SQLite4
Changes On Branch mmap-on-demand
Not logged in

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

Changes In Branch mmap-on-demand Excluding Merge-Ins

This is equivalent to a diff from e423d6c4bc to a92d659eee

2013-03-04
10:24
When in non-mmap mode, cache some page-data in memory. As sqlite does. check-in: 1f0129ac39 user: dan tags: trunk
2013-03-02
16:01
Fix some problems on this branch. Leaf check-in: a92d659eee user: dan tags: mmap-on-demand
09:15
Merge trunk changes into this branch. check-in: 6666862302 user: dan tags: mmap-on-demand
09:09
Remove a couple of unused variables from sqltest.c. check-in: e423d6c4bc user: dan tags: trunk
2013-03-01
19:29
Fix a problem with querying non-primary key indexes for blob values. check-in: 74aa63bb48 user: dan tags: trunk

Changes to lsm-test/lsmtest_tdb3.c.

299
300
301
302
303
304
305

306
307
308
309
310
311
312
313
314
315























316
317
318
319
320
321
322

static int testEnvSectorSize(lsm_file *pFile){
  lsm_env *pRealEnv = tdb_lsm_env();
  LsmFile *p = (LsmFile *)pFile;
  return pRealEnv->xSectorSize(p->pReal);
}


static int testEnvRemap(
  lsm_file *pFile, 
  lsm_i64 iMin, 
  void **ppOut,
  lsm_i64 *pnOut
){
  lsm_env *pRealEnv = tdb_lsm_env();
  LsmFile *p = (LsmFile *)pFile;
  return pRealEnv->xRemap(p->pReal, iMin, ppOut, pnOut);
}
























static int testEnvFileid(
  lsm_file *pFile, 
  void *ppOut,
  int *pnOut
){
  lsm_env *pRealEnv = tdb_lsm_env();







>










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







299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346

static int testEnvSectorSize(lsm_file *pFile){
  lsm_env *pRealEnv = tdb_lsm_env();
  LsmFile *p = (LsmFile *)pFile;
  return pRealEnv->xSectorSize(p->pReal);
}

#if 0
static int testEnvRemap(
  lsm_file *pFile, 
  lsm_i64 iMin, 
  void **ppOut,
  lsm_i64 *pnOut
){
  lsm_env *pRealEnv = tdb_lsm_env();
  LsmFile *p = (LsmFile *)pFile;
  return pRealEnv->xRemap(p->pReal, iMin, ppOut, pnOut);
}
#endif

static int testEnvMap(
  lsm_file *pFile,
  lsm_i64 iOff,
  lsm_i64 nByte,
  void **ppOut,
  lsm_i64 *pszOut
){
  lsm_env *pRealEnv = tdb_lsm_env();
  LsmFile *p = (LsmFile *)pFile;
  return pRealEnv->xMap(p->pReal, iOff, nByte, ppOut, pszOut);
}

static int testEnvUnmap(
  lsm_file *pFile, 
  void *pMap, 
  lsm_i64 nMap
){
  lsm_env *pRealEnv = tdb_lsm_env();
  LsmFile *p = (LsmFile *)pFile;
  return pRealEnv->xUnmap(p->pReal, pMap, nMap);
}

static int testEnvFileid(
  lsm_file *pFile, 
  void *ppOut,
  int *pnOut
){
  lsm_env *pRealEnv = tdb_lsm_env();
940
941
942
943
944
945
946

947



948
949
950
951
952
953
954
  pDb->env.xFullpath = testEnvFullpath;
  pDb->env.xOpen = testEnvOpen;
  pDb->env.xRead = testEnvRead;
  pDb->env.xWrite = testEnvWrite;
  pDb->env.xTruncate = testEnvTruncate;
  pDb->env.xSync = testEnvSync;
  pDb->env.xSectorSize = testEnvSectorSize;

  pDb->env.xRemap = testEnvRemap;



  pDb->env.xFileid = testEnvFileid;
  pDb->env.xClose = testEnvClose;
  pDb->env.xUnlink = testEnvUnlink;
  pDb->env.xLock = testEnvLock;
  pDb->env.xTestLock = testEnvTestLock;
  pDb->env.xShmBarrier = testEnvShmBarrier;
  pDb->env.xShmMap = testEnvShmMap;







>

>
>
>







964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
  pDb->env.xFullpath = testEnvFullpath;
  pDb->env.xOpen = testEnvOpen;
  pDb->env.xRead = testEnvRead;
  pDb->env.xWrite = testEnvWrite;
  pDb->env.xTruncate = testEnvTruncate;
  pDb->env.xSync = testEnvSync;
  pDb->env.xSectorSize = testEnvSectorSize;
#if 0
  pDb->env.xRemap = testEnvRemap;
#endif
  pDb->env.xMap = testEnvMap;
  pDb->env.xUnmap = testEnvUnmap;
  pDb->env.xFileid = testEnvFileid;
  pDb->env.xClose = testEnvClose;
  pDb->env.xUnlink = testEnvUnlink;
  pDb->env.xLock = testEnvLock;
  pDb->env.xTestLock = testEnvTestLock;
  pDb->env.xShmBarrier = testEnvShmBarrier;
  pDb->env.xShmMap = testEnvShmMap;

Changes to src/kvlsm.c.

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

    memset(pNew, 0, sizeof(KVLsm));
    pNew->base.pStoreVfunc = &kvlsmMethods;
    pNew->base.pEnv = pEnv;
    rc = lsm_new(0, &pNew->pDb);
    if( rc==SQLITE4_OK ){
      int i;
      int bMmap = 0;
      lsm_config(pNew->pDb, LSM_CONFIG_MMAP, &bMmap);
      for(i=0; i<ArraySize(aConfig); i++){
        const char *zVal = sqlite4_uri_parameter(zName, aConfig[i].zParam);
        if( zVal ){
          int nVal = sqlite4Atoi(zVal);
          lsm_config(pNew->pDb, aConfig[i].eParam, &nVal);
        }







|







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

    memset(pNew, 0, sizeof(KVLsm));
    pNew->base.pStoreVfunc = &kvlsmMethods;
    pNew->base.pEnv = pEnv;
    rc = lsm_new(0, &pNew->pDb);
    if( rc==SQLITE4_OK ){
      int i;
      int bMmap = 1;
      lsm_config(pNew->pDb, LSM_CONFIG_MMAP, &bMmap);
      for(i=0; i<ArraySize(aConfig); i++){
        const char *zVal = sqlite4_uri_parameter(zName, aConfig[i].zParam);
        if( zVal ){
          int nVal = sqlite4Atoi(zVal);
          lsm_config(pNew->pDb, aConfig[i].eParam, &nVal);
        }

Changes to src/lsm.h.

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
/* Flags for lsm_env.xOpen() */
#define LSM_OPEN_READONLY 0x0001

/*
** CAPI: Database Runtime Environment
**
** Run-time environment used by LSM

















*/
struct lsm_env {
  int nByte;                 /* Size of this structure in bytes */
  int iVersion;              /* Version number of this structure (1) */
  /****** file i/o ***********************************************/
  void *pVfsCtx;
  int (*xFullpath)(lsm_env*, const char *, char *, int *);
  int (*xOpen)(lsm_env*, const char *, int flags, lsm_file **);
  int (*xRead)(lsm_file *, lsm_i64, void *, int);
  int (*xWrite)(lsm_file *, lsm_i64, void *, int);
  int (*xTruncate)(lsm_file *, lsm_i64);
  int (*xSync)(lsm_file *);
  int (*xSectorSize)(lsm_file *);

  int (*xRemap)(lsm_file *, lsm_i64, void **, lsm_i64*);



  int (*xFileid)(lsm_file *, void *pBuf, int *pnBuf);
  int (*xClose)(lsm_file *);
  int (*xUnlink)(lsm_env*, const char *);
  int (*xLock)(lsm_file*, int, int);
  int (*xTestLock)(lsm_file*, int, int, int);
  int (*xShmMap)(lsm_file*, int, int, void **);
  void (*xShmBarrier)(void);







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













>

>
>
>







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
/* Flags for lsm_env.xOpen() */
#define LSM_OPEN_READONLY 0x0001

