aboutsummaryrefslogtreecommitdiffstats
path: root/vpp/plugins/lb-plugin/lb/lb.c
blob: 1d9b987095bcb3eaf63843491bedd9f5bb3901f6 (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
/*
 * Copyright (c) 2016 Cisco and/or its affiliates.
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at:
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

#include <lb/lb.h>
#include <vnet/plugin/plugin.h>
#include <vnet/api_errno.h>

//GC runs at most once every so many seconds
#define LB_GARBAGE_RUN 60

//After so many seconds. It is assumed that inter-core race condition will not occur.
#define LB_CONCURRENCY_TIMEOUT 10

lb_main_t lb_main;

#define lb_get_writer_lock() do {} while(__sync_lock_test_and_set (lb_main.writer_lock, 1))
#define lb_put_writer_lock() lb_main.writer_lock[0] = 0

static void lb_as_stack (lb_as_t *as);


const static char * const lb_dpo_gre4_ip4[] = { "lb4-gre4" , NULL };
const static char * const lb_dpo_gre4_ip6[] = { "lb6-gre4" , NULL };
const static char* const * const lb_dpo_gre4_nodes[DPO_PROTO_NUM] =
    {
	[DPO_PROTO_IP4]  = lb_dpo_gre4_ip4,
	[DPO_PROTO_IP6]  = lb_dpo_gre4_ip6,
    };

const static char * const lb_dpo_gre6_ip4[] = { "lb4-gre6" , NULL };
const static char * const lb_dpo_gre6_ip6[] = { "lb6-gre6" , NULL };
const static char* const * const lb_dpo_gre6_nodes[DPO_PROTO_NUM] =
    {
	[DPO_PROTO_IP4]  = lb_dpo_gre6_ip4,
	[DPO_PROTO_IP6]  = lb_dpo_gre6_ip6,
    };

u32 lb_hash_time_now(vlib_main_t * vm)
{
  return (u32) (vlib_time_now(vm) + 10000);
}

u8 *format_lb_main (u8 * s, va_list * args)
{
  vlib_thread_main_t *tm = vlib_get_thread_main();
  lb_main_t *lbm = &lb_main;
  s = format(s, "lb_main");
  s = format(s, " ip4-src-address: %U \n", format_ip4_address, &lbm->ip4_src_address);
  s = format(s, " ip6-src-address: %U \n", format_ip6_address, &lbm->ip6_src_address);
  s = format(s, " #vips: %u\n", pool_elts(lbm->vips));
  s = format(s, " #ass: %u\n", pool_elts(lbm->ass) - 1);

  u32 cpu_index;
  for(cpu_index = 0; cpu_index < tm->n_vlib_mains; cpu_index++ ) {
    lb_hash_t *h = lbm->per_cpu[cpu_index].sticky_ht;
    if (h) {
      s = format(s, "core %d\n", cpu_index);
      s = format(s, "  timeout: %ds\n", h->timeout);
      s = format(s, "  usage: %d / %d\n", lb_hash_elts(h, lb_hash_time_now(vlib_get_main())),  lb_hash_size(h));
    }
  }

  return s;
}

static char *lb_vip_type_strings[] = {
    [LB_VIP_TYPE_IP6_GRE6] = "ip6-gre6",
    [LB_VIP_TYPE_IP6_GRE4] = "ip6-gre4",
    [LB_VIP_TYPE_IP4_GRE6] = "ip4-gre6",
    [LB_VIP_TYPE_IP4_GRE4] = "ip4-gre4",
};

u8 *format_lb_vip_type (u8 * s, va_list * args)
{
  lb_vip_type_t vipt = va_arg (*args, lb_vip_type_t);
  u32 i;
  for (i=0; i<LB_VIP_N_TYPES; i++)
    if (vipt == i)
      return format(s, lb_vip_type_strings[i]);
  return format(s, "_WRONG_TYPE_");
}

