/ Check-in [30bf38d5]
Login

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

Overview
Comment:Change some LSM code internal typenames from "Blob" and "Pgno" to "LsmBlob" and "LsmPgno". This allows the LSM code to be compiled as a single unit with sqlite3.c.
Downloads: Tarball | ZIP archive | SQL archive
Timelines: family | ancestors | descendants | both | trunk
Files: files | file ages | folders
SHA3-256: 30bf38d589adf0b2eb613e184ffb03ed7e625736cb04fd5c341328f72bc5ef4d
User & Date: dan 2017-12-13 15:32:33
Context
2017-12-13
18:01
Correctly invalidate a column cache line when it is overwritten with NULL. Fix for ticket [123c9ba32130a6c9d432]. check-in: 65754dc3 user: drh tags: trunk
15:32
Change some LSM code internal typenames from "Blob" and "Pgno" to "LsmBlob" and "LsmPgno". This allows the LSM code to be compiled as a single unit with sqlite3.c. check-in: 30bf38d5 user: dan tags: trunk
14:22
Add script to amalgamate all LSM files into "lsm1.c". check-in: e32b69d7 user: dan tags: trunk
Changes
Hide Diffs Unified Diffs Ignore Whitespace Patch

Changes to ext/lsm1/lsmInt.h.

106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
...
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
...
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
...
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
...
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
...
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
...
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
...
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
typedef unsigned short int u16;
typedef unsigned int u32;
typedef lsm_i64 i64;
typedef unsigned long long int u64;
#endif

/* A page number is a 64-bit integer. */
typedef i64 Pgno;

#ifdef LSM_DEBUG
int lsmErrorBkpt(int);
#else
# define lsmErrorBkpt(x) (x)
#endif

................................................................................
  void **apShm;                   /* Shared memory chunks */
  ShmHeader *pShmhdr;             /* Live shared-memory header */
  TreeHeader treehdr;             /* Local copy of tree-header */
  u32 aSnapshot[LSM_META_PAGE_SIZE / sizeof(u32)];
};

struct Segment {
  Pgno iFirst;                     /* First page of this run */
  Pgno iLastPg;                    /* Last page of this run */
  Pgno iRoot;                      /* Root page number (if any) */
  int nSize;                       /* Size of this run in pages */

  Redirect *pRedirect;             /* Block redirects (or NULL) */
};

/*
** iSplitTopic/pSplitKey/nSplitKey:
................................................................................
** access to the associated Level struct.
**
** iOutputOff:
**   The byte offset to write to next within the last page of the 
**   output segment.
*/
struct MergeInput {
  Pgno iPg;                       /* Page on which next input is stored */
  int iCell;                      /* Cell containing next input to merge */
};
struct Merge {
  int nInput;                     /* Number of input runs being merged */
  MergeInput *aInput;             /* Array nInput entries in size */
  MergeInput splitkey;            /* Location in file of current splitkey */
  int nSkip;                      /* Number of separators entries to skip */
  int iOutputOff;                 /* Write offset on output page */
  Pgno iCurrentPtr;               /* Current pointer value */
};

/* 
** The first argument to this macro is a pointer to a Segment structure.
** Returns true if the structure instance indicates that the separators
** array is valid.
*/
................................................................................
  u32 iCmpId;                     /* Id of compression scheme */
  Level *pLevel;                  /* Pointer to level 0 of snapshot (or NULL) */
  i64 iId;                        /* Snapshot id */
  i64 iLogOff;                    /* Log file offset */
  Redirect redirect;              /* Block redirection array */

  /* Used by worker snapshots only */
  int nBlock;                     /* Number of blocks in database file */
  Pgno aiAppend[LSM_APPLIST_SZ];  /* Append point list */
  Freelist freelist;              /* Free block list */
  u32 nWrite;                     /* Total number of pages written to disk */
};
#define LSM_INITIAL_SNAPSHOT_ID 11

/*
** Functions from file "lsm_ckpt.c".
*/
int lsmCheckpointWrite(lsm_db *, u32 *);
................................................................................

int lsmFsPageSize(FileSystem *);
void lsmFsSetPageSize(FileSystem *, int);

int lsmFsFileid(lsm_db *pDb, void **ppId, int *pnId);

/* Creating, populating, gobbling and deleting sorted runs. */
void lsmFsGobble(lsm_db *, Segment *, Pgno *, int);
int lsmFsSortedDelete(FileSystem *, Snapshot *, int, Segment *);
int lsmFsSortedFinish(FileSystem *, Segment *);
int lsmFsSortedAppend(FileSystem *, Snapshot *, Level *, int, Page **);
int lsmFsSortedPadding(FileSystem *, Snapshot *, Segment *);

/* Functions to retrieve the lsm_env pointer from a FileSystem or Page object */
lsm_env *lsmFsEnv(FileSystem *);
................................................................................

int lsmFsSectorSize(FileSystem *);

void lsmSortedSplitkey(lsm_db *, Level *, int *);

/* Reading sorted run content. */
int lsmFsDbPageLast(FileSystem *pFS, Segment *pSeg, Page **ppPg);
int lsmFsDbPageGet(FileSystem *, Segment *, Pgno, Page **);
int lsmFsDbPageNext(Segment *, Page *, int eDir, Page **);

u8 *lsmFsPageData(Page *, int *);
int lsmFsPageRelease(Page *);
int lsmFsPagePersist(Page *);
void lsmFsPageRef(Page *);
Pgno lsmFsPageNumber(Page *);

int lsmFsNRead(FileSystem *);
int lsmFsNWrite(FileSystem *);

int lsmFsMetaPageGet(FileSystem *, int, int, MetaPage **);
int lsmFsMetaPageRelease(MetaPage *);
u8 *lsmFsMetaPageData(MetaPage *, int *);

#ifdef LSM_DEBUG
int lsmFsDbPageIsLast(Segment *pSeg, Page *pPg);
int lsmFsIntegrityCheck(lsm_db *);
#endif

Pgno lsmFsRedirectPage(FileSystem *, Redirect *, Pgno);

int lsmFsPageWritable(Page *);

/* Functions to read, write and sync the log file. */
int lsmFsWriteLog(FileSystem *pFS, i64 iOff, LsmString *pStr);
int lsmFsSyncLog(FileSystem *pFS);
int lsmFsReadLog(FileSystem *pFS, i64 iOff, int nRead, LsmString *pStr);
................................................................................

/* And to sync the db file */
int lsmFsSyncDb(FileSystem *, int);

void lsmFsFlushWaiting(FileSystem *, int *);

/* Used by lsm_info(ARRAY_STRUCTURE) and lsm_config(MMAP) */
int lsmInfoArrayStructure(lsm_db *pDb, int bBlock, Pgno iFirst, char **pzOut);
int lsmInfoArrayPages(lsm_db *pDb, Pgno iFirst, char **pzOut);
int lsmConfigMmap(lsm_db *pDb, int *piParam);

int lsmEnvOpen(lsm_env *, const char *, int, lsm_file **);
int lsmEnvClose(lsm_env *pEnv, lsm_file *pFile);
int lsmEnvLock(lsm_env *pEnv, lsm_file *pFile, int iLock, int eLock);
int lsmEnvTestLock(lsm_env *pEnv, lsm_file *pFile, int iLock, int nLock, int);

................................................................................
void lsmEnvShmBarrier(lsm_env *);
void lsmEnvShmUnmap(lsm_env *, lsm_file *, int);

void lsmEnvSleep(lsm_env *, int);

int lsmFsReadSyncedId(lsm_db *db, int, i64 *piVal);

int lsmFsSegmentContainsPg(FileSystem *pFS, Segment *, Pgno, int *);

void lsmFsPurgeCache(FileSystem *);

/*
** End of functions from "lsm_file.c".
**************************************************************************/

/* 
** Functions from file "lsm_sorted.c".
*/
int lsmInfoPageDump(lsm_db *, Pgno, int, char **);
void lsmSortedCleanup(lsm_db *);
int lsmSortedAutoWork(lsm_db *, int nUnit);

int lsmSortedWalkFreelist(lsm_db *, int, int (*)(void *, int, i64), void *);

int lsmSaveWorker(lsm_db *, int);








|







 







|
|
|







 







|








|







 







|
|
|
|







 







|







 







|






|













|







 







|
|







 







|










|







106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
...
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
...
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
...
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
...
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
...
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
...
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
...
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
typedef unsigned short int u16;
typedef unsigned int u32;
typedef lsm_i64 i64;
typedef unsigned long long int u64;
#endif

/* A page number is a 64-bit integer. */
typedef i64 LsmPgno;

#ifdef LSM_DEBUG
int lsmErrorBkpt(int);
#else
# define lsmErrorBkpt(x) (x)
#endif

................................................................................
  void **apShm;                   /* Shared memory chunks */
  ShmHeader *pShmhdr;             /* Live shared-memory header */
  TreeHeader treehdr;             /* Local copy of tree-header */
  u32 aSnapshot[LSM_META_PAGE_SIZE / sizeof(u32)];
};

struct Segment {
  LsmPgno iFirst;                  /* First page of this run */
  LsmPgno iLastPg;                 /* Last page of this run */
  LsmPgno iRoot;                   /* Root page number (if any) */
  int nSize;                       /* Size of this run in pages */

  Redirect *pRedirect;             /* Block redirects (or NULL) */
};

/*
** iSplitTopic/pSplitKey/nSplitKey:
................................................................................
** access to the associated Level struct.
**
** iOutputOff:
**   The byte offset to write to next within the last page of the 
**   output segment.
*/
struct MergeInput {
  LsmPgno iPg;                    /* Page on which next input is stored */
  int iCell;                      /* Cell containing next input to merge */
};
struct Merge {
  int nInput;                     /* Number of input runs being merged */
  MergeInput *aInput;             /* Array nInput entries in size */
  MergeInput splitkey;            /* Location in file of current splitkey */
  int nSkip;                      /* Number of separators entries to skip */
  int iOutputOff;                 /* Write offset on output page */
  LsmPgno iCurrentPtr;            /* Current pointer value */
};

/* 
** The first argument to this macro is a pointer to a Segment structure.
** Returns true if the structure instance indicates that the separators
** array is valid.
*/
................................................................................
  u32 iCmpId;                     /* Id of compression scheme */
  Level *pLevel;                  /* Pointer to level 0 of snapshot (or NULL) */
  i64 iId;                        /* Snapshot id */
  i64 iLogOff;                    /* Log file offset */
  Redirect redirect;              /* Block redirection array */

  /* Used by worker snapshots only */
  int nBlock;                        /* Number of blocks in database file */
  LsmPgno aiAppend[LSM_APPLIST_SZ];  /* Append point list */
  Freelist freelist;                 /* Free block list */
  u32 nWrite;                        /* Total number of pages written to disk */
};
#define LSM_INITIAL_SNAPSHOT_ID 11

/*
** Functions from file "lsm_ckpt.c".
*/
int lsmCheckpointWrite(lsm_db *, u32 *);
................................................................................

int lsmFsPageSize(FileSystem *);
void lsmFsSetPageSize(FileSystem *, int);

int lsmFsFileid(lsm_db *pDb, void **ppId, int *pnId);

/* Creating, populating, gobbling and deleting sorted runs. */
void lsmFsGobble(lsm_db *, Segment *, LsmPgno *, int);
int lsmFsSortedDelete(FileSystem *, Snapshot *, int, Segment *);
int lsmFsSortedFinish(FileSystem *, Segment *);
int lsmFsSortedAppend(FileSystem *, Snapshot *, Level *, int, Page **);
int lsmFsSortedPadding(FileSystem *, Snapshot *, Segment *);

/* Functions to retrieve the lsm_env pointer from a FileSystem or Page object */
lsm_env *lsmFsEnv(FileSystem *);
................................................................................

int lsmFsSectorSize(FileSystem *);

void lsmSortedSplitkey(lsm_db *, Level *, int *);

/* Reading sorted run content. */
int lsmFsDbPageLast(FileSystem *pFS, Segment *pSeg, Page **ppPg);
int lsmFsDbPageGet(FileSystem *, Segment *, LsmPgno, Page **);
int lsmFsDbPageNext(Segment *, Page *, int eDir, Page **);

u8 *lsmFsPageData(Page *, int *);
int lsmFsPageRelease(Page *);
int lsmFsPagePersist(Page *);
void lsmFsPageRef(Page *);
LsmPgno lsmFsPageNumber(Page *);

int lsmFsNRead(FileSystem *);
int lsmFsNWrite(FileSystem *);

int lsmFsMetaPageGet(FileSystem *, int, int, MetaPage **);
int lsmFsMetaPageRelease(MetaPage *);
u8 *lsmFsMetaPageData(MetaPage *, int *);

#ifdef LSM_DEBUG
int lsmFsDbPageIsLast(Segment *pSeg, Page *pPg);
int lsmFsIntegrityCheck(lsm_db *);
#endif

LsmPgno lsmFsRedirectPage(FileSystem *, Redirect *, LsmPgno);

int lsmFsPageWritable(Page *);

/* Functions to read, write and sync the log file. */
int lsmFsWriteLog(FileSystem *pFS, i64 iOff, LsmString *pStr);
int lsmFsSyncLog(FileSystem *pFS);
int lsmFsReadLog(FileSystem *pFS, i64 iOff, int nRead, LsmString *pStr);
................................................................................

/* And to sync the db file */
int lsmFsSyncDb(FileSystem *, int);

void lsmFsFlushWaiting(FileSystem *, int *);

/* Used by lsm_info(ARRAY_STRUCTURE) and lsm_config(MMAP) */
int lsmInfoArrayStructure(lsm_db *pDb, int bBlock, LsmPgno iFirst, char **pz);
int lsmInfoArrayPages(lsm_db *pDb, LsmPgno iFirst, char **pzOut);
int lsmConfigMmap(lsm_db *pDb, int *piParam);

int lsmEnvOpen(lsm_env *, const char *, int, lsm_file **);
int lsmEnvClose(lsm_env *pEnv, lsm_file *pFile);
int lsmEnvLock(lsm_env *pEnv, lsm_file *pFile, int iLock, int eLock);
int lsmEnvTestLock(lsm_env *pEnv, lsm_file *pFile, int iLock, int nLock, int);

................................................................................
void lsmEnvShmBarrier(lsm_env *);
void lsmEnvShmUnmap(lsm_env *, lsm_file *, int);

void lsmEnvSleep(lsm_env *, int);

int lsmFsReadSyncedId(lsm_db *db, int, i64 *piVal);

int lsmFsSegmentContainsPg(FileSystem *pFS, Segment *, LsmPgno, int *);

void lsmFsPurgeCache(FileSystem *);

/*
** End of functions from "lsm_file.c".
**************************************************************************/

/* 
** Functions from file "lsm_sorted.c".
*/
int lsmInfoPageDump(lsm_db *, LsmPgno, int, char **);
void lsmSortedCleanup(lsm_db *);
int lsmSortedAutoWork(lsm_db *, int nUnit);

int lsmSortedWalkFreelist(lsm_db *, int, int (*)(void *, int, i64), void *);

int lsmSaveWorker(lsm_db *, int);

Changes to ext/lsm1/lsm_ckpt.c.

385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
static void ckptExportAppendlist(
  lsm_db *db,                     /* Database connection */
  CkptBuffer *p,                  /* Checkpoint buffer to write to */
  int *piOut,                     /* IN/OUT: Offset within checkpoint buffer */
  int *pRc                        /* IN/OUT: Error code */
){
  int i;
  Pgno *aiAppend = db->pWorker->aiAppend;

  for(i=0; i<LSM_APPLIST_SZ; i++){
    ckptAppend64(p, piOut, aiAppend[i], pRc);
  }
};

static int ckptExportSnapshot( 







|







385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
static void ckptExportAppendlist(
  lsm_db *db,                     /* Database connection */
  CkptBuffer *p,                  /* Checkpoint buffer to write to */
  int *piOut,                     /* IN/OUT: Offset within checkpoint buffer */
  int *pRc                        /* IN/OUT: Error code */
){
  int i;
  LsmPgno *aiAppend = db->pWorker->aiAppend;

  for(i=0; i<LSM_APPLIST_SZ; i++){
    ckptAppend64(p, piOut, aiAppend[i], pRc);
  }
};

static int ckptExportSnapshot( 

Changes to ext/lsm1/lsm_file.c.

