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

Overview
Comment:Remove a layer of abstraction from the cursor object in lsm_sorted.c.
Downloads: Tarball | ZIP archive
Timelines: family | ancestors | descendants | both | trunk
Files: files | file ages | folders
SHA1: ff71b6f77869afddd4ed2eb531c0a2bd7faf4769
User & Date: dan 2012-10-02 05:19:45.146
Context
2012-10-02
12:05
Remove dead code from lsm_sorted.c. check-in: b84772a1aa user: dan tags: trunk
05:19
Remove a layer of abstraction from the cursor object in lsm_sorted.c. check-in: ff71b6f778 user: dan tags: trunk
2012-09-28
18:35
Allow deserialized snapshots to persist between transactions. check-in: fc4601f91f user: dan tags: trunk
Changes
Unified Diff Ignore Whitespace Patch
Changes to lsm-test/lsmtest1.c.
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
      testScanCompare(pDb2, pDb, 0, 0, 0,         pKey2, nKey2, &rc);
      testScanCompare(pDb2, pDb, 0, pKey1, nKey1, 0, 0,         &rc);
      testScanCompare(pDb2, pDb, 0, pKey1, nKey1, pKey2, nKey2, &rc);
      testScanCompare(pDb2, pDb, 1, 0, 0,         0, 0,         &rc);
      testScanCompare(pDb2, pDb, 1, 0, 0,         pKey2, nKey2, &rc);
      testScanCompare(pDb2, pDb, 1, pKey1, nKey1, 0, 0,         &rc);
      testScanCompare(pDb2, pDb, 1, pKey1, nKey1, pKey2, nKey2, &rc);

      testFree(pKey1);
    }
    tdb_close(pDb2);
  }

  /* Test some lookups. */
  for(j=0; rc==0 && j<nLookupTest; j++){







<







157
158
159
160
161
162
163

164
165
166
167
168
169
170
      testScanCompare(pDb2, pDb, 0, 0, 0,         pKey2, nKey2, &rc);
      testScanCompare(pDb2, pDb, 0, pKey1, nKey1, 0, 0,         &rc);
      testScanCompare(pDb2, pDb, 0, pKey1, nKey1, pKey2, nKey2, &rc);
      testScanCompare(pDb2, pDb, 1, 0, 0,         0, 0,         &rc);
      testScanCompare(pDb2, pDb, 1, 0, 0,         pKey2, nKey2, &rc);
      testScanCompare(pDb2, pDb, 1, pKey1, nKey1, 0, 0,         &rc);
      testScanCompare(pDb2, pDb, 1, pKey1, nKey1, pKey2, nKey2, &rc);

      testFree(pKey1);
    }
    tdb_close(pDb2);
  }

  /* Test some lookups. */
  for(j=0; rc==0 && j<nLookupTest; j++){
Changes to lsm-test/lsmtest_tdb3.c.
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
        }
#if 0
          if( nSleep ) printf("nLimit=%d nSleep=%d (worker)\n", nLimit, nSleep);
#endif

        rc = lsm_work(pWorker, p->lsm_work_flags, p->lsm_work_npage, &nWrite);
        if( nAuto==0 && nWrite && rc==LSM_OK ) mt_signal_worker(p->pDb, 1);
      }while( nWrite );
    }
    pthread_mutex_lock(&p->worker_mutex);

    if( rc!=LSM_OK && rc!=LSM_BUSY ){
      p->worker_rc = rc;
      break;
    }







|







958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
        }
#if 0
          if( nSleep ) printf("nLimit=%d nSleep=%d (worker)\n", nLimit, nSleep);
#endif

        rc = lsm_work(pWorker, p->lsm_work_flags, p->lsm_work_npage, &nWrite);
        if( nAuto==0 && nWrite && rc==LSM_OK ) mt_signal_worker(p->pDb, 1);
      }while( nWrite && p->pWorker );
    }
    pthread_mutex_lock(&p->worker_mutex);

    if( rc!=LSM_OK && rc!=LSM_BUSY ){
      p->worker_rc = rc;
      break;
    }
Changes to src/lsm_file.c.
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
** caller using lsmFsPageRelease().
*/
int lsmFsDbPageNext(Segment *pRun, Page *pPg, int eDir, Page **ppNext){
  FileSystem *pFS = pPg->pFS;
  int iPg = pPg->iPg;

  assert( eDir==1 || eDir==-1 );

  if( eDir<0 ){
    if( pRun && iPg==pRun->iFirst ){
      *ppNext = 0;
      return LSM_OK;
    }else if( fsIsFirst(pFS, iPg) ){
      iPg = lsmGetU32(&pPg->aData[pFS->nPagesize-4]);
    }else{







<







1037
1038
1039
1040
1041
1042
1043

1044
1045
1046
1047
1048
1049
1050
** caller using lsmFsPageRelease().
*/
int lsmFsDbPageNext(Segment *pRun, Page *pPg, int eDir, Page **ppNext){
  FileSystem *pFS = pPg->pFS;
  int iPg = pPg->iPg;

  assert( eDir==1 || eDir==-1 );

  if( eDir<0 ){
    if( pRun && iPg==pRun->iFirst ){
      *ppNext = 0;
      return LSM_OK;
    }else if( fsIsFirst(pFS, iPg) ){
      iPg = lsmGetU32(&pPg->aData[pFS->nPagesize-4]);
    }else{
Changes to src/lsm_shared.c.
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
    }

    /* Take a read-lock on the tree and snapshot just loaded. Then check
    ** that the shared-memory still contains the same values. If so, proceed.
    ** Otherwise, relinquish the read-lock and retry the whole procedure
    ** (starting with loading the in-memory tree header).  */
    if( rc==LSM_OK ){
      ShmHeader *pShm = pDb->pShmhdr;
      u32 iShmMax = pDb->treehdr.iUsedShmid;
      u32 iShmMin = pDb->treehdr.iNextShmid+1-(1<<10);
      rc = lsmReadlock(
          pDb, lsmCheckpointId(pDb->aSnapshot, 0), iShmMin, iShmMax
      );
      if( rc==LSM_OK ){
        if( lsmTreeLoadHeaderOk(pDb, iTreehdr)







<







692
693
694
695
696
697
698

699
700
701
702
703
704
705
    }

    /* Take a read-lock on the tree and snapshot just loaded. Then check
    ** that the shared-memory still contains the same values. If so, proceed.
    ** Otherwise, relinquish the read-lock and retry the whole procedure
    ** (starting with loading the in-memory tree header).  */
    if( rc==LSM_OK ){

      u32 iShmMax = pDb->treehdr.iUsedShmid;
      u32 iShmMin = pDb->treehdr.iNextShmid+1-(1<<10);
      rc = lsmReadlock(
          pDb, lsmCheckpointId(pDb->aSnapshot, 0), iShmMin, iShmMax
      );
      if( rc==LSM_OK ){
        if( lsmTreeLoadHeaderOk(pDb, iTreehdr)
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
  return rc;
}

/*
** Close the currently open read transaction.
*/
void lsmFinishReadTrans(lsm_db *pDb){
  Snapshot *pClient = pDb->pClient;

  /* Worker connections should not be closing read transactions. And
  ** read transactions should only be closed after all cursors and write
  ** transactions have been closed. Finally pClient should be non-NULL
  ** only iff pDb->iReader>=0.  */
  assert( pDb->pWorker==0 );
  assert( pDb->pCsr==0 && pDb->nTransOpen==0 );







<







742
743
744
745
746
747
748

749
750
751
752
753
754
755
  return rc;
}

/*
** Close the currently open read transaction.
*/
void lsmFinishReadTrans(lsm_db *pDb){


  /* Worker connections should not be closing read transactions. And
  ** read transactions should only be closed after all cursors and write
  ** transactions have been closed. Finally pClient should be non-NULL
  ** only iff pDb->iReader>=0.  */
  assert( pDb->pWorker==0 );
  assert( pDb->pCsr==0 && pDb->nTransOpen==0 );
Changes to src/lsm_sorted.c.
142
143
144
145
146
147
148

149
150
151
152
153
154
155
**
**   * To iterate and/or seek within a single Segment (the combination of a 
**     main run and an optional sorted run).
**
**   * To iterate through the separators array of a segment.
*/
struct SegmentPtr {

  Segment *pSeg;                /* Segment to access */

  /* Current page. See segmentPtrLoadPage(). */
  Page *pPg;                    /* Current page */
  u16 flags;                    /* Copy of page flags field */
  int nCell;                    /* Number of cells on pPg */
  int iPtr;                     /* Base cascade pointer */







>







142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
**
**   * To iterate and/or seek within a single Segment (the combination of a 
**     main run and an optional sorted run).
**
**   * To iterate through the separators array of a segment.
*/
struct SegmentPtr {
  Level *pLevel;                /* Level object segment is part of */
  Segment *pSeg;                /* Segment to access */

  /* Current page. See segmentPtrLoadPage(). */
  Page *pPg;                    /* Current page */
  u16 flags;                    /* Copy of page flags field */
  int nCell;                    /* Number of cells on pPg */
  int iPtr;                     /* Base cascade pointer */
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
**   lsmMCursorKey()
**   lsmMCursorValue()
**   lsmMCursorValid()
*/
struct MultiCursor {
  lsm_db *pDb;                    /* Connection that owns this cursor */
  MultiCursor *pNext;             /* Next cursor owned by connection pDb */

  int flags;                      /* Mask of CURSOR_XXX flags */
  int (*xCmp)(void *, int, void *, int);         /* Compare function */
  int eType;                      /* Cache of current key type */
  Blob key;                       /* Cache of current key (or NULL) */
  Blob val;                       /* Cache of current value */

  /* All the component cursors: */
  TreeCursor *apTreeCsr[2];       /* One or two tree cursors */

  int nSegCsr;                    /* Size of aSegCsr[] array */
  LevelCursor *aSegCsr;           /* Array of cursors open on sorted files */
  int nTree;

  int *aTree;

  BtreeCursor *pBtCsr;

  Snapshot *pSnap;



  /* Used by cursors flushing the in-memory tree only */
  int *pnOvfl;                    /* Number of free-list entries to store */
  void *pSystemVal;               /* Pointer to buffer to free */
};

#define CURSOR_DATA_TREE0     0   /* Current tree cursor */
#define CURSOR_DATA_TREE1     1   /* The "old" tree, if any */
#define CURSOR_DATA_SYSTEM    2
#define CURSOR_DATA_SEGMENT   3


/*
** CURSOR_IGNORE_DELETE
**   If set, this cursor will not visit SORTED_DELETE keys.
**
** CURSOR_NEW_SYSTEM
**   If set, then after all user data from the in-memory tree and any other







<

|





|
>


|
>
|
>
|

|
>
>










<







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
**   lsmMCursorKey()
**   lsmMCursorValue()
**   lsmMCursorValid()
*/
struct MultiCursor {
  lsm_db *pDb;                    /* Connection that owns this cursor */
  MultiCursor *pNext;             /* Next cursor owned by connection pDb */

  int flags;                      /* Mask of CURSOR_XXX flags */

  int eType;                      /* Cache of current key type */
  Blob key;                       /* Cache of current key (or NULL) */
  Blob val;                       /* Cache of current value */

  /* All the component cursors: */
  TreeCursor *apTreeCsr[2];       /* Up to two tree cursors */
#if 0
  int nSegCsr;                    /* Size of aSegCsr[] array */
  LevelCursor *aSegCsr;           /* Array of cursors open on sorted files */
#else
  SegmentPtr *aPtr;               /* Array of segment pointers */
  int nPtr;                       /* Size of array aPtr[] */
#endif
  BtreeCursor *pBtCsr;            /* b-tree cursor (db writes only) */

  /* Comparison results */
  int nTree;                      /* Size of aTree[] array */
  int *aTree;                     /* Array of comparison results */

  /* Used by cursors flushing the in-memory tree only */
  int *pnOvfl;                    /* Number of free-list entries to store */
  void *pSystemVal;               /* Pointer to buffer to free */
};

#define CURSOR_DATA_TREE0     0   /* Current tree cursor */
#define CURSOR_DATA_TREE1     1   /* The "old" tree, if any */
#define CURSOR_DATA_SYSTEM    2
#define CURSOR_DATA_SEGMENT   3


/*
** CURSOR_IGNORE_DELETE
**   If set, this cursor will not visit SORTED_DELETE keys.
**
** CURSOR_NEW_SYSTEM
**   If set, then after all user data from the in-memory tree and any other
294
295
296
297
298
299
300




301
302
303
304
305
306
307

308
309
310
311
312
313
314
**   If set, this cursor ignores system keys.
**
** CURSOR_NEXT_OK
**   Set if it is Ok to call lsm_csr_next().
**
** CURSOR_PREV_OK
**   Set if it is Ok to call lsm_csr_prev().