/*
** CAPI: Database Runtime Environment
**
** Run-time environment used by LSM
**
** xMap(pFile, iOff, nByte, ppOut, pnOut):
**   Memory map nByte bytes of file pFile starting at file offset iOff. If
**   the file is less than (iOff+nByte) bytes in size extend it before 
**   mapping the new region. If successful, return LSM_OK and set *ppOut to
**   point to the mapped region. Set pnOut to the number of bytes mapped
**   (always nByte in this case). Otherwise, return an LSM error code.
**
**   If parameter iOff is zero, then nByte may be negative. In this case,
**   if the absolute value of nByte is greater than the size of the file
**   in bytes, then this is identical to a request with nByte set to a
**   positive value of the same magnitude. Or, if the file is more than
**   |nByte| bytes in size, the entire file should be mapped.
**
** xUnmap(pFile, nMap, pMap):
**   Argument pMap points to a mapping nMap bytes in size returned by a
**   previous invocation of xMap() on file pFile. Unmap it.
*/
struct lsm_env {
  int nByte;                 /* Size of this structure in bytes */
  int iVersion;              /* Version number of this structure (1) */
  /****** file i/o ***********************************************/
  void *pVfsCtx;
  int (*xFullpath)(lsm_env*, const char *, char *, int *);
  int (*xOpen)(lsm_env*, const char *, int flags, lsm_file **);
  int (*xRead)(lsm_file *, lsm_i64, void *, int);
  int (*xWrite)(lsm_file *, lsm_i64, void *, int);
  int (*xTruncate)(lsm_file *, lsm_i64);
  int (*xSync)(lsm_file *);
  int (*xSectorSize)(lsm_file *);
#if 0
  int (*xRemap)(lsm_file *, lsm_i64, void **, lsm_i64*);
#endif
  int (*xMap)(lsm_file *, lsm_i64 iOff, lsm_i64 nByte, void **, lsm_i64 *);
  int (*xUnmap)(lsm_file *, void *, lsm_i64);
  int (*xFileid)(lsm_file *, void *pBuf, int *pnBuf);
  int (*xClose)(lsm_file *);
  int (*xUnlink)(lsm_env*, const char *);
  int (*xLock)(lsm_file*, int, int);
  int (*xTestLock)(lsm_file*, int, int, int);
  int (*xShmMap)(lsm_file*, int, int, void **);
  void (*xShmBarrier)(void);
290
291
292
293
294
295
296




297
298
299
300
301
302
303
#define LSM_CONFIG_GET_COMPRESSION         14
#define LSM_CONFIG_SET_COMPRESSION_FACTORY 15
#define LSM_CONFIG_READONLY                16

#define LSM_SAFETY_OFF    0
#define LSM_SAFETY_NORMAL 1
#define LSM_SAFETY_FULL   2





/*
** CAPI: Compression and/or Encryption Hooks
*/
struct lsm_compress {
  void *pCtx;
  unsigned int iId;







>
>
>
>







311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
#define LSM_CONFIG_GET_COMPRESSION         14
#define LSM_CONFIG_SET_COMPRESSION_FACTORY 15
#define LSM_CONFIG_READONLY                16

#define LSM_SAFETY_OFF    0
#define LSM_SAFETY_NORMAL 1
#define LSM_SAFETY_FULL   2

#define LSM_MMAP_OFF     0
#define LSM_MMAP_FULL    1
#define LSM_MMAP_LIMITED 2

/*
** CAPI: Compression and/or Encryption Hooks
*/
struct lsm_compress {
  void *pCtx;
  unsigned int iId;

Changes to src/lsmInt.h.

46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
#define LSM_DFLT_BLOCK_SIZE         (1 * 1024 * 1024)
#define LSM_DFLT_AUTOFLUSH          (1 * 1024 * 1024)
#define LSM_DFLT_AUTOCHECKPOINT     (i64)(2 * 1024 * 1024)
#define LSM_DFLT_AUTOWORK           1
#define LSM_DFLT_LOG_SIZE           (128*1024)
#define LSM_DFLT_AUTOMERGE          4
#define LSM_DFLT_SAFETY             LSM_SAFETY_NORMAL
#define LSM_DFLT_MMAP               LSM_IS_64_BIT
#define LSM_DFLT_MULTIPLE_PROCESSES 1
#define LSM_DFLT_USE_LOG            1

/* Initial values for log file checksums. These are only used if the 
** database file does not contain a valid checkpoint.  */
#define LSM_CKSUM0_INIT 42
#define LSM_CKSUM1_INIT 42







|







46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
#define LSM_DFLT_BLOCK_SIZE         (1 * 1024 * 1024)
#define LSM_DFLT_AUTOFLUSH          (1 * 1024 * 1024)
#define LSM_DFLT_AUTOCHECKPOINT     (i64)(2 * 1024 * 1024)
#define LSM_DFLT_AUTOWORK           1
#define LSM_DFLT_LOG_SIZE           (128*1024)
#define LSM_DFLT_AUTOMERGE          4
#define LSM_DFLT_SAFETY             LSM_SAFETY_NORMAL
#define LSM_DFLT_MMAP               (LSM_IS_64_BIT ? LSM_MMAP_FULL : 0)
#define LSM_DFLT_MULTIPLE_PROCESSES 1
#define LSM_DFLT_USE_LOG            1

/* Initial values for log file checksums. These are only used if the 
** database file does not contain a valid checkpoint.  */
#define LSM_CKSUM0_INIT 42
#define LSM_CKSUM1_INIT 42
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
  int bAutowork;                  /* Configured by LSM_CONFIG_AUTOWORK */
  int nTreeLimit;                 /* Configured by LSM_CONFIG_AUTOFLUSH */
  int nMerge;                     /* Configured by LSM_CONFIG_AUTOMERGE */
  int bUseLog;                    /* Configured by LSM_CONFIG_USE_LOG */
  int nDfltPgsz;                  /* Configured by LSM_CONFIG_PAGE_SIZE */
  int nDfltBlksz;                 /* Configured by LSM_CONFIG_BLOCK_SIZE */
  int nMaxFreelist;               /* Configured by LSM_CONFIG_MAX_FREELIST */
  int bMmap;                      /* Configured by LSM_CONFIG_MMAP */
  i64 nAutockpt;                  /* Configured by LSM_CONFIG_AUTOCHECKPOINT */
  int bMultiProc;                 /* Configured by L_C_MULTIPLE_PROCESSES */
  int bReadonly;                  /* Configured by LSM_CONFIG_READONLY */
  lsm_compress compress;          /* Compression callbacks */
  lsm_compress_factory factory;   /* Compression callback factory */

  /* Sub-system handles */







|







330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
  int bAutowork;                  /* Configured by LSM_CONFIG_AUTOWORK */
  int nTreeLimit;                 /* Configured by LSM_CONFIG_AUTOFLUSH */
  int nMerge;                     /* Configured by LSM_CONFIG_AUTOMERGE */
  int bUseLog;                    /* Configured by LSM_CONFIG_USE_LOG */
  int nDfltPgsz;                  /* Configured by LSM_CONFIG_PAGE_SIZE */
  int nDfltBlksz;                 /* Configured by LSM_CONFIG_BLOCK_SIZE */
  int nMaxFreelist;               /* Configured by LSM_CONFIG_MAX_FREELIST */
  int eMmap;                      /* Configured by LSM_CONFIG_MMAP */
  i64 nAutockpt;                  /* Configured by LSM_CONFIG_AUTOCHECKPOINT */
  int bMultiProc;                 /* Configured by L_C_MULTIPLE_PROCESSES */
  int bReadonly;                  /* Configured by LSM_CONFIG_READONLY */
  lsm_compress compress;          /* Compression callbacks */
  lsm_compress_factory factory;   /* Compression callback factory */

  /* Sub-system handles */

Changes to src/lsm_file.c.

149
150
151
152
153
154
155
156




































157
158
159
160














161
162
163
164
165
166
167
168
169
170
171
172
173


174
175
176
177
178
179
180
*/
#include "lsmInt.h"

#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>

/*




































** File-system object. Each database connection allocates a single instance
** of the following structure. It is used for all access to the database and
** log files.
**














** pLruFirst, pLruLast:
**   The first and last entries in a doubly-linked list of pages. The 
**   Page.pLruNext and Page.pLruPrev pointers are used to link the list
**   elements together.
**
**   In mmap() mode, this list contains all currently allocated pages that
**   are carrying pointers into the database file mapping (pMap/nMap). If the
**   file has to be unmapped and then remapped (required to grow the mapping
**   as the file grows), the Page.aData pointers are updated by iterating
**   through the contents of this list.
**
**   In non-mmap() mode, this list is an LRU list of cached pages with 
**   nRef==0.


*/
struct FileSystem {
  lsm_db *pDb;                    /* Database handle that owns this object */
  lsm_env *pEnv;                  /* Environment pointer */
  char *zDb;                      /* Database file name */
  char *zLog;                     /* Database file name */
  int nMetasize;                  /* Size of meta pages in bytes */








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




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













>
>







149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
*/
#include "lsmInt.h"

#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>

/*
** An object of this type provides an abstraction supporting the two 
** different mmap() strategies. See the fsMmapXXX() methods below for
** details.
*/
typedef struct MmapMgr MmapMgr;
typedef struct MmapMgrRef MmapMgrRef;

struct MmapMgr {
  int eUseMmap;                   /* LSM_MMAP_XXX mode */
  int nRef;                       /* Number of outstanding fsMmapRef() calls */

  /* Used when eUseMmap==LSM_MMAP_FULL */
  void *pMap;                     /* Current mapping of database file */
  i64 nMap;                       /* Bytes mapped at pMap */

  /* Used when eUseMmap==LSM_MMAP_LIMITED */
  int nEntry;                     /* Total number of currently mapped blocks */
  int nMapsz;                     /* Size of each mapping (default 1MB) */
  int nMaphash;                   /* Size of apHash[] in entries */
  MmapMgrRef **apHash;            /* Hash table of existing mappings */
  MmapMgrRef *pAll;               /* Linked list of all mappings */
  MmapMgrRef *pLast;              /* Last element in pAll list */
};

struct MmapMgrRef {
  void *pMap;                     /* Pointer to mapped memory */
  int nRef;                       /* Number of refs to this mapping */
  int iMap;                       /* Mapping number */
  MmapMgrRef *pNextHash;          /* Next mapped block with same hash key */
  MmapMgrRef *pNextAll;           /* Next mapped block in pAll list */
  MmapMgrRef *pPrevAll;           /* Previous mapped block in pAll list */
};

