VNET_API_ERROR_ADDRESS_FOUND_FOR_INTERFACE;
-        goto done;
+        return (VNET_API_ERROR_ADDRESS_FOUND_FOR_INTERFACE);
       }));
       /* *INDENT-ON* */
 
-      fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6,
-						     table_id);
       vec_validate (ip6_main.fib_index_by_sw_if_index, sw_if_index);
-      ip6_main.fib_index_by_sw_if_index[sw_if_index] = fib_index;
-
-      fib_index = mfib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6,
-						      table_id);
       vec_validate (ip6_main.mfib_index_by_sw_if_index, sw_if_index);
-      ip6_main.mfib_index_by_sw_if_index[sw_if_index] = fib_index;
+
+      /*
+       * tell those that are interested that the binding is changing.
+       */
+      ip6_table_bind_callback_t *cb;
+      vec_foreach (cb, ip6_main.table_bind_callbacks)
+	cb->function (&ip6_main, cb->function_opaque,
+		      sw_if_index,
+		      fib_index,
+		      ip6_main.fib_index_by_sw_if_index[sw_if_index]);
+
+      if (0 == table_id)
+	{
+	  /* reset back to default */
+	  if (0 != ip6_main.fib_index_by_sw_if_index[sw_if_index])
+	    fib_table_unlock (ip6_main.fib_index_by_sw_if_index[sw_if_index],
+			      FIB_PROTOCOL_IP6, src);
+	  if (0 != ip6_main.mfib_index_by_sw_if_index[sw_if_index])
+	    mfib_table_unlock (ip6_main.mfib_index_by_sw_if_index
+			       [sw_if_index], FIB_PROTOCOL_IP6, msrc);
+
+	}
+      else
+	{
+	  /* we need to lock the table now it's inuse */
+	  fib_table_lock (fib_index, FIB_PROTOCOL_IP6, src);
+	  mfib_table_lock (mfib_index, FIB_PROTOCOL_IP6, msrc);
+	}
+
+      ip6_main.fib_index_by_sw_if_index[sw_if_index] = fib_index;
+      ip6_main.mfib_index_by_sw_if_index[sw_if_index] = mfib_index;
     }
   else
     {
+      /*
+       * If the interface already has in IP address, then a change int
+       * VRF is not allowed. The IP address applied must first be removed.
+       * We do not do that automatically here, since VPP has no knowledge
+       * of whether thoses subnets are valid in the destination VRF.
+       */
       /* *INDENT-OFF* */
       foreach_ip_interface_address (&ip4_main.lookup_main,
 				    ia, sw_if_index,
 				    1 /* honor unnumbered */ ,
       ({
-        rv = VNET_API_ERROR_ADDRESS_FOUND_FOR_INTERFACE;
-        goto done;
+        return (VNET_API_ERROR_ADDRESS_FOUND_FOR_INTERFACE);
       }));
       /* *INDENT-ON* */
 
-      fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4,
-						     table_id);
       vec_validate (ip4_main.fib_index_by_sw_if_index, sw_if_index);
-      ip4_main.fib_index_by_sw_if_index[sw_if_index] = fib_index;
-
-      fib_index = mfib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4,
-						      table_id);
       vec_validate (ip4_main.mfib_index_by_sw_if_index, sw_if_index);
-      ip4_main.mfib_index_by_sw_if_index[sw_if_index] = fib_index;
-    }
 
-done:
-  stats_dsunlock ();
+      /*
+       * tell those that are interested that the binding is changing.
+       */
+      ip4_table_bind_callback_t *cb;
+      vec_foreach (cb, ip4_main.table_bind_callbacks)
+	cb->function (&ip4_main, cb->function_opaque,
+		      sw_if_index,
+		      fib_index,
+		      ip4_main.fib_index_by_sw_if_index[sw_if_index]);
+
+      if (0 == table_id)
+	{
+	  /* reset back to default */
+	  if (0 != ip4_main.fib_index_by_sw_if_index[sw_if_index])
+	    fib_table_unlock (ip4_main.fib_index_by_sw_if_index[sw_if_index],
+			      FIB_PROTOCOL_IP4, src);
+	  if (0 != ip4_main.mfib_index_by_sw_if_index[sw_if_index])
+	    mfib_table_unlock (ip4_main.mfib_index_by_sw_if_index
+			       [sw_if_index], FIB_PROTOCOL_IP4, msrc);
 
-  BAD_SW_IF_INDEX_LABEL;
+	}
+      else
+	{
+	  /* we need to lock the table now it's inuse */
+	  fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4,
+							 table_id, src);
 
-  REPLY_MACRO (VL_API_SW_INTERFACE_SET_TABLE_REPLY);
+	  mfib_index = mfib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4,
+							   table_id, msrc);
+	}
+
+      ip4_main.fib_index_by_sw_if_index[sw_if_index] = fib_index;
+      ip4_main.mfib_index_by_sw_if_index[sw_if_index] = mfib_index;
+    }
+
+  /*
+   * Temporary. undo the locks from the find and create at the staart
+   */
+  if (0 != table_id)
+    {
+      fib_table_unlock (fib_index, fproto, src);
+      mfib_table_unlock (mfib_index, fproto, msrc);
+    }
+
+  return (0);
 }
 
 static void
diff --git a/src/vnet/ip/ip.h b/src/vnet/ip/ip.h
index 70b4ccd8..7aae73ff 100644
--- a/src/vnet/ip/ip.h
+++ b/src/vnet/ip/ip.h
@@ -184,6 +184,13 @@ void ip_del_all_interface_addresses (vlib_main_t * vm, u32 sw_if_index);
 extern vlib_node_registration_t ip4_inacl_node;
 extern vlib_node_registration_t ip6_inacl_node;
 
+void ip_table_create (fib_protocol_t fproto, uint32_t table_id, u8 is_api);
+
+void ip_table_delete (fib_protocol_t fproto, uint32_t table_id, u8 is_api);
+
+int ip_table_bind (fib_protocol_t fproto,
+		   uint32_t sw_if_index, uint32_t table_id, u8 is_api);
+
 #endif /* included_ip_main_h */
 
 /*
diff --git a/src/vnet/ip/ip4.h b/src/vnet/ip/ip4.h
index 8f9a8e27..decb840b 100644
--- a/src/vnet/ip/ip4.h
+++ b/src/vnet/ip/ip4.h
@@ -72,6 +72,16 @@ typedef struct
   uword function_opaque;
 } ip4_add_del_interface_address_callback_t;
 
+typedef void (ip4_table_bind_function_t)
+  (struct ip4_main_t * im,
+   uword opaque, u32 sw_if_index, u32 new_fib_index, u32 old_fib_index);
+
+typedef struct
+{
+  ip4_table_bind_function_t *function;
+  uword function_opaque;
+} ip4_table_bind_callback_t;
+
 /**
  * @brief IPv4 main type.
  *
@@ -117,6 +127,9 @@ typedef struct ip4_main_t
     ip4_add_del_interface_address_callback_t
     * add_del_interface_address_callbacks;
 
+  /** Functions to call when interface to table biding changes. */
+  ip4_table_bind_callback_t *table_bind_callbacks;
+
   /** Template used to generate IP4 ARP packets. */
   vlib_packet_template_t ip4_arp_request_packet_template;
 
diff --git a/src/vnet/ip/ip4_forward.c b/src/vnet/ip/ip4_forward.c
index 2d48e8a9..ec4287bb 100755
--- a/src/vnet/ip/ip4_forward.c
+++ b/src/vnet/ip/ip4_forward.c
@@ -1198,8 +1198,10 @@ ip4_lookup_init (vlib_main_t * vm)
   ip_lookup_init (&im->lookup_main, /* is_ip6 */ 0);
 
   /* Create FIB with index 0 and table id of 0. */
-  fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, 0);
-  mfib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, 0);
+  fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, 0,
+				     FIB_SOURCE_DEFAULT_ROUTE);
+  mfib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, 0,
+				      MFIB_SOURCE_DEFAULT_ROUTE);
 
   {
     pg_node_t *pn;
@@ -2794,101 +2796,6 @@ VLIB_REGISTER_NODE (ip4_midchain_node) = {
 VLIB_NODE_FUNCTION_MULTIARCH (ip4_midchain_node, ip4_midchain);
 /* *INDENT-ON */
 
-static clib_error_t *
-add_del_interface_table (vlib_main_t * vm,
-			 unformat_input_t * input, vlib_cli_command_t * cmd)
-{
-  vnet_main_t *vnm = vnet_get_main ();
-  ip_interface_address_t *ia;
-  clib_error_t *error = 0;
-  u32 sw_if_index, table_id;
-
-  sw_if_index = ~0;
-
-  if (!unformat_user (input, unformat_vnet_sw_interface, vnm, &sw_if_index))
-    {
-      error = clib_error_return (0, "unknown interface `%U'",
-				 format_unformat_error, input);
-      goto done;
-    }
-
-  if (unformat (input, "%d", &table_id))
-    ;
-  else
-    {
-      error = clib_error_return (0, "expected table id `%U'",
-				 format_unformat_error, input);
-      goto done;
-    }
-
-  /*
-   * If the interface already has in IP address, then a change int
-   * VRF is not allowed. The IP address applied must first be removed.
-   * We do not do that automatically here, since VPP has no knowledge
-   * of whether thoses subnets are valid in the destination VRF.
-   */
-  /* *INDENT-OFF* */
-  foreach_ip_interface_address (&ip4_main.lookup_main,
-                                ia, sw_if_index,
-                                1 /* honor unnumbered */,
-  ({
-      ip4_address_t * a;
-
-      a = ip_interface_address_get_address (&ip4_main.lookup_main, ia);
-      error = clib_error_return (0, "interface %U has address %U",
-                                 format_vnet_sw_if_index_name, vnm,
-                                 sw_if_index,
-                                 format_ip4_address, a);
-      goto done;
-   }));
-   /* *INDENT-ON* */
-
-{
-  ip4_main_t *im = &ip4_main;
-  u32 fib_index;
-
-  fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, table_id);
-
-  vec_validate (im->fib_index_by_sw_if_index, sw_if_index);
-  im->fib_index_by_sw_if_index[sw_if_index] = fib_index;
-
-  fib_index = mfib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, table_id);
-  vec_validate (im->mfib_index_by_sw_if_index, sw_if_index);
-  im->mfib_index_by_sw_if_index[sw_if_index] = fib_index;
-}
-
-done:
-return error;
-}
-
-/*?
- * Place the indicated interface into the supplied IPv4 FIB table (also known
- * as a VRF). If the FIB table does not exist, this command creates it. To
- * display the current IPv4 FIB table, use the command '<em>show ip fib</em>'.
- * FIB table will only be displayed if a route has been added to the table, or
- * an IP Address is assigned to an interface in the table (which adds a route
- * automatically).
- *
- * @note IP addresses added after setting the interface IP table are added to
- * the indicated FIB table. If an IP address is added prior to changing the
- * table then this is an error. The control plane must remove these addresses
- * first and then change the table. VPP will not automatically move the
- * addresses from the old to the new table as it does not know the validity
- * of such a change.
- *
- * @cliexpar
- * Example of how to add an interface to an IPv4 FIB table (where 2 is the table-id):
- * @cliexcmd{set interface ip table GigabitEthernet2/0/0 2}
- ?*/
-/* *INDENT-OFF* */
-VLIB_CLI_COMMAND (set_interface_ip_table_command, static) =
-{
-  .path = "set interface ip table",
-  .function = add_del_interface_table,
-  .short_help = "set interface ip table <interface> <table-id>",
-};
-/* *INDENT-ON* */
-
 int
 ip4_lookup_validate (ip4_address_t * a, u32 fib_index0)
 {
diff --git a/src/vnet/ip/ip4_source_and_port_range_check.c b/src/vnet/ip/ip4_source_and_port_range_check.c
index ae836a11..9aa880ae 100644
--- a/src/vnet/ip/ip4_source_and_port_range_check.c
+++ b/src/vnet/ip/ip4_source_and_port_range_check.c
@@ -1126,6 +1126,14 @@ ip6_source_and_port_range_check_add_del (ip6_address_t * address,
 					 u16 * low_ports,
 					 u16 * high_ports, int is_add)
 {
+  uint32_t fib_index;
+
+  fib_index = fib_table_find (FIB_PROTOCOL_IP4, vrf_id);
+
+  ASSERT (~0 != fib_index);
+
+  fib_table_unlock (fib_index, FIB_PROTOCOL_IP4, FIB_SOURCE_CLASSIFY);
+
   return 0;
 }
 
@@ -1138,7 +1146,8 @@ ip4_source_and_port_range_check_add_del (ip4_address_t * address,
 {
   u32 fib_index;
 
-  fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, vrf_id);
+  fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, vrf_id,
+						 FIB_SOURCE_CLASSIFY);
 
   if (is_add == 0)
     {
diff --git a/src/vnet/ip/ip6.h b/src/vnet/ip/ip6.h
index fa922725..8aef53a9 100644
--- a/src/vnet/ip/ip6.h
+++ b/src/vnet/ip/ip6.h
@@ -103,6 +103,16 @@ typedef struct
   uword function_opaque;
 } ip6_add_del_interface_address_callback_t;
 
+typedef void (ip6_table_bind_function_t)
+  (struct ip6_main_t * im,
+   uword opaque, u32 sw_if_index, u32 new_fib_index, u32 old_fib_index);
+
+typedef struct
+{
+  ip6_table_bind_function_t *function;
+  uword function_opaque;
+} ip6_table_bind_callback_t;
+
 /**
  * Enumeration of the FIB table instance types
  */
@@ -183,6 +193,9 @@ typedef struct ip6_main_t
     ip6_add_del_interface_address_callback_t
     * add_del_interface_address_callbacks;
 
+  /** Functions to call when interface to table biding changes. */
+  ip6_table_bind_callback_t *table_bind_callbacks;
+
   /* Template used to generate IP6 neighbor solicitation packets. */
   vlib_packet_template_t discover_neighbor_packet_template;
 
diff --git a/src/vnet/ip/ip6_forward.c b/src/vnet/ip/ip6_forward.c
index 5832bd0b..1002f6b6 100644
--- a/src/vnet/ip/ip6_forward.c
+++ b/src/vnet/ip/ip6_forward.c
@@ -2999,8 +2999,10 @@ ip6_lookup_init (vlib_main_t * vm)
 			 im->lookup_table_nbuckets, im->lookup_table_size);
 
   /* Create FIB with index 0 and table id of 0. */
-  fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6, 0);
-  mfib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6, 0);
+  fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6, 0,
+				     FIB_SOURCE_DEFAULT_ROUTE);
+  mfib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6, 0,
+				      MFIB_SOURCE_DEFAULT_ROUTE);
 
   {
     pg_node_t *pn;
@@ -3045,103 +3047,6 @@ ip6_lookup_init (vlib_main_t * vm)
 
 VLIB_INIT_FUNCTION (ip6_lookup_init);
 