uword unformat_lb_vip_type (unformat_input_t * input, va_list * args)
{
  lb_vip_type_t *vipt = va_arg (*args, lb_vip_type_t *);
  u32 i;
  for (i=0; i<LB_VIP_N_TYPES; i++)
    if (unformat(input, lb_vip_type_strings[i])) {
      *vipt = i;
      return 1;
    }
  return 0;
}

u8 *format_lb_vip (u8 * s, va_list * args)
{
  lb_vip_t *vip = va_arg (*args, lb_vip_t *);
  return format(s, "%U %U new_size:%u #as:%u%s",
             format_lb_vip_type, vip->type,
             format_ip46_prefix, &vip->prefix, vip->plen, IP46_TYPE_ANY,
             vip->new_flow_table_mask + 1,
             pool_elts(vip->as_indexes),
             (vip->flags & LB_VIP_FLAGS_USED)?"":" removed");
}

u8 *format_lb_as (u8 * s, va_list * args)
{
  lb_as_t *as = va_arg (*args, lb_as_t *);
  return format(s, "%U %s", format_ip46_address,
		&as->address, IP46_TYPE_ANY,
		(as->flags & LB_AS_FLAGS_USED)?"used":"removed");
}

u8 *format_lb_vip_detailed (u8 * s, va_list * args)
{
  lb_main_t *lbm = &lb_main;
  lb_vip_t *vip = va_arg (*args, lb_vip_t *);
  uword indent = format_get_indent (s);

  s = format(s, "%U %U [%u] %U%s\n"
                   "%U  new_size:%u\n",
                  format_white_space, indent,
                  format_lb_vip_type, vip->type,
                  vip - lbm->vips, format_ip46_prefix, &vip->prefix, vip->plen, IP46_TYPE_ANY,
                  (vip->flags & LB_VIP_FLAGS_USED)?"":" removed",
                  format_white_space, indent,
                  vip->new_flow_table_mask + 1);

  //Print counters
  s = format(s, "%U  counters:\n",
             format_white_space, indent);
  u32 i;
  for (i=0; i<LB_N_VIP_COUNTERS; i++)
    s = format(s, "%U    %s: %d\n",
               format_white_space, indent,
               lbm->vip_counters[i].name,
               vlib_get_simple_counter(&lbm->vip_counters[i], vip - lbm->vips));


  s = format(s, "%U  #as:%u\n",
             format_white_space, indent,
             pool_elts(vip->as_indexes));

  //Let's count the buckets for each AS
  u32 *count = 0;
  vec_validate(count, pool_len(lbm->ass)); //Possibly big alloc for not much...
  lb_new_flow_entry_t *nfe;
  vec_foreach(nfe, vip->new_flow_table)
    count[nfe->as_index]++;

  lb_as_t *as;
  u32 *as_index;
  pool_foreach(as_index, vip->as_indexes, {
      as = &lbm->ass[*as_index];
      s = format(s, "%U    %U %d buckets   %d flows  dpo:%u %s\n",
                   format_white_space, indent,
                   format_ip46_address, &as->address, IP46_TYPE_ANY,
                   count[as - lbm->ass],
                   vlib_refcount_get(&lbm->as_refcount, as - lbm->ass),
                   as->dpo.dpoi_index,
                   (as->flags & LB_AS_FLAGS_USED)?"used":" removed");
  });

  vec_free(count);

