summaryrefslogtreecommitdiff
path: root/lib/dpif.c
blob: 64e6a0e3a3507963d5f8fd083ecd0ea9f4fdb7ce (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
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
/*
 * Copyright (c) 2008, 2009, 2010, 2011, 2012, 2013, 2014 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 "dpif-provider.h"

#include <ctype.h>
#include <errno.h>
#include <inttypes.h>
#include <stdlib.h>
#include <string.h>

#include "coverage.h"
#include "dpctl.h"
#include "dynamic-string.h"
#include "flow.h"
#include "netdev.h"
#include "netlink.h"
#include "odp-execute.h"
#include "odp-util.h"
#include "ofp-errors.h"
#include "ofp-print.h"
#include "ofp-util.h"
#include "ofpbuf.h"
#include "packet-dpif.h"
#include "packets.h"
#include "poll-loop.h"
#include "shash.h"
#include "sset.h"
#include "timeval.h"
#include "util.h"
#include "valgrind.h"
#include "vlog.h"

VLOG_DEFINE_THIS_MODULE(dpif);

COVERAGE_DEFINE(dpif_destroy);
COVERAGE_DEFINE(dpif_port_add);
COVERAGE_DEFINE(dpif_port_del);
COVERAGE_DEFINE(dpif_flow_flush);
COVERAGE_DEFINE(dpif_flow_get);
COVERAGE_DEFINE(dpif_flow_put);
COVERAGE_DEFINE(dpif_flow_del);
COVERAGE_DEFINE(dpif_execute);
COVERAGE_DEFINE(dpif_purge);
COVERAGE_DEFINE(dpif_execute_with_help);

static const struct dpif_class *base_dpif_classes[] = {
#if defined(__linux__) || defined(_WIN32)
    &dpif_netlink_class,
#endif
    &dpif_netdev_class,
};

struct registered_dpif_class {
    const struct dpif_class *dpif_class;
    int refcount;
};
static struct shash dpif_classes = SHASH_INITIALIZER(&dpif_classes);
static struct sset dpif_blacklist = SSET_INITIALIZER(&dpif_blacklist);

/* Protects 'dpif_classes', including the refcount, and 'dpif_blacklist'. */
static struct ovs_mutex dpif_mutex = OVS_MUTEX_INITIALIZER;

/* Rate limit for individual messages going to or from the datapath, output at
 * DBG level.  This is very high because, if these are enabled, it is because
 * we really need to see them. */
static struct vlog_rate_limit dpmsg_rl = VLOG_RATE_LIMIT_INIT(600, 600);

/* Not really much point in logging many dpif errors. */
static struct vlog_rate_limit error_rl = VLOG_RATE_LIMIT_INIT(60, 5);

static void log_flow_message(const struct dpif *dpif, int error,
                             const char *operation,
                             const struct nlattr *key, size_t key_len,
                             const struct nlattr *mask, size_t mask_len,
                             const struct dpif_flow_stats *stats,
                             const struct nlattr *actions, size_t actions_len);
static void log_operation(const struct dpif *, const char *operation,
                          int error);
static bool should_log_flow_message(int error);
static void log_flow_put_message(struct dpif *, const struct dpif_flow_put *,
                                 int error);
static void log_flow_del_message(struct dpif *, const struct dpif_flow_del *,
                                 int error);
static void log_execute_message(struct dpif *, const struct dpif_execute *,
                                bool subexecute, int error);
static void log_flow_get_message(const struct dpif *,
                                 const struct dpif_flow_get *, int error);

static void
dp_initialize(void)
{
    static struct ovsthread_once once = OVSTHREAD_ONCE_INITIALIZER;

    if (ovsthread_once_start(&once)) {
        int i;

        for (i = 0; i < ARRAY_SIZE(base_dpif_classes); i++) {
            dp_register_provider(base_dpif_classes[i]);
        }
        dpctl_unixctl_register();
        ovsthread_once_done(&once);
    }
}

static int
dp_register_provider__(const struct dpif_class *new_class)
{
    struct registered_dpif_class *registered_class;

    if (sset_contains(&dpif_blacklist, new_class->type)) {
        VLOG_DBG("attempted to register blacklisted provider: %s",
                 new_class->type);
        return EINVAL;
    }

    if (shash_find(&dpif_classes, new_class->type)) {
        VLOG_WARN("attempted to register duplicate datapath provider: %s",
                  new_class->type);
        return EEXIST;
    }

    registered_class = xmalloc(sizeof *registered_class);
    registered_class->dpif_class = new_class;
    registered_class->refcount = 0;

    shash_add(&dpif_classes, new_class->type, registered_class);

    return 0;
}

/* Registers a new datapath provider.  After successful registration, new
 * datapaths of that type can be opened using dpif_open(). */
int
dp_register_provider(const struct dpif_class *new_class)
{
    int error;

    ovs_mutex_lock(&dpif_mutex);
    error = dp_register_provider__(new_class);
    ovs_mutex_unlock(&dpif_mutex);

    return error;
}

/* Unregisters a datapath provider.  'type' must have been previously
 * registered and not currently be in use by any dpifs.  After unregistration
 * new datapaths of that type cannot be opened using dpif_open(). */
static int
dp_unregister_provider__(const char *type)
{
    struct shash_node *node;
    struct registered_dpif_class *registered_class;

    node = shash_find(&dpif_classes, type);
    if (!node) {
        VLOG_WARN("attempted to unregister a datapath provider that is not "
                  "registered: %s", type);
        return EAFNOSUPPORT;
    }

    registered_class = node->data;
    if (registered_class->refcount) {
        VLOG_WARN("attempted to unregister in use datapath provider: %s", type);
        return EBUSY;
    }

    shash_delete(&dpif_classes, node);
    free(registered_class);

    return 0;
}

/* Unregisters a datapath provider.  'type' must have been previously
 * registered and not currently be in use by any dpifs.  After unregistration
 * new datapaths of that type cannot be opened using dpif_open(). */
int
dp_unregister_provider(const char *type)
{
    int error;

    dp_initialize();

    ovs_mutex_lock(&dpif_mutex);
    error = dp_unregister_provider__(type);
    ovs_mutex_unlock(&dpif_mutex);

    return error;
}

/* Blacklists a provider.  Causes future calls of dp_register_provider() with
 * a dpif_class which implements 'type' to fail. */
void
dp_blacklist_provider(const char *type)
{
    ovs_mutex_lock(&dpif_mutex);
    sset_add(&dpif_blacklist, type);
    ovs_mutex_unlock(&dpif_mutex);
}

