pager.c

来自「sqlite最新源码」· C语言 代码 · 共 1,620 行 · 第 1/5 页

  i64 c = pPager->journalOff;
  if( c ){
    offset = ((c-1)/JOURNAL_HDR_SZ(pPager) + 1) * JOURNAL_HDR_SZ(pPager);
  }
  assert( offset%JOURNAL_HDR_SZ(pPager)==0 );
  assert( offset>=c );
  assert( (offset-c)<JOURNAL_HDR_SZ(pPager) );
  return offset;
}

/*
** The journal file must be open when this function is called.
**
** This function is a no-op if the journal file has not been written to
** within the current transaction (i.e. if Pager.journalOff==0).
**
** If doTruncate is non-zero or the Pager.journalSizeLimit variable is
** set to 0, then truncate the journal file to zero bytes in size. Otherwise,
** zero the 28-byte header at the start of the journal file. In either case, 
** if the pager is not in no-sync mode, sync the journal file immediately 
** after writing or truncating it.
**
** If Pager.journalSizeLimit is set to a positive, non-zero value, and
** following the truncation or zeroing described above the size of the 
** journal file in bytes is larger than this value, then truncate the
** journal file to Pager.journalSizeLimit bytes. The journal file does
** not need to be synced following this operation.
**
** If an IO error occurs, abandon processing and return the IO error code.
** Otherwise, return SQLITE_OK.
*/
static int zeroJournalHdr(Pager *pPager, int doTruncate){
  int rc = SQLITE_OK;                               /* Return code */
  assert( isOpen(pPager->jfd) );
  if( pPager->journalOff ){
    const i64 iLimit = pPager->journalSizeLimit;    /* Local cache of jsl */

    IOTRACE(("JZEROHDR %p\n", pPager))
    if( doTruncate || iLimit==0 ){
      rc = sqlite3OsTruncate(pPager->jfd, 0);
    }else{
      static const char zeroHdr[28] = {0};
      rc = sqlite3OsWrite(pPager->jfd, zeroHdr, sizeof(zeroHdr), 0);
    }
    if( rc==SQLITE_OK && !pPager->noSync ){
      rc = sqlite3OsSync(pPager->jfd, SQLITE_SYNC_DATAONLY|pPager->sync_flags);
    }

    /* At this point the transaction is committed but the write lock 
    ** is still held on the file. If there is a size limit configured for 
    ** the persistent journal and the journal file currently consumes more
    ** space than that limit allows for, truncate it now. There is no need
    ** to sync the file following this operation.
    */
    if( rc==SQLITE_OK && iLimit>0 ){
      i64 sz;
      rc = sqlite3OsFileSize(pPager->jfd, &sz);
      if( rc==SQLITE_OK && sz>iLimit ){
        rc = sqlite3OsTruncate(pPager->jfd, iLimit);
      }
    }
  }
  return rc;
}

/*
** The journal file must be open when this routine is called. A journal
** header (JOURNAL_HDR_SZ bytes) is written into the journal file at the
** current location.
**
** The format for the journal header is as follows:
** - 8 bytes: Magic identifying journal format.
** - 4 bytes: Number of records in journal, or -1 no-sync mode is on.
** - 4 bytes: Random number used for page hash.
** - 4 bytes: Initial database page count.
** - 4 bytes: Sector size used by the process that wrote this journal.
** - 4 bytes: Database page size.
** 
** Followed by (JOURNAL_HDR_SZ - 28) bytes of unused space.
*/
static int writeJournalHdr(Pager *pPager){
  int rc = SQLITE_OK;                 /* Return code */
  char *zHeader = pPager->pTmpSpace;  /* Temporary space used to build header */
  u32 nHeader = pPager->pageSize;     /* Size of buffer pointed to by zHeader */
  u32 nWrite;                         /* Bytes of header sector written */
  int ii;                             /* Loop counter */

  assert( isOpen(pPager->jfd) );      /* Journal file must be open. */

  if( nHeader>JOURNAL_HDR_SZ(pPager) ){
    nHeader = JOURNAL_HDR_SZ(pPager);
  }

  /* If there are active savepoints and any of them were created 
  ** since the most recent journal header was written, update the 
  ** PagerSavepoint.iHdrOffset fields now.
  */
  for(ii=0; ii<pPager->nSavepoint; ii++){
    if( pPager->aSavepoint[ii].iHdrOffset==0 ){
      pPager->aSavepoint[ii].iHdrOffset = pPager->journalOff;
    }
  }

  pPager->journalHdr = pPager->journalOff = journalHdrOffset(pPager);
  memcpy(zHeader, aJournalMagic, sizeof(aJournalMagic));