  /*
  s = format(s, "%U  new flows table:\n", format_white_space, indent);
  lb_new_flow_entry_t *nfe;
  vec_foreach(nfe, vip->new_flow_table) {
    s = format(s, "%U    %d: %d\n", format_white_space, indent, nfe - vip->new_flow_table, nfe->as_index);
  }
  */
  return s;
}

typedef struct {
  u32 as_index;
  u32 last;
  u32 skip;
} lb_pseudorand_t;

static int lb_pseudorand_compare(void *a, void *b)
{
  lb_as_t *asa, *asb;
  lb_main_t *lbm = &lb_main;
  asa = &lbm->ass[((lb_pseudorand_t *)a)->as_index];
  asb = &lbm->ass[((lb_pseudorand_t *)b)->as_index];
  return memcmp(&asa->address, &asb->address, sizeof(asb->address));
}

static void lb_vip_garbage_collection(lb_vip_t *vip)
{
  lb_main_t *lbm = &lb_main;
  ASSERT (lbm->writer_lock[0]);

  u32 now = (u32) vlib_time_now(vlib_get_main());
  if (!clib_u32_loop_gt(now, vip->last_garbage_collection + LB_GARBAGE_RUN))
    return;

  vip->last_garbage_collection = now;
  lb_as_t *as;
  u32 *as_index;
  pool_foreach(as_index, vip->as_indexes, {
      as = &lbm->ass[*as_index];
      if (!(as->flags & LB_AS_FLAGS_USED) && //Not used
	  clib_u32_loop_gt(now, as->last_used + LB_CONCURRENCY_TIMEOUT) && //Not recently used
	  (vlib_refcount_get(&lbm->as_refcount, as - lbm->ass) == 0))
	{ //Not referenced
	  fib_entry_child_remove(as->next_hop_fib_entry_index,
				 as->next_hop_child_index);
	  fib_table_entry_delete_index(as->next_hop_fib_entry_index,
				       FIB_SOURCE_RR);
	  as->next_hop_fib_entry_index = FIB_NODE_INDEX_INVALID;

	  pool_put(vip->as_indexes, as_index);
	  pool_put(lbm->ass, as);
	}
  });
}

void lb_garbage_collection()
{
  lb_main_t *lbm = &lb_main;
  lb_get_writer_lock();
  lb_vip_t *vip;
  u32 *to_be_removed_vips = 0, *i;
  pool_foreach(vip, lbm->vips, {
      lb_vip_garbage_collection(vip);

      if (!(vip->flags & LB_VIP_FLAGS_USED) &&
          (pool_elts(vip->as_indexes) == 0)) {
        vec_add1(to_be_removed_vips, vip - lbm->vips);
      }
  });

  vec_foreach(i, to_be_removed_vips) {
    vip = &lbm->vips[*i];
    pool_put(lbm->vips, vip);
    pool_free(vip->as_indexes);
  }

  vec_free(to_be_removed_vips);
  lb_put_writer_lock();
}

static void lb_vip_update_new_flow_table(lb_vip_t *vip)
{
  lb_main_t *lbm = &lb_main;
  lb_new_flow_entry_t *old_table;
  u32 i, *as_index;
  lb_new_flow_entry_t *new_flow_table = 0;
  lb_as_t *as;
  lb_pseudorand_t *pr, *sort_arr = 0;
  u32 count;

  ASSERT (lbm->writer_lock[0]); //We must have the lock

  //Check if some AS is configured or not
  i = 0;
  pool_foreach(as_index, vip->as_indexes, {
      as = &lbm->ass[*as_index];
      if (as->flags & LB_AS_FLAGS_USED) { //Not used anymore
        i = 1;
        goto out; //Not sure 'break' works in this macro-loop
      }
  });

out:
  if (i == 0) {
    //Only the default. i.e. no AS
    vec_validate(new_flow_table, vip->new_flow_table_mask);
    for (i=0; i<vec_len(new_flow_table); i++)
      new_flow_table[i].as_index = 0;

    goto finished;
  }

  //First, let's sort the ASs
  sort_arr = 0;
  vec_alloc(sort_arr, pool_elts(vip->as_indexes));

  i = 0;
  pool_foreach(as_index, vip->as_indexes, {
      as = &lbm->ass[*as_index];
      if (!(as->flags & LB_AS_FLAGS_USED)) //Not used anymore
        continue;

      sort_arr[i].as_index = as - lbm->ass;
      i++;
  });
  _vec_len(sort_arr) = i;

  vec_sort_with_function(sort_arr, lb_pseudorand_compare);