*/
#define CURSOR_IGNORE_DELETE    0x00000001
#define CURSOR_NEW_SYSTEM       0x00000002
#define CURSOR_AT_FREELIST      0x00000004
#define CURSOR_IGNORE_SYSTEM    0x00000010
#define CURSOR_NEXT_OK          0x00000020
#define CURSOR_PREV_OK          0x00000040


typedef struct MergeWorker MergeWorker;
typedef struct Hierarchy Hierarchy;

struct Hierarchy {
  Page **apHier;
  int nHier;







>
>
>
>







>







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
**   If set, this cursor ignores system keys.
**
** CURSOR_NEXT_OK
**   Set if it is Ok to call lsm_csr_next().
**
** CURSOR_PREV_OK
**   Set if it is Ok to call lsm_csr_prev().
**
** CURSOR_READ_SEPARATORS
**   Set if this cursor should visit the separator keys in segment 
**   aPtr[nPtr-1].
*/
#define CURSOR_IGNORE_DELETE    0x00000001
#define CURSOR_NEW_SYSTEM       0x00000002
#define CURSOR_AT_FREELIST      0x00000004
#define CURSOR_IGNORE_SYSTEM    0x00000010
#define CURSOR_NEXT_OK          0x00000020
#define CURSOR_PREV_OK          0x00000040
#define CURSOR_READ_SEPARATORS  0x00000080

typedef struct MergeWorker MergeWorker;
typedef struct Hierarchy Hierarchy;

struct Hierarchy {
  Page **apHier;
  int nHier;
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
  aCell++;
  aCell += lsmVarintGet32(aCell, &iRef);
  lsmVarintGet32(aCell, &iRef);
  assert( iRef>0 );
  return iRef;
}



static int pageGetBtreeKey(
  Page *pPg,
  int iKey, 
  int *piPtr, 
  int *piTopic, 
  void **ppKey,
  int *pnKey,
  Blob *pBlob
){
  u8 *aData;
  int nData;
  u8 *aCell;
  int eType;

  aData = fsPageData(pPg, &nData);
  assert( SEGMENT_BTREE_FLAG & pageGetFlags(aData, nData) );
  assert( iKey>=0 && iKey<pageGetNRec(aData, nData) );

  aCell = pageGetCell(aData, nData, iKey);
  eType = *aCell++;
  aCell += lsmVarintGet32(aCell, piPtr);

  if( eType==0 ){
    int rc;
    Pgno iRef;                  /* Page number of referenced page */
    Page *pRef;
    aCell += lsmVarintGet32(aCell, &iRef);
    rc = lsmFsDbPageGet(lsmPageFS(pPg), iRef, &pRef);
    if( rc!=LSM_OK ) return rc;
    pageGetKeyCopy(lsmPageEnv(pPg), pRef, 0, &eType, pBlob);
    lsmFsPageRelease(pRef);
    *ppKey = pBlob->pData;
    *pnKey = pBlob->nData;
  }else{
    aCell += lsmVarintGet32(aCell, pnKey);
    *ppKey = aCell;
  }
  if( piTopic ) *piTopic = rtTopic(eType);

  return LSM_OK;
}








>
>




















|





|







|







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
  aCell++;
  aCell += lsmVarintGet32(aCell, &iRef);
  lsmVarintGet32(aCell, &iRef);
  assert( iRef>0 );
  return iRef;
}

#define GETVARINT32(a, i) (((i)=((u8*)(a))[0])<=240?1:lsmVarintGet32((a), &(i)))

static int pageGetBtreeKey(
  Page *pPg,
  int iKey, 
  int *piPtr, 
  int *piTopic, 
  void **ppKey,
  int *pnKey,
  Blob *pBlob
){
  u8 *aData;
  int nData;
  u8 *aCell;
  int eType;

  aData = fsPageData(pPg, &nData);
  assert( SEGMENT_BTREE_FLAG & pageGetFlags(aData, nData) );
  assert( iKey>=0 && iKey<pageGetNRec(aData, nData) );

  aCell = pageGetCell(aData, nData, iKey);
  eType = *aCell++;
  aCell += GETVARINT32(aCell, *piPtr);

  if( eType==0 ){
    int rc;
    Pgno iRef;                  /* Page number of referenced page */
    Page *pRef;
    aCell += GETVARINT32(aCell, iRef);
    rc = lsmFsDbPageGet(lsmPageFS(pPg), iRef, &pRef);
    if( rc!=LSM_OK ) return rc;
    pageGetKeyCopy(lsmPageEnv(pPg), pRef, 0, &eType, pBlob);
    lsmFsPageRelease(pRef);
    *ppKey = pBlob->pData;
    *pnKey = pBlob->nData;
  }else{
    aCell += GETVARINT32(aCell, *pnKey);
    *ppKey = aCell;
  }
  if( piTopic ) *piTopic = rtTopic(eType);

  return LSM_OK;
}

1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059

1060
1061
1062
1063
1064
1065
1066

    assert( iNew<pPtr->nCell );
    pPtr->iCell = iNew;
    aData = fsPageData(pPtr->pPg, &nPgsz);
    iOff = lsmGetU16(&aData[SEGMENT_CELLPTR_OFFSET(nPgsz, pPtr->iCell)]);
    pPtr->eType = aData[iOff];
    iOff++;
    iOff += lsmVarintGet32(&aData[iOff], &pPtr->iPgPtr);
    iOff += lsmVarintGet32(&aData[iOff], &pPtr->nKey);
    if( rtIsWrite(pPtr->eType) ){
      iOff += lsmVarintGet32(&aData[iOff], &pPtr->nVal);
    }


    rc = segmentPtrReadData(
        pPtr, iOff, pPtr->nKey, &pPtr->pKey, &pPtr->blob1
    );
    if( rc==LSM_OK && rtIsWrite(pPtr->eType) ){
      rc = segmentPtrReadData(
          pPtr, iOff+pPtr->nKey, pPtr->nVal, &pPtr->pVal, &pPtr->blob2







|
|

|

>







1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078

    assert( iNew<pPtr->nCell );
    pPtr->iCell = iNew;
    aData = fsPageData(pPtr->pPg, &nPgsz);
    iOff = lsmGetU16(&aData[SEGMENT_CELLPTR_OFFSET(nPgsz, pPtr->iCell)]);
    pPtr->eType = aData[iOff];
    iOff++;
    iOff += GETVARINT32(&aData[iOff], pPtr->iPgPtr);
    iOff += GETVARINT32(&aData[iOff], pPtr->nKey);
    if( rtIsWrite(pPtr->eType) ){
      iOff += GETVARINT32(&aData[iOff], pPtr->nVal);
    }
    assert( pPtr->nKey>=0 );

    rc = segmentPtrReadData(
        pPtr, iOff, pPtr->nKey, &pPtr->pKey, &pPtr->blob1
    );
    if( rc==LSM_OK && rtIsWrite(pPtr->eType) ){
      rc = segmentPtrReadData(
          pPtr, iOff+pPtr->nKey, pPtr->nVal, &pPtr->pVal, &pPtr->blob2
1203
1204
1205
1206
1207
1208
1209
1210





1211
1212
1213
1214
1215
1216
1217
1218
1219
** be managed by the caller.
*/
static void segmentCursorClose(lsm_env *pEnv, LevelCursor *pCsr){
  segmentCursorReset(pCsr);
  lsmFree(pEnv, pCsr->aPtr);
  memset(pCsr, 0, sizeof(LevelCursor));
}






static int segmentPtrAdvance(
  LevelCursor *pCsr, 
  SegmentPtr *pPtr,
  int bReverse
){
  int eDir = (bReverse ? -1 : 1);
  do {
    int rc;
    int iCell;                    /* Number of new cell in page */








>
>
>
>
>

|







1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
** be managed by the caller.
*/
static void segmentCursorClose(lsm_env *pEnv, LevelCursor *pCsr){
  segmentCursorReset(pCsr);
  lsmFree(pEnv, pCsr->aPtr);
  memset(pCsr, 0, sizeof(LevelCursor));
}

static int segmentPtrIgnoreSeparators(MultiCursor *pCsr, SegmentPtr *pPtr){
  return (pCsr->flags & CURSOR_READ_SEPARATORS)==0
      || (pPtr!=&pCsr->aPtr[pCsr->nPtr-1]);
}

static int segmentPtrAdvance(
  MultiCursor *pCsr, 
  SegmentPtr *pPtr,
  int bReverse
){
  int eDir = (bReverse ? -1 : 1);
  do {
    int rc;
    int iCell;                    /* Number of new cell in page */
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
           && (pPtr->nCell==0 || (pPtr->flags & SEGMENT_BTREE_FLAG) ) 
      );
      if( rc!=LSM_OK ) return rc;
      iCell = bReverse ? (pPtr->nCell-1) : 0;
    }
    rc = segmentPtrLoadCell(pPtr, iCell);
    if( rc!=LSM_OK ) return rc;
  }while( pCsr && pPtr->pPg && (


        (pCsr->bIgnoreSeparators && rtIsSeparator(pPtr->eType))
     || (pCsr->bIgnoreSystem && rtTopic(pPtr->eType)!=0)
  ));


  return LSM_OK;
}

static void segmentPtrEndPage(
  FileSystem *pFS, 
  SegmentPtr *pPtr, 
  int bLast, 
  int *pRc
){
  if( *pRc==LSM_OK ){
    Page *pNew = 0;
    Pgno iPg = (bLast ? pPtr->pSeg->iLast : pPtr->pSeg->iFirst);
    *pRc = lsmFsDbPageGet(pFS, iPg, &pNew);
    segmentPtrSetPage(pPtr, pNew);
  }
}


























/*
** Try to move the segment pointer passed as the second argument so that it
** points at either the first (bLast==0) or last (bLast==1) cell in the valid
** region of the segment defined by pPtr->iFirst and pPtr->iLast.
**
** Return LSM_OK if successful or an lsm error code if something goes
** wrong (IO error, OOM etc.).
*/
static void segmentPtrEnd(
  LevelCursor *pCsr,              /* Cursor that owns this segment-pointer */
  SegmentPtr *pPtr,               /* Segment pointer to reposition */
  int bLast,                      /* True for last, false for first */
  int *pRc                        /* IN/OUT error code */
){

  if( *pRc==LSM_OK ){
    int rc = LSM_OK;

    segmentPtrEndPage(pCsr->pFS, pPtr, bLast, &rc);
    while( rc==LSM_OK 
        && pPtr->pPg 
        && (pPtr->nCell==0 || (pPtr->flags & SEGMENT_BTREE_FLAG))







|
>
>
|
<
|
<


















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














>







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
           && (pPtr->nCell==0 || (pPtr->flags & SEGMENT_BTREE_FLAG) ) 
      );
      if( rc!=LSM_OK ) return rc;
      iCell = bReverse ? (pPtr->nCell-1) : 0;
    }
    rc = segmentPtrLoadCell(pPtr, iCell);
    if( rc!=LSM_OK ) return rc;
  }while( pCsr 
       && pPtr->pPg 
       && segmentPtrIgnoreSeparators(pCsr, pPtr)
       && rtIsSeparator(pPtr->eType)

  );


  return LSM_OK;
}

static void segmentPtrEndPage(
  FileSystem *pFS, 
  SegmentPtr *pPtr, 
  int bLast, 
  int *pRc
){
  if( *pRc==LSM_OK ){
    Page *pNew = 0;
    Pgno iPg = (bLast ? pPtr->pSeg->iLast : pPtr->pSeg->iFirst);
    *pRc = lsmFsDbPageGet(pFS, iPg, &pNew);
    segmentPtrSetPage(pPtr, pNew);
  }
}


static int segmentPtrEnd2(MultiCursor *pCsr, SegmentPtr *pPtr, int bLast){
  int rc = LSM_OK;
  FileSystem *pFS = pCsr->pDb->pFS;
  int bIgnore;

  segmentPtrEndPage(pFS, pPtr, bLast, &rc);
  while( rc==LSM_OK && pPtr->pPg 
      && (pPtr->nCell==0 || (pPtr->flags & SEGMENT_BTREE_FLAG))
  ){
    rc = segmentPtrNextPage(pPtr, (bLast ? -1 : 1));
  }
  
  if( rc==LSM_OK && pPtr->pPg ){
    rc = segmentPtrLoadCell(pPtr, bLast ? (pPtr->nCell-1) : 0);
  }

  bIgnore = segmentPtrIgnoreSeparators(pCsr, pPtr);
  if( rc==LSM_OK && pPtr->pPg && bIgnore && rtIsSeparator(pPtr->eType) ){
    rc = segmentPtrAdvance(pCsr, pPtr, bLast);
  }

  return rc;
}

/*
** Try to move the segment pointer passed as the second argument so that it
** points at either the first (bLast==0) or last (bLast==1) cell in the valid
** region of the segment defined by pPtr->iFirst and pPtr->iLast.
**
** Return LSM_OK if successful or an lsm error code if something goes
** wrong (IO error, OOM etc.).
*/
static void segmentPtrEnd(
  LevelCursor *pCsr,              /* Cursor that owns this segment-pointer */
  SegmentPtr *pPtr,               /* Segment pointer to reposition */
  int bLast,                      /* True for last, false for first */
  int *pRc                        /* IN/OUT error code */
){
#if 0
  if( *pRc==LSM_OK ){
    int rc = LSM_OK;

    segmentPtrEndPage(pCsr->pFS, pPtr, bLast, &rc);
    while( rc==LSM_OK 
        && pPtr->pPg 
        && (pPtr->nCell==0 || (pPtr->flags & SEGMENT_BTREE_FLAG))
1289
1290
1291
1292
1293
1294
1295

1296
1297
1298
1299
1300
1301
1302
          (pCsr->bIgnoreSeparators && rtIsSeparator(pPtr->eType))
       || (pCsr->bIgnoreSystem && rtTopic(pPtr->eType)!=0)
    )){
      rc = segmentPtrAdvance(pCsr, pPtr, bLast);
    }
    *pRc = rc;
  }

}

static void segmentPtrKey(SegmentPtr *pPtr, void **ppKey, int *pnKey){
  assert( pPtr->pPg );
  *ppKey = pPtr->pKey;
  *pnKey = pPtr->nKey;
}







>







1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
          (pCsr->bIgnoreSeparators && rtIsSeparator(pPtr->eType))
       || (pCsr->bIgnoreSystem && rtTopic(pPtr->eType)!=0)
    )){
      rc = segmentPtrAdvance(pCsr, pPtr, bLast);
    }
    *pRc = rc;
  }
#endif
}

static void segmentPtrKey(SegmentPtr *pPtr, void **ppKey, int *pnKey){
  assert( pPtr->pPg );
  *ppKey = pPtr->pKey;
  *pnKey = pPtr->nKey;
}
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
  sortedBlobFree(&blob);
  return 1;
}
#endif

#ifndef NDEBUG
static int assertSeekResult(
  LevelCursor *pCsr,
  SegmentPtr *pPtr,
  int iTopic,
  void *pKey,
  int nKey,
  int eSeek
){
  if( pPtr->pPg ){
    int res;
    res = sortedKeyCompare(pCsr->xCmp, iTopic, pKey, nKey,
        rtTopic(pPtr->eType), pPtr->pKey, pPtr->nKey
    );

    if( eSeek==LSM_SEEK_EQ ) return (res==0);
    if( eSeek==LSM_SEEK_LE ) return (res>=0);
    if( eSeek==LSM_SEEK_GE ) return (res<=0);
  }

  return 1;
}
#endif

int segmentPtrSeek(
  LevelCursor *pCsr,              /* Cursor context */
  SegmentPtr *pPtr,               /* Pointer to seek */
  void *pKey, int nKey,           /* Key to seek to */
  int eSeek,                      /* Search bias - see above */
  int *piPtr                      /* OUT: FC pointer */
){

  int res;                        /* Result of comparison operation */
  int rc = LSM_OK;
  int iMin;
  int iMax;
  int iPtrOut = 0;

  const int iTopic = 0;







|








|













|





>







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
  sortedBlobFree(&blob);
  return 1;
}
#endif

#ifndef NDEBUG
static int assertSeekResult(
  MultiCursor *pCsr,
  SegmentPtr *pPtr,
  int iTopic,
  void *pKey,
  int nKey,
  int eSeek
){
  if( pPtr->pPg ){
    int res;
    res = sortedKeyCompare(pCsr->pDb->xCmp, iTopic, pKey, nKey,
        rtTopic(pPtr->eType), pPtr->pKey, pPtr->nKey
    );

    if( eSeek==LSM_SEEK_EQ ) return (res==0);
    if( eSeek==LSM_SEEK_LE ) return (res>=0);
    if( eSeek==LSM_SEEK_GE ) return (res<=0);
  }

  return 1;
}
#endif

int segmentPtrSeek(
  MultiCursor *pCsr,              /* Cursor context */
  SegmentPtr *pPtr,               /* Pointer to seek */
  void *pKey, int nKey,           /* Key to seek to */
  int eSeek,                      /* Search bias - see above */
  int *piPtr                      /* OUT: FC pointer */
){
  int (*xCmp)(void *, int, void *, int) = pCsr->pDb->xCmp;
  int res;                        /* Result of comparison operation */
  int rc = LSM_OK;
  int iMin;
  int iMax;
  int iPtrOut = 0;

  const int iTopic = 0;
1456
1457
1458
1459
1460
1461
1462
1463
1464

1465
1466
1467
1468
1469
1470
1471
    pLastKey = pageGetKey(
        pPtr->pPg, pPtr->nCell-1, &iLastTopic, &nLastKey, &pPtr->blob1
    );