265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
...
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
...
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
...
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
...
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
...
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
....
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
....
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
....
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
....
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
....
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
....
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
....
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
....
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
....
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
....
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
....
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
....
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
....
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
....
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
....
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
....
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
....
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
....
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
....
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
....
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
....
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
....
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
....
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
....
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
2946
2947
2948
2949
2950
2951
2952
2953
....
2966
2967
2968
2969
2970
2971
2972
2973
2974
2975
2976
2977
2978
2979
2980
....
3031
3032
3033
3034
3035
3036
3037
3038
3039
3040
3041
3042
3043
3044
3045
....
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
....
3293
3294
3295
3296
3297
3298
3299
3300
3301
3302
3303
3304
3305
3306
3307
**   The lsmFsSortedAppend() function sets the pSeg pointer to point to the
**   segment that the new page will be a part of. It is unset by
**   lsmFsPagePersist() after the page is written to disk.
*/
struct Page {
  u8 *aData;                      /* Buffer containing page data */
  int nData;                      /* Bytes of usable data at aData[] */
  Pgno iPg;                       /* Page number */
  int nRef;                       /* Number of outstanding references */
  int flags;                      /* Combination of PAGE_XXX flags */
  Page *pHashNext;                /* Next page in hash table slot */
  Page *pLruNext;                 /* Next page in LRU list */
  Page *pLruPrev;                 /* Previous page in LRU list */
  FileSystem *pFS;                /* File system that owns this page */

................................................................................
#else
# define IOERR_WRAPPER(rc) (rc)
#endif

#ifdef NDEBUG
# define assert_lists_are_ok(x)
#else
static Page *fsPageFindInHash(FileSystem *pFS, Pgno iPg, int *piHash);

static void assert_lists_are_ok(FileSystem *pFS){
#if 0
  Page *p;

  assert( pFS->nMapLimit>=0 );

................................................................................
  return LSM_OK;
}

/*
** Return true if page iReal of the database should be accessed using mmap.
** False otherwise.
*/
static int fsMmapPage(FileSystem *pFS, Pgno iReal){
  return ((i64)iReal*pFS->nPagesize <= pFS->nMapLimit);
}

/*
** Given that there are currently nHash slots in the hash table, return 
** the hash key for file iFile, page iPg.
*/
static int fsHashKey(int nHash, Pgno iPg){
  return (iPg % nHash);
}

/*
** This is a helper function for lsmFsOpen(). It opens a single file on
** disk (either the database or log file).
*/
................................................................................
** Return the page number of the first page on block iBlock. Blocks are
** numbered starting from 1.
**
** For a compressed database, page numbers are byte offsets. The first
** page on each block is the byte offset immediately following the 4-byte
** "previous block" pointer at the start of each block.
*/
static Pgno fsFirstPageOnBlock(FileSystem *pFS, int iBlock){
  Pgno iPg;
  if( pFS->pCompress ){
    if( iBlock==1 ){
      iPg = pFS->nMetasize * 2 + 4;
    }else{
      iPg = pFS->nBlocksize * (Pgno)(iBlock-1) + 4;
    }
  }else{
    const int nPagePerBlock = (pFS->nBlocksize / pFS->nPagesize);
    if( iBlock==1 ){
      iPg = 1 + ((pFS->nMetasize*2 + pFS->nPagesize - 1) / pFS->nPagesize);
    }else{
      iPg = 1 + (iBlock-1) * nPagePerBlock;
................................................................................
** Return the page number of the last page on block iBlock. Blocks are
** numbered starting from 1.
**
** For a compressed database, page numbers are byte offsets. The first
** page on each block is the byte offset of the byte immediately before 
** the 4-byte "next block" pointer at the end of each block.
*/
static Pgno fsLastPageOnBlock(FileSystem *pFS, int iBlock){
  if( pFS->pCompress ){
    return pFS->nBlocksize * (Pgno)iBlock - 1 - 4;
  }else{
    const int nPagePerBlock = (pFS->nBlocksize / pFS->nPagesize);
    return iBlock * nPagePerBlock;
  }
}

/*
** Return the block number of the block that page iPg is located on. 
** Blocks are numbered starting from 1.
*/
static int fsPageToBlock(FileSystem *pFS, Pgno iPg){
  if( pFS->pCompress ){
    return (int)((iPg / pFS->nBlocksize) + 1);
  }else{
    return (int)(1 + ((iPg-1) / (pFS->nBlocksize / pFS->nPagesize)));
  }
}

/*
** Return true if page iPg is the last page on its block.
**
** This function is only called in non-compressed database mode.
*/
static int fsIsLast(FileSystem *pFS, Pgno iPg){
  const int nPagePerBlock = (pFS->nBlocksize / pFS->nPagesize);
  assert( !pFS->pCompress );
  return ( iPg && (iPg % nPagePerBlock)==0 );
}

/*
** Return true if page iPg is the first page on its block.
**
** This function is only called in non-compressed database mode.
*/
static int fsIsFirst(FileSystem *pFS, Pgno iPg){
  const int nPagePerBlock = (pFS->nBlocksize / pFS->nPagesize);
  assert( !pFS->pCompress );
  return ( (iPg % nPagePerBlock)==1
        || (iPg<nPagePerBlock && iPg==fsFirstPageOnBlock(pFS, 1))
  );
}

................................................................................
  }
  return pPage->aData;
}

/*
** Return the page number of a page.
*/
Pgno lsmFsPageNumber(Page *pPage){
  /* assert( (pPage->flags & PAGE_DIRTY)==0 ); */
  return pPage ? pPage->iPg : 0;
}

/*
** Page pPg is currently part of the LRU list belonging to pFS. Remove
** it from the list. pPg->pLruNext and pPg->pLruPrev are cleared by this
................................................................................
/*
** Search the hash-table for page iPg. If an entry is round, return a pointer
** to it. Otherwise, return NULL.
**
** Either way, if argument piHash is not NULL set *piHash to the hash slot
** number that page iPg would be stored in before returning.
*/
static Page *fsPageFindInHash(FileSystem *pFS, Pgno iPg, int *piHash){
  Page *p;                        /* Return value */
  int iHash = fsHashKey(pFS->nHash, iPg);

  if( piHash ) *piHash = iHash;
  for(p=pFS->apHash[iHash]; p; p=p->pHashNext){
    if( p->iPg==iPg) break;
  }
................................................................................
}

/*
** If page iPg has been redirected according to the redirections in the
** object passed as the second argument, return the destination page to
** which it is redirected. Otherwise, return a copy of iPg.
*/
Pgno lsmFsRedirectPage(FileSystem *pFS, Redirect *pRedir, Pgno iPg){
  Pgno iReal = iPg;

  if( pRedir ){
    const int nPagePerBlock = (
        pFS->pCompress ? pFS->nBlocksize : (pFS->nBlocksize / pFS->nPagesize)
    );
    int iBlk = fsPageToBlock(pFS, iPg);
    int i;
    for(i=0; i<pRedir->n; i++){
      int iFrom = pRedir->a[i].iFrom;
      if( iFrom>iBlk ) break;
      if( iFrom==iBlk ){
        int iTo = pRedir->a[i].iTo;
        iReal = iPg - (Pgno)(iFrom - iTo) * nPagePerBlock;
        if( iTo==1 ){
          iReal += (fsFirstPageOnBlock(pFS, 1)-1);
        }
        break;
      }
    }
  }

  assert( iReal!=0 );
  return iReal;
}

/* Required by the circular fsBlockNext<->fsPageGet dependency. */
static int fsPageGet(FileSystem *, Segment *, Pgno, int, Page **, int *);

/*
** Parameter iBlock is a database file block. This function reads the value 
** stored in the blocks "next block" pointer and stores it in *piNext.
** LSM_OK is returned if everything is successful, or an LSM error code
** otherwise.
*/
................................................................................
  }
  return rc;
}

/*
** Return the page number of the last page on the same block as page iPg.
*/
Pgno fsLastPageOnPagesBlock(FileSystem *pFS, Pgno iPg){
  return fsLastPageOnBlock(pFS, fsPageToBlock(pFS, iPg));
}

/*
** Read nData bytes of data from offset iOff of the database file into
** buffer aData. If this means reading past the end of a block, follow
** the block pointer to the next block and continue reading.
................................................................................
** to the total number of free bytes before returning.
**
** If no error occurs, LSM_OK is returned. Otherwise, an lsm error code.
*/
static int fsPageGet(
  FileSystem *pFS,                /* File-system handle */
  Segment *pSeg,                  /* Block redirection to use (or NULL) */
  Pgno iPg,                       /* Page id */
  int noContent,                  /* True to not load content from disk */
  Page **ppPg,                    /* OUT: New page handle */
  int *pnSpace                    /* OUT: Bytes of free space */
){
  Page *p;
  int iHash;
  int rc = LSM_OK;

  /* In most cases iReal is the same as iPg. Except, if pSeg->pRedirect is 
  ** not NULL, and the block containing iPg has been redirected, then iReal
  ** is the page number after redirection.  */
  Pgno iReal = lsmFsRedirectPage(pFS, (pSeg ? pSeg->pRedirect : 0), iPg);

  assert_lists_are_ok(pFS);
  assert( iPg>=fsFirstPageOnBlock(pFS, 1) );
  assert( iReal>=fsFirstPageOnBlock(pFS, 1) );
  *ppPg = 0;

  /* Search the hash-table for the page */
................................................................................
/*
** Return true if the first or last page of segment pRun falls between iFirst
** and iLast, inclusive, and pRun is not equal to pIgnore.
*/
static int fsRunEndsBetween(
  Segment *pRun, 
  Segment *pIgnore, 
  Pgno iFirst, 
  Pgno iLast
){
  return (pRun!=pIgnore && (
        (pRun->iFirst>=iFirst && pRun->iFirst<=iLast)
     || (pRun->iLastPg>=iFirst && pRun->iLastPg<=iLast)
  ));
}

................................................................................
/*
** Return true if level pLevel contains a segment other than pIgnore for
** which the first or last page is between iFirst and iLast, inclusive.
*/
static int fsLevelEndsBetween(
  Level *pLevel, 
  Segment *pIgnore, 
  Pgno iFirst, 
  Pgno iLast
){
  int i;

  if( fsRunEndsBetween(&pLevel->lhs, pIgnore, iFirst, iLast) ){
    return 1;
  }
  for(i=0; i<pLevel->nRight; i++){
................................................................................
static int fsFreeBlock(
  FileSystem *pFS,                /* File system object */
  Snapshot *pSnapshot,            /* Worker snapshot */
  Segment *pIgnore,               /* Ignore this run when searching */
  int iBlk                        /* Block number of block to free */
){
  int rc = LSM_OK;                /* Return code */
  Pgno iFirst;                    /* First page on block iBlk */
  Pgno iLast;                     /* Last page on block iBlk */
  Level *pLevel;                  /* Used to iterate through levels */

  int iIn;                        /* Used to iterate through append points */
  int iOut = 0;                   /* Used to output append points */
  Pgno *aApp = pSnapshot->aiAppend;

  iFirst = fsFirstPageOnBlock(pFS, iBlk);
  iLast = fsLastPageOnBlock(pFS, iBlk);

  /* Check if any other run in the snapshot has a start or end page 
  ** within this block. If there is such a run, return early. */
  for(pLevel=lsmDbSnapshotLevel(pSnapshot); pLevel; pLevel=pLevel->pNext){
................................................................................
}

/*
** aPgno is an array containing nPgno page numbers. Return the smallest page
** number from the array that falls on block iBlk. Or, if none of the pages
** in aPgno[] fall on block iBlk, return 0.
*/
static Pgno firstOnBlock(FileSystem *pFS, int iBlk, Pgno *aPgno, int nPgno){





  Pgno iRet = 0;
  int i;
  for(i=0; i<nPgno; i++){
    Pgno iPg = aPgno[i];
    if( fsPageToBlock(pFS, iPg)==iBlk && (iRet==0 || iPg<iRet) ){
      iRet = iPg;
    }
  }
  return iRet;
}

#ifndef NDEBUG
/*
** Return true if page iPg, which is a part of segment p, lies on
** a redirected block. 
*/
static int fsPageRedirects(FileSystem *pFS, Segment *p, Pgno iPg){
  return (iPg!=0 && iPg!=lsmFsRedirectPage(pFS, p->pRedirect, iPg));
}

/*
** Return true if the second argument is not NULL and any of the first
** last or root pages lie on a redirected block. 
*/
................................................................................
** the segment pRun. This function gobbles from the start of the run to the
** first page that appears in aPgno[] (i.e. so that the aPgno[] entry is
** the new first page of the run).
*/
void lsmFsGobble(
  lsm_db *pDb,
  Segment *pRun, 
  Pgno *aPgno,
  int nPgno
){
  int rc = LSM_OK;
  FileSystem *pFS = pDb->pFS;
  Snapshot *pSnapshot = pDb->pWorker;
  int iBlk;

................................................................................
  assert( nPgno>0 && 0==fsPageRedirects(pFS, pRun, aPgno[0]) );

  iBlk = fsPageToBlock(pFS, pRun->iFirst);
  pRun->nSize += (int)(pRun->iFirst - fsFirstPageOnBlock(pFS, iBlk));

  while( rc==LSM_OK ){
    int iNext = 0;
    Pgno iFirst = firstOnBlock(pFS, iBlk, aPgno, nPgno);
    if( iFirst ){
      pRun->iFirst = iFirst;
      break;
    }
    rc = fsBlockNext(pFS, pRun, iBlk, &iNext);
    if( rc==LSM_OK ) rc = fsFreeBlock(pFS, pSnapshot, pRun, iBlk);
    pRun->nSize -= (int)(
................................................................................
**   *piNext = iPg + nByte;
**
** But take block overflow and redirection into account.
*/
static int fsNextPageOffset(
  FileSystem *pFS,                /* File system object */
  Segment *pSeg,                  /* Segment to move within */
  Pgno iPg,                       /* Offset of current page */
  int nByte,                      /* Size of current page including headers */
  Pgno *piNext                    /* OUT: Offset of next page. Or zero (EOF) */
){
  Pgno iNext;
  int rc;

  assert( pFS->pCompress );

  rc = fsAddOffset(pFS, pSeg, iPg, nByte-1, &iNext);
  if( pSeg && iNext==pSeg->iLastPg ){
    iNext = 0;
................................................................................
** LSM_OK is returned if no error occurs. Otherwise, an lsm error code.
** If any value other than LSM_OK is returned, then the final value of
** *piPrev is undefined.
*/
static int fsGetPageBefore(
  FileSystem *pFS, 
  Segment *pSeg, 
  Pgno iPg, 
  Pgno *piPrev
){
  u8 aSz[3];
  int rc;
  i64 iRead;

  assert( pFS->pCompress );

................................................................................
**
** Page references returned by this function should be released by the 
** caller using lsmFsPageRelease().
*/
int lsmFsDbPageNext(Segment *pRun, Page *pPg, int eDir, Page **ppNext){
  int rc = LSM_OK;
  FileSystem *pFS = pPg->pFS;
  Pgno iPg = pPg->iPg;

  assert( 0==fsSegmentRedirects(pFS, pRun) );
  if( pFS->pCompress ){
    int nSpace = pPg->nCompress + 2*3;

    do {
      if( eDir>0 ){
................................................................................
** already allocated block. If it is possible, the page number of the first
** page to use for the new segment is returned. Otherwise zero.
**
** If argument pLvl is not NULL, then this function will not attempt to
** start the new segment immediately following any segment that is part
** of the right-hand-side of pLvl.
*/
static Pgno findAppendPoint(FileSystem *pFS, Level *pLvl){
  int i;
  Pgno *aiAppend = pFS->pDb->pWorker->aiAppend;
  Pgno iRet = 0;

  for(i=LSM_APPLIST_SZ-1; iRet==0 && i>=0; i--){
    if( (iRet = aiAppend[i]) ){
      if( pLvl ){
        int iBlk = fsPageToBlock(pFS, iRet);
        int j;
        for(j=0; iRet && j<pLvl->nRight; j++){
................................................................................
  Snapshot *pSnapshot,
  Level *pLvl,
  int bDefer,
  Page **ppOut
){
  int rc = LSM_OK;
  Page *pPg = 0;
  Pgno iApp = 0;
  Pgno iNext = 0;
  Segment *p = &pLvl->lhs;
  Pgno iPrev = p->iLastPg;

  *ppOut = 0;
  assert( p->pRedirect==0 );

  if( pFS->pCompress || bDefer ){
    /* In compressed database mode the page is not assigned a page number
    ** or location in the database file at this point. This will be done
................................................................................
    ** Shift this extra block back to the free-block list. 
    **
    ** Otherwise, add the first free page in the last block used by the run
    ** to the lAppend list.
    */
    if( fsLastPageOnPagesBlock(pFS, p->iLastPg)!=p->iLastPg ){
      int i;
      Pgno *aiAppend = pFS->pDb->pWorker->aiAppend;
      for(i=0; i<LSM_APPLIST_SZ; i++){
        if( aiAppend[i]==0 ){
          aiAppend[i] = p->iLastPg+1;
          break;
        }
      }
    }else if( pFS->pCompress==0 ){
................................................................................
}

/*
** Obtain a reference to page number iPg.
**
** Return LSM_OK if successful, or an lsm error code if an error occurs.
*/
int lsmFsDbPageGet(FileSystem *pFS, Segment *pSeg, Pgno iPg, Page **ppPg){
  return fsPageGet(pFS, pSeg, iPg, 0, ppPg, 0);
}

/*
** Obtain a reference to the last page in the segment passed as the 
** second argument.
**
** Return LSM_OK if successful, or an lsm error code if an error occurs.
*/
int lsmFsDbPageLast(FileSystem *pFS, Segment *pSeg, Page **ppPg){
  int rc;
  Pgno iPg = pSeg->iLastPg;
  if( pFS->pCompress ){
    int nSpace;
    iPg++;
    do {
      nSpace = 0;
      rc = fsGetPageBefore(pFS, pSeg, iPg, &iPg);
      if( rc==LSM_OK ){
................................................................................
** number (*piPg) lies on block iFrom, then calculate the equivalent
** page on block iTo and set *piPg to this value before returning.
*/
static void fsMovePage(
  FileSystem *pFS,                /* File system object */
  int iTo,                        /* Destination block */
  int iFrom,                      /* Source block */
  Pgno *piPg                      /* IN/OUT: Page number */
){
  Pgno iPg = *piPg;
  if( iFrom==fsPageToBlock(pFS, iPg) ){
    const int nPagePerBlock = (
        pFS->pCompress ? pFS ->nBlocksize : (pFS->nBlocksize / pFS->nPagesize)
    );
    *piPg = iPg - (Pgno)(iFrom - iTo) * nPagePerBlock;
  }
}

/*
** Copy the contents of block iFrom to block iTo. 
**
** It is safe to assume that there are no outstanding references to pages 
................................................................................
/*
** Append raw data to a segment. Return the database file offset that the
** data is written to (this may be used as the page number if the data
** being appended is a new page record).
**
** This function is only used in compressed database mode.
*/
static Pgno fsAppendData(
  FileSystem *pFS,                /* File-system handle */
  Segment *pSeg,                  /* Segment to append to */
  const u8 *aData,                /* Buffer containing data to write */
  int nData,                      /* Size of buffer aData[] in bytes */
  int *pRc                        /* IN/OUT: Error code */
){
  Pgno iRet = 0;
  int rc = *pRc;
  assert( pFS->pCompress );
  if( rc==LSM_OK ){
    int nRem = 0;
    int nWrite = 0;
    Pgno iLastOnBlock;
    Pgno iApp = pSeg->iLastPg+1;

    /* If this is the first data written into the segment, find an append-point
    ** or allocate a new block.  */
    if( iApp==1 ){
      pSeg->iFirst = iApp = findAppendPoint(pFS, 0);
      if( iApp==0 ){
        int iBlk;
................................................................................
          assert( iApp==(fsPageToBlock(pFS, iApp)*pFS->nBlocksize)-4 );
          lsmPutU32(aPtr, iBlk);
          rc = lsmEnvWrite(pFS->pEnv, pFS->fdDb, iApp, aPtr, sizeof(aPtr));
        }