-static clib_error_t *
-add_del_ip6_interface_table (vlib_main_t * vm,
-			     unformat_input_t * input,
-			     vlib_cli_command_t * cmd)
-{
-  vnet_main_t *vnm = vnet_get_main ();
-  ip_interface_address_t *ia;
-  clib_error_t *error = 0;
-  u32 sw_if_index, table_id;
-
-  sw_if_index = ~0;
-
-  if (!unformat_user (input, unformat_vnet_sw_interface, vnm, &sw_if_index))
-    {
-      error = clib_error_return (0, "unknown interface `%U'",
-				 format_unformat_error, input);
-      goto done;
-    }
-
-  if (unformat (input, "%d", &table_id))
-    ;
-  else
-    {
-      error = clib_error_return (0, "expected table id `%U'",
-				 format_unformat_error, input);
-      goto done;
-    }
-
-  /*
-   * If the interface already has in IP address, then a change int
-   * VRF is not allowed. The IP address applied must first be removed.
-   * We do not do that automatically here, since VPP has no knowledge
-   * of whether thoses subnets are valid in the destination VRF.
-   */
-  /* *INDENT-OFF* */
-  foreach_ip_interface_address (&ip6_main.lookup_main,
-                                ia, sw_if_index,
-                                1 /* honor unnumbered */,
-  ({
-      ip4_address_t * a;
-
-      a = ip_interface_address_get_address (&ip6_main.lookup_main, ia);
-      error = clib_error_return (0, "interface %U has address %U",
-                                 format_vnet_sw_if_index_name, vnm,
-                                 sw_if_index,
-                                 format_ip6_address, a);
-      goto done;
-  }));
-  /* *INDENT-ON* */
-
-  {
-    u32 fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6,
-						       table_id);
-
-    vec_validate (ip6_main.fib_index_by_sw_if_index, sw_if_index);
-    ip6_main.fib_index_by_sw_if_index[sw_if_index] = fib_index;
-
-    fib_index = mfib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6,
-						    table_id);
-
-    vec_validate (ip6_main.mfib_index_by_sw_if_index, sw_if_index);
-    ip6_main.mfib_index_by_sw_if_index[sw_if_index] = fib_index;
-  }
-
-
-done:
-  return error;
-}
-
-/*?
- * Place the indicated interface into the supplied IPv6 FIB table (also known
- * as a VRF). If the FIB table does not exist, this command creates it. To
- * display the current IPv6 FIB table, use the command '<em>show ip6 fib</em>'.
- * FIB table will only be displayed if a route has been added to the table, or
- * an IP Address is assigned to an interface in the table (which adds a route
- * automatically).
- *
- * @note IP addresses added after setting the interface IP table are added to
- * the indicated FIB table. If an IP address is added prior to changing the
- * table then this is an error. The control plane must remove these addresses
- * first and then change the table. VPP will not automatically move the
- * addresses from the old to the new table as it does not know the validity
- * of such a change.
- *
- * @cliexpar
- * Example of how to add an interface to an IPv6 FIB table (where 2 is the table-id):
- * @cliexcmd{set interface ip6 table GigabitEthernet2/0/0 2}
- ?*/
-/* *INDENT-OFF* */
-VLIB_CLI_COMMAND (set_interface_ip6_table_command, static) =
-{
-  .path = "set interface ip6 table",
-  .function = add_del_ip6_interface_table,
-  .short_help = "set interface ip6 table <interface> <table-id>"
-};
-/* *INDENT-ON* */
-
 void
 ip6_link_local_address_from_ethernet_mac_address (ip6_address_t * ip,
 						  u8 * mac)
diff --git a/src/vnet/ip/ip6_neighbor.c b/src/vnet/ip/ip6_neighbor.c
index 62cf23ac..56f33ac8 100644
--- a/src/vnet/ip/ip6_neighbor.c
+++ b/src/vnet/ip/ip6_neighbor.c
@@ -250,6 +250,26 @@ format_ip6_neighbor_ip6_entry (u8 * s, va_list * va)
   return s;
 }
 
+static void
+ip6_neighbor_adj_fib_remove (ip6_neighbor_t * n, uint32_t fib_index)
+{
+  if (FIB_NODE_INDEX_INVALID != n->fib_entry_index)
+    {
+      fib_prefix_t pfx = {
+	.fp_len = 128,
+	.fp_proto = FIB_PROTOCOL_IP6,
+	.fp_addr.ip6 = n->key.ip6_address,
+      };
+      fib_table_entry_path_remove (fib_index,
+				   &pfx,
+				   FIB_SOURCE_ADJ,
+				   DPO_PROTO_IP6,
+				   &pfx.fp_addr,
+				   n->key.sw_if_index, ~0,
+				   1, FIB_ROUTE_PATH_FLAG_NONE);
+    }
+}
+
 static clib_error_t *
 ip6_neighbor_sw_interface_up_down (vnet_main_t * vnm,
 				   u32 sw_if_index, u32 flags)
@@ -273,22 +293,10 @@ ip6_neighbor_sw_interface_up_down (vnet_main_t * vnm,
 	{
 	  n = pool_elt_at_index (nm->neighbor_pool, to_delete[i]);
 	  mhash_unset (&nm->neighbor_index_by_key, &n->key, 0);
-	  if (FIB_NODE_INDEX_INVALID != n->fib_entry_index)
-	    {
-	      fib_prefix_t pfx = {
-		.fp_len = 128,
-		.fp_proto = FIB_PROTOCOL_IP6,
-		.fp_addr.ip6 = n->key.ip6_address,
-	      };
-	      fib_table_entry_path_remove
-		(ip6_fib_table_get_index_for_sw_if_index (n->key.sw_if_index),
-		 &pfx,
-		 FIB_SOURCE_ADJ,
-		 DPO_PROTO_IP6,
-		 &pfx.fp_addr,
-		 n->key.sw_if_index, ~0, 1, FIB_ROUTE_PATH_FLAG_NONE);
-	      pool_put (nm->neighbor_pool, n);
-	    }
+	  ip6_neighbor_adj_fib_remove (n,
+				       ip6_fib_table_get_index_for_sw_if_index
+				       (n->key.sw_if_index));
+	  pool_put (nm->neighbor_pool, n);
 	}
       vec_free (to_delete);
     }
@@ -579,6 +587,24 @@ ip6_ethernet_update_adjacency (vnet_main_t * vnm, u32 sw_if_index, u32 ai)
     }
 }
 
+
+static void
+ip6_neighbor_adj_fib_add (ip6_neighbor_t * n, uint32_t fib_index)
+{
+  fib_prefix_t pfx = {
+    .fp_len = 128,
+    .fp_proto = FIB_PROTOCOL_IP6,
+    .fp_addr.ip6 = n->key.ip6_address,
+  };
+
+  n->fib_entry_index =
+    fib_table_entry_path_add (fib_index, &pfx, FIB_SOURCE_ADJ,
+			      FIB_ENTRY_FLAG_ATTACHED,
+			      DPO_PROTO_IP6, &pfx.fp_addr,
+			      n->key.sw_if_index, ~0, 1, NULL,
+			      FIB_ROUTE_PATH_FLAG_NONE);
+}
+
 int
 vnet_set_ip6_ethernet_neighbor (vlib_main_t * vm,
 				u32 sw_if_index,
@@ -633,21 +659,9 @@ vnet_set_ip6_ethernet_neighbor (vlib_main_t * vm,
        */
       if (!is_no_fib_entry)
 	{
-	  fib_prefix_t pfx = {
-	    .fp_len = 128,
-	    .fp_proto = FIB_PROTOCOL_IP6,
-	    .fp_addr.ip6 = k.ip6_address,
-	  };
-	  u32 fib_index;
-
-	  fib_index =
-	    ip6_fib_table_get_index_for_sw_if_index (n->key.sw_if_index);
-	  n->fib_entry_index =
-	    fib_table_entry_path_add (fib_index, &pfx, FIB_SOURCE_ADJ,
-				      FIB_ENTRY_FLAG_ATTACHED,
-				      DPO_PROTO_IP6, &pfx.fp_addr,
-				      n->key.sw_if_index, ~0, 1, NULL,
-				      FIB_ROUTE_PATH_FLAG_NONE);
+	  ip6_neighbor_adj_fib_add (n,
+				    ip6_fib_table_get_index_for_sw_if_index
+				    (n->key.sw_if_index));
 	}
       else
 	{
@@ -3843,6 +3857,33 @@ ip6_set_neighbor_limit (u32 neighbor_limit)
   return 0;
 }
 
+static void
+ip6_neighbor_table_bind (ip6_main_t * im,
+			 uword opaque,
+			 u32 sw_if_index,
+			 u32 new_fib_index, u32 old_fib_index)
+{
+  ip6_neighbor_main_t *nm = &ip6_neighbor_main;
+  ip6_neighbor_t *n = NULL;
+  u32 i, *to_re_add = 0;
+
+  /* *INDENT-OFF* */
+  pool_foreach (n, nm->neighbor_pool,
+  ({
+    if (n->key.sw_if_index == sw_if_index)
+      vec_add1 (to_re_add, n - nm->neighbor_pool);
+  }));
+  /* *INDENT-ON* */
+
+  for (i = 0; i < vec_len (to_re_add); i++)
+    {
+      n = pool_elt_at_index (nm->neighbor_pool, to_re_add[i]);
+      ip6_neighbor_adj_fib_remove (n, old_fib_index);
+      ip6_neighbor_adj_fib_add (n, new_fib_index);
+    }
+  vec_free (to_re_add);
+}
+
 static clib_error_t *
 ip6_neighbor_init (vlib_main_t * vm)
 {
@@ -3874,6 +3915,11 @@ ip6_neighbor_init (vlib_main_t * vm)
   cb.function_opaque = 0;
   vec_add1 (im->add_del_interface_address_callbacks, cb);
 
+  ip6_table_bind_callback_t cbt;
+  cbt.function = ip6_neighbor_table_bind;
+  cbt.function_opaque = 0;
+  vec_add1 (im->table_bind_callbacks, cbt);
+
   mhash_init (&nm->pending_resolutions_by_address,
 	      /* value size */ sizeof (uword),
 	      /* key size */ sizeof (ip6_address_t));
diff --git a/src/vnet/ip/ip_api.c b/src/vnet/ip/ip_api.c
index bba65ab4..384ec3e0 100644
--- a/src/vnet/ip/ip_api.c
+++ b/src/vnet/ip/ip_api.c
@@ -699,12 +699,58 @@ vl_api_ip_neighbor_add_del_t_handler (vl_api_ip_neighbor_add_del_t * mp,
   REPLY_MACRO (VL_API_IP_NEIGHBOR_ADD_DEL_REPLY);
 }
 
+void
+ip_table_delete (fib_protocol_t fproto, u32 table_id, u8 is_api)
+{
+  u32 fib_index, mfib_index;
+
+  /*
+   * ignore action on the default table - this is always present
+   * and cannot be added nor deleted from the API
+   */
+  if (0 != table_id)
+    {
+      /*
+       * The API holds only one lock on the table.
+       * i.e. it can be added many times via the API but needs to be
+       * deleted only once.
+       * The FIB index for unicast and multicast is not necessarily the
+       * same, since internal VPP systesm (like LISP and SR) create
+       * their own unicast tables.
+       */
+      fib_index = fib_table_find (fproto, table_id);
+      mfib_index = mfib_table_find (fproto, table_id);
+
+      if (~0 != fib_index)
+	{
+	  fib_table_unlock (fib_index, fproto,
+			    (is_api ? FIB_SOURCE_API : FIB_SOURCE_CLI));
+	}
+      if (~0 != mfib_index)
+	{
+	  mfib_table_unlock (mfib_index, fproto,
+			     (is_api ? MFIB_SOURCE_API : MFIB_SOURCE_CLI));
+	}
+    }
+}
+
 void
 vl_api_ip_table_add_del_t_handler (vl_api_ip_table_add_del_t * mp)
 {
   vl_api_ip_table_add_del_reply_t *rmp;
+  fib_protocol_t fproto = (mp->is_ipv6 ? FIB_PROTOCOL_IP6 : FIB_PROTOCOL_IP4);
+  u32 table_id = ntohl (mp->table_id);
   int rv = 0;
 
+  if (mp->is_add)
+    {
+      ip_table_create (fproto, table_id, 1);
+    }
+  else
+    {
+      ip_table_delete (fproto, table_id, 1);
+    }
+
   REPLY_MACRO (VL_API_IP_TABLE_ADD_DEL_REPLY);
 }
 
@@ -866,18 +912,21 @@ add_del_route_check (fib_protocol_t table_proto,
 		     u32 next_hop_sw_if_index,
 		     dpo_proto_t next_hop_table_proto,
 		     u32 next_hop_table_id,
-		     u8 create_missing_tables,
 		     u8 is_rpf_id, u32 * fib_index, u32 * next_hop_fib_index)
 {
   vnet_main_t *vnm = vnet_get_main ();
 
+  /* Temporaray whilst I do the CSIT dance */
+  u8 create_missing_tables = 1;
+
   *fib_index = fib_table_find (table_proto, ntohl (table_id));
   if (~0 == *fib_index)
     {
       if (create_missing_tables)
 	{
 	  *fib_index = fib_table_find_or_create_and_lock (table_proto,
-							  ntohl (table_id));
+							  ntohl (table_id),
+							  FIB_SOURCE_API);
 	}
       else
 	{
@@ -918,12 +967,14 @@ add_del_route_check (fib_protocol_t table_proto,
 		*next_hop_fib_index =
 		  mfib_table_find_or_create_and_lock (fib_nh_proto,
 						      ntohl
-						      (next_hop_table_id));
+						      (next_hop_table_id),
+						      MFIB_SOURCE_API);
 	      else
 		*next_hop_fib_index =
 		  fib_table_find_or_create_and_lock (fib_nh_proto,
 						     ntohl
-						     (next_hop_table_id));
+						     (next_hop_table_id),
+						     FIB_SOURCE_API);
 	    }
 	  else
 	    {
@@ -948,8 +999,7 @@ ip4_add_del_route_t_handler (vl_api_ip_add_del_route_t * mp)
 			    mp->next_hop_sw_if_index,
 			    DPO_PROTO_IP4,
 			    mp->next_hop_table_id,
-			    mp->create_vrf_if_needed, 0,
-			    &fib_index, &next_hop_fib_index);
+			    0, &fib_index, &next_hop_fib_index);
 
   if (0 != rv)
     return (rv);
@@ -1008,8 +1058,7 @@ ip6_add_del_route_t_handler (vl_api_ip_add_del_route_t * mp)
 			    mp->next_hop_sw_if_index,
 			    DPO_PROTO_IP6,
 			    mp->next_hop_table_id,
-			    mp->create_vrf_if_needed, 0,
-			    &fib_index, &next_hop_fib_index);
+			    0, &fib_index, &next_hop_fib_index);
 
   if (0 != rv)
     return (rv);
@@ -1074,27 +1123,57 @@ vl_api_ip_add_del_route_t_handler (vl_api_ip_add_del_route_t * mp)
   REPLY_MACRO (VL_API_IP_ADD_DEL_ROUTE_REPLY);
 }
 
+void
+ip_table_create (fib_protocol_t fproto, u32 table_id, u8 is_api)
+{
+  u32 fib_index, mfib_index;
+
+  /*
+   * ignore action on the default table - this is always present
+   * and cannot be added nor deleted from the API
+   */
+  if (0 != table_id)
+    {
+      /*
+       * The API holds only one lock on the table.
+       * i.e. it can be added many times via the API but needs to be
+       * deleted only once.
+       * The FIB index for unicast and multicast is not necessarily the
+       * same, since internal VPP systesm (like LISP and SR) create
+       * their own unicast tables.
+       */
+      fib_index = fib_table_find (fproto, table_id);
+      mfib_index = mfib_table_find (fproto, table_id);
+
+      if (~0 == fib_index)
+	{
+	  fib_table_find_or_create_and_lock (fproto, table_id,
+					     (is_api ?
+					      FIB_SOURCE_API :
+					      FIB_SOURCE_CLI));
+	}
+      if (~0 == mfib_index)
+	{
+	  mfib_table_find_or_create_and_lock (fproto, table_id,
+					      (is_api ?
+					       MFIB_SOURCE_API :
+					       MFIB_SOURCE_CLI));
+	}
+    }
+}
+
 static int
 add_del_mroute_check (fib_protocol_t table_proto,
 		      u32 table_id,
-		      u32 next_hop_sw_if_index,
-		      u8 is_local, u8 create_missing_tables, u32 * fib_index)
+		      u32 next_hop_sw_if_index, u8 is_local, u32 * fib_index)
 {
   vnet_main_t *vnm = vnet_get_main ();
 
   *fib_index = mfib_table_find (table_proto, ntohl (table_id));
   if (~0 == *fib_index)
     {
-      if (create_missing_tables)
-	{
-	  *fib_index = mfib_table_find_or_create_and_lock (table_proto,
-							   ntohl (table_id));
-	}
-      else
-	{
-	  /* No such VRF, and we weren't asked to create one */
-	  return VNET_API_ERROR_NO_SUCH_FIB;
-	}
+      /* No such table */
+      return VNET_API_ERROR_NO_SUCH_FIB;
     }
 
   if (~0 != ntohl (next_hop_sw_if_index))
@@ -1163,8 +1242,7 @@ api_mroute_add_del_t_handler (vl_api_ip_mroute_add_del_t * mp)
   rv = add_del_mroute_check (fproto,
 			     mp->table_id,
 			     mp->next_hop_sw_if_index,
-			     mp->is_local,
-			     mp->create_vrf_if_needed, &fib_index);
+			     mp->is_local, &fib_index);
 
   if (0 != rv)
     return (rv);
diff --git a/src/vnet/ip/lookup.c b/src/vnet/ip/lookup.c
index 5537bb04..667c6791 100755
--- a/src/vnet/ip/lookup.c
+++ b/src/vnet/ip/lookup.c
@@ -687,6 +687,78 @@ done:
   return error;
 }
 
