aboutsummaryrefslogtreecommitdiffstats
path: root/src/plugins/map/gen-rules.py
blob: 533a8e237f75ccad3167493f23244de9da382936 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
#!/usr/bin/env python

# 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.


import ipaddress
import argparse
import sys

# map add domain ip4-pfx <pfx> ip6-pfx ::/0 ip6-src <ip6-src> ea-bits-len 0 psid-offset 6 psid-len 6
# map add rule index <0> psid <psid> ip6-dst <ip6-dst>

parser = argparse.ArgumentParser(description='MAP VPP configuration generator')
parser.add_argument('-t', action="store", dest="mapmode")
args = parser.parse_args()

#
# 1:1 Shared IPv4 address, shared BR
#
def shared11br():
    ip4_pfx = ipaddress.ip_network('20.0.0.0/16')
    ip6_dst = ipaddress.ip_network('bbbb::/32')
    psid_len = 6
    for i in range(ip4_pfx.num_addresses):
        print("map add domain ip4-pfx " + str(ip4_pfx[i]) +  "/32 ip6-pfx ::/0 ip6-shared-src cccc:bbbb::1",
              "ea-bits-len 0 psid-offset 6 psid-len", psid_len)
        for psid in range(0x1 << psid_len):
            print("map add rule index", i, "psid", psid, "ip6-dst", ip6_dst[(i * (0x1<<psid_len)) + psid])


#
# 1:1 Shared IPv4 address
#
def shared11():
    ip4_pfx = ipaddress.ip_network('20.0.0.0/16')
    ip6_src = ipaddress.ip_network('cccc:bbbb::/64')
    ip6_dst = ipaddress.ip_network('bbbb::/32')
    psid_len = 6
    for i in range(ip4_pfx.num_addresses):
        print("map add domain ip4-pfx " + str(ip4_pfx[i]) +  "/32 ip6-pfx ::/0 ip6-src", ip6_src[i],
              "ea-bits-len 0 psid-offset 6 psid-len", psid_len)
        for psid in range(0x1 << psid_len):
            print("map add rule index", i, "psid", psid, "ip6-dst", ip6_dst[(i * (0x1<<psid_len)) + psid])

#
# 1:1 Shared IPv4 address small
#
def smallshared11():
    ip4_pfx = ipaddress.ip_network('20.0.0.0/24')
    ip6_src = ipaddress.ip_network('cccc:bbbb::/64')
    ip6_dst = ipaddress.ip_network('bbbb::/32')
    psid_len = 6
    for i in range(ip4_pfx.num_addresses):
        print("map add domain ip4-pfx " + str(ip4_pfx[i]) +  "/32 ip6-pfx ::/0 ip6-src", ip6_src[i],
              "ea-bits-len 0 psid-offset 6 psid-len", psid_len)
        for psid in range(0x1 << psid_len):
            print("map add rule index", i, "psid", psid, "ip6-dst", ip6_dst[(i * (0x1<<psid_len)) + psid])

#
# 1:1 Full IPv4 address
#
def full11():
    ip4_pfx = ipaddress.ip_network('20.0.0.0/16')
    ip6_src = ipaddress.ip_network('cccc:bbbb::/64')
    ip6_dst = ipaddress.ip_network('bbbb::/32')
    psid_len = 0
    for i in range(ip4_pfx.num_addresses):
        print("map add domain ip4-pfx " + str(ip4_pfx[i]) +  "/32 ip6-pfx " + str(ip6_dst[i]) + "/128 ip6-src", ip6_src[i],
              "ea-bits-len 0 psid-offset 0 psid-len 0")
def full11br():
    ip4_pfx = ipaddress.ip_network('20.0.0.0/16')
    ip6_dst = ipaddress.ip_network('bbbb::/32')
    psid_len = 0
    for i in range(ip4_pfx.num_addresses):
        print("map add domain ip4-pfx " + str(ip4_pfx[i]) +  "/32 ip6-pfx " + str(ip6_dst[i]) + "/128 ip6-shared-src cccc:bbbb::1",
              "ea-bits-len 0 psid-offset 0 psid-len 0")

#
# Algorithmic mapping Shared IPv4 address
#
def algo():
    print("map add domain ip4-pfx 20.0.0.0/24 ip6-pfx bbbb::/32 ip6-src cccc:bbbb::1 ea-bits-len 16 psid-offset 6 psid-len 8")
    print("map add domain ip4-pfx 20.0.1.0/24 ip6-pfx bbbb:1::/32 ip6-src cccc:bbbb::2 ea-bits-len 8 psid-offset 0 psid-len 0")

#
# IP4 forwarding
#
def ip4():
    ip4_pfx = ipaddress.ip_network('20.0.0.0/16')
    for i in range(ip4_pfx.num_addresses):
        print("ip route add " + str(ip4_pfx[i]) +  "/32 via 172.16.0.2")


globals()[args.mapmode]()
882' href='#n882'>882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204
/*
 * 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 <vlib/vlib.h>
#include <vnet/dhcp/client.h>
#include <vnet/dhcp/dhcp_proxy.h>
#include <vnet/fib/fib_table.h>

dhcp_client_main_t dhcp_client_main;
static u8 *format_dhcp_client_state (u8 * s, va_list * va);
static vlib_node_registration_t dhcp_client_process_node;

#define foreach_dhcp_sent_packet_stat           \
_(DISCOVER, "DHCP discover packets sent")       \
_(OFFER, "DHCP offer packets sent")             \
_(REQUEST, "DHCP request packets sent")         \
_(ACK, "DHCP ack packets sent")

