SQLite4
Check-in [29373a8844]
Not logged in

Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

Overview
Comment:Add start of checkpointing code to btree.
Downloads: Tarball | ZIP archive | SQL archive
Timelines: family | ancestors | descendants | both | trunk
Files: files | file ages | folders
SHA1: 29373a8844687bc9c444338bcf7982376e0672b1
User & Date: dan 2013-10-25 20:20:08
Context
2013-10-26
12:44
Use sqlite4_malloc instead of malloc in test fixture for num check-in: 3f6924e784 user: peterreid tags: trunk
2013-10-25
20:20
Add start of checkpointing code to btree. check-in: 29373a8844 user: dan tags: trunk
02:37
Fix buffer overread when parsing a UTF16 exponent check-in: 9e61c90be4 user: peterreid tags: trunk
Changes
Hide Diffs Unified Diffs Ignore Whitespace Patch

Changes to lsm-test/lsmtest_main.c.

42
43
44
45
46
47
48



49
50
51
52
53
54
55
  void *pVal, int nVal,           /* Expected value */
  int *pRc                        /* IN/OUT: Error code */
){
  if( *pRc==0 ){
    void *pDbVal;
    int nDbVal;
    int rc;



    rc = tdb_fetch(pDb, pKey, nKey, &pDbVal, &nDbVal);
    testSetError(rc);
    if( rc==0 && (nVal!=nDbVal || (nVal>0 && lsm_memcmp(pVal, pDbVal, nVal))) ){
      testSetError(1);
    }
  }
}







>
>
>







42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
  void *pVal, int nVal,           /* Expected value */
  int *pRc                        /* IN/OUT: Error code */
){
  if( *pRc==0 ){
    void *pDbVal;
    int nDbVal;
    int rc;

    static int nCall = 0; nCall++;

    rc = tdb_fetch(pDb, pKey, nKey, &pDbVal, &nDbVal);
    testSetError(rc);
    if( rc==0 && (nVal!=nDbVal || (nVal>0 && lsm_memcmp(pVal, pDbVal, nVal))) ){
      testSetError(1);
    }
  }
}

Changes to src/btInt.h.

25
26
27
28
29
30
31







32
33
34
35
36
37
38
...
183
184
185
186
187
188
189

190
191
192
193
194
195
196

/* Number of elements in an array object. */
#define array_size(x) (sizeof(x)/sizeof(x[0]))

/* Number of read-lock slots in shared memory */
#define BT_NREADER 4









/*************************************************************************
** Interface to bt_pager.c functionality.
*/
typedef struct BtPage BtPage;
typedef struct BtPager BtPager;

................................................................................
int sqlite4BtLogRead(BtLog*, u32 pgno, u8 *aData);
int sqlite4BtLogWrite(BtLog*, u32 pgno, u8 *aData, int bCommit);

int sqlite4BtLogSnapshotOpen(BtLog*);
int sqlite4BtLogSnapshotClose(BtLog*);
int sqlite4BtLogSnapshotWritable(BtLog*);


int sqlite4BtLogCheckpoint(BtLog*);

/*
** End of bt_log.c interface.
*************************************************************************/

/*************************************************************************







>
>
>
>
>
>
>







 







>







25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
...
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204

/* Number of elements in an array object. */
#define array_size(x) (sizeof(x)/sizeof(x[0]))

/* Number of read-lock slots in shared memory */
#define BT_NREADER 4

#ifndef MIN
# define MIN(a,b) (((a)<(b))?(a):(b))
#endif
#ifndef MAX
# define MAX(a,b) (((a)>(b))?(a):(b))
#endif


/*************************************************************************
** Interface to bt_pager.c functionality.
*/
typedef struct BtPage BtPage;
typedef struct BtPager BtPager;

................................................................................
int sqlite4BtLogRead(BtLog*, u32 pgno, u8 *aData);
int sqlite4BtLogWrite(BtLog*, u32 pgno, u8 *aData, int bCommit);

int sqlite4BtLogSnapshotOpen(BtLog*);
int sqlite4BtLogSnapshotClose(BtLog*);
int sqlite4BtLogSnapshotWritable(BtLog*);

int sqlite4BtLogSize(BtLog*);
int sqlite4BtLogCheckpoint(BtLog*);

/*
** End of bt_log.c interface.
*************************************************************************/

/*************************************************************************

Changes to src/bt_lock.c.