+clib_error_t *
+vnet_ip_table_cmd (vlib_main_t * vm,
+		   unformat_input_t * main_input,
+		   vlib_cli_command_t * cmd, fib_protocol_t fproto)
+{
+  unformat_input_t _line_input, *line_input = &_line_input;
+  clib_error_t *error = NULL;
+  u32 table_id, is_add;
+
+  is_add = 1;
+  table_id = ~0;
+
+  /* Get a line of input. */
+  if (!unformat_user (main_input, unformat_line_input, line_input))
+    return 0;
+
+  while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
+    {
+      if (unformat (line_input, "%d", &table_id))
+	;
+      else if (unformat (line_input, "del"))
+	is_add = 0;
+      else if (unformat (line_input, "add"))
+	is_add = 1;
+      else
+	{
+	  error = unformat_parse_error (line_input);
+	  goto done;
+	}
+    }
+
+  if (~0 == table_id)
+    {
+      error = clib_error_return (0, "No table id");
+      goto done;
+    }
+  else if (0 == table_id)
+    {
+      error = clib_error_return (0, "Can't change the default table");
+      goto done;
+    }
+  else
+    {
+      if (is_add)
+	{
+	  ip_table_create (fproto, table_id, 0);
+	}
+      else
+	{
+	  ip_table_delete (fproto, table_id, 0);
+	}
+    }
+
+done:
+  unformat_free (line_input);
+  return error;
+}
+
+clib_error_t *
+vnet_ip4_table_cmd (vlib_main_t * vm,
+		    unformat_input_t * main_input, vlib_cli_command_t * cmd)
+{
+  return (vnet_ip_table_cmd (vm, main_input, cmd, FIB_PROTOCOL_IP4));
+}
+
+clib_error_t *
+vnet_ip6_table_cmd (vlib_main_t * vm,
+		    unformat_input_t * main_input, vlib_cli_command_t * cmd)
+{
+  return (vnet_ip_table_cmd (vm, main_input, cmd, FIB_PROTOCOL_IP6));
+}
+
 /* *INDENT-OFF* */
 VLIB_CLI_COMMAND (vlib_cli_ip_command, static) = {
   .path = "ip",
@@ -749,6 +821,159 @@ VLIB_CLI_COMMAND (ip_route_command, static) = {
   .function = vnet_ip_route_cmd,
   .is_mp_safe = 1,
 };
+
+/* *INDENT-ON* */
+/*?
+ * This command is used to add or delete IPv4  Tables. All
+ * Tables must be explicitly added before that can be used. Creating a
+ * table will add both unicast and multicast FIBs
+ *
+ ?*/
+/* *INDENT-OFF* */
+VLIB_CLI_COMMAND (ip4_table_command, static) = {
+  .path = "ip table",
+  .short_help = "ip table [add|del] <table-id>",
+  .function = vnet_ip4_table_cmd,
+  .is_mp_safe = 1,
+};
+/* *INDENT-ON* */
+
+/* *INDENT-ON* */
+/*?
+ * This command is used to add or delete IPv4  Tables. All
+ * Tables must be explicitly added before that can be used. Creating a
+ * table will add both unicast and multicast FIBs
+ *
+ ?*/
+/* *INDENT-OFF* */
+VLIB_CLI_COMMAND (ip6_table_command, static) = {
+  .path = "ip6 table",
+  .short_help = "ip6 table [add|del] <table-id>",
+  .function = vnet_ip6_table_cmd,
+  .is_mp_safe = 1,
+};
+
+static clib_error_t *
+ip_table_bind_cmd (vlib_main_t * vm,
+                   unformat_input_t * input,
+                   vlib_cli_command_t * cmd,
+                   fib_protocol_t fproto)
+{
+  vnet_main_t *vnm = vnet_get_main ();
+  clib_error_t *error = 0;
+  u32 sw_if_index, table_id;
+  int rv;
+
+  sw_if_index = ~0;
+
+  if (!unformat_user (input, unformat_vnet_sw_interface, vnm, &sw_if_index))
+    {
+      error = clib_error_return (0, "unknown interface `%U'",
+				 format_unformat_error, input);
+      goto done;
+    }
+
+  if (unformat (input, "%d", &table_id))
+    ;
+  else
+    {
+      error = clib_error_return (0, "expected table id `%U'",
+				 format_unformat_error, input);
+      goto done;
+    }
+
+  rv = ip_table_bind (fproto, sw_if_index, table_id, 0);
+
+  if (VNET_API_ERROR_ADDRESS_FOUND_FOR_INTERFACE == rv)
+    {
+      error = clib_error_return (0, "IP addresses are still present on %U",
+                                 format_vnet_sw_if_index_name,
+                                 vnet_get_main(),
+                                 sw_if_index);
+    }
+  else if (VNET_API_ERROR_NO_SUCH_FIB == rv)
+    {
+      error = clib_error_return (0, "no such table %d", table_id);
+    }
+  else if (0 != rv)
+    {
+      error = clib_error_return (0, "unknown error");
+    }
+
+ done:
+  return error;
+}
+
+static clib_error_t *
+ip4_table_bind_cmd (vlib_main_t * vm,
+                    unformat_input_t * input,
+                    vlib_cli_command_t * cmd)
+{
+  return (ip_table_bind_cmd (vm , input, cmd, FIB_PROTOCOL_IP4));
+}
+
+static clib_error_t *
+ip6_table_bind_cmd (vlib_main_t * vm,
+                    unformat_input_t * input,
+                    vlib_cli_command_t * cmd)
+{
+  return (ip_table_bind_cmd (vm , input, cmd, FIB_PROTOCOL_IP6));
+}
+
+/*?
+ * Place the indicated interface into the supplied IPv4 FIB table (also known
+ * as a VRF). If the FIB table does not exist, this command creates it. To
+ * display the current IPv4 FIB table, use the command '<em>show ip fib</em>'.
+ * FIB table will only be displayed if a route has been added to the table, or
+ * an IP Address is assigned to an interface in the table (which adds a route
+ * automatically).
+ *
+ * @note IP addresses added after setting the interface IP table are added to
+ * the indicated FIB table. If an IP address is added prior to changing the
+ * table then this is an error. The control plane must remove these addresses
+ * first and then change the table. VPP will not automatically move the
+ * addresses from the old to the new table as it does not know the validity
+ * of such a change.
+ *
+ * @cliexpar
+ * Example of how to add an interface to an IPv4 FIB table (where 2 is the table-id):
+ * @cliexcmd{set interface ip table GigabitEthernet2/0/0 2}
+ ?*/
+/* *INDENT-OFF* */
+VLIB_CLI_COMMAND (set_interface_ip_table_command, static) =
+{
+  .path = "set interface ip table",
+  .function = ip4_table_bind_cmd,
+  .short_help = "set interface ip table <interface> <table-id>",
+};
+/* *INDENT-ON* */
+
+/*?
+ * Place the indicated interface into the supplied IPv6 FIB table (also known
+ * as a VRF). If the FIB table does not exist, this command creates it. To
+ * display the current IPv6 FIB table, use the command '<em>show ip6 fib</em>'.
+ * FIB table will only be displayed if a route has been added to the table, or
+ * an IP Address is assigned to an interface in the table (which adds a route
+ * automatically).
+ *
+ * @note IP addresses added after setting the interface IP table are added to
+ * the indicated FIB table. If an IP address is added prior to changing the
+ * table then this is an error. The control plane must remove these addresses
+ * first and then change the table. VPP will not automatically move the
+ * addresses from the old to the new table as it does not know the validity
+ * of such a change.
+ *
+ * @cliexpar
+ * Example of how to add an interface to an IPv6 FIB table (where 2 is the table-id):
+ * @cliexcmd{set interface ip6 table GigabitEthernet2/0/0 2}
+ ?*/
+/* *INDENT-OFF* */
+VLIB_CLI_COMMAND (set_interface_ip6_table_command, static) =
+{
+  .path = "set interface ip6 table",
+  .function = ip6_table_bind_cmd,
+  .short_help = "set interface ip6 table <interface> <table-id>"
+};
 /* *INDENT-ON* */
 
 clib_error_t *
diff --git a/src/vnet/lisp-gpe/interface.c b/src/vnet/lisp-gpe/interface.c
index e832c23f..a0c05e85 100644
--- a/src/vnet/lisp-gpe/interface.c
+++ b/src/vnet/lisp-gpe/interface.c
@@ -505,12 +505,14 @@ lisp_gpe_iface_set_table (u32 sw_if_index, u32 table_id)
 {
   fib_node_index_t fib_index;
 
-  fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, table_id);
+  fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, table_id,
+						 FIB_SOURCE_LISP);
   vec_validate (ip4_main.fib_index_by_sw_if_index, sw_if_index);
   ip4_main.fib_index_by_sw_if_index[sw_if_index] = fib_index;
   ip4_sw_interface_enable_disable (sw_if_index, 1);
 
-  fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6, table_id);
+  fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6, table_id,
+						 FIB_SOURCE_LISP);
   vec_validate (ip6_main.fib_index_by_sw_if_index, sw_if_index);
   ip6_main.fib_index_by_sw_if_index[sw_if_index] = fib_index;
   ip6_sw_interface_enable_disable (sw_if_index, 1);
@@ -530,7 +532,7 @@ lisp_gpe_tenant_del_default_routes (u32 table_id)
 
     fib_index = fib_table_find (prefix.fp_proto, table_id);
     fib_table_entry_special_remove (fib_index, &prefix, FIB_SOURCE_LISP);
-    fib_table_unlock (fib_index, prefix.fp_proto);
+    fib_table_unlock (fib_index, prefix.fp_proto, FIB_SOURCE_LISP);
   }
 }
 
@@ -549,7 +551,8 @@ lisp_gpe_tenant_add_default_routes (u32 table_id)
     /*
      * Add a deafult route that results in a control plane punt DPO
      */
-    fib_index = fib_table_find_or_create_and_lock (prefix.fp_proto, table_id);
+    fib_index = fib_table_find_or_create_and_lock (prefix.fp_proto, table_id,
+						   FIB_SOURCE_LISP);
     fib_table_entry_special_dpo_add (fib_index, &prefix, FIB_SOURCE_LISP,
 				     FIB_ENTRY_FLAG_EXCLUSIVE,
 				     lisp_cp_dpo_get (fib_proto_to_dpo
diff --git a/src/vnet/lisp-gpe/lisp_gpe_fwd_entry.c b/src/vnet/lisp-gpe/lisp_gpe_fwd_entry.c
index d7d3cb86..0a8dc039 100644
--- a/src/vnet/lisp-gpe/lisp_gpe_fwd_entry.c
+++ b/src/vnet/lisp-gpe/lisp_gpe_fwd_entry.c
@@ -66,6 +66,7 @@ ip_dst_fib_add_route (u32 dst_fib_index, const ip_prefix_t * dst_prefix)
       /* create a new src FIB.  */
       src_fib_index =
 	fib_table_create_and_lock (dst_fib_prefix.fp_proto,
+				   FIB_SOURCE_LISP,
 				   "LISP-src for [%d,%U]",
 				   dst_fib_index,
 				   format_fib_prefix, &dst_fib_prefix);
@@ -180,7 +181,8 @@ ip_src_dst_fib_del_route (u32 src_fib_index,
        */
       fib_table_entry_special_remove (dst_fib_index,
 				      &dst_fib_prefix, FIB_SOURCE_LISP);
-      fib_table_unlock (src_fib_index, src_fib_prefix.fp_proto);
+      fib_table_unlock (src_fib_index, src_fib_prefix.fp_proto,
+			FIB_SOURCE_LISP);
     }
 }
 
@@ -544,7 +546,8 @@ add_ip_fwd_entry (lisp_gpe_main_t * lgm,
   lfe->tenant = lisp_gpe_tenant_find_or_create (lfe->key->vni);
   lfe->eid_table_id = a->table_id;
   lfe->eid_fib_index = fib_table_find_or_create_and_lock (fproto,
-							  lfe->eid_table_id);
+							  lfe->eid_table_id,
+							  FIB_SOURCE_LISP);
   lfe->is_src_dst = a->is_src_dst;
 
   if (LISP_GPE_FWD_ENTRY_TYPE_NEGATIVE != lfe->type)
@@ -578,7 +581,7 @@ del_ip_fwd_entry_i (lisp_gpe_main_t * lgm, lisp_gpe_fwd_entry_t * lfe)
 
   fproto = (IP4 == ip_prefix_version (&fid_addr_ippref (&lfe->key->rmt)) ?
 	    FIB_PROTOCOL_IP4 : FIB_PROTOCOL_IP6);
-  fib_table_unlock (lfe->eid_fib_index, fproto);
+  fib_table_unlock (lfe->eid_fib_index, fproto, FIB_SOURCE_LISP);
 
   hash_unset_mem (lgm->lisp_gpe_fwd_entries, lfe->key);
   clib_mem_free (lfe->key);
diff --git a/src/vnet/lisp-gpe/lisp_gpe_sub_interface.c b/src/vnet/lisp-gpe/lisp_gpe_sub_interface.c
index b234d9dc..26664f53 100644
--- a/src/vnet/lisp-gpe/lisp_gpe_sub_interface.c
+++ b/src/vnet/lisp-gpe/lisp_gpe_sub_interface.c
@@ -89,13 +89,15 @@ lisp_gpe_sub_interface_set_table (u32 sw_if_index, u32 table_id)
 {
   fib_node_index_t fib_index;
 
-  fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, table_id);
+  fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, table_id,
+						 FIB_SOURCE_LISP);
   ASSERT (FIB_NODE_INDEX_INVALID != fib_index);
 
   vec_validate (ip4_main.fib_index_by_sw_if_index, sw_if_index);
   ip4_main.fib_index_by_sw_if_index[sw_if_index] = fib_index;
 
-  fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6, table_id);
+  fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6, table_id,
+						 FIB_SOURCE_LISP);
   ASSERT (FIB_NODE_INDEX_INVALID != fib_index);
 
   vec_validate (ip6_main.fib_index_by_sw_if_index, sw_if_index);
@@ -105,9 +107,13 @@ lisp_gpe_sub_interface_set_table (u32 sw_if_index, u32 table_id)
 static void
 lisp_gpe_sub_interface_unset_table (u32 sw_if_index, u32 table_id)
 {
+  fib_table_unlock (ip4_main.fib_index_by_sw_if_index[sw_if_index],
+		    FIB_PROTOCOL_IP4, FIB_SOURCE_LISP);
   ip4_main.fib_index_by_sw_if_index[sw_if_index] = 0;
   ip4_sw_interface_enable_disable (sw_if_index, 0);
 
+  fib_table_unlock (ip6_main.fib_index_by_sw_if_index[sw_if_index],
+		    FIB_PROTOCOL_IP6, FIB_SOURCE_LISP);
   ip6_main.fib_index_by_sw_if_index[sw_if_index] = 0;
   ip6_sw_interface_enable_disable (sw_if_index, 0);
 }
@@ -185,6 +191,7 @@ lisp_gpe_sub_interface_unlock (index_t l3si)
 
   l3s = lisp_gpe_sub_interface_get_i (l3si);
 
+  ASSERT (0 != l3s->locks);
   l3s->locks--;
 
   if (0 == l3s->locks)
diff --git a/src/vnet/mfib/ip4_mfib.c b/src/vnet/mfib/ip4_mfib.c
index 1849a3a4..b2482580 100644
--- a/src/vnet/mfib/ip4_mfib.c
+++ b/src/vnet/mfib/ip4_mfib.c
@@ -33,7 +33,8 @@ static const mfib_prefix_t ip4_specials[] = {
 };
 
 static u32
-ip4_create_mfib_with_table_id (u32 table_id)
+ip4_create_mfib_with_table_id (u32 table_id,
+                               mfib_source_t src)
 {
     mfib_table_t *mfib_table;
 
@@ -53,7 +54,7 @@ ip4_create_mfib_with_table_id (u32 table_id)
         mfib_table->v4.table_id =
             table_id;
 
-    mfib_table_lock(mfib_table->mft_index, FIB_PROTOCOL_IP4);
+    mfib_table_lock(mfib_table->mft_index, FIB_PROTOCOL_IP4, src);
 