    /* If the loaded key is >= than (pKey/nKey), break out of the loop.
    ** If (pKey/nKey) is present in this array, it must be on the current 
    ** page.  */
    res = iLastTopic - iTopic;
    if( res==0 ) res = pCsr->xCmp(pLastKey, nLastKey, pKey, nKey);

    if( res>=0 ) break;

    /* Advance to the next page that contains at least one key. */
    pNext = pPtr->pPg;
    lsmFsPageRef(pNext);
    while( 1 ){
      Page *pLoad;







|
|
>







1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
    pLastKey = pageGetKey(
        pPtr->pPg, pPtr->nCell-1, &iLastTopic, &nLastKey, &pPtr->blob1
    );

    /* If the loaded key is >= than (pKey/nKey), break out of the loop.
    ** If (pKey/nKey) is present in this array, it must be on the current 
    ** page.  */
    res = sortedKeyCompare(
        xCmp, iLastTopic, pLastKey, nLastKey, iTopic, pKey, nKey
    );
    if( res>=0 ) break;

    /* Advance to the next page that contains at least one key. */
    pNext = pPtr->pPg;
    lsmFsPageRef(pNext);
    while( 1 ){
      Page *pLoad;
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537

      rc = segmentPtrLoadCell(pPtr, iTry);
      if( rc!=LSM_OK ) break;

      segmentPtrKey(pPtr, &pKeyT, &nKeyT);
      iTopicT = rtTopic(pPtr->eType);

      res = sortedKeyCompare(
          pCsr->xCmp, iTopicT, pKeyT, nKeyT, iTopic, pKey, nKey
      );
      if( res<=0 ){
        iPtrOut = pPtr->iPtr + pPtr->iPgPtr;
      }

      if( res==0 || iMin==iMax ){
        break;
      }else if( res>0 ){







<
|
<







1567
1568
1569
1570
1571
1572
1573

1574

1575
1576
1577
1578
1579
1580
1581

      rc = segmentPtrLoadCell(pPtr, iTry);
      if( rc!=LSM_OK ) break;

      segmentPtrKey(pPtr, &pKeyT, &nKeyT);
      iTopicT = rtTopic(pPtr->eType);


      res = sortedKeyCompare(xCmp, iTopicT, pKeyT, nKeyT, iTopic, pKey, nKey);

      if( res<=0 ){
        iPtrOut = pPtr->iPtr + pPtr->iPgPtr;
      }

      if( res==0 || iMin==iMax ){
        break;
      }else if( res>0 ){
1558
1559
1560
1561
1562
1563
1564


1565

1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
          case LSM_SEEK_GE:
            if( res<0 ) rc = segmentPtrAdvance(pCsr, pPtr, 0);
            break;
        }
      }
    }



    if( rc==LSM_OK && pPtr->pPg && (

          (pCsr->bIgnoreSeparators && rtIsSeparator(pPtr->eType))
       || (pCsr->bIgnoreSystem && rtTopic(pPtr->eType)!=0)
    )){
      rc = segmentPtrAdvance(pCsr, pPtr, eSeek==LSM_SEEK_LE);
    }
  }

  assert( rc!=LSM_OK || assertSeekResult(pCsr,pPtr,iTopic,pKey,nKey,eSeek) );
  *piPtr = iPtrOut;
  return rc;







>
>
|
>
|
<
|







1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613

1614
1615
1616
1617
1618
1619
1620
1621
          case LSM_SEEK_GE:
            if( res<0 ) rc = segmentPtrAdvance(pCsr, pPtr, 0);
            break;
        }
      }
    }

    /* If the cursor seek has found a separator key, and this cursor is
    ** supposed to ignore separators keys, advance to the next entry.  */
    if( rc==LSM_OK && pPtr->pPg 
     && segmentPtrIgnoreSeparators(pCsr, pPtr) 
     && rtIsSeparator(pPtr->eType)

    ){
      rc = segmentPtrAdvance(pCsr, pPtr, eSeek==LSM_SEEK_LE);
    }
  }

  assert( rc!=LSM_OK || assertSeekResult(pCsr,pPtr,iTopic,pKey,nKey,eSeek) );
  *piPtr = iPtrOut;
  return rc;
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
    if( res ) iBest = i;
  }

  pCsr->iCurrentPtr = iBest;
}

static int seekInBtree(
  LevelCursor *pCsr,
  Segment *pSeg,
  void *pKey, int nKey,           /* Key to seek to */
  Pgno *aPg,                      /* OUT: Page numbers */
  Page **ppPg                     /* OUT: Leaf (sorted-run) page reference */
){
  int i = 0;
  int rc;
  int iPg;
  Page *pPg = 0;
  Blob blob = {0, 0, 0};
  int iTopic = 0;                 /* TODO: Fix me */

  iPg = pSeg->iRoot;
  do {
    Pgno *piFirst = 0;
    if( aPg ){
      aPg[i++] = iPg;
      piFirst = &aPg[i];
    }

    rc = lsmFsDbPageGet(pCsr->pFS, iPg, &pPg);
    assert( rc==LSM_OK || pPg==0 );
    if( rc==LSM_OK ){
      u8 *aData;                  /* Buffer containing page data */
      int nData;                  /* Size of aData[] in bytes */
      int iMin;
      int iMax;
      int nRec;







|
|



















|







1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
    if( res ) iBest = i;
  }

  pCsr->iCurrentPtr = iBest;
}

static int seekInBtree(
  MultiCursor *pCsr,              /* Multi-cursor object */
  Segment *pSeg,                  /* Seek within this segment */
  void *pKey, int nKey,           /* Key to seek to */
  Pgno *aPg,                      /* OUT: Page numbers */
  Page **ppPg                     /* OUT: Leaf (sorted-run) page reference */
){
  int i = 0;
  int rc;
  int iPg;
  Page *pPg = 0;
  Blob blob = {0, 0, 0};
  int iTopic = 0;                 /* TODO: Fix me */

  iPg = pSeg->iRoot;
  do {
    Pgno *piFirst = 0;
    if( aPg ){
      aPg[i++] = iPg;
      piFirst = &aPg[i];
    }

    rc = lsmFsDbPageGet(pCsr->pDb->pFS, iPg, &pPg);
    assert( rc==LSM_OK || pPg==0 );
    if( rc==LSM_OK ){
      u8 *aData;                  /* Buffer containing page data */
      int nData;                  /* Size of aData[] in bytes */
      int iMin;
      int iMax;
      int nRec;
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
        if( rc!=LSM_OK ) break;
        if( piFirst && pKeyT==blob.pData ){
          *piFirst = pageGetBtreeRef(pPg, iTry);
          piFirst = 0;
          i++;
        }

        res = iTopic - iTopicT;
        if( res==0 ) res = pCsr->xCmp(pKey, nKey, pKeyT, nKeyT);

        if( res<0 ){
          iPg = iPtr;
          iMax = iTry-1;
        }else{
          iMin = iTry+1;
        }
      }







|
|
|







1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
        if( rc!=LSM_OK ) break;
        if( piFirst && pKeyT==blob.pData ){
          *piFirst = pageGetBtreeRef(pPg, iTry);
          piFirst = 0;
          i++;
        }

        res = sortedKeyCompare(
            pCsr->pDb->xCmp, iTopic, pKey, nKey, iTopicT, pKeyT, nKeyT
        );
        if( res<0 ){
          iPg = iPtr;
          iMax = iTry-1;
        }else{
          iMin = iTry+1;
        }
      }
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742

1743
1744
1745
1746
1747

1748
1749
1750
1751
1752
1753

1754

1755
1756
1757


1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781



1782


1783
1784










1785







1786






1787




























































































1788
1789
1790
1791

1792














1793
1794




1795
1796
1797


1798
1799
1800


1801

1802



1803

1804


1805




1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
  }else{
    lsmFsPageRelease(pPg);
  }
  return rc;
}

static int seekInSegment(
  LevelCursor *pCsr, 
  SegmentPtr *pPtr,
  void *pKey, int nKey,
  int iPg,                        /* Page to search */
  int eSeek,                      /* Search bias - see above */
  int *piPtr                      /* OUT: FC pointer */
){
  int iPtr = iPg;
  int rc = LSM_OK;

  if( pPtr->pSeg->iRoot ){
    Page *pPg;
    assert( pPtr->pSeg->iRoot!=0 );
    rc = seekInBtree(pCsr, pPtr->pSeg, pKey, nKey, 0, &pPg);
    if( rc==LSM_OK ) segmentPtrSetPage(pPtr, pPg);
  }else{
    if( iPtr==0 ){
      iPtr = pPtr->pSeg->iFirst;
    }
    if( rc==LSM_OK ){
      rc = segmentPtrLoadPage(pCsr->pFS, pPtr, iPtr);
    }
  }

  if( rc==LSM_OK ){
    rc = segmentPtrSeek(pCsr, pPtr, pKey, nKey, eSeek, piPtr);
  }
  return rc;
}

/*
** Seek the cursor.
*/
static int segmentCursorSeek(
  LevelCursor *pCsr,              /* Sorted cursor object to seek */
  void *pKey, int nKey,           /* Key to seek to */
  int iPg,                        /* Page to search */
  int eSeek,                      /* Search bias - see above */
  int *piPtr                      /* OUT: FC pointer */

){
  int rc = LSM_OK;

  int iOut = 0;                   /* Pointer to return to caller */
  int res = -1;                   /* Result of xCmp(pKey, split) */


  if( pCsr->nPtr>1 ){ 
    Level *pLevel = pCsr->pLevel;
    assert( pLevel );
    res = 0 - pLevel->iSplitTopic;
    if( res==0 ){

      res = pCsr->xCmp(pKey, nKey, pLevel->pSplitKey, pLevel->nSplitKey);

    }
  }



  if( res<0 ){
    int iPtr = 0;
    if( pCsr->nPtr==1 ) iPtr = iPg;

    rc = seekInSegment(pCsr, &pCsr->aPtr[0], pKey, nKey, iPtr, eSeek, &iOut);
    if( rc==LSM_OK 
     && pCsr->nPtr>1
     && eSeek==LSM_SEEK_GE 
     && pCsr->aPtr[0].pPg==0 
    ){
      res = 0;
    }
  }
  
  if( res>=0 ){
    int iPtr = iPg;
    int i;
    for(i=1; rc==LSM_OK && i<pCsr->nPtr; i++){
      iOut = 0;
      rc = seekInSegment(pCsr, &pCsr->aPtr[i], pKey, nKey, iPtr, eSeek, &iOut);
      iPtr = iOut;
    }

    if( eSeek==LSM_SEEK_LE ){



      segmentPtrEnd(pCsr, &pCsr->aPtr[0], 1, &rc);


    }
  }


















  segmentCursorSetCurrent(pCsr, eSeek==LSM_SEEK_LE);






  *piPtr = iOut;




























































































  return rc;
}

/*

** Advance the cursor forwards (bReverse==0) or backwards (bReverse!=0).














*/
static int segmentCursorAdvance(LevelCursor *pCsr, int bReverse){




  int rc;
  int iCurrent = pCsr->iCurrentPtr;
  SegmentPtr *pCurrent;



  pCurrent = &pCsr->aPtr[iCurrent];
  rc = segmentPtrAdvance(pCsr, pCurrent, bReverse);


  

  /* If bReverse is true and the segment pointer just moved is on the rhs



  ** of the level, check if the key it now points to is smaller than the

  ** split-key. If so, set the segment to EOF.


  **




  ** The danger here is that if the current key is smaller than the 
  ** split-key then there may have existed a delete key on a page that
  ** has already been gobbled.
  */
  if( rc==LSM_OK && bReverse && iCurrent>0 && pCurrent->pPg ){
    Level *p = pCsr->pLevel;
    int res = sortedKeyCompare(pCsr->xCmp,
        rtTopic(pCurrent->eType), pCurrent->pKey, pCurrent->nKey,
        p->iSplitTopic, p->pSplitKey, p->nSplitKey
    );
    if( res<0 ){
      segmentPtrReset(pCurrent);
    }
  }

  segmentCursorSetCurrent(pCsr, bReverse);

  if( segmentCursorValid(pCsr)==0 
   && bReverse==0 
   && iCurrent==0 
   && pCsr->nPtr>1
   && pCsr->aPtr[1].pPg==0 
  ){
    Level *p = pCsr->pLevel;







|



















|










|

|
|
|
|

|
>

|
<


>

<
|
|
|
|
>
|
>
|
|
|
>
>


|

|
|
<
<
<
<





|

|

|



|
>
>
>
|
>
>
|
|
>
>
>
>
>
>
>
>
>
>

>
>
>
>
>
>
>
|
>
>
>
>
>
>
|
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>




>
|
>
>
>
>
>
>
>
>
>
>
>
>
>
>

|
>
>
>
>

<
|
>
>

<
|
>
>
|
>
|
>
>
>
|
>
|
>
>
|
>
>
>
>
|
<
<
<
|
<
|
|
|


|



<
|







1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791

1792
1793
1794
1795

1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813




1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980

1981
1982
1983
1984

1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004



2005

2006
2007
2008
2009
2010
2011
2012
2013
2014

2015
2016
2017
2018
2019
2020
2021
2022
  }else{
    lsmFsPageRelease(pPg);
  }
  return rc;
}

static int seekInSegment(
  MultiCursor *pCsr, 
  SegmentPtr *pPtr,
  void *pKey, int nKey,
  int iPg,                        /* Page to search */
  int eSeek,                      /* Search bias - see above */
  int *piPtr                      /* OUT: FC pointer */
){
  int iPtr = iPg;
  int rc = LSM_OK;

  if( pPtr->pSeg->iRoot ){
    Page *pPg;
    assert( pPtr->pSeg->iRoot!=0 );
    rc = seekInBtree(pCsr, pPtr->pSeg, pKey, nKey, 0, &pPg);
    if( rc==LSM_OK ) segmentPtrSetPage(pPtr, pPg);
  }else{
    if( iPtr==0 ){
      iPtr = pPtr->pSeg->iFirst;
    }
    if( rc==LSM_OK ){
      rc = segmentPtrLoadPage(pCsr->pDb->pFS, pPtr, iPtr);
    }
  }

  if( rc==LSM_OK ){
    rc = segmentPtrSeek(pCsr, pPtr, pKey, nKey, eSeek, piPtr);
  }
  return rc;
}

