summaryrefslogtreecommitdiffstats
path: root/src/vnet/session/session.c
blob: 533a6c22ae02fc4a7dcfdbdbc66074b86074d3fa (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
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
/*
 * Copyright (c) 2017 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.
 */
/**
 * @file
 * @brief Session and session manager
 */

#include <vnet/session/session.h>
#include <vnet/session/session_debug.h>
#include <vnet/session/application.h>
#include <vlibmemory/api.h>
#include <vnet/dpo/load_balance.h>
#include <vnet/fib/ip4_fib.h>
#include <vnet/tcp/tcp.h>

session_manager_main_t session_manager_main;
extern transport_proto_vft_t *tp_vfts;

int
stream_session_create_i (segment_manager_t * sm, transport_connection_t * tc,
			 u8 alloc_fifos, stream_session_t ** ret_s)
{
  session_manager_main_t *smm = &session_manager_main;
  svm_fifo_t *server_rx_fifo = 0, *server_tx_fifo = 0;
  u32 fifo_segment_index;
  u32 pool_index;
  stream_session_t *s;
  u64 value;
  u32 thread_index = tc->thread_index;
  int rv;

  ASSERT (thread_index == vlib_get_thread_index ());

  /* Create the session */
  pool_get_aligned (smm->sessions[thread_index], s, CLIB_CACHE_LINE_BYTES);
  memset (s, 0, sizeof (*s));
  pool_index = s - smm->sessions[thread_index];

  /* Allocate fifos */
  if (alloc_fifos)
    {
      if ((rv = segment_manager_alloc_session_fifos (sm, &server_rx_fifo,
						     &server_tx_fifo,
						     &fifo_segment_index)))
	{
	  pool_put (smm->sessions[thread_index], s);
	  return rv;
	}
      /* Initialize backpointers */
      server_rx_fifo->master_session_index = pool_index;
      server_rx_fifo->master_thread_index = thread_index;

      server_tx_fifo->master_session_index = pool_index;
      server_tx_fifo->master_thread_index = thread_index;

      s->server_rx_fifo = server_rx_fifo;
      s->server_tx_fifo = server_tx_fifo;
      s->svm_segment_index = fifo_segment_index;
    }

  /* Initialize state machine, such as it is... */
  s->session_type = session_type_from_proto_and_ip (tc->transport_proto,
						    tc->is_ip4);
  s->session_state = SESSION_STATE_CONNECTING;
  s->thread_index = thread_index;
  s->session_index = pool_index;

  /* Attach transport to session */
  s->connection_index = tc->c_index;

  /* Attach session to transport */
  tc->s_index = s->session_index;

  /* Add to the main lookup table */
  value = stream_session_handle (s);
  stream_session_table_add_for_tc (tc, value);

  *ret_s = s;

  return 0;
}

/** Enqueue buffer chain tail */
always_inline int
session_enqueue_chain_tail (stream_session_t * s, vlib_buffer_t * b,
			    u32 offset, u8 is_in_order)
{
  vlib_buffer_t *chain_b;
  u32 chain_bi = b->next_buffer;
  vlib_main_t *vm = vlib_get_main ();
  u8 *data, len;
  u16 written = 0;
  int rv = 0;

  do
    {
      chain_b = vlib_get_buffer (vm, chain_bi);
      data = vlib_buffer_get_current (chain_b);
      len = chain_b->current_length;
      if (is_in_order)
	{
	  rv = svm_fifo_enqueue_nowait (s->server_rx_fifo, len, data);
	  if (rv < len)
	    {
	      return (rv > 0) ? (written + rv) : written;
	    }
	  written += rv;
	}
      else
	{
	  rv = svm_fifo_enqueue_with_offset (s->server_rx_fifo, offset, len,
					     data);
	  if (rv)
	    return -1;
	  offset += len;
	}
    }
  while ((chain_bi = (chain_b->flags & VLIB_BUFFER_NEXT_PRESENT)
	  ? chain_b->next_buffer : 0));

  if (is_in_order)
    return written;

