src/svm/svm.c

/*
 * map.c : MAP support
 *
 * 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/fib/fib_table.h>
#include <vnet/fib/ip6_fib.h>
#include <vnet/adj/adj.h>
#include <map/map_dpo.h>
#include <vppinfra/crc32.h>
#include <vnet/plugin/plugin.h>
#include <vpp/app/version.h>
#include "map.h"

map_main_t map_main;

/*
 * This code supports the following MAP modes:
 *
 * Algorithmic Shared IPv4 address (ea_bits_len > 0):
 *   ea_bits_len + ip4_prefix > 32
 *   psid_length > 0, ip6_prefix < 64, ip4_prefix <= 32
 * Algorithmic Full IPv4 address (ea_bits_len > 0):
 *   ea_bits_len + ip4_prefix = 32
 *   psid_length = 0, ip6_prefix < 64, ip4_prefix <= 32
 * Algorithmic IPv4 prefix (ea_bits_len > 0):
 *   ea_bits_len + ip4_prefix < 32
 *   psid_length = 0, ip6_prefix < 64, ip4_prefix <= 32
 *
 * Independent Shared IPv4 address (ea_bits_len = 0):
 *   ip4_prefix = 32
 *   psid_length > 0
 *   Rule IPv6 address = 128, Rule PSID Set
 * Independent Full IPv4 address (ea_bits_len = 0):
 *   ip4_prefix = 32
 *   psid_length = 0, ip6_prefix = 128
 * Independent IPv4 prefix (ea_bits_len = 0):
 *   ip4_prefix < 32
 *   psid_length = 0, ip6_prefix = 128
 *
 */

/*
 * This code supports MAP-T:
 *
 * With a DMR prefix length of 64 or 96 (RFC6052).
 *
 */


int
map_create_domain (ip4_address_t * ip4_prefix,
		   u8 ip4_prefix_len,
		   ip6_address_t * ip6_prefix,
		   u8 ip6_prefix_len,
		   ip6_address_t * ip6_src,
		   u8 ip6_src_len,
		   u8 ea_bits_len,
		   u8 psid_offset,
		   u8 psid_length, u32 * map_domain_index, u16 mtu, u8 flags)
{
  u8 suffix_len, suffix_shift;
  map_main_t *mm = &map_main;
  dpo_id_t dpo_v4 = DPO_INVALID;
  dpo_id_t dpo_v6 = DPO_INVALID;
  map_domain_t *d;

  /* Sanity check on the src prefix length */
  if (flags & MAP_DOMAIN_TRANSLATION)
    {
      if (ip6_src_len != 96 && ip6_src_len != 64)
	{
	  clib_warning ("MAP-T only supports prefix lengths of 64 and 96.");
	  return -1;
	}
    }
  else
    {
      if (ip6_src_len != 128)
	{
	  clib_warning
	    ("MAP-E requires a BR address, not a prefix (ip6_src_len should "
	     "be 128).");
	  return -1;
	}
    }

  /* How many, and which bits to grab from the IPv4 DA */
  if (ip4_prefix_len + ea_bits_len < 32)
    {
      if (!(flags & MAP_DOMAIN_TRANSLATION))
	flags |= MAP_DOMAIN_PREFIX;
      suffix_shift = 32 - ip4_prefix_len - ea_bits_len;
      suffix_len = ea_bits_len;
    }
  else
    {
      suffix_shift = 0;
      suffix_len = 32 - ip4_prefix_len;
    }

  /* EA bits must be within the first 64 bits */
  if (ea_bits_len > 0 && ((ip6_prefix_len + ea_bits_len) > 64 ||
			  ip6_prefix_len + suffix_len + psid_length > 64))
    {
      clib_warning
	("Embedded Address bits must be within the first 64 bits of "
	 "the IPv6 prefix");
      return -1;
    }

  if (mm->is_ce && !(flags & MAP_DOMAIN_TRANSLATION))
    {
      clib_warning ("MAP-E CE is not supported yet");
      return -1;
    }

  /* Get domain index */
  pool_get_aligned (mm->domains, d, CLIB_CACHE_LINE_BYTES);
  memset (d, 0, sizeof (*d));
  *map_domain_index = d - mm->domains;

  /* Init domain struct */
  d->ip4_prefix.as_u32 = ip4_prefix->as_u32;
  d->ip4_prefix_len = ip4_prefix_len;
  d->ip6_prefix = *ip6_prefix;
  d->ip6_prefix_len = ip6_prefix_len;
  d->ip6_src = *ip6_src;
  d->ip6_src_len = ip6_src_len;
  d->ea_bits_len = ea_bits_len;
  d->psid_offset = psid_offset;
  d->psid_length = psid_length;
  d->mtu = mtu;
  d->flags = flags;
  d->suffix_shift = suffix_shift;
  d->suffix_mask = (1 << suffix_len) - 1;

  d->psid_shift = 16 - psid_length - psid_offset;
  d->psid_mask = (1 << d->psid_length) - 1;
  d->ea_shift = 64 - ip6_prefix_len - suffix_len - d->psid_length;

  /* MAP data-plane object */
  if (d->flags & MAP_DOMAIN_TRANSLATION)
    map_t_dpo_create (DPO_PROTO_IP4, *map_domain_index, &dpo_v4);
  else
    map_dpo_create (DPO_PROTO_IP4, *map_domain_index, &dpo_v4);

  /* Create ip4 route */
  u8 ip4_pfx_len;
  ip4_address_t ip4_pfx;
  if (mm->is_ce)
    {
      ip4_pfx_len = 0;
      ip4_pfx.as_u32 = 0;
    }
  else
    {
      ip4_pfx_len = d->ip4_prefix_len;
      ip4_pfx = d->ip4_prefix;
    }
  fib_prefix_t pfx = {
    .fp_proto = FIB_PROTOCOL_IP4,
    .fp_len = ip4_pfx_len,
    .fp_addr = {
		.ip4 = ip4_pfx,
		}
    ,
  };
  fib_table_entry_special_dpo_add (0, &pfx,
				   FIB_SOURCE_MAP,
				   FIB_ENTRY_FLAG_EXCLUSIVE, &dpo_v4);
  dpo_reset (&dpo_v4);

  /*
   * construct a DPO to use the v6 domain
   */
  if (d->flags & MAP_DOMAIN_TRANSLATION)
    map_t_dpo_create (DPO_PROTO_IP6, *map_domain_index, &dpo_v6);
  else
    map_dpo_create (DPO_PROTO_IP6, *map_domain_index, &dpo_v6);

  /*
   * Multiple MAP domains may share same source IPv6 TEP. Which is just dandy.
   * We are not tracking the sharing. So a v4 lookup to find the correct
   * domain post decap/trnaslate is always done
   *
   * Create ip6 route. This is a reference counted add. If the prefix
   * already exists and is MAP sourced, it is now MAP source n+1 times
   * and will need to be removed n+1 times.
   */
  u8 ip6_pfx_len;
  ip6_address_t ip6_pfx;
  if (mm->is_ce)
    {
      ip6_pfx_len = d->ip6_prefix_len;
      ip6_pfx = d->ip6_prefix;
    }
  else
    {
      ip6_pfx_len = d->ip6_src_len;
      ip6_pfx = d->ip6_src;
    }
  fib_prefix_t pfx6 = {
    .fp_proto = FIB_PROTOCOL_IP6,
    .fp_len = ip6_pfx_len,
    .fp_addr.ip6 = ip6_pfx,
  };

  fib_table_entry_special_dpo_add (0, &pfx6,
				   FIB_SOURCE_MAP,
				   FIB_ENTRY_FLAG_EXCLUSIVE, &dpo_v6);
  dpo_reset (&dpo_v6);

  /* Validate packet/byte counters */
  map_domain_counter_lock (mm);
  int i;
  for (i = 0; i < vec_len (mm->simple_domain_counters); i++)
    {
      vlib_validate_simple_counter (&mm->simple_domain_counters[i],
				    *map_domain_index);
      vlib_zero_simple_counter (&mm->simple_domain_counters[i],
				*map_domain_index);
    }
  for (i = 0; i < vec_len (mm->domain_counters); i++)
    {
      vlib_validate_combined_counter (&mm->domain_counters[i],
				      *map_domain_index);
      vlib_zero_combined_counter (&mm->domain_counters[i], *map_domain_index);
    }
  map_domain_counter_unlock (mm);

  return 0;
}

/*
 * map_delete_domain
 */
int
map_delete_domain (u32 map_domain_index)
{
  map_main_t *mm = &map_main;
  map_domain_t *d;

  if (pool_is_free_index (mm->domains, map_domain_index))
    {
      clib_warning ("MAP domain delete: domain does not exist: %d",
		    map_domain_index);
      return -1;
    }

  d = pool_elt_at_index (mm->domains, map_domain_index);

  fib_prefix_t pfx = {
    .fp_proto = FIB_PROTOCOL_IP4,
    .fp_len = d->ip4_prefix_len,
    .fp_addr = {
		.ip4 = d->ip4_prefix,
		}
    ,
  };
  fib_table_entry_special_remove (0, &pfx, FIB_SOURCE_MAP);

  fib_prefix_t pfx6 = {
    .fp_proto = FIB_PROTOCOL_IP6,
    .fp_len = d->ip6_src_len,
    .fp_addr = {
		.ip6 = d->ip6_src,
		}
    ,
  };
  fib_table_entry_special_remove (0, &pfx6, FIB_SOURCE_MAP);

  /* Deleting rules */
  if (d->rules)
    clib_mem_free (d->rules);

  pool_put (mm->domains, d);

  return 0;
}

int
map_add_del_psid (u32 map_domain_index, u16 psid, ip6_address_t * tep,
		  u8 is_add)
{
  map_domain_t *d;
  map_main_t *mm = &map_main;

  if (pool_is_free_index (mm->domains, map_domain_index))
    {
      clib_warning ("MAP rule: domain does not exist: %d", map_domain_index);
      return -1;
    }
  d = pool_elt_at_index (mm->domains, map_domain_index);

  /* Rules are only used in 1:1 independent case */
  if (d->ea_bits_len > 0)
    return (-1);

  if (!d->rules)
    {
      u32 l = (0x1 << d->psid_length) * sizeof (ip6_address_t);
      d->rules = clib_mem_alloc_aligned (l, CLIB_CACHE_LINE_BYTES);
      if (!d->rules)
	return -1;
      memset (d->rules, 0, l);
    }

  if (psid >= (0x1 << d->psid_length))
    {
      clib_warning ("MAP rule: PSID outside bounds: %d [%d]", psid,
		    0x1 << d->psid_length);
      return -1;
    }

  if (is_add)
    {
      d->rules[psid] = *tep;
    }
  else
    {
      memset (&d->rules[psid], 0, sizeof (ip6_address_t));
    }
  return 0;
}

#ifdef MAP_SKIP_IP6_LOOKUP
/**
 * Pre-resolvd per-protocol global next-hops
 */
map_main_pre_resolved_t pre_resolved[FIB_PROTOCOL_MAX];

static void
map_pre_resolve_init (map_main_pre_resolved_t * pr)
{
  pr->fei = FIB_NODE_INDEX_INVALID;
  fib_node_init (&pr->node, FIB_NODE_TYPE_MAP_E);
}

static u8 *
format_map_pre_resolve (u8 * s, va_list * ap)
{
  map_main_pre_resolved_t *pr = va_arg (*ap, map_main_pre_resolved_t *);

  if (FIB_NODE_INDEX_INVALID != pr->fei)
    {
      const fib_prefix_t *pfx;

      pfx = fib_entry_get_prefix (pr->fei);

      return (format (s, "%U (%u)",
		      format_ip46_address, &pfx->fp_addr, IP46_TYPE_ANY,
		      pr->dpo.dpoi_index));
    }
  else
    {
      return (format (s, "un-set"));
    }
}