/*
** Seek each segment pointer in the array of (pLvl->nRight+1) at aPtr[].
*/
static int seekInLevel(
  MultiCursor *pCsr,              /* Sorted cursor object to seek */
  Level *pLvl,                    /* Level to seek within */
  SegmentPtr *aPtr,               /* Pointer to array of (nRight+1) SPs */
  int eSeek,                      /* Search bias - see above */
  void *pKey, int nKey,           /* Key to search for */
  Pgno *piPgno                    /* IN/OUT: fraction cascade pointer (or 0) */
){
  int rc = LSM_OK;                /* Return code */

  int iOut = 0;                   /* Pointer to return to caller */
  int res = -1;                   /* Result of xCmp(pKey, split) */
  int nRhs = pLvl->nRight;        /* Number of right-hand-side segments */


  /* If this is a composite level (one currently undergoing an incremental
  ** merge), figure out if the search key is larger or smaller than the
  ** levels split-key.  */
  if( nRhs ){
    res = sortedKeyCompare(pCsr->pDb->xCmp, 0, pKey, nKey, 
        pLvl->iSplitTopic, pLvl->pSplitKey, pLvl->nSplitKey
    );
  }

  /* If (res<0), then key pKey/nKey is smaller than the split-key (or this
  ** is not a composite level and there is no split-key). Search the 
  ** left-hand-side of the level in this case.  */
  if( res<0 ){
    int iPtr = 0;
    if( nRhs==0 ) iPtr = *piPgno;

    rc = seekInSegment(pCsr, &aPtr[0], pKey, nKey, iPtr, eSeek, &iOut);
    if( rc==LSM_OK && nRhs>0 && eSeek==LSM_SEEK_GE && aPtr[0].pPg==0 ){




      res = 0;
    }
  }
  
  if( res>=0 ){
    int iPtr = *piPgno;
    int i;
    for(i=1; rc==LSM_OK && i<=nRhs; i++){
      iOut = 0;
      rc = seekInSegment(pCsr, &aPtr[i], pKey, nKey, iPtr, eSeek, &iOut);
      iPtr = iOut;
    }

    if( rc==LSM_OK && eSeek==LSM_SEEK_LE ){
      rc = segmentPtrEnd2(pCsr, &aPtr[0], 1);
    }
  }

  *piPgno = iOut;
  return rc;
}

static void multiCursorGetKey(
  MultiCursor *pCsr, 
  int iKey,
  int *peType,                    /* OUT: Key type (SORTED_WRITE etc.) */
  void **ppKey,                   /* OUT: Pointer to buffer containing key */
  int *pnKey                      /* OUT: Size of *ppKey in bytes */
){
  int nKey = 0;
  void *pKey = 0;
  int eType = 0;

  switch( iKey ){
    case CURSOR_DATA_TREE0:
    case CURSOR_DATA_TREE1: {
      TreeCursor *pTreeCsr = pCsr->apTreeCsr[iKey-CURSOR_DATA_TREE0];
      if( lsmTreeCursorValid(pTreeCsr) ){
        int nVal;
        void *pVal;

        lsmTreeCursorKey(pTreeCsr, &pKey, &nKey);
        lsmTreeCursorValue(pTreeCsr, &pVal, &nVal);
        eType = (nVal<0) ? SORTED_DELETE : SORTED_WRITE;
      }
      break;
    }

    case CURSOR_DATA_SYSTEM:
      if( pCsr->flags & CURSOR_AT_FREELIST ){
        pKey = (void *)"FREELIST";
        nKey = 8;
        eType = SORTED_SYSTEM_WRITE;
      }
      break;

    default: {
      int iPtr = iKey - CURSOR_DATA_SEGMENT;
      assert( iPtr>=0 );
      if( iPtr==pCsr->nPtr ){
        if( pCsr->pBtCsr ){
          pKey = pCsr->pBtCsr->pKey;
          nKey = pCsr->pBtCsr->nKey;
          eType = pCsr->pBtCsr->eType;
        }
      }else if( iPtr<pCsr->nPtr ){
        SegmentPtr *pPtr = &pCsr->aPtr[iPtr];
        if( pPtr->pPg ){
          pKey = pPtr->pKey;
          nKey = pPtr->nKey;
          eType = pPtr->eType;
        }
      }
      break;
    }
  }

  if( peType ) *peType = eType;
  if( pnKey ) *pnKey = nKey;
  if( ppKey ) *ppKey = pKey;
}


static void multiCursorDoCompare(MultiCursor *pCsr, int iOut, int bReverse){
  int i1;
  int i2;
  int iRes;
  void *pKey1; int nKey1; int eType1;
  void *pKey2; int nKey2; int eType2;

  int mul = (bReverse ? -1 : 1);

  assert( pCsr->aTree && iOut<pCsr->nTree );
  if( iOut>=(pCsr->nTree/2) ){
    i1 = (iOut - pCsr->nTree/2) * 2;
    i2 = i1 + 1;
  }else{
    i1 = pCsr->aTree[iOut*2];
    i2 = pCsr->aTree[iOut*2+1];
  }

  multiCursorGetKey(pCsr, i1, &eType1, &pKey1, &nKey1);
  multiCursorGetKey(pCsr, i2, &eType2, &pKey2, &nKey2);

  if( pKey1==0 ){
    iRes = i2;
  }else if( pKey2==0 ){
    iRes = i1;
  }else{
    int res;

    res = (rtTopic(eType1) - rtTopic(eType2));
    if( res==0 ){
      res = pCsr->pDb->xCmp(pKey1, nKey1, pKey2, nKey2);
    }
    res = res * mul;

    if( res==0 ){
      iRes = (rtIsSeparator(eType1) ? i2 : i1);
    }else if( res<0 ){
      iRes = i1;
    }else{
      iRes = i2;
    }
  }

  pCsr->aTree[iOut] = iRes;
}

static int sortedRhsFirst(MultiCursor *pCsr, Level *pLvl, SegmentPtr *pPtr){
  int rc;
  rc = segmentPtrEnd2(pCsr, pPtr, 0);
  while( pPtr->pPg && rc==LSM_OK ){
    int res = sortedKeyCompare(pCsr->pDb->xCmp,
        pLvl->iSplitTopic, pLvl->pSplitKey, pLvl->nSplitKey,
        rtTopic(pPtr->eType), pPtr->pKey, pPtr->nKey
    );
    if( res<=0 ) break;
    rc = segmentPtrAdvance(pCsr, pPtr, 0);
  }
  return rc;
}

/*
** This function advances segment pointer iPtr belonging to multi-cursor
** pCsr forward (bReverse==0) or backward (bReverse!=0).
**
** If the segment pointer points to a segment that is part of a composite
** level, then the following special cases are handled.
**
** TODO: First point is not currently done.
**
**   * If iPtr is the lhs of a composite level, and the cursor is being
**     advanced forwards, and segment iPtr is at EOF, move all pointers
**     that correspond to rhs segments of the same level to the first
**     key in their respective data.
**
**   * If iPtr is on the rhs of a composite level, and the cursor is being
**     reversed, and the new key is smaller than the split-key, set the
**     segment pointer to point to EOF.
*/
static int segmentCursorAdvance(
  MultiCursor *pCsr, 
  int iPtr,
  int bReverse
){
  int rc;

  SegmentPtr *pPtr = &pCsr->aPtr[iPtr];
  Level *pLvl = pPtr->pLevel;
  int bComposite;


  rc = segmentPtrAdvance(pCsr, pPtr, bReverse);
  if( rc!=LSM_OK ) return rc;
  bComposite = (pLvl->nRight>0 && pCsr->nPtr>pLvl->nRight);

  if( bComposite && pPtr->pSeg==&pLvl->lhs       /* lhs of composite level */
   && bReverse==0                                /* csr advanced forwards */
   && pPtr->pPg==0                               /* segment at EOF */
  ){
    int i;

    for(i=0; rc==LSM_OK && i<pLvl->nRight; i++){
      SegmentPtr *pPtr = &pCsr->aPtr[iPtr+1+i];
      rc = sortedRhsFirst(pCsr, pLvl, &pCsr->aPtr[iPtr+1+i]);
    }

    for(i=pCsr->nTree-1; i>0; i--){
      multiCursorDoCompare(pCsr, i, 0);
    }
  }




  else if( pPtr->pPg && bComposite && bReverse && pPtr->pSeg!=&pLvl->lhs ){

    int res = sortedKeyCompare(pCsr->pDb->xCmp,
        rtTopic(pPtr->eType), pPtr->pKey, pPtr->nKey,
        pLvl->iSplitTopic, pLvl->pSplitKey, pLvl->nSplitKey
    );
    if( res<0 ){
      segmentPtrReset(pPtr);
    }
  }


#if 0
  if( segmentCursorValid(pCsr)==0 
   && bReverse==0 
   && iCurrent==0 
   && pCsr->nPtr>1
   && pCsr->aPtr[1].pPg==0 
  ){
    Level *p = pCsr->pLevel;
1850
1851
1852
1853
1854
1855
1856

1857
1858
1859
1860
1861
1862
1863
        if( res>=0 ) break;
        segmentPtrAdvance(pCsr, pPtr, 0);
      }

    }
    segmentCursorSetCurrent(pCsr, bReverse);
  }


  return rc;
}

/*
** Move the cursor to point to either the first element (if bLast==0), or
** the last element (if bLast!=0) in the sorted file.







>







2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
        if( res>=0 ) break;
        segmentPtrAdvance(pCsr, pPtr, 0);
      }

    }
    segmentCursorSetCurrent(pCsr, bReverse);
  }
#endif

  return rc;
}

/*
** Move the cursor to point to either the first element (if bLast==0), or
** the last element (if bLast!=0) in the sorted file.
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
  int i;
  lsm_env *pEnv = pCsr->pDb->pEnv;

  /* Close the tree cursor, if any. */
  lsmTreeCursorDestroy(pCsr->apTreeCsr[0]);
  lsmTreeCursorDestroy(pCsr->apTreeCsr[1]);

  /* Close the sorted file cursors */
  for(i=0; i<pCsr->nSegCsr; i++){
    segmentCursorClose(pEnv, &pCsr->aSegCsr[i]);
  }

  /* And the b-tree cursor, if any */
  btreeCursorFree(pCsr->pBtCsr);

  /* Free allocations */
  lsmFree(pEnv, pCsr->aSegCsr);
  lsmFree(pEnv, pCsr->aTree);
  lsmFree(pEnv, pCsr->pSystemVal);

  /* Zero fields */
  pCsr->nSegCsr = 0;
  pCsr->aSegCsr = 0;
  pCsr->nTree = 0;
  pCsr->aTree = 0;
  pCsr->pSystemVal = 0;
  pCsr->pSnap = 0;
  pCsr->apTreeCsr[0] = 0;
  pCsr->apTreeCsr[1] = 0;
  pCsr->pBtCsr = 0;
}

void lsmMCursorClose(MultiCursor *pCsr){
  if( pCsr ){







|
|
|






|




|
|



<







2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109

2110
2111
2112
2113
2114
2115
2116
  int i;
  lsm_env *pEnv = pCsr->pDb->pEnv;

  /* Close the tree cursor, if any. */
  lsmTreeCursorDestroy(pCsr->apTreeCsr[0]);
  lsmTreeCursorDestroy(pCsr->apTreeCsr[1]);

  /* Reset the segment pointers */
  for(i=0; i<pCsr->nPtr; i++){
    segmentPtrReset(&pCsr->aPtr[i]);
  }

  /* And the b-tree cursor, if any */
  btreeCursorFree(pCsr->pBtCsr);

  /* Free allocations */
  lsmFree(pEnv, pCsr->aPtr);
  lsmFree(pEnv, pCsr->aTree);
  lsmFree(pEnv, pCsr->pSystemVal);

  /* Zero fields */
  pCsr->nPtr = 0;
  pCsr->aPtr = 0;
  pCsr->nTree = 0;
  pCsr->aTree = 0;
  pCsr->pSystemVal = 0;

  pCsr->apTreeCsr[0] = 0;
  pCsr->apTreeCsr[1] = 0;
  pCsr->pBtCsr = 0;
}

void lsmMCursorClose(MultiCursor *pCsr){
  if( pCsr ){
1968
1969
1970
1971
1972
1973
1974

1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008



2009
2010
2011
2012
2013
2014
2015
2016
2017
2018




2019
2020
2021
2022
2023
2024
2025
2026
** when merging existing runs).
*/
int multiCursorAddLevel(
  MultiCursor *pCsr,              /* Multi-cursor to add segment to */ 
  Level *pLevel,                  /* Level to add to multi-cursor merge */
  int eMode                       /* A MULTICURSOR_ADDLEVEL_*** constant */
){

  int rc = LSM_OK;

  assert( eMode==MULTICURSOR_ADDLEVEL_ALL
       || eMode==MULTICURSOR_ADDLEVEL_RHS
       || eMode==MULTICURSOR_ADDLEVEL_LHS_SEP
  );

  if( eMode==MULTICURSOR_ADDLEVEL_LHS_SEP ){
    assert( pLevel->lhs.iRoot );
    assert( pCsr->pBtCsr==0 );
    rc = btreeCursorNew(pCsr->pDb, &pLevel->lhs, &pCsr->pBtCsr);
    assert( (rc==LSM_OK)==(pCsr->pBtCsr!=0) );
  }else{
    int i;
    int nAdd = (eMode==MULTICURSOR_ADDLEVEL_RHS ? pLevel->nRight : 1);

    for(i=0; i<nAdd; i++){
      LevelCursor *pNew;
      lsm_db *pDb = pCsr->pDb;