        /* Set the "prev" pointer on the new block */
        if( rc==LSM_OK ){
          Pgno iWrite;
          lsmPutU32(aPtr, fsPageToBlock(pFS, iApp));
          iWrite = fsFirstPageOnBlock(pFS, iBlk);
          rc = lsmEnvWrite(pFS->pEnv, pFS->fdDb, iWrite-4, aPtr, sizeof(aPtr));
          if( nRem>0 ) iApp = iWrite;
        }
      }else{
        /* The next block is already allocated. */
................................................................................
** LSM_OK is returned if successful, or an lsm error code otherwise. If
** any value other than LSM_OK is returned, then the final value of all
** output variables is undefined.
*/
static int fsAppendPage(
  FileSystem *pFS, 
  Segment *pSeg,
  Pgno *piNew,
  int *piPrev,
  int *piNext
){
  Pgno iPrev = pSeg->iLastPg;
  int rc;
  assert( iPrev!=0 );

  *piPrev = 0;
  *piNext = 0;

  if( fsIsLast(pFS, iPrev) ){
................................................................................
  }
  *pRc = rc;
}

/*
** If there exists a hash-table entry associated with page iPg, remove it.
*/
static void fsRemoveHashEntry(FileSystem *pFS, Pgno iPg){
  Page *p;
  int iHash = fsHashKey(pFS->nHash, iPg);

  for(p=pFS->apHash[iHash]; p && p->iPg!=iPg; p=p->pHashNext);

  if( p ){
    assert( p->nRef==0 || (p->flags & PAGE_FREE)==0 );
................................................................................
int lsmFsSortedPadding(
  FileSystem *pFS, 
  Snapshot *pSnapshot,
  Segment *pSeg
){
  int rc = LSM_OK;
  if( pFS->pCompress && pSeg->iFirst ){
    Pgno iLast2;
    Pgno iLast = pSeg->iLastPg;     /* Current last page of segment */
    int nPad;                       /* Bytes of padding required */
    u8 aSz[3];

    iLast2 = (1 + iLast/pFS->szSector) * pFS->szSector - 1;
    assert( fsPageToBlock(pFS, iLast)==fsPageToBlock(pFS, iLast2) );
    nPad = (int)(iLast2 - iLast);

................................................................................
int lsmFsSectorSize(FileSystem *pFS){
  return pFS->szSector;
}

/*
** Helper function for lsmInfoArrayStructure().
*/
static Segment *startsWith(Segment *pRun, Pgno iFirst){
  return (iFirst==pRun->iFirst) ? pRun : 0;
}

/*
** Return the segment that starts with page iFirst, if any. If no such segment
** can be found, return NULL.
*/
static Segment *findSegment(Snapshot *pWorker, Pgno iFirst){
  Level *pLvl;                    /* Used to iterate through db levels */
  Segment *pSeg = 0;              /* Pointer to segment to return */

  for(pLvl=lsmDbSnapshotLevel(pWorker); pLvl && pSeg==0; pLvl=pLvl->pNext){
    if( 0==(pSeg = startsWith(&pLvl->lhs, iFirst)) ){
      int i;
      for(i=0; i<pLvl->nRight; i++){
................................................................................
** eventually free the string using lsmFree().
**
** If an error occurs, *pzOut is set to NULL and an LSM error code returned.
*/
int lsmInfoArrayStructure(
  lsm_db *pDb, 
  int bBlock,                     /* True for block numbers only */
  Pgno iFirst,
  char **pzOut
){
  int rc = LSM_OK;
  Snapshot *pWorker;              /* Worker snapshot */
  Segment *pArray = 0;            /* Array to report on */
  int bUnlock = 0;

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

int lsmFsSegmentContainsPg(
  FileSystem *pFS, 
  Segment *pSeg, 
  Pgno iPg, 
  int *pbRes
){
  Redirect *pRedir = pSeg->pRedirect;
  int rc = LSM_OK;
  int iBlk;
  int iLastBlk;
  int iPgBlock;                   /* Block containing page iPg */
................................................................................
** This function implements the lsm_info(LSM_INFO_ARRAY_PAGES) request.
** If successful, *pzOut is set to point to a nul-terminated string 
** containing the array structure and LSM_OK is returned. The caller should
** eventually free the string using lsmFree().
**
** If an error occurs, *pzOut is set to NULL and an LSM error code returned.
*/
int lsmInfoArrayPages(lsm_db *pDb, Pgno iFirst, char **pzOut){
  int rc = LSM_OK;
  Snapshot *pWorker;              /* Worker snapshot */
  Segment *pSeg = 0;              /* Array to report on */
  int bUnlock = 0;

  *pzOut = 0;
  if( iFirst==0 ) return LSM_ERROR;
................................................................................
#ifndef NDEBUG
/*
** Return true if pPg happens to be the last page in segment pSeg. Or false
** otherwise. This function is only invoked as part of assert() conditions.
*/
int lsmFsDbPageIsLast(Segment *pSeg, Page *pPg){
  if( pPg->pFS->pCompress ){
    Pgno iNext = 0;
    int rc;
    rc = fsNextPageOffset(pPg->pFS, pSeg, pPg->iPg, pPg->nCompress+6, &iNext);
    return (rc!=LSM_OK || iNext==0);
  }
  return (pPg->iPg==pSeg->iLastPg);
}
#endif







|







 







|







 







|







|







 







|
|




|







 







|

|










|












|










|







 







|







 







|







 







|
|












|













|







 







|







 







|











|







 







|
|







 







|
|







 







|
|




|







 







|
>
>
>
>
>
|


|












|







 







|







 







|







 







|

|

|







 







|
|







 







|







 







|

|
|







 







|
|

|







 







|







 







|











|







 







|

|




|







 







|






|





|
|







 







|







 







|



|







 







|







 







|
|







 







|







|







 







|







 







|







 







|







 







|







265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
...
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
...
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
...
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
...
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
...
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
....
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
....
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
....
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
....
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
....
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
....
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
....
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
....
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
....
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
....
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
....
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
....
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
....
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
....
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
....
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
....
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
....
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
....
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
....
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
....
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
....
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
....
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
....
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
....
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
2946
2947
2948
2949
2950
2951
2952
2953
2954
2955
2956
2957
2958
....
2971
2972
2973
2974
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
....
3036
3037
3038
3039
3040
3041
3042
3043
3044
3045
3046
3047
3048
3049
3050
....
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
....
3298
3299
3300
3301
3302
3303
3304
3305
3306
3307
3308
3309
3310
3311
3312
**   The lsmFsSortedAppend() function sets the pSeg pointer to point to the
**   segment that the new page will be a part of. It is unset by
**   lsmFsPagePersist() after the page is written to disk.
*/
struct Page {
  u8 *aData;                      /* Buffer containing page data */
  int nData;                      /* Bytes of usable data at aData[] */
  LsmPgno iPg;                    /* Page number */
  int nRef;                       /* Number of outstanding references */
  int flags;                      /* Combination of PAGE_XXX flags */
  Page *pHashNext;                /* Next page in hash table slot */
  Page *pLruNext;                 /* Next page in LRU list */
  Page *pLruPrev;                 /* Previous page in LRU list */
  FileSystem *pFS;                /* File system that owns this page */

................................................................................
#else
# define IOERR_WRAPPER(rc) (rc)
#endif

#ifdef NDEBUG
# define assert_lists_are_ok(x)
#else
static Page *fsPageFindInHash(FileSystem *pFS, LsmPgno iPg, int *piHash);

static void assert_lists_are_ok(FileSystem *pFS){
#if 0
  Page *p;

  assert( pFS->nMapLimit>=0 );

................................................................................
  return LSM_OK;
}

/*
** Return true if page iReal of the database should be accessed using mmap.
** False otherwise.
*/
static int fsMmapPage(FileSystem *pFS, LsmPgno iReal){
  return ((i64)iReal*pFS->nPagesize <= pFS->nMapLimit);
}

/*
** Given that there are currently nHash slots in the hash table, return 
** the hash key for file iFile, page iPg.
*/
static int fsHashKey(int nHash, LsmPgno iPg){
  return (iPg % nHash);
}

/*
** This is a helper function for lsmFsOpen(). It opens a single file on
** disk (either the database or log file).
*/
................................................................................
** Return the page number of the first page on block iBlock. Blocks are
** numbered starting from 1.
**
** For a compressed database, page numbers are byte offsets. The first
** page on each block is the byte offset immediately following the 4-byte
** "previous block" pointer at the start of each block.
*/
static LsmPgno fsFirstPageOnBlock(FileSystem *pFS, int iBlock){
  LsmPgno iPg;
  if( pFS->pCompress ){
    if( iBlock==1 ){
      iPg = pFS->nMetasize * 2 + 4;
    }else{
      iPg = pFS->nBlocksize * (LsmPgno)(iBlock-1) + 4;
    }
  }else{
    const int nPagePerBlock = (pFS->nBlocksize / pFS->nPagesize);
    if( iBlock==1 ){
      iPg = 1 + ((pFS->nMetasize*2 + pFS->nPagesize - 1) / pFS->nPagesize);
    }else{
      iPg = 1 + (iBlock-1) * nPagePerBlock;
................................................................................
** Return the page number of the last page on block iBlock. Blocks are
** numbered starting from 1.
**
** For a compressed database, page numbers are byte offsets. The first
** page on each block is the byte offset of the byte immediately before 
** the 4-byte "next block" pointer at the end of each block.
*/
static LsmPgno fsLastPageOnBlock(FileSystem *pFS, int iBlock){
  if( pFS->pCompress ){
    return pFS->nBlocksize * (LsmPgno)iBlock - 1 - 4;
  }else{
    const int nPagePerBlock = (pFS->nBlocksize / pFS->nPagesize);
    return iBlock * nPagePerBlock;
  }
}

/*
** Return the block number of the block that page iPg is located on. 
** Blocks are numbered starting from 1.
*/
static int fsPageToBlock(FileSystem *pFS, LsmPgno iPg){
  if( pFS->pCompress ){
    return (int)((iPg / pFS->nBlocksize) + 1);
  }else{
    return (int)(1 + ((iPg-1) / (pFS->nBlocksize / pFS->nPagesize)));
  }
}

/*
** Return true if page iPg is the last page on its block.
**
** This function is only called in non-compressed database mode.
*/
static int fsIsLast(FileSystem *pFS, LsmPgno iPg){
  const int nPagePerBlock = (pFS->nBlocksize / pFS->nPagesize);
  assert( !pFS->pCompress );
  return ( iPg && (iPg % nPagePerBlock)==0 );
}

/*
** Return true if page iPg is the first page on its block.
**
** This function is only called in non-compressed database mode.
*/
static int fsIsFirst(FileSystem *pFS, LsmPgno iPg){
  const int nPagePerBlock = (pFS->nBlocksize / pFS->nPagesize);
  assert( !pFS->pCompress );
  return ( (iPg % nPagePerBlock)==1
        || (iPg<nPagePerBlock && iPg==fsFirstPageOnBlock(pFS, 1))
  );
}

................................................................................
  }
  return pPage->aData;
}

/*
** Return the page number of a page.
*/
LsmPgno lsmFsPageNumber(Page *pPage){
  /* assert( (pPage->flags & PAGE_DIRTY)==0 ); */
  return pPage ? pPage->iPg : 0;
}

/*
** Page pPg is currently part of the LRU list belonging to pFS. Remove
** it from the list. pPg->pLruNext and pPg->pLruPrev are cleared by this
................................................................................
/*
** Search the hash-table for page iPg. If an entry is round, return a pointer
** to it. Otherwise, return NULL.
**
** Either way, if argument piHash is not NULL set *piHash to the hash slot
** number that page iPg would be stored in before returning.
*/
static Page *fsPageFindInHash(FileSystem *pFS, LsmPgno iPg, int *piHash){
  Page *p;                        /* Return value */
  int iHash = fsHashKey(pFS->nHash, iPg);

  if( piHash ) *piHash = iHash;
  for(p=pFS->apHash[iHash]; p; p=p->pHashNext){
    if( p->iPg==iPg) break;
  }
................................................................................
}

/*
** If page iPg has been redirected according to the redirections in the
** object passed as the second argument, return the destination page to
** which it is redirected. Otherwise, return a copy of iPg.
*/
LsmPgno lsmFsRedirectPage(FileSystem *pFS, Redirect *pRedir, LsmPgno iPg){
  LsmPgno iReal = iPg;

  if( pRedir ){
    const int nPagePerBlock = (
        pFS->pCompress ? pFS->nBlocksize : (pFS->nBlocksize / pFS->nPagesize)
    );
    int iBlk = fsPageToBlock(pFS, iPg);
    int i;
    for(i=0; i<pRedir->n; i++){
      int iFrom = pRedir->a[i].iFrom;
      if( iFrom>iBlk ) break;
      if( iFrom==iBlk ){
        int iTo = pRedir->a[i].iTo;
        iReal = iPg - (LsmPgno)(iFrom - iTo) * nPagePerBlock;
        if( iTo==1 ){
          iReal += (fsFirstPageOnBlock(pFS, 1)-1);
        }
        break;
      }
    }
  }

  assert( iReal!=0 );
  return iReal;
}

/* Required by the circular fsBlockNext<->fsPageGet dependency. */
static int fsPageGet(FileSystem *, Segment *, LsmPgno, int, Page **, int *);

/*
** Parameter iBlock is a database file block. This function reads the value 
** stored in the blocks "next block" pointer and stores it in *piNext.
** LSM_OK is returned if everything is successful, or an LSM error code
** otherwise.
*/
................................................................................
  }
  return rc;
}

/*
** Return the page number of the last page on the same block as page iPg.
*/
LsmPgno fsLastPageOnPagesBlock(FileSystem *pFS, LsmPgno iPg){
  return fsLastPageOnBlock(pFS, fsPageToBlock(pFS, iPg));
}

/*
** Read nData bytes of data from offset iOff of the database file into
** buffer aData. If this means reading past the end of a block, follow
** the block pointer to the next block and continue reading.
................................................................................
** to the total number of free bytes before returning.
**
** If no error occurs, LSM_OK is returned. Otherwise, an lsm error code.
*/
static int fsPageGet(
  FileSystem *pFS,                /* File-system handle */
  Segment *pSeg,                  /* Block redirection to use (or NULL) */
  LsmPgno iPg,                    /* Page id */
  int noContent,                  /* True to not load content from disk */
  Page **ppPg,                    /* OUT: New page handle */
  int *pnSpace                    /* OUT: Bytes of free space */
){
  Page *p;
  int iHash;
  int rc = LSM_OK;

  /* In most cases iReal is the same as iPg. Except, if pSeg->pRedirect is 
  ** not NULL, and the block containing iPg has been redirected, then iReal
  ** is the page number after redirection.  */
  LsmPgno iReal = lsmFsRedirectPage(pFS, (pSeg ? pSeg->pRedirect : 0), iPg);

  assert_lists_are_ok(pFS);
  assert( iPg>=fsFirstPageOnBlock(pFS, 1) );
  assert( iReal>=fsFirstPageOnBlock(pFS, 1) );
  *ppPg = 0;

  /* Search the hash-table for the page */
................................................................................
/*
** Return true if the first or last page of segment pRun falls between iFirst
** and iLast, inclusive, and pRun is not equal to pIgnore.
*/
static int fsRunEndsBetween(
  Segment *pRun, 
  Segment *pIgnore, 
  LsmPgno iFirst, 
  LsmPgno iLast
){
  return (pRun!=pIgnore && (
        (pRun->iFirst>=iFirst && pRun->iFirst<=iLast)
     || (pRun->iLastPg>=iFirst && pRun->iLastPg<=iLast)
  ));
}

................................................................................
/*
** Return true if level pLevel contains a segment other than pIgnore for
** which the first or last page is between iFirst and iLast, inclusive.
*/
static int fsLevelEndsBetween(
  Level *pLevel, 
  Segment *pIgnore, 
  LsmPgno iFirst, 
  LsmPgno iLast
){
  int i;

  if( fsRunEndsBetween(&pLevel->lhs, pIgnore, iFirst, iLast) ){
    return 1;
  }
  for(i=0; i<pLevel->nRight; i++){
................................................................................
static int fsFreeBlock(
  FileSystem *pFS,                /* File system object */
  Snapshot *pSnapshot,            /* Worker snapshot */
  Segment *pIgnore,               /* Ignore this run when searching */
  int iBlk                        /* Block number of block to free */
){
  int rc = LSM_OK;                /* Return code */
  LsmPgno iFirst;                 /* First page on block iBlk */
  LsmPgno iLast;                  /* Last page on block iBlk */
  Level *pLevel;                  /* Used to iterate through levels */

  int iIn;                        /* Used to iterate through append points */
  int iOut = 0;                   /* Used to output append points */
  LsmPgno *aApp = pSnapshot->aiAppend;

  iFirst = fsFirstPageOnBlock(pFS, iBlk);
  iLast = fsLastPageOnBlock(pFS, iBlk);

  /* Check if any other run in the snapshot has a start or end page 
  ** within this block. If there is such a run, return early. */
  for(pLevel=lsmDbSnapshotLevel(pSnapshot); pLevel; pLevel=pLevel->pNext){
................................................................................
}

/*
** aPgno is an array containing nPgno page numbers. Return the smallest page
** number from the array that falls on block iBlk. Or, if none of the pages
** in aPgno[] fall on block iBlk, return 0.
*/
static LsmPgno firstOnBlock(
  FileSystem *pFS, 
  int iBlk, 
  LsmPgno *aPgno, 
  int nPgno
){
  LsmPgno iRet = 0;
  int i;
  for(i=0; i<nPgno; i++){
    LsmPgno iPg = aPgno[i];
    if( fsPageToBlock(pFS, iPg)==iBlk && (iRet==0 || iPg<iRet) ){
      iRet = iPg;
    }
  }
  return iRet;
}

