summaryrefslogtreecommitdiffstats
path: root/src/plugins/dpdk/buffer.h
diff options
context:
space:
mode:
authorDamjan Marion <damarion@cisco.com>2019-01-20 02:02:51 +0100
committerDamjan Marion <dmarion@me.com>2019-01-20 22:59:03 +0000
commita5c308e68d5578e5d75327619e30d73a95abb1a5 (patch)
treea1d6c591b886bfa15dcce966718e4a7faaf5378a /src/plugins/dpdk/buffer.h
parent3235382f23074e7f5b413b862f8cfed37142dbcd (diff)
dpdk: add buffer.h
Change-Id: I998658ad7860b23425444e218ce2e1ec655b885a Signed-off-by: Damjan Marion <damarion@cisco.com>
Diffstat (limited to 'src/plugins/dpdk/buffer.h')
-rw-r--r--src/plugins/dpdk/buffer.h34
1 files changed, 34 insertions, 0 deletions
diff --git a/src/plugins/dpdk/buffer.h b/src/plugins/dpdk/buffer.h
new file mode 100644
index 00000000000..56bf4fff3e6
--- /dev/null
+++ b/src/plugins/dpdk/buffer.h
@@ -0,0 +1,34 @@
+/*
+ * Copyright (c) 2017-2019 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef include_dpdk_buffer_h
+#define include_dpdk_buffer_h
+
+#define rte_mbuf_from_vlib_buffer(x) (((struct rte_mbuf *)x) - 1)
+#define vlib_buffer_from_rte_mbuf(x) ((vlib_buffer_t *)(x+1))
+
+clib_error_t *dpdk_buffer_pool_create (vlib_main_t * vm, unsigned num_mbufs,
+ unsigned socket_id);
+
+#endif /* include_dpdk_buffer_h */
+
+/** @endcond */
+/*
+ * fd.io coding-style-patch-verification: ON
+ *
+ * Local Variables:
+ * eval: (c-set-style "gnu")
+ * End:
+ */
n218' href='#n218'>218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 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 
/*
 * Copyright (c) 2019 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 "nat_ha.h"
#include <vnet/udp/udp.h>
#include <nat/nat.h>
#include <vppinfra/atomics.h>

/* number of retries */
#define NAT_HA_RETRIES 3

#define foreach_nat_ha_counter           \
_(RECV_ADD, "add-event-recv", 0)         \
_(RECV_DEL, "del-event-recv", 1)         \
_(RECV_REFRESH, "refresh-event-recv", 2) \
_(SEND_ADD, "add-event-send", 3)         \
_(SEND_DEL, "del-event-send", 4)         \
_(SEND_REFRESH, "refresh-event-send", 5) \
_(RECV_ACK, "ack-recv", 6)               \
_(SEND_ACK, "ack-send", 7)               \
_(RETRY_COUNT, "retry-count", 8)         \
_(MISSED_COUNT, "missed-count", 9)

/* NAT HA protocol version */
#define NAT_HA_VERSION 0x01

/* NAT HA protocol flags */
#define NAT_HA_FLAG_ACK 0x01

/* NAT HA event types */
typedef enum
{
  NAT_HA_ADD = 1,
  NAT_HA_DEL,
  NAT_HA_REFRESH,
} nat_ha_event_type_t;

/* NAT HA protocol header */
typedef struct
{
  /* version */
  u8 version;
  /* flags */
  u8 flags;
  /* event count */
  u16 count;
  /* sequence number */
  u32 sequence_number;
  /* thread index where events originated */
  u32 thread_index;
} __attribute__ ((packed)) nat_ha_message_header_t;

/* NAT HA protocol event data */
typedef struct
{
  /* event type */
  u8 event_type;
  /* session data */
  u8 protocol;
  u16 flags;
  u32 in_addr;
  u32 out_addr;
  u16 in_port;
  u16 out_port;
  u32 eh_addr;
  u32 ehn_addr;
  u16 eh_port;
  u16 ehn_port;
  u32 fib_index;
  u32 total_pkts;
  u64 total_bytes;
} __attribute__ ((packed)) nat_ha_event_t;

typedef enum
{
#define _(N, s, v) NAT_HA_COUNTER_##N = v,
  foreach_nat_ha_counter
#undef _
  NAT_HA_N_COUNTERS
} nat_ha_counter_t;

/* data waiting for ACK */
typedef struct
{
  /* sequence number */
  u32 seq;
  /* retry count */
  u32 retry_count;
  /* next retry time */
  f64 retry_timer;
  /* 1 if HA resync */
  u8 is_resync;
  /* packet data */
  u8 *data;
} nat_ha_resend_entry_t;

/* per thread data */
typedef struct
{
  /* buffer under construction */
  vlib_buffer_t *state_sync_buffer;
  /* frame containing NAT HA buffers */
  vlib_frame_t *state_sync_frame;
  /* number of events */
  u16 state_sync_count;
  /* next event offset */
  u32 state_sync_next_event_offset;
  /* data waiting for ACK */
  nat_ha_resend_entry_t *resend_queue;
} nat_ha_per_thread_data_t;