#define LSM_MAPPED_BLOCKS 16

/*
** File-system object. Each database connection allocates a single instance
** of the following structure. It is used for all access to the database and
** log files.
**
** eUseMmap:
**   This variable determines whether or not, and how, the lsm_env.xMap() is 
**   used to access the database file. It may be set to 0, 1 or 2. As follows:
**
**     0 - xMap() is not used at all.
**
**     1 - xMap() is used to map the entire database file.
**
**     2 - xMap() is used map 1MB blocks of the database file as required.
**         The library attempts to limit the number of mapped blocks to
**         LSM_MAPPED_BLOCKS at a time. This mode is intended for use on
**         32-bit platforms where mapping the entire database file may
**         not be possible due to the limited address space available.
**
** pLruFirst, pLruLast:
**   The first and last entries in a doubly-linked list of pages. The 
**   Page.pLruNext and Page.pLruPrev pointers are used to link the list
**   elements together.
**
**   In mmap() mode, this list contains all currently allocated pages that
**   are carrying pointers into the database file mapping (pMap/nMap). If the
**   file has to be unmapped and then remapped (required to grow the mapping
**   as the file grows), the Page.aData pointers are updated by iterating
**   through the contents of this list.
**
**   In non-mmap() mode, this list is an LRU list of cached pages with 
**   nRef==0.
**
** apHash, nHash:
*/
struct FileSystem {
  lsm_db *pDb;                    /* Database handle that owns this object */
  lsm_env *pEnv;                  /* Environment pointer */
  char *zDb;                      /* Database file name */
  char *zLog;                     /* Database file name */
  int nMetasize;                  /* Size of meta pages in bytes */
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
  /* If this is a compressed database, a pointer to the compression methods.
  ** For an uncompressed database, a NULL pointer.  */
  lsm_compress *pCompress;
  u8 *aIBuffer;                   /* Buffer to compress to */
  u8 *aOBuffer;                   /* Buffer to uncompress from */
  int nBuffer;                    /* Allocated size of aBuffer[] in bytes */

  /* mmap() mode things */
  int bUseMmap;                   /* True to use mmap() to access db file */
  void *pMap;                     /* Current mapping of database file */
  i64 nMap;                       /* Bytes mapped at pMap */
  Page *pFree;

  Page *pWaiting;                 /* b-tree pages waiting to be written */

  /* Statistics */
  int nWrite;                     /* Total number of pages written */
  int nRead;                      /* Total number of pages read */







|
|
<
<







242
243
244
245
246
247
248
249
250


251
252
253
254
255
256
257
  /* If this is a compressed database, a pointer to the compression methods.
  ** For an uncompressed database, a NULL pointer.  */
  lsm_compress *pCompress;
  u8 *aIBuffer;                   /* Buffer to compress to */
  u8 *aOBuffer;                   /* Buffer to uncompress from */
  int nBuffer;                    /* Allocated size of aBuffer[] in bytes */

  /* State variables used in mmap() mode. */
  MmapMgr mmapmgr;


  Page *pFree;

  Page *pWaiting;                 /* b-tree pages waiting to be written */

  /* Statistics */
  int nWrite;                     /* Total number of pages written */
  int nRead;                      /* Total number of pages read */
228
229
230
231
232
233
234

235
236
237
238
239
240
241
**   The lsmFsSortedAppend() function sets the pSeg pointer to point to the
**   segment that the new page will be a part of. It is unset by
**   lsmFsPagePersist() after the page is written to disk.
*/
struct Page {
  u8 *aData;                      /* Buffer containing page data */
  int nData;                      /* Bytes of usable data at aData[] */

  Pgno iPg;                       /* Page number */
  int nRef;                       /* Number of outstanding references */
  int flags;                      /* Combination of PAGE_XXX flags */
  Page *pHashNext;                /* Next page in hash table slot */
  Page *pLruNext;                 /* Next page in LRU list */
  Page *pLruPrev;                 /* Previous page in LRU list */
  FileSystem *pFS;                /* File system that owns this page */







>







278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
**   The lsmFsSortedAppend() function sets the pSeg pointer to point to the
**   segment that the new page will be a part of. It is unset by
**   lsmFsPagePersist() after the page is written to disk.
*/
struct Page {
  u8 *aData;                      /* Buffer containing page data */
  int nData;                      /* Bytes of usable data at aData[] */
  MmapMgrRef *pRef;               /* Mapping manager reference token */
  Pgno iPg;                       /* Page number */
  int nRef;                       /* Number of outstanding references */
  int flags;                      /* Combination of PAGE_XXX flags */
  Page *pHashNext;                /* Next page in hash table slot */
  Page *pLruNext;                 /* Next page in LRU list */
  Page *pLruPrev;                 /* Previous page in LRU list */
  FileSystem *pFS;                /* File system that owns this page */
253
254
255
256
257
258
259

260
261
262
263
264
265
266
** Meta-data page handle. There are two meta-data pages at the start of
** the database file, each FileSystem.nMetasize bytes in size.
*/
struct MetaPage {
  int iPg;                        /* Either 1 or 2 */
  int bWrite;                     /* Write back to db file on release */
  u8 *aData;                      /* Pointer to buffer */

  FileSystem *pFS;                /* FileSystem that owns this page */
};

/* 
** Values for LsmPage.flags 
*/
#define PAGE_DIRTY   0x00000001   /* Set if page is dirty */







>







304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
** Meta-data page handle. There are two meta-data pages at the start of
** the database file, each FileSystem.nMetasize bytes in size.
*/
struct MetaPage {
  int iPg;                        /* Either 1 or 2 */
  int bWrite;                     /* Write back to db file on release */
  u8 *aData;                      /* Pointer to buffer */
  MmapMgrRef *pRef;               /* Mmap manager reference token */
  FileSystem *pFS;                /* FileSystem that owns this page */
};

/* 
** Values for LsmPage.flags 
*/
#define PAGE_DIRTY   0x00000001   /* Set if page is dirty */
335
336
337
338
339
340
341

342
343
344
345
346
347
348
349
350



















351
352
353
354
355
356
357
}
static int lsmEnvTruncate(lsm_env *pEnv, lsm_file *pFile, lsm_i64 nByte){
  return IOERR_WRAPPER( pEnv->xTruncate(pFile, nByte) );
}
static int lsmEnvUnlink(lsm_env *pEnv, const char *zDel){
  return IOERR_WRAPPER( pEnv->xUnlink(pEnv, zDel) );
}

static int lsmEnvRemap(
  lsm_env *pEnv, 
  lsm_file *pFile, 
  i64 szMin,
  void **ppMap,
  i64 *pszMap
){
  return pEnv->xRemap(pFile, szMin, ppMap, pszMap);
}




















int lsmEnvLock(lsm_env *pEnv, lsm_file *pFile, int iLock, int eLock){
  if( pFile==0 ) return LSM_OK;
  return pEnv->xLock(pFile, iLock, eLock);
}

int lsmEnvTestLock(







>









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







387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
}
static int lsmEnvTruncate(lsm_env *pEnv, lsm_file *pFile, lsm_i64 nByte){
  return IOERR_WRAPPER( pEnv->xTruncate(pFile, nByte) );
}
static int lsmEnvUnlink(lsm_env *pEnv, const char *zDel){
  return IOERR_WRAPPER( pEnv->xUnlink(pEnv, zDel) );
}
#if 0
static int lsmEnvRemap(
  lsm_env *pEnv, 
  lsm_file *pFile, 
  i64 szMin,
  void **ppMap,
  i64 *pszMap
){
  return pEnv->xRemap(pFile, szMin, ppMap, pszMap);
}
#endif
static void lsmEnvUnmap(
  lsm_env *pEnv, 
  lsm_file *pFile, 
  void *pMap, 
  lsm_i64 nMap
){
  pEnv->xUnmap(pFile, pMap, nMap);
}
static int lsmEnvMap(
  lsm_env *pEnv, 
  lsm_file *pFile, 
  i64 iOff,
  i64 szMin,
  void **ppMap,
  i64 *pszMap
){
  return pEnv->xMap(pFile, iOff, szMin, ppMap, pszMap);
}

int lsmEnvLock(lsm_env *pEnv, lsm_file *pFile, int iLock, int eLock){
  if( pFile==0 ) return LSM_OK;
  return pEnv->xLock(pFile, iLock, eLock);
}

int lsmEnvTestLock(
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
** Close the log file. Then delete it from the file-system. This function
** is called during database shutdown only.
*/
int lsmFsCloseAndDeleteLog(FileSystem *pFS){
  char *zDel;

  if( pFS->fdLog ){
    lsmEnvClose(pFS->pEnv, pFS->fdLog );
    pFS->fdLog = 0;
  }

  zDel = lsmMallocPrintf(pFS->pEnv, "%s-log", pFS->zDb);
  if( zDel ){
    lsmEnvUnlink(pFS->pEnv, zDel);
    lsmFree(pFS->pEnv, zDel);







|







511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
** Close the log file. Then delete it from the file-system. This function
** is called during database shutdown only.
*/
int lsmFsCloseAndDeleteLog(FileSystem *pFS){
  char *zDel;

  if( pFS->fdLog ){
    lsmEnvClose(pFS->pEnv, pFS->fdLog);
    pFS->fdLog = 0;
  }

  zDel = lsmMallocPrintf(pFS->pEnv, "%s-log", pFS->zDb);
  if( zDel ){
    lsmEnvUnlink(pFS->pEnv, zDel);
    lsmFree(pFS->pEnv, zDel);
479
480
481
482
483
484
485



















































































































































































































