#ifndef NDEBUG
/*
** Return true if page iPg, which is a part of segment p, lies on
** a redirected block. 
*/
static int fsPageRedirects(FileSystem *pFS, Segment *p, LsmPgno iPg){
  return (iPg!=0 && iPg!=lsmFsRedirectPage(pFS, p->pRedirect, iPg));
}

/*
** Return true if the second argument is not NULL and any of the first
** last or root pages lie on a redirected block. 
*/
................................................................................
** the segment pRun. This function gobbles from the start of the run to the
** first page that appears in aPgno[] (i.e. so that the aPgno[] entry is
** the new first page of the run).
*/
void lsmFsGobble(
  lsm_db *pDb,
  Segment *pRun, 
  LsmPgno *aPgno,
  int nPgno
){
  int rc = LSM_OK;
  FileSystem *pFS = pDb->pFS;
  Snapshot *pSnapshot = pDb->pWorker;
  int iBlk;

................................................................................
  assert( nPgno>0 && 0==fsPageRedirects(pFS, pRun, aPgno[0]) );

  iBlk = fsPageToBlock(pFS, pRun->iFirst);
  pRun->nSize += (int)(pRun->iFirst - fsFirstPageOnBlock(pFS, iBlk));

  while( rc==LSM_OK ){
    int iNext = 0;
    LsmPgno iFirst = firstOnBlock(pFS, iBlk, aPgno, nPgno);
    if( iFirst ){
      pRun->iFirst = iFirst;
      break;
    }
    rc = fsBlockNext(pFS, pRun, iBlk, &iNext);
    if( rc==LSM_OK ) rc = fsFreeBlock(pFS, pSnapshot, pRun, iBlk);
    pRun->nSize -= (int)(
................................................................................
**   *piNext = iPg + nByte;
**
** But take block overflow and redirection into account.
*/
static int fsNextPageOffset(
  FileSystem *pFS,                /* File system object */
  Segment *pSeg,                  /* Segment to move within */
  LsmPgno iPg,                    /* Offset of current page */
  int nByte,                      /* Size of current page including headers */
  LsmPgno *piNext                 /* OUT: Offset of next page. Or zero (EOF) */
){
  LsmPgno iNext;
  int rc;

  assert( pFS->pCompress );

  rc = fsAddOffset(pFS, pSeg, iPg, nByte-1, &iNext);
  if( pSeg && iNext==pSeg->iLastPg ){
    iNext = 0;
................................................................................
** LSM_OK is returned if no error occurs. Otherwise, an lsm error code.
** If any value other than LSM_OK is returned, then the final value of
** *piPrev is undefined.
*/
static int fsGetPageBefore(
  FileSystem *pFS, 
  Segment *pSeg, 
  LsmPgno iPg, 
  LsmPgno *piPrev
){
  u8 aSz[3];
  int rc;
  i64 iRead;

  assert( pFS->pCompress );

................................................................................
**
** Page references returned by this function should be released by the 
** caller using lsmFsPageRelease().
*/
int lsmFsDbPageNext(Segment *pRun, Page *pPg, int eDir, Page **ppNext){
  int rc = LSM_OK;
  FileSystem *pFS = pPg->pFS;
  LsmPgno iPg = pPg->iPg;

  assert( 0==fsSegmentRedirects(pFS, pRun) );
  if( pFS->pCompress ){
    int nSpace = pPg->nCompress + 2*3;

    do {
      if( eDir>0 ){
................................................................................
** already allocated block. If it is possible, the page number of the first
** page to use for the new segment is returned. Otherwise zero.
**
** If argument pLvl is not NULL, then this function will not attempt to
** start the new segment immediately following any segment that is part
** of the right-hand-side of pLvl.
*/
static LsmPgno findAppendPoint(FileSystem *pFS, Level *pLvl){
  int i;
  LsmPgno *aiAppend = pFS->pDb->pWorker->aiAppend;
  LsmPgno iRet = 0;

  for(i=LSM_APPLIST_SZ-1; iRet==0 && i>=0; i--){
    if( (iRet = aiAppend[i]) ){
      if( pLvl ){
        int iBlk = fsPageToBlock(pFS, iRet);
        int j;
        for(j=0; iRet && j<pLvl->nRight; j++){
................................................................................
  Snapshot *pSnapshot,
  Level *pLvl,
  int bDefer,
  Page **ppOut
){
  int rc = LSM_OK;
  Page *pPg = 0;
  LsmPgno iApp = 0;
  LsmPgno iNext = 0;
  Segment *p = &pLvl->lhs;
  LsmPgno iPrev = p->iLastPg;

  *ppOut = 0;
  assert( p->pRedirect==0 );

  if( pFS->pCompress || bDefer ){
    /* In compressed database mode the page is not assigned a page number
    ** or location in the database file at this point. This will be done
................................................................................
    ** Shift this extra block back to the free-block list. 
    **
    ** Otherwise, add the first free page in the last block used by the run
    ** to the lAppend list.
    */
    if( fsLastPageOnPagesBlock(pFS, p->iLastPg)!=p->iLastPg ){
      int i;
      LsmPgno *aiAppend = pFS->pDb->pWorker->aiAppend;
      for(i=0; i<LSM_APPLIST_SZ; i++){
        if( aiAppend[i]==0 ){
          aiAppend[i] = p->iLastPg+1;
          break;
        }
      }
    }else if( pFS->pCompress==0 ){
................................................................................
}

/*
** Obtain a reference to page number iPg.
**
** Return LSM_OK if successful, or an lsm error code if an error occurs.
*/
int lsmFsDbPageGet(FileSystem *pFS, Segment *pSeg, LsmPgno iPg, Page **ppPg){
  return fsPageGet(pFS, pSeg, iPg, 0, ppPg, 0);
}

/*
** Obtain a reference to the last page in the segment passed as the 
** second argument.
**
** Return LSM_OK if successful, or an lsm error code if an error occurs.
*/
int lsmFsDbPageLast(FileSystem *pFS, Segment *pSeg, Page **ppPg){
  int rc;
  LsmPgno iPg = pSeg->iLastPg;
  if( pFS->pCompress ){
    int nSpace;
    iPg++;
    do {
      nSpace = 0;
      rc = fsGetPageBefore(pFS, pSeg, iPg, &iPg);
      if( rc==LSM_OK ){
................................................................................
** number (*piPg) lies on block iFrom, then calculate the equivalent
** page on block iTo and set *piPg to this value before returning.
*/
static void fsMovePage(
  FileSystem *pFS,                /* File system object */
  int iTo,                        /* Destination block */
  int iFrom,                      /* Source block */
  LsmPgno *piPg                   /* IN/OUT: Page number */
){
  LsmPgno iPg = *piPg;
  if( iFrom==fsPageToBlock(pFS, iPg) ){
    const int nPagePerBlock = (
        pFS->pCompress ? pFS ->nBlocksize : (pFS->nBlocksize / pFS->nPagesize)
    );
    *piPg = iPg - (LsmPgno)(iFrom - iTo) * nPagePerBlock;
  }
}

/*
** Copy the contents of block iFrom to block iTo. 
**
** It is safe to assume that there are no outstanding references to pages 
................................................................................
/*
** Append raw data to a segment. Return the database file offset that the
** data is written to (this may be used as the page number if the data
** being appended is a new page record).
**
** This function is only used in compressed database mode.
*/
static LsmPgno fsAppendData(
  FileSystem *pFS,                /* File-system handle */
  Segment *pSeg,                  /* Segment to append to */
  const u8 *aData,                /* Buffer containing data to write */
  int nData,                      /* Size of buffer aData[] in bytes */
  int *pRc                        /* IN/OUT: Error code */
){
  LsmPgno iRet = 0;
  int rc = *pRc;
  assert( pFS->pCompress );
  if( rc==LSM_OK ){
    int nRem = 0;
    int nWrite = 0;
    LsmPgno iLastOnBlock;
    LsmPgno iApp = pSeg->iLastPg+1;

    /* If this is the first data written into the segment, find an append-point
    ** or allocate a new block.  */
    if( iApp==1 ){
      pSeg->iFirst = iApp = findAppendPoint(pFS, 0);
      if( iApp==0 ){
        int iBlk;
................................................................................
          assert( iApp==(fsPageToBlock(pFS, iApp)*pFS->nBlocksize)-4 );
          lsmPutU32(aPtr, iBlk);
          rc = lsmEnvWrite(pFS->pEnv, pFS->fdDb, iApp, aPtr, sizeof(aPtr));
        }

        /* Set the "prev" pointer on the new block */
        if( rc==LSM_OK ){
          LsmPgno iWrite;
          lsmPutU32(aPtr, fsPageToBlock(pFS, iApp));
          iWrite = fsFirstPageOnBlock(pFS, iBlk);
          rc = lsmEnvWrite(pFS->pEnv, pFS->fdDb, iWrite-4, aPtr, sizeof(aPtr));
          if( nRem>0 ) iApp = iWrite;
        }
      }else{
        /* The next block is already allocated. */
................................................................................
** LSM_OK is returned if successful, or an lsm error code otherwise. If
** any value other than LSM_OK is returned, then the final value of all
** output variables is undefined.
*/
static int fsAppendPage(
  FileSystem *pFS, 
  Segment *pSeg,
  LsmPgno *piNew,
  int *piPrev,
  int *piNext
){
  LsmPgno iPrev = pSeg->iLastPg;
  int rc;
  assert( iPrev!=0 );

  *piPrev = 0;
  *piNext = 0;

  if( fsIsLast(pFS, iPrev) ){
................................................................................
  }
  *pRc = rc;
}

/*
** If there exists a hash-table entry associated with page iPg, remove it.
*/
static void fsRemoveHashEntry(FileSystem *pFS, LsmPgno iPg){
  Page *p;
  int iHash = fsHashKey(pFS->nHash, iPg);

  for(p=pFS->apHash[iHash]; p && p->iPg!=iPg; p=p->pHashNext);

  if( p ){
    assert( p->nRef==0 || (p->flags & PAGE_FREE)==0 );
................................................................................
int lsmFsSortedPadding(
  FileSystem *pFS, 
  Snapshot *pSnapshot,
  Segment *pSeg
){
  int rc = LSM_OK;
  if( pFS->pCompress && pSeg->iFirst ){
    LsmPgno iLast2;
    LsmPgno iLast = pSeg->iLastPg;  /* Current last page of segment */
    int nPad;                       /* Bytes of padding required */
    u8 aSz[3];

    iLast2 = (1 + iLast/pFS->szSector) * pFS->szSector - 1;
    assert( fsPageToBlock(pFS, iLast)==fsPageToBlock(pFS, iLast2) );
    nPad = (int)(iLast2 - iLast);

................................................................................
int lsmFsSectorSize(FileSystem *pFS){
  return pFS->szSector;
}

/*
** Helper function for lsmInfoArrayStructure().
*/
static Segment *startsWith(Segment *pRun, LsmPgno iFirst){
  return (iFirst==pRun->iFirst) ? pRun : 0;
}

/*
** Return the segment that starts with page iFirst, if any. If no such segment
** can be found, return NULL.
*/
static Segment *findSegment(Snapshot *pWorker, LsmPgno iFirst){
  Level *pLvl;                    /* Used to iterate through db levels */
  Segment *pSeg = 0;              /* Pointer to segment to return */

  for(pLvl=lsmDbSnapshotLevel(pWorker); pLvl && pSeg==0; pLvl=pLvl->pNext){
    if( 0==(pSeg = startsWith(&pLvl->lhs, iFirst)) ){
      int i;
      for(i=0; i<pLvl->nRight; i++){
................................................................................
** eventually free the string using lsmFree().
**
** If an error occurs, *pzOut is set to NULL and an LSM error code returned.
*/
int lsmInfoArrayStructure(
  lsm_db *pDb, 
  int bBlock,                     /* True for block numbers only */
  LsmPgno iFirst,
  char **pzOut
){
  int rc = LSM_OK;
  Snapshot *pWorker;              /* Worker snapshot */
  Segment *pArray = 0;            /* Array to report on */
  int bUnlock = 0;

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

int lsmFsSegmentContainsPg(
  FileSystem *pFS, 
  Segment *pSeg, 
  LsmPgno iPg, 
  int *pbRes
){
  Redirect *pRedir = pSeg->pRedirect;
  int rc = LSM_OK;
  int iBlk;
  int iLastBlk;
  int iPgBlock;                   /* Block containing page iPg */
................................................................................
** This function implements the lsm_info(LSM_INFO_ARRAY_PAGES) request.
** If successful, *pzOut is set to point to a nul-terminated string 
** containing the array structure and LSM_OK is returned. The caller should
** eventually free the string using lsmFree().
**
** If an error occurs, *pzOut is set to NULL and an LSM error code returned.
*/
int lsmInfoArrayPages(lsm_db *pDb, LsmPgno iFirst, char **pzOut){
  int rc = LSM_OK;
  Snapshot *pWorker;              /* Worker snapshot */
  Segment *pSeg = 0;              /* Array to report on */
  int bUnlock = 0;

  *pzOut = 0;
  if( iFirst==0 ) return LSM_ERROR;
................................................................................
#ifndef NDEBUG
/*
** Return true if pPg happens to be the last page in segment pSeg. Or false
** otherwise. This function is only invoked as part of assert() conditions.
*/
int lsmFsDbPageIsLast(Segment *pSeg, Page *pPg){
  if( pPg->pFS->pCompress ){
    LsmPgno iNext = 0;
    int rc;
    rc = fsNextPageOffset(pPg->pFS, pSeg, pPg->iPg, pPg->nCompress+6, &iNext);
    return (rc!=LSM_OK || iNext==0);
  }
  return (pPg->iPg==pSeg->iLastPg);
}
#endif

Changes to ext/lsm1/lsm_main.c.

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
    case LSM_INFO_DB_STRUCTURE: {
      char **pzVal = va_arg(ap, char **);
      rc = lsmStructList(pDb, pzVal);
      break;
    }

    case LSM_INFO_ARRAY_STRUCTURE: {
      Pgno pgno = va_arg(ap, Pgno);
      char **pzVal = va_arg(ap, char **);
      rc = lsmInfoArrayStructure(pDb, 0, pgno, pzVal);
      break;
    }

    case LSM_INFO_ARRAY_PAGES: {
      Pgno pgno = va_arg(ap, Pgno);
      char **pzVal = va_arg(ap, char **);
      rc = lsmInfoArrayPages(pDb, pgno, pzVal);
      break;
    }

    case LSM_INFO_PAGE_HEX_DUMP:
    case LSM_INFO_PAGE_ASCII_DUMP: {
      Pgno pgno = va_arg(ap, Pgno);
      char **pzVal = va_arg(ap, char **);
      int bUnlock = 0;
      rc = infoGetWorker(pDb, 0, &bUnlock);
      if( rc==LSM_OK ){
        int bHex = (eParam==LSM_INFO_PAGE_HEX_DUMP);
        rc = lsmInfoPageDump(pDb, pgno, bHex, pzVal);
      }







|






|







|







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
    case LSM_INFO_DB_STRUCTURE: {
      char **pzVal = va_arg(ap, char **);
      rc = lsmStructList(pDb, pzVal);
      break;
    }

    case LSM_INFO_ARRAY_STRUCTURE: {
      LsmPgno pgno = va_arg(ap, LsmPgno);
      char **pzVal = va_arg(ap, char **);
      rc = lsmInfoArrayStructure(pDb, 0, pgno, pzVal);
      break;
    }

    case LSM_INFO_ARRAY_PAGES: {
      LsmPgno pgno = va_arg(ap, LsmPgno);
      char **pzVal = va_arg(ap, char **);
      rc = lsmInfoArrayPages(pDb, pgno, pzVal);
      break;
    }

    case LSM_INFO_PAGE_HEX_DUMP:
    case LSM_INFO_PAGE_ASCII_DUMP: {
      LsmPgno pgno = va_arg(ap, LsmPgno);
      char **pzVal = va_arg(ap, char **);
      int bUnlock = 0;
      rc = infoGetWorker(pDb, 0, &bUnlock);
      if( rc==LSM_OK ){
        int bHex = (eParam==LSM_INFO_PAGE_HEX_DUMP);
        rc = lsmInfoPageDump(pDb, pgno, bHex, pzVal);
      }

Changes to ext/lsm1/lsm_sorted.c.