/* Clears 'types' and enumerates the types of all currently registered datapath
 * providers into it.  The caller must first initialize the sset. */
void
dp_enumerate_types(struct sset *types)
{
    struct shash_node *node;

    dp_initialize();
    sset_clear(types);

    ovs_mutex_lock(&dpif_mutex);
    SHASH_FOR_EACH(node, &dpif_classes) {
        const struct registered_dpif_class *registered_class = node->data;
        sset_add(types, registered_class->dpif_class->type);
    }
    ovs_mutex_unlock(&dpif_mutex);
}

static void
dp_class_unref(struct registered_dpif_class *rc)
{
    ovs_mutex_lock(&dpif_mutex);
    ovs_assert(rc->refcount);
    rc->refcount--;
    ovs_mutex_unlock(&dpif_mutex);
}

static struct registered_dpif_class *
dp_class_lookup(const char *type)
{
    struct registered_dpif_class *rc;

    ovs_mutex_lock(&dpif_mutex);
    rc = shash_find_data(&dpif_classes, type);
    if (rc) {
        rc->refcount++;
    }
    ovs_mutex_unlock(&dpif_mutex);

    return rc;
}

/* Clears 'names' and enumerates the names of all known created datapaths with
 * the given 'type'.  The caller must first initialize the sset.  Returns 0 if
 * successful, otherwise a positive errno value.
 *
 * Some kinds of datapaths might not be practically enumerable.  This is not
 * considered an error. */
int
dp_enumerate_names(const char *type, struct sset *names)
{
    struct registered_dpif_class *registered_class;
    const struct dpif_class *dpif_class;
    int error;

    dp_initialize();
    sset_clear(names);

    registered_class = dp_class_lookup(type);
    if (!registered_class) {
        VLOG_WARN("could not enumerate unknown type: %s", type);
        return EAFNOSUPPORT;
    }

    dpif_class = registered_class->dpif_class;
    error = (dpif_class->enumerate
             ? dpif_class->enumerate(names, dpif_class)
             : 0);
    if (error) {
        VLOG_WARN("failed to enumerate %s datapaths: %s", dpif_class->type,
                   ovs_strerror(error));
    }
    dp_class_unref(registered_class);

    return error;
}

/* Parses 'datapath_name_', which is of the form [type@]name into its
 * component pieces.  'name' and 'type' must be freed by the caller.
 *
 * The returned 'type' is normalized, as if by dpif_normalize_type(). */
void
dp_parse_name(const char *datapath_name_, char **name, char **type)
{
    char *datapath_name = xstrdup(datapath_name_);
    char *separator;

    separator = strchr(datapath_name, '@');
    if (separator) {
        *separator = '\0';
        *type = datapath_name;
        *name = xstrdup(dpif_normalize_type(separator + 1));
    } else {
        *name = datapath_name;
        *type = xstrdup(dpif_normalize_type(NULL));
    }
}

static int
do_open(const char *name, const char *type, bool create, struct dpif **dpifp)
{
    struct dpif *dpif = NULL;
    int error;
    struct registered_dpif_class *registered_class;

    dp_initialize();

    type = dpif_normalize_type(type);
    registered_class = dp_class_lookup(type);
    if (!registered_class) {
        VLOG_WARN("could not create datapath %s of unknown type %s", name,
                  type);
        error = EAFNOSUPPORT;
        goto exit;
    }

    error = registered_class->dpif_class->open(registered_class->dpif_class,
                                               name, create, &dpif);
    if (!error) {
        ovs_assert(dpif->dpif_class == registered_class->dpif_class);
    } else {
        dp_class_unref(registered_class);
    }

exit:
    *dpifp = error ? NULL : dpif;
    return error;
}

/* Tries to open an existing datapath named 'name' and type 'type'.  Will fail
 * if no datapath with 'name' and 'type' exists.  'type' may be either NULL or
 * the empty string to specify the default system type.  Returns 0 if
 * successful, otherwise a positive errno value.  On success stores a pointer
 * to the datapath in '*dpifp', otherwise a null pointer. */
int
dpif_open(const char *name, const char *type, struct dpif **dpifp)
{
    return do_open(name, type, false, dpifp);
}

/* Tries to create and open a new datapath with the given 'name' and 'type'.
 * 'type' may be either NULL or the empty string to specify the default system
 * type.  Will fail if a datapath with 'name' and 'type' already exists.
 * Returns 0 if successful, otherwise a positive errno value.  On success
 * stores a pointer to the datapath in '*dpifp', otherwise a null pointer. */
int
dpif_create(const char *name, const char *type, struct dpif **dpifp)
{
    return do_open(name, type, true, dpifp);
}

/* Tries to open a datapath with the given 'name' and 'type', creating it if it
 * does not exist.  'type' may be either NULL or the empty string to specify
 * the default system type.  Returns 0 if successful, otherwise a positive
 * errno value. On success stores a pointer to the datapath in '*dpifp',
 * otherwise a null pointer. */
int
dpif_create_and_open(const char *name, const char *type, struct dpif **dpifp)
{
    int error;

    error = dpif_create(name, type, dpifp);
    if (error == EEXIST || error == EBUSY) {
        error = dpif_open(name, type, dpifp);
        if (error) {
            VLOG_WARN("datapath %s already exists but cannot be opened: %s",
                      name, ovs_strerror(error));
        }
    } else if (error) {
        VLOG_WARN("failed to create datapath %s: %s",
                  name, ovs_strerror(error));
    }
    return error;
}

/* Closes and frees the connection to 'dpif'.  Does not destroy the datapath
 * itself; call dpif_delete() first, instead, if that is desirable. */
void
dpif_close(struct dpif *dpif)
{
    if (dpif) {
        struct registered_dpif_class *rc;

        rc = shash_find_data(&dpif_classes, dpif->dpif_class->type);
        dpif_uninit(dpif, true);
        dp_class_unref(rc);
    }
}

/* Performs periodic work needed by 'dpif'. */
void
dpif_run(struct dpif *dpif)
{
    if (dpif->dpif_class->run) {
        dpif->dpif_class->run(dpif);
    }
}

/* Arranges for poll_block() to wake up when dp_run() needs to be called for
 * 'dpif'. */
void
dpif_wait(struct dpif *dpif)
{
    if (dpif->dpif_class->wait) {
        dpif->dpif_class->wait(dpif);
    }
}

/* Returns the name of datapath 'dpif' prefixed with the type
 * (for use in log messages). */
const char *
dpif_name(const struct dpif *dpif)
{
    return dpif->full_name;
}

/* Returns the name of datapath 'dpif' without the type
 * (for use in device names). */
const char *
dpif_base_name(const struct dpif *dpif)
{
    return dpif->base_name;
}

/* Returns the type of datapath 'dpif'. */
const char *
dpif_type(const struct dpif *dpif)
{
    return dpif->dpif_class->type;
}

