summaryrefslogtreecommitdiffstats
path: root/bgpd/bgp_mpath.c
blob: 64af8a541161ddc7988bb25929f910fa88cc69b2 (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
/*
 * BGP Multipath
 * Copyright (C) 2010 Google Inc.
 *
 * This file is part of Quagga
 *
 * Quagga is free software; you can redistribute it and/or modify it
 * under the terms of the GNU General Public License as published by the
 * Free Software Foundation; either version 2, or (at your option) any
 * later version.
 *
 * Quagga is distributed in the hope that it will be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License along
 * with this program; see the file COPYING; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
 */

#include <zebra.h>

#include "command.h"
#include "prefix.h"
#include "linklist.h"
#include "sockunion.h"
#include "memory.h"
#include "queue.h"
#include "filter.h"

#include "bgpd/bgpd.h"
#include "bgpd/bgp_table.h"
#include "bgpd/bgp_route.h"
#include "bgpd/bgp_attr.h"
#include "bgpd/bgp_debug.h"
#include "bgpd/bgp_aspath.h"
#include "bgpd/bgp_community.h"
#include "bgpd/bgp_ecommunity.h"
#include "bgpd/bgp_lcommunity.h"
#include "bgpd/bgp_mpath.h"

/*
 * bgp_maximum_paths_set
 *
 * Record maximum-paths configuration for BGP instance
 */
int bgp_maximum_paths_set(struct bgp *bgp, afi_t afi, safi_t safi, int peertype,
			  uint16_t maxpaths, bool same_clusterlen)
{
	if (!bgp || (afi >= AFI_MAX) || (safi >= SAFI_MAX))
		return -1;

	switch (peertype) {
	case BGP_PEER_IBGP:
		bgp->maxpaths[afi][safi].maxpaths_ibgp = maxpaths;
		bgp->maxpaths[afi][safi].same_clusterlen = same_clusterlen;
		break;
	case BGP_PEER_EBGP:
		bgp->maxpaths[afi][safi].maxpaths_ebgp = maxpaths;
		break;
	default:
		return -1;
	}

	return 0;
}

/*
 * bgp_maximum_paths_unset
 *
 * Remove maximum-paths configuration from BGP instance
 */
int bgp_maximum_paths_unset(struct bgp *bgp, afi_t afi, safi_t safi,
			    int peertype)
{
	if (!bgp || (afi >= AFI_MAX) || (safi >= SAFI_MAX))
		return -1;

	switch (peertype) {
	case BGP_PEER_IBGP:
		bgp->maxpaths[afi][safi].maxpaths_ibgp = multipath_num;
		bgp->maxpaths[afi][safi].same_clusterlen = false;
		break;
	case BGP_PEER_EBGP:
		bgp->maxpaths[afi][safi].maxpaths_ebgp = multipath_num;
		break;
	default:
		return -1;
	}

	return 0;
}

/*
 * bgp_interface_same
 *
 * Return true if ifindex for ifp1 and ifp2 are the same, else return false.
 */
static int bgp_interface_same(struct interface *ifp1, struct interface *ifp2)
{
	if (!ifp1 && !ifp2)
		return 1;

	if (!ifp1 && ifp2)
		return 0;

	if (ifp1 && !ifp2)
		return 0;

	return (ifp1->ifindex == ifp2->ifindex);
}


/*
 * bgp_path_info_nexthop_cmp
 *
 * Compare the nexthops of two paths. Return value is less than, equal to,
 * or greater than zero if bpi1 is respectively less than, equal to,
 * or greater than bpi2.
 */
int bgp_path_info_nexthop_cmp(struct bgp_path_info *bpi1,
			      struct bgp_path_info *bpi2)
{
	int compare;
	struct in6_addr addr1, addr2;