100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
...
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
...
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
...
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
...
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
...
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
...
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
...
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
...
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
...
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
...
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
...
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
...
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
...
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
...
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
...
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
...
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
...
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
...
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
...
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
...
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
....
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
....
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
....
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
....
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
....
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
....
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
....
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
....
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
....
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
....
3051
3052
3053
3054
3055
3056
3057
3058
3059
3060
3061
3062
3063
3064
3065
....
3533
3534
3535
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
....
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
3577
3578
3579
3580
3581
3582
....
3678
3679
3680
3681
3682
3683
3684
3685
3686
3687
3688
3689
3690
3691
3692
....
3699
3700
3701
3702
3703
3704
3705
3706
3707
3708
3709
3710
3711
3712
3713
....
3730
3731
3732
3733
3734
3735
3736
3737
3738
3739
3740
3741
3742
3743
3744
....
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
....
4214
4215
4216
4217
4218
4219
4220
4221
4222
4223
4224
4225
4226
4227
4228
....
4367
4368
4369
4370
4371
4372
4373
4374
4375
4376
4377
4378
4379
4380
4381
....
4544
4545
4546
4547
4548
4549
4550
4551
4552
4553
4554
4555
4556
4557
4558
....
4626
4627
4628
4629
4630
4631
4632
4633
4634
4635
4636
4637
4638
4639
4640
4641
4642
4643
4644
4645
4646
4647
4648
4649
....
5462
5463
5464
5465
5466
5467
5468
5469
5470
5471
5472
5473
5474
5475
5476
5477
5478
....
5523
5524
5525
5526
5527
5528
5529
5530
5531
5532
5533
5534
5535
5536
5537
....
5559
5560
5561
5562
5563
5564
5565
5566
5567
5568
5569
5570
5571
5572
5573
....
5603
5604
5605
5606
5607
5608
5609
5610
5611
5612
5613
5614
5615
5616
5617
....
5621
5622
5623
5624
5625
5626
5627
5628
5629
5630
5631
5632
5633
5634
5635
....
5667
5668
5669
5670
5671
5672
5673
5674
5675
5676
5677
5678
5679
5680
5681
....
5708
5709
5710
5711
5712
5713
5714
5715
5716
5717
5718
5719
5720
5721
5722
....
5743
5744
5745
5746
5747
5748
5749
5750
5751
5752
5753
5754
5755
5756
5757
....
5809
5810
5811
5812
5813
5814
5815
5816
5817
5818
5819
5820
5821
5822
5823
....
5985
5986
5987
5988
5989
5990
5991
5992
5993
5994
5995
5996
5997
5998
5999
6000
....
6048
6049
6050
6051
6052
6053
6054
6055
6056
6057
6058
6059
6060
6061
6062
....
6071
6072
6073
6074
6075
6076
6077
6078
6079
6080
6081
6082
6083
6084
6085
....
6133
6134
6135
6136
6137
6138
6139
6140
6141
6142
6143
6144
6145
6146
6147


#ifndef LSM_SEGMENTPTR_FREE_THRESHOLD
# define LSM_SEGMENTPTR_FREE_THRESHOLD 1024
#endif

typedef struct SegmentPtr SegmentPtr;
typedef struct Blob Blob;

struct Blob {
  lsm_env *pEnv;
  void *pData;
  int nData;
  int nAlloc;
};

/*
................................................................................
  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 */
  Pgno iPtr;                    /* Base cascade pointer */

  /* Current cell. See segmentPtrLoadCell() */
  int iCell;                    /* Current record within page pPg */
  int eType;                    /* Type of current record */
  Pgno iPgPtr;                  /* Cascade pointer offset */
  void *pKey; int nKey;         /* Key associated with current record */
  void *pVal; int nVal;         /* Current record value (eType==WRITE only) */

  /* Blobs used to allocate buffers for pKey and pVal as required */
  Blob blob1;
  Blob blob2;
};

/*
** Used to iterate through the keys stored in a b-tree hierarchy from start
** to finish. Only First() and Next() operations are required.
**
**   btreeCursorNew()
................................................................................
  int iPg;                        /* Current entry in aPg[]. -1 -> EOF. */
  BtreePg *aPg;                   /* Pages from root to current location */

  /* Cache of current entry. pKey==0 for EOF. */
  void *pKey;
  int nKey;
  int eType;
  Pgno iPtr;

  /* Storage for key, if not local */
  Blob blob;
};


/*
** A cursor used for merged searches or iterations through up to one
** Tree structure and any number of sorted files.
**
................................................................................
*/
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 */
  int iFree;                      /* Next element of free-list (-ve for eof) */
  SegmentPtr *aPtr;               /* Array of segment pointers */
  int nPtr;                       /* Size of array aPtr[] */
  BtreeCursor *pBtCsr;            /* b-tree cursor (db writes only) */
................................................................................
  int nTree;                      /* Size of aTree[] array */
  int *aTree;                     /* Array of comparison results */

  /* Used by cursors flushing the in-memory tree only */
  void *pSystemVal;               /* Pointer to buffer to free */

  /* Used by worker cursors only */
  Pgno *pPrevMergePtr;
};

/*
** The following constants are used to assign integers to each component
** cursor of a multi-cursor.
*/
#define CURSOR_DATA_TREE0     0   /* Current tree cursor (apTreeCsr[0]) */
................................................................................
  lsm_db *pDb;                    /* Database handle */
  Level *pLevel;                  /* Worker snapshot Level being merged */
  MultiCursor *pCsr;              /* Cursor to read new segment contents from */
  int bFlush;                     /* True if this is an in-memory tree flush */
  Hierarchy hier;                 /* B-tree hierarchy under construction */
  Page *pPage;                    /* Current output page */
  int nWork;                      /* Number of calls to mergeWorkerNextPage() */
  Pgno *aGobble;                  /* Gobble point for each input segment */

  Pgno iIndirect;
  struct SavedPgno {
    Pgno iPgno;
    int bStore;
  } aSave[2];
};

#ifdef LSM_DEBUG_EXPENSIVE
static int assertPointersOk(lsm_db *, Segment *, Segment *, int);
static int assertBtreeOk(lsm_db *, Segment *);
................................................................................
  aOut[3] = (u8)((nVal>>32) & 0xFF);
  aOut[4] = (u8)((nVal>>24) & 0xFF);
  aOut[5] = (u8)((nVal>>16) & 0xFF);
  aOut[6] = (u8)((nVal>> 8) & 0xFF);
  aOut[7] = (u8)((nVal    ) & 0xFF);
}

static int sortedBlobGrow(lsm_env *pEnv, Blob *pBlob, int nData){
  assert( pBlob->pEnv==pEnv || (pBlob->pEnv==0 && pBlob->pData==0) );
  if( pBlob->nAlloc<nData ){
    pBlob->pData = lsmReallocOrFree(pEnv, pBlob->pData, nData);
    if( !pBlob->pData ) return LSM_NOMEM_BKPT;
    pBlob->nAlloc = nData;
    pBlob->pEnv = pEnv;
  }
  return LSM_OK;
}

static int sortedBlobSet(lsm_env *pEnv, Blob *pBlob, void *pData, int nData){
  if( sortedBlobGrow(pEnv, pBlob, nData) ) return LSM_NOMEM;
  memcpy(pBlob->pData, pData, nData);
  pBlob->nData = nData;
  return LSM_OK;
}

#if 0
static int sortedBlobCopy(Blob *pDest, Blob *pSrc){
  return sortedBlobSet(pDest, pSrc->pData, pSrc->nData);
}
#endif

static void sortedBlobFree(Blob *pBlob){
  assert( pBlob->pEnv || pBlob->pData==0 );
  if( pBlob->pData ) lsmFree(pBlob->pEnv, pBlob->pData);
  memset(pBlob, 0, sizeof(Blob));
}

static int sortedReadData(
  Segment *pSeg,
  Page *pPg,
  int iOff,
  int nByte,
  void **ppData,
  Blob *pBlob
){
  int rc = LSM_OK;
  int iEnd;
  int nData;
  int nCell;
  u8 *aData;

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

static int pageGetNRec(u8 *aData, int nData){
  return (int)lsmGetU16(&aData[SEGMENT_NRECORD_OFFSET(nData)]);
}

static Pgno pageGetPtr(u8 *aData, int nData){
  return (Pgno)lsmGetU64(&aData[SEGMENT_POINTER_OFFSET(nData)]);
}

static int pageGetFlags(u8 *aData, int nData){
  return (int)lsmGetU16(&aData[SEGMENT_FLAGS_OFFSET(nData)]);
}

static u8 *pageGetCell(u8 *aData, int nData, int iCell){
................................................................................
  return pageGetNRec(aData, nData);
}

/*
** Return the decoded (possibly relative) pointer value stored in cell 
** iCell from page aData/nData.
*/
static Pgno pageGetRecordPtr(u8 *aData, int nData, int iCell){
  Pgno iRet;                      /* Return value */
  u8 *aCell;                      /* Pointer to cell iCell */

  assert( iCell<pageGetNRec(aData, nData) && iCell>=0 );
  aCell = pageGetCell(aData, nData, iCell);
  lsmVarintGet64(&aCell[1], &iRet);
  return iRet;
}
................................................................................

static u8 *pageGetKey(
  Segment *pSeg,                  /* Segment pPg belongs to */
  Page *pPg,                      /* Page to read from */
  int iCell,                      /* Index of cell on page to read */
  int *piTopic,                   /* OUT: Topic associated with this key */
  int *pnKey,                     /* OUT: Size of key in bytes */
  Blob *pBlob                     /* If required, use this for dynamic memory */
){
  u8 *pKey;
  int nDummy;
  int eType;
  u8 *aData;
  int nData;

................................................................................

static int pageGetKeyCopy(
  lsm_env *pEnv,                  /* Environment handle */
  Segment *pSeg,                  /* Segment pPg belongs to */
  Page *pPg,                      /* Page to read from */
  int iCell,                      /* Index of cell on page to read */
  int *piTopic,                   /* OUT: Topic associated with this key */
  Blob *pBlob                     /* If required, use this for dynamic memory */
){
  int rc = LSM_OK;
  int nKey;
  u8 *aKey;

  aKey = pageGetKey(pSeg, pPg, iCell, piTopic, &nKey, pBlob);
  assert( (void *)aKey!=pBlob->pData || nKey==pBlob->nData );
................................................................................
  if( (void *)aKey!=pBlob->pData ){
    rc = sortedBlobSet(pEnv, pBlob, aKey, nKey);
  }

  return rc;
}

static Pgno pageGetBtreeRef(Page *pPg, int iKey){
  Pgno iRef;
  u8 *aData;
  int nData;
  u8 *aCell;

  aData = fsPageData(pPg, &nData);
  aCell = pageGetCell(aData, nData, iKey);
  assert( aCell[0]==0 );
................................................................................
#define GETVARINT64(a, i) (((i)=((u8*)(a))[0])<=240?1:lsmVarintGet64((a), &(i)))
#define GETVARINT32(a, i) (((i)=((u8*)(a))[0])<=240?1:lsmVarintGet32((a), &(i)))

static int pageGetBtreeKey(
  Segment *pSeg,                  /* Segment page pPg belongs to */
  Page *pPg,
  int iKey, 
  Pgno *piPtr, 
  int *piTopic, 
  void **ppKey,
  int *pnKey,
  Blob *pBlob
){
  u8 *aData;
  int nData;
  u8 *aCell;
  int eType;

  aData = fsPageData(pPg, &nData);
................................................................................

  aCell = pageGetCell(aData, nData, iKey);
  eType = *aCell++;
  aCell += GETVARINT64(aCell, *piPtr);

  if( eType==0 ){
    int rc;
    Pgno iRef;                  /* Page number of referenced page */
    Page *pRef;
    aCell += GETVARINT64(aCell, iRef);
    rc = lsmFsDbPageGet(lsmPageFS(pPg), pSeg, iRef, &pRef);
    if( rc!=LSM_OK ) return rc;
    pageGetKeyCopy(lsmPageEnv(pPg), pSeg, pRef, 0, &eType, pBlob);
    lsmFsPageRelease(pRef);
    *ppKey = pBlob->pData;
................................................................................
static int btreeCursorLoadKey(BtreeCursor *pCsr){
  int rc = LSM_OK;
  if( pCsr->iPg<0 ){
    pCsr->pKey = 0;
    pCsr->nKey = 0;
    pCsr->eType = 0;
  }else{
    Pgno dummy;
    int iPg = pCsr->iPg;
    int iCell = pCsr->aPg[iPg].iCell;
    while( iCell<0 && (--iPg)>=0 ){
      iCell = pCsr->aPg[iPg].iCell-1;
    }
    if( iPg<0 || iCell<0 ) return LSM_CORRUPT_BKPT;

................................................................................
  assert( pCsr->iPg==pCsr->nDepth-1 );

  aData = fsPageData(pPg->pPage, &nData);
  nCell = pageGetNRec(aData, nData);
  assert( pPg->iCell<=nCell );
  pPg->iCell++;
  if( pPg->iCell==nCell ){
    Pgno iLoad;

    /* Up to parent. */
    lsmFsPageRelease(pPg->pPage);
    pPg->pPage = 0;
    pCsr->iPg--;
    while( pCsr->iPg>=0 ){
      pPg = &pCsr->aPg[pCsr->iPg];
................................................................................
  MergeInput *p
){
  int rc = LSM_OK;

  if( p->iPg ){
    lsm_env *pEnv = lsmFsEnv(pCsr->pFS);
    int iCell;                    /* Current cell number on leaf page */
    Pgno iLeaf;                   /* Page number of current leaf page */
    int nDepth;                   /* Depth of b-tree structure */
    Segment *pSeg = pCsr->pSeg;

    /* Decode the MergeInput structure */
    iLeaf = p->iPg;
    nDepth = (p->iCell & 0x00FF);
    iCell = (p->iCell >> 8) - 1;
................................................................................
      pCsr->nDepth = nDepth;
      pCsr->aPg[pCsr->iPg].iCell = iCell;
      rc = lsmFsDbPageGet(pCsr->pFS, pSeg, iLeaf, pp);
    }

    /* Populate any other aPg[] array entries */
    if( rc==LSM_OK && nDepth>1 ){
      Blob blob = {0,0,0};
      void *pSeek;
      int nSeek;
      int iTopicSeek;
      int iPg = 0;
      int iLoad = (int)pSeg->iRoot;
      Page *pPg = pCsr->aPg[nDepth-1].pPage;
 
................................................................................
        ** In this case, set the iTopicSeek/pSeek/nSeek key to a value
        ** greater than any real key.  */
        assert( iCell==-1 );
        iTopicSeek = 1000;
        pSeek = 0;
        nSeek = 0;
      }else{
        Pgno dummy;
        rc = pageGetBtreeKey(pSeg, pPg,
            0, &dummy, &iTopicSeek, &pSeek, &nSeek, &pCsr->blob
        );
      }

      do {
        Page *pPg2;
................................................................................
          iMax = iCell2-1;
          iMin = 0;

          while( iMax>=iMin ){
            int iTry = (iMin+iMax)/2;
            void *pKey; int nKey;         /* Key for cell iTry */
            int iTopic;                   /* Topic for key pKeyT/nKeyT */
            Pgno iPtr;                    /* Pointer for cell iTry */
            int res;                      /* (pSeek - pKeyT) */

            rc = pageGetBtreeKey(
                pSeg, pPg2, iTry, &iPtr, &iTopic, &pKey, &nKey, &blob
            );
            if( rc!=LSM_OK ) break;

................................................................................
      u8 *aData;
      int nData;

      pBtreePg = &pCsr->aPg[pCsr->iPg];
      aData = fsPageData(pBtreePg->pPage, &nData);
      pCsr->iPtr = btreeCursorPtr(aData, nData, pBtreePg->iCell+1);
      if( pBtreePg->iCell<0 ){
        Pgno dummy;
        int i;
        for(i=pCsr->iPg-1; i>=0; i--){
          if( pCsr->aPg[i].iCell>0 ) break;
        }
        assert( i>=0 );
        rc = pageGetBtreeKey(pSeg,
            pCsr->aPg[i].pPage, pCsr->aPg[i].iCell-1,
................................................................................
}

static int segmentPtrReadData(
  SegmentPtr *pPtr,
  int iOff,
  int nByte,
  void **ppData,
  Blob *pBlob
){
  return sortedReadData(pPtr->pSeg, pPtr->pPg, iOff, nByte, ppData, pBlob);
}

static int segmentPtrNextPage(
  SegmentPtr *pPtr,              /* Load page into this SegmentPtr object */
  int eDir                       /* +1 for next(), -1 for prev() */
................................................................................

  pSeg = sortedSplitkeySegment(pLevel);
  if( rc==LSM_OK ){
    rc = lsmFsDbPageGet(pDb->pFS, pSeg, pMerge->splitkey.iPg, &pPg);
  }
  if( rc==LSM_OK ){
    int iTopic;
    Blob blob = {0, 0, 0, 0};
    u8 *aData;
    int nData;
  
    aData = lsmFsPageData(pPg, &nData);
    if( pageGetFlags(aData, nData) & SEGMENT_BTREE_FLAG ){
      void *pKey;
      int nKey;
      Pgno dummy;
      rc = pageGetBtreeKey(pSeg,
          pPg, pMerge->splitkey.iCell, &dummy, &iTopic, &pKey, &nKey, &blob
      );
      if( rc==LSM_OK && blob.pData!=pKey ){
        rc = sortedBlobSet(pEnv, &blob, pKey, nKey);
      }
    }else{
................................................................................
*/
static int assertKeyLocation(
  MultiCursor *pCsr, 
  SegmentPtr *pPtr, 
  void *pKey, int nKey
){
  lsm_env *pEnv = lsmFsEnv(pCsr->pDb->pFS);
  Blob blob = {0, 0, 0};
  int eDir;
  int iTopic = 0;                 /* TODO: Fix me */

  for(eDir=-1; eDir<=1; eDir+=2){
    Page *pTest = pPtr->pPg;

    lsmFsPageRef(pTest);
................................................................................
  return rc;
}

static int ptrFwdPointer(
  Page *pPage,
  int iCell,
  Segment *pSeg,
  Pgno *piPtr,
  int *pbFound
){
  Page *pPg = pPage;
  int iFirst = iCell;
  int rc = LSM_OK;

  do {
................................................................................
**   much better if the multi-cursor could do this lazily - only seek to the
**   level (N+1) page after the user has moved the cursor on level N passed
**   the big range-delete.
*/
static int segmentPtrFwdPointer(
  MultiCursor *pCsr,              /* Multi-cursor pPtr belongs to */
  SegmentPtr *pPtr,               /* Segment-pointer to extract FC ptr from */
  Pgno *piPtr                     /* OUT: FC pointer value */
){
  Level *pLvl = pPtr->pLevel;
  Level *pNext = pLvl->pNext;
  Page *pPg = pPtr->pPg;
  int rc;
  int bFound;
  Pgno iOut = 0;

  if( pPtr->pSeg==&pLvl->lhs || pPtr->pSeg==&pLvl->aRhs[pLvl->nRight-1] ){
    if( pNext==0 
        || (pNext->nRight==0 && pNext->lhs.iRoot)
        || (pNext->nRight!=0 && pNext->aRhs[0].iRoot)
      ){
      /* Do nothing. The pointer will not be used anyway. */
................................................................................
  int *pbStop
){
  int (*xCmp)(void *, int, void *, int) = pCsr->pDb->xCmp;
  int res = 0;                        /* Result of comparison operation */
  int rc = LSM_OK;
  int iMin;
  int iMax;
  Pgno iPtrOut = 0;

  /* If the current page contains an oversized entry, then there are no
  ** pointers to one or more of the subsequent pages in the sorted run.
  ** The following call ensures that the segment-ptr points to the correct 
  ** page in this case.  */
  rc = segmentPtrSearchOversized(pCsr, pPtr, iTopic, pKey, nKey);
  iPtrOut = pPtr->iPtr;
................................................................................
}

static int seekInBtree(
  MultiCursor *pCsr,              /* Multi-cursor object */
  Segment *pSeg,                  /* Seek within this segment */
  int iTopic,
  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};