#define foreach_dhcp_error_counter                                      \
_(NOT_FOR_US, "DHCP packets for other hosts, dropped")                  \
_(NAK, "DHCP nak packets received")                                     \
_(NON_OFFER_DISCOVER, "DHCP non-offer packets in discover state")       \
_(ODDBALL, "DHCP non-ack, non-offer packets received")                  \
_(BOUND, "DHCP bind success")

typedef enum
{
#define _(sym,str) DHCP_STAT_##sym,
  foreach_dhcp_sent_packet_stat foreach_dhcp_error_counter
#undef _
    DHCP_STAT_UNKNOWN,
  DHCP_STAT_N_STAT,
} sample_error_t;

static char *dhcp_client_process_stat_strings[] = {
#define _(sym,string) string,
  foreach_dhcp_sent_packet_stat foreach_dhcp_error_counter
#undef _
    "DHCP unknown packets sent",
};


static void
dhcp_client_acquire_address (dhcp_client_main_t * dcm, dhcp_client_t * c)
{
  /*
   * Install any/all info gleaned from dhcp, right here
   */
  ip4_add_del_interface_address (dcm->vlib_main, c->sw_if_index,
				 (void *) &c->leased_address,
				 c->subnet_mask_width, 0 /*is_del */ );
}

static void
dhcp_client_release_address (dhcp_client_main_t * dcm, dhcp_client_t * c)
{
  /*
   * Remove any/all info gleaned from dhcp, right here. Caller(s)
   * have not wiped out the info yet.
   */

  ip4_add_del_interface_address (dcm->vlib_main, c->sw_if_index,
				 (void *) &c->leased_address,
				 c->subnet_mask_width, 1 /*is_del */ );
}

static void
set_l2_rewrite (dhcp_client_main_t * dcm, dhcp_client_t * c)
{
  /* Acquire the L2 rewrite string for the indicated sw_if_index */
  c->l2_rewrite = vnet_build_rewrite_for_sw_interface (dcm->vnet_main,
						       c->sw_if_index,
						       VNET_LINK_IP4,
						       0 /* broadcast */ );
}

void vl_api_rpc_call_main_thread (void *fp, u8 * data, u32 data_length);

static void
dhcp_client_proc_callback (uword * client_index)
{
  vlib_main_t *vm = vlib_get_main ();
  ASSERT (vlib_get_thread_index () == 0);
  vlib_process_signal_event (vm, dhcp_client_process_node.index,
			     EVENT_DHCP_CLIENT_WAKEUP, *client_index);
}

static void
dhcp_client_addr_callback (dhcp_client_t * c)
{
  dhcp_client_main_t *dcm = &dhcp_client_main;

  /* disable the feature */
  vnet_feature_enable_disable ("ip4-unicast",
			       "ip4-dhcp-client-detect",
			       c->sw_if_index, 0 /* disable */ , 0, 0);
  c->client_detect_feature_enabled = 0;

  /* if renewing the lease, the address and route have already been added */
  if (c->state == DHCP_BOUND)
    return;

  /* add the address to the interface */
  dhcp_client_acquire_address (dcm, c);

  /*
   * Configure default IP route:
   */
  if (c->router_address.as_u32)
    {
      fib_prefix_t all_0s = {
	.fp_len = 0,
	.fp_addr.ip4.as_u32 = 0x0,
	.fp_proto = FIB_PROTOCOL_IP4,
      };
      ip46_address_t nh = {
	.ip4 = c->router_address,
      };

      /* *INDENT-OFF* */
      fib_table_entry_path_add (
	fib_table_get_index_for_sw_if_index (
	  FIB_PROTOCOL_IP4,
	  c->sw_if_index),
	  &all_0s,
	  FIB_SOURCE_DHCP,
	  FIB_ENTRY_FLAG_NONE,
          DPO_PROTO_IP4,
          &nh, c->sw_if_index,
          ~0, 1, NULL,	// no label stack
          FIB_ROUTE_PATH_FLAG_NONE);
      /* *INDENT-ON* */
    }

  /*
   * Call the user's event callback to report DHCP information
   */
  if (c->event_callback)
    c->event_callback (c->client_index, c);
}

/*
 * dhcp_client_for_us - server-to-client callback.
 * Called from proxy_node.c:dhcp_proxy_to_client_input().
 * This function first decides that the packet in question is
 * actually for the dhcp client code in case we're also acting as
 * a dhcp proxy. Ay caramba, what a folly!
 */
int
dhcp_client_for_us (u32 bi, vlib_buffer_t * b,
		    ip4_header_t * ip,
		    udp_header_t * udp, dhcp_header_t * dhcp)
{
  dhcp_client_main_t *dcm = &dhcp_client_main;
  vlib_main_t *vm = dcm->vlib_main;
  dhcp_client_t *c;
  uword *p;
  f64 now = vlib_time_now (dcm->vlib_main);
  u8 dhcp_message_type = 0;
  dhcp_option_t *o;

  /*
   * Doing dhcp client on this interface?
   * Presumably we will always receive dhcp clnt for-us pkts on
   * the interface that's asking for an address.
   */
  p = hash_get (dcm->client_by_sw_if_index,
		vnet_buffer (b)->sw_if_index[VLIB_RX]);
  if (p == 0)
    return 0;			/* no */

  c = pool_elt_at_index (dcm->clients, p[0]);