7
8
9
10
11
12
13
14
15
16

17
18
19
20
21
22
23
...
400
401
402
403
404
405
406
407








**    May you do good and not evil.
**    May you find forgiveness for yourself and forgive others.
**    May you share freely, never taking more than you give.
**
*************************************************************************
*/

#include "btInt.h"

#include "sqliteInt.h"


#include <string.h>
#include <assert.h>
#include <stdio.h>

#define BT_LOCK_DMS1       0      /* DMS1 */
#define BT_LOCK_DMS2_RW    1      /* DMS2/rw */
................................................................................
    }
  }
  sqlite4_mutex_leave(pShared->pClientMutex);
  
  *ppOut = pOut;
  return rc;
}
















<


>







 








>
>
>
>
>
>
>
>
7
8
9
10
11
12
13

14
15
16
17
18
19
20
21
22
23
...
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
**    May you do good and not evil.
**    May you find forgiveness for yourself and forgive others.
**    May you share freely, never taking more than you give.
**
*************************************************************************
*/



#include "sqliteInt.h"
#include "btInt.h"

#include <string.h>
#include <assert.h>
#include <stdio.h>

#define BT_LOCK_DMS1       0      /* DMS1 */
#define BT_LOCK_DMS2_RW    1      /* DMS2/rw */
................................................................................
    }
  }
  sqlite4_mutex_leave(pShared->pClientMutex);
  
  *ppOut = pOut;
  return rc;
}

int sqlite4BtLockCkpt(BtLock *pLock){
  return btLockLockop(pLock, BT_LOCK_CKPTER, BT_LOCK_EXCL, 0);
}

int sqlite4BtLockCkptUnlock(BtLock *pLock){
  return btLockLockop(pLock, BT_LOCK_CKPTER, BT_LOCK_UNLOCK, 0);
}

Changes to src/bt_log.c.

71
72
73
74
75
76
77












78
79
80

81
82
83
84
85
86
87
...
478
479
480
481
482
483
484

485
486
487
488
489
490
491
...
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
...
548
549
550
551
552
553
554




555
556
557
558
559
560
561
...
647
648
649
650
651
652
653

654
655
656
657

658
659
660
661
662
663
664
665
...
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


































































  int padding;
  u32 aCksum[2];                  /* Object checksum */
};

/*
** A single instance of this structure follows the two BtShmHdr structures 
** in shared memory.












*/
struct BtCkptHdr {
  u32 iFirstRead;                 /* First uncheckpointed frame */

  u32 iFirstRecover;              /* First recovery frame */
};

struct BtShm {
  BtShmHdr hdr1;
  BtShmHdr hdr2;
  BtCkptHdr ckpt;
................................................................................
** initialize the shared-memory accordingly.
*/
static int btLogRecover(BtLog *pLog){
  bt_env *pVfs = pLog->pLock->pVfs;
  i64 nByte = 0;                  /* Size of log file on disk */
  int rc;                         /* Return code */
  BtWalHdr *pHdr = 0;


  /* Read a log file header from the start of the file. */
  rc = pVfs->xSize(pLog->pFd, &nByte);
  if( rc==SQLITE4_OK && nByte>0 ){
    BtWalHdr hdr1;
    rc = btLogReadHeader(pLog, 0, &hdr1);
    if( rc==SQLITE4_OK ){
................................................................................
      rc = btLogReadHeader(pLog, hdr1.nSector, &hdr2);
      if( rc==SQLITE4_NOTFOUND ){
        pHdr = &hdr1;
      }else if( rc==SQLITE4_OK ){
        int aGreater[3] = {1, 2, 0};
        pHdr = ((hdr2.iCnt==aGreater[hdr1.iCnt]) ? &hdr2 : &hdr1);
      }

    }else if( rc==SQLITE4_NOTFOUND ){
      int iOff;
      for(iOff=256; iOff<=65536 && rc==SQLITE4_NOTFOUND; iOff=iOff*2){
        rc = btLogReadHeader(pLog, iOff, &hdr1);
      }
      if( rc==SQLITE4_OK ){
        pHdr = &hdr1;

      }
    }
    if( rc==SQLITE4_NOTFOUND ) rc = SQLITE4_OK;
  }

  /* If a header was successfully read from the file, scan the log file
  ** and populate the shared-memory hash tables.  */
  if( pHdr ){
    u32 iCommit = 0;
    rc = btLogTraverse(pLog, pHdr, btLogRecoverFrame, (void*)&iCommit);
    if( rc==SQLITE4_OK && iCommit>0 ){




      pLog->snapshot.aLog[4] = 1;
      pLog->snapshot.aLog[5] = iCommit;
      rc = btLogHashRollback(pLog, btLogFrameHash(pLog, iCommit), iCommit);
    }
  }

  return rc;
................................................................................

