summaryrefslogtreecommitdiff
path: root/storage/tokudb/PerconaFT/src/ydb_db.cc
blob: 6b4452325f491bd525657c0383c4fffd8d7dfd61 (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
/* -*- mode: C++; c-basic-offset: 4; indent-tabs-mode: nil -*- */
// vim: ft=cpp:expandtab:ts=8:sw=4:softtabstop=4:
#ident "$Id$"
/*======
This file is part of PerconaFT.


Copyright (c) 2006, 2015, Percona and/or its affiliates. All rights reserved.

    PerconaFT is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License, version 2,
    as published by the Free Software Foundation.

    PerconaFT is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU General Public License for more details.

    You should have received a copy of the GNU General Public License
    along with PerconaFT.  If not, see <http://www.gnu.org/licenses/>.

----------------------------------------

    PerconaFT is free software: you can redistribute it and/or modify
    it under the terms of the GNU Affero General Public License, version 3,
    as published by the Free Software Foundation.

    PerconaFT is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU Affero General Public License for more details.

    You should have received a copy of the GNU Affero General Public License
    along with PerconaFT.  If not, see <http://www.gnu.org/licenses/>.
======= */

#ident "Copyright (c) 2006, 2015, Percona and/or its affiliates. All rights reserved."

#include <my_global.h>
#include <ctype.h>

#include <db.h>
#include <locktree/locktree.h>
#include <ft/ft.h>
#include <ft/ft-flusher.h>
#include <ft/cachetable/checkpoint.h>

#include "ydb_cursor.h"
#include "ydb_row_lock.h"
#include "ydb_db.h"
#include "ydb_write.h"
#include "ydb-internal.h"
#include "ydb_load.h"
#include "indexer.h"
#include <portability/toku_atomic.h>
#include <util/status.h>
#include <ft/le-cursor.h>

static YDB_DB_LAYER_STATUS_S ydb_db_layer_status;
#ifdef STATUS_VALUE
#undef STATUS_VALUE
#endif
#define STATUS_VALUE(x) ydb_db_layer_status.status[x].value.num

#define STATUS_INIT(k,c,t,l,inc) TOKUFT_STATUS_INIT(ydb_db_layer_status, k, c, t, l, inc)

static void
ydb_db_layer_status_init (void) {
    // Note, this function initializes the keyname, type, and legend fields.
    // Value fields are initialized to zero by compiler.

    STATUS_INIT(YDB_LAYER_DIRECTORY_WRITE_LOCKS,      nullptr, UINT64,   "directory write locks", TOKU_ENGINE_STATUS);
    STATUS_INIT(YDB_LAYER_DIRECTORY_WRITE_LOCKS_FAIL, nullptr, UINT64,   "directory write locks fail", TOKU_ENGINE_STATUS);
    STATUS_INIT(YDB_LAYER_LOGSUPPRESS,                nullptr, UINT64,   "log suppress", TOKU_ENGINE_STATUS);
    STATUS_INIT(YDB_LAYER_LOGSUPPRESS_FAIL,           nullptr, UINT64,   "log suppress fail", TOKU_ENGINE_STATUS);
    ydb_db_layer_status.initialized = true;
}
#undef STATUS_INIT

void
ydb_db_layer_get_status(YDB_DB_LAYER_STATUS statp) {
    if (!ydb_db_layer_status.initialized)
        ydb_db_layer_status_init();
    *statp = ydb_db_layer_status;
}

void create_iname_hint(DB_ENV *env, const char *dname, char *hint) {
    //Requires: size of hint array must be > strlen(dname)
    //Copy alphanumeric characters only.
    //Replace strings of non-alphanumeric characters with a single underscore.
    if (env->get_dir_per_db(env) && !toku_os_is_absolute_name(dname)) {
        assert(dname);
        if (*dname == '.')
            ++dname;
        if (*dname == '/')
            ++dname;
        bool underscored = false;
        bool dbdir_is_parsed = false;
        // Do not change the first '/' because this is
        // delimiter which splits name into database dir
        // and table dir.
        while (*dname) {
            if (isalnum(*dname) || (*dname == '/' && !dbdir_is_parsed)) {
                char c = *dname++;
                *hint++ = c;
                if (c == '/')
                    dbdir_is_parsed = true;
                underscored = false;
            } else if (!dbdir_is_parsed) {
                char c = *dname++;
                *hint++ = c;
            } else {
                if (!underscored)
                    *hint++ = '_';
                dname++;
                underscored = true;
            }
        }
        *hint = '\0';
    } else {
        bool underscored = false;
        while (*dname) {
            if (isalnum(*dname)) {
                char c = *dname++;
                *hint++ = c;
                underscored = false;
            }
            else {
                if (!underscored)
                    *hint++ = '_';
                dname++;
                underscored = true;
            }
        }
        *hint = '\0';
    }
}

// n < 0  means to ignore mark and ignore n
// n >= 0 means to include mark ("_B_" or "_P_") with hex value of n in iname
// (intended for use by loader, which will create many inames using one txnid).
char *create_iname(DB_ENV *env,
                   uint64_t id1,
                   uint64_t id2,
                   char *hint,
                   const char *mark,
                   int n) {
    int bytes;
    char inamebase[strlen(hint) +
                   8 +  // hex file format version
                   24 + // hex id (normally the txnid's parent and child)
                   8  + // hex value of n if non-neg
                   sizeof("_B___.") + // extra pieces
                   strlen(toku_product_name)];
    if (n < 0)
        bytes = snprintf(inamebase, sizeof(inamebase),
                         "%s_%" PRIx64 "_%" PRIx64 "_%" PRIx32            ".%s",
                         hint, id1, id2, FT_LAYOUT_VERSION, toku_product_name);
    else {
        invariant(strlen(mark) == 1);
        bytes = snprintf(inamebase, sizeof(inamebase),
                         "%s_%" PRIx64 "_%" PRIx64 "_%" PRIx32 "_%s_%" PRIx32 ".%s",
                         hint, id1, id2, FT_LAYOUT_VERSION, mark, n, toku_product_name);
    }
    assert(bytes>0);
    assert(bytes<=(int)sizeof(inamebase)-1);
    char *rval;
    if (env->i->data_dir)
        rval = toku_construct_full_name(2, env->i->data_dir, inamebase);
    else
        rval = toku_construct_full_name(1, inamebase);
    assert(rval);
    return rval;
}