      /* Grow the pCsr->aSegCsr array if required */
      if( 0==(pCsr->nSegCsr % 16) ){
        int nByte;
        LevelCursor *aNew;
        nByte = sizeof(LevelCursor) * (pCsr->nSegCsr+16);
        aNew = (LevelCursor *)lsmRealloc(pDb->pEnv, pCsr->aSegCsr, nByte);
        if( aNew==0 ) return LSM_NOMEM_BKPT;
        memset(&aNew[pCsr->nSegCsr], 0, sizeof(LevelCursor)*16);
        pCsr->aSegCsr = aNew;
      }
      pNew = &pCsr->aSegCsr[pCsr->nSegCsr];

      switch( eMode ){
        case MULTICURSOR_ADDLEVEL_ALL:



          rc = levelCursorInit(pDb, pLevel, pCsr->xCmp, pNew);
          break;

        case MULTICURSOR_ADDLEVEL_RHS:
          rc = levelCursorInitRun(pDb, &pLevel->aRhs[i], pCsr->xCmp, pNew);
          break;
      }
      if( pCsr->flags & CURSOR_IGNORE_SYSTEM ){
        pNew->bIgnoreSystem = 1;
      }




      if( rc==LSM_OK ) pCsr->nSegCsr++;
    }
  }

  return rc;
}

#define TREE_NONE 0







>










|



|
|
<
|
<

|
<
<
<
|
|
|
|
|
|
<
|
<
|
>
>
>
|
<
|
<
|
<
|
<
<
|
>
>
>
>
|







2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184

2185

2186
2187



2188
2189
2190
2191
2192
2193

2194

2195
2196
2197
2198
2199

2200

2201

2202


2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
** when merging existing runs).
*/
int multiCursorAddLevel(
  MultiCursor *pCsr,              /* Multi-cursor to add segment to */ 
  Level *pLevel,                  /* Level to add to multi-cursor merge */
  int eMode                       /* A MULTICURSOR_ADDLEVEL_*** constant */
){
  lsm_db *pDb = pCsr->pDb;
  int rc = LSM_OK;

  assert( eMode==MULTICURSOR_ADDLEVEL_ALL
       || eMode==MULTICURSOR_ADDLEVEL_RHS
       || eMode==MULTICURSOR_ADDLEVEL_LHS_SEP
  );

  if( eMode==MULTICURSOR_ADDLEVEL_LHS_SEP ){
    assert( pLevel->lhs.iRoot );
    assert( pCsr->pBtCsr==0 );
    rc = btreeCursorNew(pDb, &pLevel->lhs, &pCsr->pBtCsr);
    assert( (rc==LSM_OK)==(pCsr->pBtCsr!=0) );
  }else{
    int i;
    int nAdd = pLevel->nRight + (eMode==MULTICURSOR_ADDLEVEL_ALL);
    int nByte;

    SegmentPtr *aNew;


    /* Grow the pCsr->aPtr array */



    nByte = sizeof(SegmentPtr) * (pCsr->nPtr + nAdd);
    aNew = (SegmentPtr *)lsmRealloc(pDb->pEnv, pCsr->aPtr, nByte);
    if( aNew==0 ) return LSM_NOMEM_BKPT;
    memset(&aNew[pCsr->nPtr], 0, nAdd * sizeof(SegmentPtr));
    pCsr->aPtr = aNew;


    /* If this is ALL, add the left-hand-side segment */

    if( eMode==MULTICURSOR_ADDLEVEL_ALL ){
      aNew[pCsr->nPtr].pSeg = &pLevel->lhs;
      aNew[pCsr->nPtr].pLevel = pLevel;
      if( pLevel->nRight && pLevel->pSplitKey==0 ){
        lsmSortedSplitkey(pDb, pLevel, &rc);

      }

      pCsr->nPtr++;

    }



    /* Add the right-hand-side segments */
    for(i=0; i<pLevel->nRight; i++){
      aNew[pCsr->nPtr].pSeg = &pLevel->aRhs[i];
      aNew[pCsr->nPtr].pLevel = pLevel;
      pCsr->nPtr++;
    }
  }

  return rc;
}

#define TREE_NONE 0
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
  if( pCsr==0 ){
    pCsr = (MultiCursor *)lsmMallocZeroRc(pDb->pEnv, sizeof(MultiCursor), &rc);
    if( pCsr ){
      pCsr->pNext = pDb->pCsr;
      pDb->pCsr = pCsr;
      if( bUserOnly ) pCsr->flags |= CURSOR_IGNORE_SYSTEM;
      pCsr->pDb = pDb;
      pCsr->pSnap = pSnap;
      pCsr->xCmp = pDb->xCmp;
    }
  }

  /* Add a tree cursor on the 'old' tree, if required. */
  if( rc==LSM_OK 
   && eTree!=TREE_NONE 
   && lsmTreeHasOld(pDb)







<
<







2234
2235
2236
2237
2238
2239
2240


2241
2242
2243
2244
2245
2246
2247
  if( pCsr==0 ){
    pCsr = (MultiCursor *)lsmMallocZeroRc(pDb->pEnv, sizeof(MultiCursor), &rc);
    if( pCsr ){
      pCsr->pNext = pDb->pCsr;
      pDb->pCsr = pCsr;
      if( bUserOnly ) pCsr->flags |= CURSOR_IGNORE_SYSTEM;
      pCsr->pDb = pDb;


    }
  }

  /* Add a tree cursor on the 'old' tree, if required. */
  if( rc==LSM_OK 
   && eTree!=TREE_NONE 
   && lsmTreeHasOld(pDb)
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
  *ppCsr = pCsr;
  return rc;
}

void lsmSortedRemap(lsm_db *pDb){
  MultiCursor *pCsr;
  for(pCsr=pDb->pCsr; pCsr; pCsr=pCsr->pNext){
    int i;
    if( pCsr->pBtCsr ){
      btreeCursorLoadKey(pCsr->pBtCsr);
    }
    for(i=0; i<pCsr->nSegCsr; i++){
      int iPtr;
      LevelCursor *p = &pCsr->aSegCsr[i];
      for(iPtr=0; iPtr<p->nPtr; iPtr++){
        segmentPtrLoadCell(&p->aPtr[iPtr], p->aPtr[iPtr].iCell);
      }
    }
  }
}

static void multiCursorReadSeparators(MultiCursor *pCsr){
  if( pCsr->nSegCsr>0 ){
    pCsr->aSegCsr[pCsr->nSegCsr-1].bIgnoreSeparators = 0;
  }
}

/*
** Have this cursor skip over SORTED_DELETE entries.
*/
static void multiCursorIgnoreDelete(MultiCursor *pCsr){







|



<
<
<
|
|
<





|
|







2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272



2273
2274

2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
  *ppCsr = pCsr;
  return rc;
}

void lsmSortedRemap(lsm_db *pDb){
  MultiCursor *pCsr;
  for(pCsr=pDb->pCsr; pCsr; pCsr=pCsr->pNext){
    int iPtr;
    if( pCsr->pBtCsr ){
      btreeCursorLoadKey(pCsr->pBtCsr);
    }



    for(iPtr=0; iPtr<pCsr->nPtr; iPtr++){
      segmentPtrLoadCell(&pCsr->aPtr[iPtr], pCsr->aPtr[iPtr].iCell);

    }
  }
}

static void multiCursorReadSeparators(MultiCursor *pCsr){
  if( pCsr->nPtr>0 ){
    pCsr->flags |= CURSOR_READ_SEPARATORS;
  }
}

/*
** Have this cursor skip over SORTED_DELETE entries.
*/
static void multiCursorIgnoreDelete(MultiCursor *pCsr){
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238





2239

2240
2241
2242
2243
2244
2245
2246



2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259

2260

2261
2262
2263
2264

2265
2266
2267




2268
2269
2270
2271
2272
2273
2274
  rc = multiCursorAllocate(pDb, 0, &pCsr);

  assert( (rc==LSM_OK)==(pCsr!=0) );
  *ppCsr = pCsr;
  return rc;
}

static void multiCursorGetKey(
  MultiCursor *pCsr, 
  int iKey,
  int *peType,                    /* OUT: Key type (SORTED_WRITE etc.) */
  void **ppKey,                   /* OUT: Pointer to buffer containing key */
  int *pnKey                      /* OUT: Size of *ppKey in bytes */
){
  int nKey = 0;
  void *pKey = 0;
  int eType = 0;

  switch( iKey ){
    case CURSOR_DATA_TREE0:
    case CURSOR_DATA_TREE1: {
      TreeCursor *pTreeCsr = pCsr->apTreeCsr[iKey-CURSOR_DATA_TREE0];
      if( lsmTreeCursorValid(pTreeCsr) ){
        int nVal;
        void *pVal;

        lsmTreeCursorKey(pTreeCsr, &pKey, &nKey);
        lsmTreeCursorValue(pTreeCsr, &pVal, &nVal);
        eType = (nVal<0) ? SORTED_DELETE : SORTED_WRITE;
      }
      break;
    }

    case CURSOR_DATA_SYSTEM:
      if( pCsr->flags & CURSOR_AT_FREELIST ){
        pKey = (void *)"FREELIST";
        nKey = 8;
        eType = SORTED_SYSTEM_WRITE;
      }
      break;

    default: {
      int iSeg = iKey - CURSOR_DATA_SEGMENT;
      if( iSeg==pCsr->nSegCsr && pCsr->pBtCsr ){
        pKey = pCsr->pBtCsr->pKey;
        nKey = pCsr->pBtCsr->nKey;
        eType = pCsr->pBtCsr->eType;
      }if( iSeg<pCsr->nSegCsr && segmentCursorValid(&pCsr->aSegCsr[iSeg]) ){
        segmentCursorKey(&pCsr->aSegCsr[iSeg], &pKey, &nKey);
        segmentCursorType(&pCsr->aSegCsr[iSeg], &eType);
      }
      break;
    }
  }

  if( peType ) *peType = eType;
  if( pnKey ) *pnKey = nKey;
  if( ppKey ) *ppKey = pKey;
}

static int multiCursorGetVal(
  MultiCursor *pCsr, 
  int iVal, 
  void **ppVal, 
  int *pnVal
){
  int rc = LSM_OK;





  if( iVal==CURSOR_DATA_TREE0 || iVal==CURSOR_DATA_TREE1 ){

    TreeCursor *pTreeCsr = pCsr->apTreeCsr[iVal-CURSOR_DATA_TREE0];
    if( lsmTreeCursorValid(pTreeCsr) ){
      lsmTreeCursorValue(pTreeCsr, ppVal, pnVal);
    }else{
      *ppVal = 0;
      *pnVal = 0;
    }



  }else if( iVal==CURSOR_DATA_SYSTEM ){
    if( pCsr->flags & CURSOR_AT_FREELIST ){
      void *aVal;
      int nVal;

      assert( pCsr->pSystemVal==0 );
      rc = lsmCheckpointOverflow(pCsr->pDb, &aVal, &nVal, pCsr->pnOvfl);
      *ppVal = pCsr->pSystemVal = aVal;
      *pnVal = nVal;
    }else{
      *ppVal = 0;
      *pnVal = 0;
    }

  }else if( iVal-CURSOR_DATA_SEGMENT<pCsr->nSegCsr 

         && segmentCursorValid(&pCsr->aSegCsr[iVal-CURSOR_DATA_SEGMENT]) 
  ){
    segmentCursorValue(&pCsr->aSegCsr[iVal-CURSOR_DATA_SEGMENT], ppVal, pnVal);
  }else{

    *ppVal = 0;
    *pnVal = 0;
  }




  assert( rc==LSM_OK || (*ppVal==0 && *pnVal==0) );
  return rc;
}

int lsmSortedLoadFreelist(
  lsm_db *pDb,                    /* Database handle (must be worker) */
  void **ppVal,                   /* OUT: Blob containing LSM free-list */







<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<







>
>
>
>
>
|
>
|
|
|
|
|
|
|
>
>
>
|
|
|
<
|
|
<
|
<
<
<
<
|
>
|
>
|
|
|
<
>
|
|
|
>
>
>
>







2355
2356
2357
2358
2359
2360
2361





















































2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388

2389
2390

2391




2392
2393
2394
2395
2396
2397
2398

2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
  rc = multiCursorAllocate(pDb, 0, &pCsr);

  assert( (rc==LSM_OK)==(pCsr!=0) );
  *ppCsr = pCsr;
  return rc;
}






















































static int multiCursorGetVal(
  MultiCursor *pCsr, 
  int iVal, 
  void **ppVal, 
  int *pnVal
){
  int rc = LSM_OK;

  *ppVal = 0;
  *pnVal = 0;

  switch( iVal ){
    case CURSOR_DATA_TREE0:
    case CURSOR_DATA_TREE1: {
      TreeCursor *pTreeCsr = pCsr->apTreeCsr[iVal-CURSOR_DATA_TREE0];
      if( lsmTreeCursorValid(pTreeCsr) ){
        lsmTreeCursorValue(pTreeCsr, ppVal, pnVal);
      }else{
        *ppVal = 0;
        *pnVal = 0;
      }
      break;
    }

    case CURSOR_DATA_SYSTEM:
      if( pCsr->flags & CURSOR_AT_FREELIST ){
        void *aVal;

        rc = lsmCheckpointOverflow(pCsr->pDb, &aVal, pnVal, pCsr->pnOvfl);
        assert( pCsr->pSystemVal==0 );

        *ppVal = pCsr->pSystemVal = aVal;




      }
      break;

    default: {
      int iPtr = iVal-CURSOR_DATA_SEGMENT;
      if( iPtr<pCsr->nPtr ){
        SegmentPtr *pPtr = &pCsr->aPtr[iPtr];

        if( pPtr->pPg ){
          *ppVal = pPtr->pVal;
          *pnVal = pPtr->nVal;
        }
      }
    }
  }

  assert( rc==LSM_OK || (*ppVal==0 && *pnVal==0) );
  return rc;
}

int lsmSortedLoadFreelist(
  lsm_db *pDb,                    /* Database handle (must be worker) */
  void **ppVal,                   /* OUT: Blob containing LSM free-list */
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369

    lsmMCursorClose(pCsr);
  }

  return rc;
}

static void multiCursorDoCompare(MultiCursor *pCsr, int iOut, int bReverse){
  int i1;
  int i2;
  int iRes;
  void *pKey1; int nKey1; int eType1;
  void *pKey2; int nKey2; int eType2;

  int mul = (bReverse ? -1 : 1);

  assert( pCsr->aTree && iOut<pCsr->nTree );
  if( iOut>=(pCsr->nTree/2) ){
    i1 = (iOut - pCsr->nTree/2) * 2;
    i2 = i1 + 1;
  }else{
    i1 = pCsr->aTree[iOut*2];
    i2 = pCsr->aTree[iOut*2+1];
  }

  multiCursorGetKey(pCsr, i1, &eType1, &pKey1, &nKey1);
  multiCursorGetKey(pCsr, i2, &eType2, &pKey2, &nKey2);

  if( pKey1==0 ){
    iRes = i2;
  }else if( pKey2==0 ){
    iRes = i1;
  }else{
    int res;

    res = (rtTopic(eType1) - rtTopic(eType2));
    if( res==0 ){
      res = pCsr->xCmp(pKey1, nKey1, pKey2, nKey2);
    }
    res = res * mul;

    if( res==0 ){
      iRes = (rtIsSeparator(eType1) ? i2 : i1);
    }else if( res<0 ){
      iRes = i1;
    }else{
      iRes = i2;
    }
  }

  pCsr->aTree[iOut] = iRes;
}

static int multiCursorAllocTree(MultiCursor *pCsr){
  int rc = LSM_OK;
  if( pCsr->aTree==0 ){
    int nByte;                    /* Bytes of space to allocate */
    int bBtree;                   /* True if b-tree cursor is present */

    bBtree = (pCsr->pBtCsr!=0);
    pCsr->nTree = 2;
    while( pCsr->nTree<(CURSOR_DATA_SEGMENT+pCsr->nSegCsr+bBtree) ){
      pCsr->nTree = pCsr->nTree*2;
    }

    nByte = sizeof(int)*pCsr->nTree*2;
    pCsr->aTree = (int *)lsmMallocZeroRc(pCsr->pDb->pEnv, nByte, &rc);
  }
  return rc;







