summaryrefslogtreecommitdiff
path: root/lib/cfm.c
blob: 71d2c02067e08d6e34366438118eefbf24868bd4 (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
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
/*
 * Copyright (c) 2010, 2011, 2012, 2013, 2014, 2015 Nicira, Inc.
 *
 * 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 <config.h>
#include "cfm.h"

#include <stdint.h>
#include <stdlib.h>
#include <string.h>

#include "byte-order.h"
#include "connectivity.h"
#include "dp-packet.h"
#include "openvswitch/dynamic-string.h"
#include "flow.h"
#include "hash.h"
#include "openvswitch/hmap.h"
#include "netdev.h"
#include "ovs-atomic.h"
#include "packets.h"
#include "openvswitch/poll-loop.h"
#include "random.h"
#include "seq.h"
#include "timer.h"
#include "timeval.h"
#include "unixctl.h"
#include "openvswitch/vlog.h"
#include "util.h"

VLOG_DEFINE_THIS_MODULE(cfm);

#define CFM_MAX_RMPS 256

/* Ethernet destination address of CCM packets. */
static const struct eth_addr eth_addr_ccm = ETH_ADDR_C(01,80,c2,00,00,30);
static const struct eth_addr eth_addr_ccm_x = ETH_ADDR_C(01,23,20,00,00,30);

#define ETH_TYPE_CFM 0x8902

/* A 'ccm' represents a Continuity Check Message from the 802.1ag
 * specification.  Continuity Check Messages are broadcast periodically so that
 * hosts can determine whom they have connectivity to.
 *
 * The minimum length of a CCM as specified by IEEE 802.1ag is 75 bytes.
 * Previous versions of Open vSwitch generated 74-byte CCM messages, so we
 * accept such messages too. */
#define CCM_LEN 75
#define CCM_ACCEPT_LEN 74
#define CCM_MAID_LEN 48
#define CCM_OPCODE 1 /* CFM message opcode meaning CCM. */
#define CCM_RDI_MASK 0x80
#define CFM_HEALTH_INTERVAL 6

OVS_PACKED(
struct ccm {
    uint8_t mdlevel_version; /* MD Level and Version */
    uint8_t opcode;
    uint8_t flags;
    uint8_t tlv_offset;
    ovs_be32 seq;
    ovs_be16 mpid;
    uint8_t maid[CCM_MAID_LEN];

    /* Defined by ITU-T Y.1731 should be zero */
    ovs_be16 interval_ms_x;      /* Transmission interval in ms. */
    ovs_be64 mpid64;             /* MPID in extended mode. */
    uint8_t opdown;              /* Operationally down. */
    uint8_t zero[5];

    /* TLV space. */
    uint8_t end_tlv;
});
BUILD_ASSERT_DECL(CCM_LEN == sizeof(struct ccm));

struct cfm {
    const char *name;           /* Name of this CFM object. */
    struct hmap_node hmap_node; /* Node in all_cfms list. */

    struct netdev *netdev;
    uint64_t rx_packets;        /* Packets received by 'netdev'. */

    uint64_t mpid;
    bool demand;           /* Demand mode. */
    bool booted;           /* A full fault interval has occurred. */
    enum cfm_fault_reason fault;  /* Connectivity fault status. */
    enum cfm_fault_reason recv_fault;  /* Bit mask of faults occurring on
                                          receive. */
    bool opup;             /* Operational State. */
    bool remote_opup;      /* Remote Operational State. */

    int fault_override;    /* Manual override of 'fault' status.
                              Ignored if negative. */

    uint32_t seq;          /* The sequence number of our last CCM. */
    uint8_t ccm_interval;  /* The CCM transmission interval. */
    int ccm_interval_ms;   /* 'ccm_interval' in milliseconds. */
    uint16_t ccm_vlan;     /* Vlan tag of CCM PDUs.  CFM_RANDOM_VLAN if
                              random. */
    uint8_t ccm_pcp;       /* Priority of CCM PDUs. */
    uint8_t maid[CCM_MAID_LEN]; /* The MAID of this CFM. */

    struct timer tx_timer;    /* Send CCM when expired. */
    struct timer fault_timer; /* Check for faults when expired. */

    struct hmap remote_mps;   /* Remote MPs. */

    /* Result of cfm_get_remote_mpids(). Updated only during fault check to
     * avoid flapping. */
    uint64_t *rmps_array;     /* Cache of remote_mps. */
    size_t rmps_array_len;    /* Number of rmps in 'rmps_array'. */

    int health;               /* Percentage of the number of CCM frames
                                 received. */
    int health_interval;      /* Number of fault_intervals since health was
                                 recomputed. */
    long long int last_tx;    /* Last CCM transmission time. */

    /* These bools are atomic to allow readers to check their values
     * without taking 'mutex'.  Such readers do not assume the values they
     * read are synchronized with any other members. */
    atomic_bool check_tnl_key; /* Verify the tunnel key of inbound packets? */
    atomic_bool extended;      /* Extended mode. */
    struct ovs_refcount ref_cnt;

    uint64_t flap_count;       /* Count the flaps since boot. */

    /* True when the variables returned by cfm_get_*() are changed
     * since last check. */
    bool status_changed;