  //Now let's pseudo-randomly generate permutations
  vec_foreach(pr, sort_arr) {
    lb_as_t *as = &lbm->ass[pr->as_index];

    u64 seed = clib_xxhash(as->address.as_u64[0] ^
                           as->address.as_u64[1]);
    /* We have 2^n buckets.
     * skip must be prime with 2^n.
     * So skip must be odd.
     * MagLev actually state that M should be prime,
     * but this has a big computation cost (% operation).
     * Using 2^n is more better (& operation).
     */
    pr->skip = ((seed & 0xffffffff) | 1) & vip->new_flow_table_mask;
    pr->last = (seed >> 32) & vip->new_flow_table_mask;
  }

  //Let's create a new flow table
  vec_validate(new_flow_table, vip->new_flow_table_mask);
  for (i=0; i<vec_len(new_flow_table); i++)
    new_flow_table[i].as_index = ~0;

  u32 done = 0;
  while (1) {
    vec_foreach(pr, sort_arr) {
      while (1) {
        u32 last = pr->last;
        pr->last = (pr->last + pr->skip) & vip->new_flow_table_mask;
        if (new_flow_table[last].as_index == ~0) {
          new_flow_table[last].as_index = pr->as_index;
          break;
        }
      }
      done++;
      if (done == vec_len(new_flow_table))
        goto finished;
    }
  }

  vec_free(sort_arr);

finished:

//Count number of changed entries
  count = 0;
  for (i=0; i<vec_len(new_flow_table); i++)
    if (vip->new_flow_table == 0 ||
        new_flow_table[i].as_index != vip->new_flow_table[i].as_index)
      count++;

  old_table = vip->new_flow_table;
  vip->new_flow_table = new_flow_table;
  vec_free(old_table);
}

int lb_conf(ip4_address_t *ip4_address, ip6_address_t *ip6_address,
           u32 per_cpu_sticky_buckets, u32 flow_timeout)
{
  lb_main_t *lbm = &lb_main;

  if (!is_pow2(per_cpu_sticky_buckets))
    return VNET_API_ERROR_INVALID_MEMORY_SIZE;

  lb_get_writer_lock(); //Not exactly necessary but just a reminder that it exists for my future self
  lbm->ip4_src_address = *ip4_address;
  lbm->ip6_src_address = *ip6_address;
  lbm->per_cpu_sticky_buckets = per_cpu_sticky_buckets;
  lbm->flow_timeout = flow_timeout;
  lb_put_writer_lock();
  return 0;
}

static
int lb_vip_find_index_with_lock(ip46_address_t *prefix, u8 plen, u32 *vip_index)
{
  lb_main_t *lbm = &lb_main;
  lb_vip_t *vip;
  ASSERT (lbm->writer_lock[0]); //This must be called with the lock owned
  ip46_prefix_normalize(prefix, plen);
  pool_foreach(vip, lbm->vips, {
      if ((vip->flags & LB_AS_FLAGS_USED) &&
          vip->plen == plen &&
          vip->prefix.as_u64[0] == prefix->as_u64[0] &&
          vip->prefix.as_u64[1] == prefix->as_u64[1]) {
        *vip_index = vip - lbm->vips;
        return 0;
      }
  });
  return VNET_API_ERROR_NO_SUCH_ENTRY;
}

int lb_vip_find_index(ip46_address_t *prefix, u8 plen, u32 *vip_index)
{
  int ret;
  lb_get_writer_lock();
  ret = lb_vip_find_index_with_lock(prefix, plen, vip_index);
  lb_put_writer_lock();
  return ret;
}