<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<




|

|

|







2440
2441
2442
2443
2444
2445
2446














































2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462

    lsmMCursorClose(pCsr);
  }

  return rc;
}















































static int multiCursorAllocTree(MultiCursor *pCsr){
  int rc = LSM_OK;
  if( pCsr->aTree==0 ){
    int nByte;                    /* Bytes of space to allocate */
    int nMin;                     /* Total number of cursors being merged */

    nMin = CURSOR_DATA_SEGMENT + pCsr->nPtr + (pCsr->pBtCsr!=0);
    pCsr->nTree = 2;
    while( pCsr->nTree<nMin ){
      pCsr->nTree = pCsr->nTree*2;
    }

    nByte = sizeof(int)*pCsr->nTree*2;
    pCsr->aTree = (int *)lsmMallocZeroRc(pCsr->pDb->pEnv, nByte, &rc);
  }
  return rc;
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397

2398


2399













2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
  if( pCsr->apTreeCsr[0] ){
    rc = lsmTreeCursorEnd(pCsr->apTreeCsr[0], bLast);
  }
  if( rc==LSM_OK && pCsr->apTreeCsr[1] ){
    rc = lsmTreeCursorEnd(pCsr->apTreeCsr[1], bLast);
  }


  if( pCsr->flags & CURSOR_NEW_SYSTEM ){
    assert( bLast==0 );
    pCsr->flags |= CURSOR_AT_FREELIST;
  }

  for(i=0; rc==LSM_OK && i<pCsr->nSegCsr; i++){


    rc = segmentCursorEnd(&pCsr->aSegCsr[i], bLast);













  }

  if( rc==LSM_OK && pCsr->pBtCsr ){
    assert( bLast==0 );
    rc = btreeCursorFirst(pCsr->pBtCsr);
  }

  if( rc==LSM_OK ){
    rc = multiCursorAllocTree(pCsr);
  }

  if( rc==LSM_OK ){
    for(i=pCsr->nTree-1; i>0; i--){
      multiCursorDoCompare(pCsr, i, bLast);
    }
    pCsr->flags |= (bLast ? CURSOR_PREV_OK : CURSOR_NEXT_OK);
  }

  multiCursorCacheKey(pCsr, &rc);
  if( rc==LSM_OK 
   && (pCsr->flags & CURSOR_IGNORE_DELETE) 
   && rtIsDelete(pCsr->eType)
  ){
    if( bLast ){
      rc = lsmMCursorPrev(pCsr);
    }else{
      rc = lsmMCursorNext(pCsr);
    }
  }








<




>
|
>
>
|
>
>
>
>
>
>
>
>
>
>
>
>
>










<








|
|
|
|







2479
2480
2481
2482
2483
2484
2485

2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517

2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
  if( pCsr->apTreeCsr[0] ){
    rc = lsmTreeCursorEnd(pCsr->apTreeCsr[0], bLast);
  }
  if( rc==LSM_OK && pCsr->apTreeCsr[1] ){
    rc = lsmTreeCursorEnd(pCsr->apTreeCsr[1], bLast);
  }


  if( pCsr->flags & CURSOR_NEW_SYSTEM ){
    assert( bLast==0 );
    pCsr->flags |= CURSOR_AT_FREELIST;
  }

  for(i=0; rc==LSM_OK && i<pCsr->nPtr; i++){
    SegmentPtr *pPtr = &pCsr->aPtr[i];
    Level *pLvl = pPtr->pLevel;

    rc = segmentPtrEnd2(pCsr, pPtr, bLast);
    if( rc==LSM_OK && bLast==0 && pLvl->nRight && pPtr->pSeg==&pLvl->lhs ){
      int iRhs;
      for(iRhs=1+i; rc==LSM_OK && iRhs<1+i+pLvl->nRight; iRhs++){
        SegmentPtr *pRhs = &pCsr->aPtr[iRhs];
        if( pPtr->pPg==0 ){
          rc = sortedRhsFirst(pCsr, pLvl, pRhs);
        }else{
          segmentPtrReset(pRhs);
        }
      }
      i += pLvl->nRight;
    }
  }

  if( rc==LSM_OK && pCsr->pBtCsr ){
    assert( bLast==0 );
    rc = btreeCursorFirst(pCsr->pBtCsr);
  }

  if( rc==LSM_OK ){
    rc = multiCursorAllocTree(pCsr);
  }

  if( rc==LSM_OK ){
    for(i=pCsr->nTree-1; i>0; i--){
      multiCursorDoCompare(pCsr, i, bLast);
    }
    pCsr->flags |= (bLast ? CURSOR_PREV_OK : CURSOR_NEXT_OK);
  }

  multiCursorCacheKey(pCsr, &rc);
  if( rc==LSM_OK && (
        ((pCsr->flags & CURSOR_IGNORE_DELETE) && rtIsDelete(pCsr->eType))
     || ((pCsr->flags & CURSOR_IGNORE_SYSTEM) && rtTopic(pCsr->eType))
  )){
    if( bLast ){
      rc = lsmMCursorPrev(pCsr);
    }else{
      rc = lsmMCursorNext(pCsr);
    }
  }

2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
  return multiCursorEnd(pCsr, 1);
}

lsm_db *lsmMCursorDb(MultiCursor *pCsr){
  return pCsr->pDb;
}



void lsmMCursorReset(MultiCursor *pCsr){
  int i;
  lsmTreeCursorReset(pCsr->apTreeCsr[0]);
  lsmTreeCursorReset(pCsr->apTreeCsr[1]);
  for(i=0; i<pCsr->nSegCsr; i++){
    segmentCursorReset(&pCsr->aSegCsr[i]);
  }
  pCsr->key.nData = 0;
}

static void treeCursorSeek(
  TreeCursor *pTreeCsr, 
  void *pKey, int nKey, 







<
<




|
|







2587
2588
2589
2590
2591
2592
2593


2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
  return multiCursorEnd(pCsr, 1);
}

lsm_db *lsmMCursorDb(MultiCursor *pCsr){
  return pCsr->pDb;
}



void lsmMCursorReset(MultiCursor *pCsr){
  int i;
  lsmTreeCursorReset(pCsr->apTreeCsr[0]);
  lsmTreeCursorReset(pCsr->apTreeCsr[1]);
  for(i=0; i<pCsr->nPtr; i++){
    segmentPtrReset(&pCsr->aPtr[i]);
  }
  pCsr->key.nData = 0;
}

static void treeCursorSeek(
  TreeCursor *pTreeCsr, 
  void *pKey, int nKey, 
2521
2522
2523
2524
2525
2526
2527













2528
2529
2530
2531
2532
2533
2534
2535


2536
2537
2538
2539
2540
2541

2542
2543
2544
2545
2546

2547



2548
2549
2550
2551
2552
2553
2554

2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
          lsmTreeCursorPrev(pTreeCsr);
        }
        break;
    }
  }
}














/*
** Seek the cursor.
*/
int lsmMCursorSeek(MultiCursor *pCsr, void *pKey, int nKey, int eSeek){
  int eESeek = eSeek;             /* Effective eSeek parameter */
  int rc = LSM_OK;
  int i;
  int iPtr = 0; 



  if( eESeek==LSM_SEEK_LEFAST ) eESeek = LSM_SEEK_LE;
  assert( eESeek==LSM_SEEK_EQ || eESeek==LSM_SEEK_LE || eESeek==LSM_SEEK_GE );

  assert( (pCsr->flags & CURSOR_NEW_SYSTEM)==0 );
  assert( (pCsr->flags & CURSOR_AT_FREELIST)==0 );


  pCsr->flags &= ~(CURSOR_NEXT_OK | CURSOR_PREV_OK);
  treeCursorSeek(pCsr->apTreeCsr[0], pKey, nKey, eESeek);
  treeCursorSeek(pCsr->apTreeCsr[1], pKey, nKey, eESeek);


  for(i=0; rc==LSM_OK && i<pCsr->nSegCsr; i++){



    rc = segmentCursorSeek(&pCsr->aSegCsr[i], pKey, nKey, iPtr, eESeek, &iPtr);
  }

  if( rc==LSM_OK ){
    rc = multiCursorAllocTree(pCsr);
  }
  if( rc==LSM_OK ){

    for(i=pCsr->nTree-1; i>0; i--){
      multiCursorDoCompare(pCsr, i, eESeek==LSM_SEEK_LE);
    }
    if( eSeek==LSM_SEEK_GE ) pCsr->flags |= CURSOR_NEXT_OK;
    if( eSeek==LSM_SEEK_LE ) pCsr->flags |= CURSOR_PREV_OK;
  }

  multiCursorCacheKey(pCsr, &rc);
  if( rc==LSM_OK 
   && eSeek!=LSM_SEEK_LEFAST
   && (pCsr->flags & CURSOR_IGNORE_DELETE)
   && rtIsDelete(pCsr->eType)
  ){
    switch( eESeek ){
      case LSM_SEEK_EQ:
        lsmMCursorReset(pCsr);
        break;
      case LSM_SEEK_GE:
        rc = lsmMCursorNext(pCsr);
        break;







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






<
|
>
>






>





>
|
>
>
>
|






>








|
<
<
<
<







2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651

2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688




2689
2690
2691
2692
2693
2694
2695
          lsmTreeCursorPrev(pTreeCsr);
        }
        break;
    }
  }
}


static int mcursorLocationOk(MultiCursor *pCsr, int bDeleteOk){
  int eType = pCsr->eType;

  if( ((pCsr->flags & CURSOR_IGNORE_DELETE) && rtIsDelete(eType) && !bDeleteOk)
   || ((pCsr->flags & CURSOR_IGNORE_SYSTEM) && rtTopic(pCsr->eType)!=0)
  ){
    return 0;
  }

  return 1;
}

/*
** Seek the cursor.
*/
int lsmMCursorSeek(MultiCursor *pCsr, void *pKey, int nKey, int eSeek){
  int eESeek = eSeek;             /* Effective eSeek parameter */
  int rc = LSM_OK;

  int iPtr = 0;
  Pgno iPgno = 0; 
  Level *pLvl;

  if( eESeek==LSM_SEEK_LEFAST ) eESeek = LSM_SEEK_LE;
  assert( eESeek==LSM_SEEK_EQ || eESeek==LSM_SEEK_LE || eESeek==LSM_SEEK_GE );

  assert( (pCsr->flags & CURSOR_NEW_SYSTEM)==0 );
  assert( (pCsr->flags & CURSOR_AT_FREELIST)==0 );
  assert( pCsr->nPtr==0 || pCsr->aPtr[0].pLevel );

  pCsr->flags &= ~(CURSOR_NEXT_OK | CURSOR_PREV_OK);
  treeCursorSeek(pCsr->apTreeCsr[0], pKey, nKey, eESeek);
  treeCursorSeek(pCsr->apTreeCsr[1], pKey, nKey, eESeek);

  /* Seek all segment pointers. */
  for(pLvl=pCsr->pDb->pClient->pLevel; rc==LSM_OK && pLvl; pLvl=pLvl->pNext){
    SegmentPtr *pPtr = &pCsr->aPtr[iPtr];
    iPtr += (1+pLvl->nRight);
    assert( pPtr->pSeg==&pLvl->lhs );
    rc = seekInLevel(pCsr, pLvl, pPtr, eESeek, pKey, nKey, &iPgno);
  }

  if( rc==LSM_OK ){
    rc = multiCursorAllocTree(pCsr);
  }
  if( rc==LSM_OK ){
    int i;
    for(i=pCsr->nTree-1; i>0; i--){
      multiCursorDoCompare(pCsr, i, eESeek==LSM_SEEK_LE);
    }
    if( eSeek==LSM_SEEK_GE ) pCsr->flags |= CURSOR_NEXT_OK;
    if( eSeek==LSM_SEEK_LE ) pCsr->flags |= CURSOR_PREV_OK;
  }

  multiCursorCacheKey(pCsr, &rc);
  if( rc==LSM_OK && 0==mcursorLocationOk(pCsr, eSeek==LSM_SEEK_LEFAST) ){




    switch( eESeek ){
      case LSM_SEEK_EQ:
        lsmMCursorReset(pCsr);
        break;
      case LSM_SEEK_GE:
        rc = lsmMCursorNext(pCsr);
        break;
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637




2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664

2665
2666
2667
2668
2669
2670
2671
  ** or less than (if bReverse!=0) the key currently cached in pCsr->key, 
  ** then the cursor has not yet been successfully advanced.  
  */
  multiCursorGetKey(pCsr, pCsr->aTree[1], &eNewType, &pNew, &nNew);
  if( pNew ){
    int res = rtTopic(eNewType) - rtTopic(pCsr->eType);
    if( res==0 ){
      res = pCsr->xCmp(pNew, nNew, pCsr->key.pData, pCsr->key.nData);
    }
    if( (bReverse==0 && res<=0) || (bReverse!=0 && res>=0) ){
      return 0;
    }
  }

  multiCursorCacheKey(pCsr, pRc);
  assert( pCsr->eType==eNewType );

  if( *pRc==LSM_OK 
   && (pCsr->flags & CURSOR_IGNORE_DELETE) 
   && rtIsDelete(eNewType)
   ){
    /* If this cursor is configured to skip deleted keys, and the current
    ** cursor points to a SORTED_DELETE entry, then the cursor has not been 
    ** successfully advanced.  */
    return 0;
  }





  return 1;
}

static int multiCursorAdvance(MultiCursor *pCsr, int bReverse){
  int rc = LSM_OK;                /* Return Code */
  if( lsmMCursorValid(pCsr) ){
    do {
      int iKey = pCsr->aTree[1];
      if( iKey==CURSOR_DATA_TREE0 || iKey==CURSOR_DATA_TREE1 ){
        TreeCursor *pTreeCsr = pCsr->apTreeCsr[iKey-CURSOR_DATA_TREE0];
        if( bReverse ){
          rc = lsmTreeCursorPrev(pTreeCsr);
        }else{
          rc = lsmTreeCursorNext(pTreeCsr);
        }
      }else if( iKey==CURSOR_DATA_SYSTEM ){
        assert( pCsr->flags & CURSOR_AT_FREELIST );
        assert( pCsr->flags & CURSOR_NEW_SYSTEM );
        assert( bReverse==0 );
        pCsr->flags &= ~CURSOR_AT_FREELIST;
      }else if( iKey==(CURSOR_DATA_SEGMENT+pCsr->nSegCsr) ){
        assert( bReverse==0 && pCsr->pBtCsr );
        rc = btreeCursorNext(pCsr->pBtCsr);
      }else{
        LevelCursor *pLevel = &pCsr->aSegCsr[iKey-CURSOR_DATA_SEGMENT];
        rc = segmentCursorAdvance(pLevel, bReverse);
      }

      if( rc==LSM_OK ){
        int i;
        for(i=(iKey+pCsr->nTree)/2; i>0; i=i/2){
          multiCursorDoCompare(pCsr, i, bReverse);
        }
      }
    }while( mcursorAdvanceOk(pCsr, bReverse, &rc)==0 );







|









<
<
<
<
|
|
|
<
<
|
>
>
>
>




















|



<
|

>







2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748




2749
2750
2751


2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780

2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
  ** or less than (if bReverse!=0) the key currently cached in pCsr->key, 
  ** then the cursor has not yet been successfully advanced.  
  */
  multiCursorGetKey(pCsr, pCsr->aTree[1], &eNewType, &pNew, &nNew);
  if( pNew ){
    int res = rtTopic(eNewType) - rtTopic(pCsr->eType);
    if( res==0 ){
      res = pCsr->pDb->xCmp(pNew, nNew, pCsr->key.pData, pCsr->key.nData);
    }
    if( (bReverse==0 && res<=0) || (bReverse!=0 && res>=0) ){
      return 0;
    }
  }

  multiCursorCacheKey(pCsr, pRc);
  assert( pCsr->eType==eNewType );





  /* If this cursor is configured to skip deleted keys, and the current
  ** cursor points to a SORTED_DELETE entry, then the cursor has not been 
  ** successfully advanced.  