  /* Mixing dhcp relay and dhcp proxy? DGMS... */
  if (c->state == DHCP_BOUND && c->retry_count == 0)
    return 0;

  /* Packet not for us? Turf it... */
  if (memcmp (dhcp->client_hardware_address, c->client_hardware_address,
	      sizeof (c->client_hardware_address)))
    {
      vlib_node_increment_counter (vm, dhcp_client_process_node.index,
				   DHCP_STAT_NOT_FOR_US, 1);
      return 0;
    }

  /* 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 */  &&
	 (u8 *) o < (b->data + b->current_data + b->current_length))
    {
      switch (o->option)
	{
	case 53:		/* dhcp message type */
	  dhcp_message_type = o->data[0];
	  break;

	case 51:		/* lease time */
	  {
	    u32 lease_time_in_seconds =
	      clib_host_to_net_u32 (o->data_as_u32[0]);
	    // for debug: lease_time_in_seconds = 20; /*$$$$*/
	    c->lease_expires = now + (f64) lease_time_in_seconds;
	    c->lease_lifetime = lease_time_in_seconds;
	    /* Set a sensible default, in case we don't get opt 58 */
	    c->lease_renewal_interval = lease_time_in_seconds / 2;
	  }
	  break;

	case 58:		/* lease renew time in seconds */
	  {
	    u32 lease_renew_time_in_seconds =
	      clib_host_to_net_u32 (o->data_as_u32[0]);
	    c->lease_renewal_interval = lease_renew_time_in_seconds;
	  }
	  break;

	case 54:		/* dhcp server address */
	  c->dhcp_server.as_u32 = o->data_as_u32[0];
	  break;

	case 1:		/* subnet mask */
	  {
	    u32 subnet_mask = clib_host_to_net_u32 (o->data_as_u32[0]);
	    c->subnet_mask_width = count_set_bits (subnet_mask);
	  }
	  break;
	case 3:		/* router address */
	  {
	    u32 router_address = o->data_as_u32[0];
	    c->router_address.as_u32 = router_address;
	  }
	  break;

	case 12:		/* hostname */
	  {
	    /* Replace the existing hostname if necessary */
	    vec_free (c->hostname);
	    vec_validate (c->hostname, o->length - 1);
	    clib_memcpy (c->hostname, o->data, o->length);
	  }
	  break;

	  /* $$$$ Your message in this space, parse more options */
	default:
	  break;
	}

      o = (dhcp_option_t *) (((uword) o) + (o->length + 2));
    }

  switch (c->state)
    {
    case DHCP_DISCOVER:
      if (dhcp_message_type != DHCP_PACKET_OFFER)
	{
	  vlib_node_increment_counter (vm, dhcp_client_process_node.index,
				       DHCP_STAT_NON_OFFER_DISCOVER, 1);
	  c->next_transmit = now + 5.0;
	  break;
	}

      /* Received an offer, go send a request */
      c->state = DHCP_REQUEST;
      c->retry_count = 0;
      c->next_transmit = 0;	/* send right now... */
      /* Poke the client process, which will send the request */
      uword client_id = c - dcm->clients;
      vl_api_rpc_call_main_thread (dhcp_client_proc_callback,
				   (u8 *) & client_id, sizeof (uword));
      break;

    case DHCP_BOUND:
    case DHCP_REQUEST:
      if (dhcp_message_type == DHCP_PACKET_NAK)
	{
	  vlib_node_increment_counter (vm, dhcp_client_process_node.index,
				       DHCP_STAT_NAK, 1);
	  /* Probably never happens in bound state, but anyhow... */
	  if (c->state == DHCP_BOUND)
	    {
	      ip4_add_del_interface_address (dcm->vlib_main, c->sw_if_index,
					     (void *) &c->leased_address,
					     c->subnet_mask_width,
					     1 /*is_del */ );
	      vnet_feature_enable_disable ("ip4-unicast",
					   "ip4-dhcp-client-detect",
					   c->sw_if_index, 1 /* enable */ ,
					   0, 0);
	      c->client_detect_feature_enabled = 1;
	    }
	  /* Wipe out any memory of the address we had... */
	  c->state = DHCP_DISCOVER;
	  c->next_transmit = now;
	  c->retry_count = 0;
	  c->leased_address.as_u32 = 0;
	  c->subnet_mask_width = 0;
	  c->router_address.as_u32 = 0;
	  c->lease_renewal_interval = 0;
	  c->dhcp_server.as_u32 = 0;
	  break;
	}

      if (dhcp_message_type != DHCP_PACKET_ACK &&
	  dhcp_message_type != DHCP_PACKET_OFFER)
	{
	  vlib_node_increment_counter (vm, dhcp_client_process_node.index,
				       DHCP_STAT_NON_OFFER_DISCOVER, 1);
	  clib_warning ("sw_if_index %d state %U message type %d",
			c->sw_if_index, format_dhcp_client_state,
			c->state, dhcp_message_type);
	  c->next_transmit = now + 5.0;
	  break;
	}
      /* OK, we own the address (etc), add to the routing table(s) */
      vl_api_rpc_call_main_thread (dhcp_client_addr_callback,
				   (u8 *) c, sizeof (*c));

      c->state = DHCP_BOUND;
      c->retry_count = 0;
      c->next_transmit = now + (f64) c->lease_renewal_interval;
      c->lease_expires = now + (f64) c->lease_lifetime;
      vlib_node_increment_counter (vm, dhcp_client_process_node.index,
				   DHCP_STAT_BOUND, 1);
      break;

    default:
      clib_warning ("client %d bogus state %d", c - dcm->clients, c->state);
      break;
    }