  return 0;
}

/*
 * Enqueue data for delivery to session peer. Does not notify peer of enqueue
 * event but on request can queue notification events for later delivery by
 * calling stream_server_flush_enqueue_events().
 *
 * @param tc Transport connection which is to be enqueued data
 * @param b Buffer to be enqueued
 * @param offset Offset at which to start enqueueing if out-of-order
 * @param queue_event Flag to indicate if peer is to be notified or if event
 *                    is to be queued. The former is useful when more data is
 *                    enqueued and only one event is to be generated.
 * @param is_in_order Flag to indicate if data is in order
 * @return Number of bytes enqueued or a negative value if enqueueing failed.
 */
int
stream_session_enqueue_data (transport_connection_t * tc, vlib_buffer_t * b,
			     u32 offset, u8 queue_event, u8 is_in_order)
{
  stream_session_t *s;
  int enqueued = 0, rv;

  s = stream_session_get (tc->s_index, tc->thread_index);

  if (is_in_order)
    {
      enqueued =
	svm_fifo_enqueue_nowait (s->server_rx_fifo, b->current_length,
				 vlib_buffer_get_current (b));
      if (PREDICT_FALSE
	  ((b->flags & VLIB_BUFFER_NEXT_PRESENT) && enqueued > 0))
	{
	  rv = session_enqueue_chain_tail (s, b, 0, 1);
	  if (rv <= 0)
	    return enqueued;
	  enqueued += rv;
	}
    }
  else
    {
      rv = svm_fifo_enqueue_with_offset (s->server_rx_fifo, offset,
					 b->current_length,
					 vlib_buffer_get_current (b));
      if (PREDICT_FALSE ((b->flags & VLIB_BUFFER_NEXT_PRESENT) && !rv))
	rv = session_enqueue_chain_tail (s, b, offset + b->current_length, 0);
      if (rv)
	return -1;
    }

  if (queue_event)
    {
      /* Queue RX event on this fifo. Eventually these will need to be flushed
       * by calling stream_server_flush_enqueue_events () */
      session_manager_main_t *smm = vnet_get_session_manager_main ();
      u32 thread_index = s->thread_index;
      u32 my_enqueue_epoch = smm->current_enqueue_epoch[thread_index];

      if (s->enqueue_epoch != my_enqueue_epoch)
	{
	  s->enqueue_epoch = my_enqueue_epoch;
	  vec_add1 (smm->session_indices_to_enqueue_by_thread[thread_index],
		    s - smm->sessions[thread_index]);
	}
    }

  if (is_in_order)
    return enqueued;

  return 0;
}

/** Check if we have space in rx fifo to push more bytes */
u8
stream_session_no_space (transport_connection_t * tc, u32 thread_index,
			 u16 data_len)
{
  stream_session_t *s = stream_session_get (tc->s_index, thread_index);

  if (PREDICT_FALSE (s->session_state != SESSION_STATE_READY))
    return 1;

  if (data_len > svm_fifo_max_enqueue (s->server_rx_fifo))
    return 1;

  return 0;
}

u32
stream_session_tx_fifo_max_dequeue (transport_connection_t * tc)
{
  stream_session_t *s = stream_session_get (tc->s_index, tc->thread_index);
  if (s->session_state != SESSION_STATE_READY)
    return 0;
  return svm_fifo_max_dequeue (s->server_tx_fifo);
}

int
stream_session_peek_bytes (transport_connection_t * tc, u8 * buffer,
			   u32 offset, u32 max_bytes)
{
  stream_session_t *s = stream_session_get (tc->s_index, tc->thread_index);
  return svm_fifo_peek (s->server_tx_fifo, offset, max_bytes, buffer);
}

u32
stream_session_dequeue_drop (transport_connection_t * tc, u32 max_bytes)
{
  stream_session_t *s = stream_session_get (tc->s_index, tc->thread_index);
  return svm_fifo_dequeue_drop (s->server_tx_fifo, max_bytes);
}

/**
 * Notify session peer that new data has been enqueued.
 *
 * @param s Stream session for which the event is to be generated.
 * @param block Flag to indicate if call should block if event queue is full.
 *
 * @return 0 on succes or negative number if failed to send notification.
 */
static int
stream_session_enqueue_notify (stream_session_t * s, u8 block)
{
  application_t *app;
  session_fifo_event_t evt;
  unix_shared_memory_queue_t *q;
  static u32 serial_number;

  if (PREDICT_FALSE (s->session_state == SESSION_STATE_CLOSED))
    return 0;