	compare = IPV4_ADDR_CMP(&bpi1->attr->nexthop, &bpi2->attr->nexthop);
	if (!compare) {
		if (bpi1->attr->mp_nexthop_len == bpi2->attr->mp_nexthop_len) {
			switch (bpi1->attr->mp_nexthop_len) {
			case BGP_ATTR_NHLEN_IPV4:
			case BGP_ATTR_NHLEN_VPNV4:
				compare = IPV4_ADDR_CMP(
					&bpi1->attr->mp_nexthop_global_in,
					&bpi2->attr->mp_nexthop_global_in);
				break;
			case BGP_ATTR_NHLEN_IPV6_GLOBAL:
			case BGP_ATTR_NHLEN_VPNV6_GLOBAL:
				compare = IPV6_ADDR_CMP(
					&bpi1->attr->mp_nexthop_global,
					&bpi2->attr->mp_nexthop_global);
				break;
			case BGP_ATTR_NHLEN_IPV6_GLOBAL_AND_LL:
				addr1 = (bpi1->attr->mp_nexthop_prefer_global)
						? bpi1->attr->mp_nexthop_global
						: bpi1->attr->mp_nexthop_local;
				addr2 = (bpi2->attr->mp_nexthop_prefer_global)
						? bpi2->attr->mp_nexthop_global
						: bpi2->attr->mp_nexthop_local;

				if (!bpi1->attr->mp_nexthop_prefer_global
				    && !bpi2->attr->mp_nexthop_prefer_global)
					compare = !bgp_interface_same(
						bpi1->peer->ifp,
						bpi2->peer->ifp);

				if (!compare)
					compare = IPV6_ADDR_CMP(&addr1, &addr2);
				break;
			}
		}

		/* This can happen if one IPv6 peer sends you global and
		 * link-local
		 * nexthops but another IPv6 peer only sends you global
		 */
		else if (bpi1->attr->mp_nexthop_len
				 == BGP_ATTR_NHLEN_IPV6_GLOBAL
			 || bpi1->attr->mp_nexthop_len
				    == BGP_ATTR_NHLEN_IPV6_GLOBAL_AND_LL) {
			compare = IPV6_ADDR_CMP(&bpi1->attr->mp_nexthop_global,
						&bpi2->attr->mp_nexthop_global);
			if (!compare) {
				if (bpi1->attr->mp_nexthop_len
				    < bpi2->attr->mp_nexthop_len)
					compare = -1;
				else
					compare = 1;
			}
		}
	}

	/*
	 * If both nexthops are same then check
	 * if they belong to same VRF
	 */
	if (!compare && bpi1->attr->nh_type != NEXTHOP_TYPE_BLACKHOLE) {
		if (bpi1->extra && bpi1->extra->bgp_orig && bpi2->extra
		    && bpi2->extra->bgp_orig) {
			if (bpi1->extra->bgp_orig->vrf_id
			    != bpi2->extra->bgp_orig->vrf_id) {
				compare = 1;
			}
		}
	}

	return compare;
}

/*
 * bgp_path_info_mpath_cmp
 *
 * This function determines our multipath list ordering. By ordering
 * the list we can deterministically select which paths are included
 * in the multipath set. The ordering also helps in detecting changes
 * in the multipath selection so we can detect whether to send an
 * update to zebra.
 *
 * The order of paths is determined first by received nexthop, and then
 * by peer address if the nexthops are the same.
 */
static int bgp_path_info_mpath_cmp(void *val1, void *val2)
{
	struct bgp_path_info *bpi1, *bpi2;
	int compare;

	bpi1 = val1;
	bpi2 = val2;

	compare = bgp_path_info_nexthop_cmp(bpi1, bpi2);

	if (!compare) {
		if (!bpi1->peer->su_remote && !bpi2->peer->su_remote)
			compare = 0;
		else if (!bpi1->peer->su_remote)
			compare = 1;
		else if (!bpi2->peer->su_remote)
			compare = -1;
		else
			compare = sockunion_cmp(bpi1->peer->su_remote,
						bpi2->peer->su_remote);
	}

	return compare;
}

/*
 * bgp_mp_list_init
 *
 * Initialize the mp_list, which holds the list of multipaths
 * selected by bgp_best_selection
 */
void bgp_mp_list_init(struct list *mp_list)
{
	assert(mp_list);
	memset(mp_list, 0, sizeof(struct list));
	mp_list->cmp = bgp_path_info_mpath_cmp;
}