  /* drop the pkt, return 1 */
  vlib_buffer_free (vm, &bi, 1);
  return 1;
}

static void
send_dhcp_pkt (dhcp_client_main_t * dcm, dhcp_client_t * c,
	       dhcp_packet_type_t type, int is_broadcast)
{
  vlib_main_t *vm = dcm->vlib_main;
  vnet_main_t *vnm = dcm->vnet_main;
  vnet_hw_interface_t *hw = vnet_get_sup_hw_interface (vnm, c->sw_if_index);
  vnet_sw_interface_t *sup_sw
    = vnet_get_sup_sw_interface (vnm, c->sw_if_index);
  vnet_sw_interface_t *sw = vnet_get_sw_interface (vnm, c->sw_if_index);
  vlib_buffer_t *b;
  u32 bi;
  ip4_header_t *ip;
  udp_header_t *udp;
  dhcp_header_t *dhcp;
  u32 *to_next;
  vlib_frame_t *f;
  dhcp_option_t *o;
  u16 udp_length, ip_length;
  u32 counter_index;

  /* Interface(s) down? */
  if ((hw->flags & VNET_HW_INTERFACE_FLAG_LINK_UP) == 0)
    return;
  if ((sup_sw->flags & VNET_SW_INTERFACE_FLAG_ADMIN_UP) == 0)
    return;
  if ((sw->flags & VNET_SW_INTERFACE_FLAG_ADMIN_UP) == 0)
    return;

  if (vlib_buffer_alloc (vm, &bi, 1) != 1)
    {
      clib_warning ("buffer allocation failure");
      c->next_transmit = 0;
      return;
    }

  /* Build a dhcpv4 pkt from whole cloth */
  b = vlib_get_buffer (vm, bi);

  ASSERT (b->current_data == 0);

  vnet_buffer (b)->sw_if_index[VLIB_RX] = c->sw_if_index;
  if (is_broadcast)
    {
      f = vlib_get_frame_to_node (vm, hw->output_node_index);
      vnet_buffer (b)->sw_if_index[VLIB_TX] = c->sw_if_index;
      clib_memcpy (b->data, c->l2_rewrite, vec_len (c->l2_rewrite));
      ip = (void *)
	(((u8 *) vlib_buffer_get_current (b)) + vec_len (c->l2_rewrite));
    }
  else
    {
      f = vlib_get_frame_to_node (vm, ip4_lookup_node.index);
      vnet_buffer (b)->sw_if_index[VLIB_TX] = ~0;	/* use interface VRF */
      ip = vlib_buffer_get_current (b);
    }

  /* Enqueue the packet right now */
  to_next = vlib_frame_vector_args (f);
  to_next[0] = bi;
  f->n_vectors = 1;

  if (is_broadcast)
    vlib_put_frame_to_node (vm, hw->output_node_index, f);
  else
    vlib_put_frame_to_node (vm, ip4_lookup_node.index, f);

  udp = (udp_header_t *) (ip + 1);
  dhcp = (dhcp_header_t *) (udp + 1);

  /* $$$ optimize, maybe */
  memset (ip, 0, sizeof (*ip) + sizeof (*udp) + sizeof (*dhcp));

  ip->ip_version_and_header_length = 0x45;
  ip->ttl = 128;
  ip->protocol = IP_PROTOCOL_UDP;

  if (is_broadcast)
    {
      /* src = 0.0.0.0, dst = 255.255.255.255 */
      ip->dst_address.as_u32 = ~0;
    }
  else
    {
      /* Renewing an active lease, plain old ip4 src/dst */
      ip->src_address.as_u32 = c->leased_address.as_u32;
      ip->dst_address.as_u32 = c->dhcp_server.as_u32;
    }

  udp->src_port = clib_host_to_net_u16 (UDP_DST_PORT_dhcp_to_client);
  udp->dst_port = clib_host_to_net_u16 (UDP_DST_PORT_dhcp_to_server);

  /* Send the interface MAC address */
  clib_memcpy (dhcp->client_hardware_address, c->l2_rewrite + 6, 6);

  /* And remember it for rx-packet-for-us checking */
  clib_memcpy (c->client_hardware_address, dhcp->client_hardware_address,
	       sizeof (c->client_hardware_address));

  /* Lease renewal, set up client_ip_address */
  if (is_broadcast == 0)
    dhcp->client_ip_address.as_u32 = c->leased_address.as_u32;

  dhcp->opcode = 1;		/* request, all we send */
  dhcp->hardware_type = 1;	/* ethernet */
  dhcp->hardware_address_length = 6;
  dhcp->transaction_identifier = c->transaction_id;
  dhcp->flags =
    clib_host_to_net_u16 (is_broadcast && c->set_broadcast_flag ?
			  DHCP_FLAG_BROADCAST : 0);
  dhcp->magic_cookie.as_u32 = DHCP_MAGIC;

  o = (dhcp_option_t *) dhcp->options;

  /* Send option 53, the DHCP message type */
  o->option = DHCP_PACKET_OPTION_MSG_TYPE;
  o->length = 1;
  o->data[0] = type;
  o = (dhcp_option_t *) (((uword) o) + (o->length + 2));

  /* Send option 57, max msg length */
  if (0 /* not needed, apparently */ )
    {
      o->option = 57;
      o->length = 2;
      {
	u16 *o2 = (u16 *) o->data;
	*o2 = clib_host_to_net_u16 (1152);
	o = (dhcp_option_t *) (((uword) o) + (o->length + 2));
      }
    }