  zWal = sqlite4BtPagerFilename(pPager, BT_PAGERFILE_LOG);
  rc = pVfs->xOpen(pEnv, pVfs, zWal, 0, &pLog->pFd);

  if( rc==SQLITE4_OK && bRecover ){
    rc = btLogMapShm(pLog, 0);
    if( rc==SQLITE4_OK ){




      btLogZeroSnapshot(pLog);
      rc = btLogRecover(pLog);
    }
    if( rc==SQLITE4_OK ){
      rc = btLogUpdateSharedHdr(pLog);
    }
  }
................................................................................
**
** If any other error occurs, an SQLite4 error code is returned. The final
** state of buffer aData[] is undefined in this case.
*/
int sqlite4BtLogRead(BtLog *pLog, u32 pgno, u8 *aData){
  const int pgsz = sqlite4BtPagerPagesize((BtPager*)(pLog->pLock));
  int iHash;

  int rc = SQLITE4_NOTFOUND;
  u32 iFrame = 0;

  iHash = btLogFrameHash(pLog, pLog->snapshot.aLog[5]);

  for( ; rc==SQLITE4_NOTFOUND && iHash>=0 && iFrame==0; iHash--){
    rc = btLogHashSearch(pLog, iHash, pgno, &iFrame);
  }

  if( rc==SQLITE4_OK && iFrame!=0 ){
    bt_env *pVfs = pLog->pLock->pVfs;
    i64 iOff;
    assert( rc==SQLITE4_OK );
................................................................................
      rc = btLogSnapshot(pLog, &shmhdr);
    }
    if( rc==SQLITE4_OK && memcmp(&shmhdr, &pLog->snapshot, sizeof(BtShmHdr)) ){
      sqlite4BtLockReaderUnlock(pLog->pLock);
      rc = SQLITE4_NOTFOUND;
    }
  }









  return rc;
}

int sqlite4BtLogSnapshotClose(BtLog *pLog){
  sqlite4BtLockReaderUnlock(pLog->pLock);
  return SQLITE4_OK;
}

int sqlite4BtLogSnapshotWritable(BtLog *pLog){
  return 1;
}












































































































int sqlite4BtLogCheckpoint(BtLog *pLog){





  return SQLITE4_OK;










}






















































































>
>
>
>
>
>
>
>
>
>
>
>



>







 







>







 







>







>











>
>
>
>







 







>
>
>
>







 







>




>
|







 








>
>
>
>
>
>
>
>












>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>

>
>
>
>
>
|
>
>
>
>
>
>
>
>
>
>
|
>
>
>
|
>
>
>
>
>
>
>
>
|
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
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
...
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
...
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
...
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
...
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
...
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
  int padding;
  u32 aCksum[2];                  /* Object checksum */
};

/*
** A single instance of this structure follows the two BtShmHdr structures 
** in shared memory.
**
** iWalHdr:
**   This field encodes two pieces of information: the location of the
**   current log file header (slot 0 or slot 1) and the value of the
**   BtWalHdr.iCnt field within it (0, 1 or 2). Encoded as:
**
**           ((iSlot << 2) + iCnt)
**
**   To decode, use:
**
**           iSlot = (ckpthdr.iWalHdr >> 2);
**           iCnt = (ckpthdr.iWalHdr & 0x03);& 0x03);
*/
struct BtCkptHdr {
  u32 iFirstRead;                 /* First uncheckpointed frame */
  u32 iWalHdr;                    /* Description of current wal header */
  u32 iFirstRecover;              /* First recovery frame */
};

struct BtShm {
  BtShmHdr hdr1;
  BtShmHdr hdr2;
  BtCkptHdr ckpt;
................................................................................
** initialize the shared-memory accordingly.
*/
static int btLogRecover(BtLog *pLog){
  bt_env *pVfs = pLog->pLock->pVfs;
  i64 nByte = 0;                  /* Size of log file on disk */
  int rc;                         /* Return code */
  BtWalHdr *pHdr = 0;
  int iSlot = 0;

  /* Read a log file header from the start of the file. */
  rc = pVfs->xSize(pLog->pFd, &nByte);
  if( rc==SQLITE4_OK && nByte>0 ){
    BtWalHdr hdr1;
    rc = btLogReadHeader(pLog, 0, &hdr1);
    if( rc==SQLITE4_OK ){
................................................................................
      rc = btLogReadHeader(pLog, hdr1.nSector, &hdr2);
      if( rc==SQLITE4_NOTFOUND ){
        pHdr = &hdr1;
      }else if( rc==SQLITE4_OK ){
        int aGreater[3] = {1, 2, 0};
        pHdr = ((hdr2.iCnt==aGreater[hdr1.iCnt]) ? &hdr2 : &hdr1);
      }
      iSlot = (pHdr==&hdr2);
    }else if( rc==SQLITE4_NOTFOUND ){
      int iOff;
      for(iOff=256; iOff<=65536 && rc==SQLITE4_NOTFOUND; iOff=iOff*2){
        rc = btLogReadHeader(pLog, iOff, &hdr1);
      }
      if( rc==SQLITE4_OK ){
        pHdr = &hdr1;
        iSlot = 1;
      }
    }
    if( rc==SQLITE4_NOTFOUND ) rc = SQLITE4_OK;
  }