    /* When 'cfm->demand' is set, at least one ccm is required to be received
     * every 100 * cfm_interval.  If ccm is not received within this interval,
     * even if data packets are received, the cfm fault will be set. */
    struct timer demand_rx_ccm_t;
};

/* Remote MPs represent foreign network entities that are configured to have
 * the same MAID as this CFM instance. */
struct remote_mp {
    uint64_t mpid;         /* The Maintenance Point ID of this 'remote_mp'. */
    struct hmap_node node; /* Node in 'remote_mps' map. */

    bool recv;           /* CCM was received since last fault check. */
    bool opup;           /* Operational State. */
    uint32_t seq;        /* Most recently received sequence number. */
    uint8_t num_health_ccm; /* Number of received ccm frames every
                               CFM_HEALTH_INTERVAL * 'fault_interval'. */
    long long int last_rx; /* Last CCM reception time. */

};

static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(20, 30);

static struct ovs_mutex mutex = OVS_MUTEX_INITIALIZER;
static struct hmap all_cfms__ = HMAP_INITIALIZER(&all_cfms__);
static struct hmap *const all_cfms OVS_GUARDED_BY(mutex) = &all_cfms__;

static unixctl_cb_func cfm_unixctl_show;
static unixctl_cb_func cfm_unixctl_set_fault;

static uint64_t
cfm_rx_packets(const struct cfm *cfm) OVS_REQUIRES(mutex)
{
    struct netdev_stats stats;

    if (!netdev_get_stats(cfm->netdev, &stats)) {
        return stats.rx_packets;
    } else {
        return 0;
    }
}

static struct eth_addr
cfm_ccm_addr(struct cfm *cfm)
{
    bool extended;

    atomic_read_relaxed(&cfm->extended, &extended);

    return extended ? eth_addr_ccm_x : eth_addr_ccm;
}

/* Returns the string representation of the given cfm_fault_reason 'reason'. */
const char *
cfm_fault_reason_to_str(int reason)
{
    switch (reason) {
#define CFM_FAULT_REASON(NAME, STR) case CFM_FAULT_##NAME: return #STR;
        CFM_FAULT_REASONS
#undef CFM_FAULT_REASON
    default: return "<unknown>";
    }
}

static void
ds_put_cfm_fault(struct ds *ds, int fault)
{
    int i;

    for (i = 0; i < CFM_FAULT_N_REASONS; i++) {
        int reason = 1 << i;

        if (fault & reason) {
            ds_put_format(ds, "%s ", cfm_fault_reason_to_str(reason));
        }
    }

    ds_chomp(ds, ' ');
}

static void
cfm_generate_maid(struct cfm *cfm) OVS_REQUIRES(mutex)
{
    const char *ovs_md_name = "ovs";
    const char *ovs_ma_name = "ovs";
    uint8_t *ma_p;
    size_t md_len, ma_len;

    memset(cfm->maid, 0, CCM_MAID_LEN);

    md_len = strlen(ovs_md_name);
    ma_len = strlen(ovs_ma_name);

    ovs_assert(md_len && ma_len && md_len + ma_len + 4 <= CCM_MAID_LEN);

    cfm->maid[0] = 4;                           /* MD name string format. */
    cfm->maid[1] = md_len;                      /* MD name size. */
    memcpy(&cfm->maid[2], ovs_md_name, md_len); /* MD name. */

    ma_p = cfm->maid + 2 + md_len;
    ma_p[0] = 2;                           /* MA name string format. */
    ma_p[1] = ma_len;                      /* MA name size. */
    memcpy(&ma_p[2], ovs_ma_name, ma_len); /* MA name. */
}

static int
ccm_interval_to_ms(uint8_t interval)
{
    switch (interval) {
    case 0:  OVS_NOT_REACHED(); /* Explicitly not supported by 802.1ag. */
    case 1:  return 3;      /* Not recommended due to timer resolution. */
    case 2:  return 10;     /* Not recommended due to timer resolution. */
    case 3:  return 100;
    case 4:  return 1000;
    case 5:  return 10000;
    case 6:  return 60000;
    case 7:  return 600000;
    default: OVS_NOT_REACHED(); /* Explicitly not supported by 802.1ag. */
    }

    OVS_NOT_REACHED();
}

static long long int
cfm_fault_interval(struct cfm *cfm) OVS_REQUIRES(mutex)
{
    /* According to the 802.1ag specification we should assume every other MP
     * with the same MAID has the same transmission interval that we have.  If
     * an MP has a different interval, cfm_process_heartbeat will register it
     * as a fault (likely due to a configuration error).  Thus we can check all
     * MPs at once making this quite a bit simpler.
     *
     * When cfm is not in demand mode, we check when (ccm_interval_ms * 3.5) ms
     * have passed.  When cfm is in demand mode, we check when
     * (MAX(ccm_interval_ms, 500) * 3.5) ms have passed.  This ensures that
     * ovs-vswitchd has enough time to pull statistics from the datapath. */

    return (MAX(cfm->ccm_interval_ms, cfm->demand ? 500 : cfm->ccm_interval_ms)
            * 7) / 2;
}