  /*
   * If server ip address is available with non-zero value,
   * option 54 (DHCP Server Identifier) is sent.
   */
  if (c->dhcp_server.as_u32)
    {
      o->option = 54;
      o->length = 4;
      clib_memcpy (o->data, &c->dhcp_server.as_u32, 4);
      o = (dhcp_option_t *) (((uword) o) + (o->length + 2));
    }

  /* send option 50, requested IP address */
  if (c->leased_address.as_u32)
    {
      o->option = 50;
      o->length = 4;
      clib_memcpy (o->data, &c->leased_address.as_u32, 4);
      o = (dhcp_option_t *) (((uword) o) + (o->length + 2));
    }

  /* send option 12, host name */
  if (vec_len (c->hostname))
    {
      o->option = 12;
      o->length = vec_len (c->hostname);
      clib_memcpy (o->data, c->hostname, vec_len (c->hostname));
      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 */

  /*
   * send option 55, parameter request list
   * The current list - see below, matches the Linux dhcp client's list
   * Any specific dhcp server config and/or dhcp server may or may
   * not yield specific options.
   */
  o->option = 55;
  o->length = vec_len (c->option_55_data);
  clib_memcpy (o->data, c->option_55_data, vec_len (c->option_55_data));
  o = (dhcp_option_t *) (((uword) o) + (o->length + 2));

  /* End of list */
  o->option = 0xff;
  o->length = 0;
  o++;

  b->current_length = ((u8 *) o) - b->data;

  /* fix ip length, checksum and udp length */
  ip_length = vlib_buffer_length_in_chain (vm, b);
  if (is_broadcast)
    ip_length -= vec_len (c->l2_rewrite);

  ip->length = clib_host_to_net_u16 (ip_length);
  ip->checksum = ip4_header_checksum (ip);

  udp_length = ip_length - (sizeof (*ip));
  udp->length = clib_host_to_net_u16 (udp_length);

  switch (type)
    {
#define _(a,b) case DHCP_PACKET_##a: {counter_index = DHCP_STAT_##a; break;}
      foreach_dhcp_sent_packet_stat
#undef _
    default:
      counter_index = DHCP_STAT_UNKNOWN;
      break;
    }

  vlib_node_increment_counter (vm, dhcp_client_process_node.index,
			       counter_index, 1);
}

static int
dhcp_discover_state (dhcp_client_main_t * dcm, dhcp_client_t * c, f64 now)
{
  /*
   * State machine "DISCOVER" state. Send a dhcp discover packet,
   * eventually back off the retry rate.
   */

  if (c->client_detect_feature_enabled == 0)
    {
      vnet_feature_enable_disable ("ip4-unicast",
				   "ip4-dhcp-client-detect",
				   c->sw_if_index, 1 /* enable */ , 0, 0);
      c->client_detect_feature_enabled = 1;
    }

  send_dhcp_pkt (dcm, c, DHCP_PACKET_DISCOVER, 1 /* is_broadcast */ );

  c->retry_count++;
  if (c->retry_count > 10)
    c->next_transmit = now + 5.0;
  else
    c->next_transmit = now + 1.0;
  return 0;
}

static int
dhcp_request_state (dhcp_client_main_t * dcm, dhcp_client_t * c, f64 now)
{
  /*
   * State machine "REQUEST" state. Send a dhcp request packet,
   * eventually drop back to the discover state.
   */
  send_dhcp_pkt (dcm, c, DHCP_PACKET_REQUEST, 1 /* is_broadcast */ );

  c->retry_count++;
  if (c->retry_count > 7 /* lucky you */ )
    {
      c->state = DHCP_DISCOVER;
      c->next_transmit = now;
      c->retry_count = 0;
      return 1;
    }
  c->next_transmit = now + 1.0;
  return 0;
}

static int
dhcp_bound_state (dhcp_client_main_t * dcm, dhcp_client_t * c, f64 now)
{
  /*
   * State machine "BOUND" state. Send a dhcp request packet to renew
   * the lease.
   * Eventually, when the lease expires, forget the dhcp data
   * and go back to the stone age.
   */

  /*
   * We disable the client detect feature when we bind a
   * DHCP address. Turn it back on again on first renew attempt.
   * Otherwise, if the DHCP server replies we'll never see it.
   */
  if (c->client_detect_feature_enabled == 0)
    {
      vnet_feature_enable_disable ("ip4-unicast",
				   "ip4-dhcp-client-detect",
				   c->sw_if_index, 1 /* enable */ , 0, 0);
      c->client_detect_feature_enabled = 1;
    }

  send_dhcp_pkt (dcm, c, DHCP_PACKET_REQUEST, 0 /* is_broadcast */ );

  c->retry_count++;
  if (c->retry_count > 10)
    c->next_transmit = now + 5.0;
  else
    c->next_transmit = now + 1.0;

  if (now > c->lease_expires)
    {
      /* Remove the default route */
      if (c->router_address.as_u32)
	{
	  fib_prefix_t all_0s = {
	    .fp_len = 0,
	    .fp_addr.ip4.as_u32 = 0x0,
	    .fp_proto = FIB_PROTOCOL_IP4,
	  };
	  ip46_address_t nh = {
	    .ip4 = c->router_address,
	  };