/**
 * Function definition to inform the FIB node that its last lock has gone.
 */
static void
map_last_lock_gone (fib_node_t * node)
{
  /*
   * The MAP is a root of the graph. As such
   * it never has children and thus is never locked.
   */
  ASSERT (0);
}

static map_main_pre_resolved_t *
map_from_fib_node (fib_node_t * node)
{
  ASSERT (FIB_NODE_TYPE_MAP_E == node->fn_type);
  return ((map_main_pre_resolved_t *)
	  (((char *) node) -
	   STRUCT_OFFSET_OF (map_main_pre_resolved_t, node)));
}

static void
map_stack (map_main_pre_resolved_t * pr)
{
  const dpo_id_t *dpo;

  dpo = fib_entry_contribute_ip_forwarding (pr->fei);

  dpo_copy (&pr->dpo, dpo);
}

/**
 * Function definition to backwalk a FIB node
 */
static fib_node_back_walk_rc_t
map_back_walk (fib_node_t * node, fib_node_back_walk_ctx_t * ctx)
{
  map_stack (map_from_fib_node (node));

  return (FIB_NODE_BACK_WALK_CONTINUE);
}

/**
 * Function definition to get a FIB node from its index
 */
static fib_node_t *
map_fib_node_get (fib_node_index_t index)
{
  return (&pre_resolved[index].node);
}

/*
 * Virtual function table registered by MPLS GRE tunnels
 * for participation in the FIB object graph.
 */
const static fib_node_vft_t map_vft = {
  .fnv_get = map_fib_node_get,
  .fnv_last_lock = map_last_lock_gone,
  .fnv_back_walk = map_back_walk,
};

static void
map_fib_resolve (map_main_pre_resolved_t * pr,
		 fib_protocol_t proto, u8 len, const ip46_address_t * addr)
{
  fib_prefix_t pfx = {
    .fp_proto = proto,
    .fp_len = len,
    .fp_addr = *addr,
  };

  pr->fei = fib_table_entry_special_add (0,	// default fib
					 &pfx,
					 FIB_SOURCE_RR, FIB_ENTRY_FLAG_NONE);
  pr->sibling = fib_entry_child_add (pr->fei, FIB_NODE_TYPE_MAP_E, proto);
  map_stack (pr);
}

static void
map_fib_unresolve (map_main_pre_resolved_t * pr,
		   fib_protocol_t proto, u8 len, const ip46_address_t * addr)
{
  fib_prefix_t pfx = {
    .fp_proto = proto,
    .fp_len = len,
    .fp_addr = *addr,
  };

  fib_entry_child_remove (pr->fei, pr->sibling);

  fib_table_entry_special_remove (0,	// default fib
				  &pfx, FIB_SOURCE_RR);
  dpo_reset (&pr->dpo);

  pr->fei = FIB_NODE_INDEX_INVALID;
  pr->sibling = FIB_NODE_INDEX_INVALID;
}

static void
map_pre_resolve (ip4_address_t * ip4, ip6_address_t * ip6, int is_del)
{
  if (ip6 && (ip6->as_u64[0] != 0 || ip6->as_u64[1] != 0))
    {
      ip46_address_t addr = {
	.ip6 = *ip6,
      };
      if (is_del)
	map_fib_unresolve (&pre_resolved[FIB_PROTOCOL_IP6],
			   FIB_PROTOCOL_IP6, 128, &addr);
      else
	map_fib_resolve (&pre_resolved[FIB_PROTOCOL_IP6],
			 FIB_PROTOCOL_IP6, 128, &addr);
    }
  if (ip4 && (ip4->as_u32 != 0))
    {
      ip46_address_t addr = {
	.ip4 = *ip4,
      };
      if (is_del)
	map_fib_unresolve (&pre_resolved[FIB_PROTOCOL_IP4],
			   FIB_PROTOCOL_IP4, 32, &addr);
      else
	map_fib_resolve (&pre_resolved[FIB_PROTOCOL_IP4],
			 FIB_PROTOCOL_IP4, 32, &addr);
    }
}
#endif

static clib_error_t *
map_security_check_command_fn (vlib_main_t * vm,
			       unformat_input_t * input,
			       vlib_cli_command_t * cmd)
{
  unformat_input_t _line_input, *line_input = &_line_input;
  map_main_t *mm = &map_main;
  clib_error_t *error = NULL;

  /* Get a line of input. */
  if (!unformat_user (input, unformat_line_input, line_input))
    return 0;

  while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
    {
      if (unformat (line_input, "off"))
	mm->sec_check = false;
      else if (unformat (line_input, "on"))
	mm->sec_check = true;
      else
	{
	  error = clib_error_return (0, "unknown input `%U'",
				     format_unformat_error, line_input);
	  goto done;
	}
    }

done:
  unformat_free (line_input);

  return error;
}

static clib_error_t *
map_security_check_frag_command_fn (vlib_main_t * vm,
				    unformat_input_t * input,
				    vlib_cli_command_t * cmd)
{
  unformat_input_t _line_input, *line_input = &_line_input;
  map_main_t *mm = &map_main;
  clib_error_t *error = NULL;

  /* Get a line of input. */
  if (!unformat_user (input, unformat_line_input, line_input))
    return 0;

  while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
    {
      if (unformat (line_input, "off"))
	mm->sec_check_frag = false;
      else if (unformat (line_input, "on"))
	mm->sec_check_frag = true;
      else
	{
	  error = clib_error_return (0, "unknown input `%U'",
				     format_unformat_error, line_input);
	  goto done;
	}
    }

done:
  unformat_free (line_input);

  return error;
}

static clib_error_t *
map_add_domain_command_fn (vlib_main_t * vm,
			   unformat_input_t * input, vlib_cli_command_t * cmd)
{
  unformat_input_t _line_input, *line_input = &_line_input;
  ip4_address_t ip4_prefix;
  ip6_address_t ip6_prefix;
  ip6_address_t ip6_src;
  u32 ip6_prefix_len = 0, ip4_prefix_len = 0, map_domain_index, ip6_src_len;
  u32 num_m_args = 0;
  /* Optional arguments */
  u32 ea_bits_len = 0, psid_offset = 0, psid_length = 0;
  u32 mtu = 0;
  u8 flags = 0;
  ip6_src_len = 128;
  clib_error_t *error = NULL;

  /* Get a line of input. */
  if (!unformat_user (input, unformat_line_input, line_input))
    return 0;

  while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
    {
      if (unformat
	  (line_input, "ip4-pfx %U/%d", unformat_ip4_address, &ip4_prefix,
	   &ip4_prefix_len))
	num_m_args++;
      else
	if (unformat
	    (line_input, "ip6-pfx %U/%d", unformat_ip6_address, &ip6_prefix,
	     &ip6_prefix_len))
	num_m_args++;
      else
	if (unformat
	    (line_input, "ip6-src %U/%d", unformat_ip6_address, &ip6_src,
	     &ip6_src_len))
	num_m_args++;
      else
	if (unformat
	    (line_input, "ip6-src %U", unformat_ip6_address, &ip6_src))
	num_m_args++;
      else if (unformat (line_input, "ea-bits-len %d", &ea_bits_len))
	num_m_args++;
      else if (unformat (line_input, "psid-offset %d", &psid_offset))
	num_m_args++;
      else if (unformat (line_input, "psid-len %d", &psid_length))
	num_m_args++;
      else if (unformat (line_input, "mtu %d", &mtu))
	num_m_args++;
      else if (unformat (line_input, "map-t"))
	flags |= MAP_DOMAIN_TRANSLATION;
      else if (unformat (line_input, "rfc6052"))
	flags |= (MAP_DOMAIN_TRANSLATION | MAP_DOMAIN_RFC6052);
      else
	{
	  error = clib_error_return (0, "unknown input `%U'",
				     format_unformat_error, line_input);
	  goto done;
	}
    }

  if (num_m_args < 3)
    {
      error = clib_error_return (0, "mandatory argument(s) missing");
      goto done;
    }

  map_create_domain (&ip4_prefix, ip4_prefix_len,
		     &ip6_prefix, ip6_prefix_len, &ip6_src, ip6_src_len,
		     ea_bits_len, psid_offset, psid_length, &map_domain_index,
		     mtu, flags);

done:
  unformat_free (line_input);

  return error;
}

static clib_error_t *
map_del_domain_command_fn (vlib_main_t * vm,
			   unformat_input_t * input, vlib_cli_command_t * cmd)
{
  unformat_input_t _line_input, *line_input = &_line_input;
  u32 num_m_args = 0;
  u32 map_domain_index;
  clib_error_t *error = NULL;

  /* Get a line of input. */
  if (!unformat_user (input, unformat_line_input, line_input))
    return 0;

  while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
    {
      if (unformat (line_input, "index %d", &map_domain_index))
	num_m_args++;
      else
	{
	  error = clib_error_return (0, "unknown input `%U'",
				     format_unformat_error, line_input);
	  goto done;
	}
    }

  if (num_m_args != 1)
    {
      error = clib_error_return (0, "mandatory argument(s) missing");
      goto done;
    }

  map_delete_domain (map_domain_index);

done:
  unformat_free (line_input);

  return error;
}

static clib_error_t *
map_add_rule_command_fn (vlib_main_t * vm,
			 unformat_input_t * input, vlib_cli_command_t * cmd)
{
  unformat_input_t _line_input, *line_input = &_line_input;
  ip6_address_t tep;
  u32 num_m_args = 0;
  u32 psid = 0, map_domain_index;
  clib_error_t *error = NULL;

  /* Get a line of input. */
  if (!unformat_user (input, unformat_line_input, line_input))
    return 0;

  while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
    {
      if (unformat (line_input, "index %d", &map_domain_index))
	num_m_args++;
      else if (unformat (line_input, "psid %d", &psid))
	num_m_args++;
      else
	if (unformat (line_input, "ip6-dst %U", unformat_ip6_address, &tep))
	num_m_args++;
      else
	{
	  error = clib_error_return (0, "unknown input `%U'",
				     format_unformat_error, line_input);
	  goto done;
	}
    }

  if (num_m_args != 3)
    {
      error = clib_error_return (0, "mandatory argument(s) missing");
      goto done;
    }

  if (map_add_del_psid (map_domain_index, psid, &tep, 1) != 0)
    {
      error = clib_error_return (0, "Failing to add Mapping Rule");
      goto done;
    }

done:
  unformat_free (line_input);

  return error;
}

#if MAP_SKIP_IP6_LOOKUP
static clib_error_t *
map_pre_resolve_command_fn (vlib_main_t * vm,
			    unformat_input_t * input,
			    vlib_cli_command_t * cmd)
{
  unformat_input_t _line_input, *line_input = &_line_input;
  ip4_address_t ip4nh, *p_v4 = NULL;
  ip6_address_t ip6nh, *p_v6 = NULL;
  clib_error_t *error = NULL;
  int is_del = 0;

  memset (&ip4nh, 0, sizeof (ip4nh));
  memset (&ip6nh, 0, sizeof (ip6nh));

  /* Get a line of input. */
  if (!unformat_user (input, unformat_line_input, line_input))
    return 0;

  while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
    {
      if (unformat (line_input, "ip4-nh %U", unformat_ip4_address, &ip4nh))
	p_v4 = &ip4nh;
      else
	if (unformat (line_input, "ip6-nh %U", unformat_ip6_address, &ip6nh))
	p_v6 = &ip6nh;
      else if (unformat (line_input, "del"))
	is_del = 1;
      else
	{
	  error = clib_error_return (0, "unknown input `%U'",
				     format_unformat_error, line_input);
	  goto done;
	}
    }

  map_pre_resolve (p_v4, p_v6, is_del);

done:
  unformat_free (line_input);

  return error;
}
#endif

static clib_error_t *
map_icmp_relay_source_address_command_fn (vlib_main_t * vm,
					  unformat_input_t * input,
					  vlib_cli_command_t * cmd)
{
  unformat_input_t _line_input, *line_input = &_line_input;
  ip4_address_t icmp_src_address;
  map_main_t *mm = &map_main;
  clib_error_t *error = NULL;

  mm->icmp4_src_address.as_u32 = 0;