static int lb_as_find_index_vip(lb_vip_t *vip, ip46_address_t *address, u32 *as_index)
{
  lb_main_t *lbm = &lb_main;
  ASSERT (lbm->writer_lock[0]); //This must be called with the lock owned
  lb_as_t *as;
  u32 *asi;
  pool_foreach(asi, vip->as_indexes, {
      as = &lbm->ass[*asi];
      if (as->vip_index == (vip - lbm->vips) &&
          as->address.as_u64[0] == address->as_u64[0] &&
          as->address.as_u64[1] == address->as_u64[1]) {
        *as_index = as - lbm->ass;
        return 0;
      }
  });
  return -1;
}

int lb_vip_add_ass(u32 vip_index, ip46_address_t *addresses, u32 n)
{
  lb_main_t *lbm = &lb_main;
  lb_get_writer_lock();
  lb_vip_t *vip;
  if (!(vip = lb_vip_get_by_index(vip_index))) {
    lb_put_writer_lock();
    return VNET_API_ERROR_NO_SUCH_ENTRY;
  }

  ip46_type_t type = lb_vip_is_gre4(vip)?IP46_TYPE_IP4:IP46_TYPE_IP6;
  u32 *to_be_added = 0;
  u32 *to_be_updated = 0;
  u32 i;
  u32 *ip;

  //Sanity check
  while (n--) {

    if (!lb_as_find_index_vip(vip, &addresses[n], &i)) {
      if (lbm->ass[i].flags & LB_AS_FLAGS_USED) {
        vec_free(to_be_added);
        vec_free(to_be_updated);
        lb_put_writer_lock();
        return VNET_API_ERROR_VALUE_EXIST;
      }
      vec_add1(to_be_updated, i);
      goto next;
    }

    if (ip46_address_type(&addresses[n]) != type) {
      vec_free(to_be_added);
      vec_free(to_be_updated);
      lb_put_writer_lock();
      return VNET_API_ERROR_INVALID_ADDRESS_FAMILY;
    }

    if (n) {
      u32 n2 = n;
      while(n2--) //Check for duplicates
        if (addresses[n2].as_u64[0] == addresses[n].as_u64[0] &&
            addresses[n2].as_u64[1] == addresses[n].as_u64[1])
          goto next;
    }

    vec_add1(to_be_added, n);

next:
    continue;
  }

  //Update reused ASs
  vec_foreach(ip, to_be_updated) {
    lbm->ass[*ip].flags = LB_AS_FLAGS_USED;
  }
  vec_free(to_be_updated);

  //Create those who have to be created
  vec_foreach(ip, to_be_added) {
    lb_as_t *as;
    u32 *as_index;
    pool_get(lbm->ass, as);
    as->address = addresses[*ip];
    as->flags = LB_AS_FLAGS_USED;
    as->vip_index = vip_index;
    pool_get(vip->as_indexes, as_index);
    *as_index = as - lbm->ass;

    /*
     * become a child of the FIB entry
     * so we are informed when its forwarding changes
     */
    fib_prefix_t nh = {};
    if (lb_vip_is_gre4(vip)) {
	nh.fp_addr.ip4 = as->address.ip4;
	nh.fp_len = 32;
	nh.fp_proto = FIB_PROTOCOL_IP4;
    } else {
	nh.fp_addr.ip6 = as->address.ip6;
	nh.fp_len = 128;
	nh.fp_proto = FIB_PROTOCOL_IP6;
    }

    as->next_hop_fib_entry_index =
	fib_table_entry_special_add(0,
				    &nh,
				    FIB_SOURCE_RR,
				    FIB_ENTRY_FLAG_NONE,
				    ADJ_INDEX_INVALID);
    as->next_hop_child_index =
	fib_entry_child_add(as->next_hop_fib_entry_index,
			    lbm->fib_node_type,
			    as - lbm->ass);

    lb_as_stack(as);
  }
  vec_free(to_be_added);

  //Recompute flows
  lb_vip_update_new_flow_table(vip);