	  fib_table_entry_path_remove (fib_table_get_index_for_sw_if_index
				       (FIB_PROTOCOL_IP4, c->sw_if_index),
				       &all_0s, FIB_SOURCE_DHCP,
				       DPO_PROTO_IP4, &nh, c->sw_if_index, ~0,
				       1, FIB_ROUTE_PATH_FLAG_NONE);
	}
      /* Remove the interface address */
      dhcp_client_release_address (dcm, c);
      c->state = DHCP_DISCOVER;
      c->next_transmit = now;
      c->retry_count = 0;
      /* Wipe out any memory of the address we had... */
      c->leased_address.as_u32 = 0;
      c->subnet_mask_width = 0;
      c->router_address.as_u32 = 0;
      c->lease_renewal_interval = 0;
      c->dhcp_server.as_u32 = 0;
      return 1;
    }
  return 0;
}

static f64
dhcp_client_sm (f64 now, f64 timeout, uword pool_index)
{
  dhcp_client_main_t *dcm = &dhcp_client_main;
  dhcp_client_t *c;

  /* deleted, pooched, yadda yadda yadda */
  if (pool_is_free_index (dcm->clients, pool_index))
    return timeout;

  c = pool_elt_at_index (dcm->clients, pool_index);

  /* Time for us to do something with this client? */
  if (now < c->next_transmit)
    return timeout;

again:
  switch (c->state)
    {
    case DHCP_DISCOVER:	/* send a discover */
      if (dhcp_discover_state (dcm, c, now))
	goto again;
      break;

    case DHCP_REQUEST:		/* send a request */
      if (dhcp_request_state (dcm, c, now))
	goto again;
      break;

    case DHCP_BOUND:		/* bound, renew needed? */
      if (dhcp_bound_state (dcm, c, now))
	goto again;
      break;

    default:
      clib_warning ("dhcp client %d bogus state %d",
		    c - dcm->clients, c->state);
      break;
    }

  if (c->next_transmit < now + timeout)
    return c->next_transmit - now;

  return timeout;
}

static uword
dhcp_client_process (vlib_main_t * vm,
		     vlib_node_runtime_t * rt, vlib_frame_t * f)
{
  f64 timeout = 100.0;
  f64 now;
  uword event_type;
  uword *event_data = 0;
  dhcp_client_main_t *dcm = &dhcp_client_main;
  dhcp_client_t *c;
  int i;

  while (1)
    {
      vlib_process_wait_for_event_or_clock (vm, timeout);

      event_type = vlib_process_get_events (vm, &event_data);

      now = vlib_time_now (vm);

      switch (event_type)
	{
	case EVENT_DHCP_CLIENT_WAKEUP:
	  for (i = 0; i < vec_len (event_data); i++)
	    timeout = dhcp_client_sm (now, timeout, event_data[i]);
	  break;

	case ~0:
          /* *INDENT-OFF* */
	  pool_foreach (c, dcm->clients,
          ({
            timeout = dhcp_client_sm (now, timeout,
                                      (uword) (c - dcm->clients));
          }));
          /* *INDENT-ON* */
	  if (pool_elts (dcm->clients) == 0)
	    timeout = 100.0;
	  break;
	}

      vec_reset_length (event_data);
    }

  /* NOTREACHED */
  return 0;
}

/* *INDENT-OFF* */
VLIB_REGISTER_NODE (dhcp_client_process_node,static) = {
    .function = dhcp_client_process,
    .type = VLIB_NODE_TYPE_PROCESS,
    .name = "dhcp-client-process",
    .process_log2_n_stack_bytes = 16,
    .n_errors = ARRAY_LEN(dhcp_client_process_stat_strings),
    .error_strings = dhcp_client_process_stat_strings,
};
/* *INDENT-ON* */

static u8 *
format_dhcp_client_state (u8 * s, va_list * va)
{
  dhcp_client_state_t state = va_arg (*va, dhcp_client_state_t);
  char *str = "BOGUS!";

  switch (state)
    {
#define _(a)                                    \
    case a:                                     \
      str = #a;                                 \
        break;
      foreach_dhcp_client_state;
#undef _
    default:
      break;
    }

  s = format (s, "%s", str);
  return s;
}

static u8 *
format_dhcp_client (u8 * s, va_list * va)
{
  dhcp_client_main_t *dcm = va_arg (*va, dhcp_client_main_t *);
  dhcp_client_t *c = va_arg (*va, dhcp_client_t *);
  int verbose = va_arg (*va, int);

  s = format (s, "[%d] %U state %U ", c - dcm->clients,
	      format_vnet_sw_if_index_name, dcm->vnet_main, c->sw_if_index,
	      format_dhcp_client_state, c->state);

  if (c->leased_address.as_u32)
    s = format (s, "addr %U/%d gw %U\n",
		format_ip4_address, &c->leased_address,
		c->subnet_mask_width, format_ip4_address, &c->router_address);
  else
    s = format (s, "no address\n");

  if (verbose)
    {
      s = format (s, "retry count %d, next xmt %.2f",
		  c->retry_count, c->next_transmit);
    }
  return s;
}

static clib_error_t *
show_dhcp_client_command_fn (vlib_main_t * vm,
			     unformat_input_t * input,
			     vlib_cli_command_t * cmd)
{
  dhcp_client_main_t *dcm = &dhcp_client_main;
  dhcp_client_t *c;
  int verbose = 0;
  u32 sw_if_index = ~0;
  uword *p;

  while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT)
    {
      if (unformat (input, "intfc %U",
		    unformat_vnet_sw_interface, dcm->vnet_main, &sw_if_index))
	;
      else if (unformat (input, "verbose"))
	verbose = 1;
      else
	break;
    }

  if (sw_if_index != ~0)
    {
      p = hash_get (dcm->client_by_sw_if_index, sw_if_index);
      if (p == 0)
	return clib_error_return (0, "dhcp client not configured");
      c = pool_elt_at_index (dcm->clients, p[0]);
      vlib_cli_output (vm, "%U", format_dhcp_client, dcm, c, verbose);
      return 0;
    }

