SQLite

}
int sqlite3OsDeviceCharacteristics(sqlite3_file *id){
  return id->pMethods->xDeviceCharacteristics(id);
}
int sqlite3OsShmOpen(sqlite3_file *id){
  return id->pMethods->xShmOpen(id);
}
int sqlite3OsShmSize(sqlite3_file *id, int reqSize, int *pNewSize){
  return id->pMethods->xShmSize(id, reqSize, pNewSize);
}
int sqlite3OsShmGet(sqlite3_file *id,int reqSize,int *pSize,void volatile **pp){
  return id->pMethods->xShmGet(id, reqSize, pSize, pp);
}
int sqlite3OsShmRelease(sqlite3_file *id){
  return id->pMethods->xShmRelease(id);
}
int sqlite3OsShmLock(sqlite3_file *id, int offset, int n, int flags){
  return id->pMethods->xShmLock(id, offset, n, flags);
}
void sqlite3OsShmBarrier(sqlite3_file *id){
  id->pMethods->xShmBarrier(id);
}
int sqlite3OsShmClose(sqlite3_file *id, int deleteFlag){
  return id->pMethods->xShmClose(id, deleteFlag);
}










/*
** The next group of routines are convenience wrappers around the
** VFS methods.
*/
int sqlite3OsOpen(
  sqlite3_vfs *pVfs, 

int sqlite3OsUnlock(sqlite3_file*, int);
int sqlite3OsCheckReservedLock(sqlite3_file *id, int *pResOut);
int sqlite3OsFileControl(sqlite3_file*,int,void*);
#define SQLITE_FCNTL_DB_UNCHANGED 0xca093fa0
int sqlite3OsSectorSize(sqlite3_file *id);
int sqlite3OsDeviceCharacteristics(sqlite3_file *id);
int sqlite3OsShmOpen(sqlite3_file *id);
int sqlite3OsShmSize(sqlite3_file *id, int, int*);
int sqlite3OsShmGet(sqlite3_file *id, int, int*, void volatile**);
int sqlite3OsShmRelease(sqlite3_file *id);
int sqlite3OsShmLock(sqlite3_file *id, int, int, int);
void sqlite3OsShmBarrier(sqlite3_file *id);
int sqlite3OsShmClose(sqlite3_file *id, int);


/* 
** Functions for accessing sqlite3_vfs methods 
*/
int sqlite3OsOpen(sqlite3_vfs *, const char *, sqlite3_file*, int, int *);
int sqlite3OsDelete(sqlite3_vfs *, const char *, int);
int sqlite3OsAccess(sqlite3_vfs *, const char *, int, int *pResOut);

** 
**      fid
**      zFilename
**
** Either unixShmNode.mutex must be held or unixShmNode.nRef==0 and
** unixMutexHeld() is true when reading or writing any other field
** in this structure.
**
** To avoid deadlocks, mutex and mutexBuf are always released in the
** reverse order that they are acquired.  mutexBuf is always acquired
** first and released last.  This invariant is check by asserting
** sqlite3_mutex_notheld() on mutex whenever mutexBuf is acquired or
** released.
*/
struct unixShmNode {
  unixInodeInfo *pInode;     /* unixInodeInfo that owns this SHM node */
  sqlite3_mutex *mutex;      /* Mutex to access this object */
  sqlite3_mutex *mutexBuf;   /* Mutex to access zBuf[] */
  char *zFilename;           /* Name of the mmapped file */
  int h;                     /* Open file descriptor */
  int szMap;                 /* Size of the mapping into memory */
  char *pMMapBuf;            /* Where currently mmapped().  NULL if unmapped */

  int nRef;                  /* Number of unixShm objects pointing to this */
  unixShm *pFirst;           /* All unixShm objects pointing to this */
#ifdef SQLITE_DEBUG
  u8 exclMask;               /* Mask of exclusive locks held */
  u8 sharedMask;             /* Mask of shared locks held */
  u8 nextShmId;              /* Next available unixShm.id value */
#endif

** All other fields are read/write.  The unixShm.pFile->mutex must be held
** while accessing any read/write fields.
*/
struct unixShm {
  unixShmNode *pShmNode;     /* The underlying unixShmNode object */
  unixShm *pNext;            /* Next unixShm with the same unixShmNode */
  u8 hasMutex;               /* True if holding the unixShmNode mutex */
  u8 hasMutexBuf;            /* True if holding pFile->mutexBuf */
  u16 sharedMask;            /* Mask of shared locks held */
  u16 exclMask;              /* Mask of exclusive locks held */
#ifdef SQLITE_DEBUG
  u8 id;                     /* Id of this connection within its unixShmNode */
#endif
};

** This is not a VFS shared-memory method; it is a utility function called
** by VFS shared-memory methods.
*/
static void unixShmPurge(unixFile *pFd){
  unixShmNode *p = pFd->pInode->pShmNode;
  assert( unixMutexHeld() );
  if( p && p->nRef==0 ){

    assert( p->pInode==pFd->pInode );
    if( p->mutex ) sqlite3_mutex_free(p->mutex);
    if( p->mutexBuf ) sqlite3_mutex_free(p->mutexBuf);

    if( p->pMMapBuf ) munmap(p->pMMapBuf, p->szMap);


    if( p->h>=0 ) close(p->h);
    p->pInode->pShmNode = 0;
    sqlite3_free(p);
  }
}

/* Forward reference */

    pDbFd->pInode->pShmNode = pShmNode;
    pShmNode->pInode = pDbFd->pInode;
    pShmNode->mutex = sqlite3_mutex_alloc(SQLITE_MUTEX_FAST);
    if( pShmNode->mutex==0 ){
      rc = SQLITE_NOMEM;
      goto shm_open_err;
    }
    pShmNode->mutexBuf = sqlite3_mutex_alloc(SQLITE_MUTEX_FAST);
    if( pShmNode->mutexBuf==0 ){
      rc = SQLITE_NOMEM;
      goto shm_open_err;
    }

    pShmNode->h = open(pShmNode->zFilename, O_RDWR|O_CREAT, 0664);
    if( pShmNode->h<0 ){
      rc = SQLITE_CANTOPEN_BKPT;
      goto shm_open_err;
    }

  /* Remove connection p from the set of connections associated
  ** with pShmNode */
  sqlite3_mutex_enter(pShmNode->mutex);
  for(pp=&pShmNode->pFirst; (*pp)!=p; pp = &(*pp)->pNext){}
  *pp = p->pNext;

  /* Free the connection p */
  assert( p->hasMutexBuf==0 );
  sqlite3_free(p);
  pDbFd->pShm = 0;
  sqlite3_mutex_leave(pShmNode->mutex);

  /* If pShmNode->nRef has reached 0, then close the underlying
  ** shared-memory file, too */
  unixEnterMutex();
  assert( pShmNode->nRef>0 );
  pShmNode->nRef--;
  if( pShmNode->nRef==0 ){
    if( deleteFlag ) unlink(pShmNode->zFilename);
    unixShmPurge(pDbFd);
  }
  unixLeaveMutex();

  return SQLITE_OK;
}

/*
** Changes the size of the underlying storage for  a shared-memory segment.
**
** The reqSize parameter is the new requested size of the shared memory.
** This implementation is free to increase the shared memory size to
** any amount greater than or equal to reqSize.  If the shared memory is
** already as big or bigger as reqSize, this routine is a no-op.
**
** The reqSize parameter is the minimum size requested.  The implementation
** is free to expand the storage to some larger amount if it chooses.
*/
static int unixShmSize(
  sqlite3_file *fd,         /* The open database file holding SHM */
  int reqSize,              /* Requested size.  -1 for query only */
  int *pNewSize             /* Write new size here */
){
  unixFile *pDbFd = (unixFile*)fd;
  unixShm *p = pDbFd->pShm;
  unixShmNode *pShmNode = p->pShmNode;
  int rc = SQLITE_OK;
  struct stat sStat;

  assert( pShmNode==pDbFd->pInode->pShmNode );
  assert( pShmNode->pInode==pDbFd->pInode );

  while( 1 ){
    if( fstat(pShmNode->h, &sStat)==0 ){
      *pNewSize = (int)sStat.st_size;
      if( reqSize<=(int)sStat.st_size ) break;
    }else{
      *pNewSize = 0;
      rc = SQLITE_IOERR_SHMSIZE;
      break;
    }
    rc = ftruncate(pShmNode->h, reqSize);
    reqSize = -1;
  }
  return rc;
}

/*
** Release the lock held on the shared memory segment to that other
** threads are free to resize it if necessary.
**
** If the lock is not currently held, this routine is a harmless no-op.
**
** If the shared-memory object is in lock state RECOVER, then we do not
** really want to release the lock, so in that case too, this routine
** is a no-op.
*/
static int unixShmRelease(sqlite3_file *fd){
  unixFile *pDbFd = (unixFile*)fd;
  unixShm *p = pDbFd->pShm;

  if( p->hasMutexBuf ){
    assert( sqlite3_mutex_notheld(p->pShmNode->mutex) );
    sqlite3_mutex_leave(p->pShmNode->mutexBuf);
    p->hasMutexBuf = 0;
  }
  return SQLITE_OK;
}

/*
** Map the shared storage into memory. 
**
** If reqMapSize is positive, then an attempt is made to make the
** mapping at least reqMapSize bytes in size.  However, the mapping
** will never be larger than the size of the underlying shared memory
** as set by prior calls to xShmSize().  
**
** *ppBuf is made to point to the memory which is a mapping of the
** underlying storage.  A mutex is acquired to prevent other threads
** from running while *ppBuf is in use in order to prevent other threads
** remapping *ppBuf out from under this thread.  The unixShmRelease()
** call will release the mutex.  However, if the lock state is CHECKPOINT,
** the mutex is not acquired because CHECKPOINT will never remap the
** buffer.  RECOVER might remap, though, so CHECKPOINT will acquire
** the mutex if and when it promotes to RECOVER.
**
** RECOVER needs to be atomic.  The same mutex that prevents *ppBuf from
** being remapped also prevents more than one thread from being in
** RECOVER at a time.  But, RECOVER sometimes wants to remap itself.
** To prevent RECOVER from losing its lock while remapping, the
** mutex is not released by unixShmRelease() when in RECOVER.
**
** *pNewMapSize is set to the size of the mapping.  Usually *pNewMapSize
** will be reqMapSize or larger, though it could be smaller if the
** underlying shared memory has never been enlarged to reqMapSize bytes
** by prior calls to xShmSize().
**
** *ppBuf might be NULL and zero if no space has
** yet been allocated to the underlying storage.
*/
static int unixShmGet(
  sqlite3_file *fd,        /* Database file holding shared memory */
  int reqMapSize,          /* Requested size of mapping. -1 means don't care */
  int *pNewMapSize,        /* Write new size of mapping here */
  void volatile **ppBuf    /* Write mapping buffer origin here */
){
  unixFile *pDbFd = (unixFile*)fd;
  unixShm *p = pDbFd->pShm;
  unixShmNode *pShmNode = p->pShmNode;
  int rc = SQLITE_OK;

  assert( pShmNode==pDbFd->pInode->pShmNode );
  assert( pShmNode->pInode==pDbFd->pInode );

  if( p->hasMutexBuf==0 ){
    assert( sqlite3_mutex_notheld(pShmNode->mutex) );
    sqlite3_mutex_enter(pShmNode->mutexBuf);
    p->hasMutexBuf = 1;
  }
  sqlite3_mutex_enter(pShmNode->mutex);
  if( pShmNode->szMap==0 || reqMapSize>pShmNode->szMap ){
    int actualSize;
    if( unixShmSize(fd, -1, &actualSize)!=SQLITE_OK ){
      actualSize = 0;
    }
    reqMapSize = actualSize;
    if( pShmNode->pMMapBuf || reqMapSize<=0 ){
      munmap(pShmNode->pMMapBuf, pShmNode->szMap);
    }
    if( reqMapSize>0 ){
      pShmNode->pMMapBuf = mmap(0, reqMapSize, PROT_READ|PROT_WRITE, MAP_SHARED,
                             pShmNode->h, 0);
      pShmNode->szMap = pShmNode->pMMapBuf ? reqMapSize : 0;
    }else{
      pShmNode->pMMapBuf = 0;
      pShmNode->szMap = 0;
    }
  }
  *pNewMapSize = pShmNode->szMap;
  *ppBuf = pShmNode->pMMapBuf;
  sqlite3_mutex_leave(pShmNode->mutex);
  if( *ppBuf==0 ){
    /* Do not hold the mutex if a NULL pointer is being returned. */
    unixShmRelease(fd);
  }
  return rc;
}


/*
** Change the lock state for a shared-memory segment.
**
** Note that the relationship between SHAREd and EXCLUSIVE locks is a little
** different here than in posix.  In xShmLock(), one can go from unlocked
** to shared and back or from unlocked to exclusive and back.  But one may

/*
** Implement a memory barrier or memory fence on shared memory.  
**
** All loads and stores begun before the barrier must complete before
** any load or store begun after the barrier.
*/
static void unixShmBarrier(
  sqlite3_file *fd           /* Database file holding the shared memory */
){
  unixEnterMutex();
  unixLeaveMutex();
}

































































































#else
# define unixShmOpen    0
# define unixShmSize    0
# define unixShmGet     0
# define unixShmRelease 0
# define unixShmLock    0
# define unixShmBarrier 0
# define unixShmClose   0

#endif /* #ifndef SQLITE_OMIT_WAL */

/*
** Here ends the implementation of all sqlite3_file methods.
**
********************** End sqlite3_file Methods *******************************
******************************************************************************/