  /* If a header was successfully read from the file, scan the log file
  ** and populate the shared-memory hash tables.  */
  if( pHdr ){
    u32 iCommit = 0;
    rc = btLogTraverse(pLog, pHdr, btLogRecoverFrame, (void*)&iCommit);
    if( rc==SQLITE4_OK && iCommit>0 ){
      BtShm *pShm = btLogShm(pLog);
      pShm->ckpt.iFirstRead = 1;
      pShm->ckpt.iWalHdr = (iSlot<<2) + pHdr->iCnt;
      pShm->ckpt.iFirstRecover = 1;
      pLog->snapshot.aLog[4] = 1;
      pLog->snapshot.aLog[5] = iCommit;
      rc = btLogHashRollback(pLog, btLogFrameHash(pLog, iCommit), iCommit);
    }
  }

  return rc;
................................................................................

  zWal = sqlite4BtPagerFilename(pPager, BT_PAGERFILE_LOG);
  rc = pVfs->xOpen(pEnv, pVfs, zWal, 0, &pLog->pFd);

  if( rc==SQLITE4_OK && bRecover ){
    rc = btLogMapShm(pLog, 0);
    if( rc==SQLITE4_OK ){
      BtShm *pShm = btLogShm(pLog);
      memset(pShm, 0, sizeof(BtShm));
      pShm->ckpt.iFirstRead = 1;
      pShm->ckpt.iFirstRecover = 1;
      btLogZeroSnapshot(pLog);
      rc = btLogRecover(pLog);
    }
    if( rc==SQLITE4_OK ){
      rc = btLogUpdateSharedHdr(pLog);
    }
  }
................................................................................
**
** If any other error occurs, an SQLite4 error code is returned. The final
** state of buffer aData[] is undefined in this case.
*/
int sqlite4BtLogRead(BtLog *pLog, u32 pgno, u8 *aData){
  const int pgsz = sqlite4BtPagerPagesize((BtPager*)(pLog->pLock));
  int iHash;
  int iHashLast;
  int rc = SQLITE4_NOTFOUND;
  u32 iFrame = 0;

  iHash = btLogFrameHash(pLog, pLog->snapshot.aLog[5]);
  iHashLast = btLogFrameHash(pLog, pLog->snapshot.aLog[4]);
  for( ; rc==SQLITE4_NOTFOUND && iHash>=iHashLast && iFrame==0; iHash--){
    rc = btLogHashSearch(pLog, iHash, pgno, &iFrame);
  }

  if( rc==SQLITE4_OK && iFrame!=0 ){
    bt_env *pVfs = pLog->pLock->pVfs;
    i64 iOff;
    assert( rc==SQLITE4_OK );
................................................................................
      rc = btLogSnapshot(pLog, &shmhdr);
    }
    if( rc==SQLITE4_OK && memcmp(&shmhdr, &pLog->snapshot, sizeof(BtShmHdr)) ){
      sqlite4BtLockReaderUnlock(pLog->pLock);
      rc = SQLITE4_NOTFOUND;
    }
  }

  if( rc==SQLITE4_OK ){
    BtShm *pShm = btLogShm(pLog);
    u32 iCkpt = pShm->ckpt.iFirstRead;
    if( iCkpt>pLog->snapshot.aLog[4] ){
      pLog->snapshot.aLog[4] = iCkpt;
    }
  }

  return rc;
}

int sqlite4BtLogSnapshotClose(BtLog *pLog){
  sqlite4BtLockReaderUnlock(pLog->pLock);
  return SQLITE4_OK;
}

int sqlite4BtLogSnapshotWritable(BtLog *pLog){
  return 1;
}

static void btLogMergeInplace(
  u32 *aLeft, int nLeft,          /* Left hand input array */
  u32 *aRight, int nRight,        /* Right hand input array */
  u32 *aSpace,                    /* Temporary space */
  int *pnOut                      /* OUT: Size of aLeft[] after merge */
){
  int iLeft = 0;
  int iRight = 0;
  int iOut = 0;