/*
 * bgp_mp_list_clear
 *
 * Clears all entries out of the mp_list
 */
void bgp_mp_list_clear(struct list *mp_list)
{
	assert(mp_list);
	list_delete_all_node(mp_list);
}

/*
 * bgp_mp_list_add
 *
 * Adds a multipath entry to the mp_list
 */
void bgp_mp_list_add(struct list *mp_list, struct bgp_path_info *mpinfo)
{
	assert(mp_list && mpinfo);
	listnode_add_sort(mp_list, mpinfo);
}

/*
 * bgp_path_info_mpath_new
 *
 * Allocate and zero memory for a new bgp_path_info_mpath element
 */
static struct bgp_path_info_mpath *bgp_path_info_mpath_new(void)
{
	struct bgp_path_info_mpath *new_mpath;
	new_mpath = XCALLOC(MTYPE_BGP_MPATH_INFO,
			    sizeof(struct bgp_path_info_mpath));
	return new_mpath;
}

/*
 * bgp_path_info_mpath_free
 *
 * Release resources for a bgp_path_info_mpath element and zero out pointer
 */
void bgp_path_info_mpath_free(struct bgp_path_info_mpath **mpath)
{
	if (mpath && *mpath) {
		if ((*mpath)->mp_attr)
			bgp_attr_unintern(&(*mpath)->mp_attr);
		XFREE(MTYPE_BGP_MPATH_INFO, *mpath);
	}
}

/*
 * bgp_path_info_mpath_get
 *
 * Fetch the mpath element for the given bgp_path_info. Used for
 * doing lazy allocation.
 */
static struct bgp_path_info_mpath *
bgp_path_info_mpath_get(struct bgp_path_info *path)
{
	struct bgp_path_info_mpath *mpath;

	if (!path)
		return NULL;

	if (!path->mpath) {
		mpath = bgp_path_info_mpath_new();
		if (!mpath)
			return NULL;
		path->mpath = mpath;
		mpath->mp_info = path;
	}
	return path->mpath;
}

/*
 * bgp_path_info_mpath_enqueue
 *
 * Enqueue a path onto the multipath list given the previous multipath
 * list entry
 */
static void bgp_path_info_mpath_enqueue(struct bgp_path_info *prev_info,
					struct bgp_path_info *path)
{
	struct bgp_path_info_mpath *prev, *mpath;

	prev = bgp_path_info_mpath_get(prev_info);
	mpath = bgp_path_info_mpath_get(path);
	if (!prev || !mpath)
		return;

	mpath->mp_next = prev->mp_next;
	mpath->mp_prev = prev;
	if (prev->mp_next)
		prev->mp_next->mp_prev = mpath;
	prev->mp_next = mpath;

	SET_FLAG(path->flags, BGP_PATH_MULTIPATH);
}

/*
 * bgp_path_info_mpath_dequeue
 *
 * Remove a path from the multipath list
 */
void bgp_path_info_mpath_dequeue(struct bgp_path_info *path)
{
	struct bgp_path_info_mpath *mpath = path->mpath;
	if (!mpath)
		return;
	if (mpath->mp_prev)
		mpath->mp_prev->mp_next = mpath->mp_next;
	if (mpath->mp_next)
		mpath->mp_next->mp_prev = mpath->mp_prev;
	mpath->mp_next = mpath->mp_prev = NULL;
	UNSET_FLAG(path->flags, BGP_PATH_MULTIPATH);
}

/*
 * bgp_path_info_mpath_next
 *
 * Given a bgp_path_info, return the next multipath entry
 */
struct bgp_path_info *bgp_path_info_mpath_next(struct bgp_path_info *path)
{
	if (!path->mpath || !path->mpath->mp_next)
		return NULL;
	return path->mpath->mp_next->mp_info;
}