/* Returns the fully spelled out name for the given datapath 'type'.
 *
 * Normalized type string can be compared with strcmp().  Unnormalized type
 * string might be the same even if they have different spellings. */
const char *
dpif_normalize_type(const char *type)
{
    return type && type[0] ? type : "system";
}

/* Destroys the datapath that 'dpif' is connected to, first removing all of its
 * ports.  After calling this function, it does not make sense to pass 'dpif'
 * to any functions other than dpif_name() or dpif_close(). */
int
dpif_delete(struct dpif *dpif)
{
    int error;

    COVERAGE_INC(dpif_destroy);

    error = dpif->dpif_class->destroy(dpif);
    log_operation(dpif, "delete", error);
    return error;
}

/* Retrieves statistics for 'dpif' into 'stats'.  Returns 0 if successful,
 * otherwise a positive errno value. */
int
dpif_get_dp_stats(const struct dpif *dpif, struct dpif_dp_stats *stats)
{
    int error = dpif->dpif_class->get_stats(dpif, stats);
    if (error) {
        memset(stats, 0, sizeof *stats);
    }
    log_operation(dpif, "get_stats", error);
    return error;
}

const char *
dpif_port_open_type(const char *datapath_type, const char *port_type)
{
    struct registered_dpif_class *rc;

    datapath_type = dpif_normalize_type(datapath_type);

    ovs_mutex_lock(&dpif_mutex);
    rc = shash_find_data(&dpif_classes, datapath_type);
    if (rc && rc->dpif_class->port_open_type) {
        port_type = rc->dpif_class->port_open_type(rc->dpif_class, port_type);
    }
    ovs_mutex_unlock(&dpif_mutex);

    return port_type;
}

/* Attempts to add 'netdev' as a port on 'dpif'.  If 'port_nop' is
 * non-null and its value is not ODPP_NONE, then attempts to use the
 * value as the port number.
 *
 * If successful, returns 0 and sets '*port_nop' to the new port's port
 * number (if 'port_nop' is non-null).  On failure, returns a positive
 * errno value and sets '*port_nop' to ODPP_NONE (if 'port_nop' is
 * non-null). */
int
dpif_port_add(struct dpif *dpif, struct netdev *netdev, odp_port_t *port_nop)
{
    const char *netdev_name = netdev_get_name(netdev);
    odp_port_t port_no = ODPP_NONE;
    int error;

    COVERAGE_INC(dpif_port_add);

    if (port_nop) {
        port_no = *port_nop;
    }

    error = dpif->dpif_class->port_add(dpif, netdev, &port_no);
    if (!error) {
        VLOG_DBG_RL(&dpmsg_rl, "%s: added %s as port %"PRIu32,
                    dpif_name(dpif), netdev_name, port_no);
    } else {
        VLOG_WARN_RL(&error_rl, "%s: failed to add %s as port: %s",
                     dpif_name(dpif), netdev_name, ovs_strerror(error));
        port_no = ODPP_NONE;
    }
    if (port_nop) {
        *port_nop = port_no;
    }
    return error;
}

/* Attempts to remove 'dpif''s port number 'port_no'.  Returns 0 if successful,
 * otherwise a positive errno value. */
int
dpif_port_del(struct dpif *dpif, odp_port_t port_no)
{
    int error;

    COVERAGE_INC(dpif_port_del);

    error = dpif->dpif_class->port_del(dpif, port_no);
    if (!error) {
        VLOG_DBG_RL(&dpmsg_rl, "%s: port_del(%"PRIu32")",
                    dpif_name(dpif), port_no);
    } else {
        log_operation(dpif, "port_del", error);
    }
    return error;
}

/* Makes a deep copy of 'src' into 'dst'. */
void
dpif_port_clone(struct dpif_port *dst, const struct dpif_port *src)
{
    dst->name = xstrdup(src->name);
    dst->type = xstrdup(src->type);
    dst->port_no = src->port_no;
}

/* Frees memory allocated to members of 'dpif_port'.
 *
 * Do not call this function on a dpif_port obtained from
 * dpif_port_dump_next(): that function retains ownership of the data in the
 * dpif_port. */
void
dpif_port_destroy(struct dpif_port *dpif_port)
{
    free(dpif_port->name);
    free(dpif_port->type);
}

/* Checks if port named 'devname' exists in 'dpif'.  If so, returns
 * true; otherwise, returns false. */
bool
dpif_port_exists(const struct dpif *dpif, const char *devname)
{
    int error = dpif->dpif_class->port_query_by_name(dpif, devname, NULL);
    if (error != 0 && error != ENOENT && error != ENODEV) {
        VLOG_WARN_RL(&error_rl, "%s: failed to query port %s: %s",
                     dpif_name(dpif), devname, ovs_strerror(error));
    }

    return !error;
}

/* Looks up port number 'port_no' in 'dpif'.  On success, returns 0 and
 * initializes '*port' appropriately; on failure, returns a positive errno
 * value.
 *
 * The caller owns the data in 'port' and must free it with
 * dpif_port_destroy() when it is no longer needed. */
int
dpif_port_query_by_number(const struct dpif *dpif, odp_port_t port_no,
                          struct dpif_port *port)
{
    int error = dpif->dpif_class->port_query_by_number(dpif, port_no, port);
    if (!error) {
        VLOG_DBG_RL(&dpmsg_rl, "%s: port %"PRIu32" is device %s",
                    dpif_name(dpif), port_no, port->name);
    } else {
        memset(port, 0, sizeof *port);
        VLOG_WARN_RL(&error_rl, "%s: failed to query port %"PRIu32": %s",
                     dpif_name(dpif), port_no, ovs_strerror(error));
    }
    return error;
}

/* Looks up port named 'devname' in 'dpif'.  On success, returns 0 and
 * initializes '*port' appropriately; on failure, returns a positive errno
 * value.
 *
 * The caller owns the data in 'port' and must free it with
 * dpif_port_destroy() when it is no longer needed. */
int
dpif_port_query_by_name(const struct dpif *dpif, const char *devname,
                        struct dpif_port *port)
{
    int error = dpif->dpif_class->port_query_by_name(dpif, devname, port);
    if (!error) {
        VLOG_DBG_RL(&dpmsg_rl, "%s: device %s is on port %"PRIu32,
                    dpif_name(dpif), devname, port->port_no);
    } else {
        memset(port, 0, sizeof *port);

        /* For ENOENT or ENODEV we use DBG level because the caller is probably
         * interested in whether 'dpif' actually has a port 'devname', so that
         * it's not an issue worth logging if it doesn't.  Other errors are
         * uncommon and more likely to indicate a real problem. */
        VLOG_RL(&error_rl,
                error == ENOENT || error == ENODEV ? VLL_DBG : VLL_WARN,
                "%s: failed to query port %s: %s",
                dpif_name(dpif), devname, ovs_strerror(error));
    }
    return error;
}