/* NAT HA settings */
typedef struct nat_ha_main_s
{
  /* local IP address and UDP port */
  ip4_address_t src_ip_address;
  u16 src_port;
  /* failvoer IP address and UDP port */
  ip4_address_t dst_ip_address;
  u16 dst_port;
  /* path MTU between local and failover */
  u32 state_sync_path_mtu;
  /* number of seconds after which to send session counters refresh */
  u32 session_refresh_interval;
  /* counters */
  vlib_simple_counter_main_t counters[NAT_HA_N_COUNTERS];
  vlib_main_t *vlib_main;
  /* sequence number counter */
  u32 sequence_number;
  /* 1 if resync in progress */
  u8 in_resync;
  /* number of remaing ACK for resync */
  u32 resync_ack_count;
  /* number of missed ACK for resync */
  u32 resync_ack_missed;
  /* resync data */
  nat_ha_resync_event_cb_t event_callback;
  u32 client_index;
  u32 pid;
  /* call back functions for received HA events on failover */
  nat_ha_sadd_cb_t sadd_cb;
  nat_ha_sdel_cb_t sdel_cb;
  nat_ha_sref_cb_t sref_cb;
  /* per thread data */
  u32 num_workers;
  nat_ha_per_thread_data_t *per_thread_data;
  /* worker handoff frame-queue index */
  u32 fq_index;
} nat_ha_main_t;

nat_ha_main_t nat_ha_main;
vlib_node_registration_t nat_ha_process_node;
vlib_node_registration_t nat_ha_worker_node;
vlib_node_registration_t nat_ha_node;
vlib_node_registration_t nat_ha_handoff_node;

static void
nat_ha_resync_fin (void)
{
  nat_ha_main_t *ha = &nat_ha_main;

  /* if no more resync ACK remainig we are done */
  if (ha->resync_ack_count)
    return;

  ha->in_resync = 0;
  if (ha->resync_ack_missed)
    {
      nat_elog_info ("resync completed with result FAILED");
    }
  else
    {
      nat_elog_info ("resync completed with result SUCCESS");
    }
  if (ha->event_callback)
    ha->event_callback (ha->client_index, ha->pid, ha->resync_ack_missed);
}

/* cache HA NAT data waiting for ACK */
static int
nat_ha_resend_queue_add (u32 seq, u8 * data, u8 data_len, u8 is_resync,
			 u32 thread_index)
{
  nat_ha_main_t *ha = &nat_ha_main;
  nat_ha_per_thread_data_t *td = &ha->per_thread_data[thread_index];
  nat_ha_resend_entry_t *entry;
  f64 now = vlib_time_now (ha->vlib_main);

  vec_add2 (td->resend_queue, entry, 1);
  clib_memset (entry, 0, sizeof (*entry));
  entry->retry_timer = now + 2.0;
  entry->seq = seq;
  entry->is_resync = is_resync;
  vec_add (entry->data, data, data_len);

  return 0;
}

static_always_inline void
nat_ha_ack_recv (u32 seq, u32 thread_index)
{
  nat_ha_main_t *ha = &nat_ha_main;
  nat_ha_per_thread_data_t *td = &ha->per_thread_data[thread_index];
  u32 i;

  vec_foreach_index (i, td->resend_queue)
  {
    if (td->resend_queue[i].seq != seq)
      continue;

    vlib_increment_simple_counter (&ha->counters[NAT_HA_COUNTER_RECV_ACK],
				   thread_index, 0, 1);
    /* ACK received remove cached data */
    if (td->resend_queue[i].is_resync)
      {
	clib_atomic_fetch_sub (&ha->resync_ack_count, 1);
	nat_ha_resync_fin ();
      }
    vec_free (td->resend_queue[i].data);
    vec_del1 (td->resend_queue, i);
    nat_elog_debug_X1 ("ACK for seq %d received", "i4",
		       clib_net_to_host_u32 (seq));

    return;
  }
}

/* scan non-ACKed HA NAT for retry */
static void
nat_ha_resend_scan (f64 now, u32 thread_index)
{
  nat_ha_main_t *ha = &nat_ha_main;
  nat_ha_per_thread_data_t *td = &ha->per_thread_data[thread_index];
  u32 i, *del, *to_delete = 0;
  vlib_main_t *vm = ha->vlib_main;
  vlib_buffer_t *b = 0;
  vlib_frame_t *f;
  u32 bi, *to_next;
  ip4_header_t *ip;

  vec_foreach_index (i, td->resend_queue)
  {
    if (td->resend_queue[i].retry_timer > now)
      continue;

    /* maximum retry reached delete cached data */
    if (td->resend_queue[i].retry_count >= NAT_HA_RETRIES)
      {
	nat_elog_notice_X1 ("seq %d missed", "i4",
			    clib_net_to_host_u32 (td->resend_queue[i].seq));
	if (td->resend_queue[i].is_resync)
	  {
	    clib_atomic_fetch_add (&ha->resync_ack_missed, 1);
	    clib_atomic_fetch_sub (&ha->resync_ack_count, 1);
	    nat_ha_resync_fin ();
	  }
	vec_add1 (to_delete, i);
	vlib_increment_simple_counter (&ha->counters
				       [NAT_HA_COUNTER_MISSED_COUNT],
				       thread_index, 0, 1);
	continue;
      }