   LOCK,                       /* xLock */                                   \
   UNLOCK,                     /* xUnlock */                                 \
   CKLOCK,                     /* xCheckReservedLock */                      \
   unixFileControl,            /* xFileControl */                            \
   unixSectorSize,             /* xSectorSize */                             \
   unixDeviceCharacteristics,  /* xDeviceCapabilities */                     \
   unixShmOpen,                /* xShmOpen */                                \
   unixShmSize,                /* xShmSize */                                \
   unixShmGet,                 /* xShmGet */                                 \
   unixShmRelease,             /* xShmRelease */                             \
   unixShmLock,                /* xShmLock */                                \
   unixShmBarrier,             /* xShmBarrier */                             \
   unixShmClose                /* xShmClose */                               \

};                                                                           \
static const sqlite3_io_methods *FINDER##Impl(const char *z, unixFile *p){   \
  UNUSED_PARAMETER(z); UNUSED_PARAMETER(p);                                  \
  return &METHOD;                                                            \
}                                                                            \
static const sqlite3_io_methods *(*const FINDER)(const char*,unixFile *p)    \
    = FINDER##Impl;

** reverse order that they are acquired.  mutexBuf is always acquired
** first and released last.  This invariant is check by asserting
** sqlite3_mutex_notheld() on mutex whenever mutexBuf is acquired or
** released.
*/
struct winShmNode {
  sqlite3_mutex *mutex;      /* Mutex to access this object */
  sqlite3_mutex *mutexBuf;   /* Mutex to access zBuf[] */
  char *zFilename;           /* Name of the file */
  winFile hFile;             /* File handle from winOpen */




  HANDLE hMap;               /* File handle from CreateFileMapping */


  DWORD lastErrno;           /* The Windows errno from the last I/O error */
  int szMap;                 /* Size of the mapping of file into memory */
  char *pMMapBuf;            /* Where currently mmapped().  NULL if unmapped */
  int nRef;                  /* Number of winShm objects pointing to this */
  winShm *pFirst;            /* All winShm objects pointing to this */
  winShmNode *pNext;         /* Next in list of all winShmNode objects */
#ifdef SQLITE_DEBUG
  u8 nextShmId;              /* Next available winShm.id value */
#endif
};

static void winShmPurge(sqlite3_vfs *pVfs, int deleteFlag){
  winShmNode **pp;
  winShmNode *p;
  assert( winShmMutexHeld() );
  pp = &winShmNodeList;
  while( (p = *pp)!=0 ){
    if( p->nRef==0 ){

      if( p->mutex ) sqlite3_mutex_free(p->mutex);
      if( p->mutexBuf ) sqlite3_mutex_free(p->mutexBuf);
      if( p->pMMapBuf ){
        UnmapViewOfFile(p->pMMapBuf);
      }
      if( INVALID_HANDLE_VALUE != p->hMap ){
        CloseHandle(p->hMap);
      }
      if( p->hFile.h != INVALID_HANDLE_VALUE ) {
        winClose((sqlite3_file *)&p->hFile);
      }
      if( deleteFlag ) winDelete(pVfs, p->zFilename, 0);
      *pp = p->pNext;

      sqlite3_free(p);
    }else{
      pp = &p->pNext;
    }
  }
}

    if( sqlite3StrICmp(pShmNode->zFilename, pNew->zFilename)==0 ) break;
  }
  if( pShmNode ){
    sqlite3_free(pNew);
  }else{
    pShmNode = pNew;
    pNew = 0;
    pShmNode->pMMapBuf = NULL;
    pShmNode->hMap = INVALID_HANDLE_VALUE;
    ((winFile*)(&pShmNode->hFile))->h = INVALID_HANDLE_VALUE;
    pShmNode->pNext = winShmNodeList;
    winShmNodeList = pShmNode;

    pShmNode->mutex = sqlite3_mutex_alloc(SQLITE_MUTEX_FAST);
    if( pShmNode->mutex==0 ){
      rc = SQLITE_NOMEM;
      goto shm_open_err;
    }
    pShmNode->mutexBuf = sqlite3_mutex_alloc(SQLITE_MUTEX_FAST);
    if( pShmNode->mutexBuf==0 ){
      rc = SQLITE_NOMEM;
      goto shm_open_err;
    }
    rc = winOpen(pDbFd->pVfs,
                 pShmNode->zFilename,             /* Name of the file (UTF-8) */
                 (sqlite3_file*)&pShmNode->hFile,  /* File handle here */
                 SQLITE_OPEN_READWRITE | SQLITE_OPEN_CREATE, /* Mode flags */
                 0);

  }
  winShmLeaveMutex();

  return SQLITE_OK;
}

/*

** Increase the size of the underlying storage for a shared-memory segment.


**
** The reqSize parameter is the new requested minimum size of the underlying
** shared memory.  This routine may choose to make the shared memory larger
** than this value (for example to round the shared memory size up to an
** operating-system dependent page size.)
**
** This routine will only grow the size of shared memory.  A request for
** a smaller size is a no-op.
*/
static int winShmSize(
  sqlite3_file *fd,         /* Database holding the shared memory */
  int reqSize,              /* Requested size.  -1 for query only */
  int *pNewSize             /* Write new size here */
){
  winFile *pDbFd = (winFile*)fd;
  winShm *p = pDbFd->pShm;
  winShmNode *pShmNode = p->pShmNode;
  int rc = SQLITE_OK;

  *pNewSize = 0;
  if( reqSize>=0 ){
    sqlite3_int64 sz;
    rc = winFileSize((sqlite3_file *)&pShmNode->hFile, &sz);
    if( SQLITE_OK==rc && reqSize>sz ){
      rc = winTruncate((sqlite3_file *)&pShmNode->hFile, reqSize);
    }
  }
  if( SQLITE_OK==rc ){
    sqlite3_int64 sz;
    rc = winFileSize((sqlite3_file *)&pShmNode->hFile, &sz);
    if( SQLITE_OK==rc ){
      *pNewSize = (int)sz;
    }else{
      rc = SQLITE_IOERR;
    }


  }
  return rc;
}


/*
** Map the shared storage into memory.  The minimum size of the
** mapping should be reqMapSize if reqMapSize is positive.  If
** reqMapSize is zero or negative, the implementation can choose
** whatever mapping size is convenient.
**
** *ppBuf is made to point to the memory which is a mapping of the
** underlying storage.  A mutex is acquired to prevent other threads
** from running while *ppBuf is in use in order to prevent other threads
** remapping *ppBuf out from under this thread.  The winShmRelease()
** call will release the mutex.  However, if the lock state is CHECKPOINT,
** the mutex is not acquired because CHECKPOINT will never remap the
** buffer.  RECOVER might remap, though, so CHECKPOINT will acquire
** the mutex if and when it promotes to RECOVER.
**
** RECOVER needs to be atomic.  The same mutex that prevents *ppBuf from
** being remapped also prevents more than one thread from being in
** RECOVER at a time.  But, RECOVER sometimes wants to remap itself.
** To prevent RECOVER from losing its lock while remapping, the
** mutex is not released by winShmRelease() when in RECOVER.
**
** *pNewMapSize is set to the size of the mapping.
**
** *ppBuf and *pNewMapSize might be NULL and zero if no space has
** yet been allocated to the underlying storage.
*/
static int winShmGet(
  sqlite3_file *fd,        /* The database file holding the shared memory */

  int reqMapSize,          /* Requested size of mapping. -1 means don't care */
  int *pNewMapSize,        /* Write new size of mapping here */
  void volatile **ppBuf    /* Write mapping buffer origin here */
){
  winFile *pDbFd = (winFile*)fd;
  winShm *p = pDbFd->pShm;
  winShmNode *pShmNode = p->pShmNode;
  int rc = SQLITE_OK;

  if( p->hasMutexBuf==0 ){
    assert( sqlite3_mutex_notheld(pShmNode->mutex) );
    sqlite3_mutex_enter(pShmNode->mutexBuf);
    p->hasMutexBuf = 1;

  }
  sqlite3_mutex_enter(pShmNode->mutex);
  if( pShmNode->szMap==0 || reqMapSize>pShmNode->szMap ){

    int actualSize;
    if( winShmSize(fd, -1, &actualSize)==SQLITE_OK
     && reqMapSize<actualSize
    ){
      reqMapSize = actualSize;

    }
    if( pShmNode->pMMapBuf ){
      if( !UnmapViewOfFile(pShmNode->pMMapBuf) ){




        pShmNode->lastErrno = GetLastError();
        rc = SQLITE_IOERR;

      }
      CloseHandle(pShmNode->hMap);
      pShmNode->hMap = INVALID_HANDLE_VALUE;
    }


    if( SQLITE_OK == rc ){





      pShmNode->pMMapBuf = 0;
      if( reqMapSize == 0 ){



        /* can't create 0 byte file mapping in Windows */


        pShmNode->szMap = 0;
      }else{
        /* create the file mapping object */





        if( INVALID_HANDLE_VALUE == pShmNode->hMap ){
          /* TBD provide an object name to each file

          ** mapping so it can be re-used across processes.

          */
          pShmNode->hMap = CreateFileMapping(pShmNode->hFile.h,
                                          NULL,
                                          PAGE_READWRITE,



                                          0,
                                          reqMapSize,
                                          NULL);
        }
        if( NULL==pShmNode->hMap ){
          pShmNode->lastErrno = GetLastError();
          rc = SQLITE_IOERR;
          pShmNode->szMap = 0;
          pShmNode->hMap = INVALID_HANDLE_VALUE;

        }else{
          pShmNode->pMMapBuf = MapViewOfFile(pShmNode->hMap,
                                          FILE_MAP_WRITE | FILE_MAP_READ,
                                          0,
                                          0,
                                          reqMapSize);
          if( !pShmNode->pMMapBuf ){
            pShmNode->lastErrno = GetLastError();
            rc = SQLITE_IOERR;
            pShmNode->szMap = 0;
          }else{
            pShmNode->szMap = reqMapSize;
          }
        }
      }




    }

  }
  *pNewMapSize = pShmNode->szMap;
  *ppBuf = pShmNode->pMMapBuf;
  sqlite3_mutex_leave(pShmNode->mutex);
  return rc;
}

/*
** Release the lock held on the shared memory segment so that other
** threads are free to resize it if necessary.
**
** If the lock is not currently held, this routine is a harmless no-op.
**
** If the shared-memory object is in lock state RECOVER, then we do not
** really want to release the lock, so in that case too, this routine
** is a no-op.
*/
static int winShmRelease(sqlite3_file *fd){
  winFile *pDbFd = (winFile*)fd;
  winShm *p = pDbFd->pShm;
  if( p->hasMutexBuf ){
    winShmNode *pShmNode = p->pShmNode;
    assert( sqlite3_mutex_notheld(pShmNode->mutex) );
    sqlite3_mutex_leave(pShmNode->mutexBuf);
    p->hasMutexBuf = 0;
  }
  return SQLITE_OK;
}

/*
** Change the lock state for a shared-memory segment.
*/
static int winShmLock(
  sqlite3_file *fd,          /* Database file holding the shared memory */
  int ofst,                  /* First lock to acquire or release */
  int n,                     /* Number of locks to acquire or release */

  winLock,
  winUnlock,
  winCheckReservedLock,
  winFileControl,
  winSectorSize,
  winDeviceCharacteristics,
  winShmOpen,              /* xShmOpen */
  winShmSize,              /* xShmSize */
  winShmGet,               /* xShmGet */
  winShmRelease,           /* xShmRelease */
  winShmLock,              /* xShmLock */
  winShmBarrier,           /* xShmBarrier */
  winShmClose              /* xShmClose */

};

/***************************************************************************
** Here ends the I/O methods that form the sqlite3_io_methods object.
**
** The next block of code implements the VFS methods.
****************************************************************************/

  int (*xUnlock)(sqlite3_file*, int);
  int (*xCheckReservedLock)(sqlite3_file*, int *pResOut);
  int (*xFileControl)(sqlite3_file*, int op, void *pArg);
  int (*xSectorSize)(sqlite3_file*);
  int (*xDeviceCharacteristics)(sqlite3_file*);
  /* Methods above are valid for version 1 */
  int (*xShmOpen)(sqlite3_file*);
  int (*xShmSize)(sqlite3_file*, int reqSize, int *pNewSize);
  int (*xShmGet)(sqlite3_file*, int reqSize, int *pSize, void volatile**);
  int (*xShmRelease)(sqlite3_file*);
  int (*xShmLock)(sqlite3_file*, int offset, int n, int flags);
  void (*xShmBarrier)(sqlite3_file*);
  int (*xShmClose)(sqlite3_file*, int deleteFlag);

  /* Methods above are valid for version 2 */
  /* Additional methods may be added in future releases */
};

