SQLite4
Changes On Branch block-redirects
Not logged in

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

Changes In Branch block-redirects Excluding Merge-Ins

This is equivalent to a diff from 4d1b506594 to 520f3729b8

2013-01-28
08:56
Merge block-redirects branch with trunk. This changes the lsm file format. check-in: 647229e983 user: dan tags: trunk
2013-01-26
20:18
Handle calls to lsm_work() with the nPage parameter set to not greater than zero. Remove some dead code. Leaf check-in: 520f3729b8 user: dan tags: block-redirects
19:17
Fix problems with redirected blocks in compressed databases. check-in: 930b7e4507 user: dan tags: block-redirects
2013-01-20
00:07
Enable the AUTOINCREMENT feature. check-in: 5442b20bf6 user: drh tags: trunk
2013-01-19
20:07
Change the lsm file-format to allow a small number of blocks belonging to the oldest segment in the database to be moved without modifying their content. This makes it easier to compact a database file to the minimum required size. check-in: 09251cee6a user: dan tags: block-redirects
19:49
Insert a value in place of NULL in an INTEGER PRIMARY KEY, even if the IPK column is omitted from the VALUES list in the INSERT statement. check-in: 4d1b506594 user: drh tags: trunk
16:14
Inserting NULL into a INTEGER PRIMARY KEY fills that key with the next available integer value. The sqlite4_last_insert_rowid() function now works for those cases. check-in: 697ee9faad user: drh tags: trunk

Changes to lsm-test/README.

24
25
26
27
28
29
30





31
32
33
34
              the system recovers and other clients proceed unaffected if
              a process fails in the middle of a write transaction.

              The difference from lsmtest2.c is that this file tests
              live-recovery (recovery from a failure that occurs while other
              clients are still running) whereas lsmtest2.c tests recovery
              from a system or power failure.
















>
>
>
>
>




24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
              the system recovers and other clients proceed unaffected if
              a process fails in the middle of a write transaction.

              The difference from lsmtest2.c is that this file tests
              live-recovery (recovery from a failure that occurs while other
              clients are still running) whereas lsmtest2.c tests recovery
              from a system or power failure.

  lsmtest9.c: More data tests. These focus on testing that calling
              lsm_work(nMerge=1) to compact the database does not corrupt it.
              In other words, that databases containing block-redirects
              can be read and written.




Changes to lsm-test/lsmtest.h.

215
216
217
218
219
220
221



222
223
224
225
226
227
228
229
230
231
232
233
234
235
236



237
238
239
240
241
242
243
244
245
246
247
248
void test_data_1(const char *, const char *, int *pRc);
void test_data_2(const char *, const char *, int *pRc);
void test_data_3(const char *, const char *, int *pRc);
void testDbContents(TestDb *, Datasource *, int, int, int, int, int, int *);
void testCaseProgress(int, int, int, int *);
int testCaseNDot(void);




typedef struct CksumDb CksumDb;
CksumDb *testCksumArrayNew(Datasource *, int, int, int);
char *testCksumArrayGet(CksumDb *, int);
void testCksumArrayFree(CksumDb *);
void testCaseStart(int *pRc, char *zFmt, ...);
void testCaseFinish(int rc);
void testCaseSkip(void);
int testCaseBegin(int *, const char *, const char *, ...);

#define TEST_CKSUM_BYTES 29
int testCksumDatabase(TestDb *pDb, char *zOut);
int testCountDatabase(TestDb *pDb);
void testCompareInt(int, int, int *);
void testCompareStr(const char *z1, const char *z2, int *pRc);





/*
** Similar to the Tcl_GetIndexFromObjStruct() Tcl library function.
*/
#define testArgSelect(w,x,y,z) testArgSelectX(w,x,sizeof(w[0]),y,z)
int testArgSelectX(void *, const char *, int, const char *, int *);

#ifdef __cplusplus
}  /* End of the 'extern "C"' block */
#endif

#endif







>
>
>















>
>
>












215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
void test_data_1(const char *, const char *, int *pRc);
void test_data_2(const char *, const char *, int *pRc);
void test_data_3(const char *, const char *, int *pRc);
void testDbContents(TestDb *, Datasource *, int, int, int, int, int, int *);
void testCaseProgress(int, int, int, int *);
int testCaseNDot(void);

void testCompareDb(Datasource *, int, int, TestDb *, TestDb *, int *);
int testControlDb(TestDb **ppDb);

typedef struct CksumDb CksumDb;
CksumDb *testCksumArrayNew(Datasource *, int, int, int);
char *testCksumArrayGet(CksumDb *, int);
void testCksumArrayFree(CksumDb *);
void testCaseStart(int *pRc, char *zFmt, ...);
void testCaseFinish(int rc);
void testCaseSkip(void);
int testCaseBegin(int *, const char *, const char *, ...);

#define TEST_CKSUM_BYTES 29
int testCksumDatabase(TestDb *pDb, char *zOut);
int testCountDatabase(TestDb *pDb);
void testCompareInt(int, int, int *);
void testCompareStr(const char *z1, const char *z2, int *pRc);

/* lsmtest9.c */
void test_data_4(const char *, const char *, int *pRc);


/*
** Similar to the Tcl_GetIndexFromObjStruct() Tcl library function.
*/
#define testArgSelect(w,x,y,z) testArgSelectX(w,x,sizeof(w[0]),y,z)
int testArgSelectX(void *, const char *, int, const char *, int *);

#ifdef __cplusplus
}  /* End of the 'extern "C"' block */
#endif

#endif

Changes to lsm-test/lsmtest1.c.

88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
  zRet = testMallocPrintf("data.%s.%s.%d.%d", 
      zSystem, zData, pTest->nRow, pTest->nVerify
  );
  testFree(zData);
  return zRet;
}

static int testControlDb(TestDb **ppDb){
#ifdef HAVE_KYOTOCABINET
  return tdb_open("kyotocabinet", "tmp.db", 1, ppDb);
#else
  return tdb_open("sqlite3", ":memory:", 1, ppDb);
#endif
}








|







88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
  zRet = testMallocPrintf("data.%s.%s.%d.%d", 
      zSystem, zData, pTest->nRow, pTest->nVerify
  );
  testFree(zData);
  return zRet;
}

int testControlDb(TestDb **ppDb){
#ifdef HAVE_KYOTOCABINET
  return tdb_open("kyotocabinet", "tmp.db", 1, ppDb);
#else
  return tdb_open("sqlite3", ":memory:", 1, ppDb);
#endif
}

344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
    if( testCaseBegin(pRc, zPattern, "%s", zName) ){
      doDataTest1(zSystem, &aTest[i], pRc);
    }
    testFree(zName);
  }
}

static void testCompareDb(
  Datasource *pData,
  int nData,
  int iSeed,
  TestDb *pControl,
  TestDb *pDb,
  int *pRc
){







|







344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
    if( testCaseBegin(pRc, zPattern, "%s", zName) ){
      doDataTest1(zSystem, &aTest[i], pRc);
    }
    testFree(zName);
  }
}

void testCompareDb(
  Datasource *pData,
  int nData,
  int iSeed,
  TestDb *pControl,
  TestDb *pDb,
  int *pRc
){

Added lsm-test/lsmtest9.c.





















































































































































































































































































>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138

#include "lsmtest.h"

#define DATA_SEQUENTIAL TEST_DATASOURCE_SEQUENCE
#define DATA_RANDOM     TEST_DATASOURCE_RANDOM

typedef struct Datatest4 Datatest4;

/*
** Test overview:
**
**   1. Insert (Datatest4.nRec) records into a database.
**
**   2. Repeat (Datatest4.nRepeat) times:
**
**      2a. Delete 2/3 of the records in the database.
**
**      2b. Run lsm_work(nMerge=1).
**
**      2c. Insert as many records as were deleted in 2a.
**
**      2d. Check database content is as expected.
**
**      2e. If (Datatest4.bReopen) is true, close and reopen the database.
*/
struct Datatest4 {
  /* Datasource definition */
  DatasourceDefn defn;

  int nRec;
  int nRepeat;
  int bReopen;
};

static void doDataTest4(
  const char *zSystem,            /* Database system to test */
  Datatest4 *p,                   /* Structure containing test parameters */
  int *pRc                        /* OUT: Error code */
){
  lsm_db *db = 0;
  TestDb *pDb;
  TestDb *pControl;
  Datasource *pData;
  int i;
  int rc = 0;
  int iDot = 0;

  int nRecOn3 = (p->nRec / 3);
  int iData = 0;

  /* Start the test case, open a database and allocate the datasource. */
  rc = testControlDb(&pControl);
  pDb = testOpen(zSystem, 1, &rc);
  pData = testDatasourceNew(&p->defn);
  if( rc==0 ) db = tdb_lsm(pDb);

  testWriteDatasourceRange(pControl, pData, iData, nRecOn3*3, &rc);
  testWriteDatasourceRange(pDb,      pData, iData, nRecOn3*3, &rc);

  for(i=0; rc==0 && i<p->nRepeat; i++){

    testDeleteDatasourceRange(pControl, pData, iData, nRecOn3*2, &rc);
    testDeleteDatasourceRange(pDb,      pData, iData, nRecOn3*2, &rc);

    if( db ){
      int nDone;
#if 0
      fprintf(stderr, "lsm_work() start...\n"); fflush(stderr);
#endif
      do {
        nDone = 0;
        rc = lsm_work(db, 1, (1<<30), &nDone);
      }while( rc==0 && nDone>0 );
#if 0 
      fprintf(stderr, "lsm_work() done...\n"); fflush(stderr);
#endif
    }

if( i+1<p->nRepeat ){
    iData += (nRecOn3*2);
    testWriteDatasourceRange(pControl, pData, iData+nRecOn3, nRecOn3*2, &rc);
    testWriteDatasourceRange(pDb,      pData, iData+nRecOn3, nRecOn3*2, &rc);

    testCompareDb(pData, nRecOn3*3, iData, pControl, pDb, &rc);

    /* If Datatest4.bReopen is true, close and reopen the database */
    if( p->bReopen ){
      testReopen(&pDb, &rc);
      if( rc==0 ) db = tdb_lsm(pDb);
    }
}

    /* Update the progress dots... */
    testCaseProgress(i, p->nRepeat, testCaseNDot(), &iDot);
  }

  testClose(&pDb);
  testClose(&pControl);
  testDatasourceFree(pData);
  testCaseFinish(rc);
  *pRc = rc;
}

static char *getName4(const char *zSystem, Datatest4 *pTest){
  char *zRet;
  char *zData;
  zData = testDatasourceName(&pTest->defn);
  zRet = testMallocPrintf("data4.%s.%s.%d.%d.%d", 
      zSystem, zData, pTest->nRec, pTest->nRepeat, pTest->bReopen
  );
  testFree(zData);
  return zRet;
}

void test_data_4(
  const char *zSystem,            /* Database system name */
  const char *zPattern,           /* Run test cases that match this pattern */
  int *pRc                        /* IN/OUT: Error code */
){
  Datatest4 aTest[] = {
      /* defn,                                 nRec, nRepeat, bReopen */
    { {DATA_RANDOM,     20,25,     500,600}, 10000,      10,       0   },
    { {DATA_RANDOM,     20,25,     500,600}, 10000,      10,       1   },
  };

  int i;

  for(i=0; *pRc==LSM_OK && i<ArraySize(aTest); i++){
    char *zName = getName4(zSystem, &aTest[i]);
    if( testCaseBegin(pRc, zPattern, "%s", zName) ){
      doDataTest4(zSystem, &aTest[i], pRc);
    }
    testFree(zName);
  }
}



Changes to lsm-test/lsmtest_func.c.

42
43
44
45
46
47
48





49
50
51
52
53
54
55
  if( rc!=LSM_OK ){
    testPrintError("lsm_open(): rc=%d\n", rc);
  }else{
    rc = lsm_open(pDb, zDb);
    if( rc!=LSM_OK ){
      testPrintError("lsm_open(): rc=%d\n", rc);
    }else{





      rc = lsm_work(pDb, nMerge, nWork, 0);
      if( rc!=LSM_OK ){
        testPrintError("lsm_work(): rc=%d\n", rc);
      }
    }
  }
  if( rc==LSM_OK ){







>
>
>
>
>







42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
  if( rc!=LSM_OK ){
    testPrintError("lsm_open(): rc=%d\n", rc);
  }else{
    rc = lsm_open(pDb, zDb);
    if( rc!=LSM_OK ){
      testPrintError("lsm_open(): rc=%d\n", rc);
    }else{
      int n = -1;
      lsm_config(pDb, LSM_CONFIG_BLOCK_SIZE, &n);
      n = n*2;
      lsm_config(pDb, LSM_CONFIG_AUTOCHECKPOINT, &n);

      rc = lsm_work(pDb, nMerge, nWork, 0);
      if( rc!=LSM_OK ){
        testPrintError("lsm_work(): rc=%d\n", rc);
      }
    }
  }
  if( rc==LSM_OK ){

Changes to lsm-test/lsmtest_main.c.

459
460
461
462
463
464
465

466
467
468
469
470
471
472

  for(j=0; tdb_system_name(j); j++){
    rc = 0;

    test_data_1(tdb_system_name(j), zPattern, &rc);
    test_data_2(tdb_system_name(j), zPattern, &rc);
    test_data_3(tdb_system_name(j), zPattern, &rc);

    test_rollback(tdb_system_name(j), zPattern, &rc);
    test_mc(tdb_system_name(j), zPattern, &rc);
    test_mt(tdb_system_name(j), zPattern, &rc);

    if( rc ) nFail++;
  }








>







459
460
461
462
463
464
465
466
467
468
469
470
471
472
473

  for(j=0; tdb_system_name(j); j++){
    rc = 0;

    test_data_1(tdb_system_name(j), zPattern, &rc);
    test_data_2(tdb_system_name(j), zPattern, &rc);
    test_data_3(tdb_system_name(j), zPattern, &rc);
    test_data_4(tdb_system_name(j), zPattern, &rc);
    test_rollback(tdb_system_name(j), zPattern, &rc);
    test_mc(tdb_system_name(j), zPattern, &rc);
    test_mt(tdb_system_name(j), zPattern, &rc);

    if( rc ) nFail++;
  }

Changes to lsm-test/lsmtest_tdb3.c.

984
985
986
987
988
989
990

991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
}

int test_lsm_lomem_open(
  const char *zFilename, 
  int bClear, 
  TestDb **ppDb
){

  const char *zCfg = 
    "page_size=256 block_size=65536 autoflush=16384 "
    "max_freelist=2 autocheckpoint=32768 "
    "mmap=0 "
  ;
  return testLsmOpen(zCfg, zFilename, bClear, ppDb);
}

int test_lsm_zip_open(
  const char *zFilename, 
  int bClear, 
  TestDb **ppDb
){
  const char *zCfg = 
    "page_size=256 block_size=65536 autoflush=16384 "
    "max_freelist=2 autocheckpoint=32768 compression=1"
    "mmap=0 "
  ;
  return testLsmOpen(zCfg, zFilename, bClear, ppDb);
}

lsm_db *tdb_lsm(TestDb *pDb){
  if( pDb->pMethods->xClose==test_lsm_close ){
    return ((LsmDb *)pDb)->db;







>


|












|
<







984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007

1008
1009
1010
1011
1012
1013
1014
}

int test_lsm_lomem_open(
  const char *zFilename, 
  int bClear, 
  TestDb **ppDb
){
    /* "max_freelist=4 autocheckpoint=32768 " */
  const char *zCfg = 
    "page_size=256 block_size=65536 autoflush=16384 "
    "autocheckpoint=32768 "
    "mmap=0 "
  ;
  return testLsmOpen(zCfg, zFilename, bClear, ppDb);
}

int test_lsm_zip_open(
  const char *zFilename, 
  int bClear, 
  TestDb **ppDb
){
  const char *zCfg = 
    "page_size=256 block_size=65536 autoflush=16384 "
    "autocheckpoint=32768 compression=1 mmap=0 "

  ;
  return testLsmOpen(zCfg, zFilename, bClear, ppDb);
}

lsm_db *tdb_lsm(TestDb *pDb){
  if( pDb->pMethods->xClose==test_lsm_close ){
    return ((LsmDb *)pDb)->db;

Changes to main.mk.

290
291
292
293
294
295
296

297
298
299
300
301
302
303
EXTHDR += \
  $(TOP)/ext/icu/sqliteicu.h

LSMTESTSRC = $(TOP)/lsm-test/lsmtest1.c $(TOP)/lsm-test/lsmtest2.c           \
             $(TOP)/lsm-test/lsmtest3.c $(TOP)/lsm-test/lsmtest4.c           \
             $(TOP)/lsm-test/lsmtest5.c $(TOP)/lsm-test/lsmtest6.c           \
             $(TOP)/lsm-test/lsmtest7.c $(TOP)/lsm-test/lsmtest8.c           \

             $(TOP)/lsm-test/lsmtest_datasource.c \
             $(TOP)/lsm-test/lsmtest_func.c $(TOP)/lsm-test/lsmtest_io.c     \
             $(TOP)/lsm-test/lsmtest_main.c $(TOP)/lsm-test/lsmtest_mem.c    \
             $(TOP)/lsm-test/lsmtest_tdb.c $(TOP)/lsm-test/lsmtest_tdb3.c    \
             $(TOP)/lsm-test/lsmtest_util.c 

LSMTESTHDR = $(TOP)/lsm-test/lsmtest.h $(TOP)/lsm-test/lsmtest_tdb.h







>







290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
EXTHDR += \
  $(TOP)/ext/icu/sqliteicu.h

LSMTESTSRC = $(TOP)/lsm-test/lsmtest1.c $(TOP)/lsm-test/lsmtest2.c           \
             $(TOP)/lsm-test/lsmtest3.c $(TOP)/lsm-test/lsmtest4.c           \
             $(TOP)/lsm-test/lsmtest5.c $(TOP)/lsm-test/lsmtest6.c           \
             $(TOP)/lsm-test/lsmtest7.c $(TOP)/lsm-test/lsmtest8.c           \
             $(TOP)/lsm-test/lsmtest9.c                                      \
             $(TOP)/lsm-test/lsmtest_datasource.c \
             $(TOP)/lsm-test/lsmtest_func.c $(TOP)/lsm-test/lsmtest_io.c     \
             $(TOP)/lsm-test/lsmtest_main.c $(TOP)/lsm-test/lsmtest_mem.c    \
             $(TOP)/lsm-test/lsmtest_tdb.c $(TOP)/lsm-test/lsmtest_tdb3.c    \
             $(TOP)/lsm-test/lsmtest_util.c 

LSMTESTHDR = $(TOP)/lsm-test/lsmtest.h $(TOP)/lsm-test/lsmtest_tdb.h

Changes to src/lsmInt.h.

79
80
81
82
83
84
85

86
87
88
89
90
91
92
typedef struct LsmString LsmString;
typedef struct Mempool Mempool;
typedef struct Merge Merge;
typedef struct MergeInput MergeInput;
typedef struct MetaPage MetaPage;
typedef struct MultiCursor MultiCursor;
typedef struct Page Page;

typedef struct Segment Segment;
typedef struct SegmentMerger SegmentMerger;
typedef struct ShmChunk ShmChunk;
typedef struct ShmHeader ShmHeader;
typedef struct ShmReader ShmReader;
typedef struct Snapshot Snapshot;
typedef struct TransMark TransMark;







>







79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
typedef struct LsmString LsmString;
typedef struct Mempool Mempool;
typedef struct Merge Merge;
typedef struct MergeInput MergeInput;
typedef struct MetaPage MetaPage;
typedef struct MultiCursor MultiCursor;
typedef struct Page Page;
typedef struct Redirect Redirect;
typedef struct Segment Segment;
typedef struct SegmentMerger SegmentMerger;
typedef struct ShmChunk ShmChunk;
typedef struct ShmHeader ShmHeader;
typedef struct ShmReader ShmReader;
typedef struct Snapshot Snapshot;
typedef struct TransMark TransMark;
139
140
141
142
143
144
145


146
147
148
149
150
151
152

/*
** Hard limit on the number of free-list entries that may be stored in 
** a checkpoint (the remainder are stored as a system record in the LSM).
** See also LSM_CONFIG_MAX_FREELIST.
*/
#define LSM_MAX_FREELIST_ENTRIES 24



#define LSM_ATTEMPTS_BEFORE_PROTOCOL 10000


/*
** Each entry stored in the LSM (or in-memory tree structure) has an
** associated mask of the following flags.







>
>







140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155

/*
** Hard limit on the number of free-list entries that may be stored in 
** a checkpoint (the remainder are stored as a system record in the LSM).
** See also LSM_CONFIG_MAX_FREELIST.
*/
#define LSM_MAX_FREELIST_ENTRIES 24

#define LSM_MAX_BLOCK_REDIRECTS 16

#define LSM_ATTEMPTS_BEFORE_PROTOCOL 10000


/*
** Each entry stored in the LSM (or in-memory tree structure) has an
** associated mask of the following flags.
186
187
188
189
190
191
192








193
194
195
196
197
198
199
*/
typedef struct IntArray IntArray;
struct IntArray {
  int nAlloc;
  int nArray;
  u32 *aArray;
};









/*
** An instance of this structure represents a point in the history of the
** tree structure to roll back to. Refer to comments in lsm_tree.c for 
** details.
*/
struct TreeMark {







>
>
>
>
>
>
>
>







189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
*/
typedef struct IntArray IntArray;
struct IntArray {
  int nAlloc;
  int nArray;
  u32 *aArray;
};

struct Redirect {
  int n;                          /* Number of redirects */
  struct RedirectEntry {
    int iFrom;
    int iTo;
  } *a;
};

/*
** An instance of this structure represents a point in the history of the
** tree structure to roll back to. Refer to comments in lsm_tree.c for 
** details.
*/
struct TreeMark {
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325

326
327
328
329
330
331
332
  lsm_compress compress;          /* Compression callbacks */

  /* Sub-system handles */
  FileSystem *pFS;                /* On-disk portion of database */
  Database *pDatabase;            /* Database shared data */

  /* Client transaction context */
  Snapshot *pClient;              /* Client snapshot (non-NULL in read trans) */
  int iReader;                    /* Read lock held (-1 == unlocked) */
  MultiCursor *pCsr;              /* List of all open cursors */
  LogWriter *pLogWriter;          /* Context for writing to the log file */
  int nTransOpen;                 /* Number of opened write transactions */
  int nTransAlloc;                /* Allocated size of aTrans[] array */
  TransMark *aTrans;              /* Array of marks for transaction rollback */
  IntArray rollback;              /* List of tree-nodes to roll back */

  /* Worker context */
  Snapshot *pWorker;              /* Worker snapshot (or NULL) */
  Freelist *pFreelist;            /* See sortedNewToplevel() */
  int bUseFreelist;               /* True to use pFreelist */


  /* Debugging message callback */
  void (*xLog)(void *, int, const char *);
  void *pLogCtx;

  /* Work done notification callback */
  void (*xWork)(lsm_db *, void *);







|












>







317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
  lsm_compress compress;          /* Compression callbacks */

  /* Sub-system handles */
  FileSystem *pFS;                /* On-disk portion of database */
  Database *pDatabase;            /* Database shared data */

  /* Client transaction context */
  Snapshot *pClient;              /* Client snapshot */
  int iReader;                    /* Read lock held (-1 == unlocked) */
  MultiCursor *pCsr;              /* List of all open cursors */
  LogWriter *pLogWriter;          /* Context for writing to the log file */
  int nTransOpen;                 /* Number of opened write transactions */
  int nTransAlloc;                /* Allocated size of aTrans[] array */
  TransMark *aTrans;              /* Array of marks for transaction rollback */
  IntArray rollback;              /* List of tree-nodes to roll back */

  /* Worker context */
  Snapshot *pWorker;              /* Worker snapshot (or NULL) */
  Freelist *pFreelist;            /* See sortedNewToplevel() */
  int bUseFreelist;               /* True to use pFreelist */
  int bIncrMerge;                 /* True if currently doing a merge */

  /* Debugging message callback */
  void (*xLog)(void *, int, const char *);
  void *pLogCtx;

  /* Work done notification callback */
  void (*xWork)(lsm_db *, void *);
343
344
345
346
347
348
349


350
351
352
353
354
355
356
};

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 */


};

/*
** iSplitTopic/pSplitKey/nSplitKey:
**   If nRight>0, this buffer contains a copy of the largest key that has
**   already been written to the left-hand-side of the level.
*/







>
>







355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
};

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:
**   If nRight>0, this buffer contains a copy of the largest key that has
**   already been written to the left-hand-side of the level.
*/
510
511
512
513
514
515
516

517
518
519
520
521
522
523
** Database below for futher details.
*/
struct Snapshot {
  Database *pDatabase;            /* Database this snapshot belongs to */
  Level *pLevel;                  /* Pointer to level 0 of snapshot (or NULL) */
  i64 iId;                        /* Snapshot id */
  i64 iLogOff;                    /* Log file offset */


  /* 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 */
};







>







524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
** Database below for futher details.
*/
struct Snapshot {
  Database *pDatabase;            /* Database this snapshot belongs to */
  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 */
};
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664

int lsmFsSectorSize(FileSystem *);

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

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







|







665
666
667
668
669
670
671
672
673
674
675
676
677
678
679

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 *);
704
705
706
707
708
709
710



711
712
713
714
715
716
717
int lsmEnvShmMap(lsm_env *, lsm_file *, int, int, void **); 
void lsmEnvShmBarrier(lsm_env *);
void lsmEnvShmUnmap(lsm_env *, lsm_file *, int);

void lsmEnvSleep(lsm_env *, int);

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




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

/* 
** Functions from file "lsm_sorted.c".







>
>
>







719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
int lsmEnvShmMap(lsm_env *, lsm_file *, int, int, void **); 
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 *);
Pgno lsmFsRedirectPage(FileSystem *, Redirect *, Pgno);

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

/* 
** Functions from file "lsm_sorted.c".
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
void lsmSnapshotSetCkptid(Snapshot *, i64);

Level *lsmDbSnapshotLevel(Snapshot *);
void lsmDbSnapshotSetLevel(Snapshot *, Level *);

void lsmDbRecoveryComplete(lsm_db *, int);

int lsmBlockAllocate(lsm_db *, int *);
int lsmBlockFree(lsm_db *, int);
int lsmBlockRefree(lsm_db *, int);

void lsmFreelistDeltaBegin(lsm_db *);
void lsmFreelistDeltaEnd(lsm_db *);
int lsmFreelistDelta(lsm_db *pDb);








|







841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
void lsmSnapshotSetCkptid(Snapshot *, i64);

Level *lsmDbSnapshotLevel(Snapshot *);
void lsmDbSnapshotSetLevel(Snapshot *, Level *);

void lsmDbRecoveryComplete(lsm_db *, int);

int lsmBlockAllocate(lsm_db *, int, int *);
int lsmBlockFree(lsm_db *, int);
int lsmBlockRefree(lsm_db *, int);

void lsmFreelistDeltaBegin(lsm_db *);
void lsmFreelistDeltaEnd(lsm_db *);
int lsmFreelistDelta(lsm_db *pDb);

Changes to src/lsm_ckpt.c.

63
64
65
66
67
68
69







70
71
72
73
74
75
76
**     6. For each segment in the merge:
**        5a. Page number of next cell to read during merge (this field
**            is 64-bits - 2 integers)
**        5b. Cell number of next cell to read during merge
**     7. Page containing current split-key (64-bits - 2 integers).
**     8. Cell within page containing current split-key.
**     9. Current pointer value (64-bits - 2 integers).







**
**   The in-memory freelist entries. Each entry is either an insert or a
**   delete. The in-memory freelist is to the free-block-list as the
**   in-memory tree is to the users database content.
**
**     1. Number of free-list entries stored in checkpoint header.
**     2. Number of free blocks (in total).