    /* retry to send non-ACKed data */
    nat_elog_debug_X1 ("state sync seq %d resend", "i4",
		       clib_net_to_host_u32 (td->resend_queue[i].seq));
    td->resend_queue[i].retry_count++;
    vlib_increment_simple_counter (&ha->counters[NAT_HA_COUNTER_RETRY_COUNT],
				   thread_index, 0, 1);
    if (vlib_buffer_alloc (vm, &bi, 1) != 1)
      {
	nat_elog_warn ("HA NAT state sync can't allocate buffer");
	return;
      }
    b = vlib_get_buffer (vm, bi);
    b->current_length = vec_len (td->resend_queue[i].data);
    b->flags |= VLIB_BUFFER_TOTAL_LENGTH_VALID;
    b->flags |= VNET_BUFFER_F_LOCALLY_ORIGINATED;
    vnet_buffer (b)->sw_if_index[VLIB_RX] = 0;
    vnet_buffer (b)->sw_if_index[VLIB_TX] = 0;
    ip = vlib_buffer_get_current (b);
    clib_memcpy (ip, td->resend_queue[i].data,
		 vec_len (td->resend_queue[i].data));
    f = vlib_get_frame_to_node (vm, ip4_lookup_node.index);
    to_next = vlib_frame_vector_args (f);
    to_next[0] = bi;
    f->n_vectors = 1;
    vlib_put_frame_to_node (vm, ip4_lookup_node.index, f);
    td->resend_queue[i].retry_timer = now + 2.0;
  }

  vec_foreach (del, to_delete)
  {
    vec_free (td->resend_queue[*del].data);
    vec_del1 (td->resend_queue, *del);
  }
  vec_free (to_delete);
}

void
nat_ha_init (vlib_main_t * vm, nat_ha_sadd_cb_t sadd_cb,
	     nat_ha_sdel_cb_t sdel_cb, nat_ha_sref_cb_t sref_cb)
{
  nat_ha_main_t *ha = &nat_ha_main;
  vlib_thread_main_t *tm = vlib_get_thread_main ();
  vlib_thread_registration_t *tr;
  uword *p;

  ha->src_ip_address.as_u32 = 0;
  ha->src_port = 0;
  ha->dst_ip_address.as_u32 = 0;
  ha->dst_port = 0;
  ha->in_resync = 0;
  ha->resync_ack_count = 0;
  ha->resync_ack_missed = 0;
  ha->vlib_main = vm;
  ha->sadd_cb = sadd_cb;
  ha->sdel_cb = sdel_cb;
  ha->sref_cb = sref_cb;
  ha->num_workers = 0;
  vec_validate (ha->per_thread_data, tm->n_vlib_mains - 1);
  ha->fq_index = ~0;
  p = hash_get_mem (tm->thread_registrations_by_name, "workers");
  if (p)
    {
      tr = (vlib_thread_registration_t *) p[0];
      if (tr)
	ha->num_workers = tr->count;
    }

#define _(N, s, v) ha->counters[v].name = s;          \
  ha->counters[v].stat_segment_name = "/nat44/ha/" s; \
  vlib_validate_simple_counter(&ha->counters[v], 0);  \
  vlib_zero_simple_counter(&ha->counters[v], 0);
  foreach_nat_ha_counter
#undef _
}

int
nat_ha_set_listener (ip4_address_t * addr, u16 port, u32 path_mtu)
{
  nat_ha_main_t *ha = &nat_ha_main;

  /* unregister previously set UDP port */
  if (ha->src_port)
    udp_unregister_dst_port (ha->vlib_main, ha->src_port, 1);

  ha->src_ip_address.as_u32 = addr->as_u32;
  ha->src_port = port;
  ha->state_sync_path_mtu = path_mtu;

  if (port)
    {
      /* if multiple worker threads first go to handoff node */
      if (ha->num_workers > 1)
	{
	  if (ha->fq_index == ~0)
	    ha->fq_index = vlib_frame_queue_main_init (nat_ha_node.index, 0);
	  udp_register_dst_port (ha->vlib_main, port,
				 nat_ha_handoff_node.index, 1);
	}
      else
	{
	  udp_register_dst_port (ha->vlib_main, port, nat_ha_node.index, 1);
	}
      nat_elog_info_X1 ("HA listening on port %d for state sync", "i4", port);
    }

  return 0;
}

void
nat_ha_get_listener (ip4_address_t * addr, u16 * port, u32 * path_mtu)
{
  nat_ha_main_t *ha = &nat_ha_main;

  addr->as_u32 = ha->src_ip_address.as_u32;
  *port = ha->src_port;
  *path_mtu = ha->state_sync_path_mtu;
}

int
nat_ha_set_failover (ip4_address_t * addr, u16 port,
		     u32 session_refresh_interval)
{
  nat_ha_main_t *ha = &nat_ha_main;

  ha->dst_ip_address.as_u32 = addr->as_u32;
  ha->dst_port = port;
  ha->session_refresh_interval = session_refresh_interval;

  vlib_process_signal_event (ha->vlib_main, nat_ha_process_node.index, 1, 0);

  return 0;
}

void
nat_ha_get_failover (ip4_address_t * addr, u16 * port,
		     u32 * session_refresh_interval)
{
  nat_ha_main_t *ha = &nat_ha_main;