  /* Get a line of input. */
  if (!unformat_user (input, unformat_line_input, line_input))
    return 0;

  while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
    {
      if (unformat
	  (line_input, "%U", unformat_ip4_address, &icmp_src_address))
	mm->icmp4_src_address = icmp_src_address;
      else
	{
	  error = clib_error_return (0, "unknown input `%U'",
				     format_unformat_error, line_input);
	  goto done;
	}
    }

done:
  unformat_free (line_input);

  return error;
}

static clib_error_t *
map_icmp_unreachables_command_fn (vlib_main_t * vm,
				  unformat_input_t * input,
				  vlib_cli_command_t * cmd)
{
  unformat_input_t _line_input, *line_input = &_line_input;
  map_main_t *mm = &map_main;
  int num_m_args = 0;
  clib_error_t *error = NULL;

  /* Get a line of input. */
  if (!unformat_user (input, unformat_line_input, line_input))
    return 0;

  while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
    {
      num_m_args++;
      if (unformat (line_input, "on"))
	mm->icmp6_enabled = true;
      else if (unformat (line_input, "off"))
	mm->icmp6_enabled = false;
      else
	{
	  error = clib_error_return (0, "unknown input `%U'",
				     format_unformat_error, line_input);
	  goto done;
	}
    }


  if (num_m_args != 1)
    error = clib_error_return (0, "mandatory argument(s) missing");

done:
  unformat_free (line_input);

  return error;
}

static clib_error_t *
map_fragment_command_fn (vlib_main_t * vm,
			 unformat_input_t * input, vlib_cli_command_t * cmd)
{
  unformat_input_t _line_input, *line_input = &_line_input;
  map_main_t *mm = &map_main;
  clib_error_t *error = NULL;

  /* Get a line of input. */
  if (!unformat_user (input, unformat_line_input, line_input))
    return 0;

  while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
    {
      if (unformat (line_input, "inner"))
	mm->frag_inner = true;
      else if (unformat (line_input, "outer"))
	mm->frag_inner = false;
      else
	{
	  error = clib_error_return (0, "unknown input `%U'",
				     format_unformat_error, line_input);
	  goto done;
	}
    }

done:
  unformat_free (line_input);

  return error;
}

static clib_error_t *
map_fragment_df_command_fn (vlib_main_t * vm,
			    unformat_input_t * input,
			    vlib_cli_command_t * cmd)
{
  unformat_input_t _line_input, *line_input = &_line_input;
  map_main_t *mm = &map_main;
  clib_error_t *error = NULL;

  /* Get a line of input. */
  if (!unformat_user (input, unformat_line_input, line_input))
    return 0;

  while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
    {
      if (unformat (line_input, "on"))
	mm->frag_ignore_df = true;
      else if (unformat (line_input, "off"))
	mm->frag_ignore_df = false;
      else
	{
	  error = clib_error_return (0, "unknown input `%U'",
				     format_unformat_error, line_input);
	  goto done;
	}
    }

done:
  unformat_free (line_input);

  return error;
}

static clib_error_t *
map_traffic_class_command_fn (vlib_main_t * vm,
			      unformat_input_t * input,
			      vlib_cli_command_t * cmd)
{
  unformat_input_t _line_input, *line_input = &_line_input;
  map_main_t *mm = &map_main;
  u32 tc = 0;
  clib_error_t *error = NULL;

  mm->tc_copy = false;

  /* Get a line of input. */
  if (!unformat_user (input, unformat_line_input, line_input))
    return 0;

  while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
    {
      if (unformat (line_input, "copy"))
	mm->tc_copy = true;
      else if (unformat (line_input, "%x", &tc))
	mm->tc = tc & 0xff;
      else
	{
	  error = clib_error_return (0, "unknown input `%U'",
				     format_unformat_error, line_input);
	  goto done;
	}
    }

done:
  unformat_free (line_input);

  return error;
}

static char *
map_flags_to_string (u32 flags)
{
  if (flags & MAP_DOMAIN_RFC6052)
    return "rfc6052";
  if (flags & MAP_DOMAIN_PREFIX)
    return "prefix";
  if (flags & MAP_DOMAIN_TRANSLATION)
    return "map-t";
  return "";
}

static u8 *
format_map_domain (u8 * s, va_list * args)
{
  map_domain_t *d = va_arg (*args, map_domain_t *);
  bool counters = va_arg (*args, int);
  map_main_t *mm = &map_main;
  ip6_address_t ip6_prefix;

  if (d->rules)
    memset (&ip6_prefix, 0, sizeof (ip6_prefix));
  else
    ip6_prefix = d->ip6_prefix;

  s = format (s,
	      "[%d] ip4-pfx %U/%d ip6-pfx %U/%d ip6-src %U/%d ea-bits-len %d "
	      "psid-offset %d psid-len %d mtu %d %s",
	      d - mm->domains,
	      format_ip4_address, &d->ip4_prefix, d->ip4_prefix_len,
	      format_ip6_address, &ip6_prefix, d->ip6_prefix_len,
	      format_ip6_address, &d->ip6_src, d->ip6_src_len,
	      d->ea_bits_len, d->psid_offset, d->psid_length, d->mtu,
	      map_flags_to_string (d->flags));

  if (counters)
    {
      map_domain_counter_lock (mm);
      vlib_counter_t v;
      vlib_get_combined_counter (&mm->domain_counters[MAP_DOMAIN_COUNTER_TX],
				 d - mm->domains, &v);
      s = format (s, "  TX: %lld/%lld", v.packets, v.bytes);
      vlib_get_combined_counter (&mm->domain_counters[MAP_DOMAIN_COUNTER_RX],
				 d - mm->domains, &v);
      s = format (s, "  RX: %lld/%lld", v.packets, v.bytes);
      map_domain_counter_unlock (mm);
    }
  s = format (s, "\n");

  if (d->rules)
    {
      int i;
      ip6_address_t dst;
      for (i = 0; i < (0x1 << d->psid_length); i++)
	{
	  dst = d->rules[i];
	  if (dst.as_u64[0] == 0 && dst.as_u64[1] == 0)
	    continue;
	  s = format (s,
		      " rule psid: %d ip6-dst %U\n", i, format_ip6_address,
		      &dst);
	}
    }
  return s;
}

static u8 *
format_map_ip4_reass (u8 * s, va_list * args)
{
  map_main_t *mm = &map_main;
  map_ip4_reass_t *r = va_arg (*args, map_ip4_reass_t *);
  map_ip4_reass_key_t *k = &r->key;
  f64 now = vlib_time_now (mm->vlib_main);
  f64 lifetime = (((f64) mm->ip4_reass_conf_lifetime_ms) / 1000);
  f64 dt = (r->ts + lifetime > now) ? (r->ts + lifetime - now) : -1;
  s = format (s,
	      "ip4-reass src=%U  dst=%U  protocol=%d  identifier=%d  port=%d  lifetime=%.3lf\n",
	      format_ip4_address, &k->src.as_u8, format_ip4_address,
	      &k->dst.as_u8, k->protocol,
	      clib_net_to_host_u16 (k->fragment_id),
	      (r->port >= 0) ? clib_net_to_host_u16 (r->port) : -1, dt);
  return s;
}

static u8 *
format_map_ip6_reass (u8 * s, va_list * args)
{
  map_main_t *mm = &map_main;
  map_ip6_reass_t *r = va_arg (*args, map_ip6_reass_t *);
  map_ip6_reass_key_t *k = &r->key;
  f64 now = vlib_time_now (mm->vlib_main);
  f64 lifetime = (((f64) mm->ip6_reass_conf_lifetime_ms) / 1000);
  f64 dt = (r->ts + lifetime > now) ? (r->ts + lifetime - now) : -1;
  s = format (s,
	      "ip6-reass src=%U  dst=%U  protocol=%d  identifier=%d  lifetime=%.3lf\n",
	      format_ip6_address, &k->src.as_u8, format_ip6_address,
	      &k->dst.as_u8, k->protocol,
	      clib_net_to_host_u32 (k->fragment_id), dt);
  return s;
}

static clib_error_t *
show_map_domain_command_fn (vlib_main_t * vm, unformat_input_t * input,
			    vlib_cli_command_t * cmd)
{
  unformat_input_t _line_input, *line_input = &_line_input;
  map_main_t *mm = &map_main;
  map_domain_t *d;
  bool counters = false;
  u32 map_domain_index = ~0;
  clib_error_t *error = NULL;

  /* Get a line of input. */
  if (!unformat_user (input, unformat_line_input, line_input))
    return 0;

  while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
    {
      if (unformat (line_input, "counters"))
	counters = true;
      else if (unformat (line_input, "index %d", &map_domain_index))
	;
      else
	{
	  error = clib_error_return (0, "unknown input `%U'",
				     format_unformat_error, line_input);
	  goto done;
	}
    }

  if (pool_elts (mm->domains) == 0)
    vlib_cli_output (vm, "No MAP domains are configured...");

  if (map_domain_index == ~0)
    {
    /* *INDENT-OFF* */
    pool_foreach(d, mm->domains, ({vlib_cli_output(vm, "%U", format_map_domain, d, counters);}));
    /* *INDENT-ON* */
    }
  else
    {
      if (pool_is_free_index (mm->domains, map_domain_index))
	{
	  error = clib_error_return (0, "MAP domain does not exists %d",
				     map_domain_index);
	  goto done;
	}

      d = pool_elt_at_index (mm->domains, map_domain_index);
      vlib_cli_output (vm, "%U", format_map_domain, d, counters);
    }

done:
  unformat_free (line_input);

  return error;
}

static clib_error_t *
show_map_fragments_command_fn (vlib_main_t * vm, unformat_input_t * input,
			       vlib_cli_command_t * cmd)
{
  map_main_t *mm = &map_main;
  map_ip4_reass_t *f4;
  map_ip6_reass_t *f6;

  /* *INDENT-OFF* */
  pool_foreach(f4, mm->ip4_reass_pool, ({vlib_cli_output (vm, "%U", format_map_ip4_reass, f4);}));
  /* *INDENT-ON* */
  /* *INDENT-OFF* */
  pool_foreach(f6, mm->ip6_reass_pool, ({vlib_cli_output (vm, "%U", format_map_ip6_reass, f6);}));
  /* *INDENT-ON* */
  return (0);
}

u64
map_error_counter_get (u32 node_index, map_error_t map_error)
{
  vlib_main_t *vm = vlib_get_main ();
  vlib_node_runtime_t *error_node = vlib_node_get_runtime (vm, node_index);
  vlib_error_main_t *em = &vm->error_main;
  vlib_error_t e = error_node->errors[map_error];
  vlib_node_t *n = vlib_get_node (vm, node_index);
  u32 ci;

  ci = vlib_error_get_code (e);
  ASSERT (ci < n->n_errors);
  ci += n->error_heap_index;

  return (em->counters[ci]);
}

static clib_error_t *
show_map_stats_command_fn (vlib_main_t * vm, unformat_input_t * input,
			   vlib_cli_command_t * cmd)
{
  map_main_t *mm = &map_main;
  map_domain_t *d;
  int domains = 0, rules = 0, domaincount = 0, rulecount = 0;
  if (pool_elts (mm->domains) == 0)
    {
      vlib_cli_output (vm, "No MAP domains are configured...");
      return 0;
    }