  /* *INDENT-OFF* */
  pool_foreach (c, dcm->clients,
  ({
    vlib_cli_output (vm, "%U",
                     format_dhcp_client, dcm,
                     c, verbose);
  }));
  /* *INDENT-ON* */

  return 0;
}

/* *INDENT-OFF* */
VLIB_CLI_COMMAND (show_dhcp_client_command, static) = {
  .path = "show dhcp client",
  .short_help = "show dhcp client [intfc <intfc>][verbose]",
  .function = show_dhcp_client_command_fn,
};
/* *INDENT-ON* */


int
dhcp_client_add_del (dhcp_client_add_del_args_t * a)
{
  dhcp_client_main_t *dcm = &dhcp_client_main;
  vlib_main_t *vm = dcm->vlib_main;
  dhcp_client_t *c;
  uword *p;
  fib_prefix_t all_0s = {
    .fp_len = 0,
    .fp_addr.ip4.as_u32 = 0x0,
    .fp_proto = FIB_PROTOCOL_IP4,
  };

  p = hash_get (dcm->client_by_sw_if_index, a->sw_if_index);

  if ((p && a->is_add) || (!p && a->is_add == 0))
    return VNET_API_ERROR_INVALID_VALUE;

  if (a->is_add)
    {
      pool_get (dcm->clients, c);
      memset (c, 0, sizeof (*c));
      c->state = DHCP_DISCOVER;
      c->sw_if_index = a->sw_if_index;
      c->client_index = a->client_index;
      c->pid = a->pid;
      c->event_callback = a->event_callback;
      c->option_55_data = a->option_55_data;
      c->hostname = a->hostname;
      c->client_identifier = a->client_identifier;
      c->set_broadcast_flag = a->set_broadcast_flag;
      do
	{
	  c->transaction_id = random_u32 (&dcm->seed);
	}
      while (c->transaction_id == 0);
      set_l2_rewrite (dcm, c);
      hash_set (dcm->client_by_sw_if_index, a->sw_if_index, c - dcm->clients);

      /*
       * In order to accept any OFFER, whether broadcasted or unicasted, we
       * need to configure the dhcp-client-detect feature as an input feature
       * so the DHCP OFFER is sent to the ip4-local node. Without this a
       * broadcasted OFFER hits the 255.255.255.255/32 address and a unicast
       * hits 0.0.0.0/0 both of which default to drop and the latter may forward
       * of box - not what we want. Nor to we want to change these route for
       * all interfaces in this table
       */
      vnet_feature_enable_disable ("ip4-unicast",
				   "ip4-dhcp-client-detect",
				   c->sw_if_index, 1 /* enable */ , 0, 0);
      c->client_detect_feature_enabled = 1;

      vlib_process_signal_event (vm, dhcp_client_process_node.index,
				 EVENT_DHCP_CLIENT_WAKEUP, c - dcm->clients);
    }
  else
    {
      c = pool_elt_at_index (dcm->clients, p[0]);

      if (c->router_address.as_u32)
	{
	  ip46_address_t nh = {
	    .ip4 = c->router_address,
	  };

	  fib_table_entry_path_remove (fib_table_get_index_for_sw_if_index
				       (FIB_PROTOCOL_IP4, c->sw_if_index),
				       &all_0s, FIB_SOURCE_DHCP,
				       DPO_PROTO_IP4, &nh, c->sw_if_index, ~0,
				       1, FIB_ROUTE_PATH_FLAG_NONE);
	}
      dhcp_client_release_address (dcm, c);

      vec_free (c->option_55_data);
      vec_free (c->hostname);
      vec_free (c->client_identifier);
      vec_free (c->l2_rewrite);
      hash_unset (dcm->client_by_sw_if_index, c->sw_if_index);
      pool_put (dcm->clients, c);
    }
  return 0;
}

int
dhcp_client_config (u32 is_add,
		    u32 client_index,
		    vlib_main_t * vm,
		    u32 sw_if_index,
		    u8 * hostname,
		    u8 * client_id,
		    dhcp_event_cb_t event_callback,
		    u8 set_broadcast_flag, u32 pid)
{
  dhcp_client_add_del_args_t _a, *a = &_a;
  int rv;

  memset (a, 0, sizeof (*a));
  a->is_add = is_add;
  a->sw_if_index = sw_if_index;
  a->client_index = client_index;
  a->pid = pid;
  a->event_callback = event_callback;
  a->set_broadcast_flag = set_broadcast_flag;
  vec_validate (a->hostname, strlen ((char *) hostname) - 1);
  strncpy ((char *) a->hostname, (char *) hostname, vec_len (a->hostname));
  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.
   */