/* Returns the Netlink PID value to supply in OVS_ACTION_ATTR_USERSPACE
 * actions as the OVS_USERSPACE_ATTR_PID attribute's value, for use in
 * flows whose packets arrived on port 'port_no'.  In the case where the
 * provider allocates multiple Netlink PIDs to a single port, it may use
 * 'hash' to spread load among them.  The caller need not use a particular
 * hash function; a 5-tuple hash is suitable.
 *
 * (The datapath implementation might use some different hash function for
 * distributing packets received via flow misses among PIDs.  This means
 * that packets received via flow misses might be reordered relative to
 * packets received via userspace actions.  This is not ordinarily a
 * problem.)
 *
 * A 'port_no' of ODPP_NONE is a special case: it returns a reserved PID, not
 * allocated to any port, that the client may use for special purposes.
 *
 * The return value is only meaningful when DPIF_UC_ACTION has been enabled in
 * the 'dpif''s listen mask.  It is allowed to change when DPIF_UC_ACTION is
 * disabled and then re-enabled, so a client that does that must be prepared to
 * update all of the flows that it installed that contain
 * OVS_ACTION_ATTR_USERSPACE actions. */
uint32_t
dpif_port_get_pid(const struct dpif *dpif, odp_port_t port_no, uint32_t hash)
{
    return (dpif->dpif_class->port_get_pid
            ? (dpif->dpif_class->port_get_pid)(dpif, port_no, hash)
            : 0);
}

/* Looks up port number 'port_no' in 'dpif'.  On success, returns 0 and copies
 * the port's name into the 'name_size' bytes in 'name', ensuring that the
 * result is null-terminated.  On failure, returns a positive errno value and
 * makes 'name' the empty string. */
int
dpif_port_get_name(struct dpif *dpif, odp_port_t port_no,
                   char *name, size_t name_size)
{
    struct dpif_port port;
    int error;

    ovs_assert(name_size > 0);

    error = dpif_port_query_by_number(dpif, port_no, &port);
    if (!error) {
        ovs_strlcpy(name, port.name, name_size);
        dpif_port_destroy(&port);
    } else {
        *name = '\0';
    }
    return error;
}

/* Initializes 'dump' to begin dumping the ports in a dpif.
 *
 * This function provides no status indication.  An error status for the entire
 * dump operation is provided when it is completed by calling
 * dpif_port_dump_done().
 */
void
dpif_port_dump_start(struct dpif_port_dump *dump, const struct dpif *dpif)
{
    dump->dpif = dpif;
    dump->error = dpif->dpif_class->port_dump_start(dpif, &dump->state);
    log_operation(dpif, "port_dump_start", dump->error);
}

/* Attempts to retrieve another port from 'dump', which must have been
 * initialized with dpif_port_dump_start().  On success, stores a new dpif_port
 * into 'port' and returns true.  On failure, returns false.
 *
 * Failure might indicate an actual error or merely that the last port has been
 * dumped.  An error status for the entire dump operation is provided when it
 * is completed by calling dpif_port_dump_done().
 *
 * The dpif owns the data stored in 'port'.  It will remain valid until at
 * least the next time 'dump' is passed to dpif_port_dump_next() or
 * dpif_port_dump_done(). */
bool
dpif_port_dump_next(struct dpif_port_dump *dump, struct dpif_port *port)
{
    const struct dpif *dpif = dump->dpif;

    if (dump->error) {
        return false;
    }

    dump->error = dpif->dpif_class->port_dump_next(dpif, dump->state, port);
    if (dump->error == EOF) {
        VLOG_DBG_RL(&dpmsg_rl, "%s: dumped all ports", dpif_name(dpif));
    } else {
        log_operation(dpif, "port_dump_next", dump->error);
    }

    if (dump->error) {
        dpif->dpif_class->port_dump_done(dpif, dump->state);
        return false;
    }
    return true;
}

/* Completes port table dump operation 'dump', which must have been initialized
 * with dpif_port_dump_start().  Returns 0 if the dump operation was
 * error-free, otherwise a positive errno value describing the problem. */
int
dpif_port_dump_done(struct dpif_port_dump *dump)
{
    const struct dpif *dpif = dump->dpif;
    if (!dump->error) {
        dump->error = dpif->dpif_class->port_dump_done(dpif, dump->state);
        log_operation(dpif, "port_dump_done", dump->error);
    }
    return dump->error == EOF ? 0 : dump->error;
}

/* Polls for changes in the set of ports in 'dpif'.  If the set of ports in
 * 'dpif' has changed, this function does one of the following:
 *
 * - Stores the name of the device that was added to or deleted from 'dpif' in
 *   '*devnamep' and returns 0.  The caller is responsible for freeing
 *   '*devnamep' (with free()) when it no longer needs it.
 *
 * - Returns ENOBUFS and sets '*devnamep' to NULL.
 *
 * This function may also return 'false positives', where it returns 0 and
 * '*devnamep' names a device that was not actually added or deleted or it
 * returns ENOBUFS without any change.
 *
 * Returns EAGAIN if the set of ports in 'dpif' has not changed.  May also
 * return other positive errno values to indicate that something has gone
 * wrong. */
int
dpif_port_poll(const struct dpif *dpif, char **devnamep)
{
    int error = dpif->dpif_class->port_poll(dpif, devnamep);
    if (error) {
        *devnamep = NULL;
    }
    return error;
}

/* Arranges for the poll loop to wake up when port_poll(dpif) will return a
 * value other than EAGAIN. */
void
dpif_port_poll_wait(const struct dpif *dpif)
{
    dpif->dpif_class->port_poll_wait(dpif);
}

/* Extracts the flow stats for a packet.  The 'flow' and 'packet'
 * arguments must have been initialized through a call to flow_extract().
 * 'used' is stored into stats->used. */
void
dpif_flow_stats_extract(const struct flow *flow, const struct ofpbuf *packet,
                        long long int used, struct dpif_flow_stats *stats)
{
    stats->tcp_flags = ntohs(flow->tcp_flags);
    stats->n_bytes = ofpbuf_size(packet);
    stats->n_packets = 1;
    stats->used = used;
}

/* Appends a human-readable representation of 'stats' to 's'. */
void
dpif_flow_stats_format(const struct dpif_flow_stats *stats, struct ds *s)
{
    ds_put_format(s, "packets:%"PRIu64", bytes:%"PRIu64", used:",
                  stats->n_packets, stats->n_bytes);
    if (stats->used) {
        ds_put_format(s, "%.3fs", (time_msec() - stats->used) / 1000.0);
    } else {
        ds_put_format(s, "never");
    }
    if (stats->tcp_flags) {
        ds_put_cstr(s, ", flags:");
        packet_format_tcp_flags(s, stats->tcp_flags);
    }
}