  /* *INDENT-OFF* */
  pool_foreach(d, mm->domains, ({
    if (d->rules) {
      rulecount+= 0x1 << d->psid_length;
      rules += sizeof(ip6_address_t) * 0x1 << d->psid_length;
    }
    domains += sizeof(*d);
    domaincount++;
  }));
  /* *INDENT-ON* */

  vlib_cli_output (vm, "MAP domains structure: %d\n", sizeof (map_domain_t));
  vlib_cli_output (vm, "MAP domains: %d (%d bytes)\n", domaincount, domains);
  vlib_cli_output (vm, "MAP rules: %d (%d bytes)\n", rulecount, rules);
  vlib_cli_output (vm, "Total: %d bytes)\n", rules + domains);

#if MAP_SKIP_IP6_LOOKUP
  vlib_cli_output (vm,
		   "MAP pre-resolve: IP6 next-hop: %U, IP4 next-hop: %U\n",
		   format_map_pre_resolve, &pre_resolved[FIB_PROTOCOL_IP6],
		   format_map_pre_resolve, &pre_resolved[FIB_PROTOCOL_IP4]);

#endif

  if (mm->tc_copy)
    vlib_cli_output (vm, "MAP traffic-class: copy");
  else
    vlib_cli_output (vm, "MAP traffic-class: %x", mm->tc);

  vlib_cli_output (vm,
		   "MAP IPv6 inbound security check: %s, fragmented packet security check: %s",
		   mm->sec_check ? "enabled" : "disabled",
		   mm->sec_check_frag ? "enabled" : "disabled");

  vlib_cli_output (vm, "ICMP-relay IPv4 source address: %U\n",
		   format_ip4_address, &mm->icmp4_src_address);
  vlib_cli_output (vm, "ICMP6 unreachables sent for unmatched packets: %s\n",
		   mm->icmp6_enabled ? "enabled" : "disabled");
  vlib_cli_output (vm, "Inner fragmentation: %s\n",
		   mm->frag_inner ? "enabled" : "disabled");
  vlib_cli_output (vm, "Fragment packets regardless of DF flag: %s\n",
		   mm->frag_ignore_df ? "enabled" : "disabled");

  /*
   * Counters
   */
  vlib_combined_counter_main_t *cm = mm->domain_counters;
  u64 total_pkts[MAP_N_DOMAIN_COUNTER];
  u64 total_bytes[MAP_N_DOMAIN_COUNTER];
  int which, i;
  vlib_counter_t v;

  memset (total_pkts, 0, sizeof (total_pkts));
  memset (total_bytes, 0, sizeof (total_bytes));

  map_domain_counter_lock (mm);
  vec_foreach (cm, mm->domain_counters)
  {
    which = cm - mm->domain_counters;

    for (i = 0; i < vlib_combined_counter_n_counters (cm); i++)
      {
	vlib_get_combined_counter (cm, i, &v);
	total_pkts[which] += v.packets;
	total_bytes[which] += v.bytes;
      }
  }
  map_domain_counter_unlock (mm);

  vlib_cli_output (vm, "Encapsulated packets: %lld bytes: %lld\n",
		   total_pkts[MAP_DOMAIN_COUNTER_TX],
		   total_bytes[MAP_DOMAIN_COUNTER_TX]);
  vlib_cli_output (vm, "Decapsulated packets: %lld bytes: %lld\n",
		   total_pkts[MAP_DOMAIN_COUNTER_RX],
		   total_bytes[MAP_DOMAIN_COUNTER_RX]);

  vlib_cli_output (vm, "ICMP relayed packets: %d\n",
		   vlib_get_simple_counter (&mm->icmp_relayed, 0));

  return 0;
}

static clib_error_t *
map_params_reass_command_fn (vlib_main_t * vm, unformat_input_t * input,
			     vlib_cli_command_t * cmd)
{
  unformat_input_t _line_input, *line_input = &_line_input;
  u32 lifetime = ~0;
  f64 ht_ratio = (MAP_IP4_REASS_CONF_HT_RATIO_MAX + 1);
  u32 pool_size = ~0;
  u64 buffers = ~(0ull);
  u8 ip4 = 0, ip6 = 0;

  if (!unformat_user (input, unformat_line_input, line_input))
    return 0;

  while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
    {
      if (unformat (line_input, "lifetime %u", &lifetime))
	;
      else if (unformat (line_input, "ht-ratio %lf", &ht_ratio))
	;
      else if (unformat (line_input, "pool-size %u", &pool_size))
	;
      else if (unformat (line_input, "buffers %llu", &buffers))
	;
      else if (unformat (line_input, "ip4"))
	ip4 = 1;
      else if (unformat (line_input, "ip6"))
	ip6 = 1;
      else
	{
	  unformat_free (line_input);
	  return clib_error_return (0, "invalid input");
	}
    }
  unformat_free (line_input);

  if (!ip4 && !ip6)
    return clib_error_return (0, "must specify ip4 and/or ip6");

  if (ip4)
    {
      if (pool_size != ~0 && pool_size > MAP_IP4_REASS_CONF_POOL_SIZE_MAX)
	return clib_error_return (0, "invalid ip4-reass pool-size ( > %d)",
				  MAP_IP4_REASS_CONF_POOL_SIZE_MAX);
      if (ht_ratio != (MAP_IP4_REASS_CONF_HT_RATIO_MAX + 1)
	  && ht_ratio > MAP_IP4_REASS_CONF_HT_RATIO_MAX)
	return clib_error_return (0, "invalid ip4-reass ht-ratio ( > %d)",
				  MAP_IP4_REASS_CONF_HT_RATIO_MAX);
      if (lifetime != ~0 && lifetime > MAP_IP4_REASS_CONF_LIFETIME_MAX)
	return clib_error_return (0, "invalid ip4-reass lifetime ( > %d)",
				  MAP_IP4_REASS_CONF_LIFETIME_MAX);
      if (buffers != ~(0ull) && buffers > MAP_IP4_REASS_CONF_BUFFERS_MAX)
	return clib_error_return (0, "invalid ip4-reass buffers ( > %ld)",
				  MAP_IP4_REASS_CONF_BUFFERS_MAX);
    }

  if (ip6)
    {
      if (pool_size != ~0 && pool_size > MAP_IP6_REASS_CONF_POOL_SIZE_MAX)
	return clib_error_return (0, "invalid ip6-reass pool-size ( > %d)",
				  MAP_IP6_REASS_CONF_POOL_SIZE_MAX);
      if (ht_ratio != (MAP_IP4_REASS_CONF_HT_RATIO_MAX + 1)
	  && ht_ratio > MAP_IP6_REASS_CONF_HT_RATIO_MAX)
	return clib_error_return (0, "invalid ip6-reass ht-log2len ( > %d)",
				  MAP_IP6_REASS_CONF_HT_RATIO_MAX);
      if (lifetime != ~0 && lifetime > MAP_IP6_REASS_CONF_LIFETIME_MAX)
	return clib_error_return (0, "invalid ip6-reass lifetime ( > %d)",
				  MAP_IP6_REASS_CONF_LIFETIME_MAX);
      if (buffers != ~(0ull) && buffers > MAP_IP6_REASS_CONF_BUFFERS_MAX)
	return clib_error_return (0, "invalid ip6-reass buffers ( > %ld)",
				  MAP_IP6_REASS_CONF_BUFFERS_MAX);
    }

  if (ip4)
    {
      u32 reass = 0, packets = 0;
      if (pool_size != ~0)
	{
	  if (map_ip4_reass_conf_pool_size (pool_size, &reass, &packets))
	    {
	      vlib_cli_output (vm, "Could not set ip4-reass pool-size");
	    }
	  else
	    {
	      vlib_cli_output (vm,
			       "Setting ip4-reass pool-size (destroyed-reassembly=%u , dropped-fragments=%u)",
			       reass, packets);
	    }
	}
      if (ht_ratio != (MAP_IP4_REASS_CONF_HT_RATIO_MAX + 1))
	{
	  if (map_ip4_reass_conf_ht_ratio (ht_ratio, &reass, &packets))
	    {
	      vlib_cli_output (vm, "Could not set ip4-reass ht-log2len");
	    }
	  else
	    {
	      vlib_cli_output (vm,
			       "Setting ip4-reass ht-log2len (destroyed-reassembly=%u , dropped-fragments=%u)",
			       reass, packets);
	    }
	}
      if (lifetime != ~0)
	{
	  if (map_ip4_reass_conf_lifetime (lifetime))
	    vlib_cli_output (vm, "Could not set ip4-reass lifetime");
	  else
	    vlib_cli_output (vm, "Setting ip4-reass lifetime");
	}
      if (buffers != ~(0ull))
	{
	  if (map_ip4_reass_conf_buffers (buffers))
	    vlib_cli_output (vm, "Could not set ip4-reass buffers");
	  else
	    vlib_cli_output (vm, "Setting ip4-reass buffers");
	}

      if (map_main.ip4_reass_conf_buffers >
	  map_main.ip4_reass_conf_pool_size *
	  MAP_IP4_REASS_MAX_FRAGMENTS_PER_REASSEMBLY)
	{
	  vlib_cli_output (vm,
			   "Note: 'ip4-reass buffers' > pool-size * max-fragments-per-reassembly.");
	}
    }

  if (ip6)
    {
      u32 reass = 0, packets = 0;
      if (pool_size != ~0)
	{
	  if (map_ip6_reass_conf_pool_size (pool_size, &reass, &packets))
	    {
	      vlib_cli_output (vm, "Could not set ip6-reass pool-size");
	    }
	  else
	    {
	      vlib_cli_output (vm,
			       "Setting ip6-reass pool-size (destroyed-reassembly=%u , dropped-fragments=%u)",
			       reass, packets);
	    }
	}
      if (ht_ratio != (MAP_IP4_REASS_CONF_HT_RATIO_MAX + 1))
	{
	  if (map_ip6_reass_conf_ht_ratio (ht_ratio, &reass, &packets))
	    {
	      vlib_cli_output (vm, "Could not set ip6-reass ht-log2len");
	    }
	  else
	    {
	      vlib_cli_output (vm,
			       "Setting ip6-reass ht-log2len (destroyed-reassembly=%u , dropped-fragments=%u)",
			       reass, packets);
	    }
	}
      if (lifetime != ~0)
	{
	  if (map_ip6_reass_conf_lifetime (lifetime))
	    vlib_cli_output (vm, "Could not set ip6-reass lifetime");
	  else
	    vlib_cli_output (vm, "Setting ip6-reass lifetime");
	}
      if (buffers != ~(0ull))
	{
	  if (map_ip6_reass_conf_buffers (buffers))
	    vlib_cli_output (vm, "Could not set ip6-reass buffers");
	  else
	    vlib_cli_output (vm, "Setting ip6-reass buffers");
	}

      if (map_main.ip6_reass_conf_buffers >
	  map_main.ip6_reass_conf_pool_size *
	  MAP_IP6_REASS_MAX_FRAGMENTS_PER_REASSEMBLY)
	{
	  vlib_cli_output (vm,
			   "Note: 'ip6-reass buffers' > pool-size * max-fragments-per-reassembly.");
	}
    }

  return 0;
}


/*
 * packet trace format function
 */
u8 *
format_map_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 *);
  map_trace_t *t = va_arg (*args, map_trace_t *);
  u32 map_domain_index = t->map_domain_index;
  u16 port = t->port;

  s =
    format (s, "MAP domain index: %d L4 port: %u", map_domain_index,
	    clib_net_to_host_u16 (port));

  return s;
}

static_always_inline map_ip4_reass_t *
map_ip4_reass_lookup (map_ip4_reass_key_t * k, u32 bucket, f64 now)
{
  map_main_t *mm = &map_main;
  u32 ri = mm->ip4_reass_hash_table[bucket];
  while (ri != MAP_REASS_INDEX_NONE)
    {
      map_ip4_reass_t *r = pool_elt_at_index (mm->ip4_reass_pool, ri);
      if (r->key.as_u64[0] == k->as_u64[0] &&
	  r->key.as_u64[1] == k->as_u64[1] &&
	  now < r->ts + (((f64) mm->ip4_reass_conf_lifetime_ms) / 1000))
	{
	  return r;
	}
      ri = r->bucket_next;
    }
  return NULL;
}