  **
  ** Similarly, if the cursor is configured to skip system keys and the
  ** current cursor points to a system key, it has not yet been advanced.
  */
  if( *pRc==LSM_OK && 0==mcursorLocationOk(pCsr, 0) ) return 0;
  return 1;
}

static int multiCursorAdvance(MultiCursor *pCsr, int bReverse){
  int rc = LSM_OK;                /* Return Code */
  if( lsmMCursorValid(pCsr) ){
    do {
      int iKey = pCsr->aTree[1];
      if( iKey==CURSOR_DATA_TREE0 || iKey==CURSOR_DATA_TREE1 ){
        TreeCursor *pTreeCsr = pCsr->apTreeCsr[iKey-CURSOR_DATA_TREE0];
        if( bReverse ){
          rc = lsmTreeCursorPrev(pTreeCsr);
        }else{
          rc = lsmTreeCursorNext(pTreeCsr);
        }
      }else if( iKey==CURSOR_DATA_SYSTEM ){
        assert( pCsr->flags & CURSOR_AT_FREELIST );
        assert( pCsr->flags & CURSOR_NEW_SYSTEM );
        assert( bReverse==0 );
        pCsr->flags &= ~CURSOR_AT_FREELIST;
      }else if( iKey==(CURSOR_DATA_SEGMENT+pCsr->nPtr) ){
        assert( bReverse==0 && pCsr->pBtCsr );
        rc = btreeCursorNext(pCsr->pBtCsr);
      }else{

        rc = segmentCursorAdvance(pCsr, iKey-CURSOR_DATA_SEGMENT, bReverse);
      }

      if( rc==LSM_OK ){
        int i;
        for(i=(iKey+pCsr->nTree)/2; i>0; i=i/2){
          multiCursorDoCompare(pCsr, i, bReverse);
        }
      }
    }while( mcursorAdvanceOk(pCsr, bReverse, &rc)==0 );
3358
3359
3360
3361
3362
3363
3364
3365
3366
3367
3368
3369
3370
3371
3372
3373
3374
3375
3376
3377
3378
3379
3380
3381
3382
3383
3384
3385
3386
3387
3388
3389
3390
3391
3392
3393
3394
3395
3396

  /* Unless the merge has finished, save the cursor position in the
  ** Merge.aInput[] array. See function mergeWorkerInit() for the 
  ** code to restore a cursor position based on aInput[].  */
  if( rc==LSM_OK && pCsr && lsmMCursorValid(pCsr) ){
    Merge *pMerge = pMW->pLevel->pMerge;
    int bBtree = (pCsr->pBtCsr!=0);
    int iSegCsr;

    /* pMerge->nInput==0 indicates that this is a FlushTree() operation. */
    assert( pMerge->nInput==0 || pMW->pLevel->nRight>0 );
    assert( pMerge->nInput==0 || pMerge->nInput==(pCsr->nSegCsr+bBtree) );

    for(i=0; i<(pMerge->nInput-bBtree); i++){
      SegmentPtr *pPtr = &pCsr->aSegCsr[i].aPtr[0];
      if( pPtr->pPg ){
        pMerge->aInput[i].iPg = lsmFsPageNumber(pPtr->pPg);
        pMerge->aInput[i].iCell = pPtr->iCell;
      }else{
        pMerge->aInput[i].iPg = 0;
        pMerge->aInput[i].iCell = 0;
      }
    }
    if( bBtree && pMerge->nInput ){
      assert( i==pCsr->nSegCsr );
      btreeCursorPosition(pCsr->pBtCsr, &pMerge->aInput[i]);
    }

    /* Store the location of the split-key */
    iSegCsr = pCsr->aTree[1] - CURSOR_DATA_SEGMENT;
    if( iSegCsr<pCsr->nSegCsr ){
      pMerge->splitkey = pMerge->aInput[iSegCsr];
    }else{
      btreeCursorSplitkey(pCsr->pBtCsr, &pMerge->splitkey);
    }
  }

  lsmMCursorClose(pCsr);
  lsmFsPageRelease(pMW->pPage);







|



|


|









|




|
|
|







3477
3478
3479
3480
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
3496
3497
3498
3499
3500
3501
3502
3503
3504
3505
3506
3507
3508
3509
3510
3511
3512
3513
3514
3515

  /* Unless the merge has finished, save the cursor position in the
  ** Merge.aInput[] array. See function mergeWorkerInit() for the 
  ** code to restore a cursor position based on aInput[].  */
  if( rc==LSM_OK && pCsr && lsmMCursorValid(pCsr) ){
    Merge *pMerge = pMW->pLevel->pMerge;
    int bBtree = (pCsr->pBtCsr!=0);
    int iPtr;

    /* pMerge->nInput==0 indicates that this is a FlushTree() operation. */
    assert( pMerge->nInput==0 || pMW->pLevel->nRight>0 );
    assert( pMerge->nInput==0 || pMerge->nInput==(pCsr->nPtr+bBtree) );

    for(i=0; i<(pMerge->nInput-bBtree); i++){
      SegmentPtr *pPtr = &pCsr->aPtr[i];
      if( pPtr->pPg ){
        pMerge->aInput[i].iPg = lsmFsPageNumber(pPtr->pPg);
        pMerge->aInput[i].iCell = pPtr->iCell;
      }else{
        pMerge->aInput[i].iPg = 0;
        pMerge->aInput[i].iCell = 0;
      }
    }
    if( bBtree && pMerge->nInput ){
      assert( i==pCsr->nPtr );
      btreeCursorPosition(pCsr->pBtCsr, &pMerge->aInput[i]);
    }

    /* Store the location of the split-key */
    iPtr = pCsr->aTree[1] - CURSOR_DATA_SEGMENT;
    if( iPtr<pCsr->nPtr ){
      pMerge->splitkey = pMerge->aInput[iPtr];
    }else{
      btreeCursorSplitkey(pCsr->pBtCsr, &pMerge->splitkey);
    }
  }

  lsmMCursorClose(pCsr);
  lsmFsPageRelease(pMW->pPage);
3421
3422
3423
3424
3425
3426
3427
3428
3429
3430
3431
3432
3433
3434
3435
  assert( pMW->pPage==0 );

  if( pCsr->pBtCsr ){
    rc = LSM_OK;
    iFPtr = pMW->pLevel->pNext->lhs.iFirst;
  }else{
    Segment *pSeg;
    pSeg = pMW->pCsr->aSegCsr[pMW->pCsr->nSegCsr-1].aPtr[0].pSeg;
    rc = lsmFsDbPageGet(pMW->pDb->pFS, pSeg->iFirst, &pPg);
    if( rc==LSM_OK ){
      u8 *aData;                    /* Buffer for page pPg */
      int nData;                    /* Size of aData[] in bytes */
      aData = fsPageData(pPg, &nData);
      iFPtr = pageGetPtr(aData, nData);
      lsmFsPageRelease(pPg);







|







3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
  assert( pMW->pPage==0 );

  if( pCsr->pBtCsr ){
    rc = LSM_OK;
    iFPtr = pMW->pLevel->pNext->lhs.iFirst;
  }else{
    Segment *pSeg;
    pSeg = pMW->pCsr->aPtr[pMW->pCsr->nPtr-1].pSeg;
    rc = lsmFsDbPageGet(pMW->pDb->pFS, pSeg->iFirst, &pPg);
    if( rc==LSM_OK ){
      u8 *aData;                    /* Buffer for page pPg */
      int nData;                    /* Size of aData[] in bytes */
      aData = fsPageData(pPg, &nData);
      iFPtr = pageGetPtr(aData, nData);
      lsmFsPageRelease(pPg);
3470
3471
3472
3473
3474
3475
3476
3477
3478
3479
3480
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
3496
    if( pBtCsr->pKey ){
      int res = rtTopic(pBtCsr->eType) - rtTopic(eType);
      if( res==0 ) res = pDb->xCmp(pBtCsr->pKey, pBtCsr->nKey, pKey, nKey);
      if( 0==res ) iPtr = pBtCsr->iPtr;

      assert( res>=0 );
    }
  }else if( pCsr->nSegCsr ){
    LevelCursor *pPtrs = &pCsr->aSegCsr[pCsr->nSegCsr-1];
    if( segmentCursorValid(pPtrs)
     && 0==pDb->xCmp(pPtrs->aPtr[0].pKey, pPtrs->aPtr[0].nKey, pKey, nKey)
    ){
      iPtr = pPtrs->aPtr[0].iPtr+pPtrs->aPtr[0].iPgPtr;
    }
  }

  if( pMW->aGobble ){
    int iGobble = pCsr->aTree[1] - CURSOR_DATA_SEGMENT;
    if( iGobble<pMW->pLevel->nRight ){
      SegmentPtr *pGobble = &pCsr->aSegCsr[iGobble].aPtr[0];
      if( (pGobble->flags & PGFTR_SKIP_THIS_FLAG)==0 ){
        pMW->aGobble[iGobble] = lsmFsPageNumber(pGobble->pPg);
      }
    }
  }

  /* If this is a separator key and we know that the output pointer has not







|
|
|
|

|





|
|







3589
3590
3591
3592
3593
3594
3595
3596
3597
3598
3599
3600
3601
3602
3603
3604
3605
3606
3607
3608
3609
3610
3611
3612
3613
3614
3615
    if( pBtCsr->pKey ){
      int res = rtTopic(pBtCsr->eType) - rtTopic(eType);
      if( res==0 ) res = pDb->xCmp(pBtCsr->pKey, pBtCsr->nKey, pKey, nKey);
      if( 0==res ) iPtr = pBtCsr->iPtr;

      assert( res>=0 );
    }
  }else if( pCsr->nPtr ){
    SegmentPtr *pPtr = &pCsr->aPtr[pCsr->nPtr-1];
    if( pPtr->pPg
     && 0==pDb->xCmp(pPtr->pKey, pPtr->nKey, pKey, nKey)
    ){
      iPtr = pPtr->iPtr+pPtr->iPgPtr;
    }
  }

  if( pMW->aGobble ){
    int iGobble = pCsr->aTree[1] - CURSOR_DATA_SEGMENT;
    if( iGobble<pCsr->nPtr ){
      SegmentPtr *pGobble = &pCsr->aPtr[iGobble];
      if( (pGobble->flags & PGFTR_SKIP_THIS_FLAG)==0 ){
        pMW->aGobble[iGobble] = lsmFsPageNumber(pGobble->pPg);
      }
    }
  }

  /* If this is a separator key and we know that the output pointer has not
3655
3656
3657
3658
3659
3660
3661
3662
3663
3664
3665
3666
3667
3668
3669
  if( rc==LSM_OK ){
    sortedInvokeWorkHook(pDb);
  }

  if( pnWrite ) *pnWrite = nWrite;
  pDb->pWorker->nWrite += nWrite;
#if 0
  lsmSortedDumpStructure(pDb, pDb->pWorker, 1, 0, "new-toplevel");
#endif
  return rc;
}

/*
** The nMerge levels in the LSM beginning with pLevel consist of a
** left-hand-side segment only. Replace these levels with a single new







|







3774
3775
3776
3777
3778
3779
3780
3781
3782
3783
3784
3785
3786
3787
3788
  if( rc==LSM_OK ){
    sortedInvokeWorkHook(pDb);
  }

  if( pnWrite ) *pnWrite = nWrite;
  pDb->pWorker->nWrite += nWrite;
#if 0
  lsmSortedDumpStructure(pDb, pDb->pWorker, 0, 0, "new-toplevel");
#endif
  return rc;
}

/*
** The nMerge levels in the LSM beginning with pLevel consist of a
** left-hand-side segment only. Replace these levels with a single new
3814
3815
3816
3817
3818
3819
3820
3821
3822
3823
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
3841
3842
3843
3844
3845
3846
3847
3848
3849
3850
3851

3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862
3863
3864
3865
3866
3867
3868
3869
3870
3871
3872
3873
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883
3884
3885
3886
3887
3888
3889
3890
3891
      Level *pNext = pLevel->pNext;
      rc = multiCursorAddLevel(pCsr, pNext, MULTICURSOR_ADDLEVEL_LHS_SEP);
    }
    multiCursorReadSeparators(pCsr);
  }else{
    multiCursorIgnoreDelete(pCsr);
  }
  assert( rc!=LSM_OK || pMerge->nInput==(pCsr->nSegCsr+(pCsr->pBtCsr!=0)) );
  pMW->pCsr = pCsr;

  /* Load the current output page into memory. */
  if( rc==LSM_OK ) rc = mergeWorkerLoadOutputPage(pMW);