/*
** CAPI3REF: Standard File Control Opcodes
**

/*
** Pass-throughs for WAL support.
*/
static int cfShmOpen(sqlite3_file *pFile){
  return sqlite3OsShmOpen(((CrashFile*)pFile)->pRealFile);
}
static int cfShmSize(sqlite3_file *pFile, int reqSize, int *pNew){
  return sqlite3OsShmSize(((CrashFile*)pFile)->pRealFile, reqSize, pNew);
}
static int cfShmGet(
  sqlite3_file *pFile,
  int reqSize,
  int *pSize,
  void volatile **pp
){
  return sqlite3OsShmGet(((CrashFile*)pFile)->pRealFile, reqSize, pSize, pp);
}
static int cfShmRelease(sqlite3_file *pFile){
  return sqlite3OsShmRelease(((CrashFile*)pFile)->pRealFile);
}
static int cfShmLock(sqlite3_file *pFile, int ofst, int n, int flags){
  return sqlite3OsShmLock(((CrashFile*)pFile)->pRealFile, ofst, n, flags);
}
static void cfShmBarrier(sqlite3_file *pFile){
  sqlite3OsShmBarrier(((CrashFile*)pFile)->pRealFile);
}
static int cfShmClose(sqlite3_file *pFile, int delFlag){
  return sqlite3OsShmClose(((CrashFile*)pFile)->pRealFile, delFlag);
}










static const sqlite3_io_methods CrashFileVtab = {
  2,                            /* iVersion */
  cfClose,                      /* xClose */
  cfRead,                       /* xRead */
  cfWrite,                      /* xWrite */
  cfTruncate,                   /* xTruncate */
  cfSync,                       /* xSync */
  cfFileSize,                   /* xFileSize */
  cfLock,                       /* xLock */
  cfUnlock,                     /* xUnlock */
  cfCheckReservedLock,          /* xCheckReservedLock */
  cfFileControl,                /* xFileControl */
  cfSectorSize,                 /* xSectorSize */
  cfDeviceCharacteristics,      /* xDeviceCharacteristics */
  cfShmOpen,                    /* xShmOpen */
  cfShmSize,                    /* xShmSize */
  cfShmGet,                     /* xShmGet */
  cfShmRelease,                 /* xShmRelease */
  cfShmLock,                    /* xShmLock */
  cfShmBarrier,                 /* xShmBarrier */
  cfShmClose                    /* xShmClose */

};

/*
** Application data for the crash VFS
*/
struct crashAppData {
  sqlite3_vfs *pOrig;                   /* Wrapped vfs structure */

static int devsymLock(sqlite3_file*, int);
static int devsymUnlock(sqlite3_file*, int);
static int devsymCheckReservedLock(sqlite3_file*, int *);
static int devsymFileControl(sqlite3_file*, int op, void *pArg);
static int devsymSectorSize(sqlite3_file*);
static int devsymDeviceCharacteristics(sqlite3_file*);
static int devsymShmOpen(sqlite3_file*);
static int devsymShmSize(sqlite3_file*,int,int*);
static int devsymShmGet(sqlite3_file*,int,int*,volatile void**);
static int devsymShmRelease(sqlite3_file*);
static int devsymShmLock(sqlite3_file*,int,int,int);
static void devsymShmBarrier(sqlite3_file*);
static int devsymShmClose(sqlite3_file*,int);


/*
** Method declarations for devsym_vfs.
*/
static int devsymOpen(sqlite3_vfs*, const char *, sqlite3_file*, int , int *);
static int devsymDelete(sqlite3_vfs*, const char *zName, int syncDir);
static int devsymAccess(sqlite3_vfs*, const char *zName, int flags, int *);

  devsymLock,                       /* xLock */
  devsymUnlock,                     /* xUnlock */
  devsymCheckReservedLock,          /* xCheckReservedLock */
  devsymFileControl,                /* xFileControl */
  devsymSectorSize,                 /* xSectorSize */
  devsymDeviceCharacteristics,      /* xDeviceCharacteristics */
  devsymShmOpen,                    /* xShmOpen */
  devsymShmSize,                    /* xShmSize */
  devsymShmGet,                     /* xShmGet */
  devsymShmRelease,                 /* xShmRelease */
  devsymShmLock,                    /* xShmLock */
  devsymShmBarrier,                 /* xShmBarrier */
  devsymShmClose                    /* xShmClose */

};

struct DevsymGlobal {
  sqlite3_vfs *pVfs;
  int iDeviceChar;
  int iSectorSize;
};

/*
** Shared-memory methods are all pass-thrus.
*/
static int devsymShmOpen(sqlite3_file *pFile){
  devsym_file *p = (devsym_file *)pFile;
  return sqlite3OsShmOpen(p->pReal);
}
static int devsymShmSize(sqlite3_file *pFile, int reqSize, int *pSize){
  devsym_file *p = (devsym_file *)pFile;
  return sqlite3OsShmSize(p->pReal, reqSize, pSize);
}
static int devsymShmGet(
  sqlite3_file *pFile,
  int reqSz,
  int *pSize,
  void volatile **pp
){
  devsym_file *p = (devsym_file *)pFile;
  return sqlite3OsShmGet(p->pReal, reqSz, pSize, pp);
}
static int devsymShmRelease(sqlite3_file *pFile){
  devsym_file *p = (devsym_file *)pFile;
  return sqlite3OsShmRelease(p->pReal);
}
static int devsymShmLock(sqlite3_file *pFile, int ofst, int n, int flags){
  devsym_file *p = (devsym_file *)pFile;
  return sqlite3OsShmLock(p->pReal, ofst, n, flags);
}
static void devsymShmBarrier(sqlite3_file *pFile){
  devsym_file *p = (devsym_file *)pFile;
  sqlite3OsShmBarrier(p->pReal);
}
static int devsymShmClose(sqlite3_file *pFile, int delFlag){
  devsym_file *p = (devsym_file *)pFile;
  return sqlite3OsShmClose(p->pReal, delFlag);
}













/*
** Open an devsym file handle.
*/
static int devsymOpen(

#define OS_SLEEP             16
#define OS_SYNC              17
#define OS_TRUNCATE          18
#define OS_UNLOCK            19
#define OS_WRITE             20
#define OS_SHMOPEN           21
#define OS_SHMCLOSE          22
#define OS_SHMGET            23
#define OS_SHMRELEASE        24
#define OS_SHMLOCK           25
#define OS_SHMBARRIER        26
#define OS_SHMSIZE           27
#define OS_ANNOTATE          28

#define OS_NUMEVENTS         29

#define VFSLOG_BUFFERSIZE 8192

typedef struct VfslogVfs VfslogVfs;

static int vfslogUnlock(sqlite3_file*, int);
static int vfslogCheckReservedLock(sqlite3_file*, int *pResOut);
static int vfslogFileControl(sqlite3_file*, int op, void *pArg);
static int vfslogSectorSize(sqlite3_file*);
static int vfslogDeviceCharacteristics(sqlite3_file*);

static int vfslogShmOpen(sqlite3_file *pFile);
static int vfslogShmSize(sqlite3_file *pFile, int reqSize, int *pNewSize);
static int vfslogShmGet(sqlite3_file *pFile, int,int*,volatile void **);
static int vfslogShmRelease(sqlite3_file *pFile);
static int vfslogShmLock(sqlite3_file *pFile, int ofst, int n, int flags);
static void vfslogShmBarrier(sqlite3_file*);
static int vfslogShmClose(sqlite3_file *pFile, int deleteFlag);


/*
** Method declarations for vfslog_vfs.
*/
static int vfslogOpen(sqlite3_vfs*, const char *, sqlite3_file*, int , int *);
static int vfslogDelete(sqlite3_vfs*, const char *zName, int syncDir);
static int vfslogAccess(sqlite3_vfs*, const char *zName, int flags, int *);

  vfslogLock,                     /* xLock */
  vfslogUnlock,                   /* xUnlock */
  vfslogCheckReservedLock,        /* xCheckReservedLock */
  vfslogFileControl,              /* xFileControl */
  vfslogSectorSize,               /* xSectorSize */
  vfslogDeviceCharacteristics,    /* xDeviceCharacteristics */
  vfslogShmOpen,                  /* xShmOpen */
  vfslogShmSize,                  /* xShmSize */
  vfslogShmGet,                   /* xShmGet */
  vfslogShmRelease,               /* xShmRelease */
  vfslogShmLock,                  /* xShmLock */
  vfslogShmBarrier,               /* xShmBarrier */
  vfslogShmClose                  /* xShmClose */

};

#if defined(SQLITE_OS_UNIX) && !defined(NO_GETTOD)
#include <sys/time.h>
static sqlite3_uint64 vfslog_time(){
  struct timeval sTime;
  gettimeofday(&sTime, 0);

  VfslogFile *p = (VfslogFile *)pFile;
  t = vfslog_time();
  rc = p->pReal->pMethods->xShmOpen(p->pReal);
  t = vfslog_time() - t;
  vfslog_call(p->pVfslog, OS_SHMOPEN, p->iFileId, t, rc, 0, 0);
  return rc;
}
static int vfslogShmSize(sqlite3_file *pFile, int reqSize, int *pNewSize){
  int rc;
  sqlite3_uint64 t;
  VfslogFile *p = (VfslogFile *)pFile;
  t = vfslog_time();
  rc = p->pReal->pMethods->xShmSize(p->pReal, reqSize, pNewSize);
  t = vfslog_time() - t;
  vfslog_call(p->pVfslog, OS_SHMSIZE, p->iFileId, t, rc, 0, 0);
  return rc;
}
static int vfslogShmGet(
  sqlite3_file *pFile,
  int req,
  int *pSize,
  volatile void **pp
){
  int rc;
  sqlite3_uint64 t;
  VfslogFile *p = (VfslogFile *)pFile;
  t = vfslog_time();
  rc = p->pReal->pMethods->xShmGet(p->pReal, req, pSize, pp);
  t = vfslog_time() - t;
  vfslog_call(p->pVfslog, OS_SHMGET, p->iFileId, t, rc, 0, 0);
  return rc;
}
static int vfslogShmRelease(sqlite3_file *pFile){
  int rc;
  sqlite3_uint64 t;
  VfslogFile *p = (VfslogFile *)pFile;
  t = vfslog_time();
  rc = p->pReal->pMethods->xShmRelease(p->pReal);
  t = vfslog_time() - t;
  vfslog_call(p->pVfslog, OS_SHMRELEASE, p->iFileId, t, rc, 0, 0);
  return rc;
}
static int vfslogShmLock(sqlite3_file *pFile, int ofst, int n, int flags){
  int rc;
  sqlite3_uint64 t;
  VfslogFile *p = (VfslogFile *)pFile;
  t = vfslog_time();
  rc = p->pReal->pMethods->xShmLock(p->pReal, ofst, n, flags);
  t = vfslog_time() - t;

  sqlite3_uint64 t;
  VfslogFile *p = (VfslogFile *)pFile;
  t = vfslog_time();
  rc = p->pReal->pMethods->xShmClose(p->pReal, deleteFlag);
  t = vfslog_time() - t;
  vfslog_call(p->pVfslog, OS_SHMCLOSE, p->iFileId, t, rc, 0, 0);
  return rc;
















}


/*
** Open an vfslog file handle.
*/
static int vfslogOpen(

    case OS_FULLPATHNAME:      zEvent = "xFullPathname"; break;
    case OS_RANDOMNESS:        zEvent = "xRandomness"; break;
    case OS_SLEEP:             zEvent = "xSleep"; break;
    case OS_CURRENTTIME:       zEvent = "xCurrentTime"; break;

    case OS_SHMCLOSE:          zEvent = "xShmClose"; break;
    case OS_SHMOPEN:           zEvent = "xShmOpen"; break;
    case OS_SHMGET:            zEvent = "xShmGet"; break;
    case OS_SHMSIZE:           zEvent = "xShmSize"; break;
    case OS_SHMRELEASE:        zEvent = "xShmRelease"; break;
    case OS_SHMLOCK:           zEvent = "xShmLock"; break;
    case OS_SHMBARRIER:        zEvent = "xShmBarrier"; break;


    case OS_ANNOTATE:          zEvent = "annotation"; break;
  }

  return zEvent;
}

** If a bit is clear in Testvfs.mask, then calls made by SQLite to the 
** corresponding VFS method is ignored for purposes of:
**
**   + Simulating IO errors, and
**   + Invoking the Tcl callback script.
*/
#define TESTVFS_SHMOPEN_MASK    0x00000001
#define TESTVFS_SHMSIZE_MASK    0x00000002
#define TESTVFS_SHMGET_MASK     0x00000004
#define TESTVFS_SHMRELEASE_MASK 0x00000008
#define TESTVFS_SHMLOCK_MASK    0x00000010
#define TESTVFS_SHMBARRIER_MASK 0x00000020
#define TESTVFS_SHMCLOSE_MASK   0x00000040


#define TESTVFS_OPEN_MASK       0x00000080
#define TESTVFS_SYNC_MASK       0x00000100
#define TESTVFS_ALL_MASK        0x000001FF




/*
** A shared-memory buffer. There is one of these objects for each shared
** memory region opened by clients. If two clients open the same file,
** there are two TestvfsFile structures but only one TestvfsBuffer structure.
*/
struct TestvfsBuffer {
  char *zFile;                    /* Associated file name */
  int n;                          /* Size of allocated buffer in bytes */
  u8 *a;                          /* Buffer allocated using ckalloc() */
  TestvfsFile *pFile;             /* List of open handles */
  TestvfsBuffer *pNext;           /* Next in linked list of all buffers */
};