  /* Get session's server */
  app = application_get_if_valid (s->app_index);

  if (PREDICT_FALSE (app == 0))
    {
      clib_warning ("invalid s->app_index = %d", s->app_index);
      return 0;
    }

  /* Built-in server? Hand event to the callback... */
  if (app->cb_fns.builtin_server_rx_callback)
    return app->cb_fns.builtin_server_rx_callback (s);

  /* If no event, send one */
  if (svm_fifo_set_event (s->server_rx_fifo))
    {
      /* Fabricate event */
      evt.fifo = s->server_rx_fifo;
      evt.event_type = FIFO_EVENT_APP_RX;
      evt.event_id = serial_number++;

      /* Add event to server's event queue */
      q = app->event_queue;

      /* Based on request block (or not) for lack of space */
      if (block || PREDICT_TRUE (q->cursize < q->maxsize))
	unix_shared_memory_queue_add (app->event_queue, (u8 *) & evt,
				      0 /* do wait for mutex */ );
      else
	{
	  clib_warning ("fifo full");
	  return -1;
	}
    }

  /* *INDENT-OFF* */
  SESSION_EVT_DBG(SESSION_EVT_ENQ, s, ({
      ed->data[0] = evt.event_id;
      ed->data[1] = svm_fifo_max_dequeue (s->server_rx_fifo);
  }));
  /* *INDENT-ON* */

  return 0;
}

/**
 * Flushes queue of sessions that are to be notified of new data
 * enqueued events.
 *
 * @param thread_index Thread index for which the flush is to be performed.
 * @return 0 on success or a positive number indicating the number of
 *         failures due to API queue being full.
 */
int
session_manager_flush_enqueue_events (u32 thread_index)
{
  session_manager_main_t *smm = &session_manager_main;
  u32 *session_indices_to_enqueue;
  int i, errors = 0;

  session_indices_to_enqueue =
    smm->session_indices_to_enqueue_by_thread[thread_index];

  for (i = 0; i < vec_len (session_indices_to_enqueue); i++)
    {
      stream_session_t *s0;

      /* Get session */
      s0 = stream_session_get_if_valid (session_indices_to_enqueue[i],
					thread_index);
      if (s0 == 0 || stream_session_enqueue_notify (s0, 0 /* don't block */ ))
	{
	  errors++;
	}
    }

  vec_reset_length (session_indices_to_enqueue);

  smm->session_indices_to_enqueue_by_thread[thread_index] =
    session_indices_to_enqueue;

  /* Increment enqueue epoch for next round */
  smm->current_enqueue_epoch[thread_index]++;

  return errors;
}

/**
 * Init fifo tail and head pointers
 *
 * Useful if transport uses absolute offsets for tracking ooo segments.
 */
void
stream_session_init_fifos_pointers (transport_connection_t * tc,
				    u32 rx_pointer, u32 tx_pointer)
{
  stream_session_t *s;
  s = stream_session_get (tc->s_index, tc->thread_index);
  svm_fifo_init_pointers (s->server_rx_fifo, rx_pointer);
  svm_fifo_init_pointers (s->server_tx_fifo, tx_pointer);
}

int
stream_session_connect_notify (transport_connection_t * tc, u8 is_fail)
{
  application_t *app;
  stream_session_t *new_s = 0;
  u64 handle;
  u32 api_context = 0;
  int error = 0;

  handle = stream_session_half_open_lookup_handle (&tc->lcl_ip, &tc->rmt_ip,
						   tc->lcl_port, tc->rmt_port,
						   tc->transport_proto);
  if (handle == HALF_OPEN_LOOKUP_INVALID_VALUE)
    {
      clib_warning ("This can't be good!");
      return -1;
    }