/*
 * bgp_path_info_mpath_first
 *
 * Given bestpath bgp_path_info, return the first multipath entry.
 */
struct bgp_path_info *bgp_path_info_mpath_first(struct bgp_path_info *path)
{
	return bgp_path_info_mpath_next(path);
}

/*
 * bgp_path_info_mpath_count
 *
 * Given the bestpath bgp_path_info, return the number of multipath entries
 */
uint32_t bgp_path_info_mpath_count(struct bgp_path_info *path)
{
	if (!path->mpath)
		return 0;
	return path->mpath->mp_count;
}

/*
 * bgp_path_info_mpath_count_set
 *
 * Sets the count of multipaths into bestpath's mpath element
 */
static void bgp_path_info_mpath_count_set(struct bgp_path_info *path,
					  uint16_t count)
{
	struct bgp_path_info_mpath *mpath;
	if (!count && !path->mpath)
		return;
	mpath = bgp_path_info_mpath_get(path);
	if (!mpath)
		return;
	mpath->mp_count = count;
}

/*
 * bgp_path_info_mpath_lb_update
 *
 * Update cumulative info related to link-bandwidth
 */
static void bgp_path_info_mpath_lb_update(struct bgp_path_info *path, bool set,
					  bool all_paths_lb, uint64_t cum_bw)
{
	struct bgp_path_info_mpath *mpath;

	mpath = path->mpath;
	if (mpath == NULL) {
		if (!set || (cum_bw == 0 && !all_paths_lb))
			return;

		mpath = bgp_path_info_mpath_get(path);
		if (!mpath)
			return;
	}
	if (set) {
		if (cum_bw)
			SET_FLAG(mpath->mp_flags, BGP_MP_LB_PRESENT);
		else
			UNSET_FLAG(mpath->mp_flags, BGP_MP_LB_PRESENT);
		if (all_paths_lb)
			SET_FLAG(mpath->mp_flags, BGP_MP_LB_ALL);
		else
			UNSET_FLAG(mpath->mp_flags, BGP_MP_LB_ALL);
		mpath->cum_bw = cum_bw;
	} else {
		mpath->mp_flags = 0;
		mpath->cum_bw = 0;
	}
}

/*
 * bgp_path_info_mpath_attr
 *
 * Given bestpath bgp_path_info, return aggregated attribute set used
 * for advertising the multipath route
 */
struct attr *bgp_path_info_mpath_attr(struct bgp_path_info *path)
{
	if (!path->mpath)
		return NULL;
	return path->mpath->mp_attr;
}

/*
 * bgp_path_info_chkwtd
 *
 * Return if we should attempt to do weighted ECMP or not
 * The path passed in is the bestpath.
 */
bool bgp_path_info_mpath_chkwtd(struct bgp *bgp, struct bgp_path_info *path)
{
	/* Check if told to ignore weights or not multipath */
	if (bgp->lb_handling == BGP_LINK_BW_IGNORE_BW || !path->mpath)
		return false;

	/* All paths in multipath should have associated weight (bandwidth)
	 * unless told explicitly otherwise.
	 */
	if (bgp->lb_handling != BGP_LINK_BW_SKIP_MISSING &&
	    bgp->lb_handling != BGP_LINK_BW_DEFWT_4_MISSING)
		return (path->mpath->mp_flags & BGP_MP_LB_ALL);

	/* At least one path should have bandwidth. */
	return (path->mpath->mp_flags & BGP_MP_LB_PRESENT);
}

/*
 * bgp_path_info_mpath_attr
 *
 * Given bestpath bgp_path_info, return cumulative bandwidth
 * computed for all multipaths with bandwidth info
 */
uint64_t bgp_path_info_mpath_cumbw(struct bgp_path_info *path)
{
	if (!path->mpath)
		return 0;
	return path->mpath->cum_bw;
}