#define PARENTVFS(x) (((Testvfs *)((x)->pAppData))->pParent)

static void tvfsDlClose(sqlite3_vfs*, void*);
#endif /* SQLITE_OMIT_LOAD_EXTENSION */
static int tvfsRandomness(sqlite3_vfs*, int nByte, char *zOut);
static int tvfsSleep(sqlite3_vfs*, int microseconds);
static int tvfsCurrentTime(sqlite3_vfs*, double*);

static int tvfsShmOpen(sqlite3_file*);
static int tvfsShmSize(sqlite3_file*, int , int *);
static int tvfsShmGet(sqlite3_file*, int , int *, volatile void **);
static int tvfsShmRelease(sqlite3_file*);
static int tvfsShmLock(sqlite3_file*, int , int, int);
static void tvfsShmBarrier(sqlite3_file*);
static int tvfsShmClose(sqlite3_file*, int);


static sqlite3_io_methods tvfs_io_methods = {
  2,                            /* iVersion */
  tvfsClose,                      /* xClose */
  tvfsRead,                       /* xRead */
  tvfsWrite,                      /* xWrite */
  tvfsTruncate,                   /* xTruncate */
  tvfsSync,                       /* xSync */
  tvfsFileSize,                   /* xFileSize */
  tvfsLock,                       /* xLock */
  tvfsUnlock,                     /* xUnlock */
  tvfsCheckReservedLock,          /* xCheckReservedLock */
  tvfsFileControl,                /* xFileControl */
  tvfsSectorSize,                 /* xSectorSize */
  tvfsDeviceCharacteristics,      /* xDeviceCharacteristics */
  tvfsShmOpen,                    /* xShmOpen */
  tvfsShmSize,                    /* xShmSize */
  tvfsShmGet,                     /* xShmGet */
  tvfsShmRelease,                 /* xShmRelease */
  tvfsShmLock,                    /* xShmLock */
  tvfsShmBarrier,                 /* xShmBarrier */
  tvfsShmClose                    /* xShmClose */

};

static int tvfsResultCode(Testvfs *p, int *pRc){
  struct errcode {
    int eCode;
    const char *zCode;
  } aCode[] = {

  rc = sqlite3OsOpen(PARENTVFS(pVfs), zName, pFd->pReal, flags, pOutFlags);
  if( pFd->pReal->pMethods ){
    sqlite3_io_methods *pMethods;
    pMethods = (sqlite3_io_methods *)ckalloc(sizeof(sqlite3_io_methods));
    memcpy(pMethods, &tvfs_io_methods, sizeof(sqlite3_io_methods));
    if( ((Testvfs *)pVfs->pAppData)->isNoshm ){
      pMethods->xShmOpen = 0;
      pMethods->xShmGet = 0;
      pMethods->xShmSize = 0;
      pMethods->xShmRelease = 0;
      pMethods->xShmClose = 0;
      pMethods->xShmLock = 0;
      pMethods->xShmBarrier = 0;

    }
    pFile->pMethods = pMethods;
  }

  return rc;
}

/*
** Return the current time as a Julian Day number in *pTimeOut.
*/
static int tvfsCurrentTime(sqlite3_vfs *pVfs, double *pTimeOut){
  return PARENTVFS(pVfs)->xCurrentTime(PARENTVFS(pVfs), pTimeOut);
}

static void tvfsGrowBuffer(TestvfsFile *pFd, int reqSize, int *pNewSize){
  TestvfsBuffer *pBuffer = pFd->pShm;
  if( reqSize>pBuffer->n ){
    pBuffer->a = (u8 *)ckrealloc((char *)pBuffer->a, reqSize);
    memset(&pBuffer->a[pBuffer->n], 0x55, reqSize-pBuffer->n);
    pBuffer->n = reqSize;
  }
  *pNewSize = pBuffer->n;
}

static int tvfsInjectIoerr(Testvfs *p){
  int ret = 0;
  if( p->ioerr ){
    p->iIoerrCnt--;
    if( p->iIoerrCnt==0 || (p->iIoerrCnt<0 && p->ioerr==2) ){
      ret = 1;
      p->nIoerrFail++;

  /* Connect the TestvfsBuffer to the new TestvfsShm handle and return. */
  pFd->pNext = pBuffer->pFile;
  pBuffer->pFile = pFd;
  pFd->pShm = pBuffer;
  return SQLITE_OK;
}

static int tvfsShmSize(
  sqlite3_file *pFile,
  int reqSize,
  int *pNewSize
){
  int rc = SQLITE_OK;
  TestvfsFile *pFd = (TestvfsFile *)pFile;
  Testvfs *p = (Testvfs *)(pFd->pVfs->pAppData);

  if( p->pScript && p->mask&TESTVFS_SHMSIZE_MASK ){
    tvfsExecTcl(p, "xShmSize", 
        Tcl_NewStringObj(pFd->pShm->zFile, -1), pFd->pShmId, 0
    );
    tvfsResultCode(p, &rc);
  }
  if( rc==SQLITE_OK && p->mask&TESTVFS_SHMSIZE_MASK && tvfsInjectIoerr(p) ){
    rc = SQLITE_IOERR;
  }
  if( rc==SQLITE_OK ){
    tvfsGrowBuffer(pFd, reqSize, pNewSize);
  }
  return rc;
}

static int tvfsShmGet(
  sqlite3_file *pFile, 
  int reqMapSize, 
  int *pMapSize, 

  volatile void **pp
){
  int rc = SQLITE_OK;
  TestvfsFile *pFd = (TestvfsFile *)pFile;
  Testvfs *p = (Testvfs *)(pFd->pVfs->pAppData);

  if( p->pScript && p->mask&TESTVFS_SHMGET_MASK ){





    tvfsExecTcl(p, "xShmGet", 
        Tcl_NewStringObj(pFd->pShm->zFile, -1), pFd->pShmId, 
        Tcl_NewIntObj(reqMapSize)
    );
    tvfsResultCode(p, &rc);

  }
  if( rc==SQLITE_OK && p->mask&TESTVFS_SHMGET_MASK && tvfsInjectIoerr(p) ){
    rc = SQLITE_IOERR;
  }

  *pMapSize = pFd->pShm->n;
  *pp = pFd->pShm->a;
  return rc;
}

static int tvfsShmRelease(sqlite3_file *pFile){
  int rc = SQLITE_OK;
  TestvfsFile *pFd = (TestvfsFile *)pFile;
  Testvfs *p = (Testvfs *)(pFd->pVfs->pAppData);

  if( p->pScript && p->mask&TESTVFS_SHMRELEASE_MASK ){
    tvfsExecTcl(p, "xShmRelease", 
        Tcl_NewStringObj(pFd->pShm->zFile, -1), pFd->pShmId, 0
    );
    tvfsResultCode(p, &rc);
  }

  return rc;
}

static int tvfsShmLock(
  sqlite3_file *pFile,
  int ofst,
  int n,
  int flags
){

  }

  for(ppFd=&pBuffer->pFile; *ppFd!=pFd; ppFd=&((*ppFd)->pNext));
  assert( (*ppFd)==pFd );
  *ppFd = pFd->pNext;

  if( pBuffer->pFile==0 ){

    TestvfsBuffer **pp;
    for(pp=&p->pBuffer; *pp!=pBuffer; pp=&((*pp)->pNext));
    *pp = (*pp)->pNext;

    ckfree((char *)pBuffer->a);

    ckfree((char *)pBuffer);
  }
  pFd->pShm = 0;

  return rc;
}

  if( Tcl_GetIndexFromObj(interp, objv[1], CMD_strs, "subcommand", 0, &i) ){
    return TCL_ERROR;
  }
  Tcl_ResetResult(interp);

  switch( (enum DB_enum)i ){
    case CMD_SHM: {


      TestvfsBuffer *pBuffer;
      char *zName;
      if( objc!=3 && objc!=4 ){
        Tcl_WrongNumArgs(interp, 2, objv, "FILE ?VALUE?");
        return TCL_ERROR;
      }


      zName = Tcl_GetString(objv[2]);


      for(pBuffer=p->pBuffer; pBuffer; pBuffer=pBuffer->pNext){
        if( 0==strcmp(pBuffer->zFile, zName) ) break;
      }

      if( !pBuffer ){
        Tcl_AppendResult(interp, "no such file: ", zName, 0);
        return TCL_ERROR;
      }
      if( objc==4 ){
        int n;
        u8 *a = Tcl_GetByteArrayFromObj(objv[3], &n);
        pBuffer->a = (u8 *)ckrealloc((char *)pBuffer->a, n);




        pBuffer->n = n;

        memcpy(pBuffer->a, a, n);
      }






      Tcl_SetObjResult(interp, Tcl_NewByteArrayObj(pBuffer->a, pBuffer->n));
      break;
    }

    case CMD_FILTER: {
      static struct VfsMethod {
        char *zName;
        int mask;
      } vfsmethod [] = {
        { "xShmOpen",    TESTVFS_SHMOPEN_MASK },
        { "xShmSize",    TESTVFS_SHMSIZE_MASK },
        { "xShmGet",     TESTVFS_SHMGET_MASK },
        { "xShmRelease", TESTVFS_SHMRELEASE_MASK },
        { "xShmLock",    TESTVFS_SHMLOCK_MASK },
        { "xShmBarrier", TESTVFS_SHMBARRIER_MASK },
        { "xShmClose",   TESTVFS_SHMCLOSE_MASK },

        { "xSync",       TESTVFS_SYNC_MASK },
        { "xOpen",       TESTVFS_OPEN_MASK },
      };
      Tcl_Obj **apElem = 0;
      int nElem = 0;
      int i;
      int mask = 0;

      if( objc==3 ){
        int nByte;
        if( p->pScript ){
          Tcl_DecrRefCount(p->pScript);
          ckfree((char *)p->apScript);
          p->apScript = 0;
          p->nScript = 0;

        }
        Tcl_GetStringFromObj(objv[2], &nByte);
        if( nByte>0 ){
          p->pScript = Tcl_DuplicateObj(objv[2]);
          Tcl_IncrRefCount(p->pScript);
        }
      }else if( objc!=2 ){

  }

  zVfs = Tcl_GetString(objv[1]);
  nByte = sizeof(Testvfs) + strlen(zVfs)+1;
  p = (Testvfs *)ckalloc(nByte);
  memset(p, 0, nByte);








  p->pParent = sqlite3_vfs_find(0);
  p->interp = interp;

  p->zName = (char *)&p[1];
  memcpy(p->zName, zVfs, strlen(zVfs)+1);

  pVfs = (sqlite3_vfs *)ckalloc(sizeof(sqlite3_vfs));
  memcpy(pVfs, &tvfs_vfs, sizeof(sqlite3_vfs));
  pVfs->pAppData = (void *)p;
  pVfs->zName = p->zName;
  pVfs->mxPathname = p->pParent->mxPathname;
  pVfs->szOsFile += p->pParent->szOsFile;
  p->pVfs = pVfs;
  p->isNoshm = isNoshm;
  p->mask = TESTVFS_ALL_MASK;

  Tcl_CreateObjCommand(interp, zVfs, testvfs_obj_cmd, p, testvfs_obj_del);
  sqlite3_vfs_register(pVfs, isDefault);

  return TCL_OK;

 bad_args:
  Tcl_WrongNumArgs(interp, 1, objv, "VFSNAME ?-noshm BOOL? ?-default BOOL?");
  return TCL_ERROR;

** a page number P, return the index of the last frame for page P in the WAL,
** or return NULL if there are no frames for page P in the WAL.
**
** The wal-index consists of a header region, followed by an one or
** more index blocks.  
**
** The wal-index header contains the total number of frames within the WAL
** in the the mxFrame field.  Each index block contains information on








** HASHTABLE_NPAGE frames.  Each index block contains two sections, a
** mapping which is a database page number for each frame, and a hash
** table used to look up frames by page number.  The mapping section is
** an array of HASHTABLE_NPAGE 32-bit page numbers.  The first entry on the
** array is the page number for the first frame; the second entry is the
** page number for the second frame; and so forth.  The last index block

** holds a total of (mxFrame%HASHTABLE_NPAGE) page numbers.  All index



** blocks other than the last are completely full with HASHTABLE_NPAGE

** page numbers.  All index blocks are the same size; the mapping section
** of the last index block merely contains unused entries if mxFrame is
** not an even multiple of HASHTABLE_NPAGE.
**
** Even without using the hash table, the last frame for page P
** can be found by scanning the mapping sections of each index block
** starting with the last index block and moving toward the first, and
** within each index block, starting at the end and moving toward the
** beginning.  The first entry that equals P corresponds to the frame
** holding the content for that page.
**
** The hash table consists of HASHTABLE_NSLOT 16-bit unsigned integers.
** HASHTABLE_NSLOT = 2*HASHTABLE_NPAGE, and there is one entry in the

** following object.
*/
struct Wal {
  sqlite3_vfs *pVfs;         /* The VFS used to create pDbFd */
  sqlite3_file *pDbFd;       /* File handle for the database file */
  sqlite3_file *pWalFd;      /* File handle for WAL file */
  u32 iCallback;             /* Value to pass to log callback (or 0) */
  int szWIndex;              /* Size of the wal-index that is mapped in mem */
  volatile u32 *pWiData;     /* Pointer to wal-index content in memory */
  u16 szPage;                /* Database page size */
  i16 readLock;              /* Which read lock is being held.  -1 for none */
  u8 exclusiveMode;          /* Non-zero if connection is in exclusive mode */
  u8 isWIndexOpen;           /* True if ShmOpen() called on pDbFd */
  u8 writeLock;              /* True if in a write transaction */
  u8 ckptLock;               /* True if holding a checkpoint lock */
  WalIndexHdr hdr;           /* Wal-index header for current transaction */
  char *zWalName;            /* Name of WAL file */
  u32 nCkpt;                 /* Checkpoint sequence counter in the wal-header */
#ifdef SQLITE_DEBUG
  u8 lockError;              /* True if a locking error has occurred */
#endif
};

/*
** Return a pointer to the WalCkptInfo structure in the wal-index.