  /* Get the app's index from the handle we stored when opening connection */
  app = application_get (handle >> 32);
  api_context = tc->s_index;

  if (!is_fail)
    {
      segment_manager_t *sm;
      u8 alloc_fifos;
      sm = application_get_connect_segment_manager (app);
      alloc_fifos = application_is_proxy (app);
      /* Create new session (svm segments are allocated if needed) */
      if (stream_session_create_i (sm, tc, alloc_fifos, &new_s))
	{
	  is_fail = 1;
	  error = -1;
	}
      else
	new_s->app_index = app->index;
    }

  /* Notify client application */
  if (app->cb_fns.session_connected_callback (app->index, api_context, new_s,
					      is_fail))
    {
      clib_warning ("failed to notify app");
      if (!is_fail)
	stream_session_disconnect (new_s);
    }
  else
    {
      if (!is_fail)
	new_s->session_state = SESSION_STATE_READY;
    }

  /* Cleanup session lookup */
  stream_session_half_open_table_del (tc);

  return error;
}

void
stream_session_accept_notify (transport_connection_t * tc)
{
  application_t *server;
  stream_session_t *s;

  s = stream_session_get (tc->s_index, tc->thread_index);
  server = application_get (s->app_index);
  server->cb_fns.session_accept_callback (s);
}

/**
 * Notification from transport that connection is being closed.
 *
 * A disconnect is sent to application but state is not removed. Once
 * disconnect is acknowledged by application, session disconnect is called.
 * Ultimately this leads to close being called on transport (passive close).
 */
void
stream_session_disconnect_notify (transport_connection_t * tc)
{
  application_t *server;
  stream_session_t *s;

  s = stream_session_get (tc->s_index, tc->thread_index);
  server = application_get (s->app_index);
  server->cb_fns.session_disconnect_callback (s);
}

/**
 * Cleans up session and associated app if needed.
 */
void
stream_session_delete (stream_session_t * s)
{
  session_manager_main_t *smm = vnet_get_session_manager_main ();
  int rv;

  /* Delete from the main lookup table. */
  if ((rv = stream_session_table_del (s)))
    clib_warning ("hash delete error, rv %d", rv);

  /* Cleanup fifo segments */
  segment_manager_dealloc_fifos (s->svm_segment_index, s->server_rx_fifo,
				 s->server_tx_fifo);

  pool_put (smm->sessions[s->thread_index], s);
  if (CLIB_DEBUG)
    memset (s, 0xFA, sizeof (*s));
}

/**
 * Notification from transport that connection is being deleted
 *
 * This should be called only on previously fully established sessions. For
 * instance failed connects should call stream_session_connect_notify and
 * indicate that the connect has failed.
 */
void
stream_session_delete_notify (transport_connection_t * tc)
{
  stream_session_t *s;

  /* App might've been removed already */
  s = stream_session_get_if_valid (tc->s_index, tc->thread_index);
  if (!s)
    {
      return;
    }
  stream_session_delete (s);
}

/**
 * Notify application that connection has been reset.
 */
void
stream_session_reset_notify (transport_connection_t * tc)
{
  stream_session_t *s;
  application_t *app;
  s = stream_session_get (tc->s_index, tc->thread_index);

  app = application_get (s->app_index);
  app->cb_fns.session_reset_callback (s);
}

/**
 * Accept a stream session. Optionally ping the server by callback.
 */
int
stream_session_accept (transport_connection_t * tc, u32 listener_index,
		       u8 sst, u8 notify)
{
  application_t *server;
  stream_session_t *s, *listener;
  segment_manager_t *sm;

  int rv;

  /* Find the server */
  listener = listen_session_get (sst, listener_index);
  server = application_get (listener->app_index);

  sm = application_get_listen_segment_manager (server, listener);
  if ((rv = stream_session_create_i (sm, tc, 1, &s)))
    return rv;

  s->app_index = server->index;
  s->listener_index = listener_index;
  s->session_state = SESSION_STATE_ACCEPTING;