     /*
      * add the special entries into the new FIB
@@ -113,14 +114,15 @@ ip4_mfib_table_destroy (ip4_mfib_t *mfib)
 }
 
 u32
-ip4_mfib_table_find_or_create_and_lock (u32 table_id)
+ip4_mfib_table_find_or_create_and_lock (u32 table_id,
+                                        mfib_source_t src)
 {
     u32 index;
 
     index = ip4_mfib_index_from_table_id(table_id);
     if (~0 == index)
-        return ip4_create_mfib_with_table_id(table_id);
-    mfib_table_lock(index, FIB_PROTOCOL_IP4);
+        return ip4_create_mfib_with_table_id(table_id, src);
+    mfib_table_lock(index, FIB_PROTOCOL_IP4, src);
 
     return (index);
 }
diff --git a/src/vnet/mfib/ip4_mfib.h b/src/vnet/mfib/ip4_mfib.h
index ea682651..e31fb744 100644
--- a/src/vnet/mfib/ip4_mfib.h
+++ b/src/vnet/mfib/ip4_mfib.h
@@ -72,8 +72,9 @@ ip4_mfib_get (u32 index)
  * @returns A pointer to the retrieved or created fib.
  *
  */
-extern u32 ip4_mfib_table_find_or_create_and_lock(u32 table_id);
-extern u32 ip4_mfib_table_create_and_lock(void);
+extern u32 ip4_mfib_table_find_or_create_and_lock(u32 table_id,
+                                                  mfib_source_t src);
+extern u32 ip4_mfib_table_create_and_lock(mfib_source_t src);
 
 static inline
 u32 ip4_mfib_index_from_table_id (u32 table_id)
diff --git a/src/vnet/mfib/ip6_mfib.c b/src/vnet/mfib/ip6_mfib.c
index 5e48e919..e4861330 100644
--- a/src/vnet/mfib/ip6_mfib.c
+++ b/src/vnet/mfib/ip6_mfib.c
@@ -151,7 +151,8 @@ static const ip6_mfib_special_t ip6_mfib_specials[] =
 
 
 static u32
-ip6_create_mfib_with_table_id (u32 table_id)
+ip6_create_mfib_with_table_id (u32 table_id,
+                               mfib_source_t src)
 {
     mfib_table_t *mfib_table;
     mfib_prefix_t pfx = {
@@ -182,7 +183,7 @@ ip6_create_mfib_with_table_id (u32 table_id)
         mfib_table->v6.table_id =
             table_id;
 
-    mfib_table_lock(mfib_table->mft_index, FIB_PROTOCOL_IP6);
+    mfib_table_lock(mfib_table->mft_index, FIB_PROTOCOL_IP6, src);
 
     mfib_table->v6.rhead =
         clib_mem_alloc_aligned (sizeof(*mfib_table->v6.rhead),
@@ -297,14 +298,15 @@ ip6_mfib_interface_enable_disable (u32 sw_if_index, int is_enable)
 }
 
 u32
-ip6_mfib_table_find_or_create_and_lock (u32 table_id)
+ip6_mfib_table_find_or_create_and_lock (u32 table_id,
+                                        mfib_source_t src)
 {
     u32 index;
 
     index = ip6_mfib_index_from_table_id(table_id);
     if (~0 == index)
-        return ip6_create_mfib_with_table_id(table_id);
-    mfib_table_lock(index, FIB_PROTOCOL_IP6);
+        return ip6_create_mfib_with_table_id(table_id, src);
+    mfib_table_lock(index, FIB_PROTOCOL_IP6, src);
 
     return (index);
 }
diff --git a/src/vnet/mfib/ip6_mfib.h b/src/vnet/mfib/ip6_mfib.h
index adaa7ec2..ea81b553 100644
--- a/src/vnet/mfib/ip6_mfib.h
+++ b/src/vnet/mfib/ip6_mfib.h
@@ -79,8 +79,9 @@ ip6_mfib_get (u32 index)
  * @returns A pointer to the retrieved or created fib.
  *
  */
-extern u32 ip6_mfib_table_find_or_create_and_lock(u32 table_id);
-extern u32 ip6_mfib_table_create_and_lock(void);
+extern u32 ip6_mfib_table_find_or_create_and_lock(u32 table_id,
+                                                  mfib_source_t src);
+extern u32 ip6_mfib_table_create_and_lock(mfib_source_t src);
 
 
 static inline
diff --git a/src/vnet/mfib/mfib_entry.c b/src/vnet/mfib/mfib_entry.c
index 804e10ab..2302b9a1 100644
--- a/src/vnet/mfib/mfib_entry.c
+++ b/src/vnet/mfib/mfib_entry.c
@@ -334,6 +334,17 @@ mfib_entry_get_best_src (const mfib_entry_t *mfib_entry)
     return (bsrc);
 }
 
+int
+mfib_entry_is_sourced (fib_node_index_t mfib_entry_index,
+                       mfib_source_t source)
+{
+    mfib_entry_t *mfib_entry;
+
+    mfib_entry = mfib_entry_get(mfib_entry_index);
+
+    return (NULL != mfib_entry_src_find(mfib_entry, source, NULL));
+}
+
 static void
 mfib_entry_src_flush (mfib_entry_src_t *msrc)
 {
diff --git a/src/vnet/mfib/mfib_entry.h b/src/vnet/mfib/mfib_entry.h
index d4377878..96ee49f7 100644
--- a/src/vnet/mfib/mfib_entry.h
+++ b/src/vnet/mfib/mfib_entry.h
@@ -130,6 +130,8 @@ extern void mfib_entry_unlock(fib_node_index_t fib_entry_index);
 extern void mfib_entry_get_prefix(fib_node_index_t fib_entry_index,
                                   mfib_prefix_t *pfx);
 extern u32 mfib_entry_get_fib_index(fib_node_index_t fib_entry_index);
+extern int mfib_entry_is_sourced(fib_node_index_t fib_entry_index,
+                                 mfib_source_t source);
 
 extern void mfib_entry_contribute_forwarding(
     fib_node_index_t mfib_entry_index,
diff --git a/src/vnet/mfib/mfib_table.c b/src/vnet/mfib/mfib_table.c
index 7ffe8941..e5550adc 100644
--- a/src/vnet/mfib/mfib_table.c
+++ b/src/vnet/mfib/mfib_table.c
@@ -424,7 +424,8 @@ mfib_table_find (fib_protocol_t proto,
 
 u32
 mfib_table_find_or_create_and_lock (fib_protocol_t proto,
-                                    u32 table_id)
+                                    u32 table_id,
+                                    mfib_source_t src)
 {
     mfib_table_t *mfib_table;
     fib_node_index_t fi;
@@ -432,10 +433,10 @@ mfib_table_find_or_create_and_lock (fib_protocol_t proto,
     switch (proto)
     {
     case FIB_PROTOCOL_IP4:
-        fi = ip4_mfib_table_find_or_create_and_lock(table_id);
+        fi = ip4_mfib_table_find_or_create_and_lock(table_id, src);
         break;
     case FIB_PROTOCOL_IP6:
-        fi = ip6_mfib_table_find_or_create_and_lock(table_id);
+        fi = ip6_mfib_table_find_or_create_and_lock(table_id, src);
         break;
     case FIB_PROTOCOL_MPLS:
     default:
@@ -451,6 +452,59 @@ mfib_table_find_or_create_and_lock (fib_protocol_t proto,
     return (fi);
 }
 
+/**
+ * @brief Table flush context. Store the indicies of matching FIB entries
+ * that need to be removed.
+ */
+typedef struct mfib_table_flush_ctx_t_
+{
+    /**
+     * The list of entries to flush
+     */
+    fib_node_index_t *mftf_entries;
+
+    /**
+     * The source we are flushing
+     */
+    mfib_source_t mftf_source;
+} mfib_table_flush_ctx_t;
+
+static int
+mfib_table_flush_cb (fib_node_index_t mfib_entry_index,
+                     void *arg)
+{
+    mfib_table_flush_ctx_t *ctx = arg;
+
+    if (mfib_entry_is_sourced(mfib_entry_index, ctx->mftf_source))
+    {
+        vec_add1(ctx->mftf_entries, mfib_entry_index);
+    }
+    return (1);
+}
+
+void
+mfib_table_flush (u32 mfib_index,
+                  fib_protocol_t proto,
+                  mfib_source_t source)
+{
+    fib_node_index_t *mfib_entry_index;
+    mfib_table_flush_ctx_t ctx = {
+        .mftf_entries = NULL,
+        .mftf_source = source,
+    };
+
+    mfib_table_walk(mfib_index, proto,
+                    mfib_table_flush_cb,
+                    &ctx);
+
+    vec_foreach(mfib_entry_index, ctx.mftf_entries)
+    {
+        mfib_table_entry_delete_index(*mfib_entry_index, source);
+    }
+
+    vec_free(ctx.mftf_entries);
+}
+
 static void
 mfib_table_destroy (mfib_table_t *mfib_table)
 {
@@ -472,27 +526,43 @@ mfib_table_destroy (mfib_table_t *mfib_table)
 
 void
 mfib_table_unlock (u32 fib_index,
-                   fib_protocol_t proto)
+                   fib_protocol_t proto,
+                   mfib_source_t source)
 {
     mfib_table_t *mfib_table;
 
     mfib_table = mfib_table_get(fib_index, proto);
-    mfib_table->mft_locks--;
+    mfib_table->mft_locks[source]--;
+    mfib_table->mft_locks[MFIB_TABLE_TOTAL_LOCKS]--;
+
+    if (0 == mfib_table->mft_locks[source])
+    {
+        /*
+         * The source no longer needs the table. flush any routes
+         * from it just in case
+         */
+        mfib_table_flush(fib_index, proto, source);
+    }
 
-    if (0 == mfib_table->mft_locks)
+    if (0 == mfib_table->mft_locks[MFIB_TABLE_TOTAL_LOCKS])
     {
-        mfib_table_destroy(mfib_table);
+        /*
+         * no more locak from any source - kill it
+         */
+	mfib_table_destroy(mfib_table);
     }
 }
 
 void
 mfib_table_lock (u32 fib_index,
-                 fib_protocol_t proto)
+                 fib_protocol_t proto,
+                 mfib_source_t source)
 {
     mfib_table_t *mfib_table;
 
     mfib_table = mfib_table_get(fib_index, proto);
-    mfib_table->mft_locks++;
+    mfib_table->mft_locks[source]++;
+    mfib_table->mft_locks[MFIB_TABLE_TOTAL_LOCKS]++;
 }
 
 void
diff --git a/src/vnet/mfib/mfib_table.h b/src/vnet/mfib/mfib_table.h
index 83aa04ef..c6b0b097 100644
--- a/src/vnet/mfib/mfib_table.h
+++ b/src/vnet/mfib/mfib_table.h
@@ -22,6 +22,12 @@
 
 #include <vnet/mfib/mfib_types.h>
 
+/**
+ * Keep a lock per-source and a total
+ */
+#define MFIB_TABLE_N_LOCKS (MFIB_N_SOURCES+1)
+#define MFIB_TABLE_TOTAL_LOCKS MFIB_N_SOURCES
+
 /**
  * @brief
  *   A protocol Independent IP multicast FIB table
@@ -47,7 +53,7 @@ typedef struct mfib_table_t_
     /**
      * number of locks on the table
      */
-    u16 mft_locks;
+    u16 mft_locks[MFIB_TABLE_N_LOCKS];
 
     /**
      * Table ID (hash key) for this FIB.
@@ -259,7 +265,8 @@ extern fib_node_index_t mfib_table_entry_special_add(u32 fib_index,
  *  the source to flush
  */
 extern void mfib_table_flush(u32 fib_index,
-                             fib_protocol_t proto);
+                             fib_protocol_t proto,
+                             mfib_source_t source);
 
 /**
  * @brief
@@ -307,9 +314,13 @@ extern u32 mfib_table_find(fib_protocol_t proto, u32 table_id);
  *
  * @return fib_index
  *  The index of the FIB
+ *
+ * @param source
+ *  The ID of the client/source.
  */
 extern u32 mfib_table_find_or_create_and_lock(fib_protocol_t proto,
-                                              u32 table_id);
+                                              u32 table_id,
+                                              mfib_source_t source);
 
 
 /**
@@ -321,9 +332,13 @@ extern u32 mfib_table_find_or_create_and_lock(fib_protocol_t proto,
  *
  * @paran proto
  *  The protocol of the FIB (and thus the entries therein)
+ *
+ * @param source
+ *  The ID of the client/source.
  */
 extern void mfib_table_unlock(u32 fib_index,
-                              fib_protocol_t proto);
+                              fib_protocol_t proto,
+                              mfib_source_t source);
 
 /**
  * @brief
@@ -335,9 +350,13 @@ extern void mfib_table_unlock(u32 fib_index,
  *
  * @paran proto
  *  The protocol of the FIB (and thus the entries therein)
+ *
+ * @param source
+ *  The ID of the client/source.
  */
 extern void mfib_table_lock(u32 fib_index,
-                            fib_protocol_t proto);
+                            fib_protocol_t proto,
+                            mfib_source_t source);
 
 /**
  * @brief
diff --git a/src/vnet/mfib/mfib_test.c b/src/vnet/mfib/mfib_test.c
index 57787eca..3055844d 100644
--- a/src/vnet/mfib/mfib_test.c
+++ b/src/vnet/mfib/mfib_test.c
@@ -22,6 +22,7 @@
 #include <vnet/fib/fib_path_list.h>
 #include <vnet/fib/fib_test.h>
 #include <vnet/fib/fib_table.h>
+#include <vnet/fib/mpls_fib.h>
 
 #include <vnet/dpo/replicate_dpo.h>
 #include <vnet/adj/adj_mcast.h>
@@ -366,7 +367,7 @@ mfib_test_i (fib_protocol_t PROTO,
     MFIB_TEST(3 == adj_mcast_db_size(), "3 MCAST adjs");
 
     /* Find or create FIB table 11 */
-    fib_index = mfib_table_find_or_create_and_lock(PROTO, 11);
+    fib_index = mfib_table_find_or_create_and_lock(PROTO, 11, MFIB_SOURCE_API);
 
     mfib_prefix_t pfx_dft = {
         .fp_len = 0,
@@ -1113,9 +1114,10 @@ mfib_test_i (fib_protocol_t PROTO,
     /*
      * MPLS enable an interface so we get the MPLS table created
      */
+    mpls_table_create(MPLS_FIB_DEFAULT_TABLE_ID, FIB_SOURCE_API);
     mpls_sw_interface_enable_disable(&mpls_main,
                                      tm->hw[0]->sw_if_index,
-                                     1);
+                                     1, 0);
 
     lfei = fib_table_entry_update_one_path(0, // default MPLS Table
                                            &pfx_3500,
@@ -1192,7 +1194,7 @@ mfib_test_i (fib_protocol_t PROTO,
     /*
      * Unlock the table - it's the last lock so should be gone thereafter
      */
-    mfib_table_unlock(fib_index, PROTO);
+    mfib_table_unlock(fib_index, PROTO, MFIB_SOURCE_API);
 
     MFIB_TEST((FIB_NODE_INDEX_INVALID ==
                mfib_table_find(PROTO, fib_index)),
@@ -1207,7 +1209,8 @@ mfib_test_i (fib_protocol_t PROTO,
      */
     mpls_sw_interface_enable_disable(&mpls_main,
                                      tm->hw[0]->sw_if_index,
-                                     0);
+                                     0, 0);
+    mpls_table_delete(MPLS_FIB_DEFAULT_TABLE_ID, FIB_SOURCE_API);
 
     /*
      * test we've leaked no resources
diff --git a/src/vnet/mfib/mfib_types.h b/src/vnet/mfib/mfib_types.h
index 863fad16..50aede04 100644
--- a/src/vnet/mfib/mfib_types.h
+++ b/src/vnet/mfib/mfib_types.h
@@ -166,9 +166,10 @@ typedef enum mfib_source_t_
     MFIB_SOURCE_VXLAN,
     MFIB_SOURCE_DHCP,
     MFIB_SOURCE_SRv6,
-    MFIB_SOURCE_DEFAULT_ROUTE,
     MFIB_SOURCE_GTPU,
     MFIB_SOURCE_VXLAN_GPE,
+    MFIB_SOURCE_RR,
+    MFIB_SOURCE_DEFAULT_ROUTE,
 } mfib_source_t;
 
 #define MFIB_SOURCE_NAMES {                        \
@@ -178,11 +179,14 @@ typedef enum mfib_source_t_
     [MFIB_SOURCE_DHCP] = "DHCP",                   \
     [MFIB_SOURCE_VXLAN] = "VXLAN",                 \
     [MFIB_SOURCE_SRv6] = "SRv6",                   \
-    [MFIB_SOURCE_DEFAULT_ROUTE] = "Default Route", \
     [MFIB_SOURCE_GTPU] = "GTPU",                   \
     [MFIB_SOURCE_VXLAN_GPE] = "VXLAN-GPE",         \
+    [MFIB_SOURCE_RR] = "Recursive-resolution",     \
+    [MFIB_SOURCE_DEFAULT_ROUTE] = "Default Route", \
 }
 