*/
static volatile WalCkptInfo *walCkptInfo(Wal *pWal){
  assert( pWal->pWiData!=0 );
  return (volatile WalCkptInfo*)&pWal->pWiData[sizeof(WalIndexHdr)/2];
}


/*
** This structure is used to implement an iterator that loops through
** all frames in the WAL in database page order. Where two or more frames
** correspond to the same database page, the iterator visits only the 
** frame most recently written to the WAL (in other words, the frame with
** the largest index).
**
** The internals of this structure are only accessed by:
**
**   walIteratorInit() - Create a new iterator,
**   walIteratorNext() - Step an iterator,
**   walIteratorFree() - Free an iterator.
**
** This functionality is used by the checkpoint code (see walCheckpoint()).
*/
struct WalIterator {
  int iPrior;           /* Last result returned from the iterator */
  int nSegment;         /* Size of the aSegment[] array */
  int nFinal;           /* Elements in aSegment[nSegment-1]  */
  struct WalSegment {
    int iNext;              /* Next slot in aIndex[] not previously returned */
    u8 *aIndex;             /* i0, i1, i2... such that aPgno[iN] ascending */
    u32 *aPgno;             /* 256 page numbers.  Pointer to Wal.pWiData */


  } aSegment[1];        /* One for every 256 entries in the WAL */
};















































































/*
** The argument to this macro must be of type u32. On a little-endian
** architecture, it returns the u32 value that results from interpreting
** the 4 bytes as a big-endian value. On a big-endian architecture, it
** returns the value that would be produced by intepreting the 4 bytes
** of the input value as a little-endian integer.

/*
** Write the header information in pWal->hdr into the wal-index.
**
** The checksum on pWal->hdr is updated before it is written.
*/
static void walIndexWriteHdr(Wal *pWal){
  WalIndexHdr *aHdr;


  assert( pWal->writeLock );
  pWal->hdr.isInit = 1;
  walChecksumBytes(1, (u8*)&pWal->hdr, offsetof(WalIndexHdr, aCksum),
                   0, pWal->hdr.aCksum);
  aHdr = (WalIndexHdr*)pWal->pWiData;
  memcpy(&aHdr[1], &pWal->hdr, sizeof(WalIndexHdr));
  sqlite3OsShmBarrier(pWal->pDbFd);
  memcpy(&aHdr[0], &pWal->hdr, sizeof(WalIndexHdr));
}

/*
** This function encodes a single frame header and writes it to a buffer
** supplied by the caller. A frame-header is made up of a series of 
** 4-byte big-endian integers, as follows:
**

  ** and the new database size.
  */
  *piPage = pgno;
  *pnTruncate = sqlite3Get4byte(&aFrame[4]);
  return 1;
}

/*
** Define the parameters of the hash tables in the wal-index file. There
** is a hash-table following every HASHTABLE_NPAGE page numbers in the
** wal-index.
**
** Changing any of these constants will alter the wal-index format and
** create incompatibilities.
*/
#define HASHTABLE_NPAGE      4096  /* Must be power of 2 and multiple of 256 */
#define HASHTABLE_DATATYPE   u16
#define HASHTABLE_HASH_1     383                  /* Should be prime */
#define HASHTABLE_NSLOT      (HASHTABLE_NPAGE*2)  /* Must be a power of 2 */
#define HASHTABLE_NBYTE      (sizeof(HASHTABLE_DATATYPE)*HASHTABLE_NSLOT)

#if defined(SQLITE_TEST) && defined(SQLITE_DEBUG)
/*
** Names of locks.  This routine is used to provide debugging output and is not
** a part of an ordinary build.
*/
static const char *walLockName(int lockIdx){

  if( pWal->exclusiveMode ) return;
  (void)sqlite3OsShmLock(pWal->pDbFd, lockIdx, n,
                         SQLITE_SHM_UNLOCK | SQLITE_SHM_EXCLUSIVE);
  WALTRACE(("WAL%p: release EXCLUSIVE-%s cnt=%d\n", pWal,
             walLockName(lockIdx), n));
}

/*
** Return the index in the Wal.pWiData array that corresponds to 
** frame iFrame.
**
** Wal.pWiData is an array of u32 elements that is the wal-index.
** The array begins with a header and is then followed by alternating
** "map" and "hash-table" blocks.  Each "map" block consists of
** HASHTABLE_NPAGE u32 elements which are page numbers corresponding
** to frames in the WAL file.  
**
** This routine returns an index X such that Wal.pWiData[X] is part
** of a "map" block that contains the page number of the iFrame-th
** frame in the WAL file.
*/
static int walIndexEntry(u32 iFrame){
  return (
      (WALINDEX_LOCK_OFFSET+WALINDEX_LOCK_RESERVED)/sizeof(u32)
    + (((iFrame-1)/HASHTABLE_NPAGE) * HASHTABLE_NBYTE)/sizeof(u32)
    + (iFrame-1)
  );
}

/*
** Return the minimum size of the shared-memory, in bytes, that is needed
** to support a wal-index containing frame iFrame.  The value returned
** includes the wal-index header and the complete "block" containing iFrame,
** including the hash table segment that follows the block.
*/
static int walMappingSize(u32 iFrame){
  const int nByte = (sizeof(u32)*HASHTABLE_NPAGE + HASHTABLE_NBYTE) ;
  return ( WALINDEX_LOCK_OFFSET 
         + WALINDEX_LOCK_RESERVED 
         + nByte * ((iFrame + HASHTABLE_NPAGE - 1)/HASHTABLE_NPAGE)
  );
}

/*
** Release our reference to the wal-index memory map, if we are holding
** it.
*/
static void walIndexUnmap(Wal *pWal){
  if( pWal->pWiData ){
    sqlite3OsShmRelease(pWal->pDbFd);
  }
  pWal->pWiData = 0;
  pWal->szWIndex = -1;
}

/*
** Map the wal-index file into memory if it isn't already. 
**
** The reqSize parameter is the requested size of the mapping.  The
** mapping will be at least this big if the underlying storage is
** that big.  But the mapping will never grow larger than the underlying
** storage.  Use the walIndexRemap() to enlarget the storage space.
*/
static int walIndexMap(Wal *pWal, int reqSize){
  int rc = SQLITE_OK;
  if( pWal->pWiData==0 || reqSize>pWal->szWIndex ){
    walIndexUnmap(pWal);
    rc = sqlite3OsShmGet(pWal->pDbFd, reqSize, &pWal->szWIndex,
                             (void volatile**)(char volatile*)&pWal->pWiData);
    if( rc!=SQLITE_OK ){
      walIndexUnmap(pWal);
    }
  }
  return rc;
}

/*
** Enlarge the wal-index to be at least enlargeTo bytes in size and
** Remap the wal-index so that the mapping covers the full size
** of the underlying file.
**
** If enlargeTo is non-negative, then increase the size of the underlying
** storage to be at least as big as enlargeTo before remapping.
*/
static int walIndexRemap(Wal *pWal, int enlargeTo){
  int rc;
  int sz;
  assert( pWal->writeLock );
  rc = sqlite3OsShmSize(pWal->pDbFd, enlargeTo, &sz);
  if( rc==SQLITE_OK && sz>pWal->szWIndex ){
    walIndexUnmap(pWal);
    rc = walIndexMap(pWal, sz);
  }
  assert( pWal->szWIndex>=enlargeTo || rc!=SQLITE_OK );
  return rc;
}

/*
** Compute a hash on a page number.  The resulting hash value must land
** between 0 and (HASHTABLE_NSLOT-1).  The walHashNext() function advances
** the hash to the next value in the event of a collision.
*/
static int walHash(u32 iPage){
  assert( iPage>0 );
  assert( (HASHTABLE_NSLOT & (HASHTABLE_NSLOT-1))==0 );
  return (iPage*HASHTABLE_HASH_1) & (HASHTABLE_NSLOT-1);
}
static int walNextHash(int iPriorHash){
  return (iPriorHash+1)&(HASHTABLE_NSLOT-1);
}


/* 
** Find the hash table and (section of the) page number array used to

** store data for WAL frame iFrame.
**
** Set output variable *paHash to point to the start of the hash table
** in the wal-index file. Set *piZero to one less than the frame 
** number of the first frame indexed by this hash table. If a
** slot in the hash table is set to N, it refers to frame number 
** (*piZero+N) in the log.
**
** Finally, set *paPgno such that for all frames F between (*piZero+1) and 
** (*piZero+HASHTABLE_NPAGE), (*paPgno)[F] is the database page number 
** associated with frame F.
*/
static void walHashFind(
  Wal *pWal,                      /* WAL handle */
  u32 iFrame,                     /* Find the hash table indexing this frame */
  volatile HASHTABLE_DATATYPE **paHash,    /* OUT: Pointer to hash index */
  volatile u32 **paPgno,          /* OUT: Pointer to page number array */
  u32 *piZero                     /* OUT: Frame associated with *paPgno[0] */
){







  u32 iZero;
  volatile u32 *aPgno;
  volatile HASHTABLE_DATATYPE *aHash;







  iZero = ((iFrame-1)/HASHTABLE_NPAGE) * HASHTABLE_NPAGE;
  aPgno = &pWal->pWiData[walIndexEntry(iZero+1)-iZero-1];
  aHash = (HASHTABLE_DATATYPE *)&aPgno[iZero+HASHTABLE_NPAGE+1];




  /* Assert that:





  **










  **   + the mapping is large enough for this hash-table, and
  **

  **   + that aPgno[iZero+1] really is the database page number associated
  **     with the first frame indexed by this hash table.
  */
  assert( (u32*)(&aHash[HASHTABLE_NSLOT])<=&pWal->pWiData[pWal->szWIndex/4] );



  assert( walIndexEntry(iZero+1)==(&aPgno[iZero+1] - pWal->pWiData) );

  *paHash = aHash;
  *paPgno = aPgno;
  *piZero = iZero;

}

/*
** Remove entries from the hash table that point to WAL slots greater
** than pWal->hdr.mxFrame.
**
** This function is called whenever pWal->hdr.mxFrame is decreased due
** to a rollback or savepoint.
**
** At most only the hash table containing pWal->hdr.mxFrame needs to be
** updated.  Any later hash tables will be automatically cleared when
** pWal->hdr.mxFrame advances to the point where those hash tables are
** actually needed.
*/
static void walCleanupHash(Wal *pWal){
  volatile HASHTABLE_DATATYPE *aHash;  /* Pointer to hash table to clear */
  volatile u32 *aPgno;                 /* Unused return from walHashFind() */
  u32 iZero;                           /* frame == (aHash[x]+iZero) */
  int iLimit = 0;                      /* Zero values greater than this */



  assert( pWal->writeLock );
  testcase( pWal->hdr.mxFrame==HASHTABLE_NPAGE-1 );
  testcase( pWal->hdr.mxFrame==HASHTABLE_NPAGE );
  testcase( pWal->hdr.mxFrame==HASHTABLE_NPAGE+1 );

  if( (pWal->hdr.mxFrame % HASHTABLE_NPAGE)>0 ){
    int nByte;                    /* Number of bytes to zero in aPgno[] */
    int i;                        /* Used to iterate through aHash[] */









    walHashFind(pWal, pWal->hdr.mxFrame+1, &aHash, &aPgno, &iZero);

    iLimit = pWal->hdr.mxFrame - iZero;
    assert( iLimit>0 );
    for(i=0; i<HASHTABLE_NSLOT; i++){
      if( aHash[i]>iLimit ){
        aHash[i] = 0;
      }
    }

    /* Zero the entries in the aPgno array that correspond to frames with
    ** frame numbers greater than pWal->hdr.mxFrame. 
    */
    nByte = sizeof(u32) * (HASHTABLE_NPAGE-iLimit);
    memset((void *)&aPgno[iZero+iLimit+1], 0, nByte);
    assert( &((u8 *)&aPgno[iZero+iLimit+1])[nByte]==(u8 *)aHash );
  }

#ifdef SQLITE_ENABLE_EXPENSIVE_ASSERT
  /* Verify that the every entry in the mapping region is still reachable
  ** via the hash table even after the cleanup.
  */
  if( iLimit ){
    int i;           /* Loop counter */
    int iKey;        /* Hash key */
    for(i=1; i<=iLimit; i++){
      for(iKey=walHash(aPgno[i+iZero]); aHash[iKey]; iKey=walNextHash(iKey)){
        if( aHash[iKey]==i ) break;
      }
      assert( aHash[iKey]==i );
    }
  }
#endif /* SQLITE_ENABLE_EXPENSIVE_ASSERT */
}


/*
** Set an entry in the wal-index that will map database page number
** pPage into WAL frame iFrame.
*/
static int walIndexAppend(Wal *pWal, u32 iFrame, u32 iPage){
  int rc;                         /* Return code */
  int nMapping;                   /* Required mapping size in bytes */
  
  /* Make sure the wal-index is mapped. Enlarge the mapping if required. */
  nMapping = walMappingSize(iFrame);
  rc = walIndexMap(pWal, nMapping);
  while( rc==SQLITE_OK && nMapping>pWal->szWIndex ){
    rc = walIndexRemap(pWal, nMapping);
  }