  addr->as_u32 = ha->dst_ip_address.as_u32;
  *port = ha->dst_port;
  *session_refresh_interval = ha->session_refresh_interval;
}

static_always_inline void
nat_ha_recv_add (nat_ha_event_t * event, f64 now, u32 thread_index)
{
  nat_ha_main_t *ha = &nat_ha_main;
  ip4_address_t in_addr, out_addr, eh_addr, ehn_addr;
  u32 fib_index;
  u16 flags;

  vlib_increment_simple_counter (&ha->counters[NAT_HA_COUNTER_RECV_ADD],
				 thread_index, 0, 1);

  in_addr.as_u32 = event->in_addr;
  out_addr.as_u32 = event->out_addr;
  eh_addr.as_u32 = event->eh_addr;
  ehn_addr.as_u32 = event->ehn_addr;
  fib_index = clib_net_to_host_u32 (event->fib_index);
  flags = clib_net_to_host_u16 (event->flags);

  ha->sadd_cb (&in_addr, event->in_port, &out_addr, event->out_port, &eh_addr,
	       event->eh_port, &ehn_addr, event->ehn_port, event->protocol,
	       fib_index, flags, thread_index);
}

static_always_inline void
nat_ha_recv_del (nat_ha_event_t * event, u32 thread_index)
{
  nat_ha_main_t *ha = &nat_ha_main;
  ip4_address_t out_addr, eh_addr;
  u32 fib_index;

  vlib_increment_simple_counter (&ha->counters[NAT_HA_COUNTER_RECV_DEL],
				 thread_index, 0, 1);

  out_addr.as_u32 = event->out_addr;
  eh_addr.as_u32 = event->eh_addr;
  fib_index = clib_net_to_host_u32 (event->fib_index);

  ha->sdel_cb (&out_addr, event->out_port, &eh_addr, event->eh_port,
	       event->protocol, fib_index, thread_index);
}

static_always_inline void
nat_ha_recv_refresh (nat_ha_event_t * event, f64 now, u32 thread_index)
{
  nat_ha_main_t *ha = &nat_ha_main;
  ip4_address_t out_addr, eh_addr;
  u32 fib_index, total_pkts;
  u64 total_bytes;

  vlib_increment_simple_counter (&ha->counters[NAT_HA_COUNTER_RECV_REFRESH],
				 thread_index, 0, 1);

  out_addr.as_u32 = event->out_addr;
  eh_addr.as_u32 = event->eh_addr;
  fib_index = clib_net_to_host_u32 (event->fib_index);
  total_pkts = clib_net_to_host_u32 (event->total_pkts);
  total_bytes = clib_net_to_host_u64 (event->total_bytes);

  ha->sref_cb (&out_addr, event->out_port, &eh_addr, event->eh_port,
	       event->protocol, fib_index, total_pkts, total_bytes,
	       thread_index);
}

/* process received NAT HA event */
static_always_inline void
nat_ha_event_process (nat_ha_event_t * event, f64 now, u32 thread_index)
{
  switch (event->event_type)
    {
    case NAT_HA_ADD:
      nat_ha_recv_add (event, now, thread_index);
      break;
    case NAT_HA_DEL:
      nat_ha_recv_del (event, thread_index);
      break;
    case NAT_HA_REFRESH:
      nat_ha_recv_refresh (event, now, thread_index);
      break;
    default:
      nat_elog_notice_X1 ("Unsupported HA event type %d", "i4",
			  event->event_type);
      break;
    }
}

static inline void
nat_ha_header_create (vlib_buffer_t * b, u32 * offset, u32 thread_index)
{
  nat_ha_main_t *ha = &nat_ha_main;
  nat_ha_message_header_t *h;
  ip4_header_t *ip;
  udp_header_t *udp;
  u32 sequence_number;

  b->current_data = 0;
  b->current_length = sizeof (*ip) + sizeof (*udp) + sizeof (*h);
  b->flags |= VLIB_BUFFER_TOTAL_LENGTH_VALID;
  b->flags |= VNET_BUFFER_F_LOCALLY_ORIGINATED;
  vnet_buffer (b)->sw_if_index[VLIB_RX] = 0;
  vnet_buffer (b)->sw_if_index[VLIB_TX] = 0;
  ip = vlib_buffer_get_current (b);
  udp = (udp_header_t *) (ip + 1);
  h = (nat_ha_message_header_t *) (udp + 1);

  /* IP header */
  ip->ip_version_and_header_length = 0x45;
  ip->ttl = 254;
  ip->protocol = IP_PROTOCOL_UDP;
  ip->flags_and_fragment_offset =
    clib_host_to_net_u16 (IP4_HEADER_FLAG_DONT_FRAGMENT);
  ip->src_address.as_u32 = ha->src_ip_address.as_u32;
  ip->dst_address.as_u32 = ha->dst_ip_address.as_u32;
  /* UDP header */
  udp->src_port = clib_host_to_net_u16 (ha->src_port);
  udp->dst_port = clib_host_to_net_u16 (ha->dst_port);
  udp->checksum = 0;

  /* NAT HA protocol header */
  h->version = NAT_HA_VERSION;
  h->flags = 0;
  h->count = 0;
  h->thread_index = clib_host_to_net_u32 (thread_index);
  sequence_number = clib_atomic_fetch_add (&ha->sequence_number, 1);
  h->sequence_number = clib_host_to_net_u32 (sequence_number);