/*
 * bgp_path_info_mpath_attr_set
 *
 * Sets the aggregated attribute into bestpath's mpath element
 */
static void bgp_path_info_mpath_attr_set(struct bgp_path_info *path,
					 struct attr *attr)
{
	struct bgp_path_info_mpath *mpath;
	if (!attr && !path->mpath)
		return;
	mpath = bgp_path_info_mpath_get(path);
	if (!mpath)
		return;
	mpath->mp_attr = attr;
}

/*
 * bgp_path_info_mpath_update
 *
 * Compare and sync up the multipath list with the mp_list generated by
 * bgp_best_selection
 */
void bgp_path_info_mpath_update(struct bgp *bgp, struct bgp_dest *dest,
				struct bgp_path_info *new_best,
				struct bgp_path_info *old_best,
				struct list *mp_list,
				struct bgp_maxpaths_cfg *mpath_cfg)
{
	uint16_t maxpaths, mpath_count, old_mpath_count;
	uint32_t bwval;
	uint64_t cum_bw, old_cum_bw;
	struct listnode *mp_node, *mp_next_node;
	struct bgp_path_info *cur_mpath, *new_mpath, *next_mpath, *prev_mpath;
	int mpath_changed, debug;
	bool all_paths_lb;
	char path_buf[PATH_ADDPATH_STR_BUFFER];

	mpath_changed = 0;
	maxpaths = multipath_num;
	mpath_count = 0;
	cur_mpath = NULL;
	old_mpath_count = 0;
	old_cum_bw = cum_bw = 0;
	prev_mpath = new_best;
	mp_node = listhead(mp_list);
	debug = bgp_debug_bestpath(dest);

	if (new_best) {
		mpath_count++;
		if (new_best != old_best)
			bgp_path_info_mpath_dequeue(new_best);
		maxpaths = (new_best->peer->sort == BGP_PEER_IBGP)
				   ? mpath_cfg->maxpaths_ibgp
				   : mpath_cfg->maxpaths_ebgp;
	}

	if (old_best) {
		cur_mpath = bgp_path_info_mpath_first(old_best);
		old_mpath_count = bgp_path_info_mpath_count(old_best);
		old_cum_bw = bgp_path_info_mpath_cumbw(old_best);
		bgp_path_info_mpath_count_set(old_best, 0);
		bgp_path_info_mpath_lb_update(old_best, false, false, 0);
		bgp_path_info_mpath_dequeue(old_best);
	}

	if (debug)
		zlog_debug(
			"%pRN(%s): starting mpath update, newbest %s num candidates %d old-mpath-count %d old-cum-bw u%" PRIu64,
			bgp_dest_to_rnode(dest), bgp->name_pretty,
			new_best ? new_best->peer->host : "NONE",
			mp_list ? listcount(mp_list) : 0, old_mpath_count,
			old_cum_bw);

	/*
	 * We perform an ordered walk through both lists in parallel.
	 * The reason for the ordered walk is that if there are paths
	 * that were previously multipaths and are still multipaths, the walk
	 * should encounter them in both lists at the same time. Otherwise
	 * there will be paths that are in one list or another, and we
	 * will deal with these separately.
	 *
	 * Note that new_best might be somewhere in the mp_list, so we need
	 * to skip over it
	 */
	all_paths_lb = true; /* We'll reset if any path doesn't have LB. */
	while (mp_node || cur_mpath) {
		struct bgp_path_info *tmp_info;

		/*
		 * We can bail out of this loop if all existing paths on the
		 * multipath list have been visited (for cleanup purposes) and
		 * the maxpath requirement is fulfulled
		 */
		if (!cur_mpath && (mpath_count >= maxpaths))
			break;

		mp_next_node = mp_node ? listnextnode(mp_node) : NULL;
		next_mpath =
			cur_mpath ? bgp_path_info_mpath_next(cur_mpath) : NULL;
		tmp_info = mp_node ? listgetdata(mp_node) : NULL;

		if (debug)
			zlog_debug(
				"%pRN(%s): comparing candidate %s with existing mpath %s",
				bgp_dest_to_rnode(dest), bgp->name_pretty,
				tmp_info ? tmp_info->peer->host : "NONE",
				cur_mpath ? cur_mpath->peer->host : "NONE");