+#define MFIB_N_SOURCES (MFIB_SOURCE_DEFAULT_ROUTE)
+
 /**
  * \brief Compare two prefixes for equality
  */
diff --git a/src/vnet/mpls/interface.c b/src/vnet/mpls/interface.c
index a085aaa2..d7c8e7d3 100644
--- a/src/vnet/mpls/interface.c
+++ b/src/vnet/mpls/interface.c
@@ -35,25 +35,33 @@ mpls_sw_interface_is_enabled (u32 sw_if_index)
     return (mm->mpls_enabled_by_sw_if_index[sw_if_index]);
 }
 
-void
+int
 mpls_sw_interface_enable_disable (mpls_main_t * mm,
                                   u32 sw_if_index,
-                                  u8 is_enable)
+                                  u8 is_enable,
+                                  u8 is_api)
 {
   fib_node_index_t lfib_index;
 
   vec_validate_init_empty (mm->mpls_enabled_by_sw_if_index, sw_if_index, 0);
 
+  lfib_index = fib_table_find(FIB_PROTOCOL_MPLS,
+                              MPLS_FIB_DEFAULT_TABLE_ID);
+
+  if (~0 == lfib_index)
+       return VNET_API_ERROR_NO_SUCH_FIB;
+
   /*
    * enable/disable only on the 1<->0 transition
    */
   if (is_enable)
     {
       if (1 != ++mm->mpls_enabled_by_sw_if_index[sw_if_index])
-        return;
+          return (0);
+
+      fib_table_lock(lfib_index, FIB_PROTOCOL_MPLS,
+                     (is_api? FIB_SOURCE_API: FIB_SOURCE_CLI));
 
-      lfib_index = fib_table_find_or_create_and_lock(FIB_PROTOCOL_MPLS,
-						     MPLS_FIB_DEFAULT_TABLE_ID);
       vec_validate(mm->fib_index_by_sw_if_index, 0);
       mm->fib_index_by_sw_if_index[sw_if_index] = lfib_index;
     }
@@ -61,15 +69,17 @@ mpls_sw_interface_enable_disable (mpls_main_t * mm,
     {
       ASSERT(mm->mpls_enabled_by_sw_if_index[sw_if_index] > 0);
       if (0 != --mm->mpls_enabled_by_sw_if_index[sw_if_index])
-        return;
+          return (0);
 
       fib_table_unlock(mm->fib_index_by_sw_if_index[sw_if_index],
-		       FIB_PROTOCOL_MPLS);
+		       FIB_PROTOCOL_MPLS,
+                       (is_api? FIB_SOURCE_API: FIB_SOURCE_CLI));
     }
 
   vnet_feature_enable_disable ("mpls-input", "mpls-not-enabled",
                                sw_if_index, !is_enable, 0, 0);
 
+  return (0);
 }
 
 static clib_error_t *
@@ -101,7 +111,7 @@ mpls_interface_enable_disable (vlib_main_t * vm,
       goto done;
     }
 
-  mpls_sw_interface_enable_disable(&mpls_main, sw_if_index, enable);
+  mpls_sw_interface_enable_disable(&mpls_main, sw_if_index, enable, 0);
 
  done:
   return error;
diff --git a/src/vnet/mpls/mpls.c b/src/vnet/mpls/mpls.c
index 5021ac23..7bdfd8c7 100644
--- a/src/vnet/mpls/mpls.c
+++ b/src/vnet/mpls/mpls.c
@@ -536,6 +536,78 @@ VLIB_CLI_COMMAND (mpls_local_label_command, static) = {
   .short_help = "Create/Delete MPL local labels",
 };
 
+clib_error_t *
+vnet_mpls_table_cmd (vlib_main_t * vm,
+                     unformat_input_t * main_input,
+                     vlib_cli_command_t * cmdo)
+{
+  unformat_input_t _line_input, *line_input = &_line_input;
+  clib_error_t *error = NULL;
+  u32 table_id, is_add;
+
+  is_add = 1;
+  table_id = ~0;
+
+  /* Get a line of input. */
+  if (!unformat_user (main_input, unformat_line_input, line_input))
+    return 0;
+
+  while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
+    {
+      if (unformat (line_input, "%d", &table_id))
+	;
+      else if (unformat (line_input, "del"))
+	is_add = 0;
+      else if (unformat (line_input, "add"))
+	is_add = 1;
+      else
+	{
+	  error = unformat_parse_error (line_input);
+	  goto done;
+	}
+    }
+
+  if (~0 == table_id)
+    {
+      error = clib_error_return (0, "No table id");
+      goto done;
+    }
+  else if (0 == table_id)
+    {
+      error = clib_error_return (0, "Can't change the default table");
+      goto done;
+    }
+  else
+    {
+      if (is_add)
+        {
+          mpls_table_create (table_id, 0);
+        }
+      else
+        {
+          mpls_table_delete (table_id, 0);
+        }
+    }
+
+ done:
+  unformat_free (line_input);
+  return error;
+}
+
+/* *INDENT-ON* */
+/*?
+ * This command is used to add or delete MPLS Tables. All
+ * Tables must be explicitly added before that can be used,
+ * Including the default table.
+ ?*/
+/* *INDENT-OFF* */
+VLIB_CLI_COMMAND (ip6_table_command, static) = {
+  .path = "mpla table",
+  .short_help = "mpls table [add|del] <table-id>",
+  .function = vnet_mpls_table_cmd,
+  .is_mp_safe = 1,
+};
+
 int
 mpls_fib_reset_labels (u32 fib_id)
 {
@@ -546,12 +618,8 @@ mpls_fib_reset_labels (u32 fib_id)
 static clib_error_t *
 mpls_init (vlib_main_t * vm)
 {
-  mpls_main_t * mm = &mpls_main;
   clib_error_t * error;
 
-  mm->vlib_main = vm;
-  mm->vnet_main = vnet_get_main();
-
   if ((error = vlib_call_init_function (vm, ip_main_init)))
     return error;
 
diff --git a/src/vnet/mpls/mpls.h b/src/vnet/mpls/mpls.h
index b0125e60..31cb1746 100644
--- a/src/vnet/mpls/mpls.h
+++ b/src/vnet/mpls/mpls.h
@@ -56,10 +56,6 @@ typedef struct {
 
   /* IP4 enabled count by software interface */
   u8 * mpls_enabled_by_sw_if_index;
-
-  /* convenience */
-  vlib_main_t * vlib_main;
-  vnet_main_t * vnet_main;
 } mpls_main_t;
 
 extern mpls_main_t mpls_main;
@@ -77,8 +73,6 @@ extern vlib_node_registration_t mpls_midchain_node;
 
 /* Parse mpls protocol as 0xXXXX or protocol name.
    In either host or network byte order. */
-unformat_function_t unformat_mpls_protocol_host_byte_order;
-unformat_function_t unformat_mpls_protocol_net_byte_order;
 unformat_function_t unformat_mpls_label_net_byte_order;
 unformat_function_t unformat_mpls_unicast_label;
 
@@ -86,9 +80,10 @@ unformat_function_t unformat_mpls_unicast_label;
 unformat_function_t unformat_mpls_header;
 unformat_function_t unformat_pg_mpls_header;
 
-void mpls_sw_interface_enable_disable (mpls_main_t * mm,
-				       u32 sw_if_index,
-				       u8 is_enable);
+int mpls_sw_interface_enable_disable (mpls_main_t * mm,
+                                      u32 sw_if_index,
+                                      u8 is_enable,
+                                      u8 is_api);
 
 u8 mpls_sw_interface_is_enabled (u32 sw_if_index);
 
@@ -103,4 +98,7 @@ mpls_fib_index_cmp(void * a1, void * a2);
 int
 mpls_label_cmp(void * a1, void * a2);
 
+void mpls_table_create(uint32_t table_id, u8 is_api);
+void mpls_table_delete(uint32_t table_id, u8 is_api);
+
 #endif /* included_vnet_mpls_h */
diff --git a/src/vnet/mpls/mpls_api.c b/src/vnet/mpls/mpls_api.c
index a44b1a25..38f5b014 100644
--- a/src/vnet/mpls/mpls_api.c
+++ b/src/vnet/mpls/mpls_api.c
@@ -58,6 +58,29 @@ _(MPLS_FIB_DUMP, mpls_fib_dump)
 extern void stats_dslock_with_hint (int hint, int tag);
 extern void stats_dsunlock (void);
 
+void
+mpls_table_delete (u32 table_id, u8 is_api)
+{
+  u32 fib_index;
+
+  /*
+   * The MPLS defult table must also be explicitly created via the API.
+   * So in contrast to IP, it gets no special treatment here.
+   *
+   * The API holds only one lock on the table.
+   * i.e. it can be added many times via the API but needs to be
+   * deleted only once.
+   */
+  fib_index = fib_table_find (FIB_PROTOCOL_MPLS, table_id);
+
+  if (~0 != fib_index)
+    {
+      fib_table_unlock (fib_index,
+			FIB_PROTOCOL_MPLS,
+			(is_api ? FIB_SOURCE_API : FIB_SOURCE_CLI));
+    }
+}
+
 void
 vl_api_mpls_table_add_del_t_handler (vl_api_mpls_table_add_del_t * mp)
 {
@@ -68,6 +91,13 @@ vl_api_mpls_table_add_del_t_handler (vl_api_mpls_table_add_del_t * mp)
   vnm = vnet_get_main ();
   vnm->api_errno = 0;
 
+  if (mp->mt_is_add)
+    mpls_table_create (ntohl (mp->mt_table_id), 1);
+  else
+    mpls_table_delete (ntohl (mp->mt_table_id), 1);
+
+  rv = (rv == 0) ? vnm->api_errno : rv;
+
   REPLY_MACRO (VL_API_MPLS_TABLE_ADD_DEL_REPLY);
 }
 
@@ -82,14 +112,7 @@ mpls_ip_bind_unbind_handler (vnet_main_t * vnm,
 
   if (~0 == mpls_fib_index)
     {
-      if (mp->mb_create_table_if_needed)
-	{
-	  mpls_fib_index =
-	    fib_table_find_or_create_and_lock (FIB_PROTOCOL_MPLS,
-					       ntohl (mp->mb_mpls_table_id));
-	}
-      else
-	return VNET_API_ERROR_NO_SUCH_FIB;
+      return VNET_API_ERROR_NO_SUCH_FIB;
     }
 
   ip_fib_index = fib_table_find ((mp->mb_is_ip4 ?
@@ -170,7 +193,6 @@ mpls_route_add_del_t_handler (vnet_main_t * vnm,
 			    mp->mr_next_hop_sw_if_index,
 			    pfx.fp_payload_proto,
 			    mp->mr_next_hop_table_id,
-			    mp->mr_create_table_if_needed,
 			    mp->mr_is_rpf_id,
 			    &fib_index, &next_hop_fib_index);
 
@@ -235,6 +257,32 @@ vl_api_mpls_route_add_del_t_handler (vl_api_mpls_route_add_del_t * mp)
   REPLY_MACRO (VL_API_MPLS_ROUTE_ADD_DEL_REPLY);
 }
 
+void
+mpls_table_create (u32 table_id, u8 is_api)
+{
+  u32 fib_index;
+
+  /*
+   * The MPLS defult table must also be explicitly created via the API.
+   * So in contrast to IP, it gets no special treatment here.
+   */
+
+  /*
+   * The API holds only one lock on the table.
+   * i.e. it can be added many times via the API but needs to be
+   * deleted only once.
+   */
+  fib_index = fib_table_find (FIB_PROTOCOL_MPLS, table_id);
+
+  if (~0 == fib_index)
+    {
+      fib_table_find_or_create_and_lock (FIB_PROTOCOL_MPLS,
+					 table_id,
+					 (is_api ?
+					  FIB_SOURCE_API : FIB_SOURCE_CLI));
+    }
+}
+
 static void
 vl_api_mpls_tunnel_add_del_t_handler (vl_api_mpls_tunnel_add_del_t * mp)
 {
diff --git a/src/vnet/srv6/sr_policy_rewrite.c b/src/vnet/srv6/sr_policy_rewrite.c
index f427bbf3..2f90993a 100755
--- a/src/vnet/srv6/sr_policy_rewrite.c
+++ b/src/vnet/srv6/sr_policy_rewrite.c
@@ -595,8 +595,10 @@ sr_policy_add (ip6_address_t * bsid, ip6_address_t * segments,
   if (sm->fib_table_ip6 == (u32) ~ 0)
     {
       sm->fib_table_ip6 = fib_table_create_and_lock (FIB_PROTOCOL_IP6,
+						     FIB_SOURCE_SR,
 						     "SRv6 steering of IP6 prefixes through BSIDs");
       sm->fib_table_ip4 = fib_table_create_and_lock (FIB_PROTOCOL_IP6,
+						     FIB_SOURCE_SR,
 						     "SRv6 steering of IP4 prefixes through BSIDs");
     }
 
@@ -684,8 +686,8 @@ sr_policy_del (ip6_address_t * bsid, u32 index)
   /* If FIB empty unlock it */
   if (!pool_elts (sm->sr_policies) && !pool_elts (sm->steer_policies))
     {
-      fib_table_unlock (sm->fib_table_ip6, FIB_PROTOCOL_IP6);
-      fib_table_unlock (sm->fib_table_ip4, FIB_PROTOCOL_IP6);
+      fib_table_unlock (sm->fib_table_ip6, FIB_PROTOCOL_IP6, FIB_SOURCE_SR);
+      fib_table_unlock (sm->fib_table_ip4, FIB_PROTOCOL_IP6, FIB_SOURCE_SR);
       sm->fib_table_ip6 = (u32) ~ 0;
       sm->fib_table_ip4 = (u32) ~ 0;
     }
diff --git a/src/vnet/srv6/sr_steering.c b/src/vnet/srv6/sr_steering.c
index 57fe21f6..cf4e81ab 100755
--- a/src/vnet/srv6/sr_steering.c
+++ b/src/vnet/srv6/sr_steering.c
@@ -159,8 +159,10 @@ sr_steering_policy (int is_del, ip6_address_t * bsid, u32 sr_policy_index,
 	  /* If no more SR policies or steering policies */
 	  if (!pool_elts (sm->sr_policies) && !pool_elts (sm->steer_policies))
 	    {
-	      fib_table_unlock (sm->fib_table_ip6, FIB_PROTOCOL_IP6);
-	      fib_table_unlock (sm->fib_table_ip4, FIB_PROTOCOL_IP6);
+	      fib_table_unlock (sm->fib_table_ip6,
+				FIB_PROTOCOL_IP6, FIB_SOURCE_SR);
+	      fib_table_unlock (sm->fib_table_ip4,
+				FIB_PROTOCOL_IP6, FIB_SOURCE_SR);
 	      sm->fib_table_ip6 = (u32) ~ 0;
 	      sm->fib_table_ip4 = (u32) ~ 0;
 	    }
diff --git a/src/vpp/api/api.c b/src/vpp/api/api.c
index f9c3129c..044ddb5b 100644
--- a/src/vpp/api/api.c
+++ b/src/vpp/api/api.c
@@ -699,8 +699,9 @@ static void
 
   VALIDATE_SW_IF_INDEX (mp);
 
-  mpls_sw_interface_enable_disable (&mpls_main,
-				    ntohl (mp->sw_if_index), mp->enable);
+  rv = mpls_sw_interface_enable_disable (&mpls_main,
+					 ntohl (mp->sw_if_index),
+					 mp->enable, 1);
 
   BAD_SW_IF_INDEX_LABEL;
   REPLY_MACRO (VL_API_SW_INTERFACE_SET_MPLS_ENABLE_REPLY);
diff --git a/src/vpp/api/custom_dump.c b/src/vpp/api/custom_dump.c
index 1353fe28..be74b83a 100644
--- a/src/vpp/api/custom_dump.c
+++ b/src/vpp/api/custom_dump.c
@@ -560,9 +560,6 @@ static void *vl_api_ip_add_del_route_t_print
   if (mp->table_id != 0)
     s = format (s, "vrf %d ", ntohl (mp->table_id));
 
-  if (mp->create_vrf_if_needed)
-    s = format (s, "create-vrf ");
-
   if (mp->next_hop_weight != 1)
     s = format (s, "weight %d ", mp->next_hop_weight);
 
diff --git a/test/test_dhcp.py b/test/test_dhcp.py
index 6fc29182..fe97f6c9 100644
--- a/test/test_dhcp.py
+++ b/test/test_dhcp.py
@@ -6,7 +6,7 @@ import struct
 
 from framework import VppTestCase, VppTestRunner
 from vpp_neighbor import VppNeighbor
-from vpp_ip_route import find_route
+from vpp_ip_route import find_route, VppIpTable
 from util import mk_ll_addr
 
 from scapy.layers.l2 import Ether, getmacbyip, ARP
@@ -34,9 +34,19 @@ class TestDHCP(VppTestCase):
 