static uint8_t
ms_to_ccm_interval(int interval_ms)
{
    uint8_t i;

    for (i = 7; i > 0; i--) {
        if (ccm_interval_to_ms(i) <= interval_ms) {
            return i;
        }
    }

    return 1;
}

static uint32_t
hash_mpid(uint64_t mpid)
{
    return hash_uint64(mpid);
}

static bool
cfm_is_valid_mpid(bool extended, uint64_t mpid)
{
    /* 802.1ag specification requires MPIDs to be within the range [1, 8191].
     * In extended mode we relax this requirement. */
    return mpid >= 1 && (extended || mpid <= 8191);
}

static struct remote_mp *
lookup_remote_mp(const struct cfm *cfm, uint64_t mpid) OVS_REQUIRES(mutex)
{
    struct remote_mp *rmp;

    HMAP_FOR_EACH_IN_BUCKET (rmp, node, hash_mpid(mpid), &cfm->remote_mps) {
        if (rmp->mpid == mpid) {
            return rmp;
        }
    }

    return NULL;
}

void
cfm_init(void)
{
    unixctl_command_register("cfm/show", "[interface]", 0, 1, cfm_unixctl_show,
                             NULL);
    unixctl_command_register("cfm/set-fault", "[interface] normal|false|true",
                             1, 2, cfm_unixctl_set_fault, NULL);
}

/* Records the status change and changes the global connectivity seq. */
static void
cfm_status_changed(struct cfm *cfm) OVS_REQUIRES(mutex)
{
    seq_change(connectivity_seq_get());
    cfm->status_changed = true;
}

/* Allocates a 'cfm' object called 'name'.  'cfm' should be initialized by
 * cfm_configure() before use. */
struct cfm *
cfm_create(const struct netdev *netdev) OVS_EXCLUDED(mutex)
{
    struct cfm *cfm;

    cfm = xzalloc(sizeof *cfm);
    cfm->netdev = netdev_ref(netdev);
    cfm->name = netdev_get_name(cfm->netdev);
    hmap_init(&cfm->remote_mps);
    cfm->remote_opup = true;
    cfm->fault_override = -1;
    cfm->health = -1;
    cfm->last_tx = 0;
    cfm->flap_count = 0;
    atomic_init(&cfm->extended, false);
    atomic_init(&cfm->check_tnl_key, false);
    ovs_refcount_init(&cfm->ref_cnt);

    ovs_mutex_lock(&mutex);
    cfm_status_changed(cfm);
    cfm_generate_maid(cfm);
    hmap_insert(all_cfms, &cfm->hmap_node, hash_string(cfm->name, 0));
    ovs_mutex_unlock(&mutex);

    return cfm;
}

void
cfm_unref(struct cfm *cfm) OVS_EXCLUDED(mutex)
{
    struct remote_mp *rmp;

    if (!cfm) {
        return;
    }

    if (ovs_refcount_unref_relaxed(&cfm->ref_cnt) != 1) {
        return;
    }

    ovs_mutex_lock(&mutex);
    cfm_status_changed(cfm);
    hmap_remove(all_cfms, &cfm->hmap_node);
    ovs_mutex_unlock(&mutex);

    HMAP_FOR_EACH_POP (rmp, node, &cfm->remote_mps) {
        free(rmp);
    }

    hmap_destroy(&cfm->remote_mps);
    netdev_close(cfm->netdev);
    free(cfm->rmps_array);

    free(cfm);
}

struct cfm *
cfm_ref(const struct cfm *cfm_)
{
    struct cfm *cfm = CONST_CAST(struct cfm *, cfm_);
    if (cfm) {
        ovs_refcount_ref(&cfm->ref_cnt);
    }
    return cfm;
}

/* Should be run periodically to update fault statistics messages. */
void
cfm_run(struct cfm *cfm) OVS_EXCLUDED(mutex)
{
    ovs_mutex_lock(&mutex);
    if (timer_expired(&cfm->fault_timer)) {
        long long int interval = cfm_fault_interval(cfm);
        struct remote_mp *rmp, *rmp_next;
        enum cfm_fault_reason old_cfm_fault = cfm->fault;
        uint64_t old_flap_count = cfm->flap_count;
        int old_health = cfm->health;
        size_t old_rmps_array_len = cfm->rmps_array_len;
        bool old_rmps_deleted = false;
        bool old_rmp_opup = cfm->remote_opup;
        bool demand_override;
        bool rmp_set_opup = false;
        bool rmp_set_opdown = false;

        cfm->fault = cfm->recv_fault;
        cfm->recv_fault = 0;

        cfm->rmps_array_len = 0;
        free(cfm->rmps_array);
        cfm->rmps_array = xmalloc(hmap_count(&cfm->remote_mps) *
                                  sizeof *cfm->rmps_array);

        if (cfm->health_interval == CFM_HEALTH_INTERVAL) {
            /* Calculate the cfm health of the interface.  If the number of
             * remote_mpids of a cfm interface is > 1, the cfm health is
             * undefined. If the number of remote_mpids is 1, the cfm health is
             * the percentage of the ccm frames received in the
             * (CFM_HEALTH_INTERVAL * 3.5)ms, else it is 0. */
            if (hmap_count(&cfm->remote_mps) > 1) {
                cfm->health = -1;
            } else if (hmap_is_empty(&cfm->remote_mps)) {
                cfm->health = 0;
            } else {
                int exp_ccm_recvd;