  *offset =
    sizeof (ip4_header_t) + sizeof (udp_header_t) +
    sizeof (nat_ha_message_header_t);
}

static inline void
nat_ha_send (vlib_frame_t * f, vlib_buffer_t * b, u8 is_resync,
	     u32 thread_index)
{
  nat_ha_main_t *ha = &nat_ha_main;
  nat_ha_per_thread_data_t *td = &ha->per_thread_data[thread_index];
  nat_ha_message_header_t *h;
  ip4_header_t *ip;
  udp_header_t *udp;
  vlib_main_t *vm = vlib_mains[thread_index];

  ip = vlib_buffer_get_current (b);
  udp = ip4_next_header (ip);
  h = (nat_ha_message_header_t *) (udp + 1);

  h->count = clib_host_to_net_u16 (td->state_sync_count);

  ip->length = clib_host_to_net_u16 (b->current_length);
  ip->checksum = ip4_header_checksum (ip);
  udp->length = clib_host_to_net_u16 (b->current_length - sizeof (*ip));

  nat_ha_resend_queue_add (h->sequence_number, (u8 *) ip, b->current_length,
			   is_resync, thread_index);

  vlib_put_frame_to_node (vm, ip4_lookup_node.index, f);
}

/* add NAT HA protocol event */
static_always_inline void
nat_ha_event_add (nat_ha_event_t * event, u8 do_flush, u32 thread_index,
		  u8 is_resync)
{
  nat_ha_main_t *ha = &nat_ha_main;
  nat_ha_per_thread_data_t *td = &ha->per_thread_data[thread_index];
  vlib_main_t *vm = vlib_mains[thread_index];
  vlib_buffer_t *b = 0;
  vlib_frame_t *f;
  u32 bi = ~0, offset;

  b = td->state_sync_buffer;

  if (PREDICT_FALSE (b == 0))
    {
      if (do_flush)
	return;

      if (vlib_buffer_alloc (vm, &bi, 1) != 1)
	{
	  nat_elog_warn ("HA NAT state sync can't allocate buffer");
	  return;
	}

      b = td->state_sync_buffer = vlib_get_buffer (vm, bi);
      clib_memset (vnet_buffer (b), 0, sizeof (*vnet_buffer (b)));
      VLIB_BUFFER_TRACE_TRAJECTORY_INIT (b);
      offset = 0;
    }
  else
    {
      bi = vlib_get_buffer_index (vm, b);
      offset = td->state_sync_next_event_offset;
    }

  f = td->state_sync_frame;
  if (PREDICT_FALSE (f == 0))
    {
      u32 *to_next;
      f = vlib_get_frame_to_node (vm, ip4_lookup_node.index);
      td->state_sync_frame = f;
      to_next = vlib_frame_vector_args (f);
      to_next[0] = bi;
      f->n_vectors = 1;
    }

  if (PREDICT_FALSE (td->state_sync_count == 0))
    nat_ha_header_create (b, &offset, thread_index);

  if (PREDICT_TRUE (do_flush == 0))
    {
      clib_memcpy_fast (b->data + offset, event, sizeof (*event));
      offset += sizeof (*event);
      td->state_sync_count++;
      b->current_length += sizeof (*event);

      switch (event->event_type)
	{
	case NAT_HA_ADD:
	  vlib_increment_simple_counter (&ha->counters
					 [NAT_HA_COUNTER_SEND_ADD],
					 thread_index, 0, 1);
	  break;
	case NAT_HA_DEL:
	  vlib_increment_simple_counter (&ha->counters
					 [NAT_HA_COUNTER_SEND_DEL],
					 thread_index, 0, 1);
	  break;
	case NAT_HA_REFRESH:
	  vlib_increment_simple_counter (&ha->counters
					 [NAT_HA_COUNTER_SEND_REFRESH],
					 thread_index, 0, 1);
	  break;
	default:
	  break;
	}
    }

  if (PREDICT_FALSE
      (do_flush || offset + (sizeof (*event)) > ha->state_sync_path_mtu))
    {
      nat_ha_send (f, b, is_resync, thread_index);
      td->state_sync_buffer = 0;
      td->state_sync_frame = 0;
      td->state_sync_count = 0;
      offset = 0;
      if (is_resync)
	{
	  clib_atomic_fetch_add (&ha->resync_ack_count, 1);
	  nat_ha_resync_fin ();
	}
    }

  td->state_sync_next_event_offset = offset;
}

#define skip_if_disabled()          \
do {                                \
  nat_ha_main_t *ha = &nat_ha_main; \
  if (PREDICT_TRUE (!ha->dst_port)) \
    return;                         \
} while (0)

void
nat_ha_flush (u8 is_resync)
{
  skip_if_disabled ();
  nat_ha_event_add (0, 1, 0, is_resync);
}

void
nat_ha_sadd (ip4_address_t * in_addr, u16 in_port, ip4_address_t * out_addr,
	     u16 out_port, ip4_address_t * eh_addr, u16 eh_port,
	     ip4_address_t * ehn_addr, u16 ehn_port, u8 proto, u32 fib_index,
	     u16 flags, u32 thread_index, u8 is_resync)
{
  nat_ha_event_t event;