#define map_ip4_reass_pool_index(r) (r - map_main.ip4_reass_pool)

void
map_ip4_reass_free (map_ip4_reass_t * r, u32 ** pi_to_drop)
{
  map_main_t *mm = &map_main;
  map_ip4_reass_get_fragments (r, pi_to_drop);

  // Unlink in hash bucket
  map_ip4_reass_t *r2 = NULL;
  u32 r2i = mm->ip4_reass_hash_table[r->bucket];
  while (r2i != map_ip4_reass_pool_index (r))
    {
      ASSERT (r2i != MAP_REASS_INDEX_NONE);
      r2 = pool_elt_at_index (mm->ip4_reass_pool, r2i);
      r2i = r2->bucket_next;
    }
  if (r2)
    {
      r2->bucket_next = r->bucket_next;
    }
  else
    {
      mm->ip4_reass_hash_table[r->bucket] = r->bucket_next;
    }

  // Unlink in list
  if (r->fifo_next == map_ip4_reass_pool_index (r))
    {
      mm->ip4_reass_fifo_last = MAP_REASS_INDEX_NONE;
    }
  else
    {
      if (mm->ip4_reass_fifo_last == map_ip4_reass_pool_index (r))
	mm->ip4_reass_fifo_last = r->fifo_prev;
      pool_elt_at_index (mm->ip4_reass_pool, r->fifo_prev)->fifo_next =
	r->fifo_next;
      pool_elt_at_index (mm->ip4_reass_pool, r->fifo_next)->fifo_prev =
	r->fifo_prev;
    }

  pool_put (mm->ip4_reass_pool, r);
  mm->ip4_reass_allocated--;
}

map_ip4_reass_t *
map_ip4_reass_get (u32 src, u32 dst, u16 fragment_id,
		   u8 protocol, u32 ** pi_to_drop)
{
  map_ip4_reass_t *r;
  map_main_t *mm = &map_main;
  map_ip4_reass_key_t k = {.src.data_u32 = src,
    .dst.data_u32 = dst,
    .fragment_id = fragment_id,
    .protocol = protocol
  };

  u32 h = 0;
#ifdef clib_crc32c_uses_intrinsics
  h = clib_crc32c ((u8 *) k.as_u32, 16);
#else
  u64 tmp = k.as_u32[0] ^ k.as_u32[1] ^ k.as_u32[2] ^ k.as_u32[3];
  h = clib_xxhash (tmp);
#endif
  h = h >> (32 - mm->ip4_reass_ht_log2len);

  f64 now = vlib_time_now (mm->vlib_main);

  //Cache garbage collection
  while (mm->ip4_reass_fifo_last != MAP_REASS_INDEX_NONE)
    {
      map_ip4_reass_t *last =
	pool_elt_at_index (mm->ip4_reass_pool, mm->ip4_reass_fifo_last);
      if (last->ts + (((f64) mm->ip4_reass_conf_lifetime_ms) / 1000) < now)
	map_ip4_reass_free (last, pi_to_drop);
      else
	break;
    }

  if ((r = map_ip4_reass_lookup (&k, h, now)))
    return r;

  if (mm->ip4_reass_allocated >= mm->ip4_reass_conf_pool_size)
    return NULL;

  pool_get (mm->ip4_reass_pool, r);
  mm->ip4_reass_allocated++;
  int i;
  for (i = 0; i < MAP_IP4_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++)
    r->fragments[i] = ~0;

  u32 ri = map_ip4_reass_pool_index (r);

  //Link in new bucket
  r->bucket = h;
  r->bucket_next = mm->ip4_reass_hash_table[h];
  mm->ip4_reass_hash_table[h] = ri;

  //Link in fifo
  if (mm->ip4_reass_fifo_last != MAP_REASS_INDEX_NONE)
    {
      r->fifo_next =
	pool_elt_at_index (mm->ip4_reass_pool,
			   mm->ip4_reass_fifo_last)->fifo_next;
      r->fifo_prev = mm->ip4_reass_fifo_last;
      pool_elt_at_index (mm->ip4_reass_pool, r->fifo_prev)->fifo_next = ri;
      pool_elt_at_index (mm->ip4_reass_pool, r->fifo_next)->fifo_prev = ri;
    }
  else
    {
      r->fifo_next = r->fifo_prev = ri;
      mm->ip4_reass_fifo_last = ri;
    }

  //Set other fields
  r->ts = now;
  r->key = k;
  r->port = -1;
#ifdef MAP_IP4_REASS_COUNT_BYTES
  r->expected_total = 0xffff;
  r->forwarded = 0;
#endif

  return r;
}

int
map_ip4_reass_add_fragment (map_ip4_reass_t * r, u32 pi)
{
  if (map_main.ip4_reass_buffered_counter >= map_main.ip4_reass_conf_buffers)
    return -1;

  int i;
  for (i = 0; i < MAP_IP4_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++)
    if (r->fragments[i] == ~0)
      {
	r->fragments[i] = pi;
	map_main.ip4_reass_buffered_counter++;
	return 0;
      }
  return -1;
}

static_always_inline map_ip6_reass_t *
map_ip6_reass_lookup (map_ip6_reass_key_t * k, u32 bucket, f64 now)
{
  map_main_t *mm = &map_main;
  u32 ri = mm->ip6_reass_hash_table[bucket];
  while (ri != MAP_REASS_INDEX_NONE)
    {
      map_ip6_reass_t *r = pool_elt_at_index (mm->ip6_reass_pool, ri);
      if (now < r->ts + (((f64) mm->ip6_reass_conf_lifetime_ms) / 1000) &&
	  r->key.as_u64[0] == k->as_u64[0] &&
	  r->key.as_u64[1] == k->as_u64[1] &&
	  r->key.as_u64[2] == k->as_u64[2] &&
	  r->key.as_u64[3] == k->as_u64[3] &&
	  r->key.as_u64[4] == k->as_u64[4])
	return r;
      ri = r->bucket_next;
    }
  return NULL;
}

#define map_ip6_reass_pool_index(r) (r - map_main.ip6_reass_pool)

void
map_ip6_reass_free (map_ip6_reass_t * r, u32 ** pi_to_drop)
{
  map_main_t *mm = &map_main;
  int i;
  for (i = 0; i < MAP_IP6_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++)
    if (r->fragments[i].pi != ~0)
      {
	vec_add1 (*pi_to_drop, r->fragments[i].pi);
	r->fragments[i].pi = ~0;
	map_main.ip6_reass_buffered_counter--;
      }

  // Unlink in hash bucket
  map_ip6_reass_t *r2 = NULL;
  u32 r2i = mm->ip6_reass_hash_table[r->bucket];
  while (r2i != map_ip6_reass_pool_index (r))
    {
      ASSERT (r2i != MAP_REASS_INDEX_NONE);
      r2 = pool_elt_at_index (mm->ip6_reass_pool, r2i);
      r2i = r2->bucket_next;
    }
  if (r2)
    {
      r2->bucket_next = r->bucket_next;
    }
  else
    {
      mm->ip6_reass_hash_table[r->bucket] = r->bucket_next;
    }

  // Unlink in list
  if (r->fifo_next == map_ip6_reass_pool_index (r))
    {
      //Single element in the list, list is now empty
      mm->ip6_reass_fifo_last = MAP_REASS_INDEX_NONE;
    }
  else
    {
      if (mm->ip6_reass_fifo_last == map_ip6_reass_pool_index (r))	//First element
	mm->ip6_reass_fifo_last = r->fifo_prev;
      pool_elt_at_index (mm->ip6_reass_pool, r->fifo_prev)->fifo_next =
	r->fifo_next;
      pool_elt_at_index (mm->ip6_reass_pool, r->fifo_next)->fifo_prev =
	r->fifo_prev;
    }

  // Free from pool if necessary
  pool_put (mm->ip6_reass_pool, r);
  mm->ip6_reass_allocated--;
}

map_ip6_reass_t *
map_ip6_reass_get (ip6_address_t * src, ip6_address_t * dst, u32 fragment_id,
		   u8 protocol, u32 ** pi_to_drop)
{
  map_ip6_reass_t *r;
  map_main_t *mm = &map_main;
  map_ip6_reass_key_t k = {
    .src = *src,
    .dst = *dst,
    .fragment_id = fragment_id,
    .protocol = protocol
  };

  u32 h = 0;
  int i;

#ifdef clib_crc32c_uses_intrinsics
  h = clib_crc32c ((u8 *) k.as_u32, 40);
#else
  u64 tmp =
    k.as_u64[0] ^ k.as_u64[1] ^ k.as_u64[2] ^ k.as_u64[3] ^ k.as_u64[4];
  h = clib_xxhash (tmp);
#endif

  h = h >> (32 - mm->ip6_reass_ht_log2len);

  f64 now = vlib_time_now (mm->vlib_main);

  //Cache garbage collection
  while (mm->ip6_reass_fifo_last != MAP_REASS_INDEX_NONE)
    {
      map_ip6_reass_t *last =
	pool_elt_at_index (mm->ip6_reass_pool, mm->ip6_reass_fifo_last);
      if (last->ts + (((f64) mm->ip6_reass_conf_lifetime_ms) / 1000) < now)
	map_ip6_reass_free (last, pi_to_drop);
      else
	break;
    }

  if ((r = map_ip6_reass_lookup (&k, h, now)))
    return r;

  if (mm->ip6_reass_allocated >= mm->ip6_reass_conf_pool_size)
    return NULL;

  pool_get (mm->ip6_reass_pool, r);
  mm->ip6_reass_allocated++;
  for (i = 0; i < MAP_IP6_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++)
    {
      r->fragments[i].pi = ~0;
      r->fragments[i].next_data_len = 0;
      r->fragments[i].next_data_offset = 0;
    }

  u32 ri = map_ip6_reass_pool_index (r);

  //Link in new bucket
  r->bucket = h;
  r->bucket_next = mm->ip6_reass_hash_table[h];
  mm->ip6_reass_hash_table[h] = ri;

  //Link in fifo
  if (mm->ip6_reass_fifo_last != MAP_REASS_INDEX_NONE)
    {
      r->fifo_next =
	pool_elt_at_index (mm->ip6_reass_pool,
			   mm->ip6_reass_fifo_last)->fifo_next;
      r->fifo_prev = mm->ip6_reass_fifo_last;
      pool_elt_at_index (mm->ip6_reass_pool, r->fifo_prev)->fifo_next = ri;
      pool_elt_at_index (mm->ip6_reass_pool, r->fifo_next)->fifo_prev = ri;
    }
  else
    {
      r->fifo_next = r->fifo_prev = ri;
      mm->ip6_reass_fifo_last = ri;
    }

  //Set other fields
  r->ts = now;
  r->key = k;
  r->ip4_header.ip_version_and_header_length = 0;
#ifdef MAP_IP6_REASS_COUNT_BYTES
  r->expected_total = 0xffff;
  r->forwarded = 0;
#endif
  return r;
}

int
map_ip6_reass_add_fragment (map_ip6_reass_t * r, u32 pi,
			    u16 data_offset, u16 next_data_offset,
			    u8 * data_start, u16 data_len)
{
  map_ip6_fragment_t *f = NULL, *prev_f = NULL;
  u16 copied_len = (data_len > 20) ? 20 : data_len;

  if (map_main.ip6_reass_buffered_counter >= map_main.ip6_reass_conf_buffers)
    return -1;