		/*
		 * If equal, the path was a multipath and is still a multipath.
		 * Insert onto new multipath list if maxpaths allows.
		 */
		if (mp_node && (listgetdata(mp_node) == cur_mpath)) {
			list_delete_node(mp_list, mp_node);
			bgp_path_info_mpath_dequeue(cur_mpath);
			if ((mpath_count < maxpaths)
			    && prev_mpath
			    && bgp_path_info_nexthop_cmp(prev_mpath,
							 cur_mpath)) {
				bgp_path_info_mpath_enqueue(prev_mpath,
							    cur_mpath);
				prev_mpath = cur_mpath;
				mpath_count++;
				if (ecommunity_linkbw_present(
					    bgp_attr_get_ecommunity(
						    cur_mpath->attr),
					    &bwval))
					cum_bw += bwval;
				else
					all_paths_lb = false;
				if (debug) {
					bgp_path_info_path_with_addpath_rx_str(
						cur_mpath, path_buf,
						sizeof(path_buf));
					zlog_debug(
						"%pRN: %s is still multipath, cur count %d",
						bgp_dest_to_rnode(dest),
						path_buf, mpath_count);
				}
			} else {
				mpath_changed = 1;
				if (debug) {
					bgp_path_info_path_with_addpath_rx_str(
						cur_mpath, path_buf,
						sizeof(path_buf));
					zlog_debug(
						"%pRN: remove mpath %s nexthop %pI4, cur count %d",
						bgp_dest_to_rnode(dest),
						path_buf,
						&cur_mpath->attr->nexthop,
						mpath_count);
				}
			}
			mp_node = mp_next_node;
			cur_mpath = next_mpath;
			continue;
		}

		if (cur_mpath
		    && (!mp_node
			|| (bgp_path_info_mpath_cmp(cur_mpath,
						    listgetdata(mp_node))
			    < 0))) {
			/*
			 * If here, we have an old multipath and either the
			 * mp_list
			 * is finished or the next mp_node points to a later
			 * multipath, so we need to purge this path from the
			 * multipath list
			 */
			bgp_path_info_mpath_dequeue(cur_mpath);
			mpath_changed = 1;
			if (debug) {
				bgp_path_info_path_with_addpath_rx_str(
					cur_mpath, path_buf, sizeof(path_buf));
				zlog_debug(
					"%pRN: remove mpath %s nexthop %pI4, cur count %d",
					bgp_dest_to_rnode(dest), path_buf,
					&cur_mpath->attr->nexthop, mpath_count);
			}
			cur_mpath = next_mpath;
		} else {
			/*
			 * If here, we have a path on the mp_list that was not
			 * previously
			 * a multipath (due to non-equivalance or maxpaths
			 * exceeded),
			 * or the matching multipath is sorted later in the
			 * multipath
			 * list. Before we enqueue the path on the new multipath
			 * list,
			 * make sure its not on the old_best multipath list or
			 * referenced
			 * via next_mpath:
			 * - If next_mpath points to this new path, update
			 * next_mpath to
			 *   point to the multipath after this one
			 * - Dequeue the path from the multipath list just to
			 * make sure
			 */
			new_mpath = listgetdata(mp_node);
			list_delete_node(mp_list, mp_node);
			assert(new_mpath);
			assert(prev_mpath);
			if ((mpath_count < maxpaths) && (new_mpath != new_best)
			    && bgp_path_info_nexthop_cmp(prev_mpath,
							 new_mpath)) {
				bgp_path_info_mpath_dequeue(new_mpath);