  skip_if_disabled ();

  clib_memset (&event, 0, sizeof (event));
  event.event_type = NAT_HA_ADD;
  event.flags = clib_host_to_net_u16 (flags);
  event.in_addr = in_addr->as_u32;
  event.in_port = in_port;
  event.out_addr = out_addr->as_u32;
  event.out_port = out_port;
  event.eh_addr = eh_addr->as_u32;
  event.eh_port = eh_port;
  event.ehn_addr = ehn_addr->as_u32;
  event.ehn_port = ehn_port;
  event.fib_index = clib_host_to_net_u32 (fib_index);
  event.protocol = proto;
  nat_ha_event_add (&event, 0, thread_index, is_resync);
}

void
nat_ha_sdel (ip4_address_t * out_addr, u16 out_port, ip4_address_t * eh_addr,
	     u16 eh_port, u8 proto, u32 fib_index, u32 thread_index)
{
  nat_ha_event_t event;

  skip_if_disabled ();

  clib_memset (&event, 0, sizeof (event));
  event.event_type = NAT_HA_DEL;
  event.out_addr = out_addr->as_u32;
  event.out_port = out_port;
  event.eh_addr = eh_addr->as_u32;
  event.eh_port = eh_port;
  event.fib_index = clib_host_to_net_u32 (fib_index);
  event.protocol = proto;
  nat_ha_event_add (&event, 0, thread_index, 0);
}

void
nat_ha_sref (ip4_address_t * out_addr, u16 out_port, ip4_address_t * eh_addr,
	     u16 eh_port, u8 proto, u32 fib_index, u32 total_pkts,
	     u64 total_bytes, u32 thread_index, f64 * last_refreshed, f64 now)
{
  nat_ha_main_t *ha = &nat_ha_main;
  nat_ha_event_t event;

  skip_if_disabled ();

  if ((*last_refreshed + ha->session_refresh_interval) > now)
    return;

  *last_refreshed = now;
  clib_memset (&event, 0, sizeof (event));
  event.event_type = NAT_HA_REFRESH;
  event.out_addr = out_addr->as_u32;
  event.out_port = out_port;
  event.eh_addr = eh_addr->as_u32;
  event.eh_port = eh_port;
  event.fib_index = clib_host_to_net_u32 (fib_index);
  event.protocol = proto;
  event.total_pkts = clib_host_to_net_u32 (total_pkts);
  event.total_bytes = clib_host_to_net_u64 (total_bytes);
  nat_ha_event_add (&event, 0, thread_index, 0);
}

/* per thread process waiting for interrupt */
static uword
nat_ha_worker_fn (vlib_main_t * vm, vlib_node_runtime_t * rt,
		  vlib_frame_t * f)
{
  u32 thread_index = vm->thread_index;
  /* flush HA NAT data under construction */
  nat_ha_event_add (0, 1, thread_index, 0);
  /* scan if we need to resend some non-ACKed data */
  nat_ha_resend_scan (vlib_time_now (vm), thread_index);
  return 0;
}

/* *INDENT-OFF* */
VLIB_REGISTER_NODE (nat_ha_worker_node) = {
    .function = nat_ha_worker_fn,
    .type = VLIB_NODE_TYPE_INPUT,
    .state = VLIB_NODE_STATE_INTERRUPT,
    .name = "nat-ha-worker",
};
/* *INDENT-ON* */

/* periodically send interrupt to each thread */
static uword
nat_ha_process (vlib_main_t * vm, vlib_node_runtime_t * rt, vlib_frame_t * f)
{
  nat_ha_main_t *ha = &nat_ha_main;
  uword event_type;
  uword *event_data = 0;
  u32 ti;

  vlib_process_wait_for_event (vm);
  event_type = vlib_process_get_events (vm, &event_data);
  if (event_type)
    nat_elog_info ("nat-ha-process: bogus kickoff event received");
  vec_reset_length (event_data);

  while (1)
    {
      vlib_process_wait_for_event_or_clock (vm, 1.0);
      event_type = vlib_process_get_events (vm, &event_data);
      vec_reset_length (event_data);
      for (ti = 0; ti < vec_len (vlib_mains); ti++)
	{
	  if (ti >= vec_len (ha->per_thread_data))
	    continue;

	  vlib_node_set_interrupt_pending (vlib_mains[ti],
					   nat_ha_worker_node.index);
	}
    }

  return 0;
}

/* *INDENT-OFF* */
VLIB_REGISTER_NODE (nat_ha_process_node) = {
    .function = nat_ha_process,
    .type = VLIB_NODE_TYPE_PROCESS,
    .name = "nat-ha-process",
};
/* *INDENT-ON* */

void
nat_ha_get_resync_status (u8 * in_resync, u32 * resync_ack_missed)
{
  nat_ha_main_t *ha = &nat_ha_main;

  *in_resync = ha->in_resync;
  *resync_ack_missed = ha->resync_ack_missed;
}

typedef struct
{
  ip4_address_t addr;
  u32 event_count;
} nat_ha_trace_t;

static u8 *
format_nat_ha_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 *);
  nat_ha_trace_t *t = va_arg (*args, nat_ha_trace_t *);

  s =
    format (s, "nat-ha: %u events from %U", t->event_count,
	    format_ip4_address, &t->addr);

  return s;
}

typedef enum
{
  NAT_HA_NEXT_IP4_LOOKUP,
  NAT_HA_NEXT_DROP,
  NAT_HA_N_NEXT,
} nat_ha_next_t;