  while( iLeft<nLeft || iRight<nRight ){
    u32 v;
    if( iRight==nRight || (iLeft<nLeft && aLeft[iLeft]<aRight[iRight]) ){
      assert( iLeft<nLeft );
      v = aLeft[iLeft++];
    }else{
      assert( iRight<nRight );
      v = aRight[iRight++];
    }
    if( v && (iOut==0 || v!=aSpace[iOut-1]) ) aSpace[iOut++] = v;
  }

  memcpy(aLeft, aSpace, iOut*sizeof(u32));
  memset(&aLeft[iOut], 0, ((nLeft+nRight)-iOut) * sizeof(u32));
  *pnOut = iOut;
}

static void btLogMergeSort(
  u32 *aPgno,                     /* Array to sort */
  int *pnPgno,                    /* IN/OUT: Number of entries in aPgno[] */
  u32 *aSpace                     /* Temporary space */
){
  int nMerge;
  int nPgno = *pnPgno;

  for(nMerge=1; nMerge<nPgno; nMerge=nMerge*2){
    int iLeft;
    for(iLeft=0; iLeft<nPgno; iLeft+=(nMerge*2)){
      u32 *aLeft = &aPgno[iLeft];
      int nLeft = MIN(nMerge, nPgno-iLeft);
      u32 *aRight = &aPgno[iLeft+nMerge];
      int nRight = MIN(nMerge, nPgno-iLeft-nLeft);
      btLogMergeInplace(aLeft, nLeft, aRight, nRight, aSpace, pnPgno);
    }
  }
}

/*
** Parameters iFirst and iLast are frame numbers for frames that are part 
** of the current log. This function scans the wal-index from iFirst to
** iLast (inclusive) and records the set of page numbers that occur once.
** This set is sorted in ascending order and returned via the output 
** variables *paPgno and *pnPgno.
** 
*/
static int btLogGatherPgno(
  BtLog *pLog,                    /* Log module handle */
  u32 iFirst,                     /* First frame to gather pgnos from */
  u32 iLast,                      /* Last frame to gather pgnos from */
  u32 **paPgno,                   /* OUT: s4_malloc'd array of sorted pgno */
  int *pnPgno                     /* OUT: Number of entries in *paPgno */
){
  u32 i;
  u32 *aPgno;                     /* Returned array */
  u32 *aSpace;                    /* Temporary space used by merge-sort */
  int nMax = (iLast - iFirst) + 1;
  int rc = SQLITE4_OK;

  /* Allocate space to collect all page numbers. */
  aPgno = (u32*)sqlite4_malloc(pLog->pLock->pEnv, sizeof(u32)*nMax*2);
  if( aPgno==0 ) rc = btErrorBkpt(SQLITE4_NOMEM);
  aSpace = &aPgno[nMax];

  /* Copy the required page numbers into the allocated array */
  for(i=iFirst; rc==SQLITE4_OK && i<=iLast; i++){
    u32 *aPage;
    ht_slot *aHash;
    u32 iZero;
    rc = btLogFindHash(pLog, btLogFrameHash(pLog, i), &aHash, &aPage, &iZero);
    if( rc==SQLITE4_OK ){
      aPgno[i-iFirst] = aPage[i-iZero];
    }
  }

  /* Sort the contents of the array in ascending order. This step also 
  ** eliminates any  duplicate page numbers. */
  if( rc==SQLITE4_OK ){
    btLogMergeSort(aPgno, &nMax, aSpace);
    *pnPgno = nMax;
    *paPgno = aPgno;
  }else{
    sqlite4_free(pLog->pLock->pEnv, aPgno);
    *paPgno = 0;
    *pnPgno = 0;
  }

  return rc;
}

/*
** Return the number of frames in the log file according to the current
** snapshot.
*/
int sqlite4BtLogSize(BtLog *pLog){
  return (int)pLog->snapshot.aLog[5] - (int)pLog->snapshot.aLog[4];
}

int sqlite4BtLogCheckpoint(BtLog *pLog){
  BtLock *pLock = pLog->pLock;
  int rc;

  /* Take the CHECKPOINTER lock. */
  rc = sqlite4BtLockCkpt(pLock);
  if( rc==SQLITE4_OK ){
    const int pgsz = sqlite4BtPagerPagesize((BtPager*)pLock);
    bt_env *pVfs = pLock->pVfs;
    bt_file *pFd = pLock->pFd;
    BtShm *pShm;                  /* Pointer to shared-memory region */
    u32 iFirst;                   /* First frame to checkpoint */
    u32 iLast;                    /* Last frame to checkpoint */
    u32 *aPgno = 0;               /* Array of page numbers to checkpoint */
    int nPgno;                    /* Number of entries in aPgno[] */
    int i;                        /* Used to loop through aPgno[] */
    u8 *aBuf;                     /* Buffer to load page data into */