>
>
>
>
>
>
>







63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
**     6. For each segment in the merge:
**        5a. Page number of next cell to read during merge (this field
**            is 64-bits - 2 integers)
**        5b. Cell number of next cell to read during merge
**     7. Page containing current split-key (64-bits - 2 integers).
**     8. Cell within page containing current split-key.
**     9. Current pointer value (64-bits - 2 integers).
**
**   The block redirect array:
**
**     1. Number of redirections (maximum LSM_MAX_BLOCK_REDIRECTS).
**     2. For each redirection:
**        a. "from" block number
**        b. "to" block number
**
**   The in-memory freelist entries. Each entry is either an insert or a
**   delete. The in-memory freelist is to the free-block-list as the
**   in-memory tree is to the users database content.
**
**     1. Number of free-list entries stored in checkpoint header.
**     2. Number of free blocks (in total).
415
416
417
418
419
420
421







422
423
424
425
426
427
428

  /* Serialize nLevel levels. */
  iLevel = 0;
  for(pLevel=lsmDbSnapshotLevel(pSnap); iLevel<nLevel; pLevel=pLevel->pNext){
    ckptExportLevel(pLevel, &ckpt, &iOut, &rc);
    iLevel++;
  }








  /* Write the freelist */
  assert( pSnap->freelist.nEntry<=pDb->nMaxFreelist );
  if( rc==LSM_OK ){
    int nFree = pSnap->freelist.nEntry;
    ckptSetValue(&ckpt, iOut++, nFree, &rc);
    for(i=0; i<nFree; i++){







>
>
>
>
>
>
>







422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442

  /* Serialize nLevel levels. */
  iLevel = 0;
  for(pLevel=lsmDbSnapshotLevel(pSnap); iLevel<nLevel; pLevel=pLevel->pNext){
    ckptExportLevel(pLevel, &ckpt, &iOut, &rc);
    iLevel++;
  }

  /* Write the block-redirect list */
  ckptSetValue(&ckpt, iOut++, pSnap->redirect.n, &rc);
  for(i=0; i<pSnap->redirect.n; i++){
    ckptSetValue(&ckpt, iOut++, pSnap->redirect.a[i].iFrom, &rc);
    ckptSetValue(&ckpt, iOut++, pSnap->redirect.a[i].iTo, &rc);
  }

  /* Write the freelist */
  assert( pSnap->freelist.nEntry<=pDb->nMaxFreelist );
  if( rc==LSM_OK ){
    int nFree = pSnap->freelist.nEntry;
    ckptSetValue(&ckpt, iOut++, nFree, &rc);
    for(i=0; i<nFree; i++){
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
      return LSM_PROTOCOL;
    }
  }

  rc = lsmCheckpointDeserialize(pDb, 1, pShm->aSnap1, &pDb->pWorker);
  if( pDb->pWorker ) pDb->pWorker->pDatabase = pDb->pDatabase;


  assert( rc!=LSM_OK || lsmFsIntegrityCheck(pDb) );

  return rc;
}

int lsmCheckpointDeserialize(
  lsm_db *pDb, 
  int bInclFreelist,              /* If true, deserialize free-list */
  u32 *aCkpt, 
  Snapshot **ppSnap
){
  int rc = LSM_OK;
  Snapshot *pNew;

  pNew = (Snapshot *)lsmMallocZeroRc(pDb->pEnv, sizeof(Snapshot), &rc);
  if( rc==LSM_OK ){

    int nFree;
    int i;
    int nLevel = (int)aCkpt[CKPT_HDR_NLEVEL];
    int iIn = CKPT_HDR_SIZE + CKPT_APPENDLIST_SIZE + CKPT_LOGPTR_SIZE;

    pNew->iId = lsmCheckpointId(aCkpt, 0);
    pNew->nBlock = aCkpt[CKPT_HDR_NBLOCK];
    pNew->nWrite = aCkpt[CKPT_HDR_NWRITE];
    rc = ckptLoadLevels(pDb, aCkpt, &iIn, nLevel, &pNew->pLevel);
    pNew->iLogOff = lsmCheckpointLogOffset(aCkpt);

    /* Make a copy of the append-list */
    for(i=0; i<LSM_APPLIST_SZ; i++){
      u32 *a = &aCkpt[CKPT_HDR_SIZE + CKPT_LOGPTR_SIZE + i*2];
      pNew->aiAppend[i] = ckptRead64(a);
    }





















    /* Copy the free-list */
    if( bInclFreelist ){
      nFree = aCkpt[iIn++];
      if( nFree ){
        pNew->freelist.aEntry = (FreelistEntry *)lsmMallocZeroRc(
            pDb->pEnv, sizeof(FreelistEntry)*nFree, &rc
        );
        if( rc==LSM_OK ){
          int i;







>

>














>

















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

|







918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
      return LSM_PROTOCOL;
    }
  }

  rc = lsmCheckpointDeserialize(pDb, 1, pShm->aSnap1, &pDb->pWorker);
  if( pDb->pWorker ) pDb->pWorker->pDatabase = pDb->pDatabase;

#if 0
  assert( rc!=LSM_OK || lsmFsIntegrityCheck(pDb) );
#endif
  return rc;
}

int lsmCheckpointDeserialize(
  lsm_db *pDb, 
  int bInclFreelist,              /* If true, deserialize free-list */
  u32 *aCkpt, 
  Snapshot **ppSnap
){
  int rc = LSM_OK;
  Snapshot *pNew;

  pNew = (Snapshot *)lsmMallocZeroRc(pDb->pEnv, sizeof(Snapshot), &rc);
  if( rc==LSM_OK ){
    Level *pLvl;
    int nFree;
    int i;
    int nLevel = (int)aCkpt[CKPT_HDR_NLEVEL];
    int iIn = CKPT_HDR_SIZE + CKPT_APPENDLIST_SIZE + CKPT_LOGPTR_SIZE;

    pNew->iId = lsmCheckpointId(aCkpt, 0);
    pNew->nBlock = aCkpt[CKPT_HDR_NBLOCK];
    pNew->nWrite = aCkpt[CKPT_HDR_NWRITE];
    rc = ckptLoadLevels(pDb, aCkpt, &iIn, nLevel, &pNew->pLevel);
    pNew->iLogOff = lsmCheckpointLogOffset(aCkpt);

    /* Make a copy of the append-list */
    for(i=0; i<LSM_APPLIST_SZ; i++){
      u32 *a = &aCkpt[CKPT_HDR_SIZE + CKPT_LOGPTR_SIZE + i*2];
      pNew->aiAppend[i] = ckptRead64(a);
    }

    /* Read the block-redirect list */
    pNew->redirect.n = aCkpt[iIn++];
    if( pNew->redirect.n ){
      pNew->redirect.a = lsmMallocZeroRc(pDb->pEnv, 
          (sizeof(struct RedirectEntry) * LSM_MAX_BLOCK_REDIRECTS), &rc
      );
      if( rc==LSM_OK ){
        for(i=0; i<pNew->redirect.n; i++){
          pNew->redirect.a[i].iFrom = aCkpt[iIn++];
          pNew->redirect.a[i].iTo = aCkpt[iIn++];
        }
      }
      for(pLvl=pNew->pLevel; pLvl->pNext; pLvl=pLvl->pNext);
      if( pLvl->nRight ){
        pLvl->aRhs[pLvl->nRight-1].pRedirect = &pNew->redirect;
      }else{
        pLvl->lhs.pRedirect = &pNew->redirect;
      }
    }

    /* Copy the free-list */
    if( rc==LSM_OK && bInclFreelist ){
      nFree = aCkpt[iIn++];
      if( nFree ){
        pNew->freelist.aEntry = (FreelistEntry *)lsmMallocZeroRc(
            pDb->pEnv, sizeof(FreelistEntry)*nFree, &rc
        );
        if( rc==LSM_OK ){
          int i;
1009
1010
1011
1012
1013
1014
1015

1016
1017
1018
1019
1020
1021
1022
1023
int lsmCheckpointSaveWorker(lsm_db *pDb, int bFlush){
  Snapshot *pSnap = pDb->pWorker;
  ShmHeader *pShm = pDb->pShmhdr;
  void *p = 0;
  int n = 0;
  int rc;


  rc = ckptExportSnapshot(pDb, bFlush, pSnap->iId+1, 1, &p, &n);
  if( rc!=LSM_OK ) return rc;
  assert( ckptChecksumOk((u32 *)p) );

  assert( n<=LSM_META_PAGE_SIZE );
  memcpy(pShm->aSnap2, p, n);
  lsmShmBarrier(pDb);
  memcpy(pShm->aSnap1, p, n);







>
|







1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
int lsmCheckpointSaveWorker(lsm_db *pDb, int bFlush){
  Snapshot *pSnap = pDb->pWorker;
  ShmHeader *pShm = pDb->pShmhdr;
  void *p = 0;
  int n = 0;
  int rc;

  pSnap->iId++;
  rc = ckptExportSnapshot(pDb, bFlush, pSnap->iId, 1, &p, &n);
  if( rc!=LSM_OK ) return rc;
  assert( ckptChecksumOk((u32 *)p) );

  assert( n<=LSM_META_PAGE_SIZE );
  memcpy(pShm->aSnap2, p, n);
  lsmShmBarrier(pDb);
  memcpy(pShm->aSnap1, p, n);

Changes to src/lsm_file.c.

875
876
877
878
879
880
881
882






































883
884
885
886
887
888
889
890
891

892
893
894
895







896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915




916
917
918
919
920
921
922
923
924
925
926
927
928
929
930

931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965

966
967
968
969
970
971
972
973
974
975
976
977
978

979
980
981
982
983
984
985
986
    i64 nMin = (i64)nBlock * (i64)pFS->nBlocksize;
    fsGrowMapping(pFS, nMin, &rc);
    if( rc!=LSM_OK ) return rc;
  }
  return lsmEnvSync(pFS->pEnv, pFS->fdDb);
}

static int fsPageGet(FileSystem *, 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.
*/
static int fsBlockNext(
  FileSystem *pFS,                /* File-system object handle */

  int iBlock,                     /* Read field from this block */
  int *piNext                     /* OUT: Next block in linked list */
){
  int rc;








  assert( pFS->bUseMmap==0 || pFS->pCompress==0 );
  if( pFS->pCompress ){
    i64 iOff;                     /* File offset to read data from */
    u8 aNext[4];                  /* 4-byte pointer read from db file */

    iOff = (i64)iBlock * pFS->nBlocksize - sizeof(aNext);
    rc = lsmEnvRead(pFS->pEnv, pFS->fdDb, iOff, aNext, sizeof(aNext));
    if( rc==LSM_OK ){
      *piNext = (int)lsmGetU32(aNext);
    }
  }else{
    const int nPagePerBlock = (pFS->nBlocksize / pFS->nPagesize);
    Page *pLast;
    rc = fsPageGet(pFS, iBlock*nPagePerBlock, 0, &pLast, 0);
    if( rc==LSM_OK ){
      *piNext = lsmGetU32(&pLast->aData[pFS->nPagesize-4]);
      lsmFsPageRelease(pLast);
    }
  }




  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));
}

/*
** This function is only called in compressed database mode.
*/
static int fsReadData(
  FileSystem *pFS,                /* File-system handle */

  i64 iOff,                       /* Read data from this offset */
  u8 *aData,                      /* Buffer to read data into */
  int nData                       /* Number of bytes to read */
){
  i64 iEob;                       /* End of block */
  int nRead;
  int rc;

  assert( pFS->pCompress );

  iEob = fsLastPageOnPagesBlock(pFS, iOff) + 1;
  nRead = LSM_MIN(iEob - iOff, nData);

  rc = lsmEnvRead(pFS->pEnv, pFS->fdDb, iOff, aData, nRead);
  if( rc==LSM_OK && nRead!=nData ){
    int iBlk;

    rc = fsBlockNext(pFS, fsPageToBlock(pFS, iOff), &iBlk);
    if( rc==LSM_OK ){
      i64 iOff2 = fsFirstPageOnBlock(pFS, iBlk);
      rc = lsmEnvRead(pFS->pEnv, pFS->fdDb, iOff2, &aData[nRead], nData-nRead);
    }
  }

  return rc;
}

/*
** Parameter iBlock is a database file block. This function reads the value 
** stored in the blocks "previous block" pointer and stores it in *piPrev.
** LSM_OK is returned if everything is successful, or an LSM error code
** otherwise.
*/
static int fsBlockPrev(
  FileSystem *pFS,                /* File-system object handle */

  int iBlock,                     /* Read field from this block */
  int *piPrev                     /* OUT: Previous block in linked list */
){
  int rc = LSM_OK;                /* Return code */

  assert( pFS->bUseMmap==0 || pFS->pCompress==0 );
  assert( iBlock>0 );

  if( pFS->pCompress ){
    i64 iOff = fsFirstPageOnBlock(pFS, iBlock) - 4;
    u8 aPrev[4];                  /* 4-byte pointer read from db file */
    rc = lsmEnvRead(pFS->pEnv, pFS->fdDb, iOff, aPrev, sizeof(aPrev));
    if( rc==LSM_OK ){

      *piPrev = (int)lsmGetU32(aPrev);
    }
  }else{
    assert( 0 );
  }
  return rc;
}








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









>




>
>
>
>
>
>
>






|







|





>
>
>
>















>

















|

















>













>
|







875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
    i64 nMin = (i64)nBlock * (i64)pFS->nBlocksize;
    fsGrowMapping(pFS, nMin, &rc);
    if( rc!=LSM_OK ) return rc;
  }
  return lsmEnvSync(pFS->pEnv, pFS->fdDb);
}

static int fsPageGet(FileSystem *, Segment *, Pgno, int, Page **, int *);

static int fsRedirectBlock(Redirect *p, int iBlk){
  if( p ){
    int i;
    for(i=0; i<p->n; i++){
      if( iBlk==p->a[i].iFrom ) return p->a[i].iTo;
    }
  }
  assert( iBlk!=0 );
  return iBlk;
}

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;
}

/*
** 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.
*/
static int fsBlockNext(
  FileSystem *pFS,                /* File-system object handle */
  Segment *pSeg,                  /* Use this segment for block redirects */
  int iBlock,                     /* Read field from this block */
  int *piNext                     /* OUT: Next block in linked list */
){
  int rc;
  int iRead;                      /* Read block from here */
  
  if( pSeg ){
    iRead = fsRedirectBlock(pSeg->pRedirect, iBlock);
  }else{
    iRead = iBlock;
  }

  assert( pFS->bUseMmap==0 || pFS->pCompress==0 );
  if( pFS->pCompress ){
    i64 iOff;                     /* File offset to read data from */
    u8 aNext[4];                  /* 4-byte pointer read from db file */

    iOff = (i64)iRead * pFS->nBlocksize - sizeof(aNext);
    rc = lsmEnvRead(pFS->pEnv, pFS->fdDb, iOff, aNext, sizeof(aNext));
    if( rc==LSM_OK ){
      *piNext = (int)lsmGetU32(aNext);
    }
  }else{
    const int nPagePerBlock = (pFS->nBlocksize / pFS->nPagesize);
    Page *pLast;
    rc = fsPageGet(pFS, 0, iRead*nPagePerBlock, 0, &pLast, 0);
    if( rc==LSM_OK ){
      *piNext = lsmGetU32(&pLast->aData[pFS->nPagesize-4]);
      lsmFsPageRelease(pLast);
    }
  }

  if( pSeg ){
    *piNext = fsRedirectBlock(pSeg->pRedirect, *piNext);
  }
  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));
}

/*
** This function is only called in compressed database mode.
*/
static int fsReadData(
  FileSystem *pFS,                /* File-system handle */
  Segment *pSeg,                  /* Block redirection */
  i64 iOff,                       /* Read data from this offset */
  u8 *aData,                      /* Buffer to read data into */
  int nData                       /* Number of bytes to read */
){
  i64 iEob;                       /* End of block */
  int nRead;
  int rc;

  assert( pFS->pCompress );

  iEob = fsLastPageOnPagesBlock(pFS, iOff) + 1;
  nRead = LSM_MIN(iEob - iOff, nData);

  rc = lsmEnvRead(pFS->pEnv, pFS->fdDb, iOff, aData, nRead);
  if( rc==LSM_OK && nRead!=nData ){
    int iBlk;

    rc = fsBlockNext(pFS, pSeg, fsPageToBlock(pFS, iOff), &iBlk);
    if( rc==LSM_OK ){
      i64 iOff2 = fsFirstPageOnBlock(pFS, iBlk);
      rc = lsmEnvRead(pFS->pEnv, pFS->fdDb, iOff2, &aData[nRead], nData-nRead);
    }
  }

  return rc;
}

/*
** Parameter iBlock is a database file block. This function reads the value 
** stored in the blocks "previous block" pointer and stores it in *piPrev.
** LSM_OK is returned if everything is successful, or an LSM error code
** otherwise.
*/
static int fsBlockPrev(
  FileSystem *pFS,                /* File-system object handle */
  Segment *pSeg,                  /* Use this segment for block redirects */
  int iBlock,                     /* Read field from this block */
  int *piPrev                     /* OUT: Previous block in linked list */
){
  int rc = LSM_OK;                /* Return code */

  assert( pFS->bUseMmap==0 || pFS->pCompress==0 );
  assert( iBlock>0 );

  if( pFS->pCompress ){
    i64 iOff = fsFirstPageOnBlock(pFS, iBlock) - 4;
    u8 aPrev[4];                  /* 4-byte pointer read from db file */
    rc = lsmEnvRead(pFS->pEnv, pFS->fdDb, iOff, aPrev, sizeof(aPrev));
    if( rc==LSM_OK ){
      Redirect *pRedir = (pSeg ? pSeg->pRedirect : 0);
      *piPrev = fsRedirectBlock(pRedir, (int)lsmGetU32(aPrev));
    }
  }else{
    assert( 0 );
  }
  return rc;
}

997
998
999
1000
1001
1002
1003
1004






1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022






1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
  nByte  = (aBuf[0] & 0x7F) << 14;
  nByte += (aBuf[1] & 0x7F) << 7;
  nByte += (aBuf[2] & 0x7F);
  *pbFree = !(aBuf[1] & 0x80);
  return nByte;
}

static int fsSubtractOffset(FileSystem *pFS, i64 iOff, int iSub, i64 *piRes){






  i64 iStart;
  int iBlk = 0;
  int rc;

  assert( pFS->pCompress );

  iStart = fsFirstPageOnBlock(pFS, fsPageToBlock(pFS, iOff));
  if( (iOff-iSub)>=iStart ){
    *piRes = (iOff-iSub);
    return LSM_OK;
  }

  rc = fsBlockPrev(pFS, fsPageToBlock(pFS, iOff), &iBlk);
  *piRes = fsLastPageOnBlock(pFS, iBlk) - iSub + (iOff - iStart + 1);
  return rc;
}

static int fsAddOffset(FileSystem *pFS, i64 iOff, int iAdd, i64 *piRes){






  i64 iEob;
  int iBlk;
  int rc;

  assert( pFS->pCompress );

  iEob = fsLastPageOnPagesBlock(pFS, iOff);
  if( (iOff+iAdd)<=iEob ){
    *piRes = (iOff+iAdd);
    return LSM_OK;
  }

  rc = fsBlockNext(pFS, fsPageToBlock(pFS, iOff), &iBlk);
  *piRes = fsFirstPageOnBlock(pFS, iBlk) + iAdd - (iEob - iOff + 1);
  return rc;
}

static int fsAllocateBuffer(FileSystem *pFS, int bWrite){
  u8 **pp;                        /* Pointer to either aIBuffer or aOBuffer */








|
>
>
>
>
>
>












|




|
>
>
>
>
>
>












|







1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
  nByte  = (aBuf[0] & 0x7F) << 14;
  nByte += (aBuf[1] & 0x7F) << 7;
  nByte += (aBuf[2] & 0x7F);
  *pbFree = !(aBuf[1] & 0x80);
  return nByte;
}

static int fsSubtractOffset(
  FileSystem *pFS, 
  Segment *pSeg,
  i64 iOff, 
  int iSub, 
  i64 *piRes
){
  i64 iStart;
  int iBlk = 0;
  int rc;

  assert( pFS->pCompress );

  iStart = fsFirstPageOnBlock(pFS, fsPageToBlock(pFS, iOff));
  if( (iOff-iSub)>=iStart ){
    *piRes = (iOff-iSub);
    return LSM_OK;
  }

  rc = fsBlockPrev(pFS, pSeg, fsPageToBlock(pFS, iOff), &iBlk);
  *piRes = fsLastPageOnBlock(pFS, iBlk) - iSub + (iOff - iStart + 1);
  return rc;
}

static int fsAddOffset(
  FileSystem *pFS, 
  Segment *pSeg,
  i64 iOff, 
  int iAdd, 
  i64 *piRes
){
  i64 iEob;
  int iBlk;
  int rc;

  assert( pFS->pCompress );

  iEob = fsLastPageOnPagesBlock(pFS, iOff);
  if( (iOff+iAdd)<=iEob ){
    *piRes = (iOff+iAdd);
    return LSM_OK;
  }

  rc = fsBlockNext(pFS, pSeg, fsPageToBlock(pFS, iOff), &iBlk);
  *piRes = fsFirstPageOnBlock(pFS, iBlk) + iAdd - (iEob - iOff + 1);
  return rc;
}

static int fsAllocateBuffer(FileSystem *pFS, int bWrite){
  u8 **pp;                        /* Pointer to either aIBuffer or aOBuffer */

1066
1067
1068
1069
1070
1071
1072

1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
** uncompresses the compressed data for page pPg from the database and
** populates the pPg->aData[] buffer and pPg->nCompress field.
**
** LSM_OK is returned if successful, or an LSM error code otherwise.
*/
static int fsReadPagedata(
  FileSystem *pFS,                /* File-system handle */

  Page *pPg,                      /* Page to read and uncompress data for */
  int *pnSpace                    /* OUT: Total bytes of free space */
){
  lsm_compress *p = pFS->pCompress;
  i64 iOff = pPg->iPg;
  u8 aSz[3];
  int rc;

  assert( p && pPg->nCompress==0 );

  if( fsAllocateBuffer(pFS, 0) ) return LSM_NOMEM;

  rc = fsReadData(pFS, iOff, aSz, sizeof(aSz));

  if( rc==LSM_OK ){
    int bFree;
    if( aSz[0] & 0x80 ){
      pPg->nCompress = (int)getRecordSize(aSz, &bFree);
    }else{
      pPg->nCompress = (int)aSz[0] - sizeof(aSz)*2;
      bFree = 1;
    }
    if( bFree ){
      if( pnSpace ){
        *pnSpace = pPg->nCompress + sizeof(aSz)*2;
      }else{
        rc = LSM_CORRUPT_BKPT;
      }
    }else{
      rc = fsAddOffset(pFS, iOff, 3, &iOff);
      if( rc==LSM_OK ){
        if( pPg->nCompress>pFS->nBuffer ){
          rc = LSM_CORRUPT_BKPT;
        }else{
          rc = fsReadData(pFS, iOff, pFS->aIBuffer, pPg->nCompress);
        }
        if( rc==LSM_OK ){
          int n = pFS->nPagesize;
          rc = p->xUncompress(p->pCtx, 
              (char *)pPg->aData, &n, 
              (const char *)pFS->aIBuffer, pPg->nCompress
          );







>












|
















|




|







1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
** uncompresses the compressed data for page pPg from the database and
** populates the pPg->aData[] buffer and pPg->nCompress field.
**
** LSM_OK is returned if successful, or an LSM error code otherwise.
*/
static int fsReadPagedata(
  FileSystem *pFS,                /* File-system handle */
  Segment *pSeg,                  /* pPg is part of this segment */
  Page *pPg,                      /* Page to read and uncompress data for */
  int *pnSpace                    /* OUT: Total bytes of free space */
){
  lsm_compress *p = pFS->pCompress;
  i64 iOff = pPg->iPg;
  u8 aSz[3];
  int rc;

  assert( p && pPg->nCompress==0 );

  if( fsAllocateBuffer(pFS, 0) ) return LSM_NOMEM;

  rc = fsReadData(pFS, pSeg, iOff, aSz, sizeof(aSz));

  if( rc==LSM_OK ){
    int bFree;
    if( aSz[0] & 0x80 ){
      pPg->nCompress = (int)getRecordSize(aSz, &bFree);
    }else{
      pPg->nCompress = (int)aSz[0] - sizeof(aSz)*2;
      bFree = 1;
    }
    if( bFree ){
      if( pnSpace ){
        *pnSpace = pPg->nCompress + sizeof(aSz)*2;
      }else{
        rc = LSM_CORRUPT_BKPT;
      }
    }else{
      rc = fsAddOffset(pFS, pSeg, iOff, 3, &iOff);
      if( rc==LSM_OK ){
        if( pPg->nCompress>pFS->nBuffer ){
          rc = LSM_CORRUPT_BKPT;
        }else{
          rc = fsReadData(pFS, pSeg, iOff, pFS->aIBuffer, pPg->nCompress);
        }
        if( rc==LSM_OK ){
          int n = pFS->nPagesize;
          rc = p->xUncompress(p->pCtx, 
              (char *)pPg->aData, &n, 
              (const char *)pFS->aIBuffer, pPg->nCompress
          );
1123
1124
1125
1126
1127
1128
1129

1130
1131
1132
1133
1134
1135
1136
1137
1138





1139

1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151

1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
}

/*
** Return a handle for a database page.
*/
static int fsPageGet(
  FileSystem *pFS,                /* File-system handle */

  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;






  assert( iPg>=fsFirstPageOnBlock(pFS, 1) );

  *ppPg = 0;

  assert( pFS->bUseMmap==0 || pFS->pCompress==0 );
  if( pFS->bUseMmap ){
    Page *pTest;
    i64 iEnd = (i64)iPg * pFS->nPagesize;
    fsGrowMapping(pFS, iEnd, &rc);
    if( rc!=LSM_OK ) return rc;

    p = 0;
    for(pTest=pFS->pWaiting; pTest; pTest=pTest->pNextWaiting){
      if( pTest->iPg==iPg ){

        p = pTest;
        p->nRef++;
        *ppPg = p;
        return LSM_OK;
      }
    }
    if( pFS->pFree ){
      p = pFS->pFree;
      pFS->pFree = p->pHashNext;
      assert( p->nRef==0 );
    }else{
      p = lsmMallocZeroRc(pFS->pEnv, sizeof(Page), &rc);
      if( rc ) return rc;
      fsPageAddToLru(pFS, p);
      p->pFS = pFS;
    }
    p->aData = &((u8 *)pFS->pMap)[pFS->nPagesize * (i64)(iPg-1)];
    p->iPg = iPg;
  }else{

    /* Search the hash-table for the page */
    iHash = fsHashKey(pFS->nHash, iPg);
    for(p=pFS->apHash[iHash]; p; p=p->pHashNext){
      if( p->iPg==iPg) break;
    }

    if( p==0 ){
      rc = fsPageBuffer(pFS, 1, &p);
      if( rc==LSM_OK ){
        int nSpace = 0;
        p->iPg = iPg;
        p->nRef = 0;
        p->pFS = pFS;
        assert( p->flags==0 || p->flags==PAGE_FREE );

#ifdef LSM_DEBUG
        memset(p->aData, 0x56, pFS->nPagesize);
#endif
        assert( p->pLruNext==0 && p->pLruPrev==0 );
        if( noContent==0 ){
          if( pFS->pCompress ){
            rc = fsReadPagedata(pFS, p, &nSpace);
          }else{
            int nByte = pFS->nPagesize;
            i64 iOff = (i64)(iPg-1) * pFS->nPagesize;
            rc = lsmEnvRead(pFS->pEnv, pFS->fdDb, iOff, p->aData, nByte);
          }
          pFS->nRead++;
        }

        /* If the xRead() call was successful (or not attempted), link the
         ** page into the page-cache hash-table. Otherwise, if it failed,







>









>
>
>
>
>

>





|





|
>
















|
|



|

|






|










|


|







1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
}

/*
** Return a handle for a database page.
*/
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( iPg>=fsFirstPageOnBlock(pFS, 1) );
  assert( iReal>=fsFirstPageOnBlock(pFS, 1) );
  *ppPg = 0;

  assert( pFS->bUseMmap==0 || pFS->pCompress==0 );
  if( pFS->bUseMmap ){
    Page *pTest;
    i64 iEnd = (i64)iReal * pFS->nPagesize;
    fsGrowMapping(pFS, iEnd, &rc);
    if( rc!=LSM_OK ) return rc;

    p = 0;
    for(pTest=pFS->pWaiting; pTest; pTest=pTest->pNextWaiting){
      if( pTest->iPg==iReal ){
        assert( iReal==iPg );
        p = pTest;
        p->nRef++;
        *ppPg = p;
        return LSM_OK;
      }
    }
    if( pFS->pFree ){
      p = pFS->pFree;
      pFS->pFree = p->pHashNext;
      assert( p->nRef==0 );
    }else{
      p = lsmMallocZeroRc(pFS->pEnv, sizeof(Page), &rc);
      if( rc ) return rc;
      fsPageAddToLru(pFS, p);
      p->pFS = pFS;
    }
    p->aData = &((u8 *)pFS->pMap)[pFS->nPagesize * (iReal-1)];
    p->iPg = iReal;
  }else{

    /* Search the hash-table for the page */
    iHash = fsHashKey(pFS->nHash, iReal);
    for(p=pFS->apHash[iHash]; p; p=p->pHashNext){
      if( p->iPg==iReal) break;
    }

    if( p==0 ){
      rc = fsPageBuffer(pFS, 1, &p);
      if( rc==LSM_OK ){
        int nSpace = 0;
        p->iPg = iReal;
        p->nRef = 0;
        p->pFS = pFS;
        assert( p->flags==0 || p->flags==PAGE_FREE );

#ifdef LSM_DEBUG
        memset(p->aData, 0x56, pFS->nPagesize);
#endif
        assert( p->pLruNext==0 && p->pLruPrev==0 );
        if( noContent==0 ){
          if( pFS->pCompress ){
            rc = fsReadPagedata(pFS, pSeg, p, &nSpace);
          }else{
            int nByte = pFS->nPagesize;
            i64 iOff = (i64)(iReal-1) * pFS->nPagesize;
            rc = lsmEnvRead(pFS->pEnv, pFS->fdDb, iOff, p->aData, nByte);
          }
          pFS->nRead++;
        }

        /* If the xRead() call was successful (or not attempted), link the
         ** page into the page-cache hash-table. Otherwise, if it failed,
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233

    assert( (rc==LSM_OK && (p || (pnSpace && *pnSpace)))
         || (rc!=LSM_OK && p==0) 
    );
  }

  if( rc==LSM_OK && p ){
    if( pFS->pCompress==0 && (fsIsLast(pFS, iPg) || fsIsFirst(pFS, iPg)) ){
      p->nData = pFS->nPagesize - 4;
      if( fsIsFirst(pFS, iPg) && p->nRef==0 ){
        p->aData += 4;
        p->flags |= PAGE_HASPREV;
      }
    }else{
      p->nData = pFS->nPagesize;
    }
    pFS->nOut += (p->nRef==0);







|

|







1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307

    assert( (rc==LSM_OK && (p || (pnSpace && *pnSpace)))
         || (rc!=LSM_OK && p==0) 
    );
  }

  if( rc==LSM_OK && p ){
    if( pFS->pCompress==0 && (fsIsLast(pFS, iReal) || fsIsFirst(pFS, iReal)) ){
      p->nData = pFS->nPagesize - 4;
      if( fsIsFirst(pFS, iReal) && p->nRef==0 ){
        p->aData += 4;
        p->flags |= PAGE_HASPREV;
      }
    }else{
      p->nData = pFS->nPagesize;
    }
    pFS->nOut += (p->nRef==0);
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372