static uint64_t nontransactional_open_id = 0;

std::unique_ptr<char[], decltype(&toku_free)> generate_iname_for_rename_or_open(
    DB_ENV *env,
    DB_TXN *txn,
    const char *dname,
    bool is_open) {
    std::unique_ptr<char[], decltype(&toku_free)> result(nullptr, &toku_free);
    char hint[strlen(dname) + 1];
    uint64_t id1 = 0;
    uint64_t id2 = 0;

    if (txn) {
        id1 = toku_txn_get_txnid(db_txn_struct_i(txn)->tokutxn).parent_id64;
        id2 = toku_txn_get_txnid(db_txn_struct_i(txn)->tokutxn).child_id64;
    } else if (is_open)
        id1 = toku_sync_fetch_and_add(&nontransactional_open_id, 1);

    create_iname_hint(env, dname, hint);

    result.reset(create_iname(env, id1, id2, hint, NULL, -1));

    return result;
}

static int toku_db_open(DB * db, DB_TXN * txn, const char *fname, const char *dbname, DBTYPE dbtype, uint32_t flags, int mode);

// Effect: Do the work required of DB->close().
// requires: the multi_operation client lock is held.
int 
toku_db_close(DB * db) {
    int r = 0;
    if (db_opened(db) && db->i->dname) {
        // internal (non-user) dictionary has no dname
        env_note_db_closed(db->dbenv, db);  // tell env that this db is no longer in use by the user of this api (user-closed, may still be in use by fractal tree internals)
    }
    // close the ft handle, and possibly close the locktree
    toku_ft_handle_close(db->i->ft_handle);
    if (db->i->lt) {
        db->dbenv->i->ltm.release_lt(db->i->lt);
    }
    toku_sdbt_cleanup(&db->i->skey);
    toku_sdbt_cleanup(&db->i->sval);
    if (db->i->dname) {
        toku_free(db->i->dname);
    }
    toku_free(db->i);
    toku_free(db);
    return r;
}

///////////
//db_getf_XXX is equivalent to c_getf_XXX, without a persistent cursor

int
db_getf_set(DB *db, DB_TXN *txn, uint32_t flags, DBT *key, YDB_CALLBACK_FUNCTION f, void *extra) {
    HANDLE_PANICKED_DB(db);
    HANDLE_DB_ILLEGAL_WORKING_PARENT_TXN(db, txn);
    DBC c;
    uint32_t create_flags = flags & (DB_ISOLATION_FLAGS | DB_RMW);
    flags &= ~DB_ISOLATION_FLAGS;
    int r = toku_db_cursor_internal(db, txn, &c, create_flags | DBC_DISABLE_PREFETCHING, 1);
    if (r==0) {
        r = toku_c_getf_set(&c, flags, key, f, extra);
        int r2 = toku_c_close_internal(&c);
        if (r==0) r = r2;
    }
    return r;
}

static inline int 
db_thread_need_flags(DBT *dbt) {
    return (dbt->flags & (DB_DBT_MALLOC+DB_DBT_REALLOC+DB_DBT_USERMEM)) == 0;
}

int 
toku_db_get (DB * db, DB_TXN * txn, DBT * key, DBT * data, uint32_t flags) {
    HANDLE_PANICKED_DB(db);
    HANDLE_DB_ILLEGAL_WORKING_PARENT_TXN(db, txn);
    int r;
    uint32_t iso_flags = flags & DB_ISOLATION_FLAGS;

    if ((db->i->open_flags & DB_THREAD) && db_thread_need_flags(data))
        return EINVAL;

    uint32_t lock_flags = flags & (DB_PRELOCKED | DB_PRELOCKED_WRITE);
    flags &= ~lock_flags;
    flags &= ~DB_ISOLATION_FLAGS;
    // And DB_GET_BOTH is no longer supported. #2862.
    if (flags != 0) return EINVAL;

    DBC dbc;
    r = toku_db_cursor_internal(db, txn, &dbc, iso_flags | DBC_DISABLE_PREFETCHING, 1);
    if (r!=0) return r;
    uint32_t c_get_flags = DB_SET;
    r = toku_c_get(&dbc, key, data, c_get_flags | lock_flags);
    int r2 = toku_c_close_internal(&dbc);
    return r ? r : r2;
}

static int
db_open_subdb(DB * db, DB_TXN * txn, const char *fname, const char *dbname, DBTYPE dbtype, uint32_t flags, int mode) {
    int r;
    if (!fname || !dbname) r = EINVAL;
    else {
        char subdb_full_name[strlen(fname) + sizeof("/") + strlen(dbname)];
        int bytes = snprintf(subdb_full_name, sizeof(subdb_full_name), "%s/%s", fname, dbname);
        assert(bytes==(int)sizeof(subdb_full_name)-1);
        const char *null_subdbname = NULL;
        r = toku_db_open(db, txn, subdb_full_name, null_subdbname, dbtype, flags, mode);
    }
    return r;
}

// inames are created here.
// algorithm:
//  begin txn
//  convert dname to iname (possibly creating new iname)
//  open file (toku_ft_handle_open() will handle logging)
//  close txn
//  if created a new iname, take full range lock
// Requires: no checkpoint may take place during this function, which is enforced by holding the multi_operation_client_lock.
static int 
toku_db_open(DB * db, DB_TXN * txn, const char *fname, const char *dbname, DBTYPE dbtype, uint32_t flags, int mode) {
    HANDLE_PANICKED_DB(db);
    HANDLE_READ_ONLY_TXN(txn);
    if (dbname != NULL) {
        return db_open_subdb(db, txn, fname, dbname, dbtype, flags, mode);
    }

    // at this point fname is the dname
    //This code ONLY supports single-db files.
    assert(dbname == NULL);
    const char * dname = fname;  // db_open_subdb() converts (fname, dbname) to dname