    /* Allocate space to load log data into */
    aBuf = sqlite4_malloc(pLock->pEnv, pgsz);
    if( aBuf==0 ) rc = btErrorBkpt(SQLITE4_NOMEM);
    
    /* Figure out the set of page numbers stored in the part of the log 
    ** file being checkpointed. Remove any duplicates and sort them in 
    ** ascending order.  */
    if( rc==SQLITE4_OK ){
      pShm = btLogShm(pLog);
      iFirst = pShm->ckpt.iFirstRead;
      iLast = pLog->snapshot.aLog[5];
      rc = btLogGatherPgno(pLog, iFirst, iLast, &aPgno, &nPgno);
    }

    /* Copy data from the log file to the database file. */
    for(i=0; rc==SQLITE4_OK && i<nPgno; i++){
      u32 pgno = aPgno[i];
      rc = sqlite4BtLogRead(pLog, pgno, aBuf);
      if( rc==SQLITE4_OK ){
        i64 iOff = (i64)pgsz * (pgno-1);
        rc = pVfs->xWrite(pFd, iOff, aBuf, pgsz);
      }
    }

    /* Update the first field of the checkpoint-header. This tells readers
    ** that they need not consider anything that in the log before this
    ** point (since the data has already been copied into the database
    ** file).  */
    if( rc==SQLITE4_OK ){
      pShm->ckpt.iFirstRead = iLast+1;
      pVfs->xShmBarrier(pLog->pFd);
    }

    /* Write a new header into the log file. This tells any future recovery
    ** where it should start reading the log. Once this new header is synced
    ** to disk, the space cleared by this checkpoint operation can be 
    ** reused.  */
    if( rc==SQLITE4_OK ){
      int iSlot = ((pShm->ckpt.iWalHdr >> 2) + 1) % 2;
      i64 iOff;
      BtWalHdr hdr;
      BtFrameHdr fhdr;

      memset(&hdr, 0, sizeof(BtWalHdr));
      hdr.iMagic = BT_WAL_MAGIC;
      hdr.iVersion = BT_WAL_VERSION;
      hdr.iCnt = (((pShm->ckpt.iWalHdr & 0x03) + 1) % 3);
      hdr.nSector = pLog->snapshot.nSector;
      hdr.nPgsz = pgsz;
      hdr.iFirstFrame = iLast+1;

      iOff = btLogFrameOffset(pLog, pgsz, iLast);
      rc = btLogReadData(pLog, iOff, (u8*)&fhdr, sizeof(BtFrameHdr));
      if( rc==SQLITE4_OK ){
        hdr.iSalt1 = fhdr.aCksum[0];
        hdr.iSalt2 = fhdr.aCksum[1];
        rc = btLogWriteHeader(pLog, iSlot, &hdr);
      }
      if( rc==SQLITE4_OK ){
        pShm->ckpt.iWalHdr = (iSlot<<2) + hdr.iCnt;
      }
    }

    /* Update the second field of the checkpoint header. This tells future
    ** writers that it is now safe to recycle pages before this point
    ** (assuming all live readers are cleared).  */
    if( rc==SQLITE4_OK ){
      pShm->ckpt.iFirstRecover = iLast+1;
      pVfs->xShmBarrier(pLog->pFd);
    }

    /* Free the buffer and drop the checkpointer lock */
    sqlite4_free(pLock->pEnv, aBuf);
    sqlite4BtLockCkptUnlock(pLog->pLock);
  }
  return rc;
}


Changes to src/bt_main.c.

21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41

/*
** Values that make up the single byte flags field at the start of
** b-tree pages. 
*/
#define BT_PGFLAGS_INTERNAL 0x01  /* True for non-leaf nodes */

#ifndef MIN
# define MIN(a,b) (((a)<(b))?(a):(b))
#endif
#ifndef MAX
# define MAX(a,b) (((a)>(b))?(a):(b))
#endif

/* #define BT_STDERR_DEBUG 1 */

struct bt_db {
  sqlite4_env *pEnv;              /* SQLite environment */
  BtPager *pPager;                /* Underlying page-based database */
  bt_cursor *pAllCsr;             /* List of all open cursors */
};







<
<
<
<
<
<
<







21
22
23
24
25
26
27







28
29
30
31
32
33
34

/*
** Values that make up the single byte flags field at the start of
** b-tree pages. 
*/
#define BT_PGFLAGS_INTERNAL 0x01  /* True for non-leaf nodes */








/* #define BT_STDERR_DEBUG 1 */

struct bt_db {
  sqlite4_env *pEnv;              /* SQLite environment */
  BtPager *pPager;                /* Underlying page-based database */
  bt_cursor *pAllCsr;             /* List of all open cursors */
};

Changes to src/bt_pager.c.