  //Lookup for fragments for the current buffer
  //and the one before that
  int i;
  for (i = 0; i < MAP_IP6_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++)
    {
      if (data_offset && r->fragments[i].next_data_offset == data_offset)
	{
	  prev_f = &r->fragments[i];	// This is buffer for previous packet
	}
      else if (r->fragments[i].next_data_offset == next_data_offset)
	{
	  f = &r->fragments[i];	// This is a buffer for the current packet
	}
      else if (r->fragments[i].next_data_offset == 0)
	{			//Available
	  if (f == NULL)
	    f = &r->fragments[i];
	  else if (prev_f == NULL)
	    prev_f = &r->fragments[i];
	}
    }

  if (!f || f->pi != ~0)
    return -1;

  if (data_offset)
    {
      if (!prev_f)
	return -1;

      clib_memcpy (prev_f->next_data, data_start, copied_len);
      prev_f->next_data_len = copied_len;
      prev_f->next_data_offset = data_offset;
    }
  else
    {
      if (((ip4_header_t *) data_start)->ip_version_and_header_length != 0x45)
	return -1;

      if (r->ip4_header.ip_version_and_header_length == 0)
	clib_memcpy (&r->ip4_header, data_start, sizeof (ip4_header_t));
    }

  if (data_len > 20)
    {
      f->next_data_offset = next_data_offset;
      f->pi = pi;
      map_main.ip6_reass_buffered_counter++;
    }
  return 0;
}

void
map_ip4_reass_reinit (u32 * trashed_reass, u32 * dropped_packets)
{
  map_main_t *mm = &map_main;
  int i;

  if (dropped_packets)
    *dropped_packets = mm->ip4_reass_buffered_counter;
  if (trashed_reass)
    *trashed_reass = mm->ip4_reass_allocated;
  if (mm->ip4_reass_fifo_last != MAP_REASS_INDEX_NONE)
    {
      u16 ri = mm->ip4_reass_fifo_last;
      do
	{
	  map_ip4_reass_t *r = pool_elt_at_index (mm->ip4_reass_pool, ri);
	  for (i = 0; i < MAP_IP4_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++)
	    if (r->fragments[i] != ~0)
	      map_ip4_drop_pi (r->fragments[i]);

	  ri = r->fifo_next;
	  pool_put (mm->ip4_reass_pool, r);
	}
      while (ri != mm->ip4_reass_fifo_last);
    }

  vec_free (mm->ip4_reass_hash_table);
  vec_resize (mm->ip4_reass_hash_table, 1 << mm->ip4_reass_ht_log2len);
  for (i = 0; i < (1 << mm->ip4_reass_ht_log2len); i++)
    mm->ip4_reass_hash_table[i] = MAP_REASS_INDEX_NONE;
  pool_free (mm->ip4_reass_pool);
  pool_alloc (mm->ip4_reass_pool, mm->ip4_reass_conf_pool_size);

  mm->ip4_reass_allocated = 0;
  mm->ip4_reass_fifo_last = MAP_REASS_INDEX_NONE;
  mm->ip4_reass_buffered_counter = 0;
}

u8
map_get_ht_log2len (f32 ht_ratio, u16 pool_size)
{
  u32 desired_size = (u32) (pool_size * ht_ratio);
  u8 i;
  for (i = 1; i < 31; i++)
    if ((1 << i) >= desired_size)
      return i;
  return 4;
}

int
map_ip4_reass_conf_ht_ratio (f32 ht_ratio, u32 * trashed_reass,
			     u32 * dropped_packets)
{
  map_main_t *mm = &map_main;
  if (ht_ratio > MAP_IP4_REASS_CONF_HT_RATIO_MAX)
    return -1;

  map_ip4_reass_lock ();
  mm->ip4_reass_conf_ht_ratio = ht_ratio;
  mm->ip4_reass_ht_log2len =
    map_get_ht_log2len (ht_ratio, mm->ip4_reass_conf_pool_size);
  map_ip4_reass_reinit (trashed_reass, dropped_packets);
  map_ip4_reass_unlock ();
  return 0;
}

int
map_ip4_reass_conf_pool_size (u16 pool_size, u32 * trashed_reass,
			      u32 * dropped_packets)
{
  map_main_t *mm = &map_main;
  if (pool_size > MAP_IP4_REASS_CONF_POOL_SIZE_MAX)
    return -1;

  map_ip4_reass_lock ();
  mm->ip4_reass_conf_pool_size = pool_size;
  map_ip4_reass_reinit (trashed_reass, dropped_packets);
  map_ip4_reass_unlock ();
  return 0;
}

int
map_ip4_reass_conf_lifetime (u16 lifetime_ms)
{
  map_main.ip4_reass_conf_lifetime_ms = lifetime_ms;
  return 0;
}

int
map_ip4_reass_conf_buffers (u32 buffers)
{
  map_main.ip4_reass_conf_buffers = buffers;
  return 0;
}

void
map_ip6_reass_reinit (u32 * trashed_reass, u32 * dropped_packets)
{
  map_main_t *mm = &map_main;
  if (dropped_packets)
    *dropped_packets = mm->ip6_reass_buffered_counter;
  if (trashed_reass)
    *trashed_reass = mm->ip6_reass_allocated;
  int i;
  if (mm->ip6_reass_fifo_last != MAP_REASS_INDEX_NONE)
    {
      u16 ri = mm->ip6_reass_fifo_last;
      do
	{
	  map_ip6_reass_t *r = pool_elt_at_index (mm->ip6_reass_pool, ri);
	  for (i = 0; i < MAP_IP6_REASS_MAX_FRAGMENTS_PER_REASSEMBLY; i++)
	    if (r->fragments[i].pi != ~0)
	      map_ip6_drop_pi (r->fragments[i].pi);

	  ri = r->fifo_next;
	  pool_put (mm->ip6_reass_pool, r);
	}
      while (ri != mm->ip6_reass_fifo_last);
      mm->ip6_reass_fifo_last = MAP_REASS_INDEX_NONE;
    }

  vec_free (mm->ip6_reass_hash_table);
  vec_resize (mm->ip6_reass_hash_table, 1 << mm->ip6_reass_ht_log2len);
  for (i = 0; i < (1 << mm->ip6_reass_ht_log2len); i++)
    mm->ip6_reass_hash_table[i] = MAP_REASS_INDEX_NONE;
  pool_free (mm->ip6_reass_pool);
  pool_alloc (mm->ip6_reass_pool, mm->ip4_reass_conf_pool_size);

  mm->ip6_reass_allocated = 0;
  mm->ip6_reass_buffered_counter = 0;
}

int
map_ip6_reass_conf_ht_ratio (f32 ht_ratio, u32 * trashed_reass,
			     u32 * dropped_packets)
{
  map_main_t *mm = &map_main;
  if (ht_ratio > MAP_IP6_REASS_CONF_HT_RATIO_MAX)
    return -1;

  map_ip6_reass_lock ();
  mm->ip6_reass_conf_ht_ratio = ht_ratio;
  mm->ip6_reass_ht_log2len =
    map_get_ht_log2len (ht_ratio, mm->ip6_reass_conf_pool_size);
  map_ip6_reass_reinit (trashed_reass, dropped_packets);
  map_ip6_reass_unlock ();
  return 0;
}

int
map_ip6_reass_conf_pool_size (u16 pool_size, u32 * trashed_reass,
			      u32 * dropped_packets)
{
  map_main_t *mm = &map_main;
  if (pool_size > MAP_IP6_REASS_CONF_POOL_SIZE_MAX)
    return -1;

  map_ip6_reass_lock ();
  mm->ip6_reass_conf_pool_size = pool_size;
  map_ip6_reass_reinit (trashed_reass, dropped_packets);
  map_ip6_reass_unlock ();
  return 0;
}

int
map_ip6_reass_conf_lifetime (u16 lifetime_ms)
{
  map_main.ip6_reass_conf_lifetime_ms = lifetime_ms;
  return 0;
}

int
map_ip6_reass_conf_buffers (u32 buffers)
{
  map_main.ip6_reass_conf_buffers = buffers;
  return 0;
}

/* *INDENT-OFF* */

/*?
 * Configure MAP reassembly behaviour
 *
 * @cliexpar
 * @cliexstart{map params reassembly}
 * @cliexend
 ?*/
VLIB_CLI_COMMAND(map_ip4_reass_lifetime_command, static) = {
  .path = "map params reassembly",
  .short_help = "map params reassembly [ip4 | ip6] [lifetime <lifetime-ms>] "
                "[pool-size <pool-size>] [buffers <buffers>] "
                "[ht-ratio <ht-ratio>]",
  .function = map_params_reass_command_fn,
};

/*?
 * Set or copy the IP TOS/Traffic Class field
 *
 * @cliexpar
 * @cliexstart{map params traffic-class}
 *
 * This command is used to set the traffic-class field in translated
 * or encapsulated packets. If copy is specifed (the default) then the
 * traffic-class/TOS field is copied from the original packet to the
 * translated / encapsulating header.
 * @cliexend
 ?*/
VLIB_CLI_COMMAND(map_traffic_class_command, static) = {
  .path = "map params traffic-class",
  .short_help = "map params traffic-class {0x0-0xff | copy}",
  .function = map_traffic_class_command_fn,
};

/*?
 * Bypass IP4/IP6 lookup
 *
 * @cliexpar
 * @cliexstart{map params pre-resolve}
 *
 * Bypass a second FIB lookup of the translated or encapsulated
 * packet, and forward the packet directly to the specified
 * next-hop. This optimization trades forwarding flexibility for
 * performance.
 * @cliexend
 ?*/
VLIB_CLI_COMMAND(map_pre_resolve_command, static) = {
  .path = "map params pre-resolve",
  .short_help = " map params pre-resolve {ip4-nh <address>} "
                "| {ip6-nh <address>}",
  .function = map_pre_resolve_command_fn,
};

/*?
 * Enable or disable the MAP-E inbound security check
 *
 * @cliexpar
 * @cliexstart{map params security-check}
 *
 * By default, a decapsulated packet's IPv4 source address will be
 * verified against the outer header's IPv6 source address. Disabling
 * this feature will allow IPv4 source address spoofing.
 * @cliexend
 ?*/
VLIB_CLI_COMMAND(map_security_check_command, static) = {
  .path = "map params security-check",
  .short_help = "map params security-check on|off",
  .function = map_security_check_command_fn,
};

/*?
 * Specifiy the IPv4 source address used for relayed ICMP error messages
 *
 * @cliexpar
 * @cliexstart{map params icmp source-address}
 *
 * This command specifies which IPv4 source address (must be local to
 * the system), that is used for relayed received IPv6 ICMP error
 * messages.
 * @cliexend
 ?*/
VLIB_CLI_COMMAND(map_icmp_relay_source_address_command, static) = {
  .path = "map params icmp source-address",
  .short_help = "map params icmp source-address <ip4-address>",
  .function = map_icmp_relay_source_address_command_fn,
};

/*?
 * Send IPv6 ICMP unreachables
 *
 * @cliexpar
 * @cliexstart{map params icmp6 unreachables}
 *
 * Send IPv6 ICMP unreachable messages back if security check fails or
 * no MAP domain exists.
 * @cliexend
 ?*/
VLIB_CLI_COMMAND(map_icmp_unreachables_command, static) = {
  .path = "map params icmp6 unreachables",
  .short_help = "map params icmp6 unreachables {on|off}",
  .function = map_icmp_unreachables_command_fn,
};