486
487
488
489
490
491
492
    const char *zPath = (bLog ? pFS->zLog : pFS->zDb);

    *pRc = lsmEnvOpen(pFS->pEnv, zPath, flags, &pFile);
  }
  return pFile;
}




















































































































































































































































/*
** If it is not already open, this function opens the log file. It returns
** LSM_OK if successful (or if the log file was already open) or an LSM
** error code otherwise.
**
** The log file must be opened before any of the following may be called:
**







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







551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
    const char *zPath = (bLog ? pFS->zLog : pFS->zDb);

    *pRc = lsmEnvOpen(pFS->pEnv, zPath, flags, &pFile);
  }
  return pFile;
}


static void fsGrowMapping(
  FileSystem *pFS,
  i64 iSz,
  int *pRc
){
  MmapMgr *p = &pFS->mmapmgr;

  /* This function won't work with compressed databases yet. */
  assert( pFS->pCompress==0 );
  assert( PAGE_HASPREV==4 );
  assert( p->eUseMmap==LSM_MMAP_FULL );

  if( *pRc==LSM_OK && iSz>p->nMap ){
    int rc;
    u8 *aOld = p->pMap;
    if( aOld ){
      lsmEnvUnmap(pFS->pEnv, pFS->fdDb, aOld, p->nMap);
      p->pMap = 0;
    }
    rc = lsmEnvMap(pFS->pEnv, pFS->fdDb, 0, iSz, &p->pMap, &p->nMap);

    if( rc==LSM_OK && p->pMap!=aOld ){
      Page *pFix;
      i64 iOff = (u8 *)p->pMap - aOld;
      for(pFix=pFS->pLruFirst; pFix; pFix=pFix->pLruNext){
        pFix->aData += iOff;
      }
      lsmSortedRemap(pFS->pDb);
    }
    *pRc = rc;
  }
}

/* 
** Remove this block from the pAll list 
*/
static void fsMmapRemoveFromAll(MmapMgr *p, MmapMgrRef *pRef){
  if( pRef->pPrevAll ){
    pRef->pPrevAll->pNextAll = pRef->pNextAll;
  }else{
    assert( pRef==p->pAll );
    p->pAll = pRef->pNextAll;
  }

  if( pRef->pNextAll ){
    pRef->pNextAll->pPrevAll = pRef->pPrevAll;
  }else{
    assert( pRef==p->pLast );
    p->pLast = pRef->pPrevAll;
  }

  pRef->pNextAll = 0;
  pRef->pPrevAll = 0;
}

/* 
** Add this block to the pAll list. To the start if bLast==0, or the end
** if bLast==1.
*/
static void fsMmapAddToAll(MmapMgr *p, MmapMgrRef *pRef, int bLast){
  if( bLast ){
    assert( p->pLast==0 || p->pLast->pNextAll==0 );
    if( p->pLast ){
      p->pLast->pNextAll = pRef;
    }else{
      p->pAll = pRef;
    }
    pRef->pPrevAll = p->pLast;
    p->pLast = pRef;
  }else{
    assert( p->pAll==0 || p->pAll->pPrevAll==0 );
    if( p->pAll ){
      p->pAll->pPrevAll = pRef;
    }else{
      p->pLast = pRef;
    }
    pRef->pNextAll = p->pAll;
    p->pAll = pRef;
  }
}

/*
** Obtain a reference to a mapping of the database file.
*/
static void *fsMmapRef(
  FileSystem *pFS,                /* Memory map manager object */
  i64 iOff,                       /* File offset to return reference to */
  int nByte,                      /* Size of referenced block */
  MmapMgrRef **ppRef,             /* OUT: Reference used to free memory */
  int *pRc                        /* IN/OUT: Error code */
){
  lsm_env *pEnv = pFS->pEnv;
  MmapMgr *p = &pFS->mmapmgr;
  void *pRet = 0;
  int rc = *pRc;

  assert( iOff>=0 && nByte>0 );
  assert( p->eUseMmap!=LSM_MMAP_OFF );
  assert( *ppRef==0 );

  if( p->eUseMmap==LSM_MMAP_FULL ){
    fsGrowMapping(pFS, LSM_MAX(iOff+nByte, 8192), &rc);
    if( rc==LSM_OK ){
      pRet = (void *)&((u8 *)p->pMap)[iOff];
      *ppRef = (MmapMgrRef *)pRet;
    }
  }else{
    MmapMgrRef *pRef = 0;         /* Mapping reference */

    /* If the hash table has not been allocated, allocate it now. */
    if( p->apHash==0 ){
      int nSz = sizeof(MmapMgrRef *) * LSM_MAPPED_BLOCKS * 2;
      p->apHash = (MmapMgrRef **)lsmMallocZeroRc(pEnv, nSz, &rc);
      p->nMaphash = LSM_MAPPED_BLOCKS * 2;
    }

    if( rc==LSM_OK ){
      int iMap = (iOff / p->nMapsz);

      /* Search the hash table for the required mapping */
      int iHash = fsHashKey(p->nMaphash, iMap);
      for(pRef=p->apHash[iHash]; pRef; pRef=pRef->pNextHash){
        if( pRef->iMap==iMap ) break;
      }

      if( pRef ){
        /* Mapping was found. */
        fsMmapRemoveFromAll(p, pRef);
      }else{
        /* We have no existing mapping for mapping block iMap. */
        if( p->nEntry>=LSM_MAPPED_BLOCKS && p->pAll->nRef==0 ){
          MmapMgrRef **pp;
          int iHash2;
          pRef = p->pAll;

          /* Unmap this blocks mapping */ 
          lsmEnvUnmap(pFS->pEnv, pFS->fdDb, pRef->pMap, p->nMapsz); 
          pRef->pMap = 0;

          /* Remove this mapped block from the hash table */
          iHash2 = fsHashKey(p->nMaphash, pRef->iMap);
          for(pp=&p->apHash[iHash2]; *pp!=pRef; pp = &((*pp)->pNextHash));
          *pp = (*pp)->pNextHash;
          pRef->pNextHash = 0;

          /* Remove this mapped block from the pAll list */
          fsMmapRemoveFromAll(p, pRef);
#if 0
          printf("recycling mapping %d -> %d\n", pRef->iMap, iMap);
#endif

        }else{
          pRef = (MmapMgrRef *)lsmMallocZeroRc(pEnv, sizeof(MmapMgrRef), &rc);
          if( pRef ) p->nEntry++;
          assert( p->nEntry<20 );
        }

        if( rc==LSM_OK ){
          i64 dummy;
          rc = lsmEnvMap(pEnv, 
              pFS->fdDb, (i64)iMap*p->nMapsz, p->nMapsz, &pRef->pMap, &dummy
          );
          if( rc==LSM_OK ){
            /* Link the mapped block into the hash table */
            pRef->pNextHash = p->apHash[iHash];
            pRef->iMap = iMap;
            p->apHash[iHash] = pRef;
          }
        }
      }

      if( rc==LSM_OK ){
        pRef->nRef++;
        pRet = (void *)&((u8 *)(pRef->pMap))[iOff % p->nMapsz];
        fsMmapAddToAll(p, pRef, 1);
        *ppRef = pRef;
      }else{
        lsmFree(pEnv, pRef);
      } 
    }
  }

  *pRc = rc;
  if( rc==LSM_OK ) p->nRef++;
  return pRet;
}

/*
** Release a reference returned by an earlier call to fsMmapRef().
*/
static void fsMmapUnref(FileSystem *pFS, MmapMgrRef **ppRef){
  MmapMgr *p = &pFS->mmapmgr;
  MmapMgrRef *pRef = *ppRef;

  assert( p->eUseMmap!=LSM_MMAP_OFF );
  if( pRef ){
    p->nRef--;

    if( p->eUseMmap==LSM_MMAP_LIMITED ){
      pRef->nRef--;
      assert( pRef->nRef>=0 );
      if( pRef->nRef==0 ){
        fsMmapRemoveFromAll(p, pRef);
        fsMmapAddToAll(p, pRef, 0);
      }
    }
  }

  *ppRef = 0;
  assert( p->nRef>=0 );
}

/*
** Unmap all currently mapped blocks. Release all allocated memory.
*/
static void fsMmapClose(FileSystem *pFS){
  MmapMgr *p = &pFS->mmapmgr;
  assert( p->nRef==0 );
  if( p->eUseMmap==LSM_MMAP_FULL ){
    if( p->pMap ){
      lsmEnvUnmap(pFS->pEnv, pFS->fdDb, p->pMap, p->nMap); 
      p->pMap = 0;
      p->nMap = 0;
    }
  }
  else if( p->eUseMmap==LSM_MMAP_LIMITED ){
    MmapMgrRef *pRef;
    MmapMgrRef *pNext;
    for(pRef=p->pAll; pRef; pRef=pNext){
      pNext = pRef->pNextAll;
      lsmEnvUnmap(pFS->pEnv, pFS->fdDb, pRef->pMap, p->nMapsz);
      lsmFree(pFS->pEnv, pRef);
    }
    lsmFree(pFS->pEnv, p->apHash);
    p->nEntry = 0;
    p->nMaphash = 0;
    p->apHash = 0;
    p->pAll = 0;
    p->pLast = 0;
  }
}