/* Deletes all flows from 'dpif'.  Returns 0 if successful, otherwise a
 * positive errno value.  */
int
dpif_flow_flush(struct dpif *dpif)
{
    int error;

    COVERAGE_INC(dpif_flow_flush);

    error = dpif->dpif_class->flow_flush(dpif);
    log_operation(dpif, "flow_flush", error);
    return error;
}

/* A dpif_operate() wrapper for performing a single DPIF_OP_FLOW_GET. */
int
dpif_flow_get(struct dpif *dpif,
              const struct nlattr *key, size_t key_len,
              struct ofpbuf *buf, struct dpif_flow *flow)
{
    struct dpif_op *opp;
    struct dpif_op op;

    op.type = DPIF_OP_FLOW_GET;
    op.u.flow_get.key = key;
    op.u.flow_get.key_len = key_len;
    op.u.flow_get.buffer = buf;
    op.u.flow_get.flow = flow;
    op.u.flow_get.flow->key = key;
    op.u.flow_get.flow->key_len = key_len;

    opp = &op;
    dpif_operate(dpif, &opp, 1);

    return op.error;
}

/* A dpif_operate() wrapper for performing a single DPIF_OP_FLOW_PUT. */
int
dpif_flow_put(struct dpif *dpif, enum dpif_flow_put_flags flags,
              const struct nlattr *key, size_t key_len,
              const struct nlattr *mask, size_t mask_len,
              const struct nlattr *actions, size_t actions_len,
              struct dpif_flow_stats *stats)
{
    struct dpif_op *opp;
    struct dpif_op op;

    op.type = DPIF_OP_FLOW_PUT;
    op.u.flow_put.flags = flags;
    op.u.flow_put.key = key;
    op.u.flow_put.key_len = key_len;
    op.u.flow_put.mask = mask;
    op.u.flow_put.mask_len = mask_len;
    op.u.flow_put.actions = actions;
    op.u.flow_put.actions_len = actions_len;
    op.u.flow_put.stats = stats;

    opp = &op;
    dpif_operate(dpif, &opp, 1);

    return op.error;
}

/* A dpif_operate() wrapper for performing a single DPIF_OP_FLOW_DEL. */
int
dpif_flow_del(struct dpif *dpif,
              const struct nlattr *key, size_t key_len,
              struct dpif_flow_stats *stats)
{
    struct dpif_op *opp;
    struct dpif_op op;

    op.type = DPIF_OP_FLOW_DEL;
    op.u.flow_del.key = key;
    op.u.flow_del.key_len = key_len;
    op.u.flow_del.stats = stats;

    opp = &op;
    dpif_operate(dpif, &opp, 1);

    return op.error;
}

/* Creates and returns a new 'struct dpif_flow_dump' for iterating through the
 * flows in 'dpif'.
 *
 * This function always successfully returns a dpif_flow_dump.  Error
 * reporting is deferred to dpif_flow_dump_destroy(). */
struct dpif_flow_dump *
dpif_flow_dump_create(const struct dpif *dpif)
{
    return dpif->dpif_class->flow_dump_create(dpif);
}

/* Destroys 'dump', which must have been created with dpif_flow_dump_create().
 * All dpif_flow_dump_thread structures previously created for 'dump' must
 * previously have been destroyed.
 *
 * Returns 0 if the dump operation was error-free, otherwise a positive errno
 * value describing the problem. */
int
dpif_flow_dump_destroy(struct dpif_flow_dump *dump)
{
    const struct dpif *dpif = dump->dpif;
    int error = dpif->dpif_class->flow_dump_destroy(dump);
    log_operation(dpif, "flow_dump_destroy", error);
    return error == EOF ? 0 : error;
}

/* Returns new thread-local state for use with dpif_flow_dump_next(). */
struct dpif_flow_dump_thread *
dpif_flow_dump_thread_create(struct dpif_flow_dump *dump)
{
    return dump->dpif->dpif_class->flow_dump_thread_create(dump);
}

/* Releases 'thread'. */
void
dpif_flow_dump_thread_destroy(struct dpif_flow_dump_thread *thread)
{
    thread->dpif->dpif_class->flow_dump_thread_destroy(thread);
}

/* Attempts to retrieve up to 'max_flows' more flows from 'thread'.  Returns 0
 * if and only if no flows remained to be retrieved, otherwise a positive
 * number reflecting the number of elements in 'flows[]' that were updated.
 * The number of flows returned might be less than 'max_flows' because
 * fewer than 'max_flows' remained, because this particular datapath does not
 * benefit from batching, or because an error occurred partway through
 * retrieval.  Thus, the caller should continue calling until a 0 return value,
 * even if intermediate return values are less than 'max_flows'.
 *
 * No error status is immediately provided.  An error status for the entire
 * dump operation is provided when it is completed by calling
 * dpif_flow_dump_destroy().
 *
 * All of the data stored into 'flows' is owned by the datapath, not by the
 * caller, and the caller must not modify or free it.  The datapath guarantees
 * that it remains accessible and unchanged until the first of:
 *  - The next call to dpif_flow_dump_next() for 'thread', or
 *  - The next rcu quiescent period. */
int
dpif_flow_dump_next(struct dpif_flow_dump_thread *thread,
                    struct dpif_flow *flows, int max_flows)
{
    struct dpif *dpif = thread->dpif;
    int n;

    ovs_assert(max_flows > 0);
    n = dpif->dpif_class->flow_dump_next(thread, flows, max_flows);
    if (n > 0) {
        struct dpif_flow *f;

        for (f = flows; f < &flows[n] && should_log_flow_message(0); f++) {
            log_flow_message(dpif, 0, "flow_dump",
                             f->key, f->key_len, f->mask, f->mask_len,
                             &f->stats, f->actions, f->actions_len);
        }
    } else {
        VLOG_DBG_RL(&dpmsg_rl, "%s: dumped all flows", dpif_name(dpif));
    }
    return n;
}

struct dpif_execute_helper_aux {
    struct dpif *dpif;
    int error;
};