    ////////////////////////////// do some level of parameter checking.
    uint32_t unused_flags = flags;
    int r;
    if (dbtype!=DB_BTREE && dbtype!=DB_UNKNOWN) return EINVAL;
    int is_db_excl    = flags & DB_EXCL;    unused_flags&=~DB_EXCL;
    int is_db_create  = flags & DB_CREATE;  unused_flags&=~DB_CREATE;
    int is_db_hot_index  = flags & DB_IS_HOT_INDEX;  unused_flags&=~DB_IS_HOT_INDEX;

    //We support READ_UNCOMMITTED and READ_COMMITTED whether or not the flag is provided.
    unused_flags&=~DB_READ_UNCOMMITTED;
    unused_flags&=~DB_READ_COMMITTED;
    unused_flags&=~DB_SERIALIZABLE;

    // DB_THREAD is implicitly supported and DB_BLACKHOLE is supported at the ft-layer
    unused_flags &= ~DB_THREAD;
    unused_flags &= ~DB_BLACKHOLE;

    // check for unknown or conflicting flags
    if (unused_flags) return EINVAL; // unknown flags
    if (is_db_excl && !is_db_create) return EINVAL;
    if (dbtype==DB_UNKNOWN && is_db_excl) return EINVAL;

    if (db_opened(db)) {
        // it was already open
        return EINVAL;
    }
    //////////////////////////////

    // convert dname to iname
    //  - look up dname, get iname
    //  - if dname does not exist, create iname and make entry in directory
    DBT dname_dbt;  // holds dname
    DBT iname_dbt;  // holds iname_in_env
    toku_fill_dbt(&dname_dbt, dname, strlen(dname)+1);
    toku_init_dbt_flags(&iname_dbt, DB_DBT_REALLOC);
    r = toku_db_get(db->dbenv->i->directory, txn, &dname_dbt, &iname_dbt, DB_SERIALIZABLE);  // allocates memory for iname
    std::unique_ptr<char[], decltype(&toku_free)> iname(
        static_cast<char *>(iname_dbt.data), &toku_free);
    if (r == DB_NOTFOUND && !is_db_create) {
        r = ENOENT;
    } else if (r==0 && is_db_excl) {
        r = EEXIST;
    } else if (r == DB_NOTFOUND) {
        iname = generate_iname_for_rename_or_open(db->dbenv, txn, dname, true);
        toku_fill_dbt(&iname_dbt, iname.get(), strlen(iname.get()) + 1);
        //
        // put_flags will be 0 for performance only, avoid unnecessary query
        // if we are creating a hot index, per #3166, we do not want the write lock  in directory grabbed.
        // directory read lock is grabbed in toku_db_get above
        //
        uint32_t put_flags = 0 | ((is_db_hot_index) ? DB_PRELOCKED_WRITE : 0); 
        r = toku_db_put(db->dbenv->i->directory, txn, &dname_dbt, &iname_dbt, put_flags, true);  
    }

    // we now have an iname
    if (r == 0) {
        r = toku_db_open_iname(db, txn, iname.get(), flags, mode);
        if (r == 0) {
            db->i->dname = toku_xstrdup(dname);
            env_note_db_opened(db->dbenv, db);  // tell env that a new db handle is open (using dname)
        }
    }

    return r;
}

// set the descriptor and cmp_descriptor to the
// descriptors from the given ft, updating the
// locktree's descriptor pointer if necessary
static void
db_set_descriptors(DB *db, FT_HANDLE ft_handle) {
    const toku::comparator &cmp = toku_ft_get_comparator(ft_handle);
    db->descriptor = toku_ft_get_descriptor(ft_handle);
    db->cmp_descriptor = toku_ft_get_cmp_descriptor(ft_handle);
    invariant(db->cmp_descriptor == cmp.get_descriptor());
    if (db->i->lt) {
        db->i->lt->set_comparator(cmp);
    }
}

// callback that sets the descriptors when 
// a dictionary is redirected at the ft layer
static void
db_on_redirect_callback(FT_HANDLE ft_handle, void* extra) {
    DB *db = (DB *) extra;
    db_set_descriptors(db, ft_handle);
}

// when a locktree is created, clone a ft handle and store it
// as userdata so we can close it later.
int toku_db_lt_on_create_callback(toku::locktree *lt, void *extra) {
    int r;
    struct lt_on_create_callback_extra *info = (struct lt_on_create_callback_extra *) extra;
    TOKUTXN ttxn = info->txn ? db_txn_struct_i(info->txn)->tokutxn : NULL;
    FT_HANDLE ft_handle = info->ft_handle;

    FT_HANDLE cloned_ft_handle;
    r = toku_ft_handle_clone(&cloned_ft_handle, ft_handle, ttxn);
    if (r == 0) {
        assert(lt->get_userdata() == NULL);
        lt->set_userdata(cloned_ft_handle);
    }
    return r;
}

// when a locktree is about to be destroyed, 
// close the ft handle stored as userdata.
void toku_db_lt_on_destroy_callback(toku::locktree *lt) {
    FT_HANDLE ft_handle = (FT_HANDLE) lt->get_userdata();
    assert(ft_handle);
    toku_ft_handle_close(ft_handle);
}

// Instruct db to use the default (built-in) key comparison function
// by setting the flag bits in the db and ft structs
int toku_db_use_builtin_key_cmp(DB *db) {
    HANDLE_PANICKED_DB(db);
    int r = 0;
    if (db_opened(db)) {
        r = toku_ydb_do_error(db->dbenv, EINVAL, "Comparison functions cannot be set after DB open.\n");
    } else if (db->i->key_compare_was_set) {
        r = toku_ydb_do_error(db->dbenv, EINVAL, "Key comparison function already set.\n");
    } else {
        uint32_t tflags;
        toku_ft_get_flags(db->i->ft_handle, &tflags);

        tflags |= TOKU_DB_KEYCMP_BUILTIN;
        toku_ft_set_flags(db->i->ft_handle, tflags);
        db->i->key_compare_was_set = true;
    }
    return r;
}