                rmp = CONTAINER_OF(hmap_first(&cfm->remote_mps),
                                   struct remote_mp, node);
                exp_ccm_recvd = (CFM_HEALTH_INTERVAL * 7) / 2;
                /* Calculate the percentage of healthy ccm frames received.
                 * Since the 'fault_interval' is (3.5 * cfm_interval), and
                 * 1 CCM packet must be received every cfm_interval,
                 * the 'remote_mpid' health reports the percentage of
                 * healthy CCM frames received every
                 * 'CFM_HEALTH_INTERVAL'th 'fault_interval'. */
                cfm->health = (rmp->num_health_ccm * 100) / exp_ccm_recvd;
                cfm->health = MIN(cfm->health, 100);
                rmp->num_health_ccm = 0;
                ovs_assert(cfm->health >= 0 && cfm->health <= 100);
            }
            cfm->health_interval = 0;
        }
        cfm->health_interval++;

        demand_override = false;
        if (cfm->demand) {
            uint64_t rx_packets = cfm_rx_packets(cfm);
            demand_override = hmap_count(&cfm->remote_mps) == 1
                && rx_packets > cfm->rx_packets
                && !timer_expired(&cfm->demand_rx_ccm_t);
            cfm->rx_packets = rx_packets;
        }

        HMAP_FOR_EACH_SAFE (rmp, rmp_next, node, &cfm->remote_mps) {
            if (!rmp->recv) {
                VLOG_INFO("%s: Received no CCM from RMP %"PRIu64" in the last"
                          " %lldms", cfm->name, rmp->mpid,
                          time_msec() - rmp->last_rx);
                if (!demand_override) {
                    old_rmps_deleted = true;
                    hmap_remove(&cfm->remote_mps, &rmp->node);
                    free(rmp);
                }
            } else {
                rmp->recv = false;

                if (rmp->opup) {
                    rmp_set_opup = true;
                } else {
                    rmp_set_opdown = true;
                }

                cfm->rmps_array[cfm->rmps_array_len++] = rmp->mpid;
            }
        }

        if (rmp_set_opdown) {
            cfm->remote_opup = false;
        }
        else if (rmp_set_opup) {
            cfm->remote_opup = true;
        }

        if (hmap_is_empty(&cfm->remote_mps)) {
            cfm->fault |= CFM_FAULT_RECV;
        }

        if (old_cfm_fault != cfm->fault) {
            if (!VLOG_DROP_INFO(&rl)) {
                struct ds ds = DS_EMPTY_INITIALIZER;

                ds_put_cstr(&ds, "from [");
                ds_put_cfm_fault(&ds, old_cfm_fault);
                ds_put_cstr(&ds, "] to [");
                ds_put_cfm_fault(&ds, cfm->fault);
                ds_put_char(&ds, ']');
                VLOG_INFO("%s: CFM faults changed %s.", cfm->name, ds_cstr(&ds));
                ds_destroy(&ds);
            }

            /* If there is a flap, increments the counter. */
            if (old_cfm_fault == 0 || cfm->fault == 0) {
                cfm->flap_count++;
            }
        }

        /* These variables represent the cfm session status, it is desirable
         * to update them to database immediately after change. */
        if (old_health != cfm->health
            || old_rmp_opup != cfm->remote_opup
            || (old_rmps_array_len != cfm->rmps_array_len || old_rmps_deleted)
            || old_cfm_fault != cfm->fault
            || old_flap_count != cfm->flap_count) {
            cfm_status_changed(cfm);
        }

        cfm->booted = true;
        timer_set_duration(&cfm->fault_timer, interval);
        VLOG_DBG("%s: new fault interval", cfm->name);
    }
    ovs_mutex_unlock(&mutex);
}

/* Should be run periodically to check if the CFM module has a CCM message it
 * wishes to send. */
bool
cfm_should_send_ccm(struct cfm *cfm) OVS_EXCLUDED(mutex)
{
    bool ret;

    ovs_mutex_lock(&mutex);
    ret = timer_expired(&cfm->tx_timer);
    ovs_mutex_unlock(&mutex);
    return ret;
}

/* Composes a CCM message into 'packet'.  Messages generated with this function
 * should be sent whenever cfm_should_send_ccm() indicates. */
void
cfm_compose_ccm(struct cfm *cfm, struct dp_packet *packet,
                const struct eth_addr eth_src) OVS_EXCLUDED(mutex)
{
    uint16_t ccm_vlan;
    struct ccm *ccm;
    bool extended;

    ovs_mutex_lock(&mutex);
    timer_set_duration(&cfm->tx_timer, cfm->ccm_interval_ms);
    eth_compose(packet, cfm_ccm_addr(cfm), eth_src, ETH_TYPE_CFM, sizeof *ccm);

    ccm_vlan = (cfm->ccm_vlan != CFM_RANDOM_VLAN
                ? cfm->ccm_vlan
                : random_uint16());
    ccm_vlan = ccm_vlan & VLAN_VID_MASK;