  /* Subnet Mask */
  vec_add1 (a->option_55_data, 1);
  /* Broadcast address */
  vec_add1 (a->option_55_data, 28);
  /* time offset */
  vec_add1 (a->option_55_data, 2);
  /* Router */
  vec_add1 (a->option_55_data, 3);
  /* Domain Name */
  vec_add1 (a->option_55_data, 15);
  /* DNS */
  vec_add1 (a->option_55_data, 6);
  /* Domain search */
  vec_add1 (a->option_55_data, 119);
  /* Host name */
  vec_add1 (a->option_55_data, 12);
  /* NetBIOS name server */
  vec_add1 (a->option_55_data, 44);
  /* NetBIOS Scope */
  vec_add1 (a->option_55_data, 47);
  /* MTU */
  vec_add1 (a->option_55_data, 26);
  /* Classless static route */
  vec_add1 (a->option_55_data, 121);
  /* NTP servers */
  vec_add1 (a->option_55_data, 42);

  rv = dhcp_client_add_del (a);

  switch (rv)
    {
    case 0:
      break;

    case VNET_API_ERROR_INVALID_VALUE:

      vec_free (a->hostname);
      vec_free (a->client_identifier);
      vec_free (a->option_55_data);

      if (is_add)
	clib_warning ("dhcp client already enabled on intf_idx %d",
		      sw_if_index);
      else
	clib_warning ("dhcp client not enabled on on intf_idx %d",
		      sw_if_index);
      break;

    default:
      clib_warning ("dhcp_client_add_del returned %d", rv);
    }

  return rv;
}

void
dhcp_client_walk (dhcp_client_walk_cb_t cb, void *ctx)
{
  dhcp_client_main_t *dcm = &dhcp_client_main;
  dhcp_client_t *c;

  /* *INDENT-OFF* */
  pool_foreach (c, dcm->clients,
  ({
    if (!cb(c, ctx))
      break;
  }));
  /* *INDENT-ON* */

}

static clib_error_t *
dhcp_client_set_command_fn (vlib_main_t * vm,
			    unformat_input_t * input,
			    vlib_cli_command_t * cmd)
{

  dhcp_client_main_t *dcm = &dhcp_client_main;
  u32 sw_if_index;
  u8 *hostname = 0;
  u8 sw_if_index_set = 0;
  u8 set_broadcast_flag = 1;
  int is_add = 1;
  dhcp_client_add_del_args_t _a, *a = &_a;
  int rv;

  while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT)
    {
      if (unformat (input, "intfc %U",
		    unformat_vnet_sw_interface, dcm->vnet_main, &sw_if_index))
	sw_if_index_set = 1;
      else if (unformat (input, "hostname %v", &hostname))
	;
      else if (unformat (input, "del"))
	is_add = 0;
      else if (unformat (input, "broadcast", &set_broadcast_flag))
	is_add = 0;
      else
	break;
    }

  if (sw_if_index_set == 0)
    return clib_error_return (0, "interface not specified");

  memset (a, 0, sizeof (*a));
  a->is_add = is_add;
  a->sw_if_index = sw_if_index;
  a->hostname = hostname;
  a->client_identifier = format (0, "vpe 1.0%c", 0);
  a->set_broadcast_flag = set_broadcast_flag;

  /*
   * Option 55 request list. These data precisely match
   * the Ubuntu dhcp client. YMMV.
   */

  /* Subnet Mask */
  vec_add1 (a->option_55_data, 1);
  /* Broadcast address */
  vec_add1 (a->option_55_data, 28);
  /* time offset */
  vec_add1 (a->option_55_data, 2);
  /* Router */
  vec_add1 (a->option_55_data, 3);
  /* Domain Name */
  vec_add1 (a->option_55_data, 15);
  /* DNS */
  vec_add1 (a->option_55_data, 6);
  /* Domain search */
  vec_add1 (a->option_55_data, 119);
  /* Host name */
  vec_add1 (a->option_55_data, 12);
  /* NetBIOS name server */
  vec_add1 (a->option_55_data, 44);
  /* NetBIOS Scope */
  vec_add1 (a->option_55_data, 47);
  /* MTU */
  vec_add1 (a->option_55_data, 26);
  /* Classless static route */
  vec_add1 (a->option_55_data, 121);
  /* NTP servers */
  vec_add1 (a->option_55_data, 42);

  rv = dhcp_client_add_del (a);

  switch (rv)
    {
    case 0:
      break;

    case VNET_API_ERROR_INVALID_VALUE:

      vec_free (a->hostname);
      vec_free (a->client_identifier);
      vec_free (a->option_55_data);
      if (is_add)
	return clib_error_return (0, "dhcp client already enabled on %U",
				  format_vnet_sw_if_index_name,
				  dcm->vnet_main, sw_if_index);
      else
	return clib_error_return (0, "dhcp client not enabled on %U",
				  format_vnet_sw_if_index_name,
				  dcm->vnet_main, sw_if_index);
      break;

    default:
      vlib_cli_output (vm, "dhcp_client_add_del returned %d", rv);
    }

  return 0;
}

/* *INDENT-OFF* */
VLIB_CLI_COMMAND (dhcp_client_set_command, static) = {
  .path = "set dhcp client",
  .short_help = "set dhcp client [del] intfc <interface> [hostname <name>]",
  .function = dhcp_client_set_command_fn,
};
/* *INDENT-ON* */

static clib_error_t *
dhcp_client_init (vlib_main_t * vm)
{
  dhcp_client_main_t *dcm = &dhcp_client_main;

  dcm->vlib_main = vm;
  dcm->vnet_main = vnet_get_main ();
  dcm->seed = 0xdeaddabe;
  return 0;
}

VLIB_INIT_FUNCTION (dhcp_client_init);

/*
 * fd.io coding-style-patch-verification: ON
 *
 * Local Variables:
 * eval: (c-set-style "gnu")
 * End:
 */