/*
** If it is not already open, this function opens the log file. It returns
** LSM_OK if successful (or if the log file was already open) or an LSM
** error code otherwise.
**
** The log file must be opened before any of the following may be called:
**
550
551
552
553
554
555
556






557
558
559
560
561
562
563
    pFS->zDb = (char *)&pFS[1];
    pFS->zLog = &pFS->zDb[nDb+1];
    pFS->nPagesize = LSM_DFLT_PAGE_SIZE;
    pFS->nBlocksize = LSM_DFLT_BLOCK_SIZE;
    pFS->nMetasize = 4 * 1024;
    pFS->pDb = pDb;
    pFS->pEnv = pDb->pEnv;







    /* Make a copy of the database and log file names. */
    memcpy(pFS->zDb, zDb, nDb+1);
    memcpy(pFS->zLog, zDb, nDb);
    memcpy(&pFS->zLog[nDb], "-log", 5);

    /* Allocate the hash-table here. At some point, it should be changed







>
>
>
>
>
>







865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
    pFS->zDb = (char *)&pFS[1];
    pFS->zLog = &pFS->zDb[nDb+1];
    pFS->nPagesize = LSM_DFLT_PAGE_SIZE;
    pFS->nBlocksize = LSM_DFLT_BLOCK_SIZE;
    pFS->nMetasize = 4 * 1024;
    pFS->pDb = pDb;
    pFS->pEnv = pDb->pEnv;
    if( !pDb->compress.xCompress ){
      pFS->mmapmgr.eUseMmap = pDb->eMmap;
      pFS->mmapmgr.nMapsz = 1*1024*1024;

      /* pFS->mmapmgr.nMapsz = 4*1024; */
    }

    /* Make a copy of the database and log file names. */
    memcpy(pFS->zDb, zDb, nDb+1);
    memcpy(pFS->zLog, zDb, nDb);
    memcpy(&pFS->zLog[nDb], "-log", 5);

    /* Allocate the hash-table here. At some point, it should be changed
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646

    /* Reset any compression/decompression buffers already allocated */
    lsmFree(pEnv, pFS->aIBuffer);
    lsmFree(pEnv, pFS->aOBuffer);
    pFS->nBuffer = 0;

    /* Unmap the file, if it is currently mapped */
    if( pFS->pMap ){
      lsmEnvRemap(pEnv, pFS->fdDb, -1, &pFS->pMap, &pFS->nMap);
      pFS->bUseMmap = 0;
    }

    /* Free all allocate page structures */
    pPg = pFS->pLruFirst;
    while( pPg ){
      Page *pNext = pPg->pLruNext;
      if( pPg->flags & PAGE_FREE ) lsmFree(pEnv, pPg->aData);
      lsmFree(pEnv, pPg);
      pPg = pNext;
    }

    /* Zero pointers that point to deleted page objects */
    pFS->nCacheAlloc = 0;
    pFS->pLruFirst = 0;
    pFS->pLruLast = 0;
    pFS->pFree = 0;

    /* Configure the FileSystem object */
    if( db->compress.xCompress ){
      pFS->pCompress = &db->compress;
      pFS->bUseMmap = 0;
    }else{
      pFS->pCompress = 0;
      pFS->bUseMmap = db->bMmap;
    }
  }

  return LSM_OK;
}

/*







<
<
|
<



















|


|







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

    /* Reset any compression/decompression buffers already allocated */
    lsmFree(pEnv, pFS->aIBuffer);
    lsmFree(pEnv, pFS->aOBuffer);
    pFS->nBuffer = 0;

    /* Unmap the file, if it is currently mapped */


    fsMmapClose(pFS);


    /* Free all allocate page structures */
    pPg = pFS->pLruFirst;
    while( pPg ){
      Page *pNext = pPg->pLruNext;
      if( pPg->flags & PAGE_FREE ) lsmFree(pEnv, pPg->aData);
      lsmFree(pEnv, pPg);
      pPg = pNext;
    }

    /* Zero pointers that point to deleted page objects */
    pFS->nCacheAlloc = 0;
    pFS->pLruFirst = 0;
    pFS->pLruLast = 0;
    pFS->pFree = 0;

    /* Configure the FileSystem object */
    if( db->compress.xCompress ){
      pFS->pCompress = &db->compress;
      pFS->mmapmgr.eUseMmap = LSM_MMAP_OFF;
    }else{
      pFS->pCompress = 0;
      pFS->mmapmgr.eUseMmap = db->eMmap;
    }
  }

  return LSM_OK;
}

/*
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
    while( pPg ){
      Page *pNext = pPg->pLruNext;
      if( pPg->flags & PAGE_FREE ) lsmFree(pEnv, pPg->aData);
      lsmFree(pEnv, pPg);
      pPg = pNext;
    }


    if( pFS->fdDb ) lsmEnvClose(pFS->pEnv, pFS->fdDb );
    if( pFS->fdLog ) lsmEnvClose(pFS->pEnv, pFS->fdLog );
    lsmFree(pEnv, pFS->pLsmFile);
    lsmFree(pEnv, pFS->apHash);
    lsmFree(pEnv, pFS->aIBuffer);
    lsmFree(pEnv, pFS->aOBuffer);
    lsmFree(pEnv, pFS);
  }
}

void lsmFsDeferClose(FileSystem *pFS, LsmFile **pp){
  LsmFile *p = pFS->pLsmFile;
  assert( p->pNext==0 );

  p->pFile = pFS->fdDb;
  pFS->fdDb = 0;
  pFS->pLsmFile = 0;
  *pp = p;
}

/*







>













>







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
    while( pPg ){
      Page *pNext = pPg->pLruNext;
      if( pPg->flags & PAGE_FREE ) lsmFree(pEnv, pPg->aData);
      lsmFree(pEnv, pPg);
      pPg = pNext;
    }

    fsMmapClose(pFS);
    if( pFS->fdDb ) lsmEnvClose(pFS->pEnv, pFS->fdDb );
    if( pFS->fdLog ) lsmEnvClose(pFS->pEnv, pFS->fdLog );
    lsmFree(pEnv, pFS->pLsmFile);
    lsmFree(pEnv, pFS->apHash);
    lsmFree(pEnv, pFS->aIBuffer);
    lsmFree(pEnv, pFS->aOBuffer);
    lsmFree(pEnv, pFS);
  }
}

void lsmFsDeferClose(FileSystem *pFS, LsmFile **pp){
  LsmFile *p = pFS->pLsmFile;
  assert( p->pNext==0 );
  fsMmapClose(pFS);
  p->pFile = pFS->fdDb;
  pFS->fdDb = 0;
  pFS->pLsmFile = 0;
  *pp = p;
}

/*
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
  Page **pp;

  iHash = fsHashKey(pFS->nHash, pPg->iPg);
  for(pp=&pFS->apHash[iHash]; *pp!=pPg; pp=&(*pp)->pHashNext);
  *pp = pPg->pHashNext;
}





static int fsPageBuffer(
  FileSystem *pFS, 
  int bRequireData,               /* True to allocate buffer as well */
  Page **ppOut
){
  int rc = LSM_OK;
  Page *pPage = 0;



  if( pFS->bUseMmap || pFS->pLruFirst==0 || pFS->nCacheAlloc<pFS->nCacheMax ){

    pPage = lsmMallocZero(pFS->pEnv, sizeof(Page));
    if( !pPage ){
      rc = LSM_NOMEM_BKPT;
    }else if( bRequireData ){
      pPage->aData = (u8 *)lsmMalloc(pFS->pEnv, pFS->nPagesize);
      pPage->flags = PAGE_FREE;
      if( !pPage->aData ){
        lsmFree(pFS->pEnv, pPage);
        rc = LSM_NOMEM_BKPT;
        pPage = 0;
      }
      pFS->nCacheAlloc++;
    }else{
      fsPageAddToLru(pFS, pPage);
    }
  }else{
    pPage = pFS->pLruFirst;
    fsPageRemoveFromLru(pFS, pPage);
    fsPageRemoveFromHash(pFS, pPage);
  }

  assert( pPage==0 || (pPage->flags & PAGE_DIRTY)==0 );
  *ppOut = pPage;
  return rc;
}

static void fsPageBufferFree(Page *pPg){



  if( pPg->flags & PAGE_FREE ){
    lsmFree(pPg->pFS->pEnv, pPg->aData);
  }
  else if( pPg->pFS->bUseMmap ){
    fsPageRemoveFromLru(pPg->pFS, pPg);
  }
  lsmFree(pPg->pFS->pEnv, pPg);
}

static void fsGrowMapping(
  FileSystem *pFS,
  i64 iSz,
  int *pRc
){
  /* This function won't work with compressed databases yet. */
  assert( pFS->pCompress==0 );
  assert( PAGE_HASPREV==4 );

  if( *pRc==LSM_OK && iSz>pFS->nMap ){
    int rc;
    u8 *aOld = pFS->pMap;
    rc = lsmEnvRemap(pFS->pEnv, pFS->fdDb, iSz, &pFS->pMap, &pFS->nMap);
    if( rc==LSM_OK && pFS->pMap!=aOld ){
      Page *pFix;
      i64 iOff = (u8 *)pFS->pMap - aOld;
      for(pFix=pFS->pLruFirst; pFix; pFix=pFix->pLruNext){
        pFix->aData += iOff;
      }
      lsmSortedRemap(pFS->pDb);
    }
    *pRc = rc;
  }
}