int toku_db_open_iname(DB * db, DB_TXN * txn, const char *iname_in_env, uint32_t flags, int mode) {
    //Set comparison functions if not yet set.
    HANDLE_READ_ONLY_TXN(txn);
    if (!db->i->key_compare_was_set && db->dbenv->i->bt_compare) {
        toku_ft_set_bt_compare(db->i->ft_handle, db->dbenv->i->bt_compare);
        db->i->key_compare_was_set = true;
    }
    if (db->dbenv->i->update_function) {
        toku_ft_set_update(db->i->ft_handle,db->dbenv->i->update_function);
    }
    toku_ft_set_redirect_callback(
        db->i->ft_handle,
        db_on_redirect_callback,
        db
        );
    bool need_locktree = (bool)((db->dbenv->i->open_flags & DB_INIT_LOCK) &&
                                (db->dbenv->i->open_flags & DB_INIT_TXN));

    int is_db_excl    = flags & DB_EXCL;    flags&=~DB_EXCL;
    int is_db_create  = flags & DB_CREATE;  flags&=~DB_CREATE;
    //We support READ_UNCOMMITTED and READ_COMMITTED whether or not the flag is provided.
                                            flags&=~DB_READ_UNCOMMITTED;
                                            flags&=~DB_READ_COMMITTED;
                                            flags&=~DB_SERIALIZABLE;
                                            flags&=~DB_IS_HOT_INDEX;
    // unknown or conflicting flags are bad
    int unknown_flags = flags & ~DB_THREAD;
    unknown_flags &= ~DB_BLACKHOLE;
    if (unknown_flags || (is_db_excl && !is_db_create)) {
        return EINVAL;
    }

    if (db_opened(db)) {
        return EINVAL;              /* It was already open. */
    }
    
    db->i->open_flags = flags;
    db->i->open_mode = mode;

    FT_HANDLE ft_handle = db->i->ft_handle;
    int r = toku_ft_handle_open(ft_handle, iname_in_env,
                      is_db_create, is_db_excl,
                      db->dbenv->i->cachetable,
                      txn ? db_txn_struct_i(txn)->tokutxn : nullptr);
    if (r != 0) {
        goto out;
    }

    // if the dictionary was opened as a blackhole, mark the
    // fractal tree as blackhole too.
    if (flags & DB_BLACKHOLE) {
        toku_ft_set_blackhole(ft_handle);
    }

    db->i->opened = 1;

    // now that the handle has successfully opened, a valid descriptor
    // is in the ft. we need to set the db's descriptor pointers
    db_set_descriptors(db, ft_handle);

    if (need_locktree) {
        db->i->dict_id = toku_ft_get_dictionary_id(db->i->ft_handle);
        struct lt_on_create_callback_extra on_create_extra = {
            .txn = txn,
            .ft_handle = db->i->ft_handle,
        };
        db->i->lt = db->dbenv->i->ltm.get_lt(db->i->dict_id,
                                             toku_ft_get_comparator(db->i->ft_handle),
                                             &on_create_extra);
        if (db->i->lt == nullptr) {
            r = errno;
            if (r == 0) {
                r = EINVAL;
            }
            goto out;
        }
    }
    r = 0;
 
out:
    if (r != 0) {
        db->i->dict_id = DICTIONARY_ID_NONE;
        db->i->opened = 0;
        if (db->i->lt) {
            db->dbenv->i->ltm.release_lt(db->i->lt);
            db->i->lt = nullptr;
        }
    }
    return r;
}

// Return the maximum key and val size in 
// *key_size and *val_size respectively
static void
toku_db_get_max_row_size(DB * UU(db), uint32_t * max_key_size, uint32_t * max_val_size) {
    *max_key_size = 0;
    *max_val_size = 0;
    toku_ft_get_maximum_advised_key_value_lengths(max_key_size, max_val_size);
}

int toku_db_pre_acquire_fileops_lock(DB *db, DB_TXN *txn) {
    // bad hack because some environment dictionaries do not have a dname
    char *dname = db->i->dname;
    if (!dname)
        return 0;

    DBT key_in_directory = { .data = dname, .size = (uint32_t) strlen(dname)+1 };
    //Left end of range == right end of range (point lock)
    int r = toku_db_get_range_lock(db->dbenv->i->directory, txn,
            &key_in_directory, &key_in_directory,
            toku::lock_request::type::WRITE);
    if (r == 0)
        STATUS_VALUE(YDB_LAYER_DIRECTORY_WRITE_LOCKS)++;  // accountability 
    else
        STATUS_VALUE(YDB_LAYER_DIRECTORY_WRITE_LOCKS_FAIL)++;  // accountability 
    return r;
}

//
// This function is used both to set an initial descriptor of a DB and to
// change a descriptor. (only way to set a descriptor of a DB)
//
// Requires:
//  - The caller must not call put_multiple, del_multiple, or update_multiple concurrently
//  - The caller must not have a hot index running concurrently on db
//  - If the caller has passed DB_UPDATE_CMP_DESCRIPTOR as a flag, then he is calling this function
//     ONLY immediately after creating the dictionary and before doing any actual work on the dictionary.
//
static int 
toku_db_change_descriptor(DB *db, DB_TXN* txn, const DBT* descriptor, uint32_t flags) {
    HANDLE_PANICKED_DB(db);
    HANDLE_READ_ONLY_TXN(txn);
    HANDLE_DB_ILLEGAL_WORKING_PARENT_TXN(db, txn);
    int r = 0;
    TOKUTXN ttxn = txn ? db_txn_struct_i(txn)->tokutxn : NULL;
    bool is_db_hot_index  = ((flags & DB_IS_HOT_INDEX) != 0);
    bool update_cmp_descriptor = ((flags & DB_UPDATE_CMP_DESCRIPTOR) != 0);

    DBT old_descriptor_dbt;
    toku_init_dbt(&old_descriptor_dbt);

    if (!db_opened(db) || !descriptor || (descriptor->size>0 && !descriptor->data)){
        r = EINVAL;
        goto cleanup;
    }
    // For a hot index, this is an initial descriptor.
    // We do not support (yet) hcad with hot index concurrently on a single table, which
    // would require changing a descriptor for a hot index.
    if (!is_db_hot_index) {
        r = toku_db_pre_acquire_table_lock(db, txn);
        if (r != 0) { goto cleanup; }    
    }

    toku_clone_dbt(&old_descriptor_dbt, db->descriptor->dbt);
    toku_ft_change_descriptor(db->i->ft_handle, &old_descriptor_dbt, descriptor, 
                              true, ttxn, update_cmp_descriptor);

cleanup:
    toku_destroy_dbt(&old_descriptor_dbt);
    return r;
}

static int 
toku_db_set_flags(DB *db, uint32_t flags) {
    HANDLE_PANICKED_DB(db);