  /* Assuming the wal-index file was successfully mapped, find the hash 
  ** table and section of of the page number array that pertain to frame 
  ** iFrame of the WAL. Then populate the page number array and the hash 
  ** table entry.
  */
  if( rc==SQLITE_OK ){
    int iKey;                     /* Hash table key */
    u32 iZero;                    /* One less than frame number of aPgno[1] */
    volatile u32 *aPgno;                 /* Page number array */
    volatile HASHTABLE_DATATYPE *aHash;  /* Hash table */
    int idx;                             /* Value to write to hash-table slot */
    TESTONLY( int nCollide = 0;          /* Number of hash collisions */ )

    walHashFind(pWal, iFrame, &aHash, &aPgno, &iZero);
    idx = iFrame - iZero;





    if( idx==1 ){

      memset((void*)&aPgno[iZero+1], 0, HASHTABLE_NPAGE*sizeof(u32));
      memset((void*)aHash, 0, HASHTABLE_NBYTE);
    }
    assert( idx <= HASHTABLE_NSLOT/2 + 1 );

    if( aPgno[iFrame] ){
      /* If the entry in aPgno[] is already set, then the previous writer
      ** must have exited unexpectedly in the middle of a transaction (after
      ** writing one or more dirty pages to the WAL to free up memory). 
      ** Remove the remnants of that writers uncommitted transaction from 
      ** the hash-table before writing any new entries.
      */

      walCleanupHash(pWal);
      assert( !aPgno[iFrame] );
    }
    aPgno[iFrame] = iPage;

    for(iKey=walHash(iPage); aHash[iKey]; iKey=walNextHash(iKey)){
      assert( nCollide++ < idx );
    }

    aHash[iKey] = idx;

#ifdef SQLITE_ENABLE_EXPENSIVE_ASSERT
    /* Verify that the number of entries in the hash table exactly equals
    ** the number of entries in the mapping region.
    */
    {

    ** via the hash table.  This turns out to be a really, really expensive
    ** thing to check, so only do this occasionally - not on every
    ** iteration.
    */
    if( (idx&0x3ff)==0 ){
      int i;           /* Loop counter */
      for(i=1; i<=idx; i++){
        for(iKey=walHash(aPgno[i+iZero]); aHash[iKey]; iKey=walNextHash(iKey)){
          if( aHash[iKey]==i ) break;
        }
        assert( aHash[iKey]==i );
      }
    }
#endif /* SQLITE_ENABLE_EXPENSIVE_ASSERT */
  }

      }
    }

    sqlite3_free(aFrame);
  }

finished:
  if( rc==SQLITE_OK && pWal->hdr.mxFrame==0 ){
    rc = walIndexRemap(pWal, walMappingSize(1));
  }
  if( rc==SQLITE_OK ){
    volatile WalCkptInfo *pInfo;
    int i;
    pWal->hdr.aFrameCksum[0] = aFrameCksum[0];
    pWal->hdr.aFrameCksum[1] = aFrameCksum[1];
    walIndexWriteHdr(pWal);

  if( !pRet ){
    return SQLITE_NOMEM;
  }

  pRet->pVfs = pVfs;
  pRet->pWalFd = (sqlite3_file *)&pRet[1];
  pRet->pDbFd = pDbFd;
  pRet->szWIndex = -1;
  pRet->readLock = -1;
  sqlite3_randomness(8, &pRet->hdr.aSalt);
  pRet->zWalName = zWal = pVfs->szOsFile + (char*)pRet->pWalFd;
  sqlite3_snprintf(nWal, zWal, "%s-wal", zDbName);
  rc = sqlite3OsShmOpen(pDbFd);

  /* Open file handle on the write-ahead log file. */

  WalIterator *p,               /* Iterator */
  u32 *piPage,                  /* OUT: The page number of the next page */
  u32 *piFrame                  /* OUT: Wal frame index of next page */
){
  u32 iMin;                     /* Result pgno must be greater than iMin */
  u32 iRet = 0xFFFFFFFF;        /* 0xffffffff is never a valid page number */
  int i;                        /* For looping through segments */
  int nBlock = p->nFinal;       /* Number of entries in current segment */

  iMin = p->iPrior;
  assert( iMin<0xffffffff );
  for(i=p->nSegment-1; i>=0; i--){
    struct WalSegment *pSegment = &p->aSegment[i];
    while( pSegment->iNext<nBlock ){
      u32 iPg = pSegment->aPgno[pSegment->aIndex[pSegment->iNext]];
      if( iPg>iMin ){
        if( iPg<iRet ){
          iRet = iPg;
          *piFrame = i*256 + 1 + pSegment->aIndex[pSegment->iNext];
        }
        break;
      }
      pSegment->iNext++;
    }
    nBlock = 256;
  }

  *piPage = p->iPrior = iRet;
  return (iRet==0xFFFFFFFF);
}


static void walMergesort8(
  Pgno *aContent,                 /* Pages in wal */
  u8 *aBuffer,                    /* Buffer of at least *pnList items to use */
  u8 *aList,                      /* IN/OUT: List to sort */
  int *pnList                     /* IN/OUT: Number of elements in aList[] */
){
  int nList = *pnList;
  if( nList>1 ){
    int nLeft = nList / 2;        /* Elements in left list */
    int nRight = nList - nLeft;   /* Elements in right list */
    u8 *aLeft = aList;            /* Left list */
    u8 *aRight = &aList[nLeft];   /* Right list */
    int iLeft = 0;                /* Current index in aLeft */
    int iRight = 0;               /* Current index in aright */
    int iOut = 0;                 /* Current index in output buffer */



    /* TODO: Change to non-recursive version. */
    walMergesort8(aContent, aBuffer, aLeft, &nLeft);
    walMergesort8(aContent, aBuffer, aRight, &nRight);

    while( iRight<nRight || iLeft<nLeft ){
      u8 logpage;
      Pgno dbpage;

      if( (iLeft<nLeft) 
       && (iRight>=nRight || aContent[aLeft[iLeft]]<aContent[aRight[iRight]])
      ){
        logpage = aLeft[iLeft++];
      }else{

    int i;
    for(i=1; i<*pnList; i++){
      assert( aContent[aList[i]] > aContent[aList[i-1]] );
    }
  }
#endif
}








/*
** Map the wal-index into memory owned by this thread, if it is not
** mapped already.  Then construct a WalInterator object that can be
** used to loop over all pages in the WAL in ascending order.  
**
** On success, make *pp point to the newly allocated WalInterator object
** return SQLITE_OK.  Otherwise, leave *pp unchanged and return an error
** code.
**
** The calling routine should invoke walIteratorFree() to destroy the
** WalIterator object when it has finished with it.  The caller must
** also unmap the wal-index.  But the wal-index must not be unmapped
** prior to the WalIterator object being destroyed.
*/
static int walIteratorInit(Wal *pWal, WalIterator **pp){
  u32 *aData;           /* Content of the wal-index file */
  WalIterator *p;       /* Return value */
  int nSegment;         /* Number of segments to merge */
  u32 iLast;            /* Last frame in log */
  int nByte;            /* Number of bytes to allocate */
  int i;                /* Iterator variable */
  int nFinal;           /* Number of unindexed entries */
  u8 *aTmp;             /* Temp space used by merge-sort */
  u8 *aSpace;           /* Surplus space on the end of the allocation */

  /* Make sure the wal-index is mapped into local memory */
  assert( pWal->pWiData && pWal->szWIndex>=walMappingSize(pWal->hdr.mxFrame) );

  /* This routine only runs while holding SQLITE_SHM_CHECKPOINT.  No other
  ** thread is able to write to shared memory while this routine is
  ** running (or, indeed, while the WalIterator object exists).  Hence,
  ** we can cast off the volatile qualifacation from shared memory
  */
  assert( pWal->ckptLock );
  aData = (u32*)pWal->pWiData;

  /* Allocate space for the WalIterator object */
  iLast = pWal->hdr.mxFrame;
  nSegment = (iLast >> 8) + 1;
  nFinal = (iLast & 0x000000FF);
  nByte = sizeof(WalIterator) + (nSegment+1)*(sizeof(struct WalSegment)+256);


  p = (WalIterator *)sqlite3_malloc(nByte);
  if( !p ){
    return SQLITE_NOMEM;
  }
  memset(p, 0, nByte);

  /* Initialize the WalIterator object.  Each 256-entry segment is
  ** presorted in order to make iterating through all entries much
  ** faster.
  */
  p->nSegment = nSegment;
  aSpace = (u8 *)&p->aSegment[nSegment];
  aTmp = &aSpace[nSegment*256];
  for(i=0; i<nSegment; i++){

    int j;

    int nIndex = (i==nSegment-1) ? nFinal : 256;


    p->aSegment[i].aPgno = &aData[walIndexEntry(i*256+1)];








    p->aSegment[i].aIndex = aSpace;
    for(j=0; j<nIndex; j++){
      aSpace[j] = j;
    }
    walMergesort8(p->aSegment[i].aPgno, aTmp, aSpace, &nIndex);


    memset(&aSpace[nIndex], aSpace[nIndex-1], 256-nIndex);

    aSpace += 256;
    p->nFinal = nIndex;
  }


  /* Return the fully initializd WalIterator object */
  *pp = p;
  return SQLITE_OK ;
}

/* 
** Free an iterator allocated by walIteratorInit().
*/
static void walIteratorFree(WalIterator *p){
  sqlite3_free(p);
}

/*
** Copy as much content as we can from the WAL back into the database file
** in response to an sqlite3_wal_checkpoint() request or the equivalent.
**
** The amount of information copies from WAL to database might be limited
** by active readers.  This routine will never overwrite a database page
** that a concurrent reader might be using.

  int rc;                         /* Return code */
  int szPage = pWal->hdr.szPage;  /* Database page-size */
  WalIterator *pIter = 0;         /* Wal iterator context */
  u32 iDbpage = 0;                /* Next database page to write */
  u32 iFrame = 0;                 /* Wal frame containing data for iDbpage */
  u32 mxSafeFrame;                /* Max frame that can be backfilled */
  int i;                          /* Loop counter */
  volatile WalIndexHdr *pHdr;     /* The actual wal-index header in SHM */
  volatile WalCkptInfo *pInfo;    /* The checkpoint status information */

  /* Allocate the iterator */
  rc = walIteratorInit(pWal, &pIter);
  if( rc!=SQLITE_OK || pWal->hdr.mxFrame==0 ){
    goto walcheckpoint_out;
  }

  /*** TODO:  Move this test out to the caller.  Make it an assert() here ***/
  if( pWal->hdr.szPage!=nBuf ){
    rc = SQLITE_CORRUPT_BKPT;
    goto walcheckpoint_out;
  }

  /* Compute in mxSafeFrame the index of the last frame of the WAL that is
  ** safe to write into the database.  Frames beyond mxSafeFrame might
  ** overwrite database pages that are in use by active readers and thus
  ** cannot be backfilled from the WAL.
  */
  mxSafeFrame = pWal->hdr.mxFrame;
  pHdr = (volatile WalIndexHdr*)pWal->pWiData;
  pInfo = (volatile WalCkptInfo*)&pHdr[2];
  assert( pInfo==walCkptInfo(pWal) );
  for(i=1; i<WAL_NREADER; i++){
    u32 y = pInfo->aReadMark[i];
    if( mxSafeFrame>=y ){
      assert( y<=pWal->hdr.mxFrame );
      rc = walLockExclusive(pWal, WAL_READ_LOCK(i), 1);
      if( rc==SQLITE_OK ){
        pInfo->aReadMark[i] = READMARK_NOT_USED;

    /* Sync the WAL to disk */
    if( sync_flags ){
      rc = sqlite3OsSync(pWal->pWalFd, sync_flags);
    }

    /* Iterate through the contents of the WAL, copying data to the db file. */
    while( rc==SQLITE_OK && 0==walIteratorNext(pIter, &iDbpage, &iFrame) ){

      if( iFrame<=nBackfill || iFrame>mxSafeFrame ) continue;
      rc = sqlite3OsRead(pWal->pWalFd, zBuf, szPage, 
          walFrameOffset(iFrame, szPage) + WAL_FRAME_HDRSIZE
      );
      if( rc!=SQLITE_OK ) break;
      rc = sqlite3OsWrite(pWal->pDbFd, zBuf, szPage, (iDbpage-1)*szPage);
      if( rc!=SQLITE_OK ) break;
    }

    /* If work was actually accomplished... */
    if( rc==SQLITE_OK ){
      if( mxSafeFrame==pHdr[0].mxFrame ){
        rc = sqlite3OsTruncate(pWal->pDbFd, ((i64)pWal->hdr.nPage*(i64)szPage));
        if( rc==SQLITE_OK && sync_flags ){
          rc = sqlite3OsSync(pWal->pDbFd, sync_flags);
        }
      }
      if( rc==SQLITE_OK ){
        pInfo->nBackfill = mxSafeFrame;

    rc = sqlite3OsLock(pWal->pDbFd, SQLITE_LOCK_EXCLUSIVE);
    if( rc==SQLITE_OK ){
      pWal->exclusiveMode = 1;
      rc = sqlite3WalCheckpoint(pWal, sync_flags, nBuf, zBuf);
      if( rc==SQLITE_OK ){
        isDelete = 1;
      }
      walIndexUnmap(pWal);
    }

    walIndexClose(pWal, isDelete);
    sqlite3OsClose(pWal->pWalFd);
    if( isDelete ){
      sqlite3OsDelete(pWal->pVfs, pWal->zWalName, 0);
    }
    WALTRACE(("WAL%p: closed\n", pWal));

    sqlite3_free(pWal);
  }
  return rc;
}

/*
** Try to read the wal-index header.  Return 0 on success and 1 if

** pWal->hdr, then pWal->hdr is updated to the content of the new header
** and *pChanged is set to 1.
**
** If the checksum cannot be verified return non-zero. If the header
** is read successfully and the checksum verified, return zero.
*/
int walIndexTryHdr(Wal *pWal, int *pChanged){
  u32 aCksum[2];               /* Checksum on the header content */
  WalIndexHdr h1, h2;          /* Two copies of the header content */
  WalIndexHdr *aHdr;           /* Header in shared memory */

  if( pWal->szWIndex < WALINDEX_HDR_SIZE ){
    /* The wal-index is not large enough to hold the header, then assume
    ** header is invalid. */
    return 1;
  }
  assert( pWal->pWiData );

