SQLite: Check-in [1dde96c9]

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

Overview

Comment:	For improved clarity of presentation, refactor some of the code associated with ZERO_DAMAGE and sector-size.
Downloads:	Tarball \| ZIP archive \| SQL archive
Timelines:	family \| ancestors \| descendants \| both \| statvfs
Files:	files \| file ages \| folders
SHA1:	1dde96c9ee88af1c4e37c2e65acb7c0fe6a20e2a
User & Date:	drh 2011-12-17 20:02:11

Context

2011-12-19
00:31		Some fixes to the test suite so that it works with ZERO_DAMAGE set to true. Still lots more problems remain. (check-in: 41891b23 user: drh tags: statvfs)
2011-12-17
20:02		For improved clarity of presentation, refactor some of the code associated with ZERO_DAMAGE and sector-size. (check-in: 1dde96c9 user: drh tags: statvfs)
19:49		Add SQLITE_IOCAP_ZERO_DAMAGE and enable it for both unix and windows. Use this device characteristic to reduce the required work in journaling. A side effect is that this changes the default page exists back to 1024 even with the use of statvfs(). (check-in: a0be6ea4 user: drh tags: statvfs)

Changes

Hide Diffs Unified Diffs Ignore Whitespace Patch

Changes to src/pager.c.

** we minimize the sector size.  For backwards compatibility of the
** rollback journal file format, we cannot reduce the effective sector
** size below 512.
*/
static void setSectorSize(Pager *pPager){
  assert( isOpen(pPager->fd) || pPager->tempFile );

  if( !pPager->tempFile
   && (sqlite3OsDeviceCharacteristics(pPager->fd)&SQLITE_IOCAP_ZERO_DAMAGE)==0
  ){
    /* Sector size doesn't matter for temporary files. Also, the file
    ** may not have been opened yet, in which case the OsSectorSize()
    ** call will segfault.
    */


    pPager->sectorSize = sqlite3OsSectorSize(pPager->fd);
  }
  if( pPager->sectorSize<32 ){
    pPager->sectorSize = 512;
  }
  if( pPager->sectorSize>MAX_SECTOR_SIZE ){
    assert( MAX_SECTOR_SIZE>=512 );
    pPager->sectorSize = MAX_SECTOR_SIZE;

  }
}

/*
** Playback the journal and thus restore the database file to
** the state it was in before we started making changes.  
**








|
|



|
<
>
>

<
|
|
|
|
|
|
>

** we minimize the sector size.  For backwards compatibility of the
** rollback journal file format, we cannot reduce the effective sector
** size below 512.
*/
static void setSectorSize(Pager *pPager){
  assert( isOpen(pPager->fd) || pPager->tempFile );

  if( pPager->tempFile
   || (sqlite3OsDeviceCharacteristics(pPager->fd)&SQLITE_IOCAP_ZERO_DAMAGE)!=0
  ){
    /* Sector size doesn't matter for temporary files. Also, the file
    ** may not have been opened yet, in which case the OsSectorSize()
    ** call will segfault. */

    pPager->sectorSize = 512;
  }else{
    pPager->sectorSize = sqlite3OsSectorSize(pPager->fd);

    if( pPager->sectorSize<32 ){
      pPager->sectorSize = 512;
    }
    if( pPager->sectorSize>MAX_SECTOR_SIZE ){
      assert( MAX_SECTOR_SIZE>=512 );
      pPager->sectorSize = MAX_SECTOR_SIZE;
    }
  }
}

/*
** Playback the journal and thus restore the database file to
** the state it was in before we started making changes.  
**

Changes to src/wal.c.

  u8 syncFlags;              /* Flags to use to sync header writes */
  u8 exclusiveMode;          /* Non-zero if connection is in exclusive mode */
  u8 writeLock;              /* True if in a write transaction */
  u8 ckptLock;               /* True if holding a checkpoint lock */
  u8 readOnly;               /* WAL_RDWR, WAL_RDONLY, or WAL_SHM_RDONLY */
  u8 truncateOnCommit;       /* True to truncate WAL file on commit */
  u8 noSyncHeader;           /* Avoid WAL header fsyncs if true */
  u8 noPadding;              /* No need to pad transactions to sector size */
  WalIndexHdr hdr;           /* Wal-index header for current transaction */
  const 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
};
................................................................................

  pRet->pVfs = pVfs;
  pRet->pWalFd = (sqlite3_file *)&pRet[1];
  pRet->pDbFd = pDbFd;
  pRet->readLock = -1;
  pRet->mxWalSize = mxWalSize;
  pRet->zWalName = zWalName;

  pRet->exclusiveMode = (bNoShm ? WAL_HEAPMEMORY_MODE: WAL_NORMAL_MODE);