    /* the following matches BDB */
    if (db_opened(db) && flags != 0) return EINVAL;

    return 0;
}

static int 
toku_db_get_flags(DB *db, uint32_t *pflags) {
    HANDLE_PANICKED_DB(db);
    if (!pflags) return EINVAL;
    *pflags = 0;
    return 0;
}

static int 
toku_db_change_pagesize(DB *db, uint32_t pagesize) {
    HANDLE_PANICKED_DB(db);
    if (!db_opened(db)) return EINVAL;
    toku_ft_handle_set_nodesize(db->i->ft_handle, pagesize);
    return 0;
}

static int 
toku_db_set_pagesize(DB *db, uint32_t pagesize) {
    HANDLE_PANICKED_DB(db);
    if (db_opened(db)) return EINVAL;
    toku_ft_handle_set_nodesize(db->i->ft_handle, pagesize);
    return 0;
}

static int 
toku_db_get_pagesize(DB *db, uint32_t *pagesize_ptr) {
    HANDLE_PANICKED_DB(db);
    toku_ft_handle_get_nodesize(db->i->ft_handle, pagesize_ptr);
    return 0;
}

static int 
toku_db_change_readpagesize(DB *db, uint32_t readpagesize) {
    HANDLE_PANICKED_DB(db);
    if (!db_opened(db)) return EINVAL;
    toku_ft_handle_set_basementnodesize(db->i->ft_handle, readpagesize);
    return 0;
}

static int 
toku_db_set_readpagesize(DB *db, uint32_t readpagesize) {
    HANDLE_PANICKED_DB(db);
    if (db_opened(db)) return EINVAL;
    toku_ft_handle_set_basementnodesize(db->i->ft_handle, readpagesize);
    return 0;
}

static int 
toku_db_get_readpagesize(DB *db, uint32_t *readpagesize_ptr) {
    HANDLE_PANICKED_DB(db);
    toku_ft_handle_get_basementnodesize(db->i->ft_handle, readpagesize_ptr);
    return 0;
}

static int 
toku_db_change_compression_method(DB *db, enum toku_compression_method compression_method) {
    HANDLE_PANICKED_DB(db);
    if (!db_opened(db)) return EINVAL;
    toku_ft_handle_set_compression_method(db->i->ft_handle, compression_method);
    return 0;
}

static int 
toku_db_set_compression_method(DB *db, enum toku_compression_method compression_method) {
    HANDLE_PANICKED_DB(db);
    if (db_opened(db)) return EINVAL;
    toku_ft_handle_set_compression_method(db->i->ft_handle, compression_method);
    return 0;
}

static int 
toku_db_get_compression_method(DB *db, enum toku_compression_method *compression_method_ptr) {
    HANDLE_PANICKED_DB(db);
    toku_ft_handle_get_compression_method(db->i->ft_handle, compression_method_ptr);
    return 0;
}

static int 
toku_db_change_fanout(DB *db, unsigned int fanout) {
    HANDLE_PANICKED_DB(db);
    if (!db_opened(db)) return EINVAL;
    toku_ft_handle_set_fanout(db->i->ft_handle, fanout);
    return 0;
}

static int 
toku_db_set_fanout(DB *db, unsigned int fanout) {
    HANDLE_PANICKED_DB(db);
    if (db_opened(db)) return EINVAL;
    toku_ft_handle_set_fanout(db->i->ft_handle, fanout);
    return 0;
}

static int 
toku_db_get_fanout(DB *db, unsigned int *fanout) {
    HANDLE_PANICKED_DB(db);
    toku_ft_handle_get_fanout(db->i->ft_handle, fanout);
    return 0;
}

static int
toku_db_set_memcmp_magic(DB *db, uint8_t magic) {
    HANDLE_PANICKED_DB(db);
    if (db_opened(db)) {
        return EINVAL;
    }
    return toku_ft_handle_set_memcmp_magic(db->i->ft_handle, magic);
}

static int
toku_db_get_fractal_tree_info64(DB *db, uint64_t *num_blocks_allocated, uint64_t *num_blocks_in_use, uint64_t *size_allocated, uint64_t *size_in_use) {
    HANDLE_PANICKED_DB(db);
    struct ftinfo64 ftinfo;
    toku_ft_handle_get_fractal_tree_info64(db->i->ft_handle, &ftinfo);
    *num_blocks_allocated = ftinfo.num_blocks_allocated;
    *num_blocks_in_use = ftinfo.num_blocks_in_use;
    *size_allocated = ftinfo.size_allocated;
    *size_in_use = ftinfo.size_in_use;
    return 0;
}

static int
toku_db_iterate_fractal_tree_block_map(DB *db, int (*iter)(uint64_t,int64_t,int64_t,int64_t,int64_t,void*), void *iter_extra) {
    HANDLE_PANICKED_DB(db);
    return toku_ft_handle_iterate_fractal_tree_block_map(db->i->ft_handle, iter, iter_extra);
}

static int 
toku_db_stat64(DB * db, DB_TXN *txn, DB_BTREE_STAT64 *s) {
    HANDLE_PANICKED_DB(db);
    HANDLE_DB_ILLEGAL_WORKING_PARENT_TXN(db, txn);
    struct ftstat64_s ftstat;
    TOKUTXN tokutxn = NULL;
    if (txn != NULL) {
        tokutxn = db_txn_struct_i(txn)->tokutxn;
    }
    toku_ft_handle_stat64(db->i->ft_handle, tokutxn, &ftstat);
    s->bt_nkeys = ftstat.nkeys;
    s->bt_ndata = ftstat.ndata;
    s->bt_dsize = ftstat.dsize;
    s->bt_fsize = ftstat.fsize;
    s->bt_create_time_sec = ftstat.create_time_sec;
    s->bt_modify_time_sec = ftstat.modify_time_sec;
    s->bt_verify_time_sec = ftstat.verify_time_sec;
    return 0;
}

static const char *
toku_db_get_dname(DB *db) {
    if (!db_opened(db)) {
        return nullptr;
    }
    if (db->i->dname == nullptr) {
        return ""; 
    }
    return db->i->dname;
}

static int 
toku_db_keys_range64(DB* db, DB_TXN* txn __attribute__((__unused__)), DBT* keyleft, DBT* keyright, uint64_t* less, uint64_t* left, uint64_t* between, uint64_t *right, uint64_t *greater, bool* middle_3_exact) {
    HANDLE_PANICKED_DB(db);
    HANDLE_DB_ILLEGAL_WORKING_PARENT_TXN(db, txn);