  /* Shoulder-tap the server */
  if (notify)
    {
      server->cb_fns.session_accept_callback (s);
    }

  return 0;
}

/**
 * Ask transport to open connection to remote transport endpoint.
 *
 * Stores handle for matching request with reply since the call can be
 * asynchronous. For instance, for TCP the 3-way handshake must complete
 * before reply comes. Session is only created once connection is established.
 *
 * @param app_index Index of the application requesting the connect
 * @param st Session type requested.
 * @param tep Remote transport endpoint
 * @param res Resulting transport connection .
 */
int
stream_session_open (u32 app_index, session_type_t st,
		     transport_endpoint_t * rmt,
		     transport_connection_t ** res)
{
  transport_connection_t *tc;
  int rv;
  u64 handle;

  rv = tp_vfts[st].open (rmt);
  if (rv < 0)
    {
      clib_warning ("Transport failed to open connection.");
      return VNET_API_ERROR_SESSION_CONNECT_FAIL;
    }

  tc = tp_vfts[st].get_half_open ((u32) rv);

  /* Save app and tc index. The latter is needed to help establish the
   * connection while the former is needed when the connect notify comes
   * and we have to notify the external app */
  handle = (((u64) app_index) << 32) | (u64) tc->c_index;

  /* Add to the half-open lookup table */
  stream_session_half_open_table_add (tc, handle);

  *res = tc;

  return 0;
}

/**
 * Ask transport to listen on local transport endpoint.
 *
 * @param s Session for which listen will be called. Note that unlike
 * 	    established sessions, listen sessions are not associated to a
 * 	    thread.
 * @param tep Local endpoint to be listened on.
 */
int
stream_session_listen (stream_session_t * s, transport_endpoint_t * tep)
{
  transport_connection_t *tc;
  u32 tci;

  /* Transport bind/listen  */
  tci = tp_vfts[s->session_type].bind (s->session_index, tep);

  if (tci == (u32) ~ 0)
    return -1;

  /* Attach transport to session */
  s->connection_index = tci;
  tc = tp_vfts[s->session_type].get_listener (tci);

  /* Weird but handle it ... */
  if (tc == 0)
    return -1;

  /* Add to the main lookup table */
  stream_session_table_add_for_tc (tc, s->session_index);

  return 0;
}

/**
 * Ask transport to stop listening on local transport endpoint.
 *
 * @param s Session to stop listening on. It must be in state LISTENING.
 */
int
stream_session_stop_listen (stream_session_t * s)
{
  transport_connection_t *tc;

  if (s->session_state != SESSION_STATE_LISTENING)
    {
      clib_warning ("not a listening session");
      return -1;
    }

  tc = tp_vfts[s->session_type].get_listener (s->connection_index);
  if (!tc)
    {
      clib_warning ("no transport");
      return VNET_API_ERROR_ADDRESS_NOT_IN_USE;
    }

  stream_session_table_del_for_tc (tc);
  tp_vfts[s->session_type].unbind (s->connection_index);
  return 0;
}

void
session_send_session_evt_to_thread (u64 session_handle,
				    fifo_event_type_t evt_type,
				    u32 thread_index)
{
  static u16 serial_number = 0;
  session_fifo_event_t evt;
  unix_shared_memory_queue_t *q;

  /* Fabricate event */
  evt.session_handle = session_handle;
  evt.event_type = evt_type;
  evt.event_id = serial_number++;

  q = session_manager_get_vpp_event_queue (thread_index);

  /* Based on request block (or not) for lack of space */
  if (PREDICT_TRUE (q->cursize < q->maxsize))
    {
      if (unix_shared_memory_queue_add (q, (u8 *) & evt,
					1 /* do wait for mutex */ ))
	{
	  clib_warning ("failed to enqueue evt");
	}
    }
  else
    {
      clib_warning ("queue full");
      return;
    }
}

/**
 * Disconnect session and propagate to transport. This should eventually
 * result in a delete notification that allows us to cleanup session state.
 * Called for both active/passive disconnects.
 *
 * Should be called from the session's thread.
 */
void
stream_session_disconnect (stream_session_t * s)
{
  s->session_state = SESSION_STATE_CLOSED;
  tp_vfts[s->session_type].close (s->connection_index, s->thread_index);
}

/**
 * Cleanup transport and session state.
 *
 * Notify transport of the cleanup, wait for a delete notify to actually
 * remove the session state.
 */
void
stream_session_cleanup (stream_session_t * s)
{
  int rv;

  s->session_state = SESSION_STATE_CLOSED;