  //Garbage collection maybe
  lb_vip_garbage_collection(vip);

  lb_put_writer_lock();
  return 0;
}

int lb_vip_del_ass_withlock(u32 vip_index, ip46_address_t *addresses, u32 n)
{
  lb_main_t *lbm = &lb_main;
  u32 now = (u32) vlib_time_now(vlib_get_main());
  u32 *ip = 0;

  lb_vip_t *vip;
  if (!(vip = lb_vip_get_by_index(vip_index))) {
    return VNET_API_ERROR_NO_SUCH_ENTRY;
  }

  u32 *indexes = NULL;
  while (n--) {
    u32 i;
    if (lb_as_find_index_vip(vip, &addresses[n], &i)) {
      vec_free(indexes);
      return VNET_API_ERROR_NO_SUCH_ENTRY;
    }

    if (n) { //Check for duplicates
      u32 n2 = n - 1;
      while(n2--) {
        if (addresses[n2].as_u64[0] == addresses[n].as_u64[0] &&
            addresses[n2].as_u64[1] == addresses[n].as_u64[1])
          goto next;
      }
    }

    vec_add1(indexes, i);
next:
  continue;
  }

  //Garbage collection maybe
  lb_vip_garbage_collection(vip);

  if (indexes != NULL) {
    vec_foreach(ip, indexes) {
      lbm->ass[*ip].flags &= ~LB_AS_FLAGS_USED;
      lbm->ass[*ip].last_used = now;
    }

    //Recompute flows
    lb_vip_update_new_flow_table(vip);
  }

  vec_free(indexes);
  return 0;
}

int lb_vip_del_ass(u32 vip_index, ip46_address_t *addresses, u32 n)
{
  lb_get_writer_lock();
  int ret = lb_vip_del_ass_withlock(vip_index, addresses, n);
  lb_put_writer_lock();
  return ret;
}

/**
 * Add the VIP adjacency to the ip4 or ip6 fib
 */
static void lb_vip_add_adjacency(lb_main_t *lbm, lb_vip_t *vip)
{
  dpo_proto_t proto = 0;
  dpo_id_t dpo = DPO_INVALID;
  fib_prefix_t pfx = {};
  if (lb_vip_is_ip4(vip)) {
      pfx.fp_addr.ip4 = vip->prefix.ip4;
      pfx.fp_len = vip->plen - 96;
      pfx.fp_proto = FIB_PROTOCOL_IP4;
      proto = DPO_PROTO_IP4;
  } else {
      pfx.fp_addr.ip6 = vip->prefix.ip6;
      pfx.fp_len = vip->plen;
      pfx.fp_proto = FIB_PROTOCOL_IP6;
      proto = DPO_PROTO_IP6;
  }
  dpo_set(&dpo, lb_vip_is_gre4(vip)?lbm->dpo_gre4_type:lbm->dpo_gre6_type,
      proto, vip - lbm->vips);
  fib_table_entry_special_dpo_add(0,
				  &pfx,
				  FIB_SOURCE_PLUGIN_HI,
				  FIB_ENTRY_FLAG_EXCLUSIVE,
				  &dpo);
  dpo_reset(&dpo);
}

/**
 * Deletes the adjacency associated with the VIP
 */
static void lb_vip_del_adjacency(lb_main_t *lbm, lb_vip_t *vip)
{
  fib_prefix_t pfx = {};
  if (lb_vip_is_ip4(vip)) {
      pfx.fp_addr.ip4 = vip->prefix.ip4;
      pfx.fp_len = vip->plen - 96;
      pfx.fp_proto = FIB_PROTOCOL_IP4;
  } else {
      pfx.fp_addr.ip6 = vip->prefix.ip6;
      pfx.fp_len = vip->plen;
      pfx.fp_proto = FIB_PROTOCOL_IP6;
  }
  fib_table_entry_special_remove(0, &pfx, FIB_SOURCE_PLUGIN_HI);
}