  /* 
  ** Write the nRec Field - the number of page records that follow this
  ** journal header. Normally, zero is written to this value at this time.
  ** After the records are added to the journal (and the journal synced, 
  ** if in full-sync mode), the zero is overwritten with the true number
  ** of records (see syncJournal()).
  **
  ** A faster alternative is to write 0xFFFFFFFF to the nRec field. When
  ** reading the journal this value tells SQLite to assume that the
  ** rest of the journal file contains valid page records. This assumption
  ** is dangerous, as if a failure occured whilst writing to the journal
  ** file it may contain some garbage data. There are two scenarios
  ** where this risk can be ignored:
  **
  **   * When the pager is in no-sync mode. Corruption can follow a
  **     power failure in this case anyway.
  **
  **   * When the SQLITE_IOCAP_SAFE_APPEND flag is set. This guarantees
  **     that garbage data is never appended to the journal file.
  */
  assert( isOpen(pPager->fd) || pPager->noSync );
  if( (pPager->noSync) || (pPager->journalMode==PAGER_JOURNALMODE_MEMORY)
   || (sqlite3OsDeviceCharacteristics(pPager->fd)&SQLITE_IOCAP_SAFE_APPEND) 
  ){
    put32bits(&zHeader[sizeof(aJournalMagic)], 0xffffffff);
  }else{
    put32bits(&zHeader[sizeof(aJournalMagic)], 0);
  }

  /* The random check-hash initialiser */ 
  sqlite3_randomness(sizeof(pPager->cksumInit), &pPager->cksumInit);
  put32bits(&zHeader[sizeof(aJournalMagic)+4], pPager->cksumInit);
  /* The initial database size */
  put32bits(&zHeader[sizeof(aJournalMagic)+8], pPager->dbOrigSize);
  /* The assumed sector size for this process */
  put32bits(&zHeader[sizeof(aJournalMagic)+12], pPager->sectorSize);

  /* The page size */
  put32bits(&zHeader[sizeof(aJournalMagic)+16], pPager->pageSize);

  /* Initializing the tail of the buffer is not necessary.  Everything
  ** works find if the following memset() is omitted.  But initializing
  ** the memory prevents valgrind from complaining, so we are willing to
  ** take the performance hit.
  */
  memset(&zHeader[sizeof(aJournalMagic)+20], 0,
         nHeader-(sizeof(aJournalMagic)+20));

  /* In theory, it is only necessary to write the 28 bytes that the 
  ** journal header consumes to the journal file here. Then increment the 
  ** Pager.journalOff variable by JOURNAL_HDR_SZ so that the next 
  ** record is written to the following sector (leaving a gap in the file
  ** that will be implicitly filled in by the OS).
  **
  ** However it has been discovered that on some systems this pattern can 
  ** be significantly slower than contiguously writing data to the file,
  ** even if that means explicitly writing data to the block of 
  ** (JOURNAL_HDR_SZ - 28) bytes that will not be used. So that is what
  ** is done. 
  **
  ** The loop is required here in case the sector-size is larger than the 
  ** database page size. Since the zHeader buffer is only Pager.pageSize
  ** bytes in size, more than one call to sqlite3OsWrite() may be required
  ** to populate the entire journal header sector.
  */ 
  for(nWrite=0; rc==SQLITE_OK&&nWrite<JOURNAL_HDR_SZ(pPager); nWrite+=nHeader){
    IOTRACE(("JHDR %p %lld %d\n", pPager, pPager->journalHdr, nHeader))
    rc = sqlite3OsWrite(pPager->jfd, zHeader, nHeader, pPager->journalOff);
    pPager->journalOff += nHeader;
  }

  return rc;
}

/*
** The journal file must be open when this is called. A journal header file
** (JOURNAL_HDR_SZ bytes) is read from the current location in the journal
** file. The current location in the journal file is given by
** pPager->journalOff. See comments above function writeJournalHdr() for
** a description of the journal header format.
**
** If the header is read successfully, *pNRec is set to the number of
** page records following this header and *pDbSize is set to the size of the
** database before the transaction began, in pages. Also, pPager->cksumInit
** is set to the value read from the journal header. SQLITE_OK is returned
** in this case.
**
** If the journal header file appears to be corrupted, SQLITE_DONE is
** returned and *pNRec and *PDbSize are undefined.  If JOURNAL_HDR_SZ bytes
** cannot be read from the journal file an error code is returned.
*/
static int readJournalHdr(
  Pager *pPager,               /* Pager object */
  i64 journalSize,             /* Size of the open journal file in bytes */
  u32 *pNRec,                  /* OUT: Value read from the nRec field */
  u32 *pDbSize                 /* OUT: Value of original database size field */
){
  int rc;                      /* Return code */
  unsigned char aMagic[8];     /* A buffer to hold the magic header */
  i64 iHdrOff;                 /* Offset of journal header being read */

  assert( isOpen(pPager->jfd) );      /* Journal file must be open. */

  /* Advance Pager.journalOff to the start of the next sector. If the
  ** journal file is too small for there to be a header stored at this
  ** point, return SQLITE_DONE.
  */
  pPager->journalOff = journalHdrOffset(pPager);
  if( pPager->journalOff+JOURNAL_HDR_SZ(pPager) > journalSize ){
    return SQLITE_DONE;
  }
  iHdrOff = pPager->journalOff;