15
16
17
18
19
20
21



22
23
24
25
26
27
28
..
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
...
192
193
194
195
196
197
198

199
200
201
202
203
204
205
...
347
348
349
350
351
352
353





354
355
356
357
358
359
360
...
373
374
375
376
377
378
379




380
381
382
383
384
385
386
...
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548

#include <string.h>
#include <assert.h>
#include <stdio.h>

/* By default pages are 1024 bytes in size. */
#define BT_DEFAULT_PGSZ 1024




typedef struct BtPageHash BtPageHash;
typedef struct BtDbhdr BtDbhdr;

/*
** Hash table for page references currently in memory. Manipulated using
** the following functions:
................................................................................
  u32 pgsz;                       /* Page size in bytes */
  u32 nPg;                        /* Number of pages in database */
  u32 cookie;                     /* User cookie value (SQL schema cookie) */
};

/*
** Pager object.




*/
struct BtPager {
  BtLock btl;                     /* Variables shared with bt_lock module */
  BtLog *pLog;                    /* Logging module */
  int iTransactionLevel;          /* Current transaction level (see bt.h) */
  char *zFile;                    /* Database file name */
  int nFile;                      /* Length of string zFile in bytes */
  BtPageHash hash;                /* Hash table */
  BtPage *pDirty;                 /* List of all dirty pages */



  BtDbhdr dbhdr;
  int nTotalRef;
};


/**************************************************************************
** Interface to BtPageHash object.
*/

................................................................................
  nByte = sizeof(BtPager) + nExtra;
  p = (BtPager*)sqlite4_malloc(pEnv, nByte);
  if( !p ) return btErrorBkpt(SQLITE4_NOMEM); 
  memset(p, 0, nByte);

  p->btl.pEnv = pEnv;
  p->btl.pVfs = sqlite4BtEnvDefault();

  *pp = p;
  return SQLITE4_OK;
}

static void btFreePage(BtPager *p, BtPage *pPg){
  if( pPg ){
    sqlite4_free(p->btl.pEnv, pPg->aData);
................................................................................
  assert( p->iTransactionLevel==0 );
  rc = sqlite4BtLogSnapshotClose(p->pLog);

  /* Purge the page cache. */
  assert( p->pDirty==0 );
  btPurgeCache(p);






  return rc;
}

/*
** Commit the current write transaction to disk.
*/
static int btCommitTransaction(BtPager *p){
................................................................................
    }
  }
  p->pDirty = 0;

  if( rc==SQLITE4_OK ){
    rc = p->btl.pVfs->xWrite(p->btl.pFd, 0, (void*)&p->dbhdr, sizeof(BtDbhdr));
  }





  return rc;
}

static int btLoadPageData(BtPager *p, BtPage *pPg){
  int rc;                         /* Return code */

................................................................................
  return p->iTransactionLevel;
}

/*
** Query for the database page size. Requires an open read transaction.
*/
int sqlite4BtPagerPagesize(BtPager *p){
  assert( p->iTransactionLevel>=1 && p->btl.pFd );
  return (int)p->dbhdr.pgsz;
}


/* 
** Query for the root page number. Requires an open read transaction.
*/
u32 sqlite4BtPagerRootpgno(BtPager *p){
  assert( p->iTransactionLevel>=1 && p->btl.pFd );
  return 2;







>
>
>







 







>
>
>
>









>
>
>

<







 







>







 







>
>
>
>
>







 







>
>
>
>







 







|


<







15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
..
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
...
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
...
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
...
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
...
550
551
552
553
554
555
556
557
558
559

560
561
562
563
564
565
566

#include <string.h>
#include <assert.h>
#include <stdio.h>

/* By default pages are 1024 bytes in size. */
#define BT_DEFAULT_PGSZ 1024

/* By default auto-checkpoint is 1000 */
#define BT_DEFAULT_AUTOCKPT 1000

typedef struct BtPageHash BtPageHash;
typedef struct BtDbhdr BtDbhdr;

/*
** Hash table for page references currently in memory. Manipulated using
** the following functions:
................................................................................
  u32 pgsz;                       /* Page size in bytes */
  u32 nPg;                        /* Number of pages in database */
  u32 cookie;                     /* User cookie value (SQL schema cookie) */
};