/*
** fsync() the database file.
*/
int lsmFsSyncDb(FileSystem *pFS, int nBlock){

  if( nBlock && pFS->bUseMmap ){
    int rc = LSM_OK;
    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 ){







>
>
>
>


<




>
>
>
|
>



|








<
<













>
>
>



<
<
<



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





>






>







1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215

1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236


1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255



1256
1257
1258

























1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
  Page **pp;

  iHash = fsHashKey(pFS->nHash, pPg->iPg);
  for(pp=&pFS->apHash[iHash]; *pp!=pPg; pp=&(*pp)->pHashNext);
  *pp = pPg->pHashNext;
}

/*
** This function is only called if FileSystem.eUseMmap==0 (i.e. in non-mmap
** mode).
*/
static int fsPageBuffer(
  FileSystem *pFS, 

  Page **ppOut
){
  int rc = LSM_OK;
  Page *pPage = 0;

  if( pFS->mmapmgr.eUseMmap!=LSM_MMAP_OFF
   || pFS->pLruFirst==0 
   || pFS->nCacheAlloc<pFS->nCacheMax 
  ){
    pPage = lsmMallocZero(pFS->pEnv, sizeof(Page));
    if( !pPage ){
      rc = LSM_NOMEM_BKPT;
    }else{
      pPage->aData = (u8 *)lsmMalloc(pFS->pEnv, pFS->nPagesize);
      pPage->flags = PAGE_FREE;
      if( !pPage->aData ){
        lsmFree(pFS->pEnv, pPage);
        rc = LSM_NOMEM_BKPT;
        pPage = 0;
      }
      pFS->nCacheAlloc++;


    }
  }else{
    pPage = pFS->pLruFirst;
    fsPageRemoveFromLru(pFS, pPage);
    fsPageRemoveFromHash(pFS, pPage);
  }

  assert( pPage==0 || (pPage->flags & PAGE_DIRTY)==0 );
  *ppOut = pPage;
  return rc;
}

static void fsPageBufferFree(Page *pPg){
  FileSystem *pFS = pPg->pFS;
  assert( pFS->mmapmgr.eUseMmap==LSM_MMAP_OFF );

  if( pPg->flags & PAGE_FREE ){
    lsmFree(pPg->pFS->pEnv, pPg->aData);
  }



  lsmFree(pPg->pFS->pEnv, pPg);
}



























/*
** fsync() the database file.
*/
int lsmFsSyncDb(FileSystem *pFS, int nBlock){
#if 0
  if( nBlock && pFS->bUseMmap ){
    int rc = LSM_OK;
    i64 nMin = (i64)nBlock * (i64)pFS->nBlocksize;
    fsGrowMapping(pFS, nMin, &rc);
    if( rc!=LSM_OK ) return rc;
  }
#endif
  return lsmEnvSync(pFS->pEnv, pFS->fdDb);
}

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

static int fsRedirectBlock(Redirect *p, int iBlk){
  if( p ){
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
  
  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 ){







|







1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
  
  if( pSeg ){
    iRead = fsRedirectBlock(pSeg->pRedirect, iBlock);
  }else{
    iRead = iBlock;
  }

  assert( pFS->mmapmgr.eUseMmap==LSM_MMAP_OFF || 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 ){
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
  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 ){







|







1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
  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->mmapmgr.eUseMmap==LSM_MMAP_OFF || 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 ){
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
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334








1335
1336
1337
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
  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;








    assert( (p->flags & PAGE_FREE)==0 );
  }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,







>





|
|

<
<
<

>
>










>
>










<

>
>
>
>
>
>
>
>










|
















<







1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
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
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
  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);
  i64 iOff = (i64)(iReal-1) * pFS->nPagesize;

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

  assert( pFS->mmapmgr.eUseMmap==LSM_MMAP_OFF || pFS->pCompress==0 );
  if( pFS->mmapmgr.eUseMmap!=LSM_MMAP_OFF ){
    Page *pTest;




    /* Check if the page is currently in the waiting list. If so, increment
    ** the refcount and return a pointer to it. No more to do in this case. */
    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;
      }
    }

    /* Allocate or recycle a Page structure */
    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->iPg = iReal;

    p->aData = fsMmapRef(pFS, iOff, pFS->nPagesize, &p->pRef, &rc);
    if( rc!=LSM_OK ){
      p->pHashNext = pFS->pFree;
      pFS->pFree = p;
      p = 0;
    }

    assert( (p->flags & PAGE_FREE)==0 );
  }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, &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;

            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,
1414
1415
1416
1417
1418
1419
1420




1421
1422
1423
1424

1425
1426
1427
1428
1429
1430
1431
** may be garbage. It is the callers responsibility to deal with this.
*/
int lsmFsReadSyncedId(lsm_db *db, int iMeta, i64 *piVal){
  FileSystem *pFS = db->pFS;
  int rc = LSM_OK;

  assert( iMeta==1 || iMeta==2 );




  if( pFS->bUseMmap ){
    fsGrowMapping(pFS, iMeta*LSM_META_PAGE_SIZE, &rc);
    if( rc==LSM_OK ){
      *piVal = (i64)lsmGetU64(&((u8 *)pFS->pMap)[(iMeta-1)*LSM_META_PAGE_SIZE]);

    }
  }else{
    MetaPage *pMeta = 0;
    rc = lsmFsMetaPageGet(pFS, 0, iMeta, &pMeta);
    if( rc==LSM_OK ){
      *piVal = (i64)lsmGetU64(pMeta->aData);
      lsmFsMetaPageRelease(pMeta);







>
>
>
>
|
|

|
>







1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
** may be garbage. It is the callers responsibility to deal with this.
*/
int lsmFsReadSyncedId(lsm_db *db, int iMeta, i64 *piVal){
  FileSystem *pFS = db->pFS;
  int rc = LSM_OK;

  assert( iMeta==1 || iMeta==2 );
  if( pFS->mmapmgr.eUseMmap!=LSM_MMAP_OFF ){
    MmapMgrRef *pRef = 0;
    u8 *pMap;                     /* Mapping of meta pages */
    int iOff = (iMeta==2 ? LSM_META_PAGE_SIZE : 0);

    pMap = (u8 *)fsMmapRef(pFS, iOff, LSM_META_PAGE_SIZE, &pRef, &rc);
    if( rc==LSM_OK ){
      *piVal = (i64)lsmGetU64(pMap);
      fsMmapUnref(pFS, &pRef);
    }
  }else{
    MetaPage *pMeta = 0;
    rc = lsmFsMetaPageGet(pFS, 0, iMeta, &pMeta);
    if( rc==LSM_OK ){
      *piVal = (i64)lsmGetU64(pMeta->aData);
      lsmFsMetaPageRelease(pMeta);
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833

  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 ){
      pPg->pFS = pFS;
      pPg->pSeg = p;
      pPg->iPg = 0;
      pPg->flags |= PAGE_DIRTY;
      pPg->nData = pFS->nPagesize;
      assert( pPg->aData );







|







2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148

  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, &pPg);
    if( rc==LSM_OK ){
      pPg->pFS = pFS;
      pPg->pSeg = p;
      pPg->iPg = 0;
      pPg->flags |= PAGE_DIRTY;
      pPg->nData = pFS->nPagesize;
      assert( pPg->aData );
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
  MetaPage *pPg;
  assert( iPg==1 || iPg==2 );

  pPg = lsmMallocZeroRc(pFS->pEnv, sizeof(Page), &rc);

  if( pPg ){
    i64 iOff = (iPg-1) * pFS->nMetasize;
    if( pFS->bUseMmap ){
      fsGrowMapping(pFS, 2*pFS->nMetasize, &rc);
      pPg->aData = (u8 *)(pFS->pMap) + iOff;
    }else{
      pPg->aData = lsmMallocRc(pFS->pEnv, pFS->nMetasize, &rc);
      if( rc==LSM_OK && bWrite==0 ){
        rc = lsmEnvRead(pFS->pEnv, pFS->fdDb, iOff, pPg->aData, pFS->nMetasize);
      }
#ifndef NDEBUG
      /* pPg->aData causes an uninitialized access via a downstreadm write().
         After discussion on this list, this memory should not, for performance
         reasons, be memset. However, tracking down "real" misuse is more
         difficult with this "false" positive, so it is set when NDEBUG.
      */
      else if( rc==LSM_OK ){
        memset( pPg->aData, 0x77, pFS->nMetasize );
      }
#endif
    }

    if( rc!=LSM_OK ){

      if( pFS->bUseMmap==0 ) lsmFree(pFS->pEnv, pPg->aData);



      lsmFree(pFS->pEnv, pPg);
      pPg = 0;
    }else{
      pPg->iPg = iPg;
      pPg->bWrite = bWrite;
      pPg->pFS = pFS;
    }







|
<
|


















>
|
>
>
>







2295
2296
2297
2298
2299
2300
2301
2302

2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
  MetaPage *pPg;
  assert( iPg==1 || iPg==2 );