/* This is called for actions that need the context of the datapath to be
 * meaningful. */
static void
dpif_execute_helper_cb(void *aux_, struct dpif_packet **packets, int cnt,
                       const struct nlattr *action, bool may_steal OVS_UNUSED)
{
    struct dpif_execute_helper_aux *aux = aux_;
    int type = nl_attr_type(action);
    struct ofpbuf *packet = &packets[0]->ofpbuf;
    struct pkt_metadata *md = &packets[0]->md;

    ovs_assert(cnt == 1);

    switch ((enum ovs_action_attr)type) {
    case OVS_ACTION_ATTR_OUTPUT:
    case OVS_ACTION_ATTR_USERSPACE:
    case OVS_ACTION_ATTR_RECIRC: {
        struct dpif_execute execute;
        struct ofpbuf execute_actions;
        uint64_t stub[256 / 8];

        if (md->tunnel.ip_dst) {
            /* The Linux kernel datapath throws away the tunnel information
             * that we supply as metadata.  We have to use a "set" action to
             * supply it. */
            ofpbuf_use_stub(&execute_actions, stub, sizeof stub);
            odp_put_tunnel_action(&md->tunnel, &execute_actions);
            ofpbuf_put(&execute_actions, action, NLA_ALIGN(action->nla_len));

            execute.actions = ofpbuf_data(&execute_actions);
            execute.actions_len = ofpbuf_size(&execute_actions);
        } else {
            execute.actions = action;
            execute.actions_len = NLA_ALIGN(action->nla_len);
        }

        execute.packet = packet;
        execute.md = *md;
        execute.needs_help = false;
        execute.probe = false;
        aux->error = dpif_execute(aux->dpif, &execute);
        log_execute_message(aux->dpif, &execute, true, aux->error);

        if (md->tunnel.ip_dst) {
            ofpbuf_uninit(&execute_actions);
        }
        break;
    }

    case OVS_ACTION_ATTR_HASH:
    case OVS_ACTION_ATTR_PUSH_VLAN:
    case OVS_ACTION_ATTR_POP_VLAN:
    case OVS_ACTION_ATTR_PUSH_MPLS:
    case OVS_ACTION_ATTR_POP_MPLS:
    case OVS_ACTION_ATTR_SET:
    case OVS_ACTION_ATTR_SET_MASKED:
    case OVS_ACTION_ATTR_SAMPLE:
    case OVS_ACTION_ATTR_UNSPEC:
    case __OVS_ACTION_ATTR_MAX:
        OVS_NOT_REACHED();
    }
}

/* Executes 'execute' by performing most of the actions in userspace and
 * passing the fully constructed packets to 'dpif' for output and userspace
 * actions.
 *
 * This helps with actions that a given 'dpif' doesn't implement directly. */
static int
dpif_execute_with_help(struct dpif *dpif, struct dpif_execute *execute)
{
    struct dpif_execute_helper_aux aux = {dpif, 0};
    struct dpif_packet packet, *pp;

    COVERAGE_INC(dpif_execute_with_help);

    packet.ofpbuf = *execute->packet;
    packet.md = execute->md;
    pp = &packet;

    odp_execute_actions(&aux, &pp, 1, false, execute->actions,
                        execute->actions_len, dpif_execute_helper_cb);

    /* Even though may_steal is set to false, some actions could modify or
     * reallocate the ofpbuf memory. We need to pass those changes to the
     * caller */
    *execute->packet = packet.ofpbuf;
    execute->md = packet.md;

    return aux.error;
}

/* Returns true if the datapath needs help executing 'execute'. */
static bool
dpif_execute_needs_help(const struct dpif_execute *execute)
{
    return execute->needs_help || nl_attr_oversized(execute->actions_len);
}

/* A dpif_operate() wrapper for performing a single DPIF_OP_EXECUTE. */
int
dpif_execute(struct dpif *dpif, struct dpif_execute *execute)
{
    if (execute->actions_len) {
        struct dpif_op *opp;
        struct dpif_op op;

        op.type = DPIF_OP_EXECUTE;
        op.u.execute = *execute;

        opp = &op;
        dpif_operate(dpif, &opp, 1);

        return op.error;
    } else {
        return 0;
    }
}

/* Executes each of the 'n_ops' operations in 'ops' on 'dpif', in the order in
 * which they are specified.  Places each operation's results in the "output"
 * members documented in comments, and 0 in the 'error' member on success or a
 * positive errno on failure. */
void
dpif_operate(struct dpif *dpif, struct dpif_op **ops, size_t n_ops)
{
    while (n_ops > 0) {
        size_t chunk;

        /* Count 'chunk', the number of ops that can be executed without
         * needing any help.  Ops that need help should be rare, so we
         * expect this to ordinarily be 'n_ops', that is, all the ops. */
        for (chunk = 0; chunk < n_ops; chunk++) {
            struct dpif_op *op = ops[chunk];

            if (op->type == DPIF_OP_EXECUTE
                && dpif_execute_needs_help(&op->u.execute)) {
                break;
            }
        }

        if (chunk) {
            /* Execute a chunk full of ops that the dpif provider can
             * handle itself, without help. */
            size_t i;

            dpif->dpif_class->operate(dpif, ops, chunk);

            for (i = 0; i < chunk; i++) {
                struct dpif_op *op = ops[i];
                int error = op->error;

                switch (op->type) {
                case DPIF_OP_FLOW_PUT: {
                    struct dpif_flow_put *put = &op->u.flow_put;

                    COVERAGE_INC(dpif_flow_put);
                    log_flow_put_message(dpif, put, error);
                    if (error && put->stats) {
                        memset(put->stats, 0, sizeof *put->stats);
                    }
                    break;
                }

                case DPIF_OP_FLOW_GET: {
                    struct dpif_flow_get *get = &op->u.flow_get;

                    COVERAGE_INC(dpif_flow_get);
                    if (error) {
                        memset(get->flow, 0, sizeof *get->flow);
                    }
                    log_flow_get_message(dpif, get, error);

                    break;
                }

                case DPIF_OP_FLOW_DEL: {
                    struct dpif_flow_del *del = &op->u.flow_del;

                    COVERAGE_INC(dpif_flow_del);
                    log_flow_del_message(dpif, del, error);
                    if (error && del->stats) {
                        memset(del->stats, 0, sizeof *del->stats);
                    }
                    break;
                }

                case DPIF_OP_EXECUTE:
                    COVERAGE_INC(dpif_execute);
                    log_execute_message(dpif, &op->u.execute, false, error);
                    break;
                }
            }

            ops += chunk;
            n_ops -= chunk;
        } else {
            /* Help the dpif provider to execute one op. */
            struct dpif_op *op = ops[0];

            COVERAGE_INC(dpif_execute);
            op->error = dpif_execute_with_help(dpif, &op->u.execute);
            ops++;
            n_ops--;
        }
    }
}

/* Returns a string that represents 'type', for use in log messages. */
const char *
dpif_upcall_type_to_string(enum dpif_upcall_type type)
{
    switch (type) {
    case DPIF_UC_MISS: return "miss";
    case DPIF_UC_ACTION: return "action";
    case DPIF_N_UC_TYPES: default: return "<unknown>";
    }
}