    if (ccm_vlan || cfm->ccm_pcp) {
        uint16_t tci = ccm_vlan | (cfm->ccm_pcp << VLAN_PCP_SHIFT);
        eth_push_vlan(packet, htons(ETH_TYPE_VLAN), htons(tci));
    }

    atomic_read_relaxed(&cfm->extended, &extended);

    ccm = dp_packet_l3(packet);
    ccm->mdlevel_version = 0;
    ccm->opcode = CCM_OPCODE;
    ccm->tlv_offset = 70;
    ccm->seq = htonl(++cfm->seq);
    ccm->flags = cfm->ccm_interval;
    memcpy(ccm->maid, cfm->maid, sizeof ccm->maid);
    memset(ccm->zero, 0, sizeof ccm->zero);
    ccm->end_tlv = 0;

    if (extended) {
        ccm->mpid = htons(hash_mpid(cfm->mpid));
        ccm->mpid64 = htonll(cfm->mpid);
        ccm->opdown = !cfm->opup;
    } else {
        ccm->mpid = htons(cfm->mpid);
        ccm->mpid64 = htonll(0);
        ccm->opdown = 0;
    }

    if (cfm->ccm_interval == 0) {
        ovs_assert(extended);
        ccm->interval_ms_x = htons(cfm->ccm_interval_ms);
    } else {
        ccm->interval_ms_x = htons(0);
    }

    if (cfm->booted && hmap_is_empty(&cfm->remote_mps)) {
        ccm->flags |= CCM_RDI_MASK;
    }

    if (cfm->last_tx) {
        long long int delay = time_msec() - cfm->last_tx;
        if (delay > (cfm->ccm_interval_ms * 3 / 2)) {
            VLOG_INFO("%s: long delay of %lldms (expected %dms) sending CCM"
                      " seq %"PRIu32, cfm->name, delay, cfm->ccm_interval_ms,
                      cfm->seq);
        }
    }
    cfm->last_tx = time_msec();
    ovs_mutex_unlock(&mutex);
}

long long int
cfm_wait(struct cfm *cfm) OVS_EXCLUDED(mutex)
{
    long long int wake_time = cfm_wake_time(cfm);
    poll_timer_wait_until(wake_time);
    return wake_time;
}


/* Returns the next cfm wakeup time. */
long long int
cfm_wake_time(struct cfm *cfm) OVS_EXCLUDED(mutex)
{
    long long int retval;

    if (!cfm) {
        return LLONG_MAX;
    }

    ovs_mutex_lock(&mutex);
    retval = MIN(cfm->tx_timer.t, cfm->fault_timer.t);
    ovs_mutex_unlock(&mutex);
    return retval;
}


/* Configures 'cfm' with settings from 's'. */
bool
cfm_configure(struct cfm *cfm, const struct cfm_settings *s)
    OVS_EXCLUDED(mutex)
{
    uint8_t interval;
    int interval_ms;

    if (!cfm_is_valid_mpid(s->extended, s->mpid) || s->interval <= 0) {
        return false;
    }

    ovs_mutex_lock(&mutex);
    cfm->mpid = s->mpid;
    cfm->opup = s->opup;
    interval = ms_to_ccm_interval(s->interval);
    interval_ms = ccm_interval_to_ms(interval);

    atomic_store_relaxed(&cfm->check_tnl_key, s->check_tnl_key);
    atomic_store_relaxed(&cfm->extended, s->extended);

    cfm->ccm_vlan = s->ccm_vlan;
    cfm->ccm_pcp = s->ccm_pcp & (VLAN_PCP_MASK >> VLAN_PCP_SHIFT);
    if (s->extended && interval_ms != s->interval) {
        interval = 0;
        interval_ms = MIN(s->interval, UINT16_MAX);
    }

    if (s->extended && s->demand) {
        if (!cfm->demand) {
            cfm->demand = true;
            cfm->rx_packets = cfm_rx_packets(cfm);
        }
    } else {
        cfm->demand = false;
    }

    if (interval != cfm->ccm_interval || interval_ms != cfm->ccm_interval_ms) {
        cfm->ccm_interval = interval;
        cfm->ccm_interval_ms = interval_ms;

        timer_set_expired(&cfm->tx_timer);
        timer_set_duration(&cfm->fault_timer, cfm_fault_interval(cfm));
    }

    ovs_mutex_unlock(&mutex);
    return true;
}

/* Must be called when the netdev owned by 'cfm' should change. */
void
cfm_set_netdev(struct cfm *cfm, const struct netdev *netdev)
    OVS_EXCLUDED(mutex)
{
    ovs_mutex_lock(&mutex);
    if (cfm->netdev != netdev) {
        netdev_close(cfm->netdev);
        cfm->netdev = netdev_ref(netdev);
    }
    ovs_mutex_unlock(&mutex);
}

/* Returns true if 'cfm' should process packets from 'flow'.  Sets
 * fields in 'wc' that were used to make the determination. */
bool
cfm_should_process_flow(const struct cfm *cfm_, const struct flow *flow,
                        struct flow_wildcards *wc)
{
    struct cfm *cfm = CONST_CAST(struct cfm *, cfm_);
    bool check_tnl_key;