1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390






















1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408



1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430














1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454





1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
    iBlk = fsPageToBlock(pFS, pDel->iFirst);
    iLastBlk = fsPageToBlock(pFS, pDel->iLastPg);

    /* Mark all blocks currently used by this sorted run as free */
    while( iBlk && rc==LSM_OK ){
      int iNext = 0;
      if( iBlk!=iLastBlk ){
        rc = fsBlockNext(pFS, iBlk, &iNext);
      }else if( bZero==0 && pDel->iLastPg!=fsLastPageOnBlock(pFS, iLastBlk) ){
        break;
      }
      rc = fsFreeBlock(pFS, pSnapshot, pDel, iBlk);
      iBlk = iNext;
    }






    if( bZero ) memset(pDel, 0, sizeof(Segment));
  }
  return LSM_OK;
}

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;
}























/*
** Argument aPgno is an array of nPgno page numbers. All pages belong to
** 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( pRun->nSize>0 );



  iBlk = fsPageToBlock(pFS, pRun->iFirst);
  pRun->nSize += (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, iBlk, &iNext);
    if( rc==LSM_OK ) rc = fsFreeBlock(pFS, pSnapshot, pRun, iBlk);
    pRun->nSize -= (
        1 + fsLastPageOnBlock(pFS, iBlk) - fsFirstPageOnBlock(pFS, iBlk)
    );
    iBlk = iNext;
  }

  pRun->nSize -= (pRun->iFirst - fsFirstPageOnBlock(pFS, iBlk));
  assert( pRun->nSize>0 );
}















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, iPg, nByte-1, &iNext);
  if( pSeg && iNext==pSeg->iLastPg ){
    iNext = 0;
  }else if( rc==LSM_OK ){
    rc = fsAddOffset(pFS, iNext, 1, &iNext);
  }

  *piNext = iNext;
  return rc;
}

static int fsGetPageBefore(FileSystem *pFS, i64 iOff, Pgno *piPrev){





  u8 aSz[3];
  int rc;
  i64 iRead;

  rc = fsSubtractOffset(pFS, iOff, sizeof(aSz), &iRead);
  if( rc==LSM_OK ) rc = fsReadData(pFS, iRead, aSz, sizeof(aSz));

  if( rc==LSM_OK ){
    int bFree;
    int nSz;
    if( aSz[2] & 0x80 ){
      nSz = getRecordSize(aSz, &bFree) + sizeof(aSz)*2;
    }else{
      nSz = (int)(aSz[2] & 0x7F);
      bFree = 1;
    }
    rc = fsSubtractOffset(pFS, iOff, nSz, piPrev);
  }

  return rc;
}

/*
** The first argument to this function is a valid reference to a database







|






>
>
>
>
>


















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


















>
>
>










|











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












|



|






|
>
>
>
>
>




|
|










|







1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
    iBlk = fsPageToBlock(pFS, pDel->iFirst);
    iLastBlk = fsPageToBlock(pFS, pDel->iLastPg);

    /* Mark all blocks currently used by this sorted run as free */
    while( iBlk && rc==LSM_OK ){
      int iNext = 0;
      if( iBlk!=iLastBlk ){
        rc = fsBlockNext(pFS, pDel, iBlk, &iNext);
      }else if( bZero==0 && pDel->iLastPg!=fsLastPageOnBlock(pFS, iLastBlk) ){
        break;
      }
      rc = fsFreeBlock(pFS, pSnapshot, pDel, iBlk);
      iBlk = iNext;
    }

    if( pDel->pRedirect ){
      assert( pDel->pRedirect==&pSnapshot->redirect );
      pSnapshot->redirect.n = 0;
    }

    if( bZero ) memset(pDel, 0, sizeof(Segment));
  }
  return LSM_OK;
}

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. 
*/
static int fsSegmentRedirects(FileSystem *pFS, Segment *p){
  return (p && (
      fsPageRedirects(pFS, p, p->iFirst)
   || fsPageRedirects(pFS, p, p->iRoot)
   || fsPageRedirects(pFS, p, p->iLastPg)
  ));
}
#endif

/*
** Argument aPgno is an array of nPgno page numbers. All pages belong to
** 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( pRun->nSize>0 );
  assert( 0==fsSegmentRedirects(pFS, pRun) );
  assert( nPgno>0 && 0==fsPageRedirects(pFS, pRun, aPgno[0]) );

  iBlk = fsPageToBlock(pFS, pRun->iFirst);
  pRun->nSize += (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 -= (
        1 + fsLastPageOnBlock(pFS, iBlk) - fsFirstPageOnBlock(pFS, iBlk)
    );
    iBlk = iNext;
  }

  pRun->nSize -= (pRun->iFirst - fsFirstPageOnBlock(pFS, iBlk));
  assert( pRun->nSize>0 );
}

/*
** This function is only used in compressed database mode.
**
** Argument iPg is the page number (byte offset) of a page within segment
** pSeg. The page record, including all headers, is nByte bytes in size.
** Before returning, set *piNext to the page number of the next page in
** the segment, or to zero if iPg is the last.
**
** In other words, do:
**
**   *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;
  }else if( rc==LSM_OK ){
    rc = fsAddOffset(pFS, pSeg, iNext, 1, &iNext);
  }

  *piNext = iNext;
  return rc;
}

static int fsGetPageBefore(
  FileSystem *pFS, 
  Segment *pSeg, 
  i64 iOff, 
  Pgno *piPrev
){
  u8 aSz[3];
  int rc;
  i64 iRead;

  rc = fsSubtractOffset(pFS, pSeg, iOff, sizeof(aSz), &iRead);
  if( rc==LSM_OK ) rc = fsReadData(pFS, pSeg, iRead, aSz, sizeof(aSz));

  if( rc==LSM_OK ){
    int bFree;
    int nSz;
    if( aSz[2] & 0x80 ){
      nSz = getRecordSize(aSz, &bFree) + sizeof(aSz)*2;
    }else{
      nSz = (int)(aSz[2] & 0x7F);
      bFree = 1;
    }
    rc = fsSubtractOffset(pFS, pSeg, iOff, nSz, piPrev);
  }

  return rc;
}

/*
** The first argument to this function is a valid reference to a database
1496
1497
1498
1499
1500
1501
1502

1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526

1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538

1539
1540
1541



1542



1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
** 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;


  if( pFS->pCompress ){
    int nSpace = pPg->nCompress + 2*3;

    do {
      if( eDir>0 ){
        rc = fsNextPageOffset(pFS, pRun, iPg, nSpace, &iPg);
      }else{
        if( iPg==pRun->iFirst ){
          iPg = 0;
        }else{
          rc = fsGetPageBefore(pFS, iPg, &iPg);
        }
      }

      nSpace = 0;
      if( iPg!=0 ){
        rc = fsPageGet(pFS, iPg, 0, ppNext, &nSpace);
        assert( (*ppNext==0)==(rc!=LSM_OK || nSpace>0) );
      }else{
        *ppNext = 0;
      }
    }while( nSpace>0 && rc==LSM_OK );

  }else{

    assert( eDir==1 || eDir==-1 );
    if( eDir<0 ){
      if( pRun && iPg==pRun->iFirst ){
        *ppNext = 0;
        return LSM_OK;
      }else if( fsIsFirst(pFS, iPg) ){
        assert( pPg->flags & PAGE_HASPREV );
        iPg = fsLastPageOnBlock(pFS, lsmGetU32(&pPg->aData[-4]));
      }else{
        iPg--;
      }
    }else{

      if( pRun && iPg==pRun->iLastPg ){
        *ppNext = 0;
        return LSM_OK;



      }else if( fsIsLast(pFS, iPg) ){



        iPg = fsFirstPageOnBlock(pFS, lsmGetU32(&pPg->aData[pFS->nPagesize-4]));
      }else{
        iPg++;
      }
    }
    rc = fsPageGet(pFS, iPg, 0, ppNext, 0);
  }

  return rc;
}

static Pgno findAppendPoint(FileSystem *pFS, Level *pLvl){
  int i;







>










|





|







>












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




|







1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
** 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 ){
        rc = fsNextPageOffset(pFS, pRun, iPg, nSpace, &iPg);
      }else{
        if( iPg==pRun->iFirst ){
          iPg = 0;
        }else{
          rc = fsGetPageBefore(pFS, pRun, iPg, &iPg);
        }
      }

      nSpace = 0;
      if( iPg!=0 ){
        rc = fsPageGet(pFS, pRun, iPg, 0, ppNext, &nSpace);
        assert( (*ppNext==0)==(rc!=LSM_OK || nSpace>0) );
      }else{
        *ppNext = 0;
      }
    }while( nSpace>0 && rc==LSM_OK );

  }else{
    Redirect *pRedir = pRun ? pRun->pRedirect : 0;
    assert( eDir==1 || eDir==-1 );
    if( eDir<0 ){
      if( pRun && iPg==pRun->iFirst ){
        *ppNext = 0;
        return LSM_OK;
      }else if( fsIsFirst(pFS, iPg) ){
        assert( pPg->flags & PAGE_HASPREV );
        iPg = fsLastPageOnBlock(pFS, lsmGetU32(&pPg->aData[-4]));
      }else{
        iPg--;
      }
    }else{
      if( pRun ){
        if( iPg==pRun->iLastPg ){
          *ppNext = 0;
          return LSM_OK;
        }
      }

      if( fsIsLast(pFS, iPg) ){
        int iBlk = fsRedirectBlock(
            pRedir, lsmGetU32(&pPg->aData[pFS->nPagesize-4])
        );
        iPg = fsFirstPageOnBlock(pFS, iBlk);
      }else{
        iPg++;
      }
    }
    rc = fsPageGet(pFS, pRun, iPg, 0, ppNext, 0);
  }

  return rc;
}

static Pgno findAppendPoint(FileSystem *pFS, Level *pLvl){
  int i;
1588
1589
1590
1591
1592
1593
1594


1595
1596
1597
1598
1599
1600
1601
  int rc = LSM_OK;
  Page *pPg = 0;
  *ppOut = 0;
  int iApp = 0;
  int iNext = 0;
  Segment *p = &pLvl->lhs;
  int iPrev = p->iLastPg;



  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
    ** by the lsmFsPagePersist() call.  */
    rc = fsPageBuffer(pFS, 1, &pPg);
    if( rc==LSM_OK ){







>
>







1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
  int rc = LSM_OK;
  Page *pPg = 0;
  *ppOut = 0;
  int iApp = 0;
  int iNext = 0;
  Segment *p = &pLvl->lhs;
  int iPrev = p->iLastPg;

  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
    ** by the lsmFsPagePersist() call.  */
    rc = fsPageBuffer(pFS, 1, &pPg);
    if( rc==LSM_OK ){
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
      pFS->nOut++;
    }
  }else{
    if( iPrev==0 ){
      iApp = findAppendPoint(pFS, pLvl);
    }else if( fsIsLast(pFS, iPrev) ){
      int iNext;
      rc = fsBlockNext(pFS, fsPageToBlock(pFS, iPrev), &iNext);
      if( rc!=LSM_OK ) return rc;
      iApp = fsFirstPageOnBlock(pFS, iNext);
    }else{
      iApp = iPrev + 1;
    }

    /* If this is the first page allocated, or if the page allocated is the
    ** last in the block, also allocate the next block here.  */
    if( iApp==0 || fsIsLast(pFS, iApp) ){
      int iNew;                     /* New block number */

      rc = lsmBlockAllocate(pFS->pDb, &iNew);
      if( rc!=LSM_OK ) return rc;
      if( iApp==0 ){
        iApp = fsFirstPageOnBlock(pFS, iNew);
      }else{
        iNext = fsFirstPageOnBlock(pFS, iNew);
      }
    }

    /* Grab the new page. */
    pPg = 0;
    rc = fsPageGet(pFS, iApp, 1, &pPg, 0);
    assert( rc==LSM_OK || pPg==0 );

    /* If this is the first or last page of a block, fill in the pointer 
     ** value at the end of the new page. */
    if( rc==LSM_OK ){
      p->nSize++;
      p->iLastPg = iApp;







|











|










|







1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
      pFS->nOut++;
    }
  }else{
    if( iPrev==0 ){
      iApp = findAppendPoint(pFS, pLvl);
    }else if( fsIsLast(pFS, iPrev) ){
      int iNext;
      rc = fsBlockNext(pFS, 0, fsPageToBlock(pFS, iPrev), &iNext);
      if( rc!=LSM_OK ) return rc;
      iApp = fsFirstPageOnBlock(pFS, iNext);
    }else{
      iApp = iPrev + 1;
    }

    /* If this is the first page allocated, or if the page allocated is the
    ** last in the block, also allocate the next block here.  */
    if( iApp==0 || fsIsLast(pFS, iApp) ){
      int iNew;                     /* New block number */

      rc = lsmBlockAllocate(pFS->pDb, 0, &iNew);
      if( rc!=LSM_OK ) return rc;
      if( iApp==0 ){
        iApp = fsFirstPageOnBlock(pFS, iNew);
      }else{
        iNext = fsFirstPageOnBlock(pFS, iNew);
      }
    }

    /* Grab the new page. */
    pPg = 0;
    rc = fsPageGet(pFS, 0, iApp, 1, &pPg, 0);
    assert( rc==LSM_OK || pPg==0 );

    /* If this is the first or last page of a block, fill in the pointer 
     ** value at the end of the new page. */
    if( rc==LSM_OK ){
      p->nSize++;
      p->iLastPg = iApp;
1663
1664
1665
1666
1667
1668
1669

1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740

/*
** Mark the sorted run passed as the second argument as finished. 
*/
int lsmFsSortedFinish(FileSystem *pFS, Segment *p){
  int rc = LSM_OK;
  if( p && p->iLastPg ){


    /* Check if the last page of this run happens to be the last of a block.
    ** If it is, then an extra block has already been allocated for this run.
    ** 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 ){
      Page *pLast;
      rc = fsPageGet(pFS, p->iLastPg, 0, &pLast, 0);
      if( rc==LSM_OK ){
        int iBlk = (int)lsmGetU32(&pLast->aData[pFS->nPagesize-4]);
        lsmBlockRefree(pFS->pDb, iBlk);
        lsmFsPageRelease(pLast);
      }
    }else{
      int iBlk = 0;
      rc = fsBlockNext(pFS, fsPageToBlock(pFS, p->iLastPg), &iBlk);
      if( rc==LSM_OK ){
        lsmBlockRefree(pFS->pDb, iBlk);
      }
    }
  }
  return rc;
}

/*
** Obtain a reference to page number iPg.
*/
int lsmFsDbPageGet(FileSystem *pFS, Pgno iPg, Page **ppPg){
  assert( pFS );
  return fsPageGet(pFS, iPg, 0, ppPg, 0);
}

/*
** Obtain a reference to the last page in the segment passed as the 
** second argument.
*/
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, iPg, &iPg);
      if( rc==LSM_OK ){
        rc = fsPageGet(pFS, iPg, 0, ppPg, &nSpace);
      }
    }while( rc==LSM_OK && nSpace>0 );

  }else{
    rc = fsPageGet(pFS, iPg, 0, ppPg, 0);
  }
  return rc;
}

/*
** Return a reference to meta-page iPg. If successful, LSM_OK is returned
** and *ppPg populated with the new page reference. The reference should







>



















|







|











|

|














|

|




|







1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875

/*
** Mark the sorted run passed as the second argument as finished. 
*/
int lsmFsSortedFinish(FileSystem *pFS, Segment *p){
  int rc = LSM_OK;
  if( p && p->iLastPg ){
    assert( p->pRedirect==0 );

    /* Check if the last page of this run happens to be the last of a block.
    ** If it is, then an extra block has already been allocated for this run.
    ** 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 ){
      Page *pLast;
      rc = fsPageGet(pFS, 0, p->iLastPg, 0, &pLast, 0);
      if( rc==LSM_OK ){
        int iBlk = (int)lsmGetU32(&pLast->aData[pFS->nPagesize-4]);
        lsmBlockRefree(pFS->pDb, iBlk);
        lsmFsPageRelease(pLast);
      }
    }else{
      int iBlk = 0;
      rc = fsBlockNext(pFS, p, fsPageToBlock(pFS, p->iLastPg), &iBlk);
      if( rc==LSM_OK ){
        lsmBlockRefree(pFS->pDb, iBlk);
      }
    }
  }
  return rc;
}

/*
** Obtain a reference to page number iPg.
*/
int lsmFsDbPageGet(FileSystem *pFS, Segment *pSeg, Pgno iPg, Page **ppPg){
  assert( pFS );
  return fsPageGet(pFS, pSeg, iPg, 0, ppPg, 0);
}

/*
** Obtain a reference to the last page in the segment passed as the 
** second argument.
*/
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 ){
        rc = fsPageGet(pFS, pSeg, iPg, 0, ppPg, &nSpace);
      }
    }while( rc==LSM_OK && nSpace>0 );

  }else{
    rc = fsPageGet(pFS, pSeg, iPg, 0, ppPg, 0);
  }
  return rc;
}

/*
** Return a reference to meta-page iPg. If successful, LSM_OK is returned
** and *ppPg populated with the new page reference. The reference should
1826
1827
1828
1829
1830
1831
1832











































































1833
1834
1835
1836
1837
1838
1839

/*
** Return true if page is currently writable.
*/
int lsmFsPageWritable(Page *pPg){
  return (pPg->flags & PAGE_DIRTY) ? 1 : 0;
}












































































/*
** Append raw data to a segment. This function is only used in compressed
** database mode.
*/
static Pgno fsAppendData(
  FileSystem *pFS,                /* File-system handle */







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







1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049

/*
** Return true if page is currently writable.
*/
int lsmFsPageWritable(Page *pPg){
  return (pPg->flags & PAGE_DIRTY) ? 1 : 0;
}

static void fsMovePage(
  FileSystem *pFS, 
  Segment *pSeg, 
  int iTo, 
  int iFrom, 
  Pgno *piPg
){
  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 
** on block iTo. And that block iFrom is not currently being written. In
** other words, the data can be read and written directly.
*/
int lsmFsMoveBlock(FileSystem *pFS, Segment *pSeg, int iTo, int iFrom){
  Snapshot *p = pFS->pDb->pWorker;
  int rc = LSM_OK;

  i64 iFromOff = (i64)(iFrom-1) * pFS->nBlocksize;
  i64 iToOff = (i64)(iTo-1) * pFS->nBlocksize;
  
  assert( iTo!=1 );
  assert( iFrom>iTo );

  if( pFS->bUseMmap ){
    fsGrowMapping(pFS, (i64)iFrom * pFS->nBlocksize, &rc);
    if( rc==LSM_OK ){
      u8 *aMap = (u8 *)(pFS->pMap);
      memcpy(&aMap[iToOff], &aMap[iFromOff], pFS->nBlocksize);
    }
  }else{
    int nSz = pFS->nPagesize;
    u8 *aData = (u8 *)lsmMallocRc(pFS->pEnv, nSz, &rc);
    if( rc==LSM_OK ){
      const int nPagePerBlock = (pFS->nBlocksize / pFS->nPagesize);
      int i;
      for(i=0; rc==LSM_OK && i<nPagePerBlock; i++){
        i64 iOff = iFromOff + i*nSz;
        rc = lsmEnvRead(pFS->pEnv, pFS->fdDb, iOff, aData, nSz);
        if( rc==LSM_OK ){
          iOff = iToOff + i*nSz;
          rc = lsmEnvWrite(pFS->pEnv, pFS->fdDb, iOff, aData, nSz);
        }
      }
    }
  }

  /* Update append-point list if necessary */
  if( rc==LSM_OK ){
    int i;
    for(i=0; i<LSM_APPLIST_SZ; i++){
      if( fsPageToBlock(pFS, p->aiAppend[i])==iFrom ){
        const int nPagePerBlock = (pFS->nBlocksize / pFS->nPagesize);
        p->aiAppend[i] -= (i64)(iFrom-iTo) * nPagePerBlock;
      }
    }
  }

  /* Update the Segment structure itself */
  fsMovePage(pFS, pSeg, iTo, iFrom, &pSeg->iFirst);
  fsMovePage(pFS, pSeg, iTo, iFrom, &pSeg->iLastPg);
  fsMovePage(pFS, pSeg, iTo, iFrom, &pSeg->iRoot);

  return rc;
}

/*
** Append raw data to a segment. This function is only used in compressed
** database mode.
*/
static Pgno fsAppendData(
  FileSystem *pFS,                /* File-system handle */
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867

    /* 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;
        rc = lsmBlockAllocate(pFS->pDb, &iBlk);
        pSeg->iFirst = iApp = fsFirstPageOnBlock(pFS, iBlk);
      }
    }
    iRet = iApp;

    /* Write as much data as is possible at iApp (usually all of it). */
    iLastOnBlock = fsLastPageOnPagesBlock(pFS, iApp);







|







2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077

    /* 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;
        rc = lsmBlockAllocate(pFS->pDb, 0, &iBlk);
        pSeg->iFirst = iApp = fsFirstPageOnBlock(pFS, iBlk);
      }
    }
    iRet = iApp;

    /* Write as much data as is possible at iApp (usually all of it). */
    iLastOnBlock = fsLastPageOnPagesBlock(pFS, iApp);
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908

1909
1910
1911
1912
1913
1914
1915
1916
    assert( nRem<=0 || (iApp-1)==iLastOnBlock );
    if( rc==LSM_OK && (iApp-1)==iLastOnBlock ){
      u8 aPtr[4];                 /* Space to serialize a u32 */
      int iBlk;                   /* New block number */

      if( nWrite>0 ){
        /* Allocate a new block. */
        rc = lsmBlockAllocate(pFS->pDb, &iBlk);

        /* Set the "next" pointer on the old block */
        if( rc==LSM_OK ){
          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. */
        assert( nRem>0 );

        rc = fsBlockNext(pFS, fsPageToBlock(pFS, iApp), &iBlk);
        iRet = iApp = fsFirstPageOnBlock(pFS, iBlk);
      }

      /* Write the remaining data into the new block */
      if( rc==LSM_OK && nRem>0 ){
        rc = lsmEnvWrite(pFS->pEnv, pFS->fdDb, iApp, &aData[nWrite], nRem);
        iApp += nRem;







|



















>
|







2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
    assert( nRem<=0 || (iApp-1)==iLastOnBlock );
    if( rc==LSM_OK && (iApp-1)==iLastOnBlock ){
      u8 aPtr[4];                 /* Space to serialize a u32 */
      int iBlk;                   /* New block number */

      if( nWrite>0 ){
        /* Allocate a new block. */
        rc = lsmBlockAllocate(pFS->pDb, 0, &iBlk);

        /* Set the "next" pointer on the old block */
        if( rc==LSM_OK ){
          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. */
        assert( nRem>0 );
        assert( pSeg->pRedirect==0 );
        rc = fsBlockNext(pFS, 0, fsPageToBlock(pFS, iApp), &iBlk);
        iRet = iApp = fsFirstPageOnBlock(pFS, iBlk);
      }

      /* Write the remaining data into the new block */
      if( rc==LSM_OK && nRem>0 ){
        rc = lsmEnvWrite(pFS->pEnv, pFS->fdDb, iApp, &aData[nWrite], nRem);
        iApp += nRem;
1963
1964
1965
1966
1967
1968
1969

1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
  if( fsIsLast(pFS, iPrev) ){
    /* Grab the first page on the next block (which has already be
    ** allocated). In this case set *piPrev to tell the caller to set
    ** the "previous block" pointer in the first 4 bytes of the page.
    */
    int iNext;
    int iBlk = fsPageToBlock(pFS, iPrev);

    rc = fsBlockNext(pFS, iBlk, &iNext);
    if( rc!=LSM_OK ) return rc;
    *piNew = fsFirstPageOnBlock(pFS, iNext);
    *piPrev = iBlk;
  }else{
    *piNew = iPrev+1;
    if( fsIsLast(pFS, *piNew) ){
      /* Allocate the next block here. */
      int iBlk;
      rc = lsmBlockAllocate(pFS->pDb, &iBlk);
      if( rc!=LSM_OK ) return rc;
      *piNext = iBlk;
    }
  }

  pSeg->nSize++;
  pSeg->iLastPg = *piNew;







>
|








|







2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
  if( fsIsLast(pFS, iPrev) ){
    /* Grab the first page on the next block (which has already be
    ** allocated). In this case set *piPrev to tell the caller to set
    ** the "previous block" pointer in the first 4 bytes of the page.
    */
    int iNext;
    int iBlk = fsPageToBlock(pFS, iPrev);
    assert( pSeg->pRedirect==0 );
    rc = fsBlockNext(pFS, 0, iBlk, &iNext);
    if( rc!=LSM_OK ) return rc;
    *piNew = fsFirstPageOnBlock(pFS, iNext);
    *piPrev = iBlk;
  }else{
    *piNew = iPrev+1;
    if( fsIsLast(pFS, *piNew) ){
      /* Allocate the next block here. */
      int iBlk;
      rc = lsmBlockAllocate(pFS->pDb, 0, &iBlk);
      if( rc!=LSM_OK ) return rc;
      *piNext = iBlk;
    }
  }

  pSeg->nSize++;
  pSeg->iLastPg = *piNew;
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
























2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
    iBlk = fsPageToBlock(pFS, pArray->iFirst);
    iLastBlk = fsPageToBlock(pFS, pArray->iLastPg);

    lsmStringInit(&str, pDb->pEnv);
    if( bBlock ){
      lsmStringAppendf(&str, "%d", iBlk);
      while( iBlk!=iLastBlk ){
        fsBlockNext(pFS, iBlk, &iBlk);
        lsmStringAppendf(&str, " %d", iBlk);
      }
    }else{
      lsmStringAppendf(&str, "%d", pArray->iFirst);
      while( iBlk!=iLastBlk ){
        lsmStringAppendf(&str, " %d", fsLastPageOnBlock(pFS, iBlk));
        fsBlockNext(pFS, iBlk, &iBlk);
        lsmStringAppendf(&str, " %d", fsFirstPageOnBlock(pFS, iBlk));
      }
      lsmStringAppendf(&str, " %d", pArray->iLastPg);
    }

    *pzOut = str.z;
  }

  if( bUnlock ){
    int rcwork = LSM_BUSY;
    lsmFinishWork(pDb, 0, &rcwork);
  }
  return rc;
}

























/*
** 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 *pArray = 0;            /* Array to report on */
  int bUnlock = 0;

  *pzOut = 0;
  if( iFirst==0 ) return LSM_ERROR;

  /* Obtain the worker snapshot */
  pWorker = pDb->pWorker;
  if( !pWorker ){
    rc = lsmBeginWork(pDb);
    if( rc!=LSM_OK ) return rc;
    pWorker = pDb->pWorker;
    bUnlock = 1;
  }

  /* Search for the array that starts on page iFirst */
  pArray = findSegment(pWorker, iFirst);

  if( pArray==0 ){
    /* Could not find the requested array. This is an error. */
    rc = LSM_ERROR;
  }else{
    Page *pPg = 0;
    FileSystem *pFS = pDb->pFS;
    LsmString str;

    lsmStringInit(&str, pDb->pEnv);
    rc = lsmFsDbPageGet(pFS, iFirst, &pPg);
    while( rc==LSM_OK && pPg ){
      Page *pNext = 0;
      lsmStringAppendf(&str, " %lld", lsmFsPageNumber(pPg));
      rc = lsmFsDbPageNext(pArray, pPg, 1, &pNext);
      lsmFsPageRelease(pPg);
      pPg = pNext;
    }

    if( rc!=LSM_OK ){
      lsmFree(pDb->pEnv, str.z);
    }else{







|






|














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












|















|

|








|



|







2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
    iBlk = fsPageToBlock(pFS, pArray->iFirst);
    iLastBlk = fsPageToBlock(pFS, pArray->iLastPg);

    lsmStringInit(&str, pDb->pEnv);
    if( bBlock ){
      lsmStringAppendf(&str, "%d", iBlk);
      while( iBlk!=iLastBlk ){
        fsBlockNext(pFS, pArray, iBlk, &iBlk);
        lsmStringAppendf(&str, " %d", iBlk);
      }
    }else{
      lsmStringAppendf(&str, "%d", pArray->iFirst);
      while( iBlk!=iLastBlk ){
        lsmStringAppendf(&str, " %d", fsLastPageOnBlock(pFS, iBlk));
        fsBlockNext(pFS, pArray, iBlk, &iBlk);
        lsmStringAppendf(&str, " %d", fsFirstPageOnBlock(pFS, iBlk));
      }
      lsmStringAppendf(&str, " %d", pArray->iLastPg);
    }