/*?
 * Configure MAP fragmentation behaviour
 *
 * @cliexpar
 * @cliexstart{map params fragment}
 * @cliexend
 ?*/
VLIB_CLI_COMMAND(map_fragment_command, static) = {
  .path = "map params fragment",
  .short_help = "map params fragment inner|outer",
  .function = map_fragment_command_fn,
};

/*?
 * Ignore the IPv4 Don't fragment bit
 *
 * @cliexpar
 * @cliexstart{map params fragment ignore-df}
 *
 * Allows fragmentation of the IPv4 packet even if the DF bit is
 * set. The choice between inner or outer fragmentation of tunnel
 * packets is complicated. The benefit of inner fragmentation is that
 * the ultimate endpoint must reassemble, instead of the tunnel
 * endpoint.
 * @cliexend
 ?*/
VLIB_CLI_COMMAND(map_fragment_df_command, static) = {
  .path = "map params fragment ignore-df",
  .short_help = "map params fragment ignore-df on|off",
  .function = map_fragment_df_command_fn,
};

/*?
 * Specifiy if the inbound security check should be done on fragments
 *
 * @cliexpar
 * @cliexstart{map params security-check fragments}
 *
 * Typically the inbound on-decapsulation security check is only done
 * on the first packet. The packet that contains the L4
 * information. While a security check on every fragment is possible,
 * it has a cost. State must be created on the first fragment.
 * @cliexend
 ?*/
VLIB_CLI_COMMAND(map_security_check_frag_command, static) = {
  .path = "map params security-check fragments",
  .short_help = "map params security-check fragments on|off",
  .function = map_security_check_frag_command_fn,
};

/*?
 * Add MAP domain
 *
 * @cliexpar
 * @cliexstart{map add domain}
 * @cliexend
 ?*/
VLIB_CLI_COMMAND(map_add_domain_command, static) = {
  .path = "map add domain",
  .short_help = "map add domain ip4-pfx <ip4-pfx> ip6-pfx <ip6-pfx> "
      "ip6-src <ip6-pfx> ea-bits-len <n> psid-offset <n> psid-len <n> "
      "[map-t] [map-ce] [mtu <mtu>]",
  .function = map_add_domain_command_fn,
};

/*?
 * Add MAP rule to a domain
 *
 * @cliexpar
 * @cliexstart{map add rule}
 * @cliexend
 ?*/
VLIB_CLI_COMMAND(map_add_rule_command, static) = {
  .path = "map add rule",
  .short_help = "map add rule index <domain> psid <psid> ip6-dst <ip6-addr>",
  .function = map_add_rule_command_fn,
};

/*?
 * Delete MAP domain
 *
 * @cliexpar
 * @cliexstart{map del domain}
 * @cliexend
 ?*/
VLIB_CLI_COMMAND(map_del_command, static) = {
  .path = "map del domain",
  .short_help = "map del domain index <domain>",
  .function = map_del_domain_command_fn,
};

/*?
 * Show MAP domains
 *
 * @cliexpar
 * @cliexstart{show map domain}
 * @cliexend
 ?*/
VLIB_CLI_COMMAND(show_map_domain_command, static) = {
  .path = "show map domain",
  .short_help = "show map domain index <n> [counters]",
  .function = show_map_domain_command_fn,
};

/*?
 * Show MAP statistics
 *
 * @cliexpar
 * @cliexstart{show map stats}
 * @cliexend
 ?*/
VLIB_CLI_COMMAND(show_map_stats_command, static) = {
  .path = "show map stats",
  .short_help = "show map stats",
  .function = show_map_stats_command_fn,
};

/*?
 * Show MAP fragmentation information
 *
 * @cliexpar
 * @cliexstart{show map fragments}
 * @cliexend
 ?*/
VLIB_CLI_COMMAND(show_map_fragments_command, static) = {
  .path = "show map fragments",
  .short_help = "show map fragments",
  .function = show_map_fragments_command_fn,
};

VLIB_PLUGIN_REGISTER() = {
  .version = VPP_BUILD_VER,
  .description = "Mapping of address and port (MAP)",
};

/* *INDENT-ON* */

static clib_error_t *
map_config (vlib_main_t * vm, unformat_input_t * input)
{
  map_main_t *mm = &map_main;
  u8 is_ce = false;

  while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT)
    {
      if (unformat (input, "customer edge"))
	is_ce = true;
      else
	return clib_error_return (0, "unknown input '%U'",
				  format_unformat_error, input);
    }

  mm->is_ce = is_ce;

  return 0;
}

VLIB_CONFIG_FUNCTION (map_config, "map");

/*
 * map_init
 */
clib_error_t *
map_init (vlib_main_t * vm)
{
  map_main_t *mm = &map_main;
  clib_error_t *error = 0;
  mm->vnet_main = vnet_get_main ();
  mm->vlib_main = vm;

#ifdef MAP_SKIP_IP6_LOOKUP
  fib_protocol_t proto;

  FOR_EACH_FIB_PROTOCOL (proto)
  {
    map_pre_resolve_init (&pre_resolved[proto]);
  }
#endif

  /* traffic class */
  mm->tc = 0;
  mm->tc_copy = true;

  /* Inbound security check */
  mm->sec_check = true;
  mm->sec_check_frag = false;

  /* ICMP6 Type 1, Code 5 for security check failure */
  mm->icmp6_enabled = false;

  mm->is_ce = false;

  /* Inner or outer fragmentation */
  mm->frag_inner = false;
  mm->frag_ignore_df = false;

  vec_validate (mm->domain_counters, MAP_N_DOMAIN_COUNTER - 1);
  mm->domain_counters[MAP_DOMAIN_COUNTER_RX].name = "/map/rx";
  mm->domain_counters[MAP_DOMAIN_COUNTER_TX].name = "/map/tx";

  vlib_validate_simple_counter (&mm->icmp_relayed, 0);
  vlib_zero_simple_counter (&mm->icmp_relayed, 0);

  /* IP4 virtual reassembly */
  mm->ip4_reass_hash_table = 0;
  mm->ip4_reass_pool = 0;
  mm->ip4_reass_lock =
    clib_mem_alloc_aligned (CLIB_CACHE_LINE_BYTES, CLIB_CACHE_LINE_BYTES);
  *mm->ip4_reass_lock = 0;
  mm->ip4_reass_conf_ht_ratio = MAP_IP4_REASS_HT_RATIO_DEFAULT;
  mm->ip4_reass_conf_lifetime_ms = MAP_IP4_REASS_LIFETIME_DEFAULT;
  mm->ip4_reass_conf_pool_size = MAP_IP4_REASS_POOL_SIZE_DEFAULT;
  mm->ip4_reass_conf_buffers = MAP_IP4_REASS_BUFFERS_DEFAULT;
  mm->ip4_reass_ht_log2len =
    map_get_ht_log2len (mm->ip4_reass_conf_ht_ratio,
			mm->ip4_reass_conf_pool_size);
  mm->ip4_reass_fifo_last = MAP_REASS_INDEX_NONE;
  map_ip4_reass_reinit (NULL, NULL);

  /* IP6 virtual reassembly */
  mm->ip6_reass_hash_table = 0;
  mm->ip6_reass_pool = 0;
  mm->ip6_reass_lock =
    clib_mem_alloc_aligned (CLIB_CACHE_LINE_BYTES, CLIB_CACHE_LINE_BYTES);
  *mm->ip6_reass_lock = 0;
  mm->ip6_reass_conf_ht_ratio = MAP_IP6_REASS_HT_RATIO_DEFAULT;
  mm->ip6_reass_conf_lifetime_ms = MAP_IP6_REASS_LIFETIME_DEFAULT;
  mm->ip6_reass_conf_pool_size = MAP_IP6_REASS_POOL_SIZE_DEFAULT;
  mm->ip6_reass_conf_buffers = MAP_IP6_REASS_BUFFERS_DEFAULT;
  mm->ip6_reass_ht_log2len =
    map_get_ht_log2len (mm->ip6_reass_conf_ht_ratio,
			mm->ip6_reass_conf_pool_size);
  mm->ip6_reass_fifo_last = MAP_REASS_INDEX_NONE;
  map_ip6_reass_reinit (NULL, NULL);

#ifdef MAP_SKIP_IP6_LOOKUP
  fib_node_register_type (FIB_NODE_TYPE_MAP_E, &map_vft);
#endif
  map_dpo_module_init ();

  error = map_plugin_api_hookup (vm);

  return error;
}

VLIB_INIT_FUNCTION (map_init);