    /* Most packets are not CFM. */
    if (OVS_LIKELY(flow->dl_type != htons(ETH_TYPE_CFM))) {
        return false;
    }

    memset(&wc->masks.dl_dst, 0xff, sizeof wc->masks.dl_dst);
    if (OVS_UNLIKELY(!eth_addr_equals(flow->dl_dst, cfm_ccm_addr(cfm)))) {
        return false;
    }

    atomic_read_relaxed(&cfm->check_tnl_key, &check_tnl_key);

    if (check_tnl_key) {
        memset(&wc->masks.tunnel.tun_id, 0xff, sizeof wc->masks.tunnel.tun_id);
        return flow->tunnel.tun_id == htonll(0);
    }
    return true;
}

/* Updates internal statistics relevant to packet 'p'.  Should be called on
 * every packet whose flow returned true when passed to
 * cfm_should_process_flow. */
void
cfm_process_heartbeat(struct cfm *cfm, const struct dp_packet *p)
    OVS_EXCLUDED(mutex)
{
    struct ccm *ccm;
    struct eth_header *eth;
    bool extended;

    ovs_mutex_lock(&mutex);

    atomic_read_relaxed(&cfm->extended, &extended);

    eth = dp_packet_eth(p);
    ccm = dp_packet_at(p, (uint8_t *)dp_packet_l3(p) - (uint8_t *)dp_packet_data(p),
                    CCM_ACCEPT_LEN);

    if (!ccm) {
        VLOG_INFO_RL(&rl, "%s: Received an unparseable 802.1ag CCM heartbeat.",
                     cfm->name);
        goto out;
    }

    if (ccm->opcode != CCM_OPCODE) {
        VLOG_INFO_RL(&rl, "%s: Received an unsupported 802.1ag message. "
                     "(opcode %u)", cfm->name, ccm->opcode);
        goto out;
    }

    /* According to the 802.1ag specification, reception of a CCM with an
     * incorrect ccm_interval, unexpected MAID, or unexpected MPID should
     * trigger a fault.  We ignore this requirement for several reasons.
     *
     * Faults can cause a controller or Open vSwitch to make potentially
     * expensive changes to the network topology.  It seems prudent to trigger
     * them judiciously, especially when CFM is used to check slave status of
     * bonds. Furthermore, faults can be maliciously triggered by crafting
     * unexpected CCMs. */
    if (memcmp(ccm->maid, cfm->maid, sizeof ccm->maid)) {
        cfm->recv_fault |= CFM_FAULT_MAID;
        VLOG_WARN_RL(&rl, "%s: Received unexpected remote MAID from MAC "
                     ETH_ADDR_FMT, cfm->name, ETH_ADDR_ARGS(eth->eth_src));
    } else {
        uint8_t ccm_interval = ccm->flags & 0x7;
        bool ccm_rdi = ccm->flags & CCM_RDI_MASK;
        uint16_t ccm_interval_ms_x = ntohs(ccm->interval_ms_x);

        struct remote_mp *rmp;
        uint64_t ccm_mpid;
        uint32_t ccm_seq;
        bool ccm_opdown;
        enum cfm_fault_reason cfm_fault = 0;

        if (extended) {
            ccm_mpid = ntohll(ccm->mpid64);
            ccm_opdown = ccm->opdown;
        } else {
            ccm_mpid = ntohs(ccm->mpid);
            ccm_opdown = false;
        }
        ccm_seq = ntohl(ccm->seq);

        if (ccm_interval != cfm->ccm_interval) {
            VLOG_WARN_RL(&rl, "%s: received a CCM with an unexpected interval"
                         " (%"PRIu8") from RMP %"PRIu64, cfm->name,
                         ccm_interval, ccm_mpid);
        }

        if (extended && ccm_interval == 0
            && ccm_interval_ms_x != cfm->ccm_interval_ms) {
            VLOG_WARN_RL(&rl, "%s: received a CCM with an unexpected extended"
                         " interval (%"PRIu16"ms) from RMP %"PRIu64, cfm->name,
                         ccm_interval_ms_x, ccm_mpid);
        }

        rmp = lookup_remote_mp(cfm, ccm_mpid);
        if (!rmp) {
            if (hmap_count(&cfm->remote_mps) < CFM_MAX_RMPS) {
                rmp = xzalloc(sizeof *rmp);
                hmap_insert(&cfm->remote_mps, &rmp->node, hash_mpid(ccm_mpid));
            } else {
                cfm_fault |= CFM_FAULT_OVERFLOW;
                VLOG_WARN_RL(&rl,
                             "%s: dropped CCM with MPID %"PRIu64" from MAC "
                             ETH_ADDR_FMT, cfm->name, ccm_mpid,
                             ETH_ADDR_ARGS(eth->eth_src));
            }
        }

        if (ccm_rdi) {
            cfm_fault |= CFM_FAULT_RDI;
            VLOG_DBG("%s: RDI bit flagged from RMP %"PRIu64, cfm->name,
                     ccm_mpid);
        }