/*
** Pager object.
**
** nAutoCkpt:
**   If a transaction is committed and there are this many frames in the
**   log file, automatically run a checkpoint operation.
*/
struct BtPager {
  BtLock btl;                     /* Variables shared with bt_lock module */
  BtLog *pLog;                    /* Logging module */
  int iTransactionLevel;          /* Current transaction level (see bt.h) */
  char *zFile;                    /* Database file name */
  int nFile;                      /* Length of string zFile in bytes */
  BtPageHash hash;                /* Hash table */
  BtPage *pDirty;                 /* List of all dirty pages */
  int nTotalRef;                  /* Total number of outstanding page refs */
  int nAutoCkpt;                  /* Auto-checkpoint when log is this large */
  int bDoAutoCkpt;                /* Do auto-checkpoint after next unlock */
  BtDbhdr dbhdr;

};


/**************************************************************************
** Interface to BtPageHash object.
*/

................................................................................
  nByte = sizeof(BtPager) + nExtra;
  p = (BtPager*)sqlite4_malloc(pEnv, nByte);
  if( !p ) return btErrorBkpt(SQLITE4_NOMEM); 
  memset(p, 0, nByte);

  p->btl.pEnv = pEnv;
  p->btl.pVfs = sqlite4BtEnvDefault();
  p->nAutoCkpt = BT_DEFAULT_AUTOCKPT;
  *pp = p;
  return SQLITE4_OK;
}

static void btFreePage(BtPager *p, BtPage *pPg){
  if( pPg ){
    sqlite4_free(p->btl.pEnv, pPg->aData);
................................................................................
  assert( p->iTransactionLevel==0 );
  rc = sqlite4BtLogSnapshotClose(p->pLog);

  /* Purge the page cache. */
  assert( p->pDirty==0 );
  btPurgeCache(p);

  if( rc==SQLITE4_OK && p->bDoAutoCkpt ){
    sqlite4BtLogCheckpoint(p->pLog);
  }
  p->bDoAutoCkpt = 0;

  return rc;
}

/*
** Commit the current write transaction to disk.
*/
static int btCommitTransaction(BtPager *p){
................................................................................
    }
  }
  p->pDirty = 0;

  if( rc==SQLITE4_OK ){
    rc = p->btl.pVfs->xWrite(p->btl.pFd, 0, (void*)&p->dbhdr, sizeof(BtDbhdr));
  }

  if( p->nAutoCkpt && sqlite4BtLogSize(p->pLog)>=p->nAutoCkpt ){
    p->bDoAutoCkpt = 1;
  }

  return rc;
}

static int btLoadPageData(BtPager *p, BtPage *pPg){
  int rc;                         /* Return code */

................................................................................
  return p->iTransactionLevel;
}

/*
** Query for the database page size. Requires an open read transaction.
*/
int sqlite4BtPagerPagesize(BtPager *p){
  /* assert( p->iTransactionLevel>=1 && p->btl.pFd ); */
  return (int)p->dbhdr.pgsz;
}


/* 
** Query for the root page number. Requires an open read transaction.
*/
u32 sqlite4BtPagerRootpgno(BtPager *p){
  assert( p->iTransactionLevel>=1 && p->btl.pFd );
  return 2;

Changes to src/bt_unix.c.

250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
  }

  sqlite4_free(pSqlEnv, zCwd);
  *pzOut = zOut;
  return rc;
}

static int btPosixOsUnlink(bt_env *pEnv, const char *zFile){
  int prc = unlink(zFile);
  return prc ? btErrorBkpt(SQLITE4_IOERR) : SQLITE4_OK;
}

int btPosixOsLock(bt_file *pFile, int iLock, int eType){
  int rc = SQLITE4_OK;
  PosixFile *p = (PosixFile *)pFile;







|







250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
  }

  sqlite4_free(pSqlEnv, zCwd);
  *pzOut = zOut;
  return rc;
}

static int btPosixOsUnlink(sqlite4_env *pEnv, bt_env *pVfs, const char *zFile){
  int prc = unlink(zFile);
  return prc ? btErrorBkpt(SQLITE4_IOERR) : SQLITE4_OK;
}

int btPosixOsLock(bt_file *pFile, int iLock, int eType){
  int rc = SQLITE4_OK;
  PosixFile *p = (PosixFile *)pFile;

Changes to test/tester.tcl.

385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
# Delete all files associated with LSM database $file. That is:
#
#     ${file}
#     ${file}-log
#     ${file}-shm
#
proc db_delete {file} {
  forcedelete $file $file-shm $file-log
}

# Create a test database
#
proc reset_db {} {
  catch {db close}
  db_delete test.db







|







385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
# Delete all files associated with LSM database $file. That is:
#
#     ${file}
#     ${file}-log
#     ${file}-shm
#
proc db_delete {file} {
  forcedelete $file $file-shm $file-log $file-wal
}

# Create a test database
#
proc reset_db {} {
  catch {db close}
  db_delete test.db