    *pzOut = str.z;
  }

  if( bUnlock ){
    int rcwork = LSM_BUSY;
    lsmFinishWork(pDb, 0, &rcwork);
  }
  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 */

  iPgBlock = fsPageToBlock(pFS, pSeg->iFirst);
  iBlk = fsRedirectBlock(pRedir, fsPageToBlock(pFS, pSeg->iFirst));
  iLastBlk = fsRedirectBlock(pRedir, fsPageToBlock(pFS, pSeg->iLastPg));

  while( iBlk!=iLastBlk && iBlk!=iPgBlock && rc==LSM_OK ){
    rc = fsBlockNext(pFS, pSeg, iBlk, &iBlk);
  }

  *pbRes = (iBlk==iPgBlock);
  return rc;
}

/*
** 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;

  /* Obtain the worker snapshot */
  pWorker = pDb->pWorker;
  if( !pWorker ){
    rc = lsmBeginWork(pDb);
    if( rc!=LSM_OK ) return rc;
    pWorker = pDb->pWorker;
    bUnlock = 1;
  }

  /* Search for the array that starts on page iFirst */
  pSeg = findSegment(pWorker, iFirst);

  if( pSeg==0 ){
    /* Could not find the requested array. This is an error. */
    rc = LSM_ERROR;
  }else{
    Page *pPg = 0;
    FileSystem *pFS = pDb->pFS;
    LsmString str;

    lsmStringInit(&str, pDb->pEnv);
    rc = lsmFsDbPageGet(pFS, pSeg, iFirst, &pPg);
    while( rc==LSM_OK && pPg ){
      Page *pNext = 0;
      lsmStringAppendf(&str, " %lld", lsmFsPageNumber(pPg));
      rc = lsmFsDbPageNext(pSeg, pPg, 1, &pNext);
      lsmFsPageRelease(pPg);
      pPg = pNext;
    }

    if( rc!=LSM_OK ){
      lsmFree(pDb->pEnv, str.z);
    }else{
2447
2448
2449
2450
2451
2452
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
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
  int bExtra,                     /* If true, count the "next" block if any */
  int nUsed,
  u8 *aUsed
){
  if( pSeg ){
    if( pSeg && pSeg->nSize>0 ){
      int rc;
      Pgno iLast = pSeg->iLastPg;
      int iBlk;
      int iLastBlk;

      int bLastIsLastOnBlock;


      iBlk = fsPageToBlock(pFS, pSeg->iFirst);
      iLastBlk = fsPageToBlock(pFS, pSeg->iLastPg);

      bLastIsLastOnBlock = (fsLastPageOnBlock(pFS, iLastBlk)==iLast);
      assert( iBlk>0 );

      /* If the first page of this run is also the first page of its first
      ** block, set the flag to indicate that the first page of iBlk is 
      ** in use.  */
      if( fsFirstPageOnBlock(pFS, iBlk)==pSeg->iFirst ){
        assert( (aUsed[iBlk-1] & INTEGRITY_CHECK_FIRST_PG)==0 );
        aUsed[iBlk-1] |= INTEGRITY_CHECK_FIRST_PG;
      }

      do {
        /* iBlk is a part of this sorted run. */
        aUsed[iBlk-1] |= INTEGRITY_CHECK_USED;









        /* Unless the sorted run finishes before the last page on this block, 
        ** the last page of this block is also in use.  */
        if( iBlk!=iLastBlk || bLastIsLastOnBlock ){
          assert( (aUsed[iBlk-1] & INTEGRITY_CHECK_LAST_PG)==0 );
          aUsed[iBlk-1] |= INTEGRITY_CHECK_LAST_PG;
        }

        /* Special case. The sorted run being scanned is the output run of
        ** a level currently undergoing an incremental merge. The sorted
        ** run ends on the last page of iBlk, but the next block has already
        ** been allocated. So mark it as in use as well.  */
        if( iBlk==iLastBlk && bLastIsLastOnBlock && bExtra ){
          int iExtra = 0;
          rc = fsBlockNext(pFS, iBlk, &iExtra);
          assert( rc==LSM_OK );

          assert( aUsed[iExtra-1]==0 );
          aUsed[iExtra-1] |= INTEGRITY_CHECK_USED;
          aUsed[iExtra-1] |= INTEGRITY_CHECK_FIRST_PG;
          aUsed[iExtra-1] |= INTEGRITY_CHECK_LAST_PG;
        }

        /* Move on to the next block in the sorted run. Or set iBlk to zero
        ** in order to break out of the loop if this was the last block in
        ** the run.  */
        if( iBlk==iLastBlk ){
          iBlk = 0;
        }else{
          rc = fsBlockNext(pFS, iBlk, &iBlk);
          assert( rc==LSM_OK );
        }
      }while( iBlk );
    }
  }
}








<
|
|
>
|

>
|

>
|

<
<
<
<
<
<
<
<




>
>
>
>
>
>
>
>














|














|







2683
2684
2685
2686
2687
2688
2689

2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700








2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
  int bExtra,                     /* If true, count the "next" block if any */
  int nUsed,
  u8 *aUsed
){
  if( pSeg ){
    if( pSeg && pSeg->nSize>0 ){
      int rc;

      int iBlk;                   /* Current block (during iteration) */
      int iLastBlk;               /* Last block of segment */
      int iFirstBlk;              /* First block of segment */
      int bLastIsLastOnBlock;     /* True iLast is the last on its block */

      assert( 0==fsSegmentRedirects(pFS, pSeg) );
      iBlk = iFirstBlk = fsPageToBlock(pFS, pSeg->iFirst);
      iLastBlk = fsPageToBlock(pFS, pSeg->iLastPg);

      bLastIsLastOnBlock = (fsLastPageOnBlock(pFS, iLastBlk)==pSeg->iLastPg);
      assert( iBlk>0 );









      do {
        /* iBlk is a part of this sorted run. */
        aUsed[iBlk-1] |= INTEGRITY_CHECK_USED;

        /* If the first page of this block is also part of the segment,
        ** set the flag to indicate that the first page of iBlk is in use.  
        */
        if( fsFirstPageOnBlock(pFS, iBlk)==pSeg->iFirst || iBlk!=iFirstBlk ){
          assert( (aUsed[iBlk-1] & INTEGRITY_CHECK_FIRST_PG)==0 );
          aUsed[iBlk-1] |= INTEGRITY_CHECK_FIRST_PG;
        }

        /* Unless the sorted run finishes before the last page on this block, 
        ** the last page of this block is also in use.  */
        if( iBlk!=iLastBlk || bLastIsLastOnBlock ){
          assert( (aUsed[iBlk-1] & INTEGRITY_CHECK_LAST_PG)==0 );
          aUsed[iBlk-1] |= INTEGRITY_CHECK_LAST_PG;
        }

        /* Special case. The sorted run being scanned is the output run of
        ** a level currently undergoing an incremental merge. The sorted
        ** run ends on the last page of iBlk, but the next block has already
        ** been allocated. So mark it as in use as well.  */
        if( iBlk==iLastBlk && bLastIsLastOnBlock && bExtra ){
          int iExtra = 0;
          rc = fsBlockNext(pFS, pSeg, iBlk, &iExtra);
          assert( rc==LSM_OK );

          assert( aUsed[iExtra-1]==0 );
          aUsed[iExtra-1] |= INTEGRITY_CHECK_USED;
          aUsed[iExtra-1] |= INTEGRITY_CHECK_FIRST_PG;
          aUsed[iExtra-1] |= INTEGRITY_CHECK_LAST_PG;
        }

        /* Move on to the next block in the sorted run. Or set iBlk to zero
        ** in order to break out of the loop if this was the last block in
        ** the run.  */
        if( iBlk==iLastBlk ){
          iBlk = 0;
        }else{
          rc = fsBlockNext(pFS, pSeg, iBlk, &iBlk);
          assert( rc==LSM_OK );
        }
      }while( iBlk );
    }
  }
}

Changes to src/lsm_main.c.

37
38
39
40
41
42
43


44
45
46
47
48
49
50

  /* If there is at least one cursor or a write transaction open, the database
  ** handle must be holding a pointer to a client snapshot. And the reverse 
  ** - if there are no open cursors and no write transactions then there must 
  ** not be a client snapshot.  */
  assert( (pDb->pCsr!=0 || pDb->nTransOpen>0)==(pDb->iReader>=0) );



  assert( pDb->nTransOpen>=0 );
}
#else
# define assert_db_state(x) 
#endif

/*







>
>







37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52

  /* If there is at least one cursor or a write transaction open, the database
  ** handle must be holding a pointer to a client snapshot. And the reverse 
  ** - if there are no open cursors and no write transactions then there must 
  ** not be a client snapshot.  */
  assert( (pDb->pCsr!=0 || pDb->nTransOpen>0)==(pDb->iReader>=0) );

  assert( pDb->iReader<0 || pDb->pClient!=0 );

  assert( pDb->nTransOpen>=0 );
}
#else
# define assert_db_state(x) 
#endif

/*
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385

  assert( *pbUnlock==0 );
  if( !pDb->pWorker ){
    rc = lsmBeginWork(pDb);
    if( rc!=LSM_OK ) return rc;
    *pbUnlock = 1;
  }
  *pp = pDb->pWorker;
  return rc;
}

static void infoFreeWorker(lsm_db *pDb, int bUnlock){
  if( bUnlock ){
    int rcdummy = LSM_BUSY;
    lsmFinishWork(pDb, 0, &rcdummy);







|







373
374
375
376
377
378
379
380
381
382
383
384
385
386
387

  assert( *pbUnlock==0 );
  if( !pDb->pWorker ){
    rc = lsmBeginWork(pDb);
    if( rc!=LSM_OK ) return rc;
    *pbUnlock = 1;
  }
  if( pp ) *pp = pDb->pWorker;
  return rc;
}

static void infoFreeWorker(lsm_db *pDb, int bUnlock){
  if( bUnlock ){
    int rcdummy = LSM_BUSY;
    lsmFinishWork(pDb, 0, &rcdummy);
529
530
531
532
533
534
535




536


537
538
539
540
541
542
543
      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 **);




      rc = lsmInfoPageDump(pDb, pgno, (eParam==LSM_INFO_PAGE_HEX_DUMP), pzVal);


      break;
    }

    case LSM_INFO_LOG_STRUCTURE: {
      char **pzVal = va_arg(ap, char **);
      rc = lsmInfoLogStructure(pDb, pzVal);
      break;







>
>
>
>
|
>
>







531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
      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);
      }
      infoFreeWorker(pDb, bUnlock);
      break;
    }

    case LSM_INFO_LOG_STRUCTURE: {
      char **pzVal = va_arg(ap, char **);
      rc = lsmInfoLogStructure(pDb, pzVal);
      break;

Changes to src/lsm_shared.c.

161
162
163
164
165
166
167
168

169
170
171
172
173
174
175
  DbTruncateCtx *p = (DbTruncateCtx *)pCtx;
  if( iBlk!=p->nBlock || (p->iInUse>=0 && iSnapshot>=p->iInUse) ) return 1;
  p->nBlock--;
  return 0;
}

static int dbTruncate(lsm_db *pDb, i64 iInUse){
  int rc;

  int i;
  DbTruncateCtx ctx;

  assert( pDb->pWorker );
  ctx.nBlock = pDb->pWorker->nBlock;
  ctx.iInUse = iInUse;








|
>







161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
  DbTruncateCtx *p = (DbTruncateCtx *)pCtx;
  if( iBlk!=p->nBlock || (p->iInUse>=0 && iSnapshot>=p->iInUse) ) return 1;
  p->nBlock--;
  return 0;
}

static int dbTruncate(lsm_db *pDb, i64 iInUse){
  int rc = LSM_OK;
#if 0
  int i;
  DbTruncateCtx ctx;

  assert( pDb->pWorker );
  ctx.nBlock = pDb->pWorker->nBlock;
  ctx.iInUse = iInUse;

184
185
186
187
188
189
190

191
192
193
194
195
196
197
      lsmLogMessage(pDb, 0, 
          "dbTruncate(): truncated db to %d blocks",ctx.nBlock
      );
    }
#endif
    pDb->pWorker->nBlock = ctx.nBlock;
  }

  return rc;
}


/*
** This function is called during database shutdown (when the number of
** connections drops from one to zero). It truncates the database file







>







185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
      lsmLogMessage(pDb, 0, 
          "dbTruncate(): truncated db to %d blocks",ctx.nBlock
      );
    }
#endif
    pDb->pWorker->nBlock = ctx.nBlock;
  }
#endif
  return rc;
}


/*
** This function is called during database shutdown (when the number of
** connections drops from one to zero). It truncates the database file
623
624
625
626
627
628
629

630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654

655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682

683

684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700


701

702
703
704
705
706
707
708
709
710
711
712
713
714
715




716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
}


typedef struct FindFreeblockCtx FindFreeblockCtx;
struct FindFreeblockCtx {
  i64 iInUse;
  int iRet;

};

static int findFreeblockCb(void *pCtx, int iBlk, i64 iSnapshot){
  FindFreeblockCtx *p = (FindFreeblockCtx *)pCtx;
  if( iSnapshot<p->iInUse ){
    p->iRet = iBlk;
    return 1;
  }
  return 0;
}

static int findFreeblock(lsm_db *pDb, i64 iInUse, int *piRet){
  int rc;                         /* Return code */
  FindFreeblockCtx ctx;           /* Context object */

#ifdef LSM_LOG_BLOCKS
  char *zList = 0;
  lsmInfoFreelist(pDb, &zList);
  lsmLogMessage(pDb, 0, 
      "findFreeblock(): iInUse=%lld freelist=%s", iInUse, zList);
  lsmFree(pDb->pEnv, zList);
#endif

  ctx.iInUse = iInUse;
  ctx.iRet = 0;

  rc = lsmWalkFreelist(pDb, 0, findFreeblockCb, (void *)&ctx);
  *piRet = ctx.iRet;

#ifdef LSM_LOG_BLOCKS
  lsmLogMessage(pDb, 0, "findFreeblock(): returning block %d", *piRet);
#endif
  return rc;
}

/*
** Allocate a new database file block to write data to, either by extending
** the database file or by recycling a free-list entry. The worker snapshot 
** must be held in order to call this function.
**
** If successful, *piBlk is set to the block number allocated and LSM_OK is
** returned. Otherwise, *piBlk is zeroed and an lsm error code returned.
*/
int lsmBlockAllocate(lsm_db *pDb, int *piBlk){
  Snapshot *p = pDb->pWorker;
  int iRet = 0;                   /* Block number of allocated block */
  int rc = LSM_OK;
  i64 iInUse = 0;                 /* Snapshot id still in use */
  i64 iSynced = 0;                /* Snapshot id synced to disk */

  assert( p );

#ifdef LSM_LOG_FREELIST
  {

    char *zFree = 0;

    rc = lsmInfoFreelist(pDb, &zFree);
    if( rc!=LSM_OK ) return rc;
    lsmLogMessage(pDb, 0, "lsmBlockAllocate(): freelist: %s", zFree);
    lsmFree(pDb->pEnv, zFree);
  }
#endif

  /* Set iInUse to the smallest snapshot id that is either:
  **
  **   * Currently in use by a database client,
  **   * May be used by a database client in the future, or
  **   * Is the most recently checkpointed snapshot (i.e. the one that will
  **     be used following recovery if a failure occurs at this point).
  */
  rc = lsmCheckpointSynced(pDb, &iSynced, 0, 0);
  if( rc==LSM_OK && iSynced==0 ) iSynced = p->iId;
  iInUse = iSynced;


  if( rc==LSM_OK && pDb->pClient ) iInUse = LSM_MIN(iInUse, pDb->pClient->iId);

  if( rc==LSM_OK ) rc = firstSnapshotInUse(pDb, &iInUse);

#ifdef LSM_LOG_FREELIST
  {
    lsmLogMessage(pDb, 0, "lsmBlockAllocate(): "
        "snapshot-in-use: %lld (iSynced=%lld) (client-id=%lld)", 
        iInUse, iSynced, (pDb->pClient ? pDb->pClient->iId : 0)
    );
  }
#endif

  /* Query the free block list for a suitable block */
  if( rc==LSM_OK ) rc = findFreeblock(pDb, iInUse, &iRet);





  /* If a block was found in the free block list, use it and remove it from 
  ** the list. Otherwise, if no suitable block was found, allocate one from
  ** the end of the file.  */
  if( rc==LSM_OK ){
    if( iRet>0 ){
#ifdef LSM_LOG_FREELIST
      lsmLogMessage(pDb, 0, 
          "reusing block %d (snapshot-in-use=%lld)", iRet, iInUse);
#endif
      rc = freelistAppend(pDb, iRet, -1);
      if( rc==LSM_OK ){
        rc = dbTruncate(pDb, iInUse);
      }
    }else{
      iRet = ++(p->nBlock);
#ifdef LSM_LOG_FREELIST
      lsmLogMessage(pDb, 0, "extending file to %d blocks", iRet);
#endif
    }
  }

  assert( iRet>0 || rc!=LSM_OK );
  *piBlk = iRet;
  return rc;
}

/*
** Free a database block. The worker snapshot must be held in order to call 
** this function.







>




|






|



<
<
<
<
<
<
<
<


>



<
<
<











|










>

>


|














>
>
|
>






|





|

>
>
>
>
|
|
|
<

















|







625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647








648
649
650
651
652
653



654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720

721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
}


typedef struct FindFreeblockCtx FindFreeblockCtx;
struct FindFreeblockCtx {
  i64 iInUse;
  int iRet;
  int bNotOne;
};

static int findFreeblockCb(void *pCtx, int iBlk, i64 iSnapshot){
  FindFreeblockCtx *p = (FindFreeblockCtx *)pCtx;
  if( iSnapshot<p->iInUse && (iBlk!=1 || p->bNotOne==0) ){
    p->iRet = iBlk;
    return 1;
  }
  return 0;
}

static int findFreeblock(lsm_db *pDb, i64 iInUse, int bNotOne, int *piRet){
  int rc;                         /* Return code */
  FindFreeblockCtx ctx;           /* Context object */









  ctx.iInUse = iInUse;
  ctx.iRet = 0;
  ctx.bNotOne = bNotOne;
  rc = lsmWalkFreelist(pDb, 0, findFreeblockCb, (void *)&ctx);
  *piRet = ctx.iRet;




  return rc;
}

/*
** Allocate a new database file block to write data to, either by extending
** the database file or by recycling a free-list entry. The worker snapshot 
** must be held in order to call this function.
**
** If successful, *piBlk is set to the block number allocated and LSM_OK is
** returned. Otherwise, *piBlk is zeroed and an lsm error code returned.
*/
int lsmBlockAllocate(lsm_db *pDb, int iBefore, int *piBlk){
  Snapshot *p = pDb->pWorker;
  int iRet = 0;                   /* Block number of allocated block */
  int rc = LSM_OK;
  i64 iInUse = 0;                 /* Snapshot id still in use */
  i64 iSynced = 0;                /* Snapshot id synced to disk */

  assert( p );

#ifdef LSM_LOG_FREELIST
  {
    static int nCall = 0;
    char *zFree = 0;
    nCall++;
    rc = lsmInfoFreelist(pDb, &zFree);
    if( rc!=LSM_OK ) return rc;
    lsmLogMessage(pDb, 0, "lsmBlockAllocate(): %d freelist: %s", nCall, zFree);
    lsmFree(pDb->pEnv, zFree);
  }
#endif

  /* Set iInUse to the smallest snapshot id that is either:
  **
  **   * Currently in use by a database client,
  **   * May be used by a database client in the future, or
  **   * Is the most recently checkpointed snapshot (i.e. the one that will
  **     be used following recovery if a failure occurs at this point).
  */
  rc = lsmCheckpointSynced(pDb, &iSynced, 0, 0);
  if( rc==LSM_OK && iSynced==0 ) iSynced = p->iId;
  iInUse = iSynced;
  if( rc==LSM_OK && pDb->iReader>=0 ){
    assert( pDb->pClient );
    iInUse = LSM_MIN(iInUse, pDb->pClient->iId);
  }
  if( rc==LSM_OK ) rc = firstSnapshotInUse(pDb, &iInUse);

#ifdef LSM_LOG_FREELIST
  {
    lsmLogMessage(pDb, 0, "lsmBlockAllocate(): "
        "snapshot-in-use: %lld (iSynced=%lld) (client-id=%lld)", 
        iInUse, iSynced, (pDb->iReader>=0 ? pDb->pClient->iId : 0)
    );
  }
#endif

  /* Query the free block list for a suitable block */
  if( rc==LSM_OK ) rc = findFreeblock(pDb, iInUse, (iBefore>0), &iRet);

  if( iBefore>0 && (iRet<=0 || iRet>=iBefore) ){
    iRet = 0;

  }else if( rc==LSM_OK ){
    /* If a block was found in the free block list, use it and remove it from 
    ** the list. Otherwise, if no suitable block was found, allocate one from
    ** the end of the file.  */

    if( iRet>0 ){
#ifdef LSM_LOG_FREELIST
      lsmLogMessage(pDb, 0, 
          "reusing block %d (snapshot-in-use=%lld)", iRet, iInUse);
#endif
      rc = freelistAppend(pDb, iRet, -1);
      if( rc==LSM_OK ){
        rc = dbTruncate(pDb, iInUse);
      }
    }else{
      iRet = ++(p->nBlock);
#ifdef LSM_LOG_FREELIST
      lsmLogMessage(pDb, 0, "extending file to %d blocks", iRet);
#endif
    }
  }

  assert( iBefore>0 || iRet>0 || rc!=LSM_OK );
  *piBlk = iRet;
  return rc;
}

/*
** Free a database block. The worker snapshot must be held in order to call 
** this function.
867
868
869
870
871
872
873

874
875
876
877
878
879
880
  return rc;
}

void lsmFreeSnapshot(lsm_env *pEnv, Snapshot *p){
  if( p ){
    lsmSortedFreeLevel(pEnv, p->pLevel);
    lsmFree(pEnv, p->freelist.aEntry);

    lsmFree(pEnv, p);
  }
}

/*
** Argument bFlush is true if the contents of the in-memory tree has just
** been flushed to disk. The significance of this is that once the snapshot







>







868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
  return rc;
}

void lsmFreeSnapshot(lsm_env *pEnv, Snapshot *p){
  if( p ){
    lsmSortedFreeLevel(pEnv, p->pLevel);
    lsmFree(pEnv, p->freelist.aEntry);
    lsmFree(pEnv, p->redirect.a);
    lsmFree(pEnv, p);
  }
}

/*
** Argument bFlush is true if the contents of the in-memory tree has just
** been flushed to disk. The significance of this is that once the snapshot

Changes to src/lsm_sorted.c.

68
69
70
71
72
73
74



75
76
77
78
79
80
81
*/

#ifndef _LSM_INT_H
# include "lsmInt.h"
#endif
#include "sqlite4.h"            /* only for sqlite4_snprintf() */




/*
** Macros to help decode record types.
*/
#define rtTopic(eType)       ((eType) & LSM_SYSTEMKEY)
#define rtIsDelete(eType)    (((eType) & 0x0F)==LSM_POINT_DELETE)

#define rtIsSeparator(eType) (((eType) & LSM_SEPARATOR)!=0)







>
>
>







68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
*/

#ifndef _LSM_INT_H
# include "lsmInt.h"
#endif
#include "sqlite4.h"            /* only for sqlite4_snprintf() */

#define LSM_LOG_STRUCTURE 0
#define LSM_LOG_DATA      0

/*
** Macros to help decode record types.
*/
#define rtTopic(eType)       ((eType) & LSM_SYSTEMKEY)
#define rtIsDelete(eType)    (((eType) & 0x0F)==LSM_POINT_DELETE)

#define rtIsSeparator(eType) (((eType) & LSM_SEPARATOR)!=0)
391
392
393
394
395
396
397

398
399
400
401
402
403
404
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(

  Page *pPg,
  int iOff,
  int nByte,
  void **ppData,
  Blob *pBlob
){
  int rc = LSM_OK;







>







394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
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;
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
      assert( nRem>=0 );
      if( nRem==0 ) break;
      i -= iEnd;

      /* Grab the next page in the segment */

      do {
        rc = lsmFsDbPageNext(0, pPg, 1, &pNext);
        if( rc==LSM_OK && pNext==0 ){
          rc = LSM_CORRUPT_BKPT;
        }
        if( rc ) break;
        lsmFsPageRelease(pPg);
        pPg = pNext;
        aData = fsPageData(pPg, &nData);







|







448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
      assert( nRem>=0 );
      if( nRem==0 ) break;
      i -= iEnd;

      /* Grab the next page in the segment */

      do {
        rc = lsmFsDbPageNext(pSeg, pPg, 1, &pNext);
        if( rc==LSM_OK && pNext==0 ){
          rc = LSM_CORRUPT_BKPT;
        }
        if( rc ) break;
        lsmFsPageRelease(pPg);
        pPg = pNext;
        aData = fsPageData(pPg, &nData);
505
506
507
508
509
510
511

512
513
514
515
516
517
518
  assert( iCell<pageGetNRec(aData, nData) && iCell>=0 );
  aCell = pageGetCell(aData, nData, iCell);
  lsmVarintGet64(&aCell[1], &iRet);
  return iRet;
}

static u8 *pageGetKey(

  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;







>







509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
  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;
531
532
533
534
535
536
537
538
539
540
541
542
543

544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
  pKey += lsmVarintGet32(pKey, &nDummy);
  pKey += lsmVarintGet32(pKey, pnKey);
  if( rtIsWrite(eType) ){
    pKey += lsmVarintGet32(pKey, &nDummy);
  }
  *piTopic = rtTopic(eType);

  sortedReadData(pPg, pKey-aData, *pnKey, (void **)&pKey, pBlob);
  return pKey;
}

static int pageGetKeyCopy(
  lsm_env *pEnv,                  /* Environment handle */

  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(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;
}







|





>









|







536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
  pKey += lsmVarintGet32(pKey, &nDummy);
  pKey += lsmVarintGet32(pKey, pnKey);
  if( rtIsWrite(eType) ){
    pKey += lsmVarintGet32(pKey, &nDummy);
  }
  *piTopic = rtTopic(eType);

  sortedReadData(pSeg, pPg, pKey-aData, *pnKey, (void **)&pKey, pBlob);
  return pKey;
}

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;
}
575
576
577
578
579
580
581

582
583
584
585
586
587
588
  return iRef;
}

#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(

  Page *pPg,
  int iKey, 
  Pgno *piPtr, 
  int *piTopic, 
  void **ppKey,
  int *pnKey,
  Blob *pBlob







>







581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
  return iRef;
}

#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
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
  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), iRef, &pRef);
    if( rc!=LSM_OK ) return rc;
    pageGetKeyCopy(lsmPageEnv(pPg), pRef, 0, &eType, pBlob);
    lsmFsPageRelease(pRef);
    *ppKey = pBlob->pData;
    *pnKey = pBlob->nData;
  }else{
    aCell += GETVARINT32(aCell, *pnKey);
    *ppKey = aCell;
  }







|

|







608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
  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;
    *pnKey = pBlob->nData;
  }else{
    aCell += GETVARINT32(aCell, *pnKey);
    *ppKey = aCell;
  }
632
633
634
635
636
637
638

639
640
641
642
643
644
645
    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;

    rc = pageGetBtreeKey(

        pCsr->aPg[iPg].pPage, iCell,
        &dummy, &pCsr->eType, &pCsr->pKey, &pCsr->nKey, &pCsr->blob
    );
    pCsr->eType |= LSM_SEPARATOR;
  }

  return rc;







>







639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
    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;

    rc = pageGetBtreeKey(
        pCsr->pSeg,
        pCsr->aPg[iPg].pPage, iCell,
        &dummy, &pCsr->eType, &pCsr->pKey, &pCsr->nKey, &pCsr->blob
    );
    pCsr->eType |= LSM_SEPARATOR;
  }

  return rc;
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
    if( pCsr->iPg>=0 ){
      pCsr->aPg[pCsr->iPg].iCell++;

      iLoad = btreeCursorPtr(aData, nData, pPg->iCell);
      do {
        Page *pLoad;
        pCsr->iPg++;
        rc = lsmFsDbPageGet(pCsr->pFS, iLoad, &pLoad);
        pCsr->aPg[pCsr->iPg].pPage = pLoad;
        pCsr->aPg[pCsr->iPg].iCell = 0;
        if( rc==LSM_OK ){
          if( pCsr->iPg==(pCsr->nDepth-1) ) break;
          aData = fsPageData(pLoad, &nData);
          iLoad = btreeCursorPtr(aData, nData, 0);
        }







|







701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
    if( pCsr->iPg>=0 ){
      pCsr->aPg[pCsr->iPg].iCell++;

      iLoad = btreeCursorPtr(aData, nData, pPg->iCell);
      do {
        Page *pLoad;
        pCsr->iPg++;
        rc = lsmFsDbPageGet(pCsr->pFS, pCsr->pSeg, iLoad, &pLoad);
        pCsr->aPg[pCsr->iPg].pPage = pLoad;
        pCsr->aPg[pCsr->iPg].iCell = 0;
        if( rc==LSM_OK ){
          if( pCsr->iPg==(pCsr->nDepth-1) ) break;
          aData = fsPageData(pLoad, &nData);
          iLoad = btreeCursorPtr(aData, nData, 0);
        }
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
  int rc;

  Page *pPg = 0;
  FileSystem *pFS = pCsr->pFS;
  int iPg = pCsr->pSeg->iRoot;

  do {
    rc = lsmFsDbPageGet(pFS, iPg, &pPg);
    assert( (rc==LSM_OK)==(pPg!=0) );
    if( rc==LSM_OK ){
      u8 *aData;
      int nData;
      int flags;

      aData = fsPageData(pPg, &nData);







|







746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
  int rc;