  pPg = lsmMallocZeroRc(pFS->pEnv, sizeof(Page), &rc);

  if( pPg ){
    i64 iOff = (iPg-1) * pFS->nMetasize;
    if( pFS->mmapmgr.eUseMmap!=LSM_MMAP_OFF ){

      pPg->aData = fsMmapRef(pFS, iOff, LSM_META_PAGE_SIZE, &pPg->pRef, &rc);
    }else{
      pPg->aData = lsmMallocRc(pFS->pEnv, pFS->nMetasize, &rc);
      if( rc==LSM_OK && bWrite==0 ){
        rc = lsmEnvRead(pFS->pEnv, pFS->fdDb, iOff, pPg->aData, pFS->nMetasize);
      }
#ifndef NDEBUG
      /* pPg->aData causes an uninitialized access via a downstreadm write().
         After discussion on this list, this memory should not, for performance
         reasons, be memset. However, tracking down "real" misuse is more
         difficult with this "false" positive, so it is set when NDEBUG.
      */
      else if( rc==LSM_OK ){
        memset( pPg->aData, 0x77, pFS->nMetasize );
      }
#endif
    }

    if( rc!=LSM_OK ){
      if( pFS->mmapmgr.eUseMmap==LSM_MMAP_OFF ){
        lsmFree(pFS->pEnv, pPg->aData);
      }else{
        fsMmapUnref(pFS, &pPg->pRef);
      }
      lsmFree(pFS->pEnv, pPg);
      pPg = 0;
    }else{
      pPg->iPg = iPg;
      pPg->bWrite = bWrite;
      pPg->pFS = pFS;
    }
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036


2037
2038
2039
2040
2041
2042
2043
** Release a meta-page reference obtained via a call to lsmFsMetaPageGet().
*/
int lsmFsMetaPageRelease(MetaPage *pPg){
  int rc = LSM_OK;
  if( pPg ){
    FileSystem *pFS = pPg->pFS;

    if( pFS->bUseMmap==0 ){
      if( pPg->bWrite ){
        i64 iOff = (pPg->iPg==2 ? pFS->nMetasize : 0);
        int nWrite = pFS->nMetasize;
        rc = lsmEnvWrite(pFS->pEnv, pFS->fdDb, iOff, pPg->aData, nWrite);
      }
      lsmFree(pFS->pEnv, pPg->aData);


    }

    lsmFree(pFS->pEnv, pPg);
  }
  return rc;
}








|






>
>







2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
** Release a meta-page reference obtained via a call to lsmFsMetaPageGet().
*/
int lsmFsMetaPageRelease(MetaPage *pPg){
  int rc = LSM_OK;
  if( pPg ){
    FileSystem *pFS = pPg->pFS;

    if( pFS->mmapmgr.eUseMmap==LSM_MMAP_OFF ){
      if( pPg->bWrite ){
        i64 iOff = (pPg->iPg==2 ? pFS->nMetasize : 0);
        int nWrite = pFS->nMetasize;
        rc = lsmEnvWrite(pFS->pEnv, pFS->fdDb, iOff, pPg->aData, nWrite);
      }
      lsmFree(pFS->pEnv, pPg->aData);
    }else{
      fsMmapUnref(pFS, &pPg->pRef);
    }

    lsmFree(pFS->pEnv, pPg);
  }
  return rc;
}

2087
2088
2089
2090
2091
2092
2093
2094



2095


2096
2097
2098
2099



2100
2101
2102
2103
2104
2105
2106

  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++){







|
>
>
>
|
>
>

<
|

>
>
>







2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421

2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433

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

  if( pFS->mmapmgr.eUseMmap!=LSM_MMAP_OFF ){
    MmapMgrRef *pRef1 = 0;
    MmapMgrRef *pRef2 = 0;
    void *pTo, *pFrom;

    pTo = fsMmapRef(pFS, iToOff, pFS->nBlocksize, &pRef1, &rc);
    pFrom = fsMmapRef(pFS, iFromOff, pFS->nBlocksize, &pRef2, &rc);
    if( rc==LSM_OK ){

      memcpy(pTo, pFrom, pFS->nBlocksize);
    }
    fsMmapUnref(pFS, &pRef1);
    fsMmapUnref(pFS, &pRef2);

  }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++){
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
        assert( pPg->flags & PAGE_FREE );
        assert( (pPg->flags & PAGE_HASPREV)==0 );
        assert( pPg->nData==pFS->nPagesize-4 );

        rc = fsAppendPage(pFS, pPg->pSeg, &pPg->iPg, &iPrev, &iNext);
        if( rc!=LSM_OK ) return rc;

        if( pFS->bUseMmap==0 ){
          int iHash = fsHashKey(pFS->nHash, pPg->iPg);
          pPg->pHashNext = pFS->apHash[iHash];
          pFS->apHash[iHash] = pPg;
          assert( pPg->pHashNext==0 || pPg->pHashNext->iPg!=pPg->iPg );
        }

        if( iPrev ){







|







2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
        assert( pPg->flags & PAGE_FREE );
        assert( (pPg->flags & PAGE_HASPREV)==0 );
        assert( pPg->nData==pFS->nPagesize-4 );

        rc = fsAppendPage(pFS, pPg->pSeg, &pPg->iPg, &iPrev, &iNext);
        if( rc!=LSM_OK ) return rc;

        if( pFS->mmapmgr.eUseMmap==LSM_MMAP_OFF ){
          int iHash = fsHashKey(pFS->nHash, pPg->iPg);
          pPg->pHashNext = pFS->apHash[iHash];
          pFS->apHash[iHash] = pPg;
          assert( pPg->pHashNext==0 || pPg->pHashNext->iPg!=pPg->iPg );
        }

        if( iPrev ){
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
        *pp = pPg;
        assert( pPg->pNextWaiting==0 );

      }else{
        i64 iOff;                   /* Offset to write within database file */

        iOff = (i64)pFS->nPagesize * (i64)(pPg->iPg-1);
        if( pFS->bUseMmap==0 ){
          u8 *aData = pPg->aData - (pPg->flags & PAGE_HASPREV);
          rc = lsmEnvWrite(pFS->pEnv, pFS->fdDb, iOff, aData, pFS->nPagesize);
        }else if( pPg->flags & PAGE_FREE ){
          fsGrowMapping(pFS, iOff + pFS->nPagesize, &rc);
          if( rc==LSM_OK ){
            u8 *aTo = &((u8 *)(pFS->pMap))[iOff];
            u8 *aFrom = pPg->aData - (pPg->flags & PAGE_HASPREV);
            memcpy(aTo, aFrom, pFS->nPagesize);
            lsmFree(pFS->pEnv, aFrom);
            pPg->aData = aTo + (pPg->flags & PAGE_HASPREV);
            pPg->flags &= ~PAGE_FREE;
            fsPageAddToLru(pFS, pPg);
          }







|



|

<







2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734

2735
2736
2737
2738
2739
2740
2741
        *pp = pPg;
        assert( pPg->pNextWaiting==0 );

      }else{
        i64 iOff;                   /* Offset to write within database file */

        iOff = (i64)pFS->nPagesize * (i64)(pPg->iPg-1);
        if( pFS->mmapmgr.eUseMmap==0 ){
          u8 *aData = pPg->aData - (pPg->flags & PAGE_HASPREV);
          rc = lsmEnvWrite(pFS->pEnv, pFS->fdDb, iOff, aData, pFS->nPagesize);
        }else if( pPg->flags & PAGE_FREE ){
          u8 *aTo = (u8 *)fsMmapRef(pFS, iOff, pFS->nPagesize, &pPg->pRef, &rc);
          if( rc==LSM_OK ){

            u8 *aFrom = pPg->aData - (pPg->flags & PAGE_HASPREV);
            memcpy(aTo, aFrom, pFS->nPagesize);
            lsmFree(pFS->pEnv, aFrom);
            pPg->aData = aTo + (pPg->flags & PAGE_HASPREV);
            pPg->flags &= ~PAGE_FREE;
            fsPageAddToLru(pFS, pPg);
          }
2507
2508
2509
2510
2511
2512
2513
2514

2515
2516
2517
2518
2519
2520
2521
      assert( pPg->pFS->pCompress 
           || fsIsFirst(pPg->pFS, pPg->iPg)==0 
           || (pPg->flags & PAGE_HASPREV)
      );
      pPg->aData -= (pPg->flags & PAGE_HASPREV);
      pPg->flags &= ~PAGE_HASPREV;

      if( pFS->bUseMmap ){

        pPg->pHashNext = pFS->pFree;
        pFS->pFree = pPg;
      }else{
        assert( pPg->pLruNext==0 );
        assert( pPg->pLruPrev==0 );
        fsPageRemoveFromHash(pFS, pPg);
        fsPageBufferFree(pPg);







|
>







2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
      assert( pPg->pFS->pCompress 
           || fsIsFirst(pPg->pFS, pPg->iPg)==0 
           || (pPg->flags & PAGE_HASPREV)
      );
      pPg->aData -= (pPg->flags & PAGE_HASPREV);
      pPg->flags &= ~PAGE_HASPREV;

      if( pFS->mmapmgr.eUseMmap!=LSM_MMAP_OFF ){
        fsMmapUnref(pFS, &pPg->pRef);
        pPg->pHashNext = pFS->pFree;
        pFS->pFree = pPg;
      }else{
        assert( pPg->pLruNext==0 );
        assert( pPg->pLruPrev==0 );
        fsPageRemoveFromHash(pFS, pPg);
        fsPageBufferFree(pPg);

Changes to src/lsm_main.c.