  /* Read in the first 8 bytes of the journal header. If they do not match
  ** the  magic string found at the start of each journal header, return
  ** SQLITE_DONE. If an IO error occurs, return an error code. Otherwise,
  ** proceed.
  */
  rc = sqlite3OsRead(pPager->jfd, aMagic, sizeof(aMagic), iHdrOff);
  if( rc ){
    return rc;
  }
  if( memcmp(aMagic, aJournalMagic, sizeof(aMagic))!=0 ){
    return SQLITE_DONE;
  }

  /* Read the first three 32-bit fields of the journal header: The nRec
  ** field, the checksum-initializer and the database size at the start
  ** of the transaction. Return an error code if anything goes wrong.
  */
  if( SQLITE_OK!=(rc = read32bits(pPager->jfd, iHdrOff+8, pNRec))
   || SQLITE_OK!=(rc = read32bits(pPager->jfd, iHdrOff+12, &pPager->cksumInit))
   || SQLITE_OK!=(rc = read32bits(pPager->jfd, iHdrOff+16, pDbSize))
  ){
    return rc;
  }

  if( pPager->journalOff==0 ){
    u32 iPageSize;               /* Page-size field of journal header */
    u32 iSectorSize;             /* Sector-size field of journal header */
    u16 iPageSize16;             /* Copy of iPageSize in 16-bit variable */

    /* Read the page-size and sector-size journal header fields. */
    if( SQLITE_OK!=(rc = read32bits(pPager->jfd, iHdrOff+20, &iSectorSize))
     || SQLITE_OK!=(rc = read32bits(pPager->jfd, iHdrOff+24, &iPageSize))
    ){
      return rc;
    }

    /* Check that the values read from the page-size and sector-size fields
    ** are within range. To be 'in range', both values need to be a power
    ** of two greater than or equal to 512, and not greater than their 
    ** respective compile time maximum limits.
    */
    if( iPageSize<512                  || iSectorSize<512
     || iPageSize>SQLITE_MAX_PAGE_SIZE || iSectorSize>MAX_SECTOR_SIZE
     || ((iPageSize-1)&iPageSize)!=0   || ((iSectorSize-1)&iSectorSize)!=0 
    ){
      /* If the either the page-size or sector-size in the journal-header is 
      ** invalid, then the process that wrote the journal-header must have 
      ** crashed before the header was synced. In this case stop reading 
      ** the journal file here.
      */
      return SQLITE_DONE;
    }

    /* Update the page-size to match the value read from the journal. 
    ** Use a testcase() macro to make sure that malloc failure within 
    ** PagerSetPagesize() is tested.
    */
    iPageSize16 = (u16)iPageSize;
    rc = sqlite3PagerSetPagesize(pPager, &iPageSize16);
    testcase( rc!=SQLITE_OK );
    assert( rc!=SQLITE_OK || iPageSize16==(u16)iPageSize );

    /* Update the assumed sector-size to match the value used by 
    ** the process that created this journal. If this journal was
    ** created by a process other than this one, then this routine
    ** is being called from within pager_playback(). The local value
    ** of Pager.sectorSize is restored at the end of that routine.
    */
    pPager->sectorSize = iSectorSize;
  }

  pPager->journalOff += JOURNAL_HDR_SZ(pPager);
  return rc;
}


/*
** Write the supplied master journal name into the journal file for pager
** pPager at the current location. The master journal name must be the last
** thing written to a journal file. If the pager is in full-sync mode, the
** journal file descriptor is advanced to the next sector boundary before
** anything is written. The format is:
**
**   + 4 bytes: PAGER_MJ_PGNO.
**   + N bytes: Master journal filename in utf-8.
**   + 4 bytes: N (length of master journal name in bytes, no nul-terminator).
**   + 4 bytes: Master journal name checksum.
**   + 8 bytes: aJournalMagic[].
**
** The master journal page checksum is the sum of the bytes in the master
** journal name, where each byte is interpreted as a signed 8-bit integer.
**
** If zMaster is a NULL pointer (occurs for a single database transaction), 
** this call is a no-op.
*/
static int writeMasterJournal(Pager *pPager, const char *zMaster){
  int rc;                          /* Return code */
  int nMaster;                     /* Length of string zMaster */
  i64 iHdrOff;                     /* Offset of header in journal file */
  i64 jrnlSize;                    /* Size of journal file on disk */
  u32 cksum = 0;                   /* Checksum of string zMaster */

  if( !zMaster || pPager->setMaster
   || pPager->journalMode==PAGER_JOURNALMODE_MEMORY 
   || pPager->journalMode==PAGER_JOURNALMODE_OFF