  Page *pPg = 0;
  FileSystem *pFS = pCsr->pFS;
  int iPg = pCsr->pSeg->iRoot;

  do {
    rc = lsmFsDbPageGet(pFS, pCsr->pSeg, iPg, &pPg);
    assert( (rc==LSM_OK)==(pPg!=0) );
    if( rc==LSM_OK ){
      u8 *aData;
      int nData;
      int flags;

      aData = fsPageData(pPg, &nData);
828
829
830
831
832
833
834

835
836
837
838
839
840
841
842
843
844
845
846

847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
  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 */


    /* Decode the MergeInput structure */
    iLeaf = p->iPg;
    nDepth = (p->iCell & 0x00FF);
    iCell = (p->iCell >> 8) - 1;

    /* Allocate the BtreeCursor.aPg[] array */
    assert( pCsr->aPg==0 );
    pCsr->aPg = (BtreePg *)lsmMallocZeroRc(pEnv, sizeof(BtreePg) * nDepth, &rc);

    /* Populate the last entry of the aPg[] array */
    if( rc==LSM_OK ){

      pCsr->iPg = nDepth-1;
      pCsr->nDepth = nDepth;
      pCsr->aPg[pCsr->iPg].iCell = iCell;
      rc = lsmFsDbPageGet(pCsr->pFS, iLeaf, &pCsr->aPg[nDepth-1].pPage);
    }

    /* 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 = pCsr->pSeg->iRoot;
      Page *pPg = pCsr->aPg[nDepth-1].pPage;
 
      if( pageObjGetNRec(pPg)==0 ){
        /* This can happen when pPg is the right-most leaf in the b-tree.
        ** 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(pPg,
            0, &dummy, &iTopicSeek, &pSeek, &nSeek, &pCsr->blob
        );
      }

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







>












>



|









|












|






|







836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
  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;

    /* Allocate the BtreeCursor.aPg[] array */
    assert( pCsr->aPg==0 );
    pCsr->aPg = (BtreePg *)lsmMallocZeroRc(pEnv, sizeof(BtreePg) * nDepth, &rc);

    /* Populate the last entry of the aPg[] array */
    if( rc==LSM_OK ){
      Page **pp = &pCsr->aPg[nDepth-1].pPage;
      pCsr->iPg = nDepth-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 = pSeg->iRoot;
      Page *pPg = pCsr->aPg[nDepth-1].pPage;
 
      if( pageObjGetNRec(pPg)==0 ){
        /* This can happen when pPg is the right-most leaf in the b-tree.
        ** 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 *pPg;
        rc = lsmFsDbPageGet(pCsr->pFS, pSeg, iLoad, &pPg);
        assert( rc==LSM_OK || pPg==0 );
        if( rc==LSM_OK ){
          u8 *aData;                  /* Buffer containing page data */
          int nData;                  /* Size of aData[] in bytes */
          int iMin;
          int iMax;
          int iCell;
897
898
899
900
901
902
903
904


905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924


925
926
927
928
929
930
931
          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(pPg, iTry, &iPtr, &iTopic, &pKey, &nKey,&blob);


            if( rc!=LSM_OK ) break;

            res = sortedKeyCompare(
                xCmp, iTopicSeek, pSeek, nSeek, iTopic, pKey, nKey
            );
            assert( res!=0 );

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

          pCsr->aPg[iPg].pPage = pPg;
          pCsr->aPg[iPg].iCell = iCell;
          iPg++;
          assert( iPg!=nDepth-1 || iLoad==iLeaf );


        }
      }while( rc==LSM_OK && iPg<(nDepth-1) );
      sortedBlobFree(&blob);
    }

    /* Load the current key and pointer */
    if( rc==LSM_OK ){







|
>
>



















|
>
>







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

            res = sortedKeyCompare(
                xCmp, iTopicSeek, pSeek, nSeek, iTopic, pKey, nKey
            );
            assert( res!=0 );

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

          pCsr->aPg[iPg].pPage = pPg;
          pCsr->aPg[iPg].iCell = iCell;
          iPg++;
          assert( iPg!=nDepth-1 
               || lsmFsRedirectPage(pCsr->pFS, pSeg->pRedirect, iLoad)==iLeaf
          );
        }
      }while( rc==LSM_OK && iPg<(nDepth-1) );
      sortedBlobFree(&blob);
    }

    /* Load the current key and pointer */
    if( rc==LSM_OK ){
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
      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(
            pCsr->aPg[i].pPage, pCsr->aPg[i].iCell-1,
            &dummy, &pCsr->eType, &pCsr->pKey, &pCsr->nKey, &pCsr->blob
        );
        pCsr->eType |= LSM_SEPARATOR;

      }else{
        rc = btreeCursorLoadKey(pCsr);







|







953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
      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,
            &dummy, &pCsr->eType, &pCsr->pKey, &pCsr->nKey, &pCsr->blob
        );
        pCsr->eType |= LSM_SEPARATOR;

      }else{
        rc = btreeCursorLoadKey(pCsr);
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
  FileSystem *pFS,
  SegmentPtr *pPtr,              /* Load page into this SegmentPtr object */
  int iNew                       /* Page number of new page */
){
  Page *pPg = 0;                 /* The new page */
  int rc;                        /* Return Code */

  rc = lsmFsDbPageGet(pFS, iNew, &pPg);
  assert( rc==LSM_OK || pPg==0 );
  segmentPtrSetPage(pPtr, pPg);

  return rc;
}

static int segmentPtrReadData(
  SegmentPtr *pPtr,
  int iOff,
  int nByte,
  void **ppData,
  Blob *pBlob
){
  return sortedReadData(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() */
){
  Page *pNext;                   /* New page to load */







|













|







1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
  FileSystem *pFS,
  SegmentPtr *pPtr,              /* Load page into this SegmentPtr object */
  int iNew                       /* Page number of new page */
){
  Page *pPg = 0;                 /* The new page */
  int rc;                        /* Return Code */

  rc = lsmFsDbPageGet(pFS, pPtr->pSeg, iNew, &pPg);
  assert( rc==LSM_OK || pPg==0 );
  segmentPtrSetPage(pPtr, pPg);

  return rc;
}

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() */
){
  Page *pNext;                   /* New page to load */
1069
1070
1071
1072
1073
1074
1075



















1076

1077
1078
1079
1080
1081

1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103


1104
1105
1106
1107
1108
1109
1110
      pPtr->pVal = 0;
    }
  }

  return rc;
}




















void lsmSortedSplitkey(lsm_db *pDb, Level *pLevel, int *pRc){

  Page *pPg = 0;
  lsm_env *pEnv = pDb->pEnv;      /* Environment handle */
  int rc = *pRc;
  Merge *pMerge = pLevel->pMerge;


  if( rc==LSM_OK ){
    rc = lsmFsDbPageGet(pDb->pFS, 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(
          pPg, pMerge->splitkey.iCell, &dummy, &iTopic, &pKey, &nKey, &blob
      );
      if( rc==LSM_OK && blob.pData!=pKey ){
        rc = sortedBlobSet(pEnv, &blob, pKey, nKey);
      }
    }else{
      rc = pageGetKeyCopy(pEnv, pPg, pMerge->splitkey.iCell, &iTopic, &blob);


    }

    pLevel->iSplitTopic = iTopic;
    pLevel->pSplitKey = blob.pData;
    pLevel->nSplitKey = blob.nData;
    lsmFsPageRelease(pPg);
  }







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





>

|












|






|
>
>







1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
      pPtr->pVal = 0;
    }
  }

  return rc;
}


static Segment *sortedSplitkeySegment(Level *pLevel){
  Merge *pMerge = pLevel->pMerge;
  MergeInput *p = &pMerge->splitkey;
  Segment *pSeg;
  int i;

  for(i=0; i<pMerge->nInput; i++){
    if( p->iPg==pMerge->aInput[i].iPg ) break;
  }
  if( pMerge->nInput==(pLevel->nRight+1) && i>=(pMerge->nInput-1) ){
    pSeg = &pLevel->pNext->lhs;
  }else{
    pSeg = &pLevel->aRhs[i];
  }

  return pSeg;
}

static void sortedSplitkey(lsm_db *pDb, Level *pLevel, int *pRc){
  Segment *pSeg;
  Page *pPg = 0;
  lsm_env *pEnv = pDb->pEnv;      /* Environment handle */
  int rc = *pRc;
  Merge *pMerge = pLevel->pMerge;

  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{
      rc = pageGetKeyCopy(
          pEnv, pSeg, pPg, pMerge->splitkey.iCell, &iTopic, &blob
      );
    }

    pLevel->iSplitTopic = iTopic;
    pLevel->pSplitKey = blob.pData;
    pLevel->nSplitKey = blob.nData;
    lsmFsPageRelease(pPg);
  }
1168
1169
1170
1171
1172
1173
1174

1175

1176
1177
1178

1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198

1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
    if( svFlags && pPtr->pPg ){
      int res = sortedKeyCompare(pCsr->pDb->xCmp,
          rtTopic(pPtr->eType), pPtr->pKey, pPtr->nKey,
          pLvl->iSplitTopic, pLvl->pSplitKey, pLvl->nSplitKey
      );
      if( res<0 ) segmentPtrReset(pPtr);
    }

    if( pPtr->pPg==0 && (svFlags & LSM_END_DELETE) ){

      rc = lsmFsDbPageGet(pCsr->pDb->pFS, pPtr->pSeg->iFirst, &pPtr->pPg);
      if( rc!=LSM_OK ) return rc;
      pPtr->eType = LSM_START_DELETE | (pLvl->iSplitTopic ? LSM_SYSTEMKEY : 0);

      pPtr->pKey = pLvl->pSplitKey;
      pPtr->nKey = pLvl->nSplitKey;
    }

  }while( pCsr 
       && pPtr->pPg 
       && segmentPtrIgnoreSeparators(pCsr, pPtr)
       && rtIsSeparator(pPtr->eType)
  );

  return LSM_OK;
}

static void segmentPtrEndPage(
  FileSystem *pFS, 
  SegmentPtr *pPtr, 
  int bLast, 
  int *pRc
){
  if( *pRc==LSM_OK ){

    Page *pNew = 0;
    if( bLast ){
      *pRc = lsmFsDbPageLast(pFS, pPtr->pSeg, &pNew);
    }else{
      *pRc = lsmFsDbPageGet(pFS, pPtr->pSeg->iFirst, &pNew);
    }
    segmentPtrSetPage(pPtr, pNew);
  }
}


/*







>

>
|

|
>




















>


|

|







1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
    if( svFlags && pPtr->pPg ){
      int res = sortedKeyCompare(pCsr->pDb->xCmp,
          rtTopic(pPtr->eType), pPtr->pKey, pPtr->nKey,
          pLvl->iSplitTopic, pLvl->pSplitKey, pLvl->nSplitKey
      );
      if( res<0 ) segmentPtrReset(pPtr);
    }

    if( pPtr->pPg==0 && (svFlags & LSM_END_DELETE) ){
      Segment *pSeg = pPtr->pSeg;
      rc = lsmFsDbPageGet(pCsr->pDb->pFS, pSeg, pSeg->iFirst, &pPtr->pPg);
      if( rc!=LSM_OK ) return rc;
      pPtr->eType = LSM_START_DELETE | LSM_POINT_DELETE;
      pPtr->eType |= (pLvl->iSplitTopic ? LSM_SYSTEMKEY : 0);
      pPtr->pKey = pLvl->pSplitKey;
      pPtr->nKey = pLvl->nSplitKey;
    }

  }while( pCsr 
       && pPtr->pPg 
       && segmentPtrIgnoreSeparators(pCsr, pPtr)
       && rtIsSeparator(pPtr->eType)
  );

  return LSM_OK;
}

static void segmentPtrEndPage(
  FileSystem *pFS, 
  SegmentPtr *pPtr, 
  int bLast, 
  int *pRc
){
  if( *pRc==LSM_OK ){
    Segment *pSeg = pPtr->pSeg;
    Page *pNew = 0;
    if( bLast ){
      *pRc = lsmFsDbPageLast(pFS, pSeg, &pNew);
    }else{
      *pRc = lsmFsDbPageGet(pFS, pSeg, pSeg->iFirst, &pNew);
    }
    segmentPtrSetPage(pPtr, pNew);
  }
}


/*
1226
1227
1228
1229
1230
1231
1232







1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251

1252
1253
1254
1255
1256
1257
1258
      && (pPtr->nCell==0 || (pPtr->flags & SEGMENT_BTREE_FLAG))
  ){
    rc = segmentPtrNextPage(pPtr, (bLast ? -1 : 1));
  }

  if( rc==LSM_OK && pPtr->pPg ){
    rc = segmentPtrLoadCell(pPtr, bLast ? (pPtr->nCell-1) : 0);







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


  if( bLast && rc==LSM_OK && pPtr->pPg
   && pPtr->pSeg==&pLvl->lhs 
   && pLvl->nRight && (pPtr->eType & LSM_START_DELETE)
  ){
    pPtr->iCell++;
    pPtr->eType = LSM_END_DELETE | (pLvl->iSplitTopic);
    pPtr->pKey = pLvl->pSplitKey;
    pPtr->nKey = pLvl->nSplitKey;
    pPtr->pVal = 0;
    pPtr->nVal = 0;
  }


  return rc;
}

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







>
>
>
>
>
>
>







|











>







1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
      && (pPtr->nCell==0 || (pPtr->flags & SEGMENT_BTREE_FLAG))
  ){
    rc = segmentPtrNextPage(pPtr, (bLast ? -1 : 1));
  }

  if( rc==LSM_OK && pPtr->pPg ){
    rc = segmentPtrLoadCell(pPtr, bLast ? (pPtr->nCell-1) : 0);
    if( rc==LSM_OK && bLast && pPtr->pSeg!=&pLvl->lhs ){
      int res = sortedKeyCompare(pCsr->pDb->xCmp,
          rtTopic(pPtr->eType), pPtr->pKey, pPtr->nKey,
          pLvl->iSplitTopic, pLvl->pSplitKey, pLvl->nSplitKey
      );
      if( res<0 ) segmentPtrReset(pPtr);
    }
  }
  
  bIgnore = segmentPtrIgnoreSeparators(pCsr, pPtr);
  if( rc==LSM_OK && pPtr->pPg && bIgnore && rtIsSeparator(pPtr->eType) ){
    rc = segmentPtrAdvance(pCsr, pPtr, bLast);
  }

#if 0
  if( bLast && rc==LSM_OK && pPtr->pPg
   && pPtr->pSeg==&pLvl->lhs 
   && pLvl->nRight && (pPtr->eType & LSM_START_DELETE)
  ){
    pPtr->iCell++;
    pPtr->eType = LSM_END_DELETE | (pLvl->iSplitTopic);
    pPtr->pKey = pLvl->pSplitKey;
    pPtr->nKey = pLvl->nSplitKey;
    pPtr->pVal = 0;
    pPtr->nVal = 0;
  }
#endif

  return rc;
}

static void segmentPtrKey(SegmentPtr *pPtr, void **ppKey, int *pnKey){
  assert( pPtr->pPg );
  *ppKey = pPtr->pKey;
1289
1290
1291
1292
1293
1294
1295

1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
  int iTopic = 0;                 /* TODO: Fix me */

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

    lsmFsPageRef(pTest);
    while( pTest ){

      Page *pNext;

      int rc = lsmFsDbPageNext(pPtr->pSeg, pTest, eDir, &pNext);
      lsmFsPageRelease(pTest);
      if( rc ) return 1;
      pTest = pNext;

      if( pTest ){
        int nData;
        u8 *aData = fsPageData(pTest, &nData);
        int nCell = pageGetNRec(aData, nData);
        int flags = pageGetFlags(aData, nData);
        if( nCell && 0==(flags&SEGMENT_BTREE_FLAG) ){
          int nPgKey;
          int iPgTopic;
          u8 *pPgKey;
          int res;
          int iCell;

          iCell = ((eDir < 0) ? (nCell-1) : 0);
          pPgKey = pageGetKey(pTest, iCell, &iPgTopic, &nPgKey, &blob);
          res = iTopic - iPgTopic;
          if( res==0 ) res = pCsr->pDb->xCmp(pKey, nKey, pPgKey, nPgKey);
          if( (eDir==1 && res>0) || (eDir==-1 && res<0) ){
            /* Taking this branch means something has gone wrong. */
            char *zMsg = lsmMallocPrintf(pEnv, "Key \"%s\" is not on page %d", 
                keyToString(pEnv, pKey, nKey), lsmFsPageNumber(pPtr->pPg)
            );







>


|

















|







1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
  int iTopic = 0;                 /* TODO: Fix me */

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

    lsmFsPageRef(pTest);
    while( pTest ){
      Segment *pSeg = pPtr->pSeg;
      Page *pNext;

      int rc = lsmFsDbPageNext(pSeg, pTest, eDir, &pNext);
      lsmFsPageRelease(pTest);
      if( rc ) return 1;
      pTest = pNext;

      if( pTest ){
        int nData;
        u8 *aData = fsPageData(pTest, &nData);
        int nCell = pageGetNRec(aData, nData);
        int flags = pageGetFlags(aData, nData);
        if( nCell && 0==(flags&SEGMENT_BTREE_FLAG) ){
          int nPgKey;
          int iPgTopic;
          u8 *pPgKey;
          int res;
          int iCell;

          iCell = ((eDir < 0) ? (nCell-1) : 0);
          pPgKey = pageGetKey(pSeg, pTest, iCell, &iPgTopic, &nPgKey, &blob);
          res = iTopic - iPgTopic;
          if( res==0 ) res = pCsr->pDb->xCmp(pKey, nKey, pPgKey, nPgKey);
          if( (eDir==1 && res>0) || (eDir==-1 && res<0) ){
            /* Taking this branch means something has gone wrong. */
            char *zMsg = lsmMallocPrintf(pEnv, "Key \"%s\" is not on page %d", 
                keyToString(pEnv, pKey, nKey), lsmFsPageNumber(pPtr->pPg)
            );
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
    u8 *pLastKey;
    int nLastKey;
    int iLastTopic;
    int res;                      /* Result of comparison */
    Page *pNext;

    /* Load the last key on the current page. */
    pLastKey = pageGetKey(
        pPtr->pPg, pPtr->nCell-1, &iLastTopic, &nLastKey, &pPtr->blob1
    );

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







|







1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
    u8 *pLastKey;
    int nLastKey;
    int iLastTopic;
    int res;                      /* Result of comparison */
    Page *pNext;

    /* Load the last key on the current page. */
    pLastKey = pageGetKey(pPtr->pSeg,
        pPtr->pPg, pPtr->nCell-1, &iLastTopic, &nLastKey, &pPtr->blob1
    );

    /* If the loaded key is >= than (pKey/nKey), break out of the loop.
    ** If (pKey/nKey) is present in this array, it must be on the current 
    ** page.  */
    res = sortedKeyCompare(
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
  do {
    Pgno *piFirst = 0;
    if( aPg ){
      aPg[i++] = iPg;
      piFirst = &aPg[i];
    }

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







|







1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
  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 );
    if( rc==LSM_OK ){
      u8 *aData;                  /* Buffer containing page data */
      int nData;                  /* Size of aData[] in bytes */
      int iMin;
      int iMax;
      int nRec;
1748
1749
1750
1751
1752
1753
1754

1755

1756
1757
1758
1759
1760
1761
1762
      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(pPg, iTry, &iPtr, &iTopicT, &pKeyT, &nKeyT, &blob);

        if( rc!=LSM_OK ) break;
        if( piFirst && pKeyT==blob.pData ){
          *piFirst = pageGetBtreeRef(pPg, iTry);
          piFirst = 0;
          i++;
        }








>
|
>







1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
      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 ){
          *piFirst = pageGetBtreeRef(pPg, iTry);
          piFirst = 0;
          i++;
        }

1869
1870
1871
1872
1873
1874
1875

1876
1877
1878

1879
1880
1881
1882
1883
1884








1885




1886
1887
1888
1889
1890
1891
1892
1893
    );
    if( rc==LSM_OK && nRhs>0 && eSeek==LSM_SEEK_GE && aPtr[0].pPg==0 ){
      res = 0;
    }
  }
  
  if( res>=0 ){

    int iPtr = *piPgno;
    int i;
    for(i=1; rc==LSM_OK && i<=nRhs && bStop==0; i++){

      iOut = 0;
      rc = seekInSegment(
          pCsr, &aPtr[i], iTopic, pKey, nKey, iPtr, eSeek, &iOut, &bStop
      );
      iPtr = iOut;
    }













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

  assert( eSeek==LSM_SEEK_EQ || bStop==0 );
  *piPgno = iOut;
  *pbStop = bStop;







>



>


|


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







1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
    );
    if( rc==LSM_OK && nRhs>0 && eSeek==LSM_SEEK_GE && aPtr[0].pPg==0 ){
      res = 0;
    }
  }
  
  if( res>=0 ){
    int bHit = 0;                 /* True if at least one rhs is not EOF */
    int iPtr = *piPgno;
    int i;
    for(i=1; rc==LSM_OK && i<=nRhs && bStop==0; i++){
      SegmentPtr *pPtr = &aPtr[i];
      iOut = 0;
      rc = seekInSegment(
          pCsr, pPtr, iTopic, pKey, nKey, iPtr, eSeek, &iOut, &bStop
      );
      iPtr = iOut;

      /* If the segment-pointer has settled on a key that is smaller than
      ** the splitkey, invalidate the segment-pointer.  */
      if( pPtr->pPg ){
        res = sortedKeyCompare(pCsr->pDb->xCmp, 
            rtTopic(pPtr->eType), pPtr->pKey, pPtr->nKey, 
            pLvl->iSplitTopic, pLvl->pSplitKey, pLvl->nSplitKey
        );
        if( res<0 ) segmentPtrReset(pPtr);
      }

      if( aPtr[i].pKey ) bHit = 1;
    }

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

  assert( eSeek==LSM_SEEK_EQ || bStop==0 );
  *piPgno = iOut;
  *pbStop = bStop;
2089
2090
2091
2092
2093
2094
2095
2096
2097

2098
2099

2100








2101





















2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113

2114
2115
2116
2117
2118
2119
2120
  MultiCursor *pCsr, 
  int iPtr,
  int bReverse
){
  int rc;
  SegmentPtr *pPtr = &pCsr->aPtr[iPtr];
  Level *pLvl = pPtr->pLevel;
  int bComposite;


  rc = segmentPtrAdvance(pCsr, pPtr, bReverse);
  if( rc!=LSM_OK ) return rc;

  bComposite = (pLvl->nRight>0 && pCsr->nPtr>pLvl->nRight);






























  if( bComposite && pPtr->pSeg==&pLvl->lhs       /* lhs of composite level */
   && bReverse==0                                /* csr advanced forwards */
   && pPtr->pPg==0                               /* segment at EOF */
  ){
    int i;
    for(i=0; rc==LSM_OK && i<pLvl->nRight; i++){
      rc = sortedRhsFirst(pCsr, pLvl, &pCsr->aPtr[iPtr+1+i]);
    }
    for(i=pCsr->nTree-1; i>0; i--){
      multiCursorDoCompare(pCsr, i, 0);
    }
  }


  return rc;
}

static void mcursorFreeComponents(MultiCursor *pCsr){
  int i;
  lsm_env *pEnv = pCsr->pDb->pEnv;







|

>


>

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












>







2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
  MultiCursor *pCsr, 
  int iPtr,
  int bReverse
){
  int rc;
  SegmentPtr *pPtr = &pCsr->aPtr[iPtr];
  Level *pLvl = pPtr->pLevel;
  int bComposite;                 /* True if pPtr is part of composite level */

  /* Advance the segment-pointer object. */
  rc = segmentPtrAdvance(pCsr, pPtr, bReverse);
  if( rc!=LSM_OK ) return rc;

  bComposite = (pLvl->nRight>0 && pCsr->nPtr>pLvl->nRight);
  if( bComposite && pPtr->pPg==0 ){
    int bFix = 0;
    if( (bReverse==0)==(pPtr->pSeg==&pLvl->lhs) ){
      int i;
      if( bReverse ){
        SegmentPtr *pLhs = &pCsr->aPtr[iPtr - 1 - (pPtr->pSeg - pLvl->aRhs)];
        for(i=0; i<pLvl->nRight; i++){
          if( pLhs[i+1].pPg ) break;
        }
        if( i==pLvl->nRight ){
          bFix = 1;
          rc = segmentPtrEnd(pCsr, pLhs, 1);
        }
      }else{
        bFix = 1;
        for(i=0; rc==LSM_OK && i<pLvl->nRight; i++){
          rc = sortedRhsFirst(pCsr, pLvl, &pCsr->aPtr[iPtr+1+i]);
        }
      }
    }

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

#if 0
  if( bComposite && pPtr->pSeg==&pLvl->lhs       /* lhs of composite level */
   && bReverse==0                                /* csr advanced forwards */
   && pPtr->pPg==0                               /* segment at EOF */
  ){
    int i;
    for(i=0; rc==LSM_OK && i<pLvl->nRight; i++){
      rc = sortedRhsFirst(pCsr, pLvl, &pCsr->aPtr[iPtr+1+i]);
    }
    for(i=pCsr->nTree-1; i>0; i--){
      multiCursorDoCompare(pCsr, i, 0);
    }
  }
#endif

  return rc;
}

static void mcursorFreeComponents(MultiCursor *pCsr){
  int i;
  lsm_env *pEnv = pCsr->pDb->pEnv;
2213
2214
2215
2216
2217
2218
2219




















2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
  for(i=0; i<nRhs; i++){
    pCsr->aPtr[i].pSeg = &pLvl->aRhs[i];
    pCsr->aPtr[i].pLevel = pLvl;
  }

  return LSM_OK;
}





















static int multiCursorAddAll(MultiCursor *pCsr, Snapshot *pSnap){
  Level *pLvl;
  int nPtr = 0;
  int iPtr = 0;
  int rc = LSM_OK;

  for(pLvl=pSnap->pLevel; pLvl; pLvl=pLvl->pNext){
    /* If the LEVEL_INCOMPLETE flag is set, then this function is being
    ** called (indirectly) from within a sortedNewToplevel() call to
    ** construct pLvl. In this case ignore pLvl - this cursor is going to
    ** be used to retrieve a freelist entry from the LSM, and the partially
    ** complete level may confuse it.  */
    if( pLvl->flags & LEVEL_INCOMPLETE ) continue;
    nPtr += (1 + pLvl->nRight);
  }

  assert( pCsr->aPtr==0 );
  pCsr->aPtr = lsmMallocZeroRc(pCsr->pDb->pEnv, sizeof(SegmentPtr) * nPtr, &rc);
  if( rc==LSM_OK ) pCsr->nPtr = nPtr;

  for(pLvl=pSnap->pLevel; pLvl && rc==LSM_OK; pLvl=pLvl->pNext){
    int i;
    if( pLvl->flags & LEVEL_INCOMPLETE ) continue;
    pCsr->aPtr[iPtr].pLevel = pLvl;
    pCsr->aPtr[iPtr].pSeg = &pLvl->lhs;
    iPtr++;
    for(i=0; i<pLvl->nRight; i++){
      pCsr->aPtr[iPtr].pLevel = pLvl;
      pCsr->aPtr[iPtr].pSeg = &pLvl->aRhs[i];
      iPtr++;
    }

    if( pLvl->nRight && pLvl->pSplitKey==0 ){
      lsmSortedSplitkey(pCsr->pDb, pLvl, &rc);
    }
  }

  return rc;
}

static int multiCursorInit(MultiCursor *pCsr, Snapshot *pSnap){







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



















<

|
<
|
<
<
<
<
<
<
<
<
<
<
|







2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356

2357
2358

2359










2360
2361
2362
2363
2364
2365
2366
2367
  for(i=0; i<nRhs; i++){
    pCsr->aPtr[i].pSeg = &pLvl->aRhs[i];
    pCsr->aPtr[i].pLevel = pLvl;
  }

  return LSM_OK;
}

static void multiCursorAddOne(MultiCursor *pCsr, Level *pLvl, int *pRc){
  if( *pRc==LSM_OK ){
    int iPtr = pCsr->nPtr;
    int i;
    pCsr->aPtr[iPtr].pLevel = pLvl;
    pCsr->aPtr[iPtr].pSeg = &pLvl->lhs;
    iPtr++;
    for(i=0; i<pLvl->nRight; i++){
      pCsr->aPtr[iPtr].pLevel = pLvl;
      pCsr->aPtr[iPtr].pSeg = &pLvl->aRhs[i];
      iPtr++;
    }

    if( pLvl->nRight && pLvl->pSplitKey==0 ){
      sortedSplitkey(pCsr->pDb, pLvl, pRc);
    }
    pCsr->nPtr = iPtr;
  }
}

static int multiCursorAddAll(MultiCursor *pCsr, Snapshot *pSnap){
  Level *pLvl;
  int nPtr = 0;
  int iPtr = 0;
  int rc = LSM_OK;

  for(pLvl=pSnap->pLevel; pLvl; pLvl=pLvl->pNext){
    /* If the LEVEL_INCOMPLETE flag is set, then this function is being
    ** called (indirectly) from within a sortedNewToplevel() call to
    ** construct pLvl. In this case ignore pLvl - this cursor is going to
    ** be used to retrieve a freelist entry from the LSM, and the partially
    ** complete level may confuse it.  */
    if( pLvl->flags & LEVEL_INCOMPLETE ) continue;
    nPtr += (1 + pLvl->nRight);
  }

  assert( pCsr->aPtr==0 );
  pCsr->aPtr = lsmMallocZeroRc(pCsr->pDb->pEnv, sizeof(SegmentPtr) * nPtr, &rc);


  for(pLvl=pSnap->pLevel; pLvl; pLvl=pLvl->pNext){

    if( (pLvl->flags & LEVEL_INCOMPLETE)==0 ){










      multiCursorAddOne(pCsr, pLvl, &rc);
    }
  }

  return rc;
}

static int multiCursorInit(MultiCursor *pCsr, Snapshot *pSnap){
2408
2409
2410
2411
2412
2413
2414