  /* Open file handle on the write-ahead log file. */
  flags = (SQLITE_OPEN_READWRITE|SQLITE_OPEN_CREATE|SQLITE_OPEN_WAL);
  rc = sqlite3OsOpen(pVfs, zWalName, pRet->pWalFd, flags, &flags);
  if( rc==SQLITE_OK && flags&SQLITE_OPEN_READONLY ){
    pRet->readOnly = WAL_RDONLY;
................................................................................
  if( rc!=SQLITE_OK ){
    walIndexClose(pRet, 0);
    sqlite3OsClose(pRet->pWalFd);
    sqlite3_free(pRet);
  }else{
    int iDC = sqlite3OsDeviceCharacteristics(pRet->pWalFd);
    if( iDC & SQLITE_IOCAP_SEQUENTIAL ){ pRet->noSyncHeader = 1; }
    if( iDC & SQLITE_IOCAP_ZERO_DAMAGE ){ pRet->noPadding = 1; }
    *ppWal = pRet;
    WALTRACE(("WAL%d: opened\n", pRet));
  }
  return rc;
}

/*
................................................................................
      return rc;
    }
    pLast = p;
  }

  /* Sync the log file if the 'isSync' flag was specified. */
  if( isCommit && (sync_flags & WAL_SYNC_TRANSACTIONS)!=0 ){
    if( !pWal->noPadding ){
      i64 iSegment = sqlite3OsSectorSize(pWal->pWalFd);
      i64 iOffset = walFrameOffset(iFrame+1, szPage);
  
      assert( iSegment>0 );
  
      iSegment = (((iOffset+iSegment-1)/iSegment) * iSegment);
      while( iOffset<iSegment ){

  u8 syncFlags;              /* Flags to use to sync header writes */
  u8 exclusiveMode;          /* Non-zero if connection is in exclusive mode */
  u8 writeLock;              /* True if in a write transaction */
  u8 ckptLock;               /* True if holding a checkpoint lock */
  u8 readOnly;               /* WAL_RDWR, WAL_RDONLY, or WAL_SHM_RDONLY */
  u8 truncateOnCommit;       /* True to truncate WAL file on commit */
  u8 noSyncHeader;           /* Avoid WAL header fsyncs if true */
  u8 padToSectorBoundary;    /* Pad transactions out to the next sector */
  WalIndexHdr hdr;           /* Wal-index header for current transaction */
  const 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
};
................................................................................

  pRet->pVfs = pVfs;
  pRet->pWalFd = (sqlite3_file *)&pRet[1];
  pRet->pDbFd = pDbFd;
  pRet->readLock = -1;
  pRet->mxWalSize = mxWalSize;
  pRet->zWalName = zWalName;
  pRet->padToSectorBoundary = 1;
  pRet->exclusiveMode = (bNoShm ? WAL_HEAPMEMORY_MODE: WAL_NORMAL_MODE);

  /* Open file handle on the write-ahead log file. */
  flags = (SQLITE_OPEN_READWRITE|SQLITE_OPEN_CREATE|SQLITE_OPEN_WAL);
  rc = sqlite3OsOpen(pVfs, zWalName, pRet->pWalFd, flags, &flags);
  if( rc==SQLITE_OK && flags&SQLITE_OPEN_READONLY ){
    pRet->readOnly = WAL_RDONLY;
................................................................................
  if( rc!=SQLITE_OK ){
    walIndexClose(pRet, 0);
    sqlite3OsClose(pRet->pWalFd);
    sqlite3_free(pRet);
  }else{
    int iDC = sqlite3OsDeviceCharacteristics(pRet->pWalFd);
    if( iDC & SQLITE_IOCAP_SEQUENTIAL ){ pRet->noSyncHeader = 1; }
    if( iDC & SQLITE_IOCAP_ZERO_DAMAGE ){ pRet->padToSectorBoundary = 0; }
    *ppWal = pRet;
    WALTRACE(("WAL%d: opened\n", pRet));
  }
  return rc;
}

/*
................................................................................
      return rc;
    }
    pLast = p;
  }

  /* Sync the log file if the 'isSync' flag was specified. */
  if( isCommit && (sync_flags & WAL_SYNC_TRANSACTIONS)!=0 ){
    if( pWal->padToSectorBoundary ){
      i64 iSegment = sqlite3OsSectorSize(pWal->pWalFd);
      i64 iOffset = walFrameOffset(iFrame+1, szPage);
  
      assert( iSegment>0 );
  
      iSegment = (((iOffset+iSegment-1)/iSegment) * iSegment);
      while( iOffset<iSegment ){