int lb_vip_add(ip46_address_t *prefix, u8 plen, lb_vip_type_t type, u32 new_length, u32 *vip_index)
{
  lb_main_t *lbm = &lb_main;
  lb_vip_t *vip;
  lb_get_writer_lock();
  ip46_prefix_normalize(prefix, plen);

  if (!lb_vip_find_index_with_lock(prefix, plen, vip_index)) {
    lb_put_writer_lock();
    return VNET_API_ERROR_VALUE_EXIST;
  }

  if (!is_pow2(new_length)) {
    lb_put_writer_lock();
    return VNET_API_ERROR_INVALID_MEMORY_SIZE;
  }

  if (ip46_prefix_is_ip4(prefix, plen) &&
      (type != LB_VIP_TYPE_IP4_GRE4) &&
      (type != LB_VIP_TYPE_IP4_GRE6))
    return VNET_API_ERROR_INVALID_ADDRESS_FAMILY;


  //Allocate
  pool_get(lbm->vips, vip);

  //Init
  vip->prefix = *prefix;
  vip->plen = plen;
  vip->last_garbage_collection = (u32) vlib_time_now(vlib_get_main());
  vip->type = type;
  vip->flags = LB_VIP_FLAGS_USED;
  vip->as_indexes = 0;

  //Validate counters
  u32 i;
  for (i = 0; i < LB_N_VIP_COUNTERS; i++) {
    vlib_validate_simple_counter(&lbm->vip_counters[i], vip - lbm->vips);
    vlib_zero_simple_counter(&lbm->vip_counters[i], vip - lbm->vips);
  }

  //Configure new flow table
  vip->new_flow_table_mask = new_length - 1;
  vip->new_flow_table = 0;

  //Create a new flow hash table full of the default entry
  lb_vip_update_new_flow_table(vip);

  //Create adjacency to direct traffic
  lb_vip_add_adjacency(lbm, vip);

  //Return result
  *vip_index = vip - lbm->vips;

  lb_put_writer_lock();
  return 0;
}

int lb_vip_del(u32 vip_index)
{
  lb_main_t *lbm = &lb_main;
  lb_vip_t *vip;
  lb_get_writer_lock();
  if (!(vip = lb_vip_get_by_index(vip_index))) {
    lb_put_writer_lock();
    return VNET_API_ERROR_NO_SUCH_ENTRY;
  }

  //FIXME: This operation is actually not working
  //We will need to remove state before performing this.

  {
    //Remove all ASs
    ip46_address_t *ass = 0;
    lb_as_t *as;
    u32 *as_index;
    pool_foreach(as_index, vip->as_indexes, {
        as = &lbm->ass[*as_index];
        vec_add1(ass, as->address);
    });
    if (vec_len(ass))
      lb_vip_del_ass_withlock(vip_index, ass, vec_len(ass));
    vec_free(ass);
  }

  //Delete adjacency
  lb_vip_del_adjacency(lbm, vip);

  //Set the VIP as unused
  vip->flags &= ~LB_VIP_FLAGS_USED;

  lb_put_writer_lock();
  return 0;
}

clib_error_t *
vlib_plugin_register (vlib_main_t * vm,
                      vnet_plugin_handoff_t * h,
                      int from_early_init)
{
  clib_error_t *error = 0;
  return error;
}


u8 *format_lb_dpo (u8 * s, va_list * va)
{
  index_t index = va_arg (*va, index_t);
  CLIB_UNUSED(u32 indent) = va_arg (*va, u32);
  lb_main_t *lbm = &lb_main;
  lb_vip_t *vip = pool_elt_at_index (lbm->vips, index);
  return format (s, "%U", format_lb_vip, vip);
}

static void lb_dpo_lock (dpo_id_t *dpo) {}
static void lb_dpo_unlock (dpo_id_t *dpo) {}