    // note that we ignore the txn param.  It would be more complicated to support it.
    // TODO(yoni): Maybe add support for txns later?  How would we do this?  ydb lock comment about db_keyrange64 is obsolete.
    toku_ft_keysrange(db->i->ft_handle, keyleft, keyright, less, left, between, right, greater, middle_3_exact);
    return 0;
}

static int 
toku_db_key_range64(DB* db, DB_TXN* txn, DBT* key, uint64_t* less_p, uint64_t* equal_p, uint64_t* greater_p, int* is_exact) {
    uint64_t less, equal_left, middle, equal_right, greater;
    bool ignore;
    int r = toku_db_keys_range64(db, txn, key, NULL, &less, &equal_left, &middle, &equal_right, &greater, &ignore);
    if (r == 0) {
        *less_p = less;
        *equal_p = equal_left;
        *greater_p = middle;
        paranoid_invariant_zero(greater);  // no keys are greater than positive infinity
        paranoid_invariant_zero(equal_right);  // no keys are equal to positive infinity
        // toku_ft_keysrange does not know when all 3 are exact, so set is_exact to false
        *is_exact = false;
    }
    return 0;
}

static int toku_db_get_key_after_bytes(DB *db, DB_TXN *txn, const DBT *start_key, uint64_t skip_len, void (*callback)(const DBT *end_key, uint64_t actually_skipped, void *extra), void *cb_extra, uint32_t UU(flags)) {
    HANDLE_PANICKED_DB(db);
    HANDLE_DB_ILLEGAL_WORKING_PARENT_TXN(db, txn);
    return toku_ft_get_key_after_bytes(db->i->ft_handle, start_key, skip_len, callback, cb_extra);
}

// needed by loader.c
int 
toku_db_pre_acquire_table_lock(DB *db, DB_TXN *txn) {
    HANDLE_PANICKED_DB(db);
    if (!db->i->lt || !txn) return 0;
    int r;
    r = toku_db_get_range_lock(db, txn, 
            toku_dbt_negative_infinity(), toku_dbt_positive_infinity(),
            toku::lock_request::type::WRITE);
    return r;
}

static int 
locked_db_close(DB * db, uint32_t UU(flags)) {
    // cannot begin a checkpoint
    toku_multi_operation_client_lock();
    int r = toku_db_close(db);
    toku_multi_operation_client_unlock();
    return r;
}

int 
autotxn_db_get(DB* db, DB_TXN* txn, DBT* key, DBT* data, uint32_t flags) {
    bool changed; int r;
    r = toku_db_construct_autotxn(db, &txn, &changed, false);
    if (r!=0) return r;
    r = toku_db_get(db, txn, key, data, flags);
    return toku_db_destruct_autotxn(txn, r, changed);
}

static inline int 
autotxn_db_getf_set (DB *db, DB_TXN *txn, uint32_t flags, DBT *key, YDB_CALLBACK_FUNCTION f, void *extra) {
    bool changed; int r;
    r = toku_db_construct_autotxn(db, &txn, &changed, false);
    if (r!=0) return r;
    r = db_getf_set(db, txn, flags, key, f, extra);
    return toku_db_destruct_autotxn(txn, r, changed);
}

static int 
locked_db_open(DB *db, DB_TXN *txn, const char *fname, const char *dbname, DBTYPE dbtype, uint32_t flags, int mode) {
    int ret, r;
    HANDLE_READ_ONLY_TXN(txn);
    HANDLE_DB_ILLEGAL_WORKING_PARENT_TXN(db, txn);

    //
    // Note that this function opens a db with a transaction. Should
    // the transaction abort, the user is responsible for closing the DB
    // before aborting the transaction. Not doing so results in undefined
    // behavior.
    //    
    DB_ENV *env = db->dbenv;
    DB_TXN *child_txn = NULL;
    int using_txns = env->i->open_flags & DB_INIT_TXN;
    if (using_txns) {
        ret = toku_txn_begin(env, txn, &child_txn, DB_TXN_NOSYNC);
        invariant_zero(ret);
    }

    // cannot begin a checkpoint
    toku_multi_operation_client_lock();
    r = toku_db_open(db, child_txn, fname, dbname, dbtype, flags & ~DB_AUTO_COMMIT, mode);
    toku_multi_operation_client_unlock();

    if (using_txns) {
        if (r == 0) {
            ret = locked_txn_commit(child_txn, DB_TXN_NOSYNC);
            invariant_zero(ret);
        } else {
            ret = locked_txn_abort(child_txn);
            invariant_zero(ret);
        }
    }
    return r;
}

static int 
locked_db_change_descriptor(DB *db, DB_TXN *txn, const DBT *descriptor, uint32_t flags) {
    // cannot begin a checkpoint
    toku_multi_operation_client_lock();
    int r = toku_db_change_descriptor(db, txn, descriptor, flags);
    toku_multi_operation_client_unlock();
    return r;
}

static int
autotxn_db_change_descriptor(DB *db, DB_TXN *txn, const DBT *descriptor, uint32_t flags) {
    bool changed; int r;
    r = toku_db_construct_autotxn(db, &txn, &changed, false);
    if (r != 0) { return r; }
    r = locked_db_change_descriptor(db, txn, descriptor, flags);
    return toku_db_destruct_autotxn(txn, r, changed);
}

static void 
toku_db_set_errfile (DB *db, FILE *errfile) {
    db->dbenv->set_errfile(db->dbenv, errfile);
}