  /* Delete from the main lookup table to avoid more enqueues */
  rv = stream_session_table_del (s);
  if (rv)
    clib_warning ("hash delete error, rv %d", rv);

  tp_vfts[s->session_type].cleanup (s->connection_index, s->thread_index);
}

/**
 * Allocate vpp event queue (once) per worker thread
 */
void
session_vpp_event_queue_allocate (session_manager_main_t * smm,
				  u32 thread_index)
{
  api_main_t *am = &api_main;
  void *oldheap;
  u32 event_queue_length = 2048;

  if (smm->vpp_event_queues[thread_index] == 0)
    {
      /* Allocate event fifo in the /vpe-api shared-memory segment */
      oldheap = svm_push_data_heap (am->vlib_rp);

      if (smm->configured_event_queue_length)
	event_queue_length = smm->configured_event_queue_length;

      smm->vpp_event_queues[thread_index] =
	unix_shared_memory_queue_init
	(event_queue_length,
	 sizeof (session_fifo_event_t), 0 /* consumer pid */ ,
	 0 /* (do not) send signal when queue non-empty */ );

      svm_pop_heap (oldheap);
    }
}

session_type_t
session_type_from_proto_and_ip (transport_proto_t proto, u8 is_ip4)
{
  if (proto == TRANSPORT_PROTO_TCP)
    {
      if (is_ip4)
	return SESSION_TYPE_IP4_TCP;
      else
	return SESSION_TYPE_IP6_TCP;
    }
  else
    {
      if (is_ip4)
	return SESSION_TYPE_IP4_UDP;
      else
	return SESSION_TYPE_IP6_UDP;
    }

  return SESSION_N_TYPES;
}

static clib_error_t *
session_manager_main_enable (vlib_main_t * vm)
{
  session_manager_main_t *smm = &session_manager_main;
  vlib_thread_main_t *vtm = vlib_get_thread_main ();
  u32 num_threads;
  u32 preallocated_sessions_per_worker;
  int i;

  num_threads = 1 /* main thread */  + vtm->n_threads;

  if (num_threads < 1)
    return clib_error_return (0, "n_thread_stacks not set");

  /* $$$ config parameters */
  svm_fifo_segment_init (0x200000000ULL /* first segment base VA */ ,
			 20 /* timeout in seconds */ );

  /* configure per-thread ** vectors */
  vec_validate (smm->sessions, num_threads - 1);
  vec_validate (smm->session_indices_to_enqueue_by_thread, num_threads - 1);
  vec_validate (smm->tx_buffers, num_threads - 1);
  vec_validate (smm->pending_event_vector, num_threads - 1);
  vec_validate (smm->free_event_vector, num_threads - 1);
  vec_validate (smm->current_enqueue_epoch, num_threads - 1);
  vec_validate (smm->vpp_event_queues, num_threads - 1);

  for (i = 0; i < num_threads; i++)
    {
      vec_validate (smm->free_event_vector[i], 0);
      _vec_len (smm->free_event_vector[i]) = 0;
      vec_validate (smm->pending_event_vector[i], 0);
      _vec_len (smm->pending_event_vector[i]) = 0;
    }

#if SESSION_DBG
  vec_validate (smm->last_event_poll_by_thread, num_threads - 1);
#endif

  /* Allocate vpp event queues */
  for (i = 0; i < vec_len (smm->vpp_event_queues); i++)
    session_vpp_event_queue_allocate (smm, i);

  /* Preallocate sessions */
  if (num_threads == 1)
    {
      for (i = 0; i < smm->preallocated_sessions; i++)
	{
	  stream_session_t *ss __attribute__ ((unused));
	  pool_get_aligned (smm->sessions[0], ss, CLIB_CACHE_LINE_BYTES);
	}

      for (i = 0; i < smm->preallocated_sessions; i++)
	pool_put_index (smm->sessions[0], i);
    }
  else
    {
      int j;
      preallocated_sessions_per_worker = smm->preallocated_sessions /
	(num_threads - 1);

      for (j = 1; j < num_threads; j++)
	{
	  for (i = 0; i < preallocated_sessions_per_worker; i++)
	    {
	      stream_session_t *ss __attribute__ ((unused));
	      pool_get_aligned (smm->sessions[j], ss, CLIB_CACHE_LINE_BYTES);
	    }
	  for (i = 0; i < preallocated_sessions_per_worker; i++)
	    pool_put_index (smm->sessions[j], i);
	}
    }

  session_lookup_init ();

  smm->is_enabled = 1;