2415
2416



2417
2418
2419
2420
2421
2422
2423
  void *pCtx                      /* First argument to pass to callback */
){
  MultiCursor *pCsr;              /* Cursor used to read db */
  int rc = LSM_OK;                /* Return Code */
  Snapshot *pSnap = 0;

  assert( pDb->pWorker );

  rc = lsmCheckpointDeserialize(pDb, 0, pDb->pShmhdr->aSnap1, &pSnap);
  if( rc!=LSM_OK ) return rc;




  pCsr = multiCursorNew(pDb, &rc);
  if( pCsr ){
    rc = multiCursorAddAll(pCsr, pSnap);
    pCsr->flags |= CURSOR_IGNORE_DELETE;
  }
  







>
|
|
>
>
>







2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
  void *pCtx                      /* First argument to pass to callback */
){
  MultiCursor *pCsr;              /* Cursor used to read db */
  int rc = LSM_OK;                /* Return Code */
  Snapshot *pSnap = 0;

  assert( pDb->pWorker );
  if( pDb->bIncrMerge ){
    rc = lsmCheckpointDeserialize(pDb, 0, pDb->pShmhdr->aSnap1, &pSnap);
    if( rc!=LSM_OK ) return rc;
  }else{
    pSnap = pDb->pWorker;
  }

  pCsr = multiCursorNew(pDb, &rc);
  if( pCsr ){
    rc = multiCursorAddAll(pCsr, pSnap);
    pCsr->flags |= CURSOR_IGNORE_DELETE;
  }
  
2444
2445
2446
2447
2448
2449
2450

2451

2452
2453
2454
2455
2456
2457
2458
        if( x(pCtx, iBlk, iSnap) ) break;
        rc = multiCursorAdvance(pCsr, !bReverse);
      }
    }
  }

  lsmMCursorClose(pCsr);

  lsmFreeSnapshot(pDb->pEnv, pSnap);


  return rc;
}

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







>
|
>







2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
        if( x(pCtx, iBlk, iSnap) ) break;
        rc = multiCursorAdvance(pCsr, !bReverse);
      }
    }
  }

  lsmMCursorClose(pCsr);
  if( pSnap!=pDb->pWorker ){
    lsmFreeSnapshot(pDb->pEnv, pSnap);
  }

  return rc;
}

int lsmSortedLoadFreelist(
  lsm_db *pDb,                    /* Database handle (must be worker) */
  void **ppVal,                   /* OUT: Blob containing LSM free-list */
2516
2517
2518
2519
2520
2521
2522






























2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535

2536



2537
2538
2539



2540
2541
2542
2543

























2544



























2545
2546
2547
2548
2549
2550
2551
  if( *pRc==LSM_OK ){
    void *pKey;
    int nKey;
    multiCursorGetKey(pCsr, pCsr->aTree[1], &pCsr->eType, &pKey, &nKey);
    *pRc = sortedBlobSet(pCsr->pDb->pEnv, &pCsr->key, pKey, nKey);
  }
}































static int mcursorLocationOk(MultiCursor *pCsr, int bDeleteOk){
  int eType = pCsr->eType;
  int iKey;
  int i;
  int rdmask = 0;
  
  assert( pCsr->flags & (CURSOR_NEXT_OK|CURSOR_PREV_OK) );
  if( pCsr->flags & CURSOR_NEXT_OK ){
    rdmask = LSM_END_DELETE;
  }else{
    rdmask = LSM_START_DELETE;
  }





  if( (pCsr->flags & CURSOR_IGNORE_DELETE) && bDeleteOk==0 ){
    if( (eType & LSM_INSERT)==0 ) return 0;
  }



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


























  /* Check if this key has already been deleted by a range-delete */



























  iKey = pCsr->aTree[1];
  for(i=0; i<iKey; i++){
    int csrflags;
    multiCursorGetKey(pCsr, i, &csrflags, 0, 0);
    if( (rdmask & csrflags) ){
      const int SD_ED = (LSM_START_DELETE|LSM_END_DELETE);
      if( (csrflags & SD_ED)==SD_ED 







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





|


|
<
<
<
|
>

>
>
>



>
>
>




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







2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673



2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
  if( *pRc==LSM_OK ){
    void *pKey;
    int nKey;
    multiCursorGetKey(pCsr, pCsr->aTree[1], &pCsr->eType, &pKey, &nKey);
    *pRc = sortedBlobSet(pCsr->pDb->pEnv, &pCsr->key, pKey, nKey);
  }
}

#ifndef NDEBUG
static void assertCursorTree(MultiCursor *pCsr){
  int bRev = !!(pCsr->flags & CURSOR_PREV_OK);
  int *aSave = pCsr->aTree;
  int nSave = pCsr->nTree;
  int rc;

  pCsr->aTree = 0;
  pCsr->nTree = 0;
  rc = multiCursorAllocTree(pCsr);
  if( rc==LSM_OK ){
    int i;
    for(i=pCsr->nTree-1; i>0; i--){
      multiCursorDoCompare(pCsr, i, bRev);
    }

    assert( nSave==pCsr->nTree 
        && 0==memcmp(aSave, pCsr->aTree, sizeof(int)*nSave)
    );

    lsmFree(pCsr->pDb->pEnv, pCsr->aTree);
  }

  pCsr->aTree = aSave;
  pCsr->nTree = nSave;
}
#else
# define assertCursorTree(x)
#endif

static int mcursorLocationOk(MultiCursor *pCsr, int bDeleteOk){
  int eType = pCsr->eType;
  int iKey;
  int i;
  int rdmask;
  
  assert( pCsr->flags & (CURSOR_NEXT_OK|CURSOR_PREV_OK) );
  assertCursorTree(pCsr);




  rdmask = (pCsr->flags & CURSOR_NEXT_OK) ? LSM_END_DELETE : LSM_START_DELETE;

  /* If the cursor does not currently point to an actual database key (i.e.
  ** it points to a delete key, or the start or end of a range-delete), and
  ** the CURSOR_IGNORE_DELETE flag is set, skip past this entry.  */
  if( (pCsr->flags & CURSOR_IGNORE_DELETE) && bDeleteOk==0 ){
    if( (eType & LSM_INSERT)==0 ) return 0;
  }

  /* If the cursor points to a system key (free-list entry), and the
  ** CURSOR_IGNORE_SYSTEM flag is set, skip thie entry.  */
  if( (pCsr->flags & CURSOR_IGNORE_SYSTEM) && rtTopic(eType)!=0 ){
    return 0;
  }

#ifndef NDEBUG
  /* This block fires assert() statements to check one of the assumptions
  ** in the comment below - that if the lhs sub-cursor of a level undergoing
  ** a merge is valid, then all the rhs sub-cursors must be at EOF. 
  **
  ** Also assert that all rhs sub-cursors are either at EOF or point to
  ** a key that is not less than the level split-key.  */
  for(i=0; i<pCsr->nPtr; i++){
    SegmentPtr *pPtr = &pCsr->aPtr[i];
    Level *pLvl = pPtr->pLevel;
    if( pLvl->nRight && pPtr->pPg ){
      if( pPtr->pSeg==&pLvl->lhs ){
        int j;
        for(j=0; j<pLvl->nRight; j++) assert( pPtr[j+1].pPg==0 );
      }else{
        int res = sortedKeyCompare(pCsr->pDb->xCmp, 
            rtTopic(pPtr->eType), pPtr->pKey, pPtr->nKey,
            pLvl->iSplitTopic, pLvl->pSplitKey, pLvl->nSplitKey
        );
        assert( res>=0 );
      }
    }
  }
#endif

  /* Now check if this key has already been deleted by a range-delete. If 
  ** so, skip past it.
  **
  ** Assume, for the moment, that the tree contains no levels currently 
  ** undergoing incremental merge, and that this cursor is iterating forwards
  ** through the database keys. The cursor currently points to a key in
  ** level L. This key has already been deleted if any of the sub-cursors
  ** that point to levels newer than L (or to the in-memory tree) point to
  ** a key greater than the current key with the LSM_END_DELETE flag set.
  **
  ** Or, if the cursor is iterating backwards through data keys, if any
  ** such sub-cursor points to a key smaller than the current key with the
  ** LSM_START_DELETE flag set.
  **
  ** Why it works with levels undergoing a merge too:
  **
  ** When a cursor iterates forwards, the sub-cursors for the rhs of a 
  ** level are only activated once the lhs reaches EOF. So when iterating
  ** forwards, the keys visited are the same as if the level was completely
  ** merged.
  **
  ** If the cursor is iterating backwards, then the lhs sub-cursor is not 
  ** initialized until the last of the rhs sub-cursors has reached EOF.
  ** Additionally, if the START_DELETE flag is set on the last entry (in
  ** reverse order - so the entry with the smallest key) of a rhs sub-cursor,
  ** then a pseudo-key equal to the levels split-key with the END_DELETE
  ** flag set is visited by the sub-cursor.
  */ 
  iKey = pCsr->aTree[1];
  for(i=0; i<iKey; i++){
    int csrflags;
    multiCursorGetKey(pCsr, i, &csrflags, 0, 0);
    if( (rdmask & csrflags) ){
      const int SD_ED = (LSM_START_DELETE|LSM_END_DELETE);
      if( (csrflags & SD_ED)==SD_ED 
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573

2574
2575

2576



2577
2578
2579
2580
2581
2582
2583


2584
2585

2586
2587
2588

2589
2590
2591
2592
2593
2594
2595

2596


2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608


2609


2610
2611
2612

2613
2614
2615
2616
2617
2618
2619



2620




2621




2622



2623
2624

2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
          continue;
        }
      }
      return 0;
    }
  }

#if 0
  if( (iKey>0 && (rdmask & lsmTreeCursorFlags(pCsr->apTreeCsr[0]))) 
   || (iKey>1 && (rdmask & lsmTreeCursorFlags(pCsr->apTreeCsr[1]))) 
  ){
    return 0;
  }
  if( iKey>CURSOR_DATA_SYSTEM && (pCsr->flags & CURSOR_FLUSH_FREELIST) ){

    int eType;
    multiCursorGetKey(pCsr, CURSOR_DATA_SYSTEM, &eType, 0, 0);

    if( rdmask & eType ) return 0;



  }

  for(i=CURSOR_DATA_SEGMENT; i<iKey; i++){
    int iPtr = i-CURSOR_DATA_SEGMENT;
    if( pCsr->aPtr[iPtr].pPg && (pCsr->aPtr[iPtr].eType & rdmask) ){
      return 0;
    }


  }
#endif


  return 1;
}


static int multiCursorEnd(MultiCursor *pCsr, int bLast){
  int rc = LSM_OK;
  int i;

  pCsr->flags &= ~(CURSOR_NEXT_OK | CURSOR_PREV_OK);
  pCsr->flags |= (bLast ? CURSOR_PREV_OK : CURSOR_NEXT_OK);




  if( pCsr->apTreeCsr[0] ){
    rc = lsmTreeCursorEnd(pCsr->apTreeCsr[0], bLast);
  }
  if( rc==LSM_OK && pCsr->apTreeCsr[1] ){
    rc = lsmTreeCursorEnd(pCsr->apTreeCsr[1], bLast);
  }

  pCsr->iFree = 0;

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





    rc = segmentPtrEnd(pCsr, pPtr, bLast);
    if( rc==LSM_OK && bLast==0 && pLvl->nRight && pPtr->pSeg==&pLvl->lhs ){
      int iRhs;

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



      }




      i += pLvl->nRight;




    }



  }


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

  if( rc==LSM_OK ){
    rc = multiCursorAllocTree(pCsr);
  }
  if( rc==LSM_OK ){
    for(i=pCsr->nTree-1; i>0; i--){
      multiCursorDoCompare(pCsr, i, bLast);
    }
  }

  multiCursorCacheKey(pCsr, &rc);
  if( rc==LSM_OK && mcursorLocationOk(pCsr, 0)==0 ){
    if( bLast ){
      rc = lsmMCursorPrev(pCsr);
    }else{
      rc = lsmMCursorNext(pCsr);
    }
  }

  return rc;
}


int mcursorSave(MultiCursor *pCsr){
  int rc = LSM_OK;
  if( pCsr->aTree ){







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

>







>

>
>
|
|
|
<
<


<
<



>
>

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


>






|

<
<
<
|
<
<
<
<
<
<
<
<
<
<
<







2758
2759
2760
2761
2762
2763
2764
2765



2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778




2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802


2803
2804


2805
2806
2807
2808
2809
2810
2811
2812
2813

2814
2815
2816
2817


2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848



2849











2850
2851
2852
2853
2854
2855
2856
          continue;
        }
      }
      return 0;
    }
  }

  /* The current cursor position is one this cursor should visit. Return 1. */



  return 1;
}

static int multiCursorSetupTree(MultiCursor *pCsr, int bRev){
  int rc;

  rc = multiCursorAllocTree(pCsr);
  if( rc==LSM_OK ){
    int i;
    for(i=pCsr->nTree-1; i>0; i--){
      multiCursorDoCompare(pCsr, i, bRev);
    }
  }





  assertCursorTree(pCsr);
  multiCursorCacheKey(pCsr, &rc);

  if( rc==LSM_OK && mcursorLocationOk(pCsr, 0)==0 ){
    rc = multiCursorAdvance(pCsr, bRev);
  }
  return rc;
}


static int multiCursorEnd(MultiCursor *pCsr, int bLast){
  int rc = LSM_OK;
  int i;

  pCsr->flags &= ~(CURSOR_NEXT_OK | CURSOR_PREV_OK);
  pCsr->flags |= (bLast ? CURSOR_PREV_OK : CURSOR_NEXT_OK);
  pCsr->iFree = 0;

  /* Position the two in-memory tree cursors */
  for(i=0; rc==LSM_OK && i<2; i++){
    if( pCsr->apTreeCsr[i] ){
      rc = lsmTreeCursorEnd(pCsr->apTreeCsr[i], bLast);
    }


  }



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

    if( bLast ){
      for(iRhs=0; iRhs<pLvl->nRight && rc==LSM_OK; iRhs++){
        rc = segmentPtrEnd(pCsr, &pPtr[iRhs+1], 1);

        if( pPtr[iRhs+1].pPg ) bHit = 1;
      }
      if( bHit==0 && rc==LSM_OK ){
        rc = segmentPtrEnd(pCsr, pPtr, 1);


      }else{
        segmentPtrReset(pPtr);
      }
    }else{
      int bLhs = (pPtr->pSeg==&pLvl->lhs);
      assert( pPtr->pSeg==&pLvl->lhs || pPtr->pSeg==&pLvl->aRhs[0] );

      if( bLhs ){
        rc = segmentPtrEnd(pCsr, pPtr, 0);
        if( pPtr->pKey ) bHit = 1;
      }
      for(iRhs=0; iRhs<pLvl->nRight && rc==LSM_OK; iRhs++){
        if( bHit ){
          segmentPtrReset(&pPtr[iRhs+1]);
        }else{
          rc = sortedRhsFirst(pCsr, pLvl, &pPtr[iRhs+bLhs]);
        }
      }
    }
    i += pLvl->nRight;
  }

  /* And the b-tree cursor, if applicable */
  if( rc==LSM_OK && pCsr->pBtCsr ){
    assert( bLast==0 );
    rc = btreeCursorFirst(pCsr->pBtCsr);
  }

  if( rc==LSM_OK ){
    rc = multiCursorSetupTree(pCsr, bLast);
  }



  











  return rc;
}


int mcursorSave(MultiCursor *pCsr){
  int rc = LSM_OK;
  if( pCsr->aTree ){
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896

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

static void flCsrAdvance(MultiCursor *pCsr){
  assert( pCsr->flags & CURSOR_FLUSH_FREELIST );







|







3084
3085
3086
3087
3088
3089
3090
3091
3092
3093
3094
3095
3096
3097
3098

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

static void flCsrAdvance(MultiCursor *pCsr){
  assert( pCsr->flags & CURSOR_FLUSH_FREELIST );
2920
2921
2922
2923
2924
2925
2926


2927
2928
2929
2930
2931
2932
2933
}

static int multiCursorAdvance(MultiCursor *pCsr, int bReverse){
  int rc = LSM_OK;                /* Return Code */
  if( lsmMCursorValid(pCsr) ){
    do {
      int iKey = pCsr->aTree[1];



      /* If this multi-cursor is advancing forwards, and the sub-cursor
      ** being advanced is the one that separator keys may be being read
      ** from, record the current absolute pointer value.  */
      if( pCsr->pPrevMergePtr ){
        if( iKey==(CURSOR_DATA_SEGMENT+pCsr->nPtr) ){
          assert( pCsr->pBtCsr );







>
>







3122
3123
3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
}

static int multiCursorAdvance(MultiCursor *pCsr, int bReverse){
  int rc = LSM_OK;                /* Return Code */
  if( lsmMCursorValid(pCsr) ){
    do {
      int iKey = pCsr->aTree[1];

      assertCursorTree(pCsr);

      /* If this multi-cursor is advancing forwards, and the sub-cursor
      ** being advanced is the one that separator keys may be being read
      ** from, record the current absolute pointer value.  */
      if( pCsr->pPrevMergePtr ){
        if( iKey==(CURSOR_DATA_SEGMENT+pCsr->nPtr) ){
          assert( pCsr->pBtCsr );
2958
2959
2960
2961
2962
2963
2964

2965
2966
2967
2968
2969
2970
2971
        rc = segmentCursorAdvance(pCsr, iKey-CURSOR_DATA_SEGMENT, bReverse);
      }
      if( rc==LSM_OK ){
        int i;
        for(i=(iKey+pCsr->nTree)/2; i>0; i=i/2){
          multiCursorDoCompare(pCsr, i, bReverse);
        }

      }
    }while( mcursorAdvanceOk(pCsr, bReverse, &rc)==0 );
  }
  return rc;
}

int lsmMCursorNext(MultiCursor *pCsr){







>







3162
3163
3164
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
        rc = segmentCursorAdvance(pCsr, iKey-CURSOR_DATA_SEGMENT, bReverse);
      }
      if( rc==LSM_OK ){
        int i;
        for(i=(iKey+pCsr->nTree)/2; i>0; i=i/2){
          multiCursorDoCompare(pCsr, i, bReverse);
        }
        assertCursorTree(pCsr);
      }
    }while( mcursorAdvanceOk(pCsr, bReverse, &rc)==0 );
  }
  return rc;
}

int lsmMCursorNext(MultiCursor *pCsr){
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
3208
3209
3210
3211

    do {
      Page *pPg = 0;
      u8 *aData;
      int nData;
      int flags;

      rc = lsmFsDbPageGet(pFS, iPg, &pPg);
      if( rc!=LSM_OK ) break;

      aData = fsPageData(pPg, &nData);
      flags = pageGetFlags(aData, nData);
      if( flags&SEGMENT_BTREE_FLAG ){
        Page **apNew = (Page **)lsmRealloc(
            pEnv, apHier, sizeof(Page *)*(nHier+1)







|







3402
3403
3404
3405
3406
3407
3408
3409
3410
3411
3412
3413
3414
3415
3416

    do {
      Page *pPg = 0;
      u8 *aData;
      int nData;
      int flags;

      rc = lsmFsDbPageGet(pFS, pSeg, iPg, &pPg);
      if( rc!=LSM_OK ) break;

      aData = fsPageData(pPg, &nData);
      flags = pageGetFlags(aData, nData);
      if( flags&SEGMENT_BTREE_FLAG ){
        Page **apNew = (Page **)lsmRealloc(
            pEnv, apHier, sizeof(Page *)*(nHier+1)
3643
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657

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







|







3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862

  if( pCsr->pBtCsr ){
    rc = LSM_OK;
    iFPtr = pMW->pLevel->pNext->lhs.iFirst;
  }else if( pCsr->nPtr>0 ){
    Segment *pSeg;
    pSeg = pCsr->aPtr[pCsr->nPtr-1].pSeg;
    rc = lsmFsDbPageGet(pMW->pDb->pFS, pSeg, pSeg->iFirst, &pPg);
    if( rc==LSM_OK ){
      u8 *aData;                    /* Buffer for page pPg */
      int nData;                    /* Size of aData[] in bytes */
      aData = fsPageData(pPg, &nData);
      iFPtr = pageGetPtr(aData, nData);
      lsmFsPageRelease(pPg);
    }
3779
3780
3781
3782
3783
3784
3785
3786
3787
3788
3789
3790
3791
3792
3793
3794
3795
3796
3797
3798
3799
3800
3801
3802
3803
3804
3805
3806
3807
3808
3809
    }
  }

  return rc;
}


static int multiCursorSetupTree(MultiCursor *pCsr, int bRev){
  int rc;

  assert( pCsr->aTree==0 );

  rc = multiCursorAllocTree(pCsr);
  if( rc==LSM_OK ){
    int i;
    for(i=pCsr->nTree-1; i>0; i--){
      multiCursorDoCompare(pCsr, i, bRev);
    }
  }

  multiCursorCacheKey(pCsr, &rc);
  return rc;
}

/*
** Free all resources allocated by mergeWorkerInit().
*/
static void mergeWorkerShutdown(MergeWorker *pMW, int *pRc){
  int i;                          /* Iterator variable */
  int rc = *pRc;
  MultiCursor *pCsr = pMW->pCsr;







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







3984
3985
3986
3987
3988
3989
3990

















3991
3992
3993
3994
3995
3996
3997
    }
  }

  return rc;
}



















/*
** Free all resources allocated by mergeWorkerInit().
*/
static void mergeWorkerShutdown(MergeWorker *pMW, int *pRc){
  int i;                          /* Iterator variable */
  int rc = *pRc;
  MultiCursor *pCsr = pMW->pCsr;
3985
3986
3987
3988
3989
3990
3991
3992
3993
3994
3995
3996
3997
3998
3999

  iVal = pCsr->aTree[1];
  mergeRangeDeletes(pCsr, &iVal, &eType);

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








|







4173
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
4184
4185
4186
4187

  iVal = pCsr->aTree[1];
  mergeRangeDeletes(pCsr, &iVal, &eType);

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

4015
4016
4017
4018
4019
4020
4021
4022
4023
4024
4025
4026
4027
4028
4029
4030
4031
4032
4033
4034
4035
4036
4037
4038
4039
4040
4041
4042
4043
4044
4045
4046
4047
4048
4049
4050
4051
4052
    }
  }

  /* Advance the cursor to the next input record (assuming one exists). */
  assert( lsmMCursorValid(pMW->pCsr) );
  if( rc==LSM_OK ) rc = lsmMCursorNext(pMW->pCsr);

  /* If the cursor is at EOF, the merge is finished. Release all page
  ** references currently held by the merge worker and inform the 
  ** FileSystem object that no further pages will be appended to either 
  ** the main or separators array. 
  */
  if( rc==LSM_OK && !lsmMCursorValid(pMW->pCsr) ){

    mergeWorkerShutdown(pMW, &rc);
    if( pSeg->iFirst ){
      rc = lsmFsSortedFinish(pDb->pFS, pSeg);
    }

#ifdef LSM_DEBUG_EXPENSIVE
    if( rc==LSM_OK ){
#if 0
      rc = assertBtreeOk(pDb, pSeg);
      if( pMW->pCsr->pBtCsr ){
        Segment *pNext = &pMW->pLevel->pNext->lhs;
        rc = assertPointersOk(pDb, pSeg, pNext, 0);
      }
#endif
    }
#endif
  }
  return rc;
}

static int mergeWorkerDone(MergeWorker *pMW){
  return pMW->pCsr==0 || !lsmMCursorValid(pMW->pCsr);
}








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







4203
4204
4205
4206
4207
4208
4209
























4210
4211
4212
4213
4214
4215
4216
    }
  }

  /* Advance the cursor to the next input record (assuming one exists). */
  assert( lsmMCursorValid(pMW->pCsr) );
  if( rc==LSM_OK ) rc = lsmMCursorNext(pMW->pCsr);

























  return rc;
}

static int mergeWorkerDone(MergeWorker *pMW){
  return pMW->pCsr==0 || !lsmMCursorValid(pMW->pCsr);
}

4070
4071
4072
4073
4074
4075
4076
4077

4078
4079
4080
4081
4082
4083
4084
  int eTree,                      /* One of the TREE_XXX constants */
  int *pnWrite                    /* OUT: Number of database pages written */
){
  int rc = LSM_OK;                /* Return Code */
  MultiCursor *pCsr = 0;
  Level *pNext = 0;               /* The current top level */
  Level *pNew;                    /* The new level itself */
  Segment *pDel = 0;              /* Delete separators from this segment */

  int nWrite = 0;                 /* Number of database pages written */
  Freelist freelist;

  if( eTree!=TREE_NONE ){
    rc = lsmShmCacheChunks(pDb, pDb->treehdr.nChunk);
  }








|
>







4234
4235
4236
4237
4238
4239
4240
4241
4242
4243
4244
4245
4246
4247
4248
4249
  int eTree,                      /* One of the TREE_XXX constants */
  int *pnWrite                    /* OUT: Number of database pages written */
){
  int rc = LSM_OK;                /* Return Code */
  MultiCursor *pCsr = 0;
  Level *pNext = 0;               /* The current top level */
  Level *pNew;                    /* The new level itself */
  Segment *pLinked = 0;           /* Delete separators from this segment */
  Level *pDel = 0;                /* Delete this entire level */
  int nWrite = 0;                 /* Number of database pages written */
  Freelist freelist;

  if( eTree!=TREE_NONE ){
    rc = lsmShmCacheChunks(pDb, pDb->treehdr.nChunk);
  }

4101
4102
4103
4104
4105
4106
4107
4108
4109
4110

4111


4112
4113

4114
4115
4116
4117
4118
4119
4120
  pCsr = multiCursorNew(pDb, &rc);
  if( pCsr ){
    pCsr->pDb = pDb;
    rc = multiCursorVisitFreelist(pCsr);
    if( rc==LSM_OK ){
      rc = multiCursorAddTree(pCsr, pDb->pWorker, eTree);
    }
    if( rc==LSM_OK 
        && pNext && pNext->pMerge==0 && pNext->lhs.iRoot 
        && (eTree!=TREE_NONE || (pNext->flags & LEVEL_FREELIST_ONLY))

    ){


      pDel = &pNext->lhs;
      rc = btreeCursorNew(pDb, pDel, &pCsr->pBtCsr);

    }

    /* If this will be the only segment in the database, discard any delete
    ** markers present in the in-memory tree.  */
    if( pNext==0 ){
      multiCursorIgnoreDelete(pCsr);
    }







|
<
|
>
|
>
>
|
|
>







4266
4267
4268
4269
4270
4271
4272
4273

4274
4275
4276
4277
4278
4279
4280
4281
4282
4283
4284
4285
4286
4287
4288
  pCsr = multiCursorNew(pDb, &rc);
  if( pCsr ){
    pCsr->pDb = pDb;
    rc = multiCursorVisitFreelist(pCsr);
    if( rc==LSM_OK ){
      rc = multiCursorAddTree(pCsr, pDb->pWorker, eTree);
    }
    if( rc==LSM_OK && pNext && pNext->pMerge==0 ){

      if( (pNext->flags & LEVEL_FREELIST_ONLY) ){
        pDel = pNext;
        pCsr->aPtr = lsmMallocZeroRc(pDb->pEnv, sizeof(SegmentPtr), &rc);
        multiCursorAddOne(pCsr, pNext, &rc);
      }else if( eTree!=TREE_NONE && pNext->lhs.iRoot ){
        pLinked = &pNext->lhs;
        rc = btreeCursorNew(pDb, pLinked, &pCsr->pBtCsr);
      }
    }

    /* If this will be the only segment in the database, discard any delete
    ** markers present in the in-memory tree.  */
    if( pNext==0 ){
      multiCursorIgnoreDelete(pCsr);
    }
4141
4142
4143
4144
4145
4146
4147

4148


4149
4150
4151
4152
4153
4154
4155
4156
4157
4158
4159
4160
4161
4162
4163

4164





4165
4166

4167
4168
4169
4170
4171
4172
4173
4174
    mergeworker.bFlush = 1;

    /* Do the work to create the new merged segment on disk */
    if( rc==LSM_OK ) rc = lsmMCursorFirst(pCsr);
    while( rc==LSM_OK && mergeWorkerDone(&mergeworker)==0 ){
      rc = mergeWorkerStep(&mergeworker);
    }

    assert( rc!=LSM_OK || mergeworker.nWork==0 || pNew->lhs.iFirst );



    nWrite = mergeworker.nWork;
    mergeWorkerShutdown(&mergeworker, &rc);
    pNew->flags &= ~LEVEL_INCOMPLETE;
    if( eTree==TREE_NONE ){
      pNew->flags |= LEVEL_FREELIST_ONLY;
    }
    pNew->pMerge = 0;
  }

  if( rc!=LSM_OK || pNew->lhs.iFirst==0 ){
    assert( rc!=LSM_OK || pDb->pWorker->freelist.nEntry==0 );
    lsmDbSnapshotSetLevel(pDb->pWorker, pNext);
    sortedFreeLevel(pDb->pEnv, pNew);
  }else{

    if( pDel ) pDel->iRoot = 0;