#define foreach_nat_ha_error   \
_(PROCESSED, "pkts-processed") \
_(BAD_VERSION, "bad-version")

typedef enum
{
#define _(sym, str) NAT_HA_ERROR_##sym,
  foreach_nat_ha_error
#undef _
    NAT_HA_N_ERROR,
} nat_ha_error_t;

static char *nat_ha_error_strings[] = {
#define _(sym, str) str,
  foreach_nat_ha_error
#undef _
};

/* process received HA NAT protocol messages */
static uword
nat_ha_node_fn (vlib_main_t * vm, vlib_node_runtime_t * node,
		vlib_frame_t * frame)
{
  u32 n_left_from, *from, next_index, *to_next;
  f64 now = vlib_time_now (vm);
  u32 thread_index = vm->thread_index;
  u32 pkts_processed = 0;
  ip4_main_t *i4m = &ip4_main;
  u8 host_config_ttl = i4m->host_config.ttl;
  nat_ha_main_t *ha = &nat_ha_main;

  from = vlib_frame_vector_args (frame);
  n_left_from = frame->n_vectors;
  next_index = node->cached_next_index;

  while (n_left_from > 0)
    {
      u32 n_left_to_next;

      vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next);

      while (n_left_from > 0 && n_left_to_next > 0)
	{
	  u32 bi0, next0, src_addr0, dst_addr0;;
	  vlib_buffer_t *b0;
	  nat_ha_message_header_t *h0;
	  nat_ha_event_t *e0;
	  u16 event_count0, src_port0, dst_port0, old_len0;
	  ip4_header_t *ip0;
	  udp_header_t *udp0;
	  ip_csum_t sum0;

	  bi0 = from[0];
	  to_next[0] = bi0;
	  from += 1;
	  to_next += 1;
	  n_left_from -= 1;
	  n_left_to_next -= 1;

	  b0 = vlib_get_buffer (vm, bi0);
	  h0 = vlib_buffer_get_current (b0);
	  vlib_buffer_advance (b0, -sizeof (*udp0));
	  udp0 = vlib_buffer_get_current (b0);
	  vlib_buffer_advance (b0, -sizeof (*ip0));
	  ip0 = vlib_buffer_get_current (b0);

	  next0 = NAT_HA_NEXT_DROP;

	  if (h0->version != NAT_HA_VERSION)
	    {
	      b0->error = node->errors[NAT_HA_ERROR_BAD_VERSION];
	      goto done0;
	    }

	  event_count0 = clib_net_to_host_u16 (h0->count);
	  /* ACK for previously send data */
	  if (!event_count0 && (h0->flags & NAT_HA_FLAG_ACK))
	    {
	      nat_ha_ack_recv (h0->sequence_number, thread_index);
	      b0->error = node->errors[NAT_HA_ERROR_PROCESSED];
	      goto done0;
	    }

	  e0 = (nat_ha_event_t *) (h0 + 1);

	  /* process each event */
	  while (event_count0)
	    {
	      nat_ha_event_process (e0, now, thread_index);
	      event_count0--;
	      e0 = (nat_ha_event_t *) ((u8 *) e0 + sizeof (nat_ha_event_t));
	    }

	  next0 = NAT_HA_NEXT_IP4_LOOKUP;
	  pkts_processed++;

	  /* reply with ACK */
	  b0->current_length = sizeof (*ip0) + sizeof (*udp0) + sizeof (*h0);

	  src_addr0 = ip0->src_address.data_u32;
	  dst_addr0 = ip0->dst_address.data_u32;
	  ip0->src_address.data_u32 = dst_addr0;
	  ip0->dst_address.data_u32 = src_addr0;
	  old_len0 = ip0->length;
	  ip0->length = clib_host_to_net_u16 (b0->current_length);

	  sum0 = ip0->checksum;
	  sum0 = ip_csum_update (sum0, ip0->ttl, host_config_ttl,
				 ip4_header_t, ttl);
	  ip0->ttl = host_config_ttl;
	  sum0 =
	    ip_csum_update (sum0, old_len0, ip0->length, ip4_header_t,
			    length);
	  ip0->checksum = ip_csum_fold (sum0);

	  udp0->checksum = 0;
	  src_port0 = udp0->src_port;
	  dst_port0 = udp0->dst_port;
	  udp0->src_port = dst_port0;
	  udp0->dst_port = src_port0;
	  udp0->length =
	    clib_host_to_net_u16 (b0->current_length - sizeof (*ip0));

	  h0->flags = NAT_HA_FLAG_ACK;
	  h0->count = 0;
	  vlib_increment_simple_counter (&ha->counters
					 [NAT_HA_COUNTER_SEND_ACK],
					 thread_index, 0, 1);

	done0:
	  if (PREDICT_FALSE ((node->flags & VLIB_NODE_FLAG_TRACE)
			     && (b0->flags & VLIB_BUFFER_IS_TRACED)))
	    {
	      nat_ha_trace_t *t = vlib_add_trace (vm, node, b0, sizeof (*t));
	      ip4_header_t *ip =
		(void *) (b0->data + vnet_buffer (b0)->l3_hdr_offset);
	      t->event_count = clib_net_to_host_u16 (h0->count);
	      t->addr.as_u32 = ip->src_address.data_u32;
	    }