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

    rc = lsmFsDbPageGet(pCsr->pDb->pFS, pSeg, iPg, &pPg);
    assert( rc==LSM_OK || pPg==0 );
................................................................................

      iMin = 0;
      iMax = nRec-1;
      while( iMax>=iMin ){
        int iTry = (iMin+iMax)/2;
        void *pKeyT; int nKeyT;       /* Key for cell iTry */
        int iTopicT;                  /* Topic for key pKeyT/nKeyT */
        Pgno iPtr;                    /* Pointer associated with cell iTry */
        int res;                      /* (pKey - pKeyT) */

        rc = pageGetBtreeKey(
            pSeg, pPg, iTry, &iPtr, &iTopicT, &pKeyT, &nKeyT, &blob
        );
        if( rc!=LSM_OK ) break;
        if( piFirst && pKeyT==blob.pData ){
................................................................................
*/
static int seekInLevel(
  MultiCursor *pCsr,              /* Sorted cursor object to seek */
  SegmentPtr *aPtr,               /* Pointer to array of (nRhs+1) SPs */
  int eSeek,                      /* Search bias - see above */
  int iTopic,                     /* Key topic to search for */
  void *pKey, int nKey,           /* Key to search for */
  Pgno *piPgno,                   /* IN/OUT: fraction cascade pointer (or 0) */
  int *pbStop                     /* OUT: See above */
){
  Level *pLvl = aPtr[0].pLevel;   /* Level to seek within */
  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 */
................................................................................
  void *pKey, int nKey, 
  int eSeek
){
  int eESeek = eSeek;             /* Effective eSeek parameter */
  int bStop = 0;                  /* Set to true to halt search operation */
  int rc = LSM_OK;                /* Return code */
  int iPtr = 0;                   /* Used to iterate through pCsr->aPtr[] */
  Pgno iPgno = 0;                 /* FC pointer value */

  assert( pCsr->apTreeCsr[0]==0 || iTopic==0 );
  assert( pCsr->apTreeCsr[1]==0 || iTopic==0 );

  if( eESeek==LSM_SEEK_LEFAST ) eESeek = LSM_SEEK_LE;

  assert( eESeek==LSM_SEEK_EQ || eESeek==LSM_SEEK_LE || eESeek==LSM_SEEK_GE );
................................................................................
** differences are:
**
**   1. The record format is (usually, see below) as follows:
**
**         + Type byte (always SORTED_SEPARATOR or SORTED_SYSTEM_SEPARATOR),
**         + Absolute pointer value (varint),
**         + Number of bytes in key (varint),
**         + Blob containing key data.
**
**   2. All pointer values are stored as absolute values (not offsets 
**      relative to the footer pointer value).
**
**   3. Each pointer that is part of a record points to a page that 
**      contains keys smaller than the records key (note: not "equal to or
**      smaller than - smaller than").
................................................................................
**
** See function seekInBtree() for the code that traverses b-tree pages.
*/

static int mergeWorkerBtreeWrite(
  MergeWorker *pMW,
  u8 eType,
  Pgno iPtr,
  Pgno iKeyPg,
  void *pKey,
  int nKey
){
  Hierarchy *p = &pMW->hier;
  lsm_db *pDb = pMW->pDb;         /* Database handle */
  int rc = LSM_OK;                /* Return Code */
  int iLevel;                     /* Level of b-tree hierachy to write to */
................................................................................

  return rc;
}

static int mergeWorkerBtreeIndirect(MergeWorker *pMW){
  int rc = LSM_OK;
  if( pMW->iIndirect ){
    Pgno iKeyPg = pMW->aSave[1].iPgno;
    rc = mergeWorkerBtreeWrite(pMW, 0, pMW->iIndirect, iKeyPg, 0, 0);
    pMW->iIndirect = 0;
  }
  return rc;
}

/*
................................................................................
static int mergeWorkerPushHierarchy(
  MergeWorker *pMW,               /* Merge worker object */
  int iTopic,                     /* Topic value for this key */
  void *pKey,                     /* Pointer to key buffer */
  int nKey                        /* Size of pKey buffer in bytes */
){
  int rc = LSM_OK;                /* Return Code */
  Pgno iPtr;                      /* Pointer value to accompany pKey/nKey */

  assert( pMW->aSave[0].bStore==0 );
  assert( pMW->aSave[1].bStore==0 );
  rc = mergeWorkerBtreeIndirect(pMW);

  /* Obtain the absolute pointer value to store along with the key in the
  ** page body. This pointer points to a page that contains keys that are
................................................................................
}

static int mergeWorkerFinishHierarchy(
  MergeWorker *pMW                /* Merge worker object */
){
  int i;                          /* Used to loop through apHier[] */
  int rc = LSM_OK;                /* Return code */
  Pgno iPtr;                      /* New right-hand-child pointer value */

  iPtr = pMW->aSave[0].iPgno;
  for(i=0; i<pMW->hier.nHier && rc==LSM_OK; i++){
    Page *pPg = pMW->hier.apHier[i];
    int nData;                    /* Size of aData[] in bytes */
    u8 *aData;                    /* Page data for pPg */

................................................................................
** zero records. The flags field is cleared. The page footer pointer field
** is set to iFPtr.
**
** If successful, LSM_OK is returned. Otherwise, an error code.
*/
static int mergeWorkerNextPage(
  MergeWorker *pMW,               /* Merge worker object to append page to */
  Pgno iFPtr                      /* Pointer value for footer of new page */
){
  int rc = LSM_OK;                /* Return code */
  Page *pNext = 0;                /* New page appended to run */
  lsm_db *pDb = pMW->pDb;         /* Database handle */

  rc = lsmFsSortedAppend(pDb->pFS, pDb->pWorker, pMW->pLevel, 0, &pNext);
  assert( rc || pMW->pLevel->lhs.iFirst>0 || pMW->pDb->compress.xCompress );
................................................................................

static int mergeWorkerStep(MergeWorker *pMW){
  lsm_db *pDb = pMW->pDb;       /* Database handle */
  MultiCursor *pCsr;            /* Cursor to read input data from */
  int rc = LSM_OK;              /* Return code */
  int eType;                    /* SORTED_SEPARATOR, WRITE or DELETE */
  void *pKey; int nKey;         /* Key */
  Pgno iPtr;
  int iVal;

  pCsr = pMW->pCsr;

  /* Pull the next record out of the source cursor. */
  lsmMCursorKey(pCsr, &pKey, &nKey);
  eType = pCsr->eType;
................................................................................
      multiCursorIgnoreDelete(pCsr);
    }
  }

  if( rc!=LSM_OK ){
    lsmMCursorClose(pCsr, 0);
  }else{
    Pgno iLeftPtr = 0;
    Merge merge;                  /* Merge object used to create new level */
    MergeWorker mergeworker;      /* MergeWorker object for the same purpose */

    memset(&merge, 0, sizeof(Merge));
    memset(&mergeworker, 0, sizeof(MergeWorker));

    pNew->pMerge = &merge;
................................................................................
  assert( pDb->pWorker );
  assert( pLevel->pMerge );
  assert( pLevel->nRight>0 );

  memset(pMW, 0, sizeof(MergeWorker));
  pMW->pDb = pDb;
  pMW->pLevel = pLevel;
  pMW->aGobble = lsmMallocZeroRc(pDb->pEnv, sizeof(Pgno) * pLevel->nRight, &rc);

  /* Create a multi-cursor to read the data to write to the new
  ** segment. The new segment contains:
  **
  **   1. Records from LHS of each of the nMerge levels being merged.
  **   2. Separators from either the last level being merged, or the
  **      separators attached to the LHS of the following level, or neither.
................................................................................
  lsm_db *pDb,                    /* Worker connection */
  MultiCursor *pCsr,              /* Multi-cursor being used for a merge */
  int iGobble                     /* pCsr->aPtr[] entry to operate on */
){
  int rc = LSM_OK;
  if( rtTopic(pCsr->eType)==0 ){
    Segment *pSeg = pCsr->aPtr[iGobble].pSeg;
    Pgno *aPg;
    int nPg;

    /* Seek from the root of the b-tree to the segment leaf that may contain
    ** a key equal to the one multi-cursor currently points to. Record the
    ** page number of each b-tree page and the leaf. The segment may be
    ** gobbled up to (but not including) the first of these page numbers.
    */
    assert( pSeg->iRoot>0 );
    aPg = lsmMallocZeroRc(pDb->pEnv, sizeof(Pgno)*32, &rc);
    if( rc==LSM_OK ){
      rc = seekInBtree(pCsr, pSeg, 
          rtTopic(pCsr->eType), pCsr->key.pData, pCsr->key.nData, aPg, 0
      ); 
    }

    if( rc==LSM_OK ){
................................................................................
/*
** Return a string representation of the segment passed as the only argument.
** Space for the returned string is allocated using lsmMalloc(), and should
** be freed by the caller using lsmFree().
*/
static char *segToString(lsm_env *pEnv, Segment *pSeg, int nMin){
  int nSize = pSeg->nSize;
  Pgno iRoot = pSeg->iRoot;
  Pgno iFirst = pSeg->iFirst;
  Pgno iLast = pSeg->iLastPg;
  char *z;

  char *z1;
  char *z2;
  int nPad;

  z1 = lsmMallocPrintf(pEnv, "%d.%d", iFirst, iLast);
................................................................................
    aBuf[0] = '\0';
  }

  return i;
}

void sortedDumpPage(lsm_db *pDb, Segment *pRun, Page *pPg, int bVals){
  Blob blob = {0, 0, 0};         /* Blob used for keys */
  LsmString s;
  int i;

  int nRec;
  int iPtr;
  int flags;
  u8 *aData;
................................................................................

    aCell = pageGetCell(aData, nData, i);
    eType = *aCell++;
    assert( (flags & SEGMENT_BTREE_FLAG) || eType!=0 );
    aCell += lsmVarintGet32(aCell, &iPgPtr);

    if( eType==0 ){
      Pgno iRef;                  /* Page number of referenced page */
      aCell += lsmVarintGet64(aCell, &iRef);
      lsmFsDbPageGet(pDb->pFS, pRun, iRef, &pRef);
      aKey = pageGetKey(pRun, pRef, 0, &iTopic, &nKey, &blob);
    }else{
      aCell += lsmVarintGet32(aCell, &nKey);
      if( rtIsWrite(eType) ) aCell += lsmVarintGet32(aCell, &nVal);
      sortedReadData(0, pPg, (aCell-aData), nKey+nVal, (void **)&aKey, &blob);
................................................................................
  int bIndirect,                  /* True to follow indirect refs */
  Page *pPg,
  int iCell,
  int *peType,
  int *piPgPtr,
  u8 **paKey, int *pnKey,
  u8 **paVal, int *pnVal,
  Blob *pBlob
){
  u8 *aData; int nData;           /* Page data */
  u8 *aKey; int nKey = 0;         /* Key */
  u8 *aVal = 0; int nVal = 0;     /* Value */
  int eType;
  int iPgPtr;
  Page *pRef = 0;                 /* Pointer to page iRef */
................................................................................

  aCell = pageGetCell(aData, nData, iCell);
  eType = *aCell++;
  aCell += lsmVarintGet32(aCell, &iPgPtr);

  if( eType==0 ){
    int dummy;
    Pgno iRef;                  /* Page number of referenced page */
    aCell += lsmVarintGet64(aCell, &iRef);
    if( bIndirect ){
      lsmFsDbPageGet(pDb->pFS, pSeg, iRef, &pRef);
      pageGetKeyCopy(pDb->pEnv, pSeg, pRef, 0, &dummy, pBlob);
      aKey = (u8 *)pBlob->pData;
      nKey = pBlob->nData;
      lsmFsPageRelease(pRef);
................................................................................
#define INFO_PAGE_DUMP_DATA     0x01
#define INFO_PAGE_DUMP_VALUES   0x02
#define INFO_PAGE_DUMP_HEX      0x04
#define INFO_PAGE_DUMP_INDIRECT 0x08

static int infoPageDump(
  lsm_db *pDb,                    /* Database handle */
  Pgno iPg,                       /* Page number of page to dump */
  int flags,
  char **pzOut                    /* OUT: lsmMalloc'd string */
){
  int rc = LSM_OK;                /* Return code */
  Page *pPg = 0;                  /* Handle for page iPg */
  int i, j;                       /* Loop counters */
  const int perLine = 16;         /* Bytes per line in the raw hex dump */
................................................................................
  ** to pass a NULL in place of the segment pointer as the second argument
  ** to lsmFsDbPageGet() here.  */
  if( rc==LSM_OK ){
    rc = lsmFsDbPageGet(pDb->pFS, 0, iPg, &pPg);
  }

  if( rc==LSM_OK ){
    Blob blob = {0, 0, 0, 0};
    int nKeyWidth = 0;
    LsmString str;
    int nRec;
    int iPtr;
    int flags2;
    int iCell;
    u8 *aData; int nData;         /* Page data and size thereof */
................................................................................
    if( bHex ) nKeyWidth = nKeyWidth * 2;

    for(iCell=0; iCell<nRec; iCell++){
      u8 *aKey; int nKey = 0;       /* Key */
      u8 *aVal; int nVal = 0;       /* Value */
      int iPgPtr;
      int eType;
      Pgno iAbsPtr;
      char zFlags[8];

      infoCellDump(pDb, pSeg, bIndirect, pPg, iCell, &eType, &iPgPtr,
          &aKey, &nKey, &aVal, &nVal, &blob
      );
      iAbsPtr = iPgPtr + ((flags2 & SEGMENT_BTREE_FLAG) ? 0 : iPtr);

................................................................................
  }

  return rc;
}

int lsmInfoPageDump(
  lsm_db *pDb,                    /* Database handle */
  Pgno iPg,                       /* Page number of page to dump */
  int bHex,                       /* True to output key/value in hex form */
  char **pzOut                    /* OUT: lsmMalloc'd string */
){
  int flags = INFO_PAGE_DUMP_DATA | INFO_PAGE_DUMP_VALUES;
  if( bHex ) flags |= INFO_PAGE_DUMP_HEX;
  return infoPageDump(pDb, iPg, flags, pzOut);
}
................................................................................
  iHdr = SEGMENT_EOF(nOrig, nEntry);
  memmove(&aData[iHdr + (nData-nOrig)], &aData[iHdr], nOrig-iHdr);
}

#ifdef LSM_DEBUG_EXPENSIVE
static void assertRunInOrder(lsm_db *pDb, Segment *pSeg){
  Page *pPg = 0;
  Blob blob1 = {0, 0, 0, 0};
  Blob blob2 = {0, 0, 0, 0};

  lsmFsDbPageGet(pDb->pFS, pSeg, pSeg->iFirst, &pPg);
  while( pPg ){
    u8 *aData; int nData;
    Page *pNext;

    aData = lsmFsPageData(pPg, &nData);
................................................................................
  Segment *pOne,                  /* Segment containing pointers */
  Segment *pTwo,                  /* Segment containing pointer targets */
  int bRhs                        /* True if pTwo may have been Gobble()d */
){
  int rc = LSM_OK;                /* Error code */
  SegmentPtr ptr1;                /* Iterates through pOne */
  SegmentPtr ptr2;                /* Iterates through pTwo */
  Pgno iPrev;

  assert( pOne && pTwo );

  memset(&ptr1, 0, sizeof(ptr1));
  memset(&ptr2, 0, sizeof(ptr1));
  ptr1.pSeg = pOne;
  ptr2.pSeg = pTwo;
................................................................................
  }

  if( rc==LSM_OK && ptr1.nCell>0 ){
    rc = segmentPtrLoadCell(&ptr1, 0);
  }
      
  while( rc==LSM_OK && ptr2.pPg ){
    Pgno iThis;

    /* Advance to the next page of segment pTwo that contains at least
    ** one cell. Break out of the loop if the iterator reaches EOF.  */
    do{
      rc = segmentPtrNextPage(&ptr2, 1);
      assert( rc==LSM_OK );
    }while( rc==LSM_OK && ptr2.pPg && ptr2.nCell==0 );
................................................................................
*/
static int assertBtreeOk(
  lsm_db *pDb,
  Segment *pSeg
){
  int rc = LSM_OK;                /* Return code */
  if( pSeg->iRoot ){
    Blob blob = {0, 0, 0};        /* Buffer used to cache overflow keys */
    FileSystem *pFS = pDb->pFS;   /* File system to read from */
    Page *pPg = 0;                /* Main run page */
    BtreeCursor *pCsr = 0;        /* Btree cursor */

    rc = btreeCursorNew(pDb, pSeg, &pCsr);
    if( rc==LSM_OK ){
      rc = btreeCursorFirst(pCsr);







|

|







 







|




|




|
|







 







|


|







 







|
|







 







|







 







|

|

|







 







|










|







|




|


|








|







 







|
|







 







|
|







 







|







 







|







 







|
|







 







|



|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







|







 







|







 







|







 







|






|







 







|







 







|






|



|







 







|







 







|







 







|







 







|







 







|
|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|








|







 







|
|
|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|
|







 







|







 







|







 







|







100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
...
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
...
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
...
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
...
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
...
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
...
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
...
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
...
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
...
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
...
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
...
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
...
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
...
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
...
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
...
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
...
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
...
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
...
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
...
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
...
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
....
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
....
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
....
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
....
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
....
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
....
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
....
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
....
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
....
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
....
3051
3052
3053
3054
3055
3056
3057
3058
3059
3060
3061
3062
3063
3064
3065
....
3533
3534
3535
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
....
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
3577
3578
3579
3580
3581
3582
....
3678
3679
3680
3681
3682
3683
3684
3685
3686
3687
3688
3689
3690
3691
3692
....
3699
3700
3701
3702
3703
3704
3705
3706
3707
3708
3709
3710
3711
3712
3713
....
3730
3731
3732
3733
3734
3735
3736
3737
3738
3739
3740
3741
3742
3743
3744
....
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
....
4214
4215
4216
4217
4218
4219
4220
4221
4222
4223
4224
4225
4226
4227
4228
....
4367
4368
4369
4370
4371
4372
4373
4374
4375
4376
4377
4378
4379
4380
4381
....
4544
4545
4546
4547
4548
4549
4550
4551
4552
4553
4554
4555
4556
4557
4558
....
4626
4627
4628
4629
4630
4631
4632
4633
4634
4635
4636
4637
4638
4639
4640
4641
4642
4643
4644
4645
4646
4647
4648
4649
....
5462
5463
5464
5465
5466
5467
5468
5469
5470
5471
5472
5473
5474
5475
5476
5477
5478
....
5523
5524
5525
5526
5527
5528
5529
5530
5531
5532
5533
5534
5535
5536
5537
....
5559
5560
5561
5562
5563
5564
5565
5566
5567
5568
5569
5570
5571
5572
5573
....
5603
5604
5605
5606
5607
5608
5609
5610
5611
5612
5613
5614
5615
5616
5617
....
5621
5622
5623
5624
5625
5626
5627
5628
5629
5630
5631
5632
5633
5634
5635
....
5667
5668
5669
5670
5671
5672
5673
5674
5675
5676
5677
5678
5679
5680
5681
....
5708
5709
5710
5711
5712
5713
5714
5715
5716
5717
5718
5719
5720
5721
5722
....
5743
5744
5745
5746
5747
5748
5749
5750
5751
5752
5753
5754
5755
5756
5757
....
5809
5810
5811
5812
5813
5814
5815
5816
5817
5818
5819
5820
5821
5822
5823
....
5985
5986
5987
5988
5989
5990
5991
5992
5993
5994
5995
5996
5997
5998
5999
6000
....
6048
6049
6050
6051
6052
6053
6054
6055
6056
6057
6058
6059
6060
6061
6062
....
6071
6072
6073
6074
6075
6076
6077
6078
6079
6080
6081
6082
6083
6084
6085
....
6133
6134
6135
6136
6137
6138
6139
6140
6141
6142
6143
6144
6145
6146
6147