/* Enables or disables receiving packets with dpif_recv() on 'dpif'.  Returns 0
 * if successful, otherwise a positive errno value.
 *
 * Turning packet receive off and then back on may change the Netlink PID
 * assignments returned by dpif_port_get_pid().  If the client does this, it
 * must update all of the flows that have OVS_ACTION_ATTR_USERSPACE actions
 * using the new PID assignment. */
int
dpif_recv_set(struct dpif *dpif, bool enable)
{
    int error = 0;

    if (dpif->dpif_class->recv_set) {
        error = dpif->dpif_class->recv_set(dpif, enable);
        log_operation(dpif, "recv_set", error);
    }
    return error;
}

/* Refreshes the poll loops and Netlink sockets associated to each port,
 * when the number of upcall handlers (upcall receiving thread) is changed
 * to 'n_handlers' and receiving packets for 'dpif' is enabled by
 * recv_set().
 *
 * Since multiple upcall handlers can read upcalls simultaneously from
 * 'dpif', each port can have multiple Netlink sockets, one per upcall
 * handler.  So, handlers_set() is responsible for the following tasks:
 *
 *    When receiving upcall is enabled, extends or creates the
 *    configuration to support:
 *
 *        - 'n_handlers' Netlink sockets for each port.
 *
 *        - 'n_handlers' poll loops, one for each upcall handler.
 *
 *        - registering the Netlink sockets for the same upcall handler to
 *          the corresponding poll loop.
 *
 * Returns 0 if successful, otherwise a positive errno value. */
int
dpif_handlers_set(struct dpif *dpif, uint32_t n_handlers)
{
    int error = 0;

    if (dpif->dpif_class->handlers_set) {
        error = dpif->dpif_class->handlers_set(dpif, n_handlers);
        log_operation(dpif, "handlers_set", error);
    }
    return error;
}

void
dpif_register_upcall_cb(struct dpif *dpif, upcall_callback *cb, void *aux)
{
    if (dpif->dpif_class->register_upcall_cb) {
        dpif->dpif_class->register_upcall_cb(dpif, cb, aux);
    }
}

void
dpif_enable_upcall(struct dpif *dpif)
{
    if (dpif->dpif_class->enable_upcall) {
        dpif->dpif_class->enable_upcall(dpif);
    }
}

void
dpif_disable_upcall(struct dpif *dpif)
{
    if (dpif->dpif_class->disable_upcall) {
        dpif->dpif_class->disable_upcall(dpif);
    }
}

void
dpif_print_packet(struct dpif *dpif, struct dpif_upcall *upcall)
{
    if (!VLOG_DROP_DBG(&dpmsg_rl)) {
        struct ds flow;
        char *packet;

        packet = ofp_packet_to_string(ofpbuf_data(&upcall->packet),
                                      ofpbuf_size(&upcall->packet));

        ds_init(&flow);
        odp_flow_key_format(upcall->key, upcall->key_len, &flow);

        VLOG_DBG("%s: %s upcall:\n%s\n%s",
                 dpif_name(dpif), dpif_upcall_type_to_string(upcall->type),
                 ds_cstr(&flow), packet);

        ds_destroy(&flow);
        free(packet);
    }
}

/* If 'dpif' creates its own I/O polling threads, refreshes poll threads
 * configuration. */
int
dpif_poll_threads_set(struct dpif *dpif, unsigned int n_rxqs,
                      const char *cmask)
{
    int error = 0;

    if (dpif->dpif_class->poll_threads_set) {
        error = dpif->dpif_class->poll_threads_set(dpif, n_rxqs, cmask);
        if (error) {
            log_operation(dpif, "poll_threads_set", error);
        }
    }

    return error;
}

/* Polls for an upcall from 'dpif' for an upcall handler.  Since there
 * there can be multiple poll loops, 'handler_id' is needed as index to
 * identify the corresponding poll loop.  If successful, stores the upcall
 * into '*upcall', using 'buf' for storage.  Should only be called if
 * 'recv_set' has been used to enable receiving packets from 'dpif'.
 *
 * 'upcall->key' and 'upcall->userdata' point into data in the caller-provided
 * 'buf', so their memory cannot be freed separately from 'buf'.
 *
 * The caller owns the data of 'upcall->packet' and may modify it.  If
 * packet's headroom is exhausted as it is manipulated, 'upcall->packet'
 * will be reallocated.  This requires the data of 'upcall->packet' to be
 * released with ofpbuf_uninit() before 'upcall' is destroyed.  However,
 * when an error is returned, the 'upcall->packet' may be uninitialized
 * and should not be released.
 *
 * Returns 0 if successful, otherwise a positive errno value.  Returns EAGAIN
 * if no upcall is immediately available. */
int
dpif_recv(struct dpif *dpif, uint32_t handler_id, struct dpif_upcall *upcall,
          struct ofpbuf *buf)
{
    int error = EAGAIN;

    if (dpif->dpif_class->recv) {
        error = dpif->dpif_class->recv(dpif, handler_id, upcall, buf);
        if (!error) {
            dpif_print_packet(dpif, upcall);
        } else if (error != EAGAIN) {
            log_operation(dpif, "recv", error);
        }
    }
    return error;
}

/* Discards all messages that would otherwise be received by dpif_recv() on
 * 'dpif'. */
void
dpif_recv_purge(struct dpif *dpif)
{
    COVERAGE_INC(dpif_purge);
    if (dpif->dpif_class->recv_purge) {
        dpif->dpif_class->recv_purge(dpif);
    }
}

/* Arranges for the poll loop for an upcall handler to wake up when 'dpif'
 * 'dpif' has a message queued to be received with the recv member
 * function.  Since there can be multiple poll loops, 'handler_id' is
 * needed as index to identify the corresponding poll loop. */
void
dpif_recv_wait(struct dpif *dpif, uint32_t handler_id)
{
    if (dpif->dpif_class->recv_wait) {
        dpif->dpif_class->recv_wait(dpif, handler_id);
    }
}

/* Obtains the NetFlow engine type and engine ID for 'dpif' into '*engine_type'
 * and '*engine_id', respectively. */
void
dpif_get_netflow_ids(const struct dpif *dpif,
                     uint8_t *engine_type, uint8_t *engine_id)
{
    *engine_type = dpif->netflow_engine_type;
    *engine_id = dpif->netflow_engine_id;
}

/* Translates OpenFlow queue ID 'queue_id' (in host byte order) into a priority
 * value used for setting packet priority.
 * On success, returns 0 and stores the priority into '*priority'.
 * On failure, returns a positive errno value and stores 0 into '*priority'. */
int
dpif_queue_to_priority(const struct dpif *dpif, uint32_t queue_id,
                       uint32_t *priority)
{
    int error = (dpif->dpif_class->queue_to_priority
                 ? dpif->dpif_class->queue_to_priority(dpif, queue_id,
                                                       priority)
                 : EOPNOTSUPP);
    if (error) {
        *priority = 0;
    }
    log_operation(dpif, "queue_to_priority", error);
    return error;
}

void
dpif_init(struct dpif *dpif, const struct dpif_class *dpif_class,
          const char *name,
          uint8_t netflow_engine_type, uint8_t netflow_engine_id)
{
    dpif->dpif_class = dpif_class;
    dpif->base_name = xstrdup(name);
    dpif->full_name = xasprintf("%s@%s", dpif_class->type, name);
    dpif->netflow_engine_type = netflow_engine_type;
    dpif->netflow_engine_id = netflow_engine_id;
}