         # create 3 pg interfaces
         self.create_pg_interfaces(range(4))
+        self.tables = []
 
         # pg0 and 1 are IP configured in VRF 0 and 1.
         # pg2 and 3 are non IP-configured in VRF 0 and 1
+        table_id = 0
+        for table_id in range(1, 4):
+            tbl4 = VppIpTable(self, table_id)
+            tbl4.add_vpp_config()
+            self.tables.append(tbl4)
+            tbl6 = VppIpTable(self, table_id, is_ip6=1)
+            tbl6.add_vpp_config()
+            self.tables.append(tbl6)
+
         table_id = 0
         for i in self.pg_interfaces[:2]:
             i.admin_up()
@@ -56,11 +66,15 @@ class TestDHCP(VppTestCase):
             table_id += 1
 
     def tearDown(self):
-        super(TestDHCP, self).tearDown()
-        for i in self.pg_interfaces:
+        for i in self.pg_interfaces[:2]:
             i.unconfig_ip4()
             i.unconfig_ip6()
+
+        for i in self.pg_interfaces:
+            i.set_table_ip4(0)
+            i.set_table_ip6(0)
             i.admin_down()
+        super(TestDHCP, self).tearDown()
 
     def send_and_assert_no_replies(self, intf, pkts, remark):
         intf.add_stream(pkts)
@@ -667,6 +681,8 @@ class TestDHCP(VppTestCase):
                                         "DHCP cleanup VRF 0")
         self.send_and_assert_no_replies(self.pg3, pkts_disc_vrf1,
                                         "DHCP cleanup VRF 1")
+        self.pg2.unconfig_ip4()
+        self.pg3.unconfig_ip4()
 
     def test_dhcp6_proxy(self):
         """ DHCPv6 Proxy"""
@@ -1045,6 +1061,8 @@ class TestDHCP(VppTestCase):
                                     server_table_id=0,
                                     is_ipv6=1,
                                     is_add=0)
+        self.pg2.unconfig_ip6()
+        self.pg3.unconfig_ip6()
 
     def test_dhcp_client(self):
         """ DHCP Client"""
diff --git a/test/test_gre.py b/test/test_gre.py
index 1afc44fb..9046b05f 100644
--- a/test/test_gre.py
+++ b/test/test_gre.py
@@ -6,7 +6,7 @@ from logging import *
 from framework import VppTestCase, VppTestRunner
 from vpp_sub_interface import VppDot1QSubint
 from vpp_gre_interface import VppGreInterface, VppGre6Interface
-from vpp_ip_route import VppIpRoute, VppRoutePath, DpoProto
+from vpp_ip_route import VppIpRoute, VppRoutePath, DpoProto, VppIpTable
 from vpp_papi_provider import L2_VTR_OP
 
 from scapy.packet import Raw
@@ -30,6 +30,9 @@ class TestGRE(VppTestCase):
 
         # create 3 pg interfaces - set one in a non-default table.
         self.create_pg_interfaces(range(3))
+
+        self.tbl = VppIpTable(self, 1)
+        self.tbl.add_vpp_config()
         self.pg1.set_table_ip4(1)
 
         for i in self.pg_interfaces:
@@ -43,11 +46,12 @@ class TestGRE(VppTestCase):
         self.pg2.resolve_ndp()
 
     def tearDown(self):
-        super(TestGRE, self).tearDown()
         for i in self.pg_interfaces:
             i.unconfig_ip4()
             i.unconfig_ip6()
             i.admin_down()
+        self.pg1.set_table_ip4(0)
+        super(TestGRE, self).tearDown()
 
     def create_stream_ip4(self, src_if, src_ip, dst_ip):
         pkts = []
diff --git a/test/test_ip4.py b/test/test_ip4.py
index 7a7098c3..55d16735 100644
--- a/test/test_ip4.py
+++ b/test/test_ip4.py
@@ -6,7 +6,8 @@ import unittest
 from framework import VppTestCase, VppTestRunner
 from vpp_sub_interface import VppSubInterface, VppDot1QSubint, VppDot1ADSubint
 from vpp_ip_route import VppIpRoute, VppRoutePath, VppIpMRoute, \
-    VppMRoutePath, MRouteItfFlags, MRouteEntryFlags, VppMplsIpBind
+    VppMRoutePath, MRouteItfFlags, MRouteEntryFlags, VppMplsIpBind, \
+    VppMplsTable
 
 from scapy.packet import Raw
 from scapy.layers.l2 import Ether, Dot1Q, ARP
@@ -774,6 +775,8 @@ class TestIPLoadBalance(VppTestCase):
         super(TestIPLoadBalance, self).setUp()
 
         self.create_pg_interfaces(range(5))
+        mpls_tbl = VppMplsTable(self, 0)
+        mpls_tbl.add_vpp_config()
 
         for i in self.pg_interfaces:
             i.admin_up()
@@ -782,11 +785,11 @@ class TestIPLoadBalance(VppTestCase):
             i.enable_mpls()
 
     def tearDown(self):
-        super(TestIPLoadBalance, self).tearDown()
         for i in self.pg_interfaces:
             i.disable_mpls()
             i.unconfig_ip4()
             i.admin_down()
+        super(TestIPLoadBalance, self).tearDown()
 
     def send_and_expect_load_balancing(self, input, pkts, outputs):
         input.add_stream(pkts)
@@ -966,6 +969,8 @@ class TestIPVlan0(VppTestCase):
         super(TestIPVlan0, self).setUp()
 
         self.create_pg_interfaces(range(2))
+        mpls_tbl = VppMplsTable(self, 0)
+        mpls_tbl.add_vpp_config()
 
         for i in self.pg_interfaces:
             i.admin_up()
@@ -974,11 +979,11 @@ class TestIPVlan0(VppTestCase):
             i.enable_mpls()
 
     def tearDown(self):
-        super(TestIPVlan0, self).tearDown()
         for i in self.pg_interfaces:
             i.disable_mpls()
             i.unconfig_ip4()
             i.admin_down()
+        super(TestIPVlan0, self).tearDown()
 
     def send_and_expect(self, input, pkts, output):
         input.add_stream(pkts)
diff --git a/test/test_ip4_vrf_multi_instance.py b/test/test_ip4_vrf_multi_instance.py
index b73ac948..5a8d6760 100644
--- a/test/test_ip4_vrf_multi_instance.py
+++ b/test/test_ip4_vrf_multi_instance.py
@@ -172,9 +172,10 @@ class TestIp4VrfMultiInst(VppTestCase):
             pg_if = self.pg_if_by_vrf_id[vrf_id][0]
             dest_addr = pg_if.remote_hosts[0].ip4n
             dest_addr_len = 24
+            self.vapi.ip_table_add_del(vrf_id, is_add=1)
             self.vapi.ip_add_del_route(
                 dest_addr, dest_addr_len, pg_if.local_ip4n,
-                table_id=vrf_id, create_vrf_if_needed=1, is_multipath=1)
+                table_id=vrf_id, is_multipath=1)
             self.logger.info("IPv4 VRF ID %d created" % vrf_id)
             if vrf_id not in self.vrf_list:
                 self.vrf_list.append(vrf_id)
@@ -216,6 +217,7 @@ class TestIp4VrfMultiInst(VppTestCase):
         self.logger.info("IPv4 VRF ID %d reset" % vrf_id)
         self.logger.debug(self.vapi.ppcli("show ip fib"))
         self.logger.debug(self.vapi.ppcli("show ip arp"))
+        self.vapi.ip_table_add_del(vrf_id, is_add=0)
 
     def create_stream(self, src_if, packet_sizes):
         """
diff --git a/test/test_ip6.py b/test/test_ip6.py
index 285ce181..aad3713c 100644
--- a/test/test_ip6.py
+++ b/test/test_ip6.py
@@ -8,7 +8,7 @@ from vpp_sub_interface import VppSubInterface, VppDot1QSubint
 from vpp_pg_interface import is_ipv6_misc
 from vpp_ip_route import VppIpRoute, VppRoutePath, find_route, VppIpMRoute, \
     VppMRoutePath, MRouteItfFlags, MRouteEntryFlags, VppMplsIpBind, \
-    VppMplsRoute, DpoProto
+    VppMplsRoute, DpoProto, VppMplsTable
 from vpp_neighbor import find_nbr, VppNeighbor
 
 from scapy.packet import Raw
@@ -1260,6 +1260,9 @@ class TestIP6LoadBalance(VppTestCase):
 
         self.create_pg_interfaces(range(5))
 
+        mpls_tbl = VppMplsTable(self, 0)
+        mpls_tbl.add_vpp_config()
+
         for i in self.pg_interfaces:
             i.admin_up()
             i.config_ip6()
@@ -1267,11 +1270,11 @@ class TestIP6LoadBalance(VppTestCase):
             i.enable_mpls()
 
     def tearDown(self):
-        super(TestIP6LoadBalance, self).tearDown()
         for i in self.pg_interfaces:
             i.unconfig_ip6()
             i.admin_down()
             i.disable_mpls()
+        super(TestIP6LoadBalance, self).tearDown()
 
     def send_and_expect_load_balancing(self, input, pkts, outputs):
         input.add_stream(pkts)
diff --git a/test/test_ip6_vrf_multi_instance.py b/test/test_ip6_vrf_multi_instance.py
index af80b5ba..769cb2e5 100644
--- a/test/test_ip6_vrf_multi_instance.py
+++ b/test/test_ip6_vrf_multi_instance.py
@@ -187,9 +187,10 @@ class TestIP6VrfMultiInst(VppTestCase):
             pg_if = self.pg_if_by_vrf_id[vrf_id][0]
             dest_addr = pg_if.remote_hosts[0].ip6n
             dest_addr_len = 64
+            self.vapi.ip_table_add_del(vrf_id, is_add=1, is_ipv6=1)
             self.vapi.ip_add_del_route(
                 dest_addr, dest_addr_len, pg_if.local_ip6n, is_ipv6=1,
-                table_id=vrf_id, create_vrf_if_needed=1, is_multipath=1)
+                table_id=vrf_id, is_multipath=1)
             self.logger.info("IPv6 VRF ID %d created" % vrf_id)
             if vrf_id not in self.vrf_list:
                 self.vrf_list.append(vrf_id)
@@ -232,6 +233,7 @@ class TestIP6VrfMultiInst(VppTestCase):
         self.logger.info("IPv6 VRF ID %d reset" % vrf_id)
         self.logger.debug(self.vapi.ppcli("show ip6 fib"))
         self.logger.debug(self.vapi.ppcli("show ip6 neighbors"))
+        self.vapi.ip_table_add_del(vrf_id, is_add=0, is_ipv6=1)
 
     def create_stream(self, src_if, packet_sizes):
         """
diff --git a/test/test_ip_mcast.py b/test/test_ip_mcast.py
index 276555d6..7cad683c 100644
--- a/test/test_ip_mcast.py
+++ b/test/test_ip_mcast.py
@@ -5,7 +5,7 @@ import unittest
 from framework import VppTestCase, VppTestRunner
 from vpp_sub_interface import VppSubInterface, VppDot1QSubint, VppDot1ADSubint
 from vpp_ip_route import VppIpMRoute, VppMRoutePath, VppMFibSignal, \
-    MRouteItfFlags, MRouteEntryFlags
+    MRouteItfFlags, MRouteEntryFlags, VppIpTable
 
 from scapy.packet import Raw
 from scapy.layers.l2 import Ether
@@ -44,16 +44,37 @@ class TestIPMcast(VppTestCase):
         super(TestIPMcast, self).setUp()
 
         # create 8 pg interfaces
-        self.create_pg_interfaces(range(8))
+        self.create_pg_interfaces(range(9))
 
         # setup interfaces
-        for i in self.pg_interfaces:
+        for i in self.pg_interfaces[:8]:
             i.admin_up()
             i.config_ip4()
             i.config_ip6()
             i.resolve_arp()
             i.resolve_ndp()
 
+        # one more in a vrf
+        tbl4 = VppIpTable(self, 10)
+        tbl4.add_vpp_config()
+        self.pg8.set_table_ip4(10)
+        self.pg8.config_ip4()
+
+        tbl6 = VppIpTable(self, 10, is_ip6=1)
+        tbl6.add_vpp_config()
+        self.pg8.set_table_ip6(10)
+        self.pg8.config_ip6()
+
+    def tearDown(self):
+        for i in self.pg_interfaces:
+            i.unconfig_ip4()
+            i.unconfig_ip6()
+            i.admin_down()
+
+        self.pg8.set_table_ip4(0)
+        self.pg8.set_table_ip6(0)
+        super(TestIPMcast, self).tearDown()
+
     def create_stream_ip4(self, src_if, src_ip, dst_ip, payload_size=0):
         pkts = []
         # default to small packet sizes
@@ -663,6 +684,77 @@ class TestIPMcast(VppTestCase):
         #
         route_232_1_1_1.remove_vpp_config()
 
+    def test_ip_mcast_vrf(self):
+        """ IP Multicast Replication in non-default table"""
+
+        #
+        # An (S,G).
+        # one accepting interface, pg0, 2 forwarding interfaces
+        #
+        route_1_1_1_1_232_1_1_1 = VppIpMRoute(
+            self,
+            "1.1.1.1",
+            "232.1.1.1", 64,
+            MRouteEntryFlags.MFIB_ENTRY_FLAG_NONE,
+            [VppMRoutePath(self.pg8.sw_if_index,
+                           MRouteItfFlags.MFIB_ITF_FLAG_ACCEPT),
+             VppMRoutePath(self.pg1.sw_if_index,
+                           MRouteItfFlags.MFIB_ITF_FLAG_FORWARD),
+             VppMRoutePath(self.pg2.sw_if_index,
+                           MRouteItfFlags.MFIB_ITF_FLAG_FORWARD)],
+            table_id=10)
+        route_1_1_1_1_232_1_1_1.add_vpp_config()
+
+        #
+        # a stream that matches the route for (1.1.1.1,232.1.1.1)
+        #  small packets
+        #
+        self.vapi.cli("clear trace")
+        tx = self.create_stream_ip4(self.pg8, "1.1.1.1", "232.1.1.1")
+        self.pg8.add_stream(tx)
+
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        # We expect replications on Pg1 & 2
+        self.verify_capture_ip4(self.pg1, tx)
+        self.verify_capture_ip4(self.pg2, tx)
+
+    def test_ip6_mcast_vrf(self):
+        """ IPv6 Multicast Replication in non-default table"""
+
+        #
+        # An (S,G).
+        # one accepting interface, pg0, 2 forwarding interfaces
+        #
+        route_2001_ff01_1 = VppIpMRoute(
+            self,
+            "2001::1",
+            "ff01::1", 256,
+            MRouteEntryFlags.MFIB_ENTRY_FLAG_NONE,
+            [VppMRoutePath(self.pg8.sw_if_index,
+                           MRouteItfFlags.MFIB_ITF_FLAG_ACCEPT),
+             VppMRoutePath(self.pg1.sw_if_index,
+                           MRouteItfFlags.MFIB_ITF_FLAG_FORWARD),
+             VppMRoutePath(self.pg2.sw_if_index,
+                           MRouteItfFlags.MFIB_ITF_FLAG_FORWARD)],
+            table_id=10,
+            is_ip6=1)
+        route_2001_ff01_1.add_vpp_config()
+
+        #
+        # a stream that matches the route for (2001::1, ff00::1)
+        #
+        self.vapi.cli("clear trace")
+        tx = self.create_stream_ip6(self.pg8, "2001::1", "ff01::1")
+        self.pg8.add_stream(tx)
+
+        self.pg_enable_capture(self.pg_interfaces)
+        self.pg_start()
+
+        # We expect replications on Pg1, 2,
+        self.verify_capture_ip6(self.pg1, tx)
+        self.verify_capture_ip6(self.pg2, tx)
 
 if __name__ == '__main__':
     unittest.main(testRunner=VppTestRunner)
diff --git a/test/test_mpls.py b/test/test_mpls.py
index b2226a74..460a32d1 100644
--- a/test/test_mpls.py
+++ b/test/test_mpls.py
@@ -6,7 +6,7 @@ import socket
 from framework import VppTestCase, VppTestRunner
 from vpp_ip_route import VppIpRoute, VppRoutePath, VppMplsRoute, \
     VppMplsIpBind, VppIpMRoute, VppMRoutePath, \
-    MRouteItfFlags, MRouteEntryFlags, DpoProto
+    MRouteItfFlags, MRouteEntryFlags, DpoProto, VppIpTable, VppMplsTable
 from vpp_mpls_tunnel_interface import VppMPLSTunnelInterface
 
 from scapy.packet import Raw
@@ -60,9 +60,23 @@ class TestMPLS(VppTestCase):
         # setup both interfaces
         # assign them different tables.
         table_id = 0
+        self.tables = []
+
+        tbl = VppMplsTable(self, 0)
+        tbl.add_vpp_config()
+        self.tables.append(tbl)
 
         for i in self.pg_interfaces:
             i.admin_up()
+
+            if table_id != 0:
+                tbl = VppIpTable(self, table_id)
+                tbl.add_vpp_config()
+                self.tables.append(tbl)
+                tbl = VppIpTable(self, table_id, is_ip6=1)
+                tbl.add_vpp_config()
+                self.tables.append(tbl)
+
             i.set_table_ip4(table_id)
             i.set_table_ip6(table_id)
             i.config_ip4()
@@ -73,12 +87,15 @@ class TestMPLS(VppTestCase):
             table_id += 1
 
     def tearDown(self):
-        super(TestMPLS, self).tearDown()
         for i in self.pg_interfaces:
             i.unconfig_ip4()
             i.unconfig_ip6()
             i.ip6_disable()
+            i.set_table_ip4(0)
+            i.set_table_ip6(0)
+            i.disable_mpls()
             i.admin_down()
+        super(TestMPLS, self).tearDown()
 