  /* Read the header. This might happen currently with a write to the
  ** same area of shared memory on a different CPU in a SMP,
  ** meaning it is possible that an inconsistent snapshot is read
  ** from the file. If this happens, return non-zero.
  **
  ** There are two copies of the header at the beginning of the wal-index.
  ** When reading, read [0] first then [1].  Writes are in the reverse order.
  ** Memory barriers are used to prevent the compiler or the hardware from
  ** reordering the reads and writes.
  */
  aHdr = (WalIndexHdr*)pWal->pWiData;
  memcpy(&h1, &aHdr[0], sizeof(h1));
  sqlite3OsShmBarrier(pWal->pDbFd);
  memcpy(&h2, &aHdr[1], sizeof(h2));

  if( memcmp(&h1, &h2, sizeof(h1))!=0 ){
    return 1;   /* Dirty read */
  }  
  if( h1.isInit==0 ){
    return 1;   /* Malformed header - probably all zeros */
  }

**
** If the wal-index header is successfully read, return SQLITE_OK. 
** Otherwise an SQLite error code.
*/
static int walIndexReadHdr(Wal *pWal, int *pChanged){
  int rc;                         /* Return code */
  int badHdr;                     /* True if a header read failed */





  assert( pChanged );
  rc = walIndexMap(pWal, walMappingSize(1));
  if( rc!=SQLITE_OK ){
    return rc;
  }



  /* Try once to read the header straight out.  This works most of the

  ** time.
  */
  badHdr = walIndexTryHdr(pWal, pChanged);

  /* If the first attempt failed, it might have been due to a race
  ** with a writer.  So get a WRITE lock and try again.
  */
  assert( badHdr==0 || pWal->writeLock==0 );
  if( badHdr ){
    rc = walLockExclusive(pWal, WAL_WRITE_LOCK, 1);
    if( rc==SQLITE_OK ){
      pWal->writeLock = 1;

      badHdr = walIndexTryHdr(pWal, pChanged);
      if( badHdr ){
        /* If the wal-index header is still malformed even while holding
        ** a WRITE lock, it can only mean that the header is corrupted and
        ** needs to be reconstructed.  So run recovery to do exactly that.
        */
        rc = walIndexRecover(pWal);
        *pChanged = 1;
      }
      walUnlockExclusive(pWal, WAL_WRITE_LOCK, 1);
      pWal->writeLock = 0;
    }
  }

  /* Make sure the mapping is large enough to cover the entire wal-index */
  if( rc==SQLITE_OK ){
    int szWanted = walMappingSize(pWal->hdr.mxFrame);
    if( pWal->szWIndex<szWanted ){
      rc = walIndexMap(pWal, szWanted);
    }
  }

  return rc;
}

/*
** This is the value that walTryBeginRead returns when it needs to

** to select a particular WAL_READ_LOCK() that strives to let the
** checkpoint process do as much work as possible.  This routine might
** update values of the aReadMark[] array in the header, but if it does
** so it takes care to hold an exclusive lock on the corresponding
** WAL_READ_LOCK() while changing values.
*/
static int walTryBeginRead(Wal *pWal, int *pChanged, int useWal, int cnt){
  volatile WalIndexHdr *pHdr;     /* Header of the wal-index */
  volatile WalCkptInfo *pInfo;    /* Checkpoint information in wal-index */
  u32 mxReadMark;                 /* Largest aReadMark[] value */
  int mxI;                        /* Index of largest aReadMark[] value */
  int i;                          /* Loop counter */
  int rc;                         /* Return code  */

  assert( pWal->readLock<0 );     /* Not currently locked */

  /* Take steps to avoid spinning forever if there is a protocol error. */
  if( cnt>5 ){
    if( cnt>100 ) return SQLITE_PROTOCOL;
    sqlite3OsSleep(pWal->pVfs, 1);

      if( rc==SQLITE_OK ){
        walUnlockShared(pWal, WAL_RECOVER_LOCK);
        rc = WAL_RETRY;
      }else if( rc==SQLITE_BUSY ){
        rc = SQLITE_BUSY_RECOVERY;
      }
    }
  }else{
    rc = walIndexMap(pWal, walMappingSize(pWal->hdr.mxFrame));
  }
  if( rc!=SQLITE_OK ){
    return rc;
  }

  pHdr = (volatile WalIndexHdr*)pWal->pWiData;
  pInfo = (volatile WalCkptInfo*)&pHdr[2];
  assert( pInfo==walCkptInfo(pWal) );
  if( !useWal && pInfo->nBackfill==pWal->hdr.mxFrame ){
    /* The WAL has been completely backfilled (or it is empty).
    ** and can be safely ignored.
    */
    rc = walLockShared(pWal, WAL_READ_LOCK(0));
    sqlite3OsShmBarrier(pWal->pDbFd);
    if( rc==SQLITE_OK ){
      if( memcmp((void *)pHdr, &pWal->hdr, sizeof(WalIndexHdr)) ){
        /* It is not safe to allow the reader to continue here if frames
        ** may have been appended to the log before READ_LOCK(0) was obtained.
        ** When holding READ_LOCK(0), the reader ignores the entire log file,
        ** which implies that the database file contains a trustworthy
        ** snapshoT. Since holding READ_LOCK(0) prevents a checkpoint from
        ** happening, this is usually correct.
        **

    ** date before proceeding. That would not be possible without somehow
    ** blocking writers. It only guarantees that a dangerous checkpoint or 
    ** log-wrap (either of which would require an exclusive lock on
    ** WAL_READ_LOCK(mxI)) has not occurred since the snapshot was valid.
    */
    sqlite3OsShmBarrier(pWal->pDbFd);
    if( pInfo->aReadMark[mxI]!=mxReadMark
     || memcmp((void *)pHdr, &pWal->hdr, sizeof(WalIndexHdr))
    ){
      walUnlockShared(pWal, WAL_READ_LOCK(mxI));
      return WAL_RETRY;
    }else{
      assert( mxReadMark<=pWal->hdr.mxFrame );
      pWal->readLock = mxI;
    }

int sqlite3WalBeginReadTransaction(Wal *pWal, int *pChanged){
  int rc;                         /* Return code */
  int cnt = 0;                    /* Number of TryBeginRead attempts */

  do{
    rc = walTryBeginRead(pWal, pChanged, 0, ++cnt);
  }while( rc==WAL_RETRY );
  walIndexUnmap(pWal);
  return rc;
}

/*
** Finish with a read transaction.  All this does is release the
** read-lock.
*/

int sqlite3WalRead(
  Wal *pWal,                      /* WAL handle */
  Pgno pgno,                      /* Database page number to read data for */
  int *pInWal,                    /* OUT: True if data is read from WAL */
  int nOut,                       /* Size of buffer pOut in bytes */
  u8 *pOut                        /* Buffer to write page data to */
){
  int rc;                         /* Return code */
  u32 iRead = 0;                  /* If !=0, WAL frame to return data from */
  u32 iLast = pWal->hdr.mxFrame;  /* Last page in WAL for this reader */
  int iHash;                      /* Used to loop through N hash tables */

  /* This routine is only be called from within a read transaction. */
  assert( pWal->readLock>=0 || pWal->lockError );

  /* If the "last page" field of the wal-index header snapshot is 0, then
  ** no data will be read from the wal under any circumstances. Return early
  ** in this case to avoid the walIndexMap/Unmap overhead.  Likewise, if
  ** pWal->readLock==0, then the WAL is ignored by the reader so
  ** return early, as if the WAL were empty.
  */
  if( iLast==0 || pWal->readLock==0 ){
    *pInWal = 0;
    return SQLITE_OK;
  }

  /* Ensure the wal-index is mapped. */
  rc = walIndexMap(pWal, walMappingSize(iLast));
  if( rc!=SQLITE_OK ){
    return rc;
  }

  /* Search the hash table or tables for an entry matching page number
  ** pgno. Each iteration of the following for() loop searches one
  ** hash table (each hash table indexes up to HASHTABLE_NPAGE frames).
  **
  ** This code may run concurrently to the code in walIndexAppend()
  ** that adds entries to the wal-index (and possibly to this hash 
  ** table). This means the value just read from the hash 

  **   (aPgno[iFrame]==pgno): 
  **     This condition filters out normal hash-table collisions.
  **
  **   (iFrame<=iLast): 
  **     This condition filters out entries that were added to the hash
  **     table after the current read-transaction had started.
  */
  for(iHash=iLast; iHash>0 && iRead==0; iHash-=HASHTABLE_NPAGE){
    volatile HASHTABLE_DATATYPE *aHash;  /* Pointer to hash table */
    volatile u32 *aPgno;                 /* Pointer to array of page numbers */
    u32 iZero;                    /* Frame number corresponding to aPgno[0] */
    int iKey;                     /* Hash slot index */
    int mxHash;                   /* upper bound on aHash[] values */

    walHashFind(pWal, iHash, &aHash, &aPgno, &iZero);
    mxHash = iLast - iZero;
    if( mxHash > HASHTABLE_NPAGE )  mxHash = HASHTABLE_NPAGE;


    for(iKey=walHash(pgno); aHash[iKey]; iKey=walNextHash(iKey)){
      u32 iFrame = aHash[iKey] + iZero;
      if( iFrame<=iLast && aPgno[iFrame]==pgno ){
        assert( iFrame>iRead );
        iRead = iFrame;
      }
    }
  }
  assert( iRead==0 || pWal->pWiData[walIndexEntry(iRead)]==pgno );

#ifdef SQLITE_ENABLE_EXPENSIVE_ASSERT
  /* If expensive assert() statements are available, do a linear search
  ** of the wal-index file content. Make sure the results agree with the
  ** result obtained using the hash indexes above.  */
  {
    u32 iRead2 = 0;
    u32 iTest;
    for(iTest=iLast; iTest>0; iTest--){
      if( pWal->pWiData[walIndexEntry(iTest)]==pgno ){
        iRead2 = iTest;
        break;
      }
    }
    assert( iRead==iRead2 );
  }
#endif

  /* If iRead is non-zero, then it is the log frame number that contains the
  ** required page. Read and return data from the log file.
  */
  walIndexUnmap(pWal);
  if( iRead ){
    i64 iOffset = walFrameOffset(iRead, pWal->hdr.szPage) + WAL_FRAME_HDRSIZE;
    *pInWal = 1;
    return sqlite3OsRead(pWal->pWalFd, pOut, nOut, iOffset);
  }

  *pInWal = 0;

  }
  pWal->writeLock = 1;

  /* If another connection has written to the database file since the
  ** time the read transaction on this connection was started, then
  ** the write is disallowed.
  */
  rc = walIndexMap(pWal, walMappingSize(pWal->hdr.mxFrame));
  if( rc ){
    walUnlockExclusive(pWal, WAL_WRITE_LOCK, 1);
    pWal->writeLock = 0;
    return rc;
  }
  if( memcmp(&pWal->hdr, (void*)pWal->pWiData, sizeof(WalIndexHdr))!=0 ){
    walUnlockExclusive(pWal, WAL_WRITE_LOCK, 1);
    pWal->writeLock = 0;
    rc = SQLITE_BUSY;
  }

  walIndexUnmap(pWal);
  return rc;
}

/*
** End a write transaction.  The commit has already been done.  This
** routine merely releases the lock.
*/

**
** Otherwise, if the callback function does not return an error, this
** function returns SQLITE_OK.
*/
int sqlite3WalUndo(Wal *pWal, int (*xUndo)(void *, Pgno), void *pUndoCtx){
  int rc = SQLITE_OK;
  if( pWal->writeLock ){
    int unused;
    Pgno iMax = pWal->hdr.mxFrame;
    Pgno iFrame;
  

    assert( pWal->pWiData==0 );

    rc = walIndexReadHdr(pWal, &unused);
    if( rc==SQLITE_OK ){
      rc = walIndexMap(pWal, walMappingSize(iMax));
    }
    if( rc==SQLITE_OK ){
      for(iFrame=pWal->hdr.mxFrame+1; 
          ALWAYS(rc==SQLITE_OK) && iFrame<=iMax; 
          iFrame++
      ){
        /* This call cannot fail. Unless the page for which the page number
        ** is passed as the second argument is (a) in the cache and 
        ** (b) has an outstanding reference, then xUndo is either a no-op
        ** (if (a) is false) or simply expels the page from the cache (if (b)
        ** is false).
        **
        ** If the upper layer is doing a rollback, it is guaranteed that there
        ** are no outstanding references to any page other than page 1. And
        ** page 1 is never written to the log until the transaction is
        ** committed. As a result, the call to xUndo may not fail.
        */
        assert( pWal->writeLock );
        assert( pWal->pWiData[walIndexEntry(iFrame)]!=1 );
        rc = xUndo(pUndoCtx, pWal->pWiData[walIndexEntry(iFrame)]);
      }
      walCleanupHash(pWal);
    }
    walIndexUnmap(pWal);
  }

  return rc;
}

/* 
** Argument aWalData must point to an array of WAL_SAVEPOINT_NDATA u32 
** values. This function populates the array with values required to 
** "rollback" the write position of the WAL handle back to the current 