#if 0

    lsmSortedDumpStructure(pDb, pDb->pWorker, 0, 0, "new-toplevel");
#endif

    if( freelist.nEntry ){
      Freelist *p = &pDb->pWorker->freelist;
      lsmFree(pDb->pEnv, p->aEntry);
      memcpy(p, &freelist, sizeof(freelist));
      freelist.aEntry = 0;







>

>
>
|

<












>
|
>
>
>
>
>
|
|
>
|







4309
4310
4311
4312
4313
4314
4315
4316
4317
4318
4319
4320
4321

4322
4323
4324
4325
4326
4327
4328
4329
4330
4331
4332
4333
4334
4335
4336
4337
4338
4339
4340
4341
4342
4343
4344
4345
4346
4347
4348
4349
4350
4351
    mergeworker.bFlush = 1;

    /* Do the work to create the new merged segment on disk */
    if( rc==LSM_OK ) rc = lsmMCursorFirst(pCsr);
    while( rc==LSM_OK && mergeWorkerDone(&mergeworker)==0 ){
      rc = mergeWorkerStep(&mergeworker);
    }
    mergeWorkerShutdown(&mergeworker, &rc);
    assert( rc!=LSM_OK || mergeworker.nWork==0 || pNew->lhs.iFirst );
    if( rc==LSM_OK && pNew->lhs.iFirst ){
      rc = lsmFsSortedFinish(pDb->pFS, &pNew->lhs);
    }
    nWrite = mergeworker.nWork;

    pNew->flags &= ~LEVEL_INCOMPLETE;
    if( eTree==TREE_NONE ){
      pNew->flags |= LEVEL_FREELIST_ONLY;
    }
    pNew->pMerge = 0;
  }

  if( rc!=LSM_OK || pNew->lhs.iFirst==0 ){
    assert( rc!=LSM_OK || pDb->pWorker->freelist.nEntry==0 );
    lsmDbSnapshotSetLevel(pDb->pWorker, pNext);
    sortedFreeLevel(pDb->pEnv, pNew);
  }else{
    if( pLinked ){
      pLinked->iRoot = 0;
    }else if( pDel ){
      assert( pNew->pNext==pDel );
      pNew->pNext = pDel->pNext;
      lsmFsSortedDelete(pDb->pFS, pDb->pWorker, 1, &pDel->lhs);
      sortedFreeLevel(pDb->pEnv, pDel);
    }

#if LSM_LOG_STRUCTURE
    lsmSortedDumpStructure(pDb, pDb->pWorker, LSM_LOG_DATA, 0, "new-toplevel");
#endif

    if( freelist.nEntry ){
      Freelist *p = &pDb->pWorker->freelist;
      lsmFree(pDb->pEnv, p->aEntry);
      memcpy(p, &freelist, sizeof(freelist));
      freelist.aEntry = 0;
4250
4251
4252
4253
4254
4255
4256
4257


4258
4259
4260
4261
4262
4263
4264
    pTopLevel = lsmDbSnapshotLevel(pDb->pWorker);
    pNew->pNext = p;
    for(pp=&pTopLevel; *pp!=pLevel; pp=&((*pp)->pNext));
    *pp = pNew;
    lsmDbSnapshotSetLevel(pDb->pWorker, pTopLevel);

    /* Determine whether or not the next separators will be linked in */
    if( pNext && pNext->pMerge==0 && pNext->lhs.iRoot ){


      bUseNext = 1;
    }
  }

  /* Allocate the merge object */
  nByte = sizeof(Merge) + sizeof(MergeInput) * (nMerge + bUseNext);
  pMerge = (Merge *)lsmMallocZeroRc(pDb->pEnv, nByte, &rc);







|
>
>







4427
4428
4429
4430
4431
4432
4433
4434
4435
4436
4437
4438
4439
4440
4441
4442
4443
    pTopLevel = lsmDbSnapshotLevel(pDb->pWorker);
    pNew->pNext = p;
    for(pp=&pTopLevel; *pp!=pLevel; pp=&((*pp)->pNext));
    *pp = pNew;
    lsmDbSnapshotSetLevel(pDb->pWorker, pTopLevel);

    /* Determine whether or not the next separators will be linked in */
    if( pNext && pNext->pMerge==0 && pNext->lhs.iRoot && pNext 
     && (bFreeOnly==0 || (pNext->flags & LEVEL_FREELIST_ONLY))
    ){
      bUseNext = 1;
    }
  }

  /* Allocate the merge object */
  nByte = sizeof(Merge) + sizeof(MergeInput) * (nMerge + bUseNext);
  pMerge = (Merge *)lsmMallocZeroRc(pDb->pEnv, nByte, &rc);
4299
4300
4301
4302
4303
4304
4305

4306
4307
4308
4309
4310
4311
4312
  **      separators attached to the LHS of the following level, or neither.
  **
  ** If the new level is the lowest (oldest) in the db, discard any
  ** delete keys. Key annihilation.
  */
  pCsr = multiCursorNew(pDb, &rc);
  if( pCsr ){

    rc = multiCursorAddRhs(pCsr, pLevel);
  }
  if( rc==LSM_OK && pMerge->nInput > pLevel->nRight ){
    rc = btreeCursorNew(pDb, &pNext->lhs, &pCsr->pBtCsr);
  }else if( pNext ){
    multiCursorReadSeparators(pCsr);
  }else{







>







4478
4479
4480
4481
4482
4483
4484
4485
4486
4487
4488
4489
4490
4491
4492
  **      separators attached to the LHS of the following level, or neither.
  **
  ** If the new level is the lowest (oldest) in the db, discard any
  ** delete keys. Key annihilation.
  */
  pCsr = multiCursorNew(pDb, &rc);
  if( pCsr ){
    pCsr->flags |= CURSOR_NEXT_OK;
    rc = multiCursorAddRhs(pCsr, pLevel);
  }
  if( rc==LSM_OK && pMerge->nInput > pLevel->nRight ){
    rc = btreeCursorNew(pDb, &pNext->lhs, &pCsr->pBtCsr);
  }else if( pNext ){
    multiCursorReadSeparators(pCsr);
  }else{
4427
4428
4429
4430
4431
4432
4433
4434
4435
4436
4437
4438
4439
4440
4441
4442
  ** with the largest number of right-hand segments. Work on it. */
  for(pLevel=pTopLevel; pLevel; pLevel=pLevel->pNext){
    if( pLevel->nRight==0 && pThis && pLevel->iAge==pThis->iAge ){
      nThis++;
    }else{
      if( nThis>nBest ){
        if( (pLevel->iAge!=pThis->iAge+1)
            || (pLevel->nRight==0 && sortedCountLevels(pLevel)<=pDb->nMerge)
          ){
          pBest = pThis;
          nBest = nThis;
        }
      }
      if( pLevel->nRight ){
        if( pLevel->nRight>nBest ){
          nBest = pLevel->nRight;







|
|







4607
4608
4609
4610
4611
4612
4613
4614
4615
4616
4617
4618
4619
4620
4621
4622
  ** with the largest number of right-hand segments. Work on it. */
  for(pLevel=pTopLevel; pLevel; pLevel=pLevel->pNext){
    if( pLevel->nRight==0 && pThis && pLevel->iAge==pThis->iAge ){
      nThis++;
    }else{
      if( nThis>nBest ){
        if( (pLevel->iAge!=pThis->iAge+1)
         || (pLevel->nRight==0 && sortedCountLevels(pLevel)<=pDb->nMerge)
        ){
          pBest = pThis;
          nBest = nThis;
        }
      }
      if( pLevel->nRight ){
        if( pLevel->nRight>nBest ){
          nBest = pLevel->nRight;
4491
4492
4493
4494
4495
4496
4497

























































































































































4498
4499
4500
4501
4502
4503
4504
  if( pTop && pTop->iAge==0
   && (pTop->nRight || sortedCountLevels(pTop)>=pDb->nMerge)
  ){
    return 1;
  }
  return 0;
}


























































































































































static int sortedWork(
  lsm_db *pDb,                    /* Database handle. Must be worker. */
  int nWork,                      /* Number of pages of work to do */
  int nMerge,                     /* Try to merge this many levels at once */
  int bFlush,                     /* Set if call is to make room for a flush */
  int *pnWrite                    /* OUT: Actual number of pages written */







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







4671
4672
4673
4674
4675
4676
4677
4678
4679
4680
4681
4682
4683
4684
4685
4686
4687
4688
4689
4690
4691
4692
4693
4694
4695
4696
4697
4698
4699
4700
4701
4702
4703
4704
4705
4706
4707
4708
4709
4710
4711
4712
4713
4714
4715
4716
4717
4718
4719
4720
4721
4722
4723
4724
4725
4726
4727
4728
4729
4730
4731
4732
4733
4734
4735
4736
4737
4738
4739
4740
4741
4742
4743
4744
4745
4746
4747
4748
4749
4750
4751
4752
4753
4754
4755
4756
4757
4758
4759
4760
4761
4762
4763
4764
4765
4766
4767
4768
4769
4770
4771
4772
4773
4774
4775
4776
4777
4778
4779
4780
4781
4782
4783
4784
4785
4786
4787
4788
4789
4790
4791
4792
4793
4794
4795
4796
4797
4798
4799
4800
4801
4802
4803
4804
4805
4806
4807
4808
4809
4810
4811
4812
4813
4814
4815
4816
4817
4818
4819
4820
4821
4822
4823
4824
4825
4826
4827
4828
4829
4830
4831
4832
4833
4834
4835
4836
4837
  if( pTop && pTop->iAge==0
   && (pTop->nRight || sortedCountLevels(pTop)>=pDb->nMerge)
  ){
    return 1;
  }
  return 0;
}

typedef struct MoveBlockCtx MoveBlockCtx;
struct MoveBlockCtx {
  int iSeen;                      /* Previous free block on list */
  int iFrom;                      /* Total number of blocks in file */
};

static int moveBlockCb(void *pCtx, int iBlk, i64 iSnapshot){
  MoveBlockCtx *p = (MoveBlockCtx *)pCtx;
  assert( p->iFrom==0 );
  if( iBlk==(p->iSeen-1) ){
    p->iSeen = iBlk;
    return 0;
  }
  p->iFrom = p->iSeen-1;
  return 1;
}

/*
** This function is called to further compact a database for which all 
** of the content has already been merged into a single segment. If 
** possible, it moves the contents of a single block from the end of the
** file to a free-block that lies closer to the start of the file (allowing
** the file to be eventually truncated).
*/
static int sortedMoveBlock(lsm_db *pDb, int *pnWrite){
  Snapshot *p = pDb->pWorker;
  Level *pLvl = lsmDbSnapshotLevel(p);
  int iFrom;                      /* Block to move */
  int iTo;                        /* Destination to move block to */
  int rc;                         /* Return code */

  MoveBlockCtx sCtx;

  assert( pLvl->pNext==0 && pLvl->nRight==0 );
  assert( p->redirect.n<=LSM_MAX_BLOCK_REDIRECTS );

  *pnWrite = 0;

  /* Check that the redirect array is not already full. If it is, return
  ** without moving any database content.  */
  if( p->redirect.n>=LSM_MAX_BLOCK_REDIRECTS ) return LSM_OK;

  /* Find the last block of content in the database file. Do this by 
  ** traversing the free-list in reverse (descending block number) order.
  ** The first block not on the free list is the one that will be moved.
  ** Since the db consists of a single segment, there is no ambiguity as
  ** to which segment the block belongs to.  */
  sCtx.iSeen = p->nBlock+1;
  sCtx.iFrom = 0;
  rc = lsmWalkFreelist(pDb, 1, moveBlockCb, &sCtx);
  if( rc!=LSM_OK || sCtx.iFrom==0 ) return rc;
  iFrom = sCtx.iFrom;

  /* Find the first free block in the database, ignoring block 1. Block
  ** 1 is tricky as it is smaller than the other blocks.  */
  rc = lsmBlockAllocate(pDb, iFrom, &iTo);
  if( rc!=LSM_OK || iTo==0 ) return rc;
  assert( iTo!=1 && iTo<iFrom );

  rc = lsmFsMoveBlock(pDb->pFS, &pLvl->lhs, iTo, iFrom);
  if( rc==LSM_OK ){
    if( p->redirect.a==0 ){
      int nByte = sizeof(struct RedirectEntry) * LSM_MAX_BLOCK_REDIRECTS;
      p->redirect.a = lsmMallocZeroRc(pDb->pEnv, nByte, &rc);
    }
    if( rc==LSM_OK ){
      memmove(&p->redirect.a[1], &p->redirect.a[0], 
          sizeof(struct RedirectEntry) * p->redirect.n
      );
      p->redirect.a[0].iFrom = iFrom;
      p->redirect.a[0].iTo = iTo;
      p->redirect.n++;

      rc = lsmBlockFree(pDb, iFrom);

      *pnWrite = lsmFsBlockSize(pDb->pFS) / lsmFsPageSize(pDb->pFS);
      pLvl->lhs.pRedirect = &p->redirect;
    }
  }

#if LSM_LOG_STRUCTURE
  if( rc==LSM_OK ){
    char aBuf[64];
    sprintf(aBuf, "move-block %d/%d", p->redirect.n-1, LSM_MAX_BLOCK_REDIRECTS);
    lsmSortedDumpStructure(pDb, pDb->pWorker, LSM_LOG_DATA, 0, aBuf);
  }
#endif
  return rc;
}

/*
*/
static int mergeInsertFreelistSegments(
  lsm_db *pDb, 
  int nFree,
  MergeWorker *pMW
){
  int rc = LSM_OK;
  if( nFree>0 ){
    MultiCursor *pCsr = pMW->pCsr;
    Level *pLvl = pMW->pLevel;
    SegmentPtr *aNew1;
    Segment *aNew2;

    Level *pIter;
    Level *pNext;
    int i = 0;

    aNew1 = (SegmentPtr *)lsmMallocZeroRc(
        pDb->pEnv, sizeof(SegmentPtr) * (pCsr->nPtr+nFree), &rc
    );
    if( rc ) return rc;
    memcpy(&aNew1[nFree], pCsr->aPtr, sizeof(SegmentPtr)*pCsr->nPtr);
    pCsr->nPtr += nFree;
    lsmFree(pDb->pEnv, pCsr->aTree);
    lsmFree(pDb->pEnv, pCsr->aPtr);
    pCsr->aTree = 0;
    pCsr->aPtr = aNew1;

    aNew2 = (Segment *)lsmMallocZeroRc(
        pDb->pEnv, sizeof(Segment) * (pLvl->nRight+nFree), &rc
    );
    if( rc ) return rc;
    memcpy(&aNew2[nFree], pLvl->aRhs, sizeof(Segment)*pLvl->nRight);
    pLvl->nRight += nFree;
    lsmFree(pDb->pEnv, pLvl->aRhs);
    pLvl->aRhs = aNew2;

    for(pIter=pDb->pWorker->pLevel; rc==LSM_OK && pIter!=pLvl; pIter=pNext){
      Segment *pSeg = &pLvl->aRhs[i];
      memcpy(pSeg, &pIter->lhs, sizeof(Segment));

      pCsr->aPtr[i].pSeg = pSeg;
      pCsr->aPtr[i].pLevel = pLvl;
      rc = segmentPtrEnd(pCsr, &pCsr->aPtr[i], 0);

      pDb->pWorker->pLevel = pNext = pIter->pNext;
      sortedFreeLevel(pDb->pEnv, pIter);
      i++;
    }
    assert( i==nFree );
    assert( rc!=LSM_OK || pDb->pWorker->pLevel==pLvl );

    for(i=nFree; i<pCsr->nPtr; i++){
      pCsr->aPtr[i].pSeg = &pLvl->aRhs[i];
    }

    lsmFree(pDb->pEnv, pMW->aGobble);
    pMW->aGobble = 0;
  }
  return rc;
}

static int sortedWork(
  lsm_db *pDb,                    /* Database handle. Must be worker. */
  int nWork,                      /* Number of pages of work to do */
  int nMerge,                     /* Try to merge this many levels at once */
  int bFlush,                     /* Set if call is to make room for a flush */
  int *pnWrite                    /* OUT: Actual number of pages written */
4514
4515
4516
4517
4518
4519
4520







4521
4522
4523


4524
4525




4526
4527

4528
4529
4530
4531

4532




































4533
4534
4535

4536
4537
4538
4539
4540
4541
4542
4543
4544
4545
4546
4547
4548
4549
4550
4551
4552
4553




4554



















4555
4556
4557
4558
4559
4560
4561
4562
4563
4564
4565
4566
4567
4568
4569
4570
4571
4572
4573
4574
4575
4576
4577
4578
4579
4580
4581
4582
4583
4584
4585
4586
4587
4588
4589
4590
4591
4592
4593
4594
4595
4596
4597






4598
4599
4600
4601
4602
4603
4604

4605
4606
4607
4608
4609
4610
4611
4612
4613
4614
4615
    Level *pLevel = 0;

    /* Find a level to work on. */
    rc = sortedSelectLevel(pDb, nMerge, &pLevel);
    assert( rc==LSM_OK || pLevel==0 );

    if( pLevel==0 ){







      /* Could not find any work to do. Finished. */
      break;
    }else{


      MergeWorker mergeworker;    /* State used to work on the level merge */





      rc = mergeWorkerInit(pDb, pLevel, &mergeworker);
      assert( mergeworker.nWork==0 );

      while( rc==LSM_OK 
          && 0==mergeWorkerDone(&mergeworker) 
          && mergeworker.nWork<nRemaining 
      ){

        rc = mergeWorkerStep(&mergeworker);




































      }
      nRemaining -= LSM_MAX(mergeworker.nWork, 1);


      /* Check if the merge operation is completely finished. If so, the
      ** Merge object and the right-hand-side of the level can be deleted. 
      **
      ** Otherwise, gobble up (declare eligible for recycling) any pages
      ** from rhs segments for which the content has been completely merged
      ** into the lhs of the level.
      */
      if( rc==LSM_OK ){
        if( mergeWorkerDone(&mergeworker)==0 ){
          int i;
          for(i=0; i<pLevel->nRight; i++){
            SegmentPtr *pGobble = &mergeworker.pCsr->aPtr[i];
            if( pGobble->pSeg->iRoot ){
              rc = sortedBtreeGobble(pDb, mergeworker.pCsr, i);
            }else if( mergeworker.aGobble[i] ){
              lsmFsGobble(pDb, pGobble->pSeg, &mergeworker.aGobble[i], 1);
            }
          }




        }else if( pLevel->lhs.iFirst==0 ){



















          /* If the new level is completely empty, remove it from the 
          ** database snapshot. This can only happen if all input keys were
          ** annihilated. Since keys are only annihilated if the new level
          ** is the last in the linked list (contains the most ancient of
          ** database content), this guarantees that pLevel->pNext==0.  */ 

          Level *pTop;          /* Top level of worker snapshot */
          Level **pp;           /* Read/write iterator for Level.pNext list */
          int i;
          assert( pLevel->pNext==0 );

          /* Remove the level from the worker snapshot. */
          pTop = lsmDbSnapshotLevel(pWorker);
          for(pp=&pTop; *pp!=pLevel; pp=&((*pp)->pNext));
          *pp = pLevel->pNext;
          lsmDbSnapshotSetLevel(pWorker, pTop);

          /* Free the Level structure. */
          lsmFsSortedDelete(pDb->pFS, pWorker, 1, &pLevel->lhs);
          for(i=0; i<pLevel->nRight; i++){
            lsmFsSortedDelete(pDb->pFS, pWorker, 1, &pLevel->aRhs[i]);
          }
          sortedFreeLevel(pDb->pEnv, pLevel);
        }else{
          int i;

          /* Free the separators of the next level, if required. */
          if( pLevel->pMerge->nInput > pLevel->nRight ){
            assert( pLevel->pNext->lhs.iRoot );
            pLevel->pNext->lhs.iRoot = 0;
          }

          /* Free the right-hand-side of pLevel */
          for(i=0; i<pLevel->nRight; i++){
            lsmFsSortedDelete(pDb->pFS, pWorker, 1, &pLevel->aRhs[i]);
          }
          lsmFree(pDb->pEnv, pLevel->aRhs);
          pLevel->nRight = 0;
          pLevel->aRhs = 0;

          /* Free the Merge object */
          lsmFree(pDb->pEnv, pLevel->pMerge);
          pLevel->pMerge = 0;






        }
      }

      /* Clean up the MergeWorker object initialized above. If no error
      ** has occurred, invoke the work-hook to inform the application that
      ** the database structure has changed. */
      mergeWorkerShutdown(&mergeworker, &rc);

      if( rc==LSM_OK ) sortedInvokeWorkHook(pDb);

#if 0
      lsmSortedDumpStructure(pDb, pDb->pWorker, 0, 0, "work");
#endif
      assertBtreeOk(pDb, &pLevel->lhs);
      assertRunInOrder(pDb, &pLevel->lhs);

      /* If bFlush is true and the database is no longer considered "full",
      ** break out of the loop even if nRemaining is still greater than
      ** zero. The caller has an in-memory tree to flush to disk.  */







>
>
>
>
>
>
>

|

>
>


>
>
>
>


>


|

>

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



>
|
<
<
|
|
|
<
<










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

|
|
|
|
|

|
<
<
<
<
|
|
<

|
|
|
|
|

|
<
<
<
|
|
|

|
|
|
>
>
>
>
>
>







>


|
|







4847
4848
4849
4850
4851
4852
4853
4854
4855
4856
4857
4858
4859
4860
4861
4862
4863
4864
4865
4866
4867
4868
4869
4870
4871
4872
4873
4874
4875
4876
4877
4878
4879
4880
4881
4882
4883
4884
4885
4886
4887
4888
4889
4890
4891
4892
4893
4894
4895
4896
4897
4898
4899
4900
4901
4902
4903
4904
4905
4906
4907
4908
4909
4910
4911
4912
4913
4914
4915
4916
4917
4918
4919
4920
4921


4922
4923
4924


4925
4926
4927
4928
4929
4930
4931
4932
4933
4934
4935
4936
4937
4938
4939
4940
4941
4942
4943
4944
4945
4946
4947
4948
4949
4950
4951
4952
4953
4954
4955
4956
4957
4958
4959
4960
4961
4962
4963

4964
4965
4966
4967
4968
4969
4970
4971
4972
4973
4974
4975




4976
4977

4978
4979
4980
4981
4982
4983
4984
4985



4986
4987
4988
4989
4990
4991
4992
4993
4994
4995
4996
4997
4998
4999
5000
5001
5002
5003
5004
5005
5006
5007
5008
5009
5010
5011
5012
5013
5014
5015
5016
5017
    Level *pLevel = 0;

    /* Find a level to work on. */
    rc = sortedSelectLevel(pDb, nMerge, &pLevel);
    assert( rc==LSM_OK || pLevel==0 );

    if( pLevel==0 ){
      int nDone = 0;
      Level *pTopLevel = lsmDbSnapshotLevel(pDb->pWorker);
      if( bFlush==0 && nMerge==1 && pTopLevel && pTopLevel->pNext==0 ){
        rc = sortedMoveBlock(pDb, &nDone);
      }
      nRemaining -= nDone;

      /* Could not find any work to do. Finished. */
      if( nDone==0 ) break;
    }else{
      int bSave = 0;
      Freelist freelist = {0, 0, 0};
      MergeWorker mergeworker;    /* State used to work on the level merge */

      assert( pDb->bIncrMerge==0 );
      assert( pDb->pFreelist==0 && pDb->bUseFreelist==0 );

      pDb->bIncrMerge = 1;
      rc = mergeWorkerInit(pDb, pLevel, &mergeworker);
      assert( mergeworker.nWork==0 );
      
      while( rc==LSM_OK 
          && 0==mergeWorkerDone(&mergeworker) 
          && (mergeworker.nWork<nRemaining || pDb->bUseFreelist)
      ){
        int eType = rtTopic(mergeworker.pCsr->eType);
        rc = mergeWorkerStep(&mergeworker);

        /* If the cursor now points at the first entry past the end of the
        ** user data (i.e. either to EOF or to the first free-list entry
        ** that will be added to the run), then check if it is possible to
        ** merge in any free-list entries that are either in-memory or in
        ** free-list-only blocks.  */
        if( rc==LSM_OK && nMerge==1 && eType==0
         && (rtTopic(mergeworker.pCsr->eType) || mergeWorkerDone(&mergeworker))
        ){
          int nFree = 0;          /* Number of free-list-only levels to merge */
          Level *pLvl;
          assert( pDb->pFreelist==0 && pDb->bUseFreelist==0 );

          /* Now check if all levels containing data newer than this one
          ** are single-segment free-list only levels. If so, they will be
          ** merged in now.  */
          for(pLvl=pDb->pWorker->pLevel; 
              pLvl!=mergeworker.pLevel && (pLvl->flags & LEVEL_FREELIST_ONLY); 
              pLvl=pLvl->pNext
          ){
            assert( pLvl->nRight==0 );
            nFree++;
          }
          if( pLvl==mergeworker.pLevel ){

            rc = mergeInsertFreelistSegments(pDb, nFree, &mergeworker);
            if( rc==LSM_OK ){
              rc = multiCursorVisitFreelist(mergeworker.pCsr);
            }
            if( rc==LSM_OK ){
              rc = multiCursorSetupTree(mergeworker.pCsr, 0);
              pDb->pFreelist = &freelist;
              pDb->bUseFreelist = 1;
            }
          }
        }
      }
      nRemaining -= LSM_MAX(mergeworker.nWork, 1);

      if( rc==LSM_OK ){
        /* Check if the merge operation is completely finished. If not,


        ** gobble up (declare eligible for recycling) any pages from rhs
        ** segments for which the content has been completely merged into 
        ** the lhs of the level.  */


        if( mergeWorkerDone(&mergeworker)==0 ){
          int i;
          for(i=0; i<pLevel->nRight; i++){
            SegmentPtr *pGobble = &mergeworker.pCsr->aPtr[i];
            if( pGobble->pSeg->iRoot ){
              rc = sortedBtreeGobble(pDb, mergeworker.pCsr, i);
            }else if( mergeworker.aGobble[i] ){
              lsmFsGobble(pDb, pGobble->pSeg, &mergeworker.aGobble[i], 1);
            }
          }
        }else{
          int i;
          int bEmpty;
          mergeWorkerShutdown(&mergeworker, &rc);
          bEmpty = (pLevel->lhs.iFirst==0);

          if( bEmpty==0 && rc==LSM_OK ){
            rc = lsmFsSortedFinish(pDb->pFS, &pLevel->lhs);
          }

          if( pDb->bUseFreelist ){
            Freelist *p = &pDb->pWorker->freelist;
            lsmFree(pDb->pEnv, p->aEntry);
            memcpy(p, &freelist, sizeof(freelist));
            pDb->bUseFreelist = 0;
            pDb->pFreelist = 0;
            bSave = 1;
          }

          for(i=0; i<pLevel->nRight; i++){
            lsmFsSortedDelete(pDb->pFS, pWorker, 1, &pLevel->aRhs[i]);
          }

          if( bEmpty ){
            /* If the new level is completely empty, remove it from the 
            ** database snapshot. This can only happen if all input keys were
            ** annihilated. Since keys are only annihilated if the new level
            ** is the last in the linked list (contains the most ancient of
            ** database content), this guarantees that pLevel->pNext==0.  */ 

            Level *pTop;          /* Top level of worker snapshot */
            Level **pp;           /* Read/write iterator for Level.pNext list */

            assert( pLevel->pNext==0 );

            /* Remove the level from the worker snapshot. */
            pTop = lsmDbSnapshotLevel(pWorker);
            for(pp=&pTop; *pp!=pLevel; pp=&((*pp)->pNext));
            *pp = pLevel->pNext;
            lsmDbSnapshotSetLevel(pWorker, pTop);

            /* Free the Level structure. */




            sortedFreeLevel(pDb->pEnv, pLevel);
          }else{


            /* Free the separators of the next level, if required. */
            if( pLevel->pMerge->nInput > pLevel->nRight ){
              assert( pLevel->pNext->lhs.iRoot );
              pLevel->pNext->lhs.iRoot = 0;
            }

            /* Zero the right-hand-side of pLevel */



            lsmFree(pDb->pEnv, pLevel->aRhs);
            pLevel->nRight = 0;
            pLevel->aRhs = 0;

            /* Free the Merge object */
            lsmFree(pDb->pEnv, pLevel->pMerge);
            pLevel->pMerge = 0;
          }

          if( bSave && rc==LSM_OK ){
            pDb->bIncrMerge = 0;
            rc = lsmSaveWorker(pDb, 0);
          }
        }
      }

      /* Clean up the MergeWorker object initialized above. If no error
      ** has occurred, invoke the work-hook to inform the application that
      ** the database structure has changed. */
      mergeWorkerShutdown(&mergeworker, &rc);
      pDb->bIncrMerge = 0;
      if( rc==LSM_OK ) sortedInvokeWorkHook(pDb);

#if LSM_LOG_STRUCTURE
      lsmSortedDumpStructure(pDb, pDb->pWorker, LSM_LOG_DATA, 0, "work");
#endif
      assertBtreeOk(pDb, &pLevel->lhs);
      assertRunInOrder(pDb, &pLevel->lhs);