static fib_node_t *
lb_fib_node_get_node (fib_node_index_t index)
{
  lb_main_t *lbm = &lb_main;
  lb_as_t *as = pool_elt_at_index (lbm->ass, index);
  return (&as->fib_node);
}

static void
lb_fib_node_last_lock_gone (fib_node_t *node)
{
}

static lb_as_t *
lb_as_from_fib_node (fib_node_t *node)
{
  return ((lb_as_t*)(((char*)node) -
      STRUCT_OFFSET_OF(lb_as_t, fib_node)));
}

static void
lb_as_stack (lb_as_t *as)
{
  lb_main_t *lbm = &lb_main;
  lb_vip_t *vip = &lbm->vips[as->vip_index];
  dpo_stack(lb_vip_is_gre4(vip)?lbm->dpo_gre4_type:lbm->dpo_gre6_type,
	    lb_vip_is_ip4(vip)?DPO_PROTO_IP4:DPO_PROTO_IP6,
	    &as->dpo,
	    fib_entry_contribute_ip_forwarding(
		as->next_hop_fib_entry_index));
}

static fib_node_back_walk_rc_t
lb_fib_node_back_walk_notify (fib_node_t *node,
			       fib_node_back_walk_ctx_t *ctx)
{
    lb_as_stack(lb_as_from_fib_node(node));
    return (FIB_NODE_BACK_WALK_CONTINUE);
}

clib_error_t *
lb_init (vlib_main_t * vm)
{
  vlib_thread_main_t *tm = vlib_get_thread_main ();
  lb_main_t *lbm = &lb_main;
  lb_as_t *default_as;
  fib_node_vft_t lb_fib_node_vft = {
      .fnv_get = lb_fib_node_get_node,
      .fnv_last_lock = lb_fib_node_last_lock_gone,
      .fnv_back_walk = lb_fib_node_back_walk_notify,
  };
  dpo_vft_t lb_vft = {
      .dv_lock = lb_dpo_lock,
      .dv_unlock = lb_dpo_unlock,
      .dv_format = format_lb_dpo,
  };

  lbm->vips = 0;
  lbm->per_cpu = 0;
  vec_validate(lbm->per_cpu, tm->n_vlib_mains - 1);
  lbm->writer_lock = clib_mem_alloc_aligned (CLIB_CACHE_LINE_BYTES,  CLIB_CACHE_LINE_BYTES);
  lbm->writer_lock[0] = 0;
  lbm->per_cpu_sticky_buckets = LB_DEFAULT_PER_CPU_STICKY_BUCKETS;
  lbm->flow_timeout = LB_DEFAULT_FLOW_TIMEOUT;
  lbm->ip4_src_address.as_u32 = 0xffffffff;
  lbm->ip6_src_address.as_u64[0] = 0xffffffffffffffffL;
  lbm->ip6_src_address.as_u64[1] = 0xffffffffffffffffL;
  lbm->dpo_gre4_type = dpo_register_new_type(&lb_vft, lb_dpo_gre4_nodes);
  lbm->dpo_gre6_type = dpo_register_new_type(&lb_vft, lb_dpo_gre6_nodes);
  lbm->fib_node_type = fib_node_register_new_type(&lb_fib_node_vft);

  //Init AS reference counters
  vlib_refcount_init(&lbm->as_refcount);

  //Allocate and init default AS.
  lbm->ass = 0;
  pool_get(lbm->ass, default_as);
  default_as->flags = 0;
  default_as->dpo.dpoi_next_node = LB_NEXT_DROP;
  default_as->vip_index = ~0;
  default_as->address.ip6.as_u64[0] = 0xffffffffffffffffL;
  default_as->address.ip6.as_u64[1] = 0xffffffffffffffffL;

#define _(a,b,c) lbm->vip_counters[c].name = b;
  lb_foreach_vip_counter
#undef _
  return NULL;
}

VLIB_INIT_FUNCTION (lb_init);