  /* Enable TCP transport */
  vnet_tcp_enable_disable (vm, 1);

  return 0;
}

void
session_node_enable_disable (u8 is_en)
{
  u8 state = is_en ? VLIB_NODE_STATE_POLLING : VLIB_NODE_STATE_DISABLED;
  /* *INDENT-OFF* */
  foreach_vlib_main (({
    vlib_node_set_state (this_vlib_main, session_queue_node.index,
                         state);
  }));
  /* *INDENT-ON* */
}

clib_error_t *
vnet_session_enable_disable (vlib_main_t * vm, u8 is_en)
{
  if (is_en)
    {
      if (session_manager_main.is_enabled)
	return 0;

      session_node_enable_disable (is_en);

      return session_manager_main_enable (vm);
    }
  else
    {
      session_manager_main.is_enabled = 0;
      session_node_enable_disable (is_en);
    }

  return 0;
}

clib_error_t *
session_manager_main_init (vlib_main_t * vm)
{
  session_manager_main_t *smm = &session_manager_main;
  smm->is_enabled = 0;
  return 0;
}

VLIB_INIT_FUNCTION (session_manager_main_init);

static clib_error_t *
session_config_fn (vlib_main_t * vm, unformat_input_t * input)
{
  session_manager_main_t *smm = &session_manager_main;
  u32 nitems;
  uword tmp;

  while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT)
    {
      if (unformat (input, "event-queue-length %d", &nitems))
	{
	  if (nitems >= 2048)
	    smm->configured_event_queue_length = nitems;
	  else
	    clib_warning ("event queue length %d too small, ignored", nitems);
	}
      else if (unformat (input, "preallocated-sessions %d",
			 &smm->preallocated_sessions))
	;
      else if (unformat (input, "v4-session-table-buckets %d",
			 &smm->configured_v4_session_table_buckets))
	;
      else if (unformat (input, "v4-halfopen-table-buckets %d",
			 &smm->configured_v4_halfopen_table_buckets))
	;
      else if (unformat (input, "v6-session-table-buckets %d",
			 &smm->configured_v6_session_table_buckets))
	;
      else if (unformat (input, "v6-halfopen-table-buckets %d",
			 &smm->configured_v6_halfopen_table_buckets))
	;
      else if (unformat (input, "v4-session-table-memory %U",
			 unformat_memory_size, &tmp))
	{
	  if (tmp >= 0x100000000)
	    return clib_error_return (0, "memory size %llx (%lld) too large",
				      tmp, tmp);
	  smm->configured_v4_session_table_memory = tmp;
	}
      else if (unformat (input, "v4-halfopen-table-memory %U",
			 unformat_memory_size, &tmp))
	{
	  if (tmp >= 0x100000000)
	    return clib_error_return (0, "memory size %llx (%lld) too large",
				      tmp, tmp);
	  smm->configured_v4_halfopen_table_memory = tmp;
	}
      else if (unformat (input, "v6-session-table-memory %U",
			 unformat_memory_size, &tmp))
	{
	  if (tmp >= 0x100000000)
	    return clib_error_return (0, "memory size %llx (%lld) too large",
				      tmp, tmp);
	  smm->configured_v6_session_table_memory = tmp;
	}
      else if (unformat (input, "v6-halfopen-table-memory %U",
			 unformat_memory_size, &tmp))
	{
	  if (tmp >= 0x100000000)
	    return clib_error_return (0, "memory size %llx (%lld) too large",
				      tmp, tmp);
	  smm->configured_v6_halfopen_table_memory = tmp;
	}
      else
	return clib_error_return (0, "unknown input `%U'",
				  format_unformat_error, input);
    }
  return 0;
}

VLIB_CONFIG_FUNCTION (session_config_fn, "session");

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