#ifndef LSM_SEGMENTPTR_FREE_THRESHOLD
# define LSM_SEGMENTPTR_FREE_THRESHOLD 1024
#endif

typedef struct SegmentPtr SegmentPtr;
typedef struct LsmBlob LsmBlob;

struct LsmBlob {
  lsm_env *pEnv;
  void *pData;
  int nData;
  int nAlloc;
};

/*
................................................................................
  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 */
  LsmPgno iPtr;                 /* Base cascade pointer */

  /* Current cell. See segmentPtrLoadCell() */
  int iCell;                    /* Current record within page pPg */
  int eType;                    /* Type of current record */
  LsmPgno iPgPtr;               /* Cascade pointer offset */
  void *pKey; int nKey;         /* Key associated with current record */
  void *pVal; int nVal;         /* Current record value (eType==WRITE only) */

  /* Blobs used to allocate buffers for pKey and pVal as required */
  LsmBlob blob1;
  LsmBlob blob2;
};

/*
** Used to iterate through the keys stored in a b-tree hierarchy from start
** to finish. Only First() and Next() operations are required.
**
**   btreeCursorNew()
................................................................................
  int iPg;                        /* Current entry in aPg[]. -1 -> EOF. */
  BtreePg *aPg;                   /* Pages from root to current location */

  /* Cache of current entry. pKey==0 for EOF. */
  void *pKey;
  int nKey;
  int eType;
  LsmPgno iPtr;

  /* Storage for key, if not local */
  LsmBlob blob;
};


/*
** A cursor used for merged searches or iterations through up to one
** Tree structure and any number of sorted files.
**
................................................................................
*/
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 */
  LsmBlob key;                    /* Cache of current key (or NULL) */
  LsmBlob val;                    /* Cache of current value */

  /* All the component cursors: */
  TreeCursor *apTreeCsr[2];       /* Up to two tree cursors */
  int iFree;                      /* Next element of free-list (-ve for eof) */
  SegmentPtr *aPtr;               /* Array of segment pointers */
  int nPtr;                       /* Size of array aPtr[] */
  BtreeCursor *pBtCsr;            /* b-tree cursor (db writes only) */
................................................................................
  int nTree;                      /* Size of aTree[] array */
  int *aTree;                     /* Array of comparison results */

  /* Used by cursors flushing the in-memory tree only */
  void *pSystemVal;               /* Pointer to buffer to free */

  /* Used by worker cursors only */
  LsmPgno *pPrevMergePtr;
};

/*
** The following constants are used to assign integers to each component
** cursor of a multi-cursor.
*/
#define CURSOR_DATA_TREE0     0   /* Current tree cursor (apTreeCsr[0]) */
................................................................................
  lsm_db *pDb;                    /* Database handle */
  Level *pLevel;                  /* Worker snapshot Level being merged */
  MultiCursor *pCsr;              /* Cursor to read new segment contents from */
  int bFlush;                     /* True if this is an in-memory tree flush */
  Hierarchy hier;                 /* B-tree hierarchy under construction */
  Page *pPage;                    /* Current output page */
  int nWork;                      /* Number of calls to mergeWorkerNextPage() */
  LsmPgno *aGobble;               /* Gobble point for each input segment */

  LsmPgno iIndirect;
  struct SavedPgno {
    LsmPgno iPgno;
    int bStore;
  } aSave[2];
};

#ifdef LSM_DEBUG_EXPENSIVE
static int assertPointersOk(lsm_db *, Segment *, Segment *, int);
static int assertBtreeOk(lsm_db *, Segment *);
................................................................................
  aOut[3] = (u8)((nVal>>32) & 0xFF);
  aOut[4] = (u8)((nVal>>24) & 0xFF);
  aOut[5] = (u8)((nVal>>16) & 0xFF);
  aOut[6] = (u8)((nVal>> 8) & 0xFF);
  aOut[7] = (u8)((nVal    ) & 0xFF);
}

static int sortedBlobGrow(lsm_env *pEnv, LsmBlob *pBlob, int nData){
  assert( pBlob->pEnv==pEnv || (pBlob->pEnv==0 && pBlob->pData==0) );
  if( pBlob->nAlloc<nData ){
    pBlob->pData = lsmReallocOrFree(pEnv, pBlob->pData, nData);
    if( !pBlob->pData ) return LSM_NOMEM_BKPT;
    pBlob->nAlloc = nData;
    pBlob->pEnv = pEnv;
  }
  return LSM_OK;
}

static int sortedBlobSet(lsm_env *pEnv, LsmBlob *pBlob, void *pData, int nData){
  if( sortedBlobGrow(pEnv, pBlob, nData) ) return LSM_NOMEM;
  memcpy(pBlob->pData, pData, nData);
  pBlob->nData = nData;
  return LSM_OK;
}

#if 0
static int sortedBlobCopy(LsmBlob *pDest, LsmBlob *pSrc){
  return sortedBlobSet(pDest, pSrc->pData, pSrc->nData);
}
#endif

static void sortedBlobFree(LsmBlob *pBlob){
  assert( pBlob->pEnv || pBlob->pData==0 );
  if( pBlob->pData ) lsmFree(pBlob->pEnv, pBlob->pData);
  memset(pBlob, 0, sizeof(LsmBlob));
}

static int sortedReadData(
  Segment *pSeg,
  Page *pPg,
  int iOff,
  int nByte,
  void **ppData,
  LsmBlob *pBlob
){
  int rc = LSM_OK;
  int iEnd;
  int nData;
  int nCell;
  u8 *aData;

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

static int pageGetNRec(u8 *aData, int nData){
  return (int)lsmGetU16(&aData[SEGMENT_NRECORD_OFFSET(nData)]);
}

static LsmPgno pageGetPtr(u8 *aData, int nData){
  return (LsmPgno)lsmGetU64(&aData[SEGMENT_POINTER_OFFSET(nData)]);
}

static int pageGetFlags(u8 *aData, int nData){
  return (int)lsmGetU16(&aData[SEGMENT_FLAGS_OFFSET(nData)]);
}

static u8 *pageGetCell(u8 *aData, int nData, int iCell){
................................................................................
  return pageGetNRec(aData, nData);
}

/*
** Return the decoded (possibly relative) pointer value stored in cell 
** iCell from page aData/nData.
*/
static LsmPgno pageGetRecordPtr(u8 *aData, int nData, int iCell){
  LsmPgno iRet;                   /* Return value */
  u8 *aCell;                      /* Pointer to cell iCell */

  assert( iCell<pageGetNRec(aData, nData) && iCell>=0 );
  aCell = pageGetCell(aData, nData, iCell);
  lsmVarintGet64(&aCell[1], &iRet);
  return iRet;
}
................................................................................

static u8 *pageGetKey(
  Segment *pSeg,                  /* Segment pPg belongs to */
  Page *pPg,                      /* Page to read from */
  int iCell,                      /* Index of cell on page to read */
  int *piTopic,                   /* OUT: Topic associated with this key */
  int *pnKey,                     /* OUT: Size of key in bytes */
  LsmBlob *pBlob                  /* If required, use this for dynamic memory */
){
  u8 *pKey;
  int nDummy;
  int eType;
  u8 *aData;
  int nData;

................................................................................

static int pageGetKeyCopy(
  lsm_env *pEnv,                  /* Environment handle */
  Segment *pSeg,                  /* Segment pPg belongs to */
  Page *pPg,                      /* Page to read from */
  int iCell,                      /* Index of cell on page to read */
  int *piTopic,                   /* OUT: Topic associated with this key */
  LsmBlob *pBlob                  /* If required, use this for dynamic memory */
){
  int rc = LSM_OK;
  int nKey;
  u8 *aKey;

  aKey = pageGetKey(pSeg, pPg, iCell, piTopic, &nKey, pBlob);
  assert( (void *)aKey!=pBlob->pData || nKey==pBlob->nData );
................................................................................
  if( (void *)aKey!=pBlob->pData ){
    rc = sortedBlobSet(pEnv, pBlob, aKey, nKey);
  }

  return rc;
}

static LsmPgno pageGetBtreeRef(Page *pPg, int iKey){
  LsmPgno iRef;
  u8 *aData;
  int nData;
  u8 *aCell;

  aData = fsPageData(pPg, &nData);
  aCell = pageGetCell(aData, nData, iKey);
  assert( aCell[0]==0 );
................................................................................
#define GETVARINT64(a, i) (((i)=((u8*)(a))[0])<=240?1:lsmVarintGet64((a), &(i)))
#define GETVARINT32(a, i) (((i)=((u8*)(a))[0])<=240?1:lsmVarintGet32((a), &(i)))

static int pageGetBtreeKey(
  Segment *pSeg,                  /* Segment page pPg belongs to */
  Page *pPg,
  int iKey, 
  LsmPgno *piPtr, 
  int *piTopic, 
  void **ppKey,
  int *pnKey,
  LsmBlob *pBlob
){
  u8 *aData;
  int nData;
  u8 *aCell;
  int eType;

  aData = fsPageData(pPg, &nData);
................................................................................

  aCell = pageGetCell(aData, nData, iKey);
  eType = *aCell++;
  aCell += GETVARINT64(aCell, *piPtr);

  if( eType==0 ){
    int rc;
    LsmPgno iRef;               /* Page number of referenced page */
    Page *pRef;
    aCell += GETVARINT64(aCell, iRef);
    rc = lsmFsDbPageGet(lsmPageFS(pPg), pSeg, iRef, &pRef);
    if( rc!=LSM_OK ) return rc;
    pageGetKeyCopy(lsmPageEnv(pPg), pSeg, pRef, 0, &eType, pBlob);
    lsmFsPageRelease(pRef);
    *ppKey = pBlob->pData;
................................................................................
static int btreeCursorLoadKey(BtreeCursor *pCsr){
  int rc = LSM_OK;
  if( pCsr->iPg<0 ){
    pCsr->pKey = 0;
    pCsr->nKey = 0;
    pCsr->eType = 0;
  }else{
    LsmPgno dummy;
    int iPg = pCsr->iPg;
    int iCell = pCsr->aPg[iPg].iCell;
    while( iCell<0 && (--iPg)>=0 ){
      iCell = pCsr->aPg[iPg].iCell-1;
    }
    if( iPg<0 || iCell<0 ) return LSM_CORRUPT_BKPT;

................................................................................
  assert( pCsr->iPg==pCsr->nDepth-1 );

  aData = fsPageData(pPg->pPage, &nData);
  nCell = pageGetNRec(aData, nData);
  assert( pPg->iCell<=nCell );
  pPg->iCell++;
  if( pPg->iCell==nCell ){
    LsmPgno iLoad;

    /* Up to parent. */
    lsmFsPageRelease(pPg->pPage);
    pPg->pPage = 0;
    pCsr->iPg--;
    while( pCsr->iPg>=0 ){
      pPg = &pCsr->aPg[pCsr->iPg];
................................................................................
  MergeInput *p
){
  int rc = LSM_OK;

  if( p->iPg ){
    lsm_env *pEnv = lsmFsEnv(pCsr->pFS);
    int iCell;                    /* Current cell number on leaf page */
    LsmPgno iLeaf;                /* Page number of current leaf page */
    int nDepth;                   /* Depth of b-tree structure */
    Segment *pSeg = pCsr->pSeg;

    /* Decode the MergeInput structure */
    iLeaf = p->iPg;
    nDepth = (p->iCell & 0x00FF);
    iCell = (p->iCell >> 8) - 1;
................................................................................
      pCsr->nDepth = nDepth;
      pCsr->aPg[pCsr->iPg].iCell = iCell;
      rc = lsmFsDbPageGet(pCsr->pFS, pSeg, iLeaf, pp);
    }

    /* Populate any other aPg[] array entries */
    if( rc==LSM_OK && nDepth>1 ){
      LsmBlob blob = {0,0,0};
      void *pSeek;
      int nSeek;
      int iTopicSeek;
      int iPg = 0;
      int iLoad = (int)pSeg->iRoot;
      Page *pPg = pCsr->aPg[nDepth-1].pPage;
 
................................................................................
        ** In this case, set the iTopicSeek/pSeek/nSeek key to a value
        ** greater than any real key.  */
        assert( iCell==-1 );
        iTopicSeek = 1000;
        pSeek = 0;
        nSeek = 0;
      }else{
        LsmPgno dummy;
        rc = pageGetBtreeKey(pSeg, pPg,
            0, &dummy, &iTopicSeek, &pSeek, &nSeek, &pCsr->blob
        );
      }

      do {
        Page *pPg2;
................................................................................
          iMax = iCell2-1;
          iMin = 0;

          while( iMax>=iMin ){
            int iTry = (iMin+iMax)/2;
            void *pKey; int nKey;         /* Key for cell iTry */
            int iTopic;                   /* Topic for key pKeyT/nKeyT */
            LsmPgno iPtr;                 /* Pointer for cell iTry */
            int res;                      /* (pSeek - pKeyT) */

            rc = pageGetBtreeKey(
                pSeg, pPg2, iTry, &iPtr, &iTopic, &pKey, &nKey, &blob
            );
            if( rc!=LSM_OK ) break;

................................................................................
      u8 *aData;
      int nData;

      pBtreePg = &pCsr->aPg[pCsr->iPg];
      aData = fsPageData(pBtreePg->pPage, &nData);
      pCsr->iPtr = btreeCursorPtr(aData, nData, pBtreePg->iCell+1);
      if( pBtreePg->iCell<0 ){
        LsmPgno dummy;
        int i;
        for(i=pCsr->iPg-1; i>=0; i--){
          if( pCsr->aPg[i].iCell>0 ) break;
        }
        assert( i>=0 );
        rc = pageGetBtreeKey(pSeg,
            pCsr->aPg[i].pPage, pCsr->aPg[i].iCell-1,
................................................................................
}

static int segmentPtrReadData(
  SegmentPtr *pPtr,
  int iOff,
  int nByte,
  void **ppData,
  LsmBlob *pBlob
){
  return sortedReadData(pPtr->pSeg, pPtr->pPg, iOff, nByte, ppData, pBlob);
}

static int segmentPtrNextPage(
  SegmentPtr *pPtr,              /* Load page into this SegmentPtr object */
  int eDir                       /* +1 for next(), -1 for prev() */
................................................................................

  pSeg = sortedSplitkeySegment(pLevel);
  if( rc==LSM_OK ){
    rc = lsmFsDbPageGet(pDb->pFS, pSeg, pMerge->splitkey.iPg, &pPg);
  }
  if( rc==LSM_OK ){
    int iTopic;
    LsmBlob blob = {0, 0, 0, 0};
    u8 *aData;
    int nData;
  
    aData = lsmFsPageData(pPg, &nData);
    if( pageGetFlags(aData, nData) & SEGMENT_BTREE_FLAG ){
      void *pKey;
      int nKey;
      LsmPgno dummy;
      rc = pageGetBtreeKey(pSeg,
          pPg, pMerge->splitkey.iCell, &dummy, &iTopic, &pKey, &nKey, &blob
      );
      if( rc==LSM_OK && blob.pData!=pKey ){
        rc = sortedBlobSet(pEnv, &blob, pKey, nKey);
      }
    }else{
................................................................................
*/
static int assertKeyLocation(
  MultiCursor *pCsr, 
  SegmentPtr *pPtr, 
  void *pKey, int nKey
){
  lsm_env *pEnv = lsmFsEnv(pCsr->pDb->pFS);
  LsmBlob blob = {0, 0, 0};
  int eDir;
  int iTopic = 0;                 /* TODO: Fix me */

  for(eDir=-1; eDir<=1; eDir+=2){
    Page *pTest = pPtr->pPg;

    lsmFsPageRef(pTest);
................................................................................
  return rc;
}

static int ptrFwdPointer(
  Page *pPage,
  int iCell,
  Segment *pSeg,
  LsmPgno *piPtr,
  int *pbFound
){
  Page *pPg = pPage;
  int iFirst = iCell;
  int rc = LSM_OK;

  do {
................................................................................
**   much better if the multi-cursor could do this lazily - only seek to the
**   level (N+1) page after the user has moved the cursor on level N passed
**   the big range-delete.
*/
static int segmentPtrFwdPointer(
  MultiCursor *pCsr,              /* Multi-cursor pPtr belongs to */
  SegmentPtr *pPtr,               /* Segment-pointer to extract FC ptr from */
  LsmPgno *piPtr                  /* OUT: FC pointer value */
){
  Level *pLvl = pPtr->pLevel;
  Level *pNext = pLvl->pNext;
  Page *pPg = pPtr->pPg;
  int rc;
  int bFound;
  LsmPgno iOut = 0;

  if( pPtr->pSeg==&pLvl->lhs || pPtr->pSeg==&pLvl->aRhs[pLvl->nRight-1] ){
    if( pNext==0 
        || (pNext->nRight==0 && pNext->lhs.iRoot)
        || (pNext->nRight!=0 && pNext->aRhs[0].iRoot)
      ){
      /* Do nothing. The pointer will not be used anyway. */
................................................................................
  int *pbStop
){
  int (*xCmp)(void *, int, void *, int) = pCsr->pDb->xCmp;
  int res = 0;                        /* Result of comparison operation */
  int rc = LSM_OK;
  int iMin;
  int iMax;
  LsmPgno iPtrOut = 0;