				bgp_path_info_mpath_enqueue(prev_mpath,
							    new_mpath);
				prev_mpath = new_mpath;
				mpath_changed = 1;
				mpath_count++;
				if (ecommunity_linkbw_present(
					    bgp_attr_get_ecommunity(
						    new_mpath->attr),
					    &bwval))
					cum_bw += bwval;
				else
					all_paths_lb = false;
				if (debug) {
					bgp_path_info_path_with_addpath_rx_str(
						new_mpath, path_buf,
						sizeof(path_buf));
					zlog_debug(
						"%pRN: add mpath %s nexthop %pI4, cur count %d",
						bgp_dest_to_rnode(dest),
						path_buf,
						&new_mpath->attr->nexthop,
						mpath_count);
				}
			}
			mp_node = mp_next_node;
		}
	}

	if (new_best) {
		bgp_path_info_mpath_count_set(new_best, mpath_count - 1);
		if (mpath_count <= 1 ||
		    !ecommunity_linkbw_present(
			    bgp_attr_get_ecommunity(new_best->attr), &bwval))
			all_paths_lb = false;
		else
			cum_bw += bwval;
		bgp_path_info_mpath_lb_update(new_best, true,
					      all_paths_lb, cum_bw);

		if (debug)
			zlog_debug(
				"%pRN(%s): New mpath count (incl newbest) %d mpath-change %s all_paths_lb %d cum_bw u%" PRIu64,
				bgp_dest_to_rnode(dest), bgp->name_pretty,
				mpath_count, mpath_changed ? "YES" : "NO",
				all_paths_lb, cum_bw);

		if (mpath_changed
		    || (bgp_path_info_mpath_count(new_best) != old_mpath_count))
			SET_FLAG(new_best->flags, BGP_PATH_MULTIPATH_CHG);
		if ((mpath_count - 1) != old_mpath_count ||
		    old_cum_bw != cum_bw)
			SET_FLAG(new_best->flags, BGP_PATH_LINK_BW_CHG);
	}
}

/*
 * bgp_mp_dmed_deselect
 *
 * Clean up multipath information for BGP_PATH_DMED_SELECTED path that
 * is not selected as best path
 */
void bgp_mp_dmed_deselect(struct bgp_path_info *dmed_best)
{
	struct bgp_path_info *mpinfo, *mpnext;

	if (!dmed_best)
		return;

	for (mpinfo = bgp_path_info_mpath_first(dmed_best); mpinfo;
	     mpinfo = mpnext) {
		mpnext = bgp_path_info_mpath_next(mpinfo);
		bgp_path_info_mpath_dequeue(mpinfo);
	}

	bgp_path_info_mpath_count_set(dmed_best, 0);
	UNSET_FLAG(dmed_best->flags, BGP_PATH_MULTIPATH_CHG);
	UNSET_FLAG(dmed_best->flags, BGP_PATH_LINK_BW_CHG);
	assert(bgp_path_info_mpath_first(dmed_best) == NULL);
}

/*
 * bgp_path_info_mpath_aggregate_update
 *
 * Set the multipath aggregate attribute. We need to see if the
 * aggregate has changed and then set the ATTR_CHANGED flag on the
 * bestpath info so that a peer update will be generated. The
 * change is detected by generating the current attribute,
 * interning it, and then comparing the interned pointer with the
 * current value. We can skip this generate/compare step if there
 * is no change in multipath selection and no attribute change in
 * any multipath.
 */
void bgp_path_info_mpath_aggregate_update(struct bgp_path_info *new_best,
					  struct bgp_path_info *old_best)
{
	struct bgp_path_info *mpinfo;
	struct aspath *aspath;
	struct aspath *asmerge;
	struct attr *new_attr, *old_attr;
	uint8_t origin;
	struct community *community, *commerge;
	struct ecommunity *ecomm, *ecommerge;
	struct lcommunity *lcomm, *lcommerge;
	struct attr attr = {0};

	if (old_best && (old_best != new_best)
	    && (old_attr = bgp_path_info_mpath_attr(old_best))) {
		bgp_attr_unintern(&old_attr);
		bgp_path_info_mpath_attr_set(old_best, NULL);
	}