// TODO 2216 delete this
static int 
toku_db_fd(DB * UU(db), int * UU(fdp)) {
    return 0;
}

static const DBT* toku_db_dbt_pos_infty(void) __attribute__((pure));
static const DBT*
toku_db_dbt_pos_infty(void) {
    return toku_dbt_positive_infinity();
}

static const DBT* toku_db_dbt_neg_infty(void) __attribute__((pure));
static const DBT* 
toku_db_dbt_neg_infty(void) {
    return toku_dbt_negative_infinity();
}

static int
toku_db_optimize(DB *db) {
    HANDLE_PANICKED_DB(db);
    toku_ft_optimize(db->i->ft_handle);
    return 0;
}

static int
toku_db_hot_optimize(DB *db, DBT* left, DBT* right,
                     int (*progress_callback)(void *extra, float progress),
                     void *progress_extra, uint64_t* loops_run)
{
    HANDLE_PANICKED_DB(db);
    int r = 0;
    r = toku_ft_hot_optimize(db->i->ft_handle, left, right,
                              progress_callback,
                              progress_extra, loops_run);

    return r;
}

static int 
locked_db_optimize(DB *db) {
    // need to protect from checkpointing because
    // toku_db_optimize does a message injection
    toku_multi_operation_client_lock(); //Cannot begin checkpoint
    int r = toku_db_optimize(db);
    toku_multi_operation_client_unlock();
    return r;
}


struct last_key_extra {
    YDB_CALLBACK_FUNCTION func;
    void* extra;
};

static int
db_get_last_key_callback(uint32_t keylen, const void *key, uint32_t vallen UU(), const void *val UU(), void *extra, bool lock_only) {
    if (!lock_only) {
        DBT keydbt;
        toku_fill_dbt(&keydbt, key, keylen);
        struct last_key_extra * CAST_FROM_VOIDP(info, extra);
        info->func(&keydbt, NULL, info->extra);
    }
    return 0;
}

static int
toku_db_get_last_key(DB * db, DB_TXN *txn, YDB_CALLBACK_FUNCTION func, void* extra) {
    int r;
    LE_CURSOR cursor = nullptr;
    struct last_key_extra last_extra = { .func = func, .extra = extra };

    r = toku_le_cursor_create(&cursor, db->i->ft_handle, db_txn_struct_i(txn)->tokutxn);
    if (r != 0) { goto cleanup; }

    // Goes in reverse order.  First key returned is last in dictionary.
    r = toku_le_cursor_next(cursor, db_get_last_key_callback, &last_extra);
    if (r != 0) { goto cleanup; }

cleanup:
    if (cursor) {
        toku_le_cursor_close(cursor);
    }
    return r;
}

static int
autotxn_db_get_last_key(DB* db, YDB_CALLBACK_FUNCTION func, void* extra) {
    bool changed; int r;
    DB_TXN *txn = nullptr;
    // Cursors inside require transactions, but this is _not_ a transactional function.
    // Create transaction in a wrapper and then later close it.
    r = toku_db_construct_autotxn(db, &txn, &changed, false);
    if (r!=0) return r;
    r = toku_db_get_last_key(db, txn, func, extra);
    return toku_db_destruct_autotxn(txn, r, changed);
}

static int
toku_db_get_fragmentation(DB * db, TOKU_DB_FRAGMENTATION report) {
    HANDLE_PANICKED_DB(db);
    int r;
    if (!db_opened(db))
        r = toku_ydb_do_error(db->dbenv, EINVAL, "Fragmentation report available only on open DBs.\n");
    else
        r = toku_ft_get_fragmentation(db->i->ft_handle, report);
    return r;
}

int 
toku_db_set_indexer(DB *db, DB_INDEXER * indexer) {
    int r = 0;
    if ( db->i->indexer != NULL && indexer != NULL ) {
        // you are trying to overwrite a valid indexer
        r = EINVAL;
    }
    else {
        db->i->indexer = indexer;
    }
    return r;
}

DB_INDEXER *
toku_db_get_indexer(DB *db) {
    return db->i->indexer;
}

static void 
db_get_indexer(DB *db, DB_INDEXER **indexer_ptr) {
    *indexer_ptr = toku_db_get_indexer(db);
}

struct ydb_verify_context {
    int (*progress_callback)(void *extra, float progress);
    void *progress_extra;
};

static int
ydb_verify_progress_callback(void *extra, float progress) {
    struct ydb_verify_context *context = (struct ydb_verify_context *) extra;
    int r = 0;
    if (context->progress_callback) {
        r = context->progress_callback(context->progress_extra, progress);
    }
    return r;
}

static int
toku_db_verify_with_progress(DB *db, int (*progress_callback)(void *extra, float progress), void *progress_extra, int verbose, int keep_going) {
    struct ydb_verify_context context = { progress_callback, progress_extra };
    int r = toku_verify_ft_with_progress(db->i->ft_handle, ydb_verify_progress_callback, &context, verbose, keep_going);
    return r;
}


static int
toku_db_recount_rows(DB* db, int (*progress_callback)(uint64_t count,
                                                      uint64_t deleted,
                                                      void* progress_extra),
                     void* progress_extra) {

    HANDLE_PANICKED_DB(db);
    int r = 0;
    r =
        toku_ft_recount_rows(
            db->i->ft_handle,
            progress_callback,
            progress_extra);

    return r;
}


int toku_setup_db_internal (DB **dbp, DB_ENV *env, uint32_t flags, FT_HANDLE ft_handle, bool is_open) {
    if (flags || env == NULL) 
        return EINVAL;

    if (!env_opened(env))
        return EINVAL;
    
    DB *MALLOC(result);
    if (result == 0) {
        return ENOMEM;
    }
    memset(result, 0, sizeof *result);
    result->dbenv = env;
    MALLOC(result->i);
    if (result->i == 0) {
        toku_free(result);
        return ENOMEM;
    }
    memset(result->i, 0, sizeof *result->i);
    result->i->ft_handle = ft_handle;
    result->i->opened = is_open;
    *dbp = result;
    return 0;
}

int 
toku_db_create(DB ** db, DB_ENV * env, uint32_t flags) {
    if (flags || env == NULL) 
        return EINVAL;

    if (!env_opened(env))
        return EINVAL;
    