  /* If the current page contains an oversized entry, then there are no
  ** pointers to one or more of the subsequent pages in the sorted run.
  ** The following call ensures that the segment-ptr points to the correct 
  ** page in this case.  */
  rc = segmentPtrSearchOversized(pCsr, pPtr, iTopic, pKey, nKey);
  iPtrOut = pPtr->iPtr;
................................................................................
}

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

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

    rc = lsmFsDbPageGet(pCsr->pDb->pFS, pSeg, iPg, &pPg);
    assert( rc==LSM_OK || pPg==0 );
................................................................................

      iMin = 0;
      iMax = nRec-1;
      while( iMax>=iMin ){
        int iTry = (iMin+iMax)/2;
        void *pKeyT; int nKeyT;       /* Key for cell iTry */
        int iTopicT;                  /* Topic for key pKeyT/nKeyT */
        LsmPgno iPtr;                 /* Pointer associated with cell iTry */
        int res;                      /* (pKey - pKeyT) */

        rc = pageGetBtreeKey(
            pSeg, pPg, iTry, &iPtr, &iTopicT, &pKeyT, &nKeyT, &blob
        );
        if( rc!=LSM_OK ) break;
        if( piFirst && pKeyT==blob.pData ){
................................................................................
*/
static int seekInLevel(
  MultiCursor *pCsr,              /* Sorted cursor object to seek */
  SegmentPtr *aPtr,               /* Pointer to array of (nRhs+1) SPs */
  int eSeek,                      /* Search bias - see above */
  int iTopic,                     /* Key topic to search for */
  void *pKey, int nKey,           /* Key to search for */
  LsmPgno *piPgno,                /* IN/OUT: fraction cascade pointer (or 0) */
  int *pbStop                     /* OUT: See above */
){
  Level *pLvl = aPtr[0].pLevel;   /* Level to seek within */
  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 */
................................................................................
  void *pKey, int nKey, 
  int eSeek
){
  int eESeek = eSeek;             /* Effective eSeek parameter */
  int bStop = 0;                  /* Set to true to halt search operation */
  int rc = LSM_OK;                /* Return code */
  int iPtr = 0;                   /* Used to iterate through pCsr->aPtr[] */
  LsmPgno iPgno = 0;              /* FC pointer value */

  assert( pCsr->apTreeCsr[0]==0 || iTopic==0 );
  assert( pCsr->apTreeCsr[1]==0 || iTopic==0 );

  if( eESeek==LSM_SEEK_LEFAST ) eESeek = LSM_SEEK_LE;

  assert( eESeek==LSM_SEEK_EQ || eESeek==LSM_SEEK_LE || eESeek==LSM_SEEK_GE );
................................................................................
** differences are:
**
**   1. The record format is (usually, see below) as follows:
**
**         + Type byte (always SORTED_SEPARATOR or SORTED_SYSTEM_SEPARATOR),
**         + Absolute pointer value (varint),
**         + Number of bytes in key (varint),
**         + LsmBlob containing key data.
**
**   2. All pointer values are stored as absolute values (not offsets 
**      relative to the footer pointer value).
**
**   3. Each pointer that is part of a record points to a page that 
**      contains keys smaller than the records key (note: not "equal to or
**      smaller than - smaller than").
................................................................................
**
** See function seekInBtree() for the code that traverses b-tree pages.
*/

static int mergeWorkerBtreeWrite(
  MergeWorker *pMW,
  u8 eType,
  LsmPgno iPtr,
  LsmPgno iKeyPg,
  void *pKey,
  int nKey
){
  Hierarchy *p = &pMW->hier;
  lsm_db *pDb = pMW->pDb;         /* Database handle */
  int rc = LSM_OK;                /* Return Code */
  int iLevel;                     /* Level of b-tree hierachy to write to */
................................................................................

  return rc;
}

static int mergeWorkerBtreeIndirect(MergeWorker *pMW){
  int rc = LSM_OK;
  if( pMW->iIndirect ){
    LsmPgno iKeyPg = pMW->aSave[1].iPgno;
    rc = mergeWorkerBtreeWrite(pMW, 0, pMW->iIndirect, iKeyPg, 0, 0);
    pMW->iIndirect = 0;
  }
  return rc;
}

/*
................................................................................
static int mergeWorkerPushHierarchy(
  MergeWorker *pMW,               /* Merge worker object */
  int iTopic,                     /* Topic value for this key */
  void *pKey,                     /* Pointer to key buffer */
  int nKey                        /* Size of pKey buffer in bytes */
){
  int rc = LSM_OK;                /* Return Code */
  LsmPgno iPtr;                   /* Pointer value to accompany pKey/nKey */

  assert( pMW->aSave[0].bStore==0 );
  assert( pMW->aSave[1].bStore==0 );
  rc = mergeWorkerBtreeIndirect(pMW);

  /* Obtain the absolute pointer value to store along with the key in the
  ** page body. This pointer points to a page that contains keys that are
................................................................................
}

static int mergeWorkerFinishHierarchy(
  MergeWorker *pMW                /* Merge worker object */
){
  int i;                          /* Used to loop through apHier[] */
  int rc = LSM_OK;                /* Return code */
  LsmPgno iPtr;                   /* New right-hand-child pointer value */

  iPtr = pMW->aSave[0].iPgno;
  for(i=0; i<pMW->hier.nHier && rc==LSM_OK; i++){
    Page *pPg = pMW->hier.apHier[i];
    int nData;                    /* Size of aData[] in bytes */
    u8 *aData;                    /* Page data for pPg */

................................................................................
** zero records. The flags field is cleared. The page footer pointer field
** is set to iFPtr.
**
** If successful, LSM_OK is returned. Otherwise, an error code.
*/
static int mergeWorkerNextPage(
  MergeWorker *pMW,               /* Merge worker object to append page to */
  LsmPgno iFPtr                   /* Pointer value for footer of new page */
){
  int rc = LSM_OK;                /* Return code */
  Page *pNext = 0;                /* New page appended to run */
  lsm_db *pDb = pMW->pDb;         /* Database handle */

  rc = lsmFsSortedAppend(pDb->pFS, pDb->pWorker, pMW->pLevel, 0, &pNext);
  assert( rc || pMW->pLevel->lhs.iFirst>0 || pMW->pDb->compress.xCompress );
................................................................................

static int mergeWorkerStep(MergeWorker *pMW){
  lsm_db *pDb = pMW->pDb;       /* Database handle */
  MultiCursor *pCsr;            /* Cursor to read input data from */
  int rc = LSM_OK;              /* Return code */
  int eType;                    /* SORTED_SEPARATOR, WRITE or DELETE */
  void *pKey; int nKey;         /* Key */
  LsmPgno iPtr;
  int iVal;

  pCsr = pMW->pCsr;

  /* Pull the next record out of the source cursor. */
  lsmMCursorKey(pCsr, &pKey, &nKey);
  eType = pCsr->eType;
................................................................................
      multiCursorIgnoreDelete(pCsr);
    }
  }

  if( rc!=LSM_OK ){
    lsmMCursorClose(pCsr, 0);
  }else{
    LsmPgno iLeftPtr = 0;
    Merge merge;                  /* Merge object used to create new level */
    MergeWorker mergeworker;      /* MergeWorker object for the same purpose */

    memset(&merge, 0, sizeof(Merge));
    memset(&mergeworker, 0, sizeof(MergeWorker));

    pNew->pMerge = &merge;
................................................................................
  assert( pDb->pWorker );
  assert( pLevel->pMerge );
  assert( pLevel->nRight>0 );

  memset(pMW, 0, sizeof(MergeWorker));
  pMW->pDb = pDb;
  pMW->pLevel = pLevel;
  pMW->aGobble = lsmMallocZeroRc(pDb->pEnv, sizeof(LsmPgno)*pLevel->nRight,&rc);

  /* Create a multi-cursor to read the data to write to the new
  ** segment. The new segment contains:
  **
  **   1. Records from LHS of each of the nMerge levels being merged.
  **   2. Separators from either the last level being merged, or the
  **      separators attached to the LHS of the following level, or neither.
................................................................................
  lsm_db *pDb,                    /* Worker connection */
  MultiCursor *pCsr,              /* Multi-cursor being used for a merge */
  int iGobble                     /* pCsr->aPtr[] entry to operate on */
){
  int rc = LSM_OK;
  if( rtTopic(pCsr->eType)==0 ){
    Segment *pSeg = pCsr->aPtr[iGobble].pSeg;
    LsmPgno *aPg;
    int nPg;

    /* Seek from the root of the b-tree to the segment leaf that may contain
    ** a key equal to the one multi-cursor currently points to. Record the
    ** page number of each b-tree page and the leaf. The segment may be
    ** gobbled up to (but not including) the first of these page numbers.
    */
    assert( pSeg->iRoot>0 );
    aPg = lsmMallocZeroRc(pDb->pEnv, sizeof(LsmPgno)*32, &rc);
    if( rc==LSM_OK ){
      rc = seekInBtree(pCsr, pSeg, 
          rtTopic(pCsr->eType), pCsr->key.pData, pCsr->key.nData, aPg, 0
      ); 
    }

    if( rc==LSM_OK ){
................................................................................
/*
** Return a string representation of the segment passed as the only argument.
** Space for the returned string is allocated using lsmMalloc(), and should
** be freed by the caller using lsmFree().
*/
static char *segToString(lsm_env *pEnv, Segment *pSeg, int nMin){
  int nSize = pSeg->nSize;
  LsmPgno iRoot = pSeg->iRoot;
  LsmPgno iFirst = pSeg->iFirst;
  LsmPgno iLast = pSeg->iLastPg;
  char *z;

  char *z1;
  char *z2;
  int nPad;

  z1 = lsmMallocPrintf(pEnv, "%d.%d", iFirst, iLast);
................................................................................
    aBuf[0] = '\0';
  }

  return i;
}

void sortedDumpPage(lsm_db *pDb, Segment *pRun, Page *pPg, int bVals){
  LsmBlob blob = {0, 0, 0};       /* LsmBlob used for keys */
  LsmString s;
  int i;

  int nRec;
  int iPtr;
  int flags;
  u8 *aData;
................................................................................

    aCell = pageGetCell(aData, nData, i);
    eType = *aCell++;
    assert( (flags & SEGMENT_BTREE_FLAG) || eType!=0 );
    aCell += lsmVarintGet32(aCell, &iPgPtr);

    if( eType==0 ){
      LsmPgno iRef;               /* Page number of referenced page */
      aCell += lsmVarintGet64(aCell, &iRef);
      lsmFsDbPageGet(pDb->pFS, pRun, iRef, &pRef);
      aKey = pageGetKey(pRun, pRef, 0, &iTopic, &nKey, &blob);
    }else{
      aCell += lsmVarintGet32(aCell, &nKey);
      if( rtIsWrite(eType) ) aCell += lsmVarintGet32(aCell, &nVal);
      sortedReadData(0, pPg, (aCell-aData), nKey+nVal, (void **)&aKey, &blob);
................................................................................
  int bIndirect,                  /* True to follow indirect refs */
  Page *pPg,
  int iCell,
  int *peType,
  int *piPgPtr,
  u8 **paKey, int *pnKey,
  u8 **paVal, int *pnVal,
  LsmBlob *pBlob
){
  u8 *aData; int nData;           /* Page data */
  u8 *aKey; int nKey = 0;         /* Key */
  u8 *aVal = 0; int nVal = 0;     /* Value */
  int eType;
  int iPgPtr;
  Page *pRef = 0;                 /* Pointer to page iRef */
................................................................................

  aCell = pageGetCell(aData, nData, iCell);
  eType = *aCell++;
  aCell += lsmVarintGet32(aCell, &iPgPtr);

  if( eType==0 ){
    int dummy;
    LsmPgno iRef;                 /* Page number of referenced page */
    aCell += lsmVarintGet64(aCell, &iRef);
    if( bIndirect ){
      lsmFsDbPageGet(pDb->pFS, pSeg, iRef, &pRef);
      pageGetKeyCopy(pDb->pEnv, pSeg, pRef, 0, &dummy, pBlob);
      aKey = (u8 *)pBlob->pData;
      nKey = pBlob->nData;
      lsmFsPageRelease(pRef);
................................................................................
#define INFO_PAGE_DUMP_DATA     0x01
#define INFO_PAGE_DUMP_VALUES   0x02
#define INFO_PAGE_DUMP_HEX      0x04
#define INFO_PAGE_DUMP_INDIRECT 0x08

static int infoPageDump(
  lsm_db *pDb,                    /* Database handle */
  LsmPgno iPg,                    /* Page number of page to dump */
  int flags,
  char **pzOut                    /* OUT: lsmMalloc'd string */
){
  int rc = LSM_OK;                /* Return code */
  Page *pPg = 0;                  /* Handle for page iPg */
  int i, j;                       /* Loop counters */
  const int perLine = 16;         /* Bytes per line in the raw hex dump */
................................................................................
  ** to pass a NULL in place of the segment pointer as the second argument
  ** to lsmFsDbPageGet() here.  */
  if( rc==LSM_OK ){
    rc = lsmFsDbPageGet(pDb->pFS, 0, iPg, &pPg);
  }

  if( rc==LSM_OK ){
    LsmBlob blob = {0, 0, 0, 0};
    int nKeyWidth = 0;
    LsmString str;
    int nRec;
    int iPtr;
    int flags2;
    int iCell;
    u8 *aData; int nData;         /* Page data and size thereof */
................................................................................
    if( bHex ) nKeyWidth = nKeyWidth * 2;

    for(iCell=0; iCell<nRec; iCell++){
      u8 *aKey; int nKey = 0;       /* Key */
      u8 *aVal; int nVal = 0;       /* Value */
      int iPgPtr;
      int eType;
      LsmPgno iAbsPtr;
      char zFlags[8];

      infoCellDump(pDb, pSeg, bIndirect, pPg, iCell, &eType, &iPgPtr,
          &aKey, &nKey, &aVal, &nVal, &blob
      );
      iAbsPtr = iPgPtr + ((flags2 & SEGMENT_BTREE_FLAG) ? 0 : iPtr);

................................................................................
  }

  return rc;
}

int lsmInfoPageDump(
  lsm_db *pDb,                    /* Database handle */
  LsmPgno iPg,                    /* Page number of page to dump */
  int bHex,                       /* True to output key/value in hex form */
  char **pzOut                    /* OUT: lsmMalloc'd string */
){
  int flags = INFO_PAGE_DUMP_DATA | INFO_PAGE_DUMP_VALUES;
  if( bHex ) flags |= INFO_PAGE_DUMP_HEX;
  return infoPageDump(pDb, iPg, flags, pzOut);
}
................................................................................
  iHdr = SEGMENT_EOF(nOrig, nEntry);
  memmove(&aData[iHdr + (nData-nOrig)], &aData[iHdr], nOrig-iHdr);
}

#ifdef LSM_DEBUG_EXPENSIVE
static void assertRunInOrder(lsm_db *pDb, Segment *pSeg){
  Page *pPg = 0;
  LsmBlob blob1 = {0, 0, 0, 0};
  LsmBlob blob2 = {0, 0, 0, 0};

  lsmFsDbPageGet(pDb->pFS, pSeg, pSeg->iFirst, &pPg);
  while( pPg ){
    u8 *aData; int nData;
    Page *pNext;

    aData = lsmFsPageData(pPg, &nData);
................................................................................
  Segment *pOne,                  /* Segment containing pointers */
  Segment *pTwo,                  /* Segment containing pointer targets */
  int bRhs                        /* True if pTwo may have been Gobble()d */
){
  int rc = LSM_OK;                /* Error code */
  SegmentPtr ptr1;                /* Iterates through pOne */
  SegmentPtr ptr2;                /* Iterates through pTwo */
  LsmPgno iPrev;

  assert( pOne && pTwo );

  memset(&ptr1, 0, sizeof(ptr1));
  memset(&ptr2, 0, sizeof(ptr1));
  ptr1.pSeg = pOne;
  ptr2.pSeg = pTwo;
................................................................................
  }

  if( rc==LSM_OK && ptr1.nCell>0 ){
    rc = segmentPtrLoadCell(&ptr1, 0);
  }
      
  while( rc==LSM_OK && ptr2.pPg ){
    LsmPgno iThis;

    /* Advance to the next page of segment pTwo that contains at least
    ** one cell. Break out of the loop if the iterator reaches EOF.  */
    do{
      rc = segmentPtrNextPage(&ptr2, 1);
      assert( rc==LSM_OK );
    }while( rc==LSM_OK && ptr2.pPg && ptr2.nCell==0 );
................................................................................
*/
static int assertBtreeOk(
  lsm_db *pDb,
  Segment *pSeg
){
  int rc = LSM_OK;                /* Return code */
  if( pSeg->iRoot ){
    LsmBlob blob = {0, 0, 0};     /* Buffer used to cache overflow keys */
    FileSystem *pFS = pDb->pFS;   /* File system to read from */
    Page *pPg = 0;                /* Main run page */
    BtreeCursor *pCsr = 0;        /* Btree cursor */

    rc = btreeCursorNew(pDb, pSeg, &pCsr);
    if( rc==LSM_OK ){
      rc = btreeCursorFirst(pCsr);