      /* If bFlush is true and the database is no longer considered "full",
      ** break out of the loop even if nRemaining is still greater than
      ** zero. The caller has an in-memory tree to flush to disk.  */
4649
4650
4651
4652
4653
4654
4655








4656
4657
4658
4659
4660
4661
4662
4663
4664
4665
4666
4667
4668
4669
4670
4671
4672
4673

4674
4675
4676
4677
4678


4679
4680
4681
4682
4683
4684

4685
4686
4687
4688
4689
4690
4691
      bRet = 0;
    }
    *pRc = rc;
  }
  assert( *pRc==LSM_OK || bRet==0 );
  return bRet;
}









int lsmSaveWorker(lsm_db *pDb, int bFlush){
  Snapshot *p = pDb->pWorker;
  if( p->freelist.nEntry>pDb->nMaxFreelist ){
    int rc = sortedNewToplevel(pDb, TREE_NONE, 0);
    if( rc!=LSM_OK ) return rc;
  }
  return lsmCheckpointSaveWorker(pDb, bFlush);
}

static int doLsmSingleWork(
  lsm_db *pDb, 
  int bShutdown,
  int nMerge,                     /* Minimum segments to merge together */
  int nPage,                      /* Number of pages to write to disk */
  int *pnWrite,                   /* OUT: Pages actually written to disk */
  int *pbCkpt                     /* OUT: True if an auto-checkpoint is req. */
){

  int rc = LSM_OK;                /* Return code */
  int bDirty = 0;
  int nMax = nPage;               /* Maximum pages to write to disk */
  int nRem = nPage;
  int bCkpt = 0;



  /* Open the worker 'transaction'. It will be closed before this function
  ** returns.  */
  assert( pDb->pWorker==0 );
  rc = lsmBeginWork(pDb);
  if( rc!=LSM_OK ) return rc;


  /* If this connection is doing auto-checkpoints, set nMax (and nRem) so
  ** that this call stops writing when the auto-checkpoint is due. The
  ** caller will do the checkpoint, then possibly call this function again. */
  if( bShutdown==0 && pDb->nAutockpt ){
    u32 nSync;
    u32 nUnsync;







>
>
>
>
>
>
>
>




|













>





>
>






>







5051
5052
5053
5054
5055
5056
5057
5058
5059
5060
5061
5062
5063
5064
5065
5066
5067
5068
5069
5070
5071
5072
5073
5074
5075
5076
5077
5078
5079
5080
5081
5082
5083
5084
5085
5086
5087
5088
5089
5090
5091
5092
5093
5094
5095
5096
5097
5098
5099
5100
5101
5102
5103
5104
5105
      bRet = 0;
    }
    *pRc = rc;
  }
  assert( *pRc==LSM_OK || bRet==0 );
  return bRet;
}

/*
** Create a new free-list only top-level segment. Return LSM_OK if successful
** or an LSM error code if some error occurs.
*/
static int sortedNewFreelistOnly(lsm_db *pDb){
  return sortedNewToplevel(pDb, TREE_NONE, 0);
}

int lsmSaveWorker(lsm_db *pDb, int bFlush){
  Snapshot *p = pDb->pWorker;
  if( p->freelist.nEntry>pDb->nMaxFreelist ){
    int rc = sortedNewFreelistOnly(pDb);
    if( rc!=LSM_OK ) return rc;
  }
  return lsmCheckpointSaveWorker(pDb, bFlush);
}

static int doLsmSingleWork(
  lsm_db *pDb, 
  int bShutdown,
  int nMerge,                     /* Minimum segments to merge together */
  int nPage,                      /* Number of pages to write to disk */
  int *pnWrite,                   /* OUT: Pages actually written to disk */
  int *pbCkpt                     /* OUT: True if an auto-checkpoint is req. */
){
  Snapshot *pWorker;              /* Worker snapshot */
  int rc = LSM_OK;                /* Return code */
  int bDirty = 0;
  int nMax = nPage;               /* Maximum pages to write to disk */
  int nRem = nPage;
  int bCkpt = 0;

  assert( nPage>0 );

  /* Open the worker 'transaction'. It will be closed before this function
  ** returns.  */
  assert( pDb->pWorker==0 );
  rc = lsmBeginWork(pDb);
  if( rc!=LSM_OK ) return rc;
  pWorker = pDb->pWorker;

  /* If this connection is doing auto-checkpoints, set nMax (and nRem) so
  ** that this call stops writing when the auto-checkpoint is due. The
  ** caller will do the checkpoint, then possibly call this function again. */
  if( bShutdown==0 && pDb->nAutockpt ){
    u32 nSync;
    u32 nUnsync;
4747
4748
4749
4750
4751
4752
4753
4754
4755
4756
4757
4758












4759
4760
4761
4762
4763
4764
4765
4766
4767
4768
4769
4770
4771
4772
4773
4774
4775
4776
4777
4778
4779
4780
4781
4782
4783
4784



4785
4786
4787
4788
4789
4790
4791
4792
4793

4794
4795
4796
4797
4798
4799
4800
  if( rc==LSM_OK && pDb->pWorker->freelist.nEntry > pDb->nMaxFreelist ){
    int nPg = 0;
    while( rc==LSM_OK && lsmDatabaseFull(pDb) ){
      rc = sortedWork(pDb, 16, nMerge, 1, &nPg);
      nRem -= nPg;
    }
    if( rc==LSM_OK ){
      rc = sortedNewToplevel(pDb, TREE_NONE, &nPg);
    }
    nRem -= nPg;
    if( nPg ) bDirty = 1;
  }













  if( rc==LSM_OK && bDirty ){
    lsmFinishWork(pDb, 0, &rc);
  }else{
    int rcdummy = LSM_BUSY;
    lsmFinishWork(pDb, 0, &rcdummy);
  }
  assert( pDb->pWorker==0 );

  if( rc==LSM_OK ){
    *pnWrite = (nMax - nRem);
    *pbCkpt = (bCkpt && nRem<=0);
  }else{
    *pnWrite = 0;
    *pbCkpt = 0;
  }

  return rc;
}

static int doLsmWork(lsm_db *pDb, int nMerge, int nPage, int *pnWrite){
  int rc;
  int nWrite = 0;
  int bCkpt = 0;

  assert( nMerge>=1 );



  do {
    int nThis = 0;
    bCkpt = 0;
    rc = doLsmSingleWork(pDb, 0, nMerge, nPage-nWrite, &nThis, &bCkpt);
    nWrite += nThis;
    if( rc==LSM_OK && bCkpt ){
      rc = lsm_checkpoint(pDb, 0);
    }
  }while( rc==LSM_OK && (nWrite<nPage && bCkpt) );


  if( pnWrite ){
    if( rc==LSM_OK ){
      *pnWrite = nWrite;
    }else{
      *pnWrite = 0;
    }







|




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






<
<
<
<
<
<
<

<

|




|
|
<


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







5161
5162
5163
5164
5165
5166
5167
5168
5169
5170
5171
5172
5173
5174
5175
5176
5177
5178
5179
5180
5181
5182
5183
5184
5185
5186
5187
5188
5189
5190







5191

5192
5193
5194
5195
5196
5197
5198
5199

5200
5201
5202
5203
5204
5205
5206
5207
5208
5209
5210
5211
5212
5213
5214
5215
5216
5217
5218
5219
5220
5221
  if( rc==LSM_OK && pDb->pWorker->freelist.nEntry > pDb->nMaxFreelist ){
    int nPg = 0;
    while( rc==LSM_OK && lsmDatabaseFull(pDb) ){
      rc = sortedWork(pDb, 16, nMerge, 1, &nPg);
      nRem -= nPg;
    }
    if( rc==LSM_OK ){
      rc = sortedNewFreelistOnly(pDb);
    }
    nRem -= nPg;
    if( nPg ) bDirty = 1;
  }

  if( rc==LSM_OK ){
    *pnWrite = (nMax - nRem);
    *pbCkpt = (bCkpt && nRem<=0);
    if( nMerge==1 && pDb->nAutockpt>0 && *pnWrite>0
     && pWorker->pLevel 
     && pWorker->pLevel->nRight==0 
     && pWorker->pLevel->pNext==0 
    ){
      *pbCkpt = 1;
    }
  }

  if( rc==LSM_OK && bDirty ){
    lsmFinishWork(pDb, 0, &rc);
  }else{
    int rcdummy = LSM_BUSY;
    lsmFinishWork(pDb, 0, &rcdummy);







    *pnWrite = 0;

  }
  assert( pDb->pWorker==0 );
  return rc;
}

static int doLsmWork(lsm_db *pDb, int nMerge, int nPage, int *pnWrite){
  int rc = LSM_OK;                /* Return code */
  int nWrite = 0;                 /* Number of pages written */


  assert( nMerge>=1 );

  if( nPage>0 ){
    int bCkpt = 0;
    do {
      int nThis = 0;
      bCkpt = 0;
      rc = doLsmSingleWork(pDb, 0, nMerge, nPage-nWrite, &nThis, &bCkpt);
      nWrite += nThis;
      if( rc==LSM_OK && bCkpt ){
        rc = lsm_checkpoint(pDb, 0);
      }
    }while( rc==LSM_OK && (nWrite<nPage && bCkpt) );
  }

  if( pnWrite ){
    if( rc==LSM_OK ){
      *pnWrite = nWrite;
    }else{
      *pnWrite = 0;
    }
5018
5019
5020
5021
5022
5023
5024
5025
5026
5027
5028
5029
5030
5031
5032
5033
5034
5035
5036
5037
    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, iRef, &pRef);
      aKey = pageGetKey(pRef, 0, &iTopic, &nKey, &blob);
    }else{
      aCell += lsmVarintGet32(aCell, &nKey);
      if( rtIsWrite(eType) ) aCell += lsmVarintGet32(aCell, &nVal);
      sortedReadData(pPg, (aCell-aData), nKey+nVal, (void **)&aKey, &blob);
      aVal = &aKey[nKey];
      iTopic = eType;
    }

    lsmStringAppendf(&s, "%s%2X:", (i==0?"":" "), iTopic);
    for(iChar=0; iChar<nKey; iChar++){
      lsmStringAppendf(&s, "%c", isalnum(aKey[iChar]) ? aKey[iChar] : '.');







|
|



|







5439
5440
5441
5442
5443
5444
5445
5446
5447
5448
5449
5450
5451
5452
5453
5454
5455
5456
5457
5458
    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);
      aVal = &aKey[nKey];
      iTopic = eType;
    }

    lsmStringAppendf(&s, "%s%2X:", (i==0?"":" "), iTopic);
    for(iChar=0; iChar<nKey; iChar++){
      lsmStringAppendf(&s, "%c", isalnum(aKey[iChar]) ? aKey[iChar] : '.');
5051
5052
5053
5054
5055
5056
5057
5058

5059
5060
5061
5062
5063
5064
5065
  lsmLogMessage(pDb, LSM_OK, "      Page %d: %s", lsmFsPageNumber(pPg), s.z);
  lsmStringClear(&s);

  sortedBlobFree(&blob);
}

static void infoCellDump(
  lsm_db *pDb,

  int bIndirect,                  /* True to follow indirect refs */
  Page *pPg,
  int iCell,
  int *peType,
  int *piPgPtr,
  u8 **paKey, int *pnKey,
  u8 **paVal, int *pnVal,







|
>







5472
5473
5474
5475
5476
5477
5478
5479
5480
5481
5482
5483
5484
5485
5486
5487
  lsmLogMessage(pDb, LSM_OK, "      Page %d: %s", lsmFsPageNumber(pPg), s.z);
  lsmStringClear(&s);

  sortedBlobFree(&blob);
}

static void infoCellDump(
  lsm_db *pDb,                    /* Database handle */
  Segment *pSeg,                  /* Segment page belongs to */
  int bIndirect,                  /* True to follow indirect refs */
  Page *pPg,
  int iCell,
  int *peType,
  int *piPgPtr,
  u8 **paKey, int *pnKey,
  u8 **paVal, int *pnVal,
5080
5081
5082
5083
5084
5085
5086
5087
5088
5089
5090
5091
5092
5093
5094
5095
5096
5097
5098
5099
5100
5101
5102
5103
5104
5105
5106
  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, iRef, &pRef);
      pageGetKeyCopy(pDb->pEnv, pRef, 0, &dummy, pBlob);
      aKey = (u8 *)pBlob->pData;
      nKey = pBlob->nData;
      lsmFsPageRelease(pRef);
    }else{
      aKey = (u8 *)"<indirect>";
      nKey = 11;
    }
  }else{
    aCell += lsmVarintGet32(aCell, &nKey);
    if( rtIsWrite(eType) ) aCell += lsmVarintGet32(aCell, &nVal);
    sortedReadData(pPg, (aCell-aData), nKey+nVal, (void **)&aKey, pBlob);
    aVal = &aKey[nKey];
  }

  if( peType ) *peType = eType;
  if( piPgPtr ) *piPgPtr = iPgPtr;
  if( paKey ) *paKey = aKey;
  if( paVal ) *paVal = aVal;







|
|










|







5502
5503
5504
5505
5506
5507
5508
5509
5510
5511
5512
5513
5514
5515
5516
5517
5518
5519
5520
5521
5522
5523
5524
5525
5526
5527
5528
  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);
    }else{
      aKey = (u8 *)"<indirect>";
      nKey = 11;
    }
  }else{
    aCell += lsmVarintGet32(aCell, &nKey);
    if( rtIsWrite(eType) ) aCell += lsmVarintGet32(aCell, &nVal);
    sortedReadData(pSeg, pPg, (aCell-aData), nKey+nVal, (void **)&aKey, pBlob);
    aVal = &aKey[nKey];
  }

  if( peType ) *peType = eType;
  if( piPgPtr ) *piPgPtr = iPgPtr;
  if( paKey ) *paKey = aKey;
  if( paVal ) *paVal = aVal;
5131
5132
5133
5134
5135
5136
5137


5138
5139
5140
5141
5142
5143
5144
5145
5146


















5147


5148
5149
5150
5151
5152
5153
5154
  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 */



  int bValues = (flags & INFO_PAGE_DUMP_VALUES);
  int bHex = (flags & INFO_PAGE_DUMP_HEX);
  int bData = (flags & INFO_PAGE_DUMP_DATA);
  int bIndirect = (flags & INFO_PAGE_DUMP_INDIRECT);

  *pzOut = 0;
  if( iPg==0 ) return LSM_ERROR;



















  rc = lsmFsDbPageGet(pDb->pFS, iPg, &pPg);


  if( rc==LSM_OK ){
    Blob blob = {0, 0, 0, 0};
    int nKeyWidth = 0;
    LsmString str;
    int nRec;
    int iPtr;
    int flags;







>
>









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







5553
5554
5555
5556
5557
5558
5559
5560
5561
5562
5563
5564
5565
5566
5567
5568
5569
5570
5571
5572
5573
5574
5575
5576
5577
5578
5579
5580
5581
5582
5583
5584
5585
5586
5587
5588
5589
5590
5591
5592
5593
5594
5595
5596
5597
5598
  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 */
  Segment *pSeg = 0;
  Snapshot *pSnap;

  int bValues = (flags & INFO_PAGE_DUMP_VALUES);
  int bHex = (flags & INFO_PAGE_DUMP_HEX);
  int bData = (flags & INFO_PAGE_DUMP_DATA);
  int bIndirect = (flags & INFO_PAGE_DUMP_INDIRECT);

  *pzOut = 0;
  if( iPg==0 ) return LSM_ERROR;

  assert( pDb->pClient || pDb->pWorker );
  pSnap = pDb->pClient;
  if( pSnap==0 ) pSnap = pDb->pWorker;
  if( pSnap->redirect.n>0 ){
    Level *pLvl;
    int bUse = 0;
    for(pLvl=pSnap->pLevel; pLvl->pNext; pLvl=pLvl->pNext);
    pSeg = (pLvl->nRight==0 ? &pLvl->lhs : &pLvl->aRhs[pLvl->nRight-1]);
    rc = lsmFsSegmentContainsPg(pDb->pFS, pSeg, iPg, &bUse);
    if( bUse==0 ){
      pSeg = 0;
    }
  }

  /* iPg is a real page number (not subject to redirection). So it is safe 
  ** 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 flags;
5165
5166
5167
5168
5169
5170
5171

5172

5173
5174
5175
5176
5177
5178
5179
5180
5181
5182
5183
5184
5185
5186
5187
5188
5189
5190
5191
5192
5193
    lsmStringAppendf(&str, "nRec : %d\n", nRec);
    lsmStringAppendf(&str, "iPtr : %d\n", iPtr);
    lsmStringAppendf(&str, "flags: %04x\n", flags);
    lsmStringAppendf(&str, "\n");

    for(iCell=0; iCell<nRec; iCell++){
      int nKey;

      infoCellDump(pDb, bIndirect, pPg, iCell, 0, 0, 0, &nKey, 0, 0, &blob);

      if( nKey>nKeyWidth ) nKeyWidth = nKey;
    }
    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;
      char cType = '?';
      Pgno iAbsPtr;
      char zFlags[8];

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

      lsmFlagsToString(eType, zFlags);
      lsmStringAppendf(&str, "%s %d (%s) ", 
          zFlags, iAbsPtr, (rtTopic(eType) ? "sys" : "usr")







>
|
>









<



|







5609
5610
5611
5612
5613
5614
5615
5616
5617
5618
5619
5620
5621
5622
5623
5624
5625
5626
5627

5628
5629
5630
5631
5632
5633
5634
5635
5636
5637
5638
    lsmStringAppendf(&str, "nRec : %d\n", nRec);
    lsmStringAppendf(&str, "iPtr : %d\n", iPtr);
    lsmStringAppendf(&str, "flags: %04x\n", flags);
    lsmStringAppendf(&str, "\n");

    for(iCell=0; iCell<nRec; iCell++){
      int nKey;
      infoCellDump(
          pDb, pSeg, bIndirect, pPg, iCell, 0, 0, 0, &nKey, 0, 0, &blob
      );
      if( nKey>nKeyWidth ) nKeyWidth = nKey;
    }
    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 + ((flags & SEGMENT_BTREE_FLAG) ? 0 : iPtr);

      lsmFlagsToString(eType, zFlags);
      lsmStringAppendf(&str, "%s %d (%s) ", 
          zFlags, iAbsPtr, (rtTopic(eType) ? "sys" : "usr")
5262
5263
5264
5265
5266
5267
5268
5269
5270
5271
5272
5273
5274
5275
5276
    char *zSeg;
    Page *pPg;

    zSeg = segToString(pDb->pEnv, pRun, 0);
    lsmLogMessage(pDb, LSM_OK, "Segment: %s", zSeg);
    lsmFree(pDb->pEnv, zSeg);

    lsmFsDbPageGet(pDb->pFS, pRun->iFirst, &pPg);
    while( pPg ){
      Page *pNext;
      char *z = 0;
      infoPageDump(pDb, lsmFsPageNumber(pPg), flags, &z);
      lsmLogMessage(pDb, LSM_OK, "%s", z);
      lsmFree(pDb->pEnv, z);
#if 0







|







5707
5708
5709
5710
5711
5712
5713
5714
5715
5716
5717
5718
5719
5720
5721
    char *zSeg;
    Page *pPg;

    zSeg = segToString(pDb->pEnv, pRun, 0);
    lsmLogMessage(pDb, LSM_OK, "Segment: %s", zSeg);
    lsmFree(pDb->pEnv, zSeg);

    lsmFsDbPageGet(pDb->pFS, pRun, pRun->iFirst, &pPg);
    while( pPg ){
      Page *pNext;
      char *z = 0;
      infoPageDump(pDb, lsmFsPageNumber(pPg), flags, &z);
      lsmLogMessage(pDb, LSM_OK, "%s", z);
      lsmFree(pDb->pEnv, z);
#if 0
5292
5293
5294
5295
5296
5297
5298


5299
5300
5301
5302

5303
5304
5305

5306




5307
5308
5309
5310
5311
5312
5313
  Snapshot *pSnap,                /* Snapshot to dump */
  int bKeys,                      /* Output the keys from each segment */
  int bVals,                      /* Output the values from each segment */
  const char *zWhy                /* Caption to print near top of dump */
){
  Snapshot *pDump = pSnap;
  Level *pTopLevel;



  assert( pSnap );
  pTopLevel = lsmDbSnapshotLevel(pDump);
  if( pDb->xLog && pTopLevel ){

    Level *pLevel;
    int iLevel = 0;


    lsmLogMessage(pDb, LSM_OK, "Database structure (%s)", zWhy);





    for(pLevel=pTopLevel; pLevel; pLevel=pLevel->pNext){
      char zLeft[1024];
      char zRight[1024];
      int i = 0;

      Segment *aLeft[24];  







>
>




>



>
|
>
>
>
>







5737
5738
5739
5740
5741
5742
5743
5744
5745
5746
5747
5748
5749
5750
5751
5752
5753
5754
5755
5756
5757
5758
5759
5760
5761
5762
5763
5764
5765
5766
  Snapshot *pSnap,                /* Snapshot to dump */
  int bKeys,                      /* Output the keys from each segment */
  int bVals,                      /* Output the values from each segment */
  const char *zWhy                /* Caption to print near top of dump */
){
  Snapshot *pDump = pSnap;
  Level *pTopLevel;
  char *zFree = 0;


  assert( pSnap );
  pTopLevel = lsmDbSnapshotLevel(pDump);
  if( pDb->xLog && pTopLevel ){
    static int nCall = 0;
    Level *pLevel;
    int iLevel = 0;

    nCall++;
    lsmLogMessage(pDb, LSM_OK, "Database structure %d (%s)", nCall, zWhy);

#if 0
    if( nCall==1031 || nCall==1032 ) bKeys=1;
#endif

    for(pLevel=pTopLevel; pLevel; pLevel=pLevel->pNext){
      char zLeft[1024];
      char zRight[1024];
      int i = 0;

      Segment *aLeft[24];  
5370
5371
5372
5373
5374
5375
5376






5377
5378
5379
5380
5381
5382
5383
        sortedDumpSegment(pDb, &pLevel->lhs, bVals);
        for(i=0; i<pLevel->nRight; i++){
          sortedDumpSegment(pDb, &pLevel->aRhs[i], bVals);
        }
      }
    }
  }






}

void lsmSortedFreeLevel(lsm_env *pEnv, Level *pLevel){
  Level *pNext;
  Level *p;

  for(p=pLevel; p; p=pNext){







>
>
>
>
>
>







5823
5824
5825
5826
5827
5828
5829
5830
5831
5832
5833
5834
5835
5836
5837
5838
5839
5840
5841
5842
        sortedDumpSegment(pDb, &pLevel->lhs, bVals);
        for(i=0; i<pLevel->nRight; i++){
          sortedDumpSegment(pDb, &pLevel->aRhs[i], bVals);
        }
      }
    }
  }

  lsmInfoFreelist(pDb, &zFree);
  lsmLogMessage(pDb, LSM_OK, "Freelist: %s", zFree);
  lsmFree(pDb->pEnv, zFree);

  assert( lsmFsIntegrityCheck(pDb) );
}

void lsmSortedFreeLevel(lsm_env *pEnv, Level *pLevel){
  Level *pNext;
  Level *p;

  for(p=pLevel; p; p=pNext){
5408
5409
5410
5411
5412
5413
5414
5415
5416
5417
5418
5419
5420
5421
5422
5423
5424
5425
5426
5427
5428
5429
5430
5431
5432
5433
5434
5435
5436
5437
5438
5439
5440
5441
5442
5443

#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->iFirst, &pPg);
  while( pPg ){
    u8 *aData; int nData;
    Page *pNext;

    aData = lsmFsPageData(pPg, &nData);
    if( 0==(pageGetFlags(aData, nData) & SEGMENT_BTREE_FLAG) ){
      int i;
      int nRec = pageGetNRec(aData, nData);
      for(i=0; i<nRec; i++){
        int iTopic1, iTopic2;
        pageGetKeyCopy(pDb->pEnv, pPg, i, &iTopic1, &blob1);

        if( i==0 && blob2.nData ){
          assert( sortedKeyCompare(
                pDb->xCmp, iTopic2, blob2.pData, blob2.nData,
                iTopic1, blob1.pData, blob1.nData
          )<0 );
        }

        if( i<(nRec-1) ){
          pageGetKeyCopy(pDb->pEnv, pPg, i+1, &iTopic2, &blob2);
          assert( sortedKeyCompare(
                pDb->xCmp, iTopic1, blob1.pData, blob1.nData,
                iTopic2, blob2.pData, blob2.nData
          )<0 );
        }
      }
    }







|










|









|







5867
5868
5869
5870
5871
5872
5873
5874
5875
5876
5877
5878
5879
5880
5881
5882
5883
5884
5885
5886
5887
5888
5889
5890
5891
5892
5893
5894
5895
5896
5897
5898
5899
5900
5901
5902

#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);
    if( 0==(pageGetFlags(aData, nData) & SEGMENT_BTREE_FLAG) ){
      int i;
      int nRec = pageGetNRec(aData, nData);
      for(i=0; i<nRec; i++){
        int iTopic1, iTopic2;
        pageGetKeyCopy(pDb->pEnv, pSeg, pPg, i, &iTopic1, &blob1);

        if( i==0 && blob2.nData ){
          assert( sortedKeyCompare(
                pDb->xCmp, iTopic2, blob2.pData, blob2.nData,
                iTopic1, blob1.pData, blob1.nData
          )<0 );
        }

        if( i<(nRec-1) ){
          pageGetKeyCopy(pDb->pEnv, pSeg, pPg, i+1, &iTopic2, &blob2);
          assert( sortedKeyCompare(
                pDb->xCmp, iTopic1, blob1.pData, blob1.nData,
                iTopic2, blob2.pData, blob2.nData
          )<0 );
        }
      }
    }
5563
5564
5565
5566
5567
5568
5569
5570
5571
5572
5573
5574
5575
5576
5577
    BtreeCursor *pCsr = 0;        /* Btree cursor */

    rc = btreeCursorNew(pDb, pSeg, &pCsr);
    if( rc==LSM_OK ){
      rc = btreeCursorFirst(pCsr);
    }
    if( rc==LSM_OK ){
      rc = lsmFsDbPageGet(pFS, pSeg->iFirst, &pPg);
    }

    while( rc==LSM_OK ){
      Page *pNext;
      u8 *aData;
      int nData;
      int flags;







|







6022
6023
6024
6025
6026
6027
6028
6029
6030
6031
6032
6033
6034
6035
6036
    BtreeCursor *pCsr = 0;        /* Btree cursor */

    rc = btreeCursorNew(pDb, pSeg, &pCsr);
    if( rc==LSM_OK ){
      rc = btreeCursorFirst(pCsr);
    }
    if( rc==LSM_OK ){
      rc = lsmFsDbPageGet(pFS, pSeg, pSeg->iFirst, &pPg);
    }

    while( rc==LSM_OK ){
      Page *pNext;
      u8 *aData;
      int nData;
      int flags;
5585
5586
5587
5588
5589
5590
5591
5592
5593
5594
5595
5596
5597
5598
5599
      if( rc==LSM_OK 
       && 0==((SEGMENT_BTREE_FLAG|PGFTR_SKIP_THIS_FLAG) & flags)
       && 0!=pageGetNRec(aData, nData)
      ){
        u8 *pKey;
        int nKey;
        int iTopic;
        pKey = pageGetKey(pPg, 0, &iTopic, &nKey, &blob);
        assert( nKey==pCsr->nKey && 0==memcmp(pKey, pCsr->pKey, nKey) );
        assert( lsmFsPageNumber(pPg)==pCsr->iPtr );
        rc = btreeCursorNext(pCsr);
      }
    }
    assert( rc!=LSM_OK || pCsr->pKey==0 );








|







6044
6045
6046
6047
6048
6049
6050
6051
6052
6053
6054
6055
6056
6057
6058
      if( rc==LSM_OK 
       && 0==((SEGMENT_BTREE_FLAG|PGFTR_SKIP_THIS_FLAG) & flags)
       && 0!=pageGetNRec(aData, nData)
      ){
        u8 *pKey;
        int nKey;
        int iTopic;
        pKey = pageGetKey(pSeg, pPg, 0, &iTopic, &nKey, &blob);
        assert( nKey==pCsr->nKey && 0==memcmp(pKey, pCsr->pKey, nKey) );
        assert( lsmFsPageNumber(pPg)==pCsr->iPtr );
        rc = btreeCursorNext(pCsr);
      }
    }
    assert( rc!=LSM_OK || pCsr->pKey==0 );

Changes to tool/lsmview.tcl.

39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
  }
  namespace export scrollable
}
namespace import ::autoscroll::*
#############################################################################

proc exec_lsmtest_show {args} {
  set fd [open [list |lsmtest show {*}$args]]
  set res ""
  while {![eof $fd]} { 
    set line [gets $fd]
    if {[regexp {^\#.*} $line]} continue
    if {[regexp {^Leaked*} $line]} continue
    append res $line
    append res "\n"







|







39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
  }
  namespace export scrollable
}
namespace import ::autoscroll::*
#############################################################################

proc exec_lsmtest_show {args} {
  set fd [open [list |lsmtest show {*}$args 2>/dev/null]]
  set res ""
  while {![eof $fd]} { 
    set line [gets $fd]
    if {[regexp {^\#.*} $line]} continue
    if {[regexp {^Leaked*} $line]} continue
    append res $line
    append res "\n"