     # the default of 64 matches the IP packet TTL default
     def create_stream_labelled_ip4(
@@ -1092,6 +1109,9 @@ class TestMPLSDisabled(VppTestCase):
         # create 2 pg interfaces
         self.create_pg_interfaces(range(2))
 
+        self.tbl = VppMplsTable(self, 0)
+        self.tbl.add_vpp_config()
+
         # PG0 is MPLS enalbed
         self.pg0.admin_up()
         self.pg0.config_ip4()
@@ -1102,11 +1122,13 @@ class TestMPLSDisabled(VppTestCase):
         self.pg1.admin_up()
 
     def tearDown(self):
-        super(TestMPLSDisabled, self).tearDown()
         for i in self.pg_interfaces:
             i.unconfig_ip4()
             i.admin_down()
 
+        self.pg0.disable_mpls()
+        super(TestMPLSDisabled, self).tearDown()
+
     def send_and_assert_no_replies(self, intf, pkts, remark):
         intf.add_stream(pkts)
         self.pg_enable_capture(self.pg_interfaces)
@@ -1174,6 +1196,13 @@ class TestMPLSPIC(VppTestCase):
         # create 2 pg interfaces
         self.create_pg_interfaces(range(4))
 
+        mpls_tbl = VppMplsTable(self, 0)
+        mpls_tbl.add_vpp_config()
+        tbl4 = VppIpTable(self, 1)
+        tbl4.add_vpp_config()
+        tbl6 = VppIpTable(self, 1, is_ip6=1)
+        tbl6.add_vpp_config()
+
         # core links
         self.pg0.admin_up()
         self.pg0.config_ip4()
@@ -1201,14 +1230,15 @@ class TestMPLSPIC(VppTestCase):
         self.pg3.resolve_ndp()
 
     def tearDown(self):
-        super(TestMPLSPIC, self).tearDown()
         self.pg0.disable_mpls()
+        self.pg1.disable_mpls()
         for i in self.pg_interfaces:
             i.unconfig_ip4()
             i.unconfig_ip6()
             i.set_table_ip4(0)
             i.set_table_ip6(0)
             i.admin_down()
+        super(TestMPLSPIC, self).tearDown()
 
     def test_mpls_ibgp_pic(self):
         """ MPLS iBGP PIC edge convergence
@@ -1534,24 +1564,30 @@ class TestMPLSL2(VppTestCase):
         # create 2 pg interfaces
         self.create_pg_interfaces(range(2))
 
+        # create the default MPLS table
+        self.tables = []
+        tbl = VppMplsTable(self, 0)
+        tbl.add_vpp_config()
+        self.tables.append(tbl)
+
         # use pg0 as the core facing interface
         self.pg0.admin_up()
         self.pg0.config_ip4()
         self.pg0.resolve_arp()
         self.pg0.enable_mpls()
 
-        # use the other 2 for customer facg L2 links
+        # use the other 2 for customer facing L2 links
         for i in self.pg_interfaces[1:]:
             i.admin_up()
 
     def tearDown(self):
-        super(TestMPLSL2, self).tearDown()
         for i in self.pg_interfaces[1:]:
             i.admin_down()
 
         self.pg0.disable_mpls()
         self.pg0.unconfig_ip4()
         self.pg0.admin_down()
+        super(TestMPLSL2, self).tearDown()
 
     def verify_capture_tunneled_ethernet(self, capture, sent, mpls_labels,
                                          ttl=255, top=None):
diff --git a/test/test_nat.py b/test/test_nat.py
index 1f2d17ab..73e9e217 100644
--- a/test/test_nat.py
+++ b/test/test_nat.py
@@ -549,6 +549,8 @@ class TestNAT44(MethodHolder):
             cls.pg0.configure_ipv4_neighbors()
 
             cls.overlapping_interfaces = list(list(cls.pg_interfaces[4:7]))
+            cls.vapi.ip_table_add_del(10, is_add=1)
+            cls.vapi.ip_table_add_del(20, is_add=1)
 
             cls.pg4._local_ip4 = "172.16.255.1"
             cls.pg4._local_ip4n = socket.inet_pton(socket.AF_INET, i.local_ip4)
@@ -1797,6 +1799,8 @@ class TestNAT44(MethodHolder):
 
         self.pg0.unconfig_ip4()
         self.pg1.unconfig_ip4()
+        self.vapi.ip_table_add_del(vrf_id1, is_add=1)
+        self.vapi.ip_table_add_del(vrf_id2, is_add=1)
         self.pg0.set_table_ip4(vrf_id1)
         self.pg1.set_table_ip4(vrf_id2)
         self.pg0.config_ip4()
@@ -1825,6 +1829,13 @@ class TestNAT44(MethodHolder):
         capture = self.pg2.get_capture(len(pkts))
         self.verify_capture_out(capture, nat_ip2)
 
+        self.pg0.unconfig_ip4()
+        self.pg1.unconfig_ip4()
+        self.pg0.set_table_ip4(0)
+        self.pg1.set_table_ip4(0)
+        self.vapi.ip_table_add_del(vrf_id1, is_add=0)
+        self.vapi.ip_table_add_del(vrf_id2, is_add=0)
+
     def test_vrf_feature_independent(self):
         """ NAT44 tenant VRF independent address pool mode """
 
@@ -3042,6 +3053,8 @@ class TestNAT64(MethodHolder):
             cls.ip6_interfaces.append(cls.pg_interfaces[2])
             cls.ip4_interfaces = list(cls.pg_interfaces[1:2])
 
+            cls.vapi.ip_table_add_del(cls.vrf1_id, is_add=1, is_ipv6=1)
+
             cls.pg_interfaces[2].set_table_ip6(cls.vrf1_id)
 
             cls.pg0.generate_remote_hosts(2)
diff --git a/test/test_neighbor.py b/test/test_neighbor.py
index 1c7cc267..68dde2fb 100644
--- a/test/test_neighbor.py
+++ b/test/test_neighbor.py
@@ -5,7 +5,8 @@ from socket import AF_INET, AF_INET6, inet_pton
 
 from framework import VppTestCase, VppTestRunner
 from vpp_neighbor import VppNeighbor, find_nbr
-from vpp_ip_route import VppIpRoute, VppRoutePath, find_route
+from vpp_ip_route import VppIpRoute, VppRoutePath, find_route, \
+    VppIpTable
 
 from scapy.packet import Raw
 from scapy.layers.l2 import Ether, ARP, Dot1Q
@@ -39,11 +40,13 @@ class ARPTestCase(VppTestCase):
         self.pg1.config_ip6()
 
         # pg3 in a different VRF
+        self.tbl = VppIpTable(self, 1)
+        self.tbl.add_vpp_config()
+
         self.pg3.set_table_ip4(1)
         self.pg3.config_ip4()
 
     def tearDown(self):
-        super(ARPTestCase, self).tearDown()
         self.pg0.unconfig_ip4()
         self.pg0.unconfig_ip6()
 
@@ -51,10 +54,13 @@ class ARPTestCase(VppTestCase):
         self.pg1.unconfig_ip6()
 
         self.pg3.unconfig_ip4()
+        self.pg3.set_table_ip4(0)
 
         for i in self.pg_interfaces:
             i.admin_down()
 
+        super(ARPTestCase, self).tearDown()
+
     def verify_arp_req(self, rx, smac, sip, dip):
         ether = rx[Ether]
         self.assertEqual(ether.dst, "ff:ff:ff:ff:ff:ff")
@@ -1080,6 +1086,62 @@ class ARPTestCase(VppTestCase):
                        self.pg0.remote_ip4,
                        self.pg1.remote_hosts[1].ip4)
 
+    def test_arp_static(self):
+        """ ARP Static"""
+        self.pg2.generate_remote_hosts(3)
+
+        #
+        # Add a static ARP entry
+        #
+        static_arp = VppNeighbor(self,
+                                 self.pg2.sw_if_index,
+                                 self.pg2.remote_hosts[1].mac,
+                                 self.pg2.remote_hosts[1].ip4,
+                                 is_static=1)
+        static_arp.add_vpp_config()
+
+        #
+        # Add the connected prefix to the interface
+        #
+        self.pg2.config_ip4()
+
+        #
+        # We should now find the adj-fib
+        #
+        self.assertTrue(find_nbr(self,
+                                 self.pg2.sw_if_index,
+                                 self.pg2.remote_hosts[1].ip4,
+                                 is_static=1))
+        self.assertTrue(find_route(self,
+                                   self.pg2.remote_hosts[1].ip4,
+                                   32))
+
+        #
+        # remove the connected
+        #
+        self.pg2.unconfig_ip4()
+
+        #
+        # put the interface into table 1
+        #
+        self.pg2.set_table_ip4(1)
+
+        #
+        # configure the same connected and expect to find the
+        # adj fib in the new table
+        #
+        self.pg2.config_ip4()
+        self.assertTrue(find_route(self,
+                                   self.pg2.remote_hosts[1].ip4,
+                                   32,
+                                   table_id=1))
+
+        #
+        # clean-up
+        #
+        self.pg2.unconfig_ip4()
+        self.pg2.set_table_ip4(0)
+
 
 if __name__ == '__main__':
     unittest.main(testRunner=VppTestRunner)
diff --git a/test/vpp_ip_route.py b/test/vpp_ip_route.py
index 2c489e3c..b7993793 100644
--- a/test/vpp_ip_route.py
+++ b/test/vpp_ip_route.py
@@ -54,6 +54,46 @@ def find_route(test, ip_addr, len, table_id=0, inet=AF_INET):
     return False
 
 
+class VppIpTable(VppObject):
+
+    def __init__(self,
+                 test,
+                 table_id,
+                 is_ip6=0):
+        self._test = test
+        self.table_id = table_id
+        self.is_ip6 = is_ip6
+
+    def add_vpp_config(self):
+        self._test.vapi.ip_table_add_del(
+            self.table_id,
+            is_ipv6=self.is_ip6,
+            is_add=1)
+        self._test.registry.register(self, self._test.logger)
+
+    def remove_vpp_config(self):
+        self._test.vapi.ip_table_add_del(
+            self.table_id,
+            is_ipv6=self.is_ip6,
+            is_add=0)
+
+    def query_vpp_config(self):
+        # find the default route
+        return find_route(self._test,
+                          "::" if self.is_ip6 else "0.0.0.0",
+                          0,
+                          self.table_id,
+                          inet=AF_INET6 if self.is_ip6 == 1 else AF_INET)
+
+    def __str__(self):
+        return self.object_id()
+
+    def object_id(self):
+        return ("table-%s-%d" %
+                ("v6" if self.is_ip6 == 1 else "v4",
+                 self.table_id))
+
+
 class VppRoutePath(object):
 
     def __init__(
@@ -391,6 +431,39 @@ class VppMplsIpBind(VppObject):
                    self.dest_addr_len))
 
 
+class VppMplsTable(VppObject):
+
+    def __init__(self,
+                 test,
+                 table_id):
+        self._test = test
+        self.table_id = table_id
+
+    def add_vpp_config(self):
+        self._test.vapi.mpls_table_add_del(
+            self.table_id,
+            is_add=1)
+        self._test.registry.register(self, self._test.logger)
+
+    def remove_vpp_config(self):
+        self._test.vapi.mpls_table_add_del(
+            self.table_id,
+            is_add=0)
+
+    def query_vpp_config(self):
+        # find the default route
+        dump = self._test.vapi.mpls_fib_dump()
+        if len(dump):
+            return True
+        return False
+
+    def __str__(self):
+        return self.object_id()
+
+    def object_id(self):
+        return ("table-mpls-%d" % (self.table_id))
+
+
 class VppMplsRoute(VppObject):
     """
     MPLS Route/LSP
diff --git a/test/vpp_papi_provider.py b/test/vpp_papi_provider.py
index b70da026..519aff80 100644
--- a/test/vpp_papi_provider.py
+++ b/test/vpp_papi_provider.py
@@ -652,6 +652,24 @@ class VppPapiProvider(object):
         return self.api(self.papi.delete_loopback,
                         {'sw_if_index': sw_if_index, })
 
+    def ip_table_add_del(self,
+                         table_id,
+                         is_add=1,
+                         is_ipv6=0):
+        """
+
+        :param table_id
+        :param is_add:  (Default value = 1)
+        :param is_ipv6:  (Default value = 0)
+
+        """
+
+        return self.api(
+            self.papi.ip_table_add_del,
+            {'table_id': table_id,
+             'is_add': is_add,
+             'is_ipv6': is_ipv6})
+
     def ip_add_del_route(
             self,
             dst_address,
@@ -664,7 +682,6 @@ class VppPapiProvider(object):
             next_hop_n_out_labels=0,
             next_hop_out_label_stack=[],
             next_hop_via_label=MPLS_LABEL_INVALID,
-            create_vrf_if_needed=0,
             is_resolve_host=0,
             is_resolve_attached=0,
             classify_table_index=0xFFFFFFFF,
@@ -687,7 +704,6 @@ class VppPapiProvider(object):
         :param vrf_id:  (Default value = 0)
         :param lookup_in_vrf:  (Default value = 0)
         :param classify_table_index:  (Default value = 0xFFFFFFFF)
-        :param create_vrf_if_needed:  (Default value = 0)
         :param is_add:  (Default value = 1)
         :param is_drop:  (Default value = 0)
         :param is_ipv6:  (Default value = 0)
@@ -707,7 +723,6 @@ class VppPapiProvider(object):
              'table_id': table_id,
              'classify_table_index': classify_table_index,
              'next_hop_table_id': next_hop_table_id,
-             'create_vrf_if_needed': create_vrf_if_needed,
              'is_add': is_add,
              'is_drop': is_drop,
              'is_unreach': is_unreach,
@@ -912,6 +927,22 @@ class VppPapiProvider(object):
     def mpls_fib_dump(self):
         return self.api(self.papi.mpls_fib_dump, {})
 