        VLOG_DBG("%s: received CCM (seq %"PRIu32") (mpid %"PRIu64")"
                 " (interval %"PRIu8") (RDI %s)", cfm->name, ccm_seq,
                 ccm_mpid, ccm_interval, ccm_rdi ? "true" : "false");

        if (rmp) {
            if (rmp->mpid == cfm->mpid) {
                cfm_fault |= CFM_FAULT_LOOPBACK;
                VLOG_WARN_RL(&rl,"%s: received CCM with local MPID"
                             " %"PRIu64, cfm->name, rmp->mpid);
            }

            if (rmp->seq && ccm_seq != (rmp->seq + 1)) {
                VLOG_WARN_RL(&rl, "%s: (mpid %"PRIu64") detected sequence"
                             " numbers which indicate possible connectivity"
                             " problems (previous %"PRIu32") (current %"PRIu32
                             ")", cfm->name, ccm_mpid, rmp->seq, ccm_seq);
            }

            rmp->mpid = ccm_mpid;
            if (!cfm_fault) {
                rmp->num_health_ccm++;
                if (cfm->demand) {
                    timer_set_duration(&cfm->demand_rx_ccm_t,
                                       100 * cfm->ccm_interval_ms);
                }
            }
            rmp->recv = true;
            cfm->recv_fault |= cfm_fault;
            rmp->seq = ccm_seq;
            rmp->opup = !ccm_opdown;
            rmp->last_rx = time_msec();
        }
    }

out:
    ovs_mutex_unlock(&mutex);
}

/* Returns and resets the 'cfm->status_changed'. */
bool
cfm_check_status_change(struct cfm *cfm) OVS_EXCLUDED(mutex)
{
    bool ret;

    ovs_mutex_lock(&mutex);
    ret = cfm->status_changed;
    cfm->status_changed = false;
    ovs_mutex_unlock(&mutex);

    return ret;
}

static int
cfm_get_fault__(const struct cfm *cfm) OVS_REQUIRES(mutex)
{
    if (cfm->fault_override >= 0) {
        return cfm->fault_override ? CFM_FAULT_OVERRIDE : 0;
    }
    return cfm->fault;
}

/* Gets the fault status of 'cfm'.  Returns a bit mask of 'cfm_fault_reason's
 * indicating the cause of the connectivity fault, or zero if there is no
 * fault. */
int
cfm_get_fault(const struct cfm *cfm) OVS_EXCLUDED(mutex)
{
    int fault;

    ovs_mutex_lock(&mutex);
    fault = cfm_get_fault__(cfm);
    ovs_mutex_unlock(&mutex);
    return fault;
}

/* Gets the number of cfm fault flapping since start. */
uint64_t
cfm_get_flap_count(const struct cfm *cfm) OVS_EXCLUDED(mutex)
{
    uint64_t flap_count;
    ovs_mutex_lock(&mutex);
    flap_count = cfm->flap_count;
    ovs_mutex_unlock(&mutex);
    return flap_count;
}

/* Gets the health of 'cfm'.  Returns an integer between 0 and 100 indicating
 * the health of the link as a percentage of ccm frames received in
 * CFM_HEALTH_INTERVAL * 'fault_interval' if there is only 1 remote_mpid,
 * returns 0 if there are no remote_mpids, and returns -1 if there are more
 * than 1 remote_mpids. */
int
cfm_get_health(const struct cfm *cfm) OVS_EXCLUDED(mutex)
{
    int health;

    ovs_mutex_lock(&mutex);
    health = cfm->health;
    ovs_mutex_unlock(&mutex);
    return health;
}

static int
cfm_get_opup__(const struct cfm *cfm_) OVS_REQUIRES(mutex)
{
    struct cfm *cfm = CONST_CAST(struct cfm *, cfm_);
    bool extended;

    atomic_read_relaxed(&cfm->extended, &extended);

    return extended ? cfm->remote_opup : -1;
}

/* Gets the operational state of 'cfm'.  'cfm' is considered operationally down
 * if it has received a CCM with the operationally down bit set from any of its
 * remote maintenance points. Returns 1 if 'cfm' is operationally up, 0 if
 * 'cfm' is operationally down, or -1 if 'cfm' has no operational state
 * (because it isn't in extended mode). */
int
cfm_get_opup(const struct cfm *cfm) OVS_EXCLUDED(mutex)
{
    int opup;

    ovs_mutex_lock(&mutex);
    opup = cfm_get_opup__(cfm);
    ovs_mutex_unlock(&mutex);

    return opup;
}

static void
cfm_get_remote_mpids__(const struct cfm *cfm, uint64_t **rmps, size_t *n_rmps)
    OVS_REQUIRES(mutex)
{
    *rmps = xmemdup(cfm->rmps_array, cfm->rmps_array_len * sizeof **rmps);
    *n_rmps = cfm->rmps_array_len;
}

/* Populates 'rmps' with an array of remote maintenance points reachable by
 * 'cfm'. The number of remote maintenance points is written to 'n_rmps'.
 * 'cfm' retains ownership of the array written to 'rmps' */
void
cfm_get_remote_mpids(const struct cfm *cfm, uint64_t **rmps, size_t *n_rmps)
    OVS_EXCLUDED(mutex)
{
    ovs_mutex_lock(&mutex);
    cfm_get_remote_mpids__(cfm, rmps, n_rmps);
    ovs_mutex_unlock(&mutex);
}