92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
  pDb->nDfltBlksz = LSM_DFLT_BLOCK_SIZE;
  pDb->nMerge = LSM_DFLT_AUTOMERGE;
  pDb->nMaxFreelist = LSM_MAX_FREELIST_ENTRIES;
  pDb->bUseLog = LSM_DFLT_USE_LOG;
  pDb->iReader = -1;
  pDb->iRwclient = -1;
  pDb->bMultiProc = LSM_DFLT_MULTIPLE_PROCESSES;
  pDb->bMmap = LSM_DFLT_MMAP;
  pDb->xLog = xLog;
  pDb->compress.iId = LSM_COMPRESSION_NONE;
  return LSM_OK;
}

lsm_env *lsm_get_env(lsm_db *pDb){
  assert( pDb->pEnv );







|







92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
  pDb->nDfltBlksz = LSM_DFLT_BLOCK_SIZE;
  pDb->nMerge = LSM_DFLT_AUTOMERGE;
  pDb->nMaxFreelist = LSM_MAX_FREELIST_ENTRIES;
  pDb->bUseLog = LSM_DFLT_USE_LOG;
  pDb->iReader = -1;
  pDb->iRwclient = -1;
  pDb->bMultiProc = LSM_DFLT_MULTIPLE_PROCESSES;
  pDb->eMmap = LSM_DFLT_MMAP;
  pDb->xLog = xLog;
  pDb->compress.iId = LSM_COMPRESSION_NONE;
  return LSM_OK;
}

lsm_env *lsm_get_env(lsm_db *pDb){
  assert( pDb->pEnv );
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
      }
      *piVal = pDb->eSafety;
      break;
    }

    case LSM_CONFIG_MMAP: {
      int *piVal = va_arg(ap, int *);
      if( pDb->iReader<0 && *piVal>=0 && *piVal<=1 ){
        pDb->bMmap = *piVal;
        rc = lsmFsConfigure(pDb);
      }
      *piVal = pDb->bMmap;
      break;
    }

    case LSM_CONFIG_USE_LOG: {
      int *piVal = va_arg(ap, int *);
      if( pDb->nTransOpen==0 && (*piVal==0 || *piVal==1) ){
        pDb->bUseLog = *piVal;







|
|


|







325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
      }
      *piVal = pDb->eSafety;
      break;
    }

    case LSM_CONFIG_MMAP: {
      int *piVal = va_arg(ap, int *);
      if( pDb->iReader<0 && *piVal>=0 && *piVal<=2 ){
        pDb->eMmap = *piVal;
        rc = lsmFsConfigure(pDb);
      }
      *piVal = pDb->eMmap;
      break;
    }

    case LSM_CONFIG_USE_LOG: {
      int *piVal = va_arg(ap, int *);
      if( pDb->nTransOpen==0 && (*piVal==0 || *piVal==1) ){
        pDb->bUseLog = *piVal;

Changes to src/lsm_sorted.c.

700
701
702
703
704
705
706
707
708

709
710
711
712
713
714
715
    ** the left-most most descendent. */
    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);
        }







<

>







700
701
702
703
704
705
706

707
708
709
710
711
712
713
714
715
    ** the left-most most descendent. */
    if( pCsr->iPg>=0 ){
      pCsr->aPg[pCsr->iPg].iCell++;

      iLoad = btreeCursorPtr(aData, nData, pPg->iCell);
      do {
        Page *pLoad;

        rc = lsmFsDbPageGet(pCsr->pFS, pCsr->pSeg, iLoad, &pLoad);
        pCsr->iPg++;
        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);
        }

Changes to src/lsm_unix.c.

186
187
188
189
190
191
192















































193
194
195
196
197
198
199

  return rc;
}

static int lsmPosixOsSectorSize(lsm_file *pFile){
  return 512;
}
















































static int lsmPosixOsRemap(
  lsm_file *pFile, 
  lsm_i64 iMin, 
  void **ppOut,
  lsm_i64 *pnOut
){







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







186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246

  return rc;
}

static int lsmPosixOsSectorSize(lsm_file *pFile){
  return 512;
}

static int lsmPosixOsMap(
  lsm_file *pFile,
  lsm_i64 iOff,
  lsm_i64 nByte,
  void **ppOut,
  lsm_i64 *pszOut
){
  PosixFile *p = (PosixFile *)pFile;
  off_t off = (off_t)iOff;
  int prc;                        /* Posix Return Code */
  struct stat buf;
  size_t sz;

  memset(&buf, 0, sizeof(buf));
  prc = fstat(p->fd, &buf);
  if( prc!=0 ) return LSM_IOERR_BKPT;

  if( nByte<=0 ){
    sz = (size_t)(LSM_MAX(nByte*-1, (i64)buf.st_size));
  }else{
    sz = (size_t)nByte;
  }

  if( (off+sz)>buf.st_size ){
    prc = ftruncate(p->fd, (off+sz));
    if( prc!=0 ) return LSM_IOERR_BKPT;
  }

  *ppOut = mmap(0, sz, PROT_READ|PROT_WRITE, MAP_SHARED, p->fd, off);
  if( *ppOut==MAP_FAILED ){
    return LSM_IOERR_BKPT;
  }

  *pszOut = (i64)sz;
  return LSM_OK;
}

static int lsmPosixOsUnmap(
  lsm_file *pFile, 
  void *pMap, 
  lsm_i64 nMap
){
  PosixFile *p = (PosixFile *)pFile;
  munmap(pMap, (size_t)nMap);
  return LSM_OK;
}

static int lsmPosixOsRemap(
  lsm_file *pFile, 
  lsm_i64 iMin, 
  void **ppOut,
  lsm_i64 *pnOut
){
700
701
702
703
704
705
706

707



708
709
710
711
712
713
714
    lsmPosixOsFullpath,      /* xFullpath */
    lsmPosixOsOpen,          /* xOpen */
    lsmPosixOsRead,          /* xRead */
    lsmPosixOsWrite,         /* xWrite */
    lsmPosixOsTruncate,      /* xTruncate */
    lsmPosixOsSync,          /* xSync */
    lsmPosixOsSectorSize,    /* xSectorSize */

    lsmPosixOsRemap,         /* xRemap */



    lsmPosixOsFileid,        /* xFileid */
    lsmPosixOsClose,         /* xClose */
    lsmPosixOsUnlink,        /* xUnlink */
    lsmPosixOsLock,          /* xLock */
    lsmPosixOsTestLock,      /* xTestLock */
    lsmPosixOsShmMap,        /* xShmMap */
    lsmPosixOsShmBarrier,    /* xShmBarrier */







>

>
>
>







747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
    lsmPosixOsFullpath,      /* xFullpath */
    lsmPosixOsOpen,          /* xOpen */
    lsmPosixOsRead,          /* xRead */
    lsmPosixOsWrite,         /* xWrite */
    lsmPosixOsTruncate,      /* xTruncate */
    lsmPosixOsSync,          /* xSync */
    lsmPosixOsSectorSize,    /* xSectorSize */
#if 0
    lsmPosixOsRemap,         /* xRemap */
#endif
    lsmPosixOsMap,           /* xMap */
    lsmPosixOsUnmap,         /* xUnmap */
    lsmPosixOsFileid,        /* xFileid */
    lsmPosixOsClose,         /* xClose */
    lsmPosixOsUnlink,        /* xUnlink */
    lsmPosixOsLock,          /* xLock */
    lsmPosixOsTestLock,      /* xTestLock */
    lsmPosixOsShmMap,        /* xShmMap */
    lsmPosixOsShmBarrier,    /* xShmBarrier */

Changes to tool/lsmperf.tcl.

377
378
379
380
381
382
383

384
385
386
387
388
389
390
391
392
393
394
395
396
397
398









399
400
401
    puts "sleeping 20 seconds..." ; after $nSleep
    run_speed_test res.db 900 $nInsert $nSelect $nPause 200 $config $name
  }
}

#run_all_tests


generate_chart png res.db 1 2
update
capture_photo lsmperf1.gif
destroy .c

generate_chart png res.db 3 4
update
capture_photo lsmperf2.gif
destroy .c

generate_chart png res.db 5 6
update
capture_photo lsmperf3.gif
destroy .c










exit









>














|
>
>
>
>
>
>
>
>
>
|


377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
    puts "sleeping 20 seconds..." ; after $nSleep
    run_speed_test res.db 900 $nInsert $nSelect $nPause 200 $config $name
  }
}

#run_all_tests

if 0 {
generate_chart png res.db 1 2
update
capture_photo lsmperf1.gif
destroy .c

generate_chart png res.db 3 4
update
capture_photo lsmperf2.gif
destroy .c

generate_chart png res.db 5 6
update
capture_photo lsmperf3.gif
destroy .c
}

run_speed_test res.db 900 20000 20000 0 100 "mmap=2" mmap=2
after 10000
run_speed_test res.db 900 20000 20000 0 100 "mmap=0" mmap=0
generate_chart png res.db 1 2
update
#capture_photo lsmperf.gif
#destroy .c

#exit