/*
 * fd.io coding-style-patch-verification: ON
 *
 * Local Variables:
 * eval: (c-set-style "gnu")
 * End:
 */
Age	Commit message (Collapse)	Author	Files	Lines
2022-04-05	vppinfra: refactor address sanitizer	Damjan Marion	1	-7/+7
	Type: refactor Change-Id: I5ca142ec1557d5b5c3806b43553ad9d3b5ea1112 Signed-off-by: Damjan Marion <damarion@cisco.com>
2021-06-24	svm: fix asan check failed @svm_map_region on arm	Tianyu Li	1	-4/+3
	==283032==AddressSanitizer CHECK failed: compiler-rt/lib/asan/asan_mapping.h:366 "((AddrIsInMem(p))) != (0)" (0x0, 0x0) #0 0x49c128 in __asan::AsanCheckFailed #1 0x4ae8dc in __sanitizer::CheckFailed #2 0x495dec in __asan::ShadowSegmentEndpoint::ShadowSegmentEndpoint #3 0x495e48 in __asan_unpoison_memory_region #4 0xfffff4e851f8 in svm_map_region /home/vpp/src/svm/svm.c:611:7 #5 0xfffff4e86d9c in svm_region_init_internal /home/vpp/src/svm/svm.c:797:8 #6 0xfffff4e87ce4 in svm_region_init_args /home/vpp/src/svm/svm.c:880:3 #7 0xfffff7f30d30 in vlibmemory_init /home/vpp/src/vlibmemory/memory_api.c:974:3 #8 0xfffff4fd5368 in vlib_main /home/vpp/src/vlib/main.c:1986:16 svm_global_region_base_va 0x200000000000 is not in the aarch64 mapping range, leading check failure and vpp cannot start. aarch64 asan mapping \|\| `[0x201000000000, 0xffffffffffff]` \|\| HighMem \|\| \|\| `[0x041200000000, 0x200fffffffff]` \|\| HighShadow \|\| \|\| `[0x001200000000, 0x0411ffffffff]` \|\| ShadowGap \|\| \|\| `[0x001000000000, 0x0011ffffffff]` \|\| LowShadow \|\| \|\| `[0x000000000000, 0x000fffffffff]` \|\| LowMem \|\| x86 asan mapping \|\| `[0x10007fff8000, 0x7fffffffffff]` \|\| HighMem \|\| \|\| `[0x02008fff7000, 0x10007fff7fff]` \|\| HighShadow \|\| \|\| `[0x00008fff7000, 0x02008fff6fff]` \|\| ShadowGap \|\| \|\| `[0x00007fff8000, 0x00008fff6fff]` \|\| LowShadow \|\| \|\| `[0x000000000000, 0x00007fff7fff]` \|\| LowMem \|\| Type: fix Signed-off-by: Tianyu Li <tianyu.li@arm.com> Change-Id: I55ddbdcd361d66d4cfaf6459b2fa20fd8b64af37
2021-01-24	svm: use standard function to reset stale mutex	Benoît Ganne	1	-2/+7
	Avoid accessing the private data structure of mutexes which is implementation-dependent, eg. musl is different from glibc. Type: improvement Change-Id: I20ec0c1c9faef0749d89a1969cd2430c80ac04b3 Signed-off-by: Benoît Ganne <bganne@cisco.com>
2020-12-14	misc: move to new pool_foreach macros	Damjan Marion	1	-2/+2
	Type: refactor Change-Id: Ie67dc579e88132ddb1ee4a34cb69f96920101772 Signed-off-by: Damjan Marion <damarion@cisco.com>
2020-09-28	vppinfra: don't call dlmalloc API directly from the code	Damjan Marion	1	-8/+5
	- it is confusing from end consumer perspective that some thing is somewhere called heap and somewhere mspace - this is base for additional work where heap pointer is not the same thing like mspace Type: improvement Change-Id: I644d5a0de17690d65d164d8cec3c5654571629ef Signed-off-by: Damjan Marion <damarion@cisco.com>
2020-09-09	svm: fix non-null-terminated string	Benoît Ganne	1	-1/+1
	Type: fix Change-Id: Ia9cd71ce12584cd8ef90bfe357cf762dd7653f71 Signed-off-by: Benoît Ganne <bganne@cisco.com>
2020-04-22	svm: asan: fix asan support	Benoît Ganne	1	-0/+10
	- restrict the unittests SVM address space to what is supported by ASan - mark SVM mmap()ed address space accessible for ASan - SVM shared memory heap scheme means some allocation can happen outside the current process. Lazily mark those accessible for ASan Type: fix Change-Id: I7c196c80b2a5297651d0afa54f1a8e478fcf59b1 Signed-off-by: Benoît Ganne <bganne@cisco.com>
2020-04-02	misc: fix static analysis warnings	Dave Barach	1	-3/+3
	One actual bugfix. Type: fix Ticket: VPP-1837 Signed-off-by: Dave Barach <dave@barachs.net> Change-Id: Icef25167f97a70cc795c0a481174de319ed79ad5
2020-02-11	vppinfra: remove the historical mheap memory allocator	Dave Barach	1	-28/+0
	The mheap allocator has been turned off for several releases. This commit removes the cmake config parameter, parallel support for dlmalloc and mheap, and the mheap allocator itself. Type: refactor Signed-off-by: Dave Barach <dave@barachs.net> Change-Id: I104f88a1f06e47e90e5f7fb3e11cd1ca66467903
2019-12-18	svm: decrease svm map verbosity	Florin Coras	1	-1/+1
	Type: fix Change-Id: I833b6c23c376f1e806f94f0780be365c82ad1f88 Signed-off-by: Florin Coras <fcoras@cisco.com>
2019-12-10	api: multiple connections per process	Dave Barach	1	-1/+1
	Type: feature Signed-off-by: Dave Barach <dave@barachs.net> Change-Id: I2272521d6e69edcd385ef684af6dd4eea5eaa953
2019-11-27	misc: add address sanitizer heap instrumentation	Benoît Ganne	1	-0/+4
	Introduce AddressSanitizer support: https://github.com/google/sanitizers/ This starts with heap instrumentation. vlib_buffer, bihash and stack instrumentation should follow. Type: feature Change-Id: I7f20e235b2f79db72efd0e756f22c75f717a9884 Signed-off-by: Benoît Ganne <bganne@cisco.com>
2019-05-01	svm: remove dead code triggering overflow	Benoît Ganne	1	-26/+0
	split_point is incremented twice per iteration whereas we only check for *split_point != '\0' at the beginning of the iteration. This code seems dead so remove it altogether. Change-Id: Iabaf26cd5c0c8f747e267bcff044540a2376d3f6 Signed-off-by: Benoît Ganne <bganne@cisco.com>
2018-10-23	c11 safe string handling support	Dave Barach	1	-4/+4
	Change-Id: Ied34720ca5a6e6e717eea4e86003e854031b6eab Signed-off-by: Dave Barach <dave@barachs.net>
2018-08-28	32/64 shmem bihash interoperability	Dave Barach	1	-1/+1
	Move the binary api segment above 4gb Change-Id: I40e8aa7a97722a32397f5a538b5ff8344c50d408 Signed-off-by: Dave Barach <dave@barachs.net>
2018-07-18	Add config option to use dlmalloc instead of mheap	Dave Barach	1	-0/+18
	Configure w/ --enable-dlmalloc, see .../build-data/platforms/vpp.mk src/vppinfra/dlmalloc.[ch] are slightly modified versions of the well-known Doug Lea malloc. Main advantage: dlmalloc mspaces have no inherent size limit. Change-Id: I19b3f43f3c65bcfb82c1a265a97922d01912446e Signed-off-by: Dave Barach <dave@barachs.net>
2018-06-05	VPP API: Memory trace	Ole Troan	1	-0/+4
	if you plan to put a hash into shared memory, the key sum and key equal functions MUST be set to constants such as KEY_FUNC_STRING, KEY_FUNC_MEM, etc. -lvppinfra is PIC, which means that the process which set up the hash won't have the same idea where the key sum and key compare functions live in other processes. Change-Id: Ib3b5963a0d2fb467b91e1f16274df66ac74009e9 Signed-off-by: Ole Troan <ot@cisco.com> Signed-off-by: Dave Barach <dave@barachs.net> Signed-off-by: Ole Troan <ot@cisco.com>
2018-05-31	VPP-1299: Reset uid/gid on shared memory segment if client starts first.	Ole Troan	1	-0/+4
	Change-Id: I2b5bc7d864f8829e1dff63fec3c9484078601309 Signed-off-by: Ole Troan <ot@cisco.com>
2018-05-24	VPP-1286: close the fd after mmap-ing svm segments	Dave Barach	1	-0/+2
	Broken for years. Duh. Change-Id: Ie5fb8e802f143aacd3301c45b136b24a8d4f6d74 Signed-off-by: Dave Barach <dave@barachs.net>
2018-05-10	vppinfra: use count_trailing_zeros in sparse_vec_index	Damjan Marion	1	-1/+1
	It is much cheaper to use ctzll than to do shift,subtract and mask in likely case when we are looking for 1st set bit in the uword. Change-Id: I31954081571978878c7098bafad0c85a91755fa2 Signed-off-by: Damjan Marion <damarion@cisco.com>
2018-02-02	vlmemory/svm: fix client detach from svm region	Florin Coras	1	-5/+34
	Clients cannot know at svm region detach time if the shm backing files have been recreated (e.g., if vpp restarts) and therefore should not try to unlink them. Otherwise, terminating clients attached to previous instantiations of a re-allocated region end up making the new instance un-mappable by removing its backing file. Change-Id: Idcd0cab776e63fd75b821bc9f0fac58217b9ccbe Signed-off-by: Florin Coras <fcoras@cisco.com>
2018-01-14	vabits - subtract from the 'end' address instead of from 'bits'.	Gabriel Ganne	1	-2/+3
	This is a fixup for the commits on calc base address on AArch64 based on autodetected VA space size As reported by: Brian Brooks <brian.brooks@arm.com> Signed-off-by: Gabriel Ganne <gabriel.ganne@enea.com> Change-Id: Id1bd7b7d7e5c188d8547c46134082bd4563b92db
2018-01-10	svm: calc base address on AArch64 based on autodetected VA space size	Gabriel Ganne	1	-4/+3
	fix proposal for aec8f8984771cabc79a8ed64f56afcf61465d00a * fix 0/1 bit count * fix memory leak Change-Id: I488229917c463be10b8a5a1b0a3d7723f05061d0 Signed-off-by: Gabriel Ganne <gabriel.ganne@enea.com>
2018-01-10	svm: calc base address on AArch64 based on autodetected VA space size	Damjan Marion	1	-3/+40
	Change-Id: I7487eb74b8deebff849d662b55a6708566ccd9ef Signed-off-by: Damjan Marion <damarion@cisco.com>
2017-10-03	Repair vlib API socket server	Dave Barach	1	-91/+102
	- Teach vpp_api_test to send/receive API messages over sockets - Add memfd-based shared memory - Add api messages to create memfd-based shared memory segments - vpp_api_test supports both socket and shared memory segment connections - vpp_api_test pivot from socket to shared memory API messaging - add socket client support to libvlibclient.so - dead client reaper sends ping messages, container-friendly - dead client reaper falls back to kill (<pid>, 0) live checking if e.g. a python app goes silent for tens of seconds - handle ping messages in python client support code - teach show api ring about pairwise shared-memory segments - fix ip probing of already resolved destinations (VPP-998) We'll need this work to implement proper host-stack client isolation Change-Id: Ic23b65f75c854d0393d9a2e9d6b122a9551be769 Signed-off-by: Dave Barach <dave@barachs.net> Signed-off-by: Dave Wallace <dwallacelf@gmail.com> Signed-off-by: Florin Coras <fcoras@cisco.com>
2017-09-06	Set uid/gid on ssvm segment file.	Dave Wallace	1	-1/+3
	Change-Id: I482bb9654f4dfe240bace5c2b61056cfd04cf018 Signed-off-by: Dave Wallace <dwallacelf@gmail.com>
2017-08-25	jvpp: introducing callback api and future api tests for all plugins (VPP-591)	Matej Perina	1	-1/+1
	test can be run with: make test TEST=test_jvpp memory_shared.c: declaring and assigning variable in if statement makes it usage outside statement impossible. Looks like memory space assigned to variable declared in statement is freed when statement ends svm.c: - fixed case when root path can have a "/" at beggining - added option for test to operate over shared memory space with /vpe-api name and not create new one with name consisting of root path and region name which would require root permisions Change-Id: Iff1170dc6a5c1be134c152f2757c7ab9b919a8ed Signed-off-by: Matej Perina <mperina@cisco.com>
2017-08-18	API: More gracefully fail when opening shared memory segment fails.	Ole Troan	1	-5/+8
	API clients would fail with an ASSERT (and core dump) whenever the API shared memory segment could not be opened. This returns an error value to the client's connect instead. Change-Id: Id122a3a090b24b139c382ae09f341bde61fd2540 Signed-off-by: Ole Troan <ot@cisco.com>
2017-08-14	jvpp: make shm_prefix configurable (VPP-591)	Jan Srnicek	1	-1/+2
	svm.c - set default map region root path only if root path is not already present memory_shared.c - added option for tests to send memory region name and root path in one variable, if so name and root path are separated here and set to map region structure so find function can find it properly jvpp-registry.c - added parameters shmPrefix to be able pass + removed sudo restriction specific shared memory prefix that is used while starting python tests(see framework.py) JVppRegistyImpl - added option to specify shmPrefix VppJNIConnection - added option to specify shmPrefix Change-Id: I3f89f867fb9b20eef00fbd497cb0e41b25d6eab7 Signed-off-by: Jan Srnicek <jsrnicek@cisco.com> Signed-off-by: Matej Perina <mperina@cisco.com>
2017-07-17	Fix unlinking of /dev/shm files.	Dave Wallace	1	-7/+31
	- api-segment prefix not used when unlinking shm files - unlink root region on exit if no clients referenced - stale reference to freed segment name - don't add fake client to /db unless CLIB_DEBUG > 2 - turn off the gmond plugin - clean up unused vars in vpp/api Change-Id: I66451fcfd6ee64a12466c2d6c209050e3cdb74b7 Signed-off-by: Dave Wallace <dwallacelf@gmail.com> Signed-off-by: Dave Barach <dave@barachs.net>
2017-05-09	Fix remaining 32-bit compile issues	Damjan Marion	1	-5/+6
	Change-Id: I9664214652229b663c3e3ba7406b4ede96bfb123 Signed-off-by: Damjan Marion <damarion@cisco.com>
2017-05-03	A sprinkling of const in vlibmemory/api.h and friends	Neale Ranns	1	-3/+3
	Change-Id: I953ebb37eeec7de0c4a6b00258c3c67a83cbc020 Signed-off-by: Neale Ranns <nranns@cisco.com>
2016-12-28	Reorganize source tree to use single autotools instance	Damjan Marion	1	-0/+1237
	Change-Id: I7b51f88292e057c6443b12224486f2d0c9f8ae23 Signed-off-by: Damjan Marion <damarion@cisco.com>