+    def mpls_table_add_del(
+            self,
+            table_id,
+            is_add=1):
+        """
+
+        :param table_id
+        :param is_add:  (Default value = 1)
+
+        """
+
+        return self.api(
+            self.papi.mpls_table_add_del,
+            {'mt_table_id': table_id,
+             'mt_is_add': is_add})
+
     def mpls_route_add_del(
             self,
             label,
@@ -925,7 +956,6 @@ class VppPapiProvider(object):
             next_hop_n_out_labels=0,
             next_hop_out_label_stack=[],
             next_hop_via_label=MPLS_LABEL_INVALID,
-            create_vrf_if_needed=0,
             is_resolve_host=0,
             is_resolve_attached=0,
             is_interface_rx=0,
@@ -947,7 +977,6 @@ class VppPapiProvider(object):
         :param vrf_id:  (Default value = 0)
         :param lookup_in_vrf:  (Default value = 0)
         :param classify_table_index:  (Default value = 0xFFFFFFFF)
-        :param create_vrf_if_needed:  (Default value = 0)
         :param is_add:  (Default value = 1)
         :param is_drop:  (Default value = 0)
         :param is_ipv6:  (Default value = 0)
@@ -968,7 +997,6 @@ class VppPapiProvider(object):
              'mr_eos': eos,
              'mr_table_id': table_id,
              'mr_classify_table_index': classify_table_index,
-             'mr_create_table_if_needed': create_vrf_if_needed,
              'mr_is_add': is_add,
              'mr_is_classify': is_classify,
              'mr_is_multipath': is_multipath,
@@ -994,7 +1022,6 @@ class VppPapiProvider(object):
             table_id=0,
             ip_table_id=0,
             is_ip4=1,
-            create_vrf_if_needed=0,
             is_bind=1):
         """
         """
@@ -1003,7 +1030,6 @@ class VppPapiProvider(object):
             {'mb_mpls_table_id': table_id,
              'mb_label': label,
              'mb_ip_table_id': ip_table_id,
-             'mb_create_table_if_needed': create_vrf_if_needed,
              'mb_is_bind': is_bind,
              'mb_is_ip4': is_ip4,
              'mb_address_length': dst_address_length,
@@ -1020,7 +1046,6 @@ class VppPapiProvider(object):
             next_hop_n_out_labels=0,
             next_hop_out_label_stack=[],
             next_hop_via_label=MPLS_LABEL_INVALID,
-            create_vrf_if_needed=0,
             is_add=1,
             l2_only=0,
             is_multicast=0):
@@ -1034,7 +1059,6 @@ class VppPapiProvider(object):
         :param vrf_id:  (Default value = 0)
         :param lookup_in_vrf:  (Default value = 0)
         :param classify_table_index:  (Default value = 0xFFFFFFFF)
-        :param create_vrf_if_needed:  (Default value = 0)
         :param is_add:  (Default value = 1)
         :param is_drop:  (Default value = 0)
         :param is_ipv6:  (Default value = 0)
@@ -1844,7 +1868,6 @@ class VppPapiProvider(object):
                           i_flags,
                           rpf_id=0,
                           table_id=0,
-                          create_vrf_if_needed=0,
                           is_add=1,
                           is_ipv6=0,
                           is_local=0):
@@ -1857,7 +1880,6 @@ class VppPapiProvider(object):
              'itf_flags': i_flags,
              'table_id': table_id,
              'rpf_id': rpf_id,
-             'create_vrf_if_needed': create_vrf_if_needed,
              'is_add': is_add,
              'is_ipv6': is_ipv6,
              'is_local': is_local,
-- 
cgit 1.2.3-korg


From 088f042400fe104c86c86fb0de04aeb4b8013e74 Mon Sep 17 00:00:00 2001
From: Ole Troan <ot@cisco.com>
Date: Fri, 20 Oct 2017 13:28:20 +0200
Subject: VPP-1033: Python API support arbitrary sized input parameters.

Dynamically calculate the required buffer size to pack into based on
message definition. Also add input parameter length checking.

Change-Id: I7633bec596e4833bb328fbf63a65b866c7985de5
Signed-off-by: Ole Troan <ot@cisco.com>
(cherry picked from commit 895b6e8b4408108a9b5cea99dcb378c3524b18b2)
---
 src/vpp-api/python/vpp_papi.py | 62 +++++++++++++++++++++++++++++++++---------
 test/test_acl_plugin.py        |  4 +--
 test/test_dhcp.py              |  3 +-
 test/test_nat.py               | 17 +++++++-----
 test/test_papi.py              | 31 +++++++++++++++++++++
 test/vpp_papi_provider.py      |  4 +--
 6 files changed, 96 insertions(+), 25 deletions(-)
 create mode 100644 test/test_papi.py

(limited to 'test/test_dhcp.py')

diff --git a/src/vpp-api/python/vpp_papi.py b/src/vpp-api/python/vpp_papi.py
index 14f367c2..7b66c0f4 100644
--- a/src/vpp-api/python/vpp_papi.py
+++ b/src/vpp-api/python/vpp_papi.py
@@ -129,7 +129,6 @@ class VPP():
         self.messages = {}
         self.id_names = []
         self.id_msgdef = []
-        self.buffersize = 10000
         self.connected = False
         self.header = struct.Struct('>HI')
         self.apifiles = []
@@ -199,35 +198,45 @@ class VPP():
             if not vl:
                 if e > 0 and t == 'u8':
                     # Fixed byte array
-                    return struct.Struct('>' + str(e) + 's')
+                    s = struct.Struct('>' + str(e) + 's')
+                    return s.size, s
                 if e > 0:
                     # Fixed array of base type
-                    return [e, struct.Struct('>' + base_types[t])]
+                    s = struct.Struct('>' + base_types[t])
+                    return s.size, [e, s]
                 elif e == 0:
                     # Old style variable array
-                    return [-1, struct.Struct('>' + base_types[t])]
+                    s = struct.Struct('>' + base_types[t])
+                    return s.size, [-1, s]
             else:
                 # Variable length array
-                return [vl, struct.Struct('>s')] if t == 'u8' else \
-                    [vl, struct.Struct('>' + base_types[t])]
+                if t == 'u8':
+                    s = struct.Struct('>s')
+                    return s.size, [vl, s]
+                else:
+                    s = struct.Struct('>' + base_types[t])
+                return s.size, [vl, s]
 
-            return struct.Struct('>' + base_types[t])
+            s = struct.Struct('>' + base_types[t])
+            return s.size, s
 
         if t in self.messages:
+            size = self.messages[t]['sizes'][0]
+
             # Return a list in case of array
             if e > 0 and not vl:
-                return [e, lambda self, encode, buf, offset, args: (
+                return size, [e, lambda self, encode, buf, offset, args: (
                     self.__struct_type(encode, self.messages[t], buf, offset,
                                        args))]
             if vl:
-                return [vl, lambda self, encode, buf, offset, args: (
+                return size, [vl, lambda self, encode, buf, offset, args: (
                     self.__struct_type(encode, self.messages[t], buf, offset,
                                        args))]
             elif e == 0:
                 # Old style VLA
                 raise NotImplementedError(1,
                                           'No support for compound types ' + t)
-            return lambda self, encode, buf, offset, args: (
+            return size, lambda self, encode, buf, offset, args: (
                 self.__struct_type(encode, self.messages[t], buf, offset, args)
             )
 
@@ -250,13 +259,14 @@ class VPP():
                                  ' used in call to: ' + \
                                  self.id_names[kwargs['_vl_msg_id']] + '()' )
 
-
         for k, v in vpp_iterator(msgdef['args']):
             off += size
             if k in kwargs:
                 if type(v) is list:
                     if callable(v[1]):
                         e = kwargs[v[0]] if v[0] in kwargs else v[0]
+                        if e != len(kwargs[k]):
+                            raise (ValueError(1, 'Input list length mismatch: %s (%s != %s)' %  (k, e, len(kwargs[k]))))
                         size = 0
                         for i in range(e):
                             size += v[1](self, True, buf, off + size,
@@ -264,6 +274,8 @@ class VPP():
                     else:
                         if v[0] in kwargs:
                             l = kwargs[v[0]]
+                            if l != len(kwargs[k]):
+                                raise ValueError(1, 'Input list length mistmatch: %s (%s != %s)' % (k, l, len(kwargs[k])))
                         else:
                             l = len(kwargs[k])
                         if v[1].size == 1:
@@ -278,6 +290,8 @@ class VPP():
                     if callable(v):
                         size = v(self, True, buf, off, kwargs[k])
                     else:
+                        if type(kwargs[k]) is str and v.size < len(kwargs[k]):
+                            raise ValueError(1, 'Input list length mistmatch: %s (%s < %s)' % (k, v.size, len(kwargs[k])))
                         v.pack_into(buf, off, kwargs[k])
                         size = v.size
             else:
@@ -290,9 +304,17 @@ class VPP():
             return self.messages[name]
         return None
 
+    def get_size(self, sizes, kwargs):
+        total_size = sizes[0]
+        for e in sizes[1]:
+            if e in kwargs and type(kwargs[e]) is list:
+                total_size += len(kwargs[e]) * sizes[1][e]
+        return total_size
+
     def encode(self, msgdef, kwargs):
         # Make suitably large buffer
-        buf = bytearray(self.buffersize)
+    	size = self.get_size(msgdef['sizes'], kwargs)
+        buf = bytearray(size)
         offset = 0
         size = self.__struct_type(True, msgdef, buf, offset, kwargs)
         return buf[:offset + size]
@@ -360,6 +382,8 @@ class VPP():
         argtypes = collections.OrderedDict()
         fields = []
         msg = {}
+        total_size = 0
+        sizes = {}
         for i, f in enumerate(msgdef):
             if type(f) is dict and 'crc' in f:
                 msg['crc'] = f['crc']
@@ -368,7 +392,18 @@ class VPP():
             field_name = f[1]
             if len(f) == 3 and f[2] == 0 and i != len(msgdef) - 2:
                 raise ValueError('Variable Length Array must be last: ' + name)
-            args[field_name] = self.__struct(*f)
+            size, s = self.__struct(*f)
+            args[field_name] = s
+            if type(s) == list and type(s[0]) == int and type(s[1]) == struct.Struct:
+                if s[0] < 0:
+                    sizes[field_name] = size
+                else:
+                    sizes[field_name] = size
+                    total_size += s[0] * size
+            else:
+                sizes[field_name] = size
+                total_size += size
+
             argtypes[field_name] = field_type
             if len(f) == 4:  # Find offset to # elements field
                 idx = list(args.keys()).index(f[3]) - i
@@ -380,6 +415,7 @@ class VPP():
         self.messages[name]['args'] = args
         self.messages[name]['argtypes'] = argtypes
         self.messages[name]['typeonly'] = typeonly
+        self.messages[name]['sizes'] = [total_size, sizes]
         return self.messages[name]
 
     def add_type(self, name, typedef):
diff --git a/test/test_acl_plugin.py b/test/test_acl_plugin.py
index 97fca1a1..cd375a2c 100644
--- a/test/test_acl_plugin.py
+++ b/test/test_acl_plugin.py
@@ -532,7 +532,7 @@ class TestACLplugin(VppTestCase):
                                  r[i_rule][rule_key])
 
         # Add a deny-1234 ACL
-        r_deny = ({'is_permit': 0, 'is_ipv6': 0, 'proto': 17,
+        r_deny = [{'is_permit': 0, 'is_ipv6': 0, 'proto': 17,
                    'srcport_or_icmptype_first': 1234,
                    'srcport_or_icmptype_last': 1235,
                    'src_ip_prefix_len': 0,
@@ -549,7 +549,7 @@ class TestACLplugin(VppTestCase):
                    'dstport_or_icmpcode_first': 0,
                    'dstport_or_icmpcode_last': 0,
                    'dst_ip_addr': '\x00\x00\x00\x00',
-                   'dst_ip_prefix_len': 0})
+                   'dst_ip_prefix_len': 0}]
 
         reply = self.vapi.acl_add_replace(acl_index=4294967295, r=r_deny,
                                           tag="deny 1234;permit all")
diff --git a/test/test_dhcp.py b/test/test_dhcp.py
index fe97f6c9..42b80af3 100644
--- a/test/test_dhcp.py
+++ b/test/test_dhcp.py
@@ -19,6 +19,7 @@ from scapy.layers.dhcp6 import DHCP6, DHCP6_Solicit, DHCP6_RelayForward, \
 from socket import AF_INET, AF_INET6
 from scapy.utils import inet_pton, inet_ntop
 from scapy.utils6 import in6_ptop
+from util import mactobinary
 
 DHCP4_CLIENT_PORT = 68
 DHCP4_SERVER_PORT = 67
@@ -1134,7 +1135,7 @@ class TestDHCP(VppTestCase):
 
         # remove the left over ARP entry
         self.vapi.ip_neighbor_add_del(self.pg2.sw_if_index,
-                                      self.pg2.remote_mac,
+                                      mactobinary(self.pg2.remote_mac),
                                       self.pg2.remote_ip4,
                                       is_add=0)
         #
diff --git a/test/test_nat.py b/test/test_nat.py
index 73e9e217..44758906 100644
--- a/test/test_nat.py
+++ b/test/test_nat.py
@@ -16,6 +16,7 @@ from util import ppp
 from ipfix import IPFIX, Set, Template, Data, IPFIXDecoder
 from time import sleep
 from util import ip4_range
+from util import mactobinary
 
 
 class MethodHolder(VppTestCase):
@@ -643,7 +644,9 @@ class TestNAT44(MethodHolder):
                 lb_sm.external_port,
                 lb_sm.protocol,
                 lb_sm.vrf_id,
-                is_add=0)
+                is_add=0,
+                local_num=0,
+                locals=[])
 
         adresses = self.vapi.nat44_address_dump()
         for addr in adresses:
@@ -1869,11 +1872,11 @@ class TestNAT44(MethodHolder):
         """ NAT44 interfaces without configured IP address """
 
         self.vapi.ip_neighbor_add_del(self.pg7.sw_if_index,
-                                      self.pg7.remote_mac,
+                                      mactobinary(self.pg7.remote_mac),
                                       self.pg7.remote_ip4n,
                                       is_static=1)
         self.vapi.ip_neighbor_add_del(self.pg8.sw_if_index,
-                                      self.pg8.remote_mac,
+                                      mactobinary(self.pg8.remote_mac),
                                       self.pg8.remote_ip4n,
                                       is_static=1)
 
@@ -1911,11 +1914,11 @@ class TestNAT44(MethodHolder):
         """ NAT44 interfaces without configured IP address - 1:1 NAT """
 
         self.vapi.ip_neighbor_add_del(self.pg7.sw_if_index,
-                                      self.pg7.remote_mac,
+                                      mactobinary(self.pg7.remote_mac),
                                       self.pg7.remote_ip4n,
                                       is_static=1)
         self.vapi.ip_neighbor_add_del(self.pg8.sw_if_index,
-                                      self.pg8.remote_mac,
+                                      mactobinary(self.pg8.remote_mac),
                                       self.pg8.remote_ip4n,
                                       is_static=1)
 
@@ -1957,11 +1960,11 @@ class TestNAT44(MethodHolder):
         self.icmp_id_out = 30608
 
         self.vapi.ip_neighbor_add_del(self.pg7.sw_if_index,
-                                      self.pg7.remote_mac,
+                                      mactobinary(self.pg7.remote_mac),
                                       self.pg7.remote_ip4n,
                                       is_static=1)
         self.vapi.ip_neighbor_add_del(self.pg8.sw_if_index,
-                                      self.pg8.remote_mac,
+                                      mactobinary(self.pg8.remote_mac),
                                       self.pg8.remote_ip4n,
                                       is_static=1)
 
diff --git a/test/test_papi.py b/test/test_papi.py
new file mode 100644
index 00000000..1a5f6ae6
--- /dev/null
+++ b/test/test_papi.py
@@ -0,0 +1,31 @@
+import binascii
+from framework import VppTestCase
+
+""" TestPAPI is a subclass of  VPPTestCase classes.
+
+Basic test for sanity check of the Python API binding.
+
+"""
+
+
+class TestPAPI(VppTestCase):
+    """ PAPI Test Case """
+
+    @classmethod
+    def setUpClass(cls):
+        super(TestPAPI, cls).setUpClass()
+        cls.v = cls.vapi.papi
+
+    def test_show_version(self):
+        rv = self.v.show_version()
+        self.assertEqual(rv.retval, 0)
+
+    def test_show_version_invalid_param(self):
+        self.assertRaises(ValueError, self.v.show_version, foobar='foo')
+
+    def test_u8_array(self):
+        rv = self.v.get_node_index(node_name='ip4-lookup')
+        self.assertEqual(rv.retval, 0)
+        node_name = 'X' * 100
+        self.assertRaises(ValueError, self.v.get_node_index,
+                          node_name=node_name)
diff --git a/test/vpp_papi_provider.py b/test/vpp_papi_provider.py
index b6759ec3..2eab6c6e 100644
--- a/test/vpp_papi_provider.py
+++ b/test/vpp_papi_provider.py
@@ -1290,7 +1290,7 @@ class VppPapiProvider(object):
             protocol,
             vrf_id=0,
             local_num=0,
-            locals=None,
+            locals=[],
             is_add=1):
         """Add/delete NAT44 load balancing static mapping
 
@@ -2004,7 +2004,7 @@ class VppPapiProvider(object):
                             eid,
                             eid_prefix_len=0,
                             vni=0,
-                            rlocs=None,
+                            rlocs=[],
                             rlocs_num=0,
                             is_src_dst=0,
                             is_add=1):
-- 
cgit 1.2.3-korg