    FT_HANDLE ft_handle;
    toku_ft_handle_create(&ft_handle);

    int r = toku_setup_db_internal(db, env, flags, ft_handle, false);
    if (r != 0) return r;

    DB *result=*db;
    // methods that grab the ydb lock
#define SDB(name) result->name = locked_db_ ## name
    SDB(close);
    SDB(open);
    SDB(optimize);
#undef SDB
    // methods that do not take the ydb lock
#define USDB(name) result->name = toku_db_ ## name
    USDB(set_errfile);
    USDB(set_pagesize);
    USDB(get_pagesize);
    USDB(change_pagesize);
    USDB(set_readpagesize);
    USDB(get_readpagesize);
    USDB(change_readpagesize);
    USDB(set_compression_method);
    USDB(get_compression_method);
    USDB(change_compression_method);
    USDB(set_fanout);
    USDB(get_fanout);
    USDB(set_memcmp_magic);
    USDB(change_fanout);
    USDB(set_flags);
    USDB(get_flags);
    USDB(fd);
    USDB(get_max_row_size);
    USDB(set_indexer);
    USDB(pre_acquire_table_lock);
    USDB(pre_acquire_fileops_lock);
    USDB(key_range64);
    USDB(keys_range64);
    USDB(get_key_after_bytes);
    USDB(hot_optimize);
    USDB(stat64);
    USDB(get_fractal_tree_info64);
    USDB(iterate_fractal_tree_block_map);
    USDB(get_dname);
    USDB(verify_with_progress);
    USDB(cursor);
    USDB(dbt_pos_infty);
    USDB(dbt_neg_infty);
    USDB(get_fragmentation);
    USDB(recount_rows);
#undef USDB
    result->get_indexer = db_get_indexer;
    result->del = autotxn_db_del;
    result->put = autotxn_db_put;
    result->update = autotxn_db_update;
    result->update_broadcast = autotxn_db_update_broadcast;
    result->change_descriptor = autotxn_db_change_descriptor;
    result->get_last_key = autotxn_db_get_last_key;
    
    // unlocked methods
    result->get = autotxn_db_get;
    result->getf_set = autotxn_db_getf_set;
    
    result->i->dict_id = DICTIONARY_ID_NONE;
    result->i->opened = 0;
    result->i->open_flags = 0;
    result->i->open_mode = 0;
    result->i->indexer = NULL;
    *db = result;
    return 0;
}

// When the loader is created, it makes this call (toku_env_load_inames).
// For each dictionary to be loaded, replace old iname in directory
// with a newly generated iname.  This will also take a write lock
// on the directory entries.  The write lock will be released when
// the transaction of the loader is completed.
// If the transaction commits, the new inames are in place.
// If the transaction aborts, the old inames will be restored.
// The new inames are returned to the caller.  
// It is the caller's responsibility to free them.
// If "mark_as_loader" is true, then include a mark in the iname
// to indicate that the file is created by the ft loader.
// Return 0 on success (could fail if write lock not available).
static int
load_inames(DB_ENV * env, DB_TXN * txn, int N, DB * dbs[/*N*/], const char * new_inames_in_env[/*N*/], LSN *load_lsn, bool mark_as_loader) {
    int rval = 0;
    int i;
    
    TXNID_PAIR xid = TXNID_PAIR_NONE;
    DBT dname_dbt;  // holds dname
    DBT iname_dbt;  // holds new iname
    
    const char *mark;

    if (mark_as_loader) {
        mark = "B";
    } else {
        mark = "P";
    }

    for (i=0; i<N; i++) {
        new_inames_in_env[i] = NULL;
    }

    if (txn) {
        xid = toku_txn_get_txnid(db_txn_struct_i(txn)->tokutxn);
    }
    for (i = 0; i < N; i++) {
        char * dname = dbs[i]->i->dname;
        toku_fill_dbt(&dname_dbt, dname, strlen(dname)+1);

        // now create new iname
        char hint[strlen(dname) + 1];
        create_iname_hint(env, dname, hint);

        // allocates memory for iname_in_env
        const char *new_iname =
            create_iname(env, xid.parent_id64, xid.child_id64, hint, mark, i);
        new_inames_in_env[i] = new_iname;

        // iname_in_env goes in directory
        toku_fill_dbt(&iname_dbt, new_iname, strlen(new_iname) + 1);
        rval = toku_db_put(env->i->directory, txn, &dname_dbt, &iname_dbt, 0, true);
        if (rval) break;
    }

    // Generate load log entries.
    if (!rval && txn) {
        TOKUTXN ttxn = db_txn_struct_i(txn)->tokutxn;
        int do_fsync = 0;
        LSN *get_lsn = NULL;
        for (i = 0; i < N; i++) {
            FT_HANDLE ft_handle  = dbs[i]->i->ft_handle;
            //Fsync is necessary for the last one only.
            if (i==N-1) {
                do_fsync = 1; //We only need a single fsync of logs.
                get_lsn  = load_lsn; //Set pointer to capture the last lsn.
            }
            toku_ft_load(ft_handle, ttxn, new_inames_in_env[i], do_fsync, get_lsn);
        }
    }
    return rval;
}

int
locked_load_inames(DB_ENV * env, DB_TXN * txn, int N, DB * dbs[/*N*/], char * new_inames_in_env[/*N*/], LSN *load_lsn, bool mark_as_loader) {
    int r;
    HANDLE_READ_ONLY_TXN(txn);

    // cannot begin a checkpoint
    toku_multi_operation_client_lock();
    r = load_inames(env, txn, N, dbs, (const char **) new_inames_in_env, load_lsn, mark_as_loader);
    toku_multi_operation_client_unlock();

    return r;

}

#undef STATUS_VALUE

#include <toku_race_tools.h>
void __attribute__((constructor)) toku_ydb_db_helgrind_ignore(void);
void
toku_ydb_db_helgrind_ignore(void) {
    TOKU_VALGRIND_HG_DISABLE_CHECKING(&ydb_db_layer_status, sizeof ydb_db_layer_status);
}