    ** to the start of the log. Update the savepoint values to match.
    */
    aWalData[0] = 0;
    aWalData[3] = pWal->nCkpt;
  }

  if( aWalData[0]<pWal->hdr.mxFrame ){
    rc = walIndexMap(pWal, walMappingSize(pWal->hdr.mxFrame));
    pWal->hdr.mxFrame = aWalData[0];
    pWal->hdr.aFrameCksum[0] = aWalData[1];
    pWal->hdr.aFrameCksum[1] = aWalData[2];
    if( rc==SQLITE_OK ){
      walCleanupHash(pWal);
    }
  }

  walIndexUnmap(pWal);
  return rc;
}

/*
** This function is called just before writing a set of frames to the log
** file (see sqlite3WalFrames()). It checks to see if, instead of appending
** to the current log file, it is possible to overwrite the start of the
** existing log file with the new frames (i.e. "reset" the log). If so,
** it sets pWal->hdr.mxFrame to 0. Otherwise, pWal->hdr.mxFrame is left
** unchanged.
**
** SQLITE_OK is returned if no error is encountered (regardless of whether
** or not pWal->hdr.mxFrame is modified). An SQLite error code is returned
** if some error 
*/
static int walRestartLog(Wal *pWal){
  int rc = SQLITE_OK;
  int cnt;

  if( pWal->readLock==0 
   && SQLITE_OK==(rc = walIndexMap(pWal, walMappingSize(pWal->hdr.mxFrame)))
  ){
    volatile WalCkptInfo *pInfo = walCkptInfo(pWal);
    assert( pInfo->nBackfill==pWal->hdr.mxFrame );
    if( pInfo->nBackfill>0 ){
      rc = walLockExclusive(pWal, WAL_READ_LOCK(1), WAL_NREADER-1);
      if( rc==SQLITE_OK ){
        /* If all readers are using WAL_READ_LOCK(0) (in other words if no
        ** readers are currently using the WAL), then the transactions

    walUnlockShared(pWal, WAL_READ_LOCK(0));
    pWal->readLock = -1;
    cnt = 0;
    do{
      int notUsed;
      rc = walTryBeginRead(pWal, &notUsed, 1, ++cnt);
    }while( rc==WAL_RETRY );

    /* Unmap the wal-index before returning. Otherwise the VFS layer may
    ** hold a mutex for the duration of the IO performed by WalFrames().
    */
    walIndexUnmap(pWal);
  }
  return rc;
}

/* 
** Write a set of frames to the log. The caller must hold the write-lock
** on the log file (obtained using sqlite3WalBeginWriteTransaction()).

  u8 aFrame[WAL_FRAME_HDRSIZE];   /* Buffer to assemble frame-header in */
  PgHdr *p;                       /* Iterator to run through pList with. */
  PgHdr *pLast = 0;               /* Last frame in list */
  int nLast = 0;                  /* Number of extra copies of last page */

  assert( pList );
  assert( pWal->writeLock );
  assert( pWal->pWiData==0 );

#if defined(SQLITE_TEST) && defined(SQLITE_DEBUG)
  { int cnt; for(cnt=0, p=pList; p; p=p->pDirty, cnt++){}
    WALTRACE(("WAL%p: frame write begin. %d frames. mxFrame=%d. %s\n",
              pWal, cnt, pWal->hdr.mxFrame, isCommit ? "Commit" : "Spill"));
  }
#endif

  /* See if it is possible to write these frames into the start of the
  ** log file, instead of appending to it at pWal->hdr.mxFrame.
  */
  if( SQLITE_OK!=(rc = walRestartLog(pWal)) ){
    assert( pWal->pWiData==0 );
    return rc;
  }
  assert( pWal->pWiData==0 && pWal->readLock>0 );

  /* If this is the first frame written into the log, write the WAL
  ** header to the start of the WAL file. See comments at the top of
  ** this source file for a description of the WAL header format.
  */
  iFrame = pWal->hdr.mxFrame;
  if( iFrame==0 ){

      }
      nLast++;
      iOffset += szPage;
    }

    rc = sqlite3OsSync(pWal->pWalFd, sync_flags);
  }
  assert( pWal->pWiData==0 );

  /* Append data to the wal-index. It is not necessary to lock the 
  ** wal-index to do this as the SQLITE_SHM_WRITE lock held on the wal-index
  ** guarantees that there are no other writers, and no data that may
  ** be in use by existing readers is being overwritten.
  */
  iFrame = pWal->hdr.mxFrame;

    /* If this is a commit, update the wal-index header too. */
    if( isCommit ){
      walIndexWriteHdr(pWal);
      pWal->iCallback = iFrame;
    }
  }

  walIndexUnmap(pWal);
  WALTRACE(("WAL%p: frame write %s\n", pWal, rc ? "failed" : "ok"));
  return rc;
}

/* 
** This routine is called to implement sqlite3_wal_checkpoint() and
** related interfaces.
**
** Obtain a CHECKPOINT lock and then backfill as much information as
** we can from WAL into the database.
*/
int sqlite3WalCheckpoint(
  Wal *pWal,                      /* Wal connection */
  int sync_flags,                 /* Flags to sync db file with (or 0) */
  int nBuf,                       /* Size of temporary buffer */
  u8 *zBuf                        /* Temporary buffer to use */
){
  int rc;                         /* Return code */
  int isChanged = 0;              /* True if a new wal-index header is loaded */

  assert( pWal->pWiData==0 );
  assert( pWal->ckptLock==0 );

  WALTRACE(("WAL%p: checkpoint begins\n", pWal));
  rc = walLockExclusive(pWal, WAL_CKPT_LOCK, 1);
  if( rc ){
    /* Usually this is SQLITE_BUSY meaning that another thread or process
    ** is already running a checkpoint, or maybe a recovery.  But it might

    ** next time the pager opens a snapshot on this database it knows that
    ** the cache needs to be reset.
    */
    memset(&pWal->hdr, 0, sizeof(WalIndexHdr));
  }

  /* Release the locks. */
  walIndexUnmap(pWal);
  walUnlockExclusive(pWal, WAL_CKPT_LOCK, 1);
  pWal->ckptLock = 0;
  WALTRACE(("WAL%p: checkpoint %s\n", pWal, rc ? "failed" : "ok"));
  return rc;
}

/* Return the value to pass to a sqlite3_wal_hook callback, the

#
#***********************************************************************
#
# $Id: permutations.test,v 1.51 2009/07/01 18:09:02 danielk1977 Exp $

set testdir [file dirname $argv0]
source $testdir/tester.tcl


# Argument processing.
#
#puts "PERM-DEBUG: argv=$argv"
namespace eval ::perm {
  variable testmode [lindex $::argv 0]
  variable testfile [lindex $::argv 1]

#      of the the integer fields (so that the reader ends up with a corrupted
#      header).
#
#   3. Check that the reader recovers the wal-index and reads the correct
#      database content.
#
do_test wal2-1.0 {
  proc tvfs_cb {method args} { return SQLITE_OK }




  testvfs tvfs
  tvfs script tvfs_cb


  sqlite3 db  test.db -vfs tvfs
  sqlite3 db2 test.db -vfs tvfs

  execsql {
    PRAGMA journal_mode = WAL;
    CREATE TABLE t1(a);

        10   13   {13 91}   8             {$RECOVER $READ}
        11   14   {14 105}  9             {$RECOVER $READ}
        12   15   {15 120}  -1            {$READ}
" {

  do_test wal2-1.$tn.1 {
    execsql { INSERT INTO t1 VALUES($iInsert) }

    set ::locks [list]
    set ::cb_done 0

    proc tvfs_cb {method args} {
      if {$::cb_done == 0 && $method == "xShmGet"} {
        set ::cb_done 1
        if {$::wal_index_hdr_mod >= 0} {
          incr_tvfs_hdr [lindex $args 0] $::wal_index_hdr_mod 1
        }
      }
      if {$method == "xShmLock"} { lappend ::locks [lindex $args 2] }
      return SQLITE_OK
    }




    execsql { SELECT count(a), sum(a) FROM t1 } db2
  } $res

  do_test wal2-1.$tn.2 {
    set ::locks
  } $wal_locks
}

  {4 1 lock exclusive} {4 1 unlock exclusive} \
  {4 1 lock shared}    {4 1 unlock shared}    \
]
do_test wal2-2.0 {

  testvfs tvfs
  tvfs script tvfs_cb

  proc tvfs_cb {method args} {
    if {$method == "xShmOpen"} { set ::shm_file [lindex $args 0] }
    return SQLITE_OK
  }

  sqlite3 db  test.db -vfs tvfs
  sqlite3 db2 test.db -vfs tvfs

  execsql {

         4    7   {6 21}   {7 28}    2
         5    8   {7 28}   {8 36}    3
         6    9   {8 36}   {9 45}    4
         7   10   {9 45}   {10 55}   5
         8   11   {10 55}  {11 66}   6
         9   12   {11 66}  {12 78}   7
} {


  do_test wal2-2.$tn.1 {
    set oldhdr [set_tvfs_hdr $::shm_file]
    execsql { INSERT INTO t1 VALUES($iInsert) }
    execsql { SELECT count(a), sum(a) FROM t1 }
  } $res1

  do_test wal2-2.$tn.2 {
    set ::locks [list]
    set ::cb_done 0
    proc tvfs_cb {method args} {
      if {$::cb_done == 0 && $method == "xShmGet"} {
        set ::cb_done 1
        if {$::wal_index_hdr_mod >= 0} {
          incr_tvfs_hdr $::shm_file $::wal_index_hdr_mod 1
        }
      }
      if {$method == "xShmLock"} {
        set lock [lindex $args 2]
        lappend ::locks $lock
        if {$lock == $::WRITER} {
          set_tvfs_hdr $::shm_file $::oldhdr
        }
      }
      return SQLITE_OK
    }




    execsql { SELECT count(a), sum(a) FROM t1 } db2
  } $res0

  do_test wal2-2.$tn.3 {
    set ::locks
  } $LOCKS

  do_test wal2-2.$tn.4 {
    set ::locks [list]
    set ::cb_done 0
    proc tvfs_cb {method args} {
      if {$::cb_done == 0 && $method == "xShmGet"} {
        set ::cb_done 1
        if {$::wal_index_hdr_mod >= 0} {
          incr_tvfs_hdr $::shm_file $::wal_index_hdr_mod 1
        }
      }
      if {$method == "xShmLock"} {
        set lock [lindex $args 2]
        lappend ::locks $lock
      }
      return SQLITE_OK
    }




    execsql { SELECT count(a), sum(a) FROM t1 } db2
  } $res1
}
db close
db2 close
tvfs delete
file delete -force test.db test.db-wal test.db-journal

testvfs T -default 1
T script method_callback

proc method_callback {method args} {
  if {$method == "xShmBarrier"} {
    incr ::barrier_count
    if {$::barrier_count == 1} {
      # This code is executed within the xShmBarrier() callback invoked
      # by the client running recovery as part of writing the recovered
      # wal-index header. If a second client attempts to access the 
      # database now, it reads a corrupt (partially written) wal-index
      # header. But it cannot even get that far, as the first client
      # is still holding all the locks (recovery takes an exclusive lock
      # on *all* db locks, preventing access by any other client).

  db eval {
    DELETE FROM abc;
    PRAGMA wal_checkpoint;
  }
} -test {
  faultsim_test_result {0 {}}
}


#--------------------------------------------------------------------------
#
faultsim_delete_and_reopen
faultsim_save_and_close
do_faultsim_test walfault-4 -prep {
  faultsim_restore_and_reopen

    PRAGMA journal_mode = WAL;
  }
  faultsim_save_and_close
} {}
do_faultsim_test walfault-5 -faults shmerr* -prep {
  faultsim_restore_and_reopen
  execsql { PRAGMA wal_autocheckpoint = 0 }
  shmfault filter xShmSize
} -body {
  execsql {
    CREATE TABLE t1(x);
    BEGIN;
      INSERT INTO t1 VALUES(randomblob(400));           /* 1 */
      INSERT INTO t1 SELECT randomblob(400) FROM t1;    /* 2 */
      INSERT INTO t1 SELECT randomblob(400) FROM t1;    /* 4 */

      INSERT INTO t1 SELECT randomblob(400) FROM t1;    /* 16384 */
    COMMIT;
  }
  faultsim_save_and_close
} {}
do_faultsim_test walfault-6 -faults shmerr* -prep {
  faultsim_restore_and_reopen
  shmfault filter xShmSize
} -body {
  execsql { SELECT count(*) FROM t1 }
} -test {
  faultsim_test_result {0 16384}
  faultsim_integrity_check
  set n [db one {SELECT count(*) FROM t1}]
  if {$n != 16384 && $n != 0} { error "Incorrect number of rows: $n" }

  if {$n != 1 && $n != 2} { error "Incorrect number of rows: $n" }
}

do_test walfault-10-pre1 {
  faultsim_delete_and_reopen
  execsql {
    PRAGMA journal_mode = WAL;
    PRAGMA wal_checkpoint = 0;
    CREATE TABLE z(zz INTEGER PRIMARY KEY, zzz BLOB);
    CREATE INDEX zzzz ON z(zzz);
    INSERT INTO z VALUES(NULL, randomblob(800));
    INSERT INTO z VALUES(NULL, randomblob(800));
    INSERT INTO z SELECT NULL, randomblob(800) FROM z;
    INSERT INTO z SELECT NULL, randomblob(800) FROM z;
    INSERT INTO z SELECT NULL, randomblob(800) FROM z;

  faultsim_test_result {0 {}}
  catch { db eval { ROLLBACK } }
  faultsim_integrity_check

  set n [db eval {SELECT count(*), sum(length(zzz)) FROM z}]
  if {$n != "64 51200"} { error "Incorrect data: $n" }
}
















































































finish_test