  /* Position the cursor. */
  if( rc==LSM_OK ){
    if( pMW->pPage==0 ){
      /* The output array is still empty. So position the cursor at the very 
      ** start of the input.  */
      rc = multiCursorEnd(pCsr, 0);
    }else{
      /* The output array is non-empty. Position the cursor based on the
      ** page/cell data saved in the Merge.aInput[] array.  */
      int i;
      for(i=0; rc==LSM_OK && i<pCsr->nSegCsr; i++){
        MergeInput *pInput = &pMerge->aInput[i];
        if( pInput->iPg ){
          SegmentPtr *pPtr;
          assert( pCsr->aSegCsr[i].nPtr==1 );
          assert( pCsr->aSegCsr[i].aPtr[0].pPg==0 );
          pPtr = &pCsr->aSegCsr[i].aPtr[0];
          rc = segmentPtrLoadPage(pDb->pFS, pPtr, pInput->iPg);
          if( rc==LSM_OK && pPtr->nCell>0 ){
            rc = segmentPtrLoadCell(pPtr, pInput->iCell);
          }
        }
      }

      if( rc==LSM_OK && pCsr->pBtCsr ){

        assert( i==pCsr->nSegCsr );
        rc = btreeCursorRestore(pCsr->pBtCsr, pCsr->xCmp, &pMerge->aInput[i]);
      }

      if( rc==LSM_OK ){
        rc = multiCursorSetupTree(pCsr, 0);
      }
    }
    pCsr->flags |= CURSOR_NEXT_OK;
  }

  return rc;
}

static int sortedBtreeGobble(
  lsm_db *pDb, 
  MultiCursor *pCsr, 
  int iGobble
){
  int rc = LSM_OK;
  if( rtTopic(pCsr->eType)==0 ){
    LevelCursor *pLvlcsr = &pCsr->aSegCsr[iGobble];
    Segment *pSeg = pLvlcsr->aPtr[0].pSeg;
    Blob *p = &pCsr->key;
    Pgno *aPg;
    int nPg;

    assert( pLvlcsr->nPtr==1 );
    assert( pSeg->iRoot>0 );

    aPg = lsmMallocZeroRc(pDb->pEnv, sizeof(Pgno)*32, &rc);
    if( rc==LSM_OK ){
      rc = seekInBtree(pLvlcsr, pSeg, p->pData, p->nData, aPg, 0); 
    }

    for(nPg=0; aPg[nPg]; nPg++);
#if 1
    lsmFsGobble(pDb, pSeg, aPg, nPg);
#endif








|















|



|
<
|








>
|
|



















<
|




<

<


|







3933
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
3950
3951
3952
3953
3954
3955
3956
3957
3958
3959
3960

3961
3962
3963
3964
3965
3966
3967
3968
3969
3970
3971
3972
3973
3974
3975
3976
3977
3978
3979
3980
3981
3982
3983
3984
3985
3986
3987
3988
3989
3990
3991

3992
3993
3994
3995
3996

3997

3998
3999
4000
4001
4002
4003
4004
4005
4006
4007
      Level *pNext = pLevel->pNext;
      rc = multiCursorAddLevel(pCsr, pNext, MULTICURSOR_ADDLEVEL_LHS_SEP);
    }
    multiCursorReadSeparators(pCsr);
  }else{
    multiCursorIgnoreDelete(pCsr);
  }
  assert( rc!=LSM_OK || pMerge->nInput==(pCsr->nPtr+(pCsr->pBtCsr!=0)) );
  pMW->pCsr = pCsr;

  /* Load the current output page into memory. */
  if( rc==LSM_OK ) rc = mergeWorkerLoadOutputPage(pMW);

  /* Position the cursor. */
  if( rc==LSM_OK ){
    if( pMW->pPage==0 ){
      /* The output array is still empty. So position the cursor at the very 
      ** start of the input.  */
      rc = multiCursorEnd(pCsr, 0);
    }else{
      /* The output array is non-empty. Position the cursor based on the
      ** page/cell data saved in the Merge.aInput[] array.  */
      int i;
      for(i=0; rc==LSM_OK && i<pCsr->nPtr; i++){
        MergeInput *pInput = &pMerge->aInput[i];
        if( pInput->iPg ){
          SegmentPtr *pPtr;
          assert( pCsr->aPtr[i].pPg==0 );

          pPtr = &pCsr->aPtr[i];
          rc = segmentPtrLoadPage(pDb->pFS, pPtr, pInput->iPg);
          if( rc==LSM_OK && pPtr->nCell>0 ){
            rc = segmentPtrLoadCell(pPtr, pInput->iCell);
          }
        }
      }

      if( rc==LSM_OK && pCsr->pBtCsr ){
        int (*xCmp)(void *, int, void *, int) = pCsr->pDb->xCmp;
        assert( i==pCsr->nPtr );
        rc = btreeCursorRestore(pCsr->pBtCsr, xCmp, &pMerge->aInput[i]);
      }

      if( rc==LSM_OK ){
        rc = multiCursorSetupTree(pCsr, 0);
      }
    }
    pCsr->flags |= CURSOR_NEXT_OK;
  }

  return rc;
}

static int sortedBtreeGobble(
  lsm_db *pDb, 
  MultiCursor *pCsr, 
  int iGobble
){
  int rc = LSM_OK;
  if( rtTopic(pCsr->eType)==0 ){

    Segment *pSeg = pCsr->aPtr[iGobble].pSeg;
    Blob *p = &pCsr->key;
    Pgno *aPg;
    int nPg;


    assert( pSeg->iRoot>0 );

    aPg = lsmMallocZeroRc(pDb->pEnv, sizeof(Pgno)*32, &rc);
    if( rc==LSM_OK ){
      rc = seekInBtree(pCsr, pSeg, p->pData, p->nData, aPg, 0); 
    }

    for(nPg=0; aPg[nPg]; nPg++);
#if 1
    lsmFsGobble(pDb, pSeg, aPg, nPg);
#endif

4023
4024
4025
4026
4027
4028
4029
4030
4031
4032
4033
4034
4035
4036
4037
      ** from rhs segments for which the content has been completely merged
      ** into the lhs of the level.
      */
      if( rc==LSM_OK ){
        if( mergeWorkerDone(&mergeworker)==0 ){
          int i;
          for(i=0; i<pLevel->nRight; i++){
            SegmentPtr *pGobble = &mergeworker.pCsr->aSegCsr[i].aPtr[0];
            if( pGobble->pSeg->iRoot ){
              rc = sortedBtreeGobble(pDb, mergeworker.pCsr, i);
            }else if( mergeworker.aGobble[i] ){
              lsmFsGobble(pDb, pGobble->pSeg, &mergeworker.aGobble[i], 1);
            }
          }
#if 0







|







4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
4149
4150
4151
4152
4153
      ** from rhs segments for which the content has been completely merged
      ** into the lhs of the level.
      */
      if( rc==LSM_OK ){
        if( mergeWorkerDone(&mergeworker)==0 ){
          int i;
          for(i=0; i<pLevel->nRight; i++){
            SegmentPtr *pGobble = &mergeworker.pCsr->aPtr[i];
            if( pGobble->pSeg->iRoot ){
              rc = sortedBtreeGobble(pDb, mergeworker.pCsr, i);
            }else if( mergeworker.aGobble[i] ){
              lsmFsGobble(pDb, pGobble->pSeg, &mergeworker.aGobble[i], 1);
            }
          }
#if 0
4345
4346
4347
4348
4349
4350
4351
4352
4353
4354
4355
4356
4357
4358
4359
4360
4361
4362
4363
4364
4365
4366
4367
4368
4369
4370
4371
4372
4373
4374



4375
4376
4377
4378
4379
4380
4381
4382
4383
4384
4385
4386
4387
4388
4389
4390
4391

4392
4393
4394
4395
4396
4397
4398
** nUnit database pages worth of data have been written to the database
** (i.e. the in-memory tree) since the last call.
*/
int lsmSortedAutoWork(
  lsm_db *pDb,                    /* Database handle */
  int nUnit                       /* Pages of data written to in-memory tree */
){
  int rc;                         /* Return code */
  int nRemaining;                 /* Units of work to do before returning */
  int nDepth = 0;                 /* Current height of tree (longest path) */
  Level *pLevel;                  /* Used to iterate through levels */
  int bRestore = 0;

  assert( pDb->pWorker==0 );
  assert( pDb->nTransOpen>0 );

  /* Determine how many units of work to do before returning. One unit of
  ** work is achieved by writing one page (~4KB) of merged data.  */
  nRemaining = 0;
  for(pLevel=lsmDbSnapshotLevel(pDb->pClient); pLevel; pLevel=pLevel->pNext){
    /* nDepth += LSM_MAX(1, pLevel->nRight); */
    nDepth += 1;
  }
  if( lsmTreeHasOld(pDb) ){
    nDepth += 1;
    bRestore = 1;
    rc = lsmSaveCursors(pDb);
    if( rc!=LSM_OK ) return rc;
  }




  nRemaining = nUnit * nDepth;
#ifdef LSM_LOG_WORK
  lsmLogMessage(pDb, rc, "lsmSortedAutoWork(): %d*%d = %d pages", 
      nUnit, nDepth, nRemaining);
#endif
  rc = doLsmWork(pDb, LSM_WORK_FLUSH, nRemaining, 0);
  if( rc==LSM_BUSY ) rc = LSM_OK;

  if( bRestore && pDb->pCsr ){
    lsmFreeSnapshot(pDb->pEnv, pDb->pClient);
    pDb->pClient = 0;
    rc = lsmCheckpointLoad(pDb, 0);
    if( rc==LSM_OK ){
      rc = lsmCheckpointDeserialize(pDb, 0, pDb->aSnapshot, &pDb->pClient);
    }
    if( rc==LSM_OK ){
      rc = lsmRestoreCursors(pDb);

    }
  }

  return rc;
}

/*







|
<









<











>
>
>
|

|
|

|
|

|
|
|
|
|
|
|
|
|
>







4461
4462
4463
4464
4465
4466
4467
4468

4469
4470
4471
4472
4473
4474
4475
4476
4477

4478
4479
4480
4481
4482
4483
4484
4485
4486
4487
4488
4489
4490
4491
4492
4493
4494
4495
4496
4497
4498
4499
4500
4501
4502
4503
4504
4505
4506
4507
4508
4509
4510
4511
4512
4513
4514
4515
4516
** nUnit database pages worth of data have been written to the database
** (i.e. the in-memory tree) since the last call.
*/
int lsmSortedAutoWork(
  lsm_db *pDb,                    /* Database handle */
  int nUnit                       /* Pages of data written to in-memory tree */
){
  int rc = LSM_OK;                /* Return code */

  int nDepth = 0;                 /* Current height of tree (longest path) */
  Level *pLevel;                  /* Used to iterate through levels */
  int bRestore = 0;

  assert( pDb->pWorker==0 );
  assert( pDb->nTransOpen>0 );

  /* Determine how many units of work to do before returning. One unit of
  ** work is achieved by writing one page (~4KB) of merged data.  */

  for(pLevel=lsmDbSnapshotLevel(pDb->pClient); pLevel; pLevel=pLevel->pNext){
    /* nDepth += LSM_MAX(1, pLevel->nRight); */
    nDepth += 1;
  }
  if( lsmTreeHasOld(pDb) ){
    nDepth += 1;
    bRestore = 1;
    rc = lsmSaveCursors(pDb);
    if( rc!=LSM_OK ) return rc;
  }

  if( nDepth>0 ){
    int nRemaining;               /* Units of work to do before returning */

    nRemaining = nUnit * nDepth;
#ifdef LSM_LOG_WORK
    lsmLogMessage(pDb, rc, "lsmSortedAutoWork(): %d*%d = %d pages", 
        nUnit, nDepth, nRemaining);
#endif
    rc = doLsmWork(pDb, LSM_WORK_FLUSH, nRemaining, 0);
    if( rc==LSM_BUSY ) rc = LSM_OK;

    if( bRestore && pDb->pCsr ){
      lsmFreeSnapshot(pDb->pEnv, pDb->pClient);
      pDb->pClient = 0;
      rc = lsmCheckpointLoad(pDb, 0);
      if( rc==LSM_OK ){
        rc = lsmCheckpointDeserialize(pDb, 0, pDb->aSnapshot, &pDb->pClient);
      }
      if( rc==LSM_OK ){
        rc = lsmRestoreCursors(pDb);
      }
    }
  }

  return rc;
}

/*
Changes to src/lsm_varint.c.
150
151
152
153
154
155
156
157

158
159

















160
161
162
163
164
165
166
167
168
169
  return lsmSqlite4GetVarint64(aData, (u64 *)piVal);
}

int lsmVarintPut32(u8 *aData, int iVal){
  return lsmSqlite4PutVarint64(aData, (u64)iVal);
}

int lsmVarintGet32(u8 *aData, int *piVal){

  int nByte;
  u64 iVal64;

















  nByte = lsmSqlite4GetVarint64(aData, &iVal64);
  *piVal = (int)iVal64;
  return nByte;
}

int lsmVarintLen32(int n){
  u8 aData[9];
  return lsmVarintPut32(aData, n);
}








|
>
|
|
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
|
|
|







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
  return lsmSqlite4GetVarint64(aData, (u64 *)piVal);
}

int lsmVarintPut32(u8 *aData, int iVal){
  return lsmSqlite4PutVarint64(aData, (u64)iVal);
}

int lsmVarintGet32(u8 *z, int *piVal){
  u64 i;
  int ret;

  if( z[0]<=240 ){
    *piVal = z[0];
    return 1;
  }
  if( z[0]<=248 ){
    *piVal = (z[0]-241)*256 + z[1] + 240;
    return 2;
  }
  if( z[0]==249 ){
    *piVal = 2288 + 256*z[1] + z[2];
    return 3;
  }
  if( z[0]==250 ){
    *piVal = (z[1]<<16) + (z[2]<<8) + z[3];
    return 4;
  }

  ret = lsmSqlite4GetVarint64(z, &i);
  *piVal = i;
  return ret;
}

int lsmVarintLen32(int n){
  u8 aData[9];
  return lsmVarintPut32(aData, n);
}

Changes to tool/lsmperf.tcl.
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
  append script $data2
  append script $data3

  append script "pause -1\n"
  exec_gnuplot_script $script $zPng
}

do_write_test x.png 120 50000 50000 30 {
  lsm-mt     "mmap=1 multi_proc=0 safety=1 threads=3 autowork=0"
  LevelDB leveldb
}

# lsm-mt    "mmap=1 multi_proc=0 threads=2 autowork=0 autocheckpoint=8192000"
# lsm-mt     "mmap=1 multi_proc=0 safety=1 threads=3 autowork=0"
# lsm-st     "mmap=1 multi_proc=0 safety=1 threads=1 autowork=1"
# lsm-mt     "mmap=1 multi_proc=0 safety=1 threads=3 autowork=0"
# lsm-mt     "mmap=1 multi_proc=0 safety=1 threads=3 autowork=0"







|
|
|







164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
  append script $data2
  append script $data3

  append script "pause -1\n"
  exec_gnuplot_script $script $zPng
}

do_write_test x.png 200 50000 0 20 {
  lsm-mt     "mmap=1 multi_proc=0 safety=1 threads=3 autowork=0 nmerge=4"
  leveldb leveldb
}

# lsm-mt    "mmap=1 multi_proc=0 threads=2 autowork=0 autocheckpoint=8192000"
# lsm-mt     "mmap=1 multi_proc=0 safety=1 threads=3 autowork=0"
# lsm-st     "mmap=1 multi_proc=0 safety=1 threads=1 autowork=1"
# lsm-mt     "mmap=1 multi_proc=0 safety=1 threads=3 autowork=0"
# lsm-mt     "mmap=1 multi_proc=0 safety=1 threads=3 autowork=0"