	  vlib_validate_buffer_enqueue_x1 (vm, node, next_index,
					   to_next, n_left_to_next,
					   bi0, next0);
	}

      vlib_put_next_frame (vm, node, next_index, n_left_to_next);
    }

  vlib_node_increment_counter (vm, nat_ha_node.index,
			       NAT_HA_ERROR_PROCESSED, pkts_processed);

  return frame->n_vectors;
}

/* *INDENT-OFF* */
VLIB_REGISTER_NODE (nat_ha_node) = {
  .function = nat_ha_node_fn,
  .name = "nat-ha",
  .vector_size = sizeof (u32),
  .format_trace = format_nat_ha_trace,
  .type = VLIB_NODE_TYPE_INTERNAL,
  .n_errors = ARRAY_LEN (nat_ha_error_strings),
  .error_strings = nat_ha_error_strings,
  .n_next_nodes = NAT_HA_N_NEXT,
  .next_nodes = {
     [NAT_HA_NEXT_IP4_LOOKUP] = "ip4-lookup",
     [NAT_HA_NEXT_DROP] = "error-drop",
  },
};
/* *INDENT-ON* */

typedef struct
{
  u32 next_worker_index;
  u8 in2out;
} nat_ha_handoff_trace_t;

#define foreach_nat_ha_handoff_error  \
_(CONGESTION_DROP, "congestion drop") \
_(SAME_WORKER, "same worker")         \
_(DO_HANDOFF, "do handoff")

typedef enum
{
#define _(sym,str) NAT_HA_HANDOFF_ERROR_##sym,
  foreach_nat_ha_handoff_error
#undef _
    NAT44_HANDOFF_N_ERROR,
} nat_ha_handoff_error_t;

static char *nat_ha_handoff_error_strings[] = {
#define _(sym,string) string,
  foreach_nat_ha_handoff_error
#undef _
};

static u8 *
format_nat_ha_handoff_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 *);
  nat_ha_handoff_trace_t *t = va_arg (*args, nat_ha_handoff_trace_t *);

  s =
    format (s, "NAT_HA_WORKER_HANDOFF: next-worker %d", t->next_worker_index);

  return s;
}

/* do worker handoff based on thread_index in NAT HA protcol header */
static uword
nat_ha_handoff_node_fn (vlib_main_t * vm, vlib_node_runtime_t * node,
			vlib_frame_t * frame)
{
  nat_ha_main_t *ha = &nat_ha_main;
  vlib_buffer_t *bufs[VLIB_FRAME_SIZE], **b;
  u32 n_enq, n_left_from, *from;
  u16 thread_indices[VLIB_FRAME_SIZE], *ti;
  u32 thread_index = vm->thread_index;
  u32 do_handoff = 0, same_worker = 0;

  from = vlib_frame_vector_args (frame);
  n_left_from = frame->n_vectors;
  vlib_get_buffers (vm, from, bufs, n_left_from);

  b = bufs;
  ti = thread_indices;

  while (n_left_from > 0)
    {
      nat_ha_message_header_t *h0;

      h0 = vlib_buffer_get_current (b[0]);
      ti[0] = clib_net_to_host_u32 (h0->thread_index);

      if (ti[0] != thread_index)
	do_handoff++;
      else
	same_worker++;

      if (PREDICT_FALSE ((node->flags & VLIB_NODE_FLAG_TRACE)
			 && (b[0]->flags & VLIB_BUFFER_IS_TRACED)))
	{
	  nat_ha_handoff_trace_t *t =
	    vlib_add_trace (vm, node, b[0], sizeof (*t));
	  t->next_worker_index = ti[0];
	}

      n_left_from -= 1;
      ti += 1;
      b += 1;
    }

  n_enq =
    vlib_buffer_enqueue_to_thread (vm, ha->fq_index, from, thread_indices,
				   frame->n_vectors, 1);

  if (n_enq < frame->n_vectors)
    vlib_node_increment_counter (vm, node->node_index,
				 NAT_HA_HANDOFF_ERROR_CONGESTION_DROP,
				 frame->n_vectors - n_enq);
  vlib_node_increment_counter (vm, node->node_index,
			       NAT_HA_HANDOFF_ERROR_SAME_WORKER, same_worker);
  vlib_node_increment_counter (vm, node->node_index,
			       NAT_HA_HANDOFF_ERROR_DO_HANDOFF, do_handoff);
  return frame->n_vectors;
}

/* *INDENT-OFF* */
VLIB_REGISTER_NODE (nat_ha_handoff_node) = {
  .function = nat_ha_handoff_node_fn,
  .name = "nat-ha-handoff",
  .vector_size = sizeof (u32),
  .format_trace = format_nat_ha_handoff_trace,
  .type = VLIB_NODE_TYPE_INTERNAL,
  .n_errors = ARRAY_LEN(nat_ha_handoff_error_strings),
  .error_strings = nat_ha_handoff_error_strings,
  .n_next_nodes = 1,
  .next_nodes = {
    [0] = "error-drop",
  },
};
/* *INDENT-ON* */

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