/* Extracts the status of 'cfm' and fills in the 's'. */
void
cfm_get_status(const struct cfm *cfm, struct cfm_status *s) OVS_EXCLUDED(mutex)
{
    ovs_mutex_lock(&mutex);
    s->faults = cfm_get_fault__(cfm);
    s->remote_opstate = cfm_get_opup__(cfm);
    s->flap_count = cfm->flap_count;
    s->health = cfm->health;
    cfm_get_remote_mpids__(cfm, &s->rmps, &s->n_rmps);
    ovs_mutex_unlock(&mutex);
}

static struct cfm *
cfm_find(const char *name) OVS_REQUIRES(mutex)
{
    struct cfm *cfm;

    HMAP_FOR_EACH_WITH_HASH (cfm, hmap_node, hash_string(name, 0), all_cfms) {
        if (!strcmp(cfm->name, name)) {
            return cfm;
        }
    }
    return NULL;
}

static void
cfm_print_details(struct ds *ds, struct cfm *cfm) OVS_REQUIRES(mutex)
{
    struct remote_mp *rmp;
    bool extended;
    int fault;

    atomic_read_relaxed(&cfm->extended, &extended);

    ds_put_format(ds, "---- %s ----\n", cfm->name);
    ds_put_format(ds, "MPID %"PRIu64":%s%s\n", cfm->mpid,
                  extended ? " extended" : "",
                  cfm->fault_override >= 0 ? " fault_override" : "");

    fault = cfm_get_fault__(cfm);
    if (fault) {
        ds_put_cstr(ds, "  fault: ");
        ds_put_cfm_fault(ds, fault);
        ds_put_cstr(ds, "\n");
    }

    if (cfm->health == -1) {
        ds_put_format(ds, "  average health: undefined\n");
    } else {
        ds_put_format(ds, "  average health: %d\n", cfm->health);
    }
    ds_put_format(ds, "  opstate: %s\n", cfm->opup ? "up" : "down");
    ds_put_format(ds, "  remote_opstate: %s\n",
                  cfm->remote_opup ? "up" : "down");
    ds_put_format(ds, "  interval: %dms\n", cfm->ccm_interval_ms);
    ds_put_format(ds, "  next CCM tx: %lldms\n",
                  timer_msecs_until_expired(&cfm->tx_timer));
    ds_put_format(ds, "  next fault check: %lldms\n",
                  timer_msecs_until_expired(&cfm->fault_timer));

    HMAP_FOR_EACH (rmp, node, &cfm->remote_mps) {
        ds_put_format(ds, "Remote MPID %"PRIu64"\n", rmp->mpid);
        ds_put_format(ds, "  recv since check: %s\n",
                      rmp->recv ? "true" : "false");
        ds_put_format(ds, "  opstate: %s\n", rmp->opup? "up" : "down");
    }
}

static void
cfm_unixctl_show(struct unixctl_conn *conn, int argc, const char *argv[],
                 void *aux OVS_UNUSED) OVS_EXCLUDED(mutex)
{
    struct ds ds = DS_EMPTY_INITIALIZER;
    struct cfm *cfm;

    ovs_mutex_lock(&mutex);
    if (argc > 1) {
        cfm = cfm_find(argv[1]);
        if (!cfm) {
            unixctl_command_reply_error(conn, "no such CFM object");
            goto out;
        }
        cfm_print_details(&ds, cfm);
    } else {
        HMAP_FOR_EACH (cfm, hmap_node, all_cfms) {
            cfm_print_details(&ds, cfm);
        }
    }

    unixctl_command_reply(conn, ds_cstr(&ds));
    ds_destroy(&ds);
out:
    ovs_mutex_unlock(&mutex);
}

static void
cfm_unixctl_set_fault(struct unixctl_conn *conn, int argc, const char *argv[],
                      void *aux OVS_UNUSED) OVS_EXCLUDED(mutex)
{
    const char *fault_str = argv[argc - 1];
    int fault_override;
    struct cfm *cfm;

    ovs_mutex_lock(&mutex);
    if (!strcasecmp("true", fault_str)) {
        fault_override = 1;
    } else if (!strcasecmp("false", fault_str)) {
        fault_override = 0;
    } else if (!strcasecmp("normal", fault_str)) {
        fault_override = -1;
    } else {
        unixctl_command_reply_error(conn, "unknown fault string");
        goto out;
    }

    if (argc > 2) {
        cfm = cfm_find(argv[1]);
        if (!cfm) {
            unixctl_command_reply_error(conn, "no such CFM object");
            goto out;
        }
        cfm->fault_override = fault_override;
        cfm_status_changed(cfm);
    } else {
        HMAP_FOR_EACH (cfm, hmap_node, all_cfms) {
            cfm->fault_override = fault_override;
            cfm_status_changed(cfm);
        }
    }

    unixctl_command_reply(conn, "OK");

out:
    ovs_mutex_unlock(&mutex);
}