/* Undoes the results of initialization.
 *
 * Normally this function only needs to be called from dpif_close().
 * However, it may be called by providers due to an error on opening
 * that occurs after initialization.  It this case dpif_close() would
 * never be called. */
void
dpif_uninit(struct dpif *dpif, bool close)
{
    char *base_name = dpif->base_name;
    char *full_name = dpif->full_name;

    if (close) {
        dpif->dpif_class->close(dpif);
    }

    free(base_name);
    free(full_name);
}

static void
log_operation(const struct dpif *dpif, const char *operation, int error)
{
    if (!error) {
        VLOG_DBG_RL(&dpmsg_rl, "%s: %s success", dpif_name(dpif), operation);
    } else if (ofperr_is_valid(error)) {
        VLOG_WARN_RL(&error_rl, "%s: %s failed (%s)",
                     dpif_name(dpif), operation, ofperr_get_name(error));
    } else {
        VLOG_WARN_RL(&error_rl, "%s: %s failed (%s)",
                     dpif_name(dpif), operation, ovs_strerror(error));
    }
}

static enum vlog_level
flow_message_log_level(int error)
{
    /* If flows arrive in a batch, userspace may push down multiple
     * unique flow definitions that overlap when wildcards are applied.
     * Kernels that support flow wildcarding will reject these flows as
     * duplicates (EEXIST), so lower the log level to debug for these
     * types of messages. */
    return (error && error != EEXIST) ? VLL_WARN : VLL_DBG;
}

static bool
should_log_flow_message(int error)
{
    return !vlog_should_drop(THIS_MODULE, flow_message_log_level(error),
                             error ? &error_rl : &dpmsg_rl);
}

static void
log_flow_message(const struct dpif *dpif, int error, const char *operation,
                 const struct nlattr *key, size_t key_len,
                 const struct nlattr *mask, size_t mask_len,
                 const struct dpif_flow_stats *stats,
                 const struct nlattr *actions, size_t actions_len)
{
    struct ds ds = DS_EMPTY_INITIALIZER;
    ds_put_format(&ds, "%s: ", dpif_name(dpif));
    if (error) {
        ds_put_cstr(&ds, "failed to ");
    }
    ds_put_format(&ds, "%s ", operation);
    if (error) {
        ds_put_format(&ds, "(%s) ", ovs_strerror(error));
    }
    odp_flow_format(key, key_len, mask, mask_len, NULL, &ds, true);
    if (stats) {
        ds_put_cstr(&ds, ", ");
        dpif_flow_stats_format(stats, &ds);
    }
    if (actions || actions_len) {
        ds_put_cstr(&ds, ", actions:");
        format_odp_actions(&ds, actions, actions_len);
    }
    vlog(THIS_MODULE, flow_message_log_level(error), "%s", ds_cstr(&ds));
    ds_destroy(&ds);
}

static void
log_flow_put_message(struct dpif *dpif, const struct dpif_flow_put *put,
                     int error)
{
    if (should_log_flow_message(error) && !(put->flags & DPIF_FP_PROBE)) {
        struct ds s;

        ds_init(&s);
        ds_put_cstr(&s, "put");
        if (put->flags & DPIF_FP_CREATE) {
            ds_put_cstr(&s, "[create]");
        }
        if (put->flags & DPIF_FP_MODIFY) {
            ds_put_cstr(&s, "[modify]");
        }
        if (put->flags & DPIF_FP_ZERO_STATS) {
            ds_put_cstr(&s, "[zero]");
        }
        log_flow_message(dpif, error, ds_cstr(&s),
                         put->key, put->key_len, put->mask, put->mask_len,
                         put->stats, put->actions, put->actions_len);
        ds_destroy(&s);
    }
}

static void
log_flow_del_message(struct dpif *dpif, const struct dpif_flow_del *del,
                     int error)
{
    if (should_log_flow_message(error)) {
        log_flow_message(dpif, error, "flow_del", del->key, del->key_len,
                         NULL, 0, !error ? del->stats : NULL, NULL, 0);
    }
}

/* Logs that 'execute' was executed on 'dpif' and completed with errno 'error'
 * (0 for success).  'subexecute' should be true if the execution is a result
 * of breaking down a larger execution that needed help, false otherwise.
 *
 *
 * XXX In theory, the log message could be deceptive because this function is
 * called after the dpif_provider's '->execute' function, which is allowed to
 * modify execute->packet and execute->md.  In practice, though:
 *
 *     - dpif-netlink doesn't modify execute->packet or execute->md.
 *
 *     - dpif-netdev does modify them but it is less likely to have problems
 *       because it is built into ovs-vswitchd and cannot have version skew,
 *       etc.
 *
 * It would still be better to avoid the potential problem.  I don't know of a
 * good way to do that, though, that isn't expensive. */
static void
log_execute_message(struct dpif *dpif, const struct dpif_execute *execute,
                    bool subexecute, int error)
{
    if (!(error ? VLOG_DROP_WARN(&error_rl) : VLOG_DROP_DBG(&dpmsg_rl))
        && !execute->probe) {
        struct ds ds = DS_EMPTY_INITIALIZER;
        char *packet;

        packet = ofp_packet_to_string(ofpbuf_data(execute->packet),
                                      ofpbuf_size(execute->packet));
        ds_put_format(&ds, "%s: %sexecute ",
                      dpif_name(dpif),
                      (subexecute ? "sub-"
                       : dpif_execute_needs_help(execute) ? "super-"
                       : ""));
        format_odp_actions(&ds, execute->actions, execute->actions_len);
        if (error) {
            ds_put_format(&ds, " failed (%s)", ovs_strerror(error));
        }
        ds_put_format(&ds, " on packet %s", packet);
        vlog(THIS_MODULE, error ? VLL_WARN : VLL_DBG, "%s", ds_cstr(&ds));
        ds_destroy(&ds);
        free(packet);
    }
}

static void
log_flow_get_message(const struct dpif *dpif, const struct dpif_flow_get *get,
                     int error)
{
    if (should_log_flow_message(error)) {
        log_flow_message(dpif, error, "flow_get",
                         get->key, get->key_len,
                         get->flow->mask, get->flow->mask_len,
                         &get->flow->stats,
                         get->flow->actions, get->flow->actions_len);
    }
}