	if (!new_best)
		return;

	if (!bgp_path_info_mpath_count(new_best)) {
		if ((new_attr = bgp_path_info_mpath_attr(new_best))) {
			bgp_attr_unintern(&new_attr);
			bgp_path_info_mpath_attr_set(new_best, NULL);
			SET_FLAG(new_best->flags, BGP_PATH_ATTR_CHANGED);
		}
		return;
	}

	attr = *new_best->attr;

	if (new_best->peer
	    && CHECK_FLAG(new_best->peer->bgp->flags,
			  BGP_FLAG_MULTIPATH_RELAX_AS_SET)) {

		/* aggregate attribute from multipath constituents */
		aspath = aspath_dup(attr.aspath);
		origin = attr.origin;
		community =
			bgp_attr_get_community(&attr)
				? community_dup(bgp_attr_get_community(&attr))
				: NULL;
		ecomm = (bgp_attr_get_ecommunity(&attr))
				? ecommunity_dup(bgp_attr_get_ecommunity(&attr))
				: NULL;
		lcomm = (bgp_attr_get_lcommunity(&attr))
				? lcommunity_dup(bgp_attr_get_lcommunity(&attr))
				: NULL;

		for (mpinfo = bgp_path_info_mpath_first(new_best); mpinfo;
		     mpinfo = bgp_path_info_mpath_next(mpinfo)) {
			asmerge =
				aspath_aggregate(aspath, mpinfo->attr->aspath);
			aspath_free(aspath);
			aspath = asmerge;

			if (origin < mpinfo->attr->origin)
				origin = mpinfo->attr->origin;

			if (bgp_attr_get_community(mpinfo->attr)) {
				if (community) {
					commerge = community_merge(
						community,
						bgp_attr_get_community(
							mpinfo->attr));
					community =
						community_uniq_sort(commerge);
					community_free(&commerge);
				} else
					community = community_dup(
						bgp_attr_get_community(
							mpinfo->attr));
			}

			if (bgp_attr_get_ecommunity(mpinfo->attr)) {
				if (ecomm) {
					ecommerge = ecommunity_merge(
						ecomm, bgp_attr_get_ecommunity(
							       mpinfo->attr));
					ecomm = ecommunity_uniq_sort(ecommerge);
					ecommunity_free(&ecommerge);
				} else
					ecomm = ecommunity_dup(
						bgp_attr_get_ecommunity(
							mpinfo->attr));
			}
			if (bgp_attr_get_lcommunity(mpinfo->attr)) {
				if (lcomm) {
					lcommerge = lcommunity_merge(
						lcomm, bgp_attr_get_lcommunity(
							       mpinfo->attr));
					lcomm = lcommunity_uniq_sort(lcommerge);
					lcommunity_free(&lcommerge);
				} else
					lcomm = lcommunity_dup(
						bgp_attr_get_lcommunity(
							mpinfo->attr));
			}
		}

		attr.aspath = aspath;
		attr.origin = origin;
		if (community)
			bgp_attr_set_community(&attr, community);
		if (ecomm)
			bgp_attr_set_ecommunity(&attr, ecomm);
		if (lcomm)
			bgp_attr_set_lcommunity(&attr, lcomm);

		/* Zap multipath attr nexthop so we set nexthop to self */
		attr.nexthop.s_addr = INADDR_ANY;
		memset(&attr.mp_nexthop_global, 0, sizeof(struct in6_addr));

		/* TODO: should we set ATOMIC_AGGREGATE and AGGREGATOR? */
	}

	new_attr = bgp_attr_intern(&attr);

	if (new_attr != bgp_path_info_mpath_attr(new_best)) {
		if ((old_attr = bgp_path_info_mpath_attr(new_best)))
			bgp_attr_unintern(&old_attr);
		bgp_path_info_mpath_attr_set(new_best, new_attr);
		SET_FLAG(new_best->flags, BGP_PATH_ATTR_CHANGED);
	} else
		bgp_attr_unintern(&new_attr);
}