SQLite

**   The flag is cleared as soon as the journal file is finalized (either
**   by PagerCommitPhaseTwo or PagerRollback). If an IO error prevents the
**   journal file from being successfully finalized, the setMaster flag
**   is cleared anyway (and the pager will move to ERROR state).
**
** doNotSpill, doNotSyncSpill
**
**   These two boolean variables control the behaviour of cache-spills
**   (calls made by the pcache module to the pagerStress() routine to
**   write cached data to the file-system in order to free up memory).
**
**   When doNotSpill is non-zero, writing to the database from pagerStress()
**   is disabled altogether. This is done in a very obscure case that
**   comes up during savepoint rollback that requires the pcache module
**   to allocate a new page to prevent the journal file from being written
**   while it is being traversed by code in pager_playback().
** 
**   If doNotSyncSpill is non-zero, writing to the database from pagerStress()
**   is permitted, but syncing the journal file is not. This flag is set
**   by sqlite3PagerWrite() when the file-system sector-size is larger than
**   the database page-size in order to prevent a journal sync from happening 
**   in between the journalling of two pages on the same sector. 


**
** subjInMemory
**
**   This is a boolean variable. If true, then any required sub-journal
**   is opened as an in-memory journal file. If false, then in-memory
**   sub-journals are only used for in-memory pager files.
**

/*
** This routine ends a transaction. A transaction is usually ended by 
** either a COMMIT or a ROLLBACK operation. This routine may be called 
** after rollback of a hot-journal, or if an error occurs while opening
** the journal file or writing the very first journal-header of a
** database transaction.
** 
** This routine is never called in PAGER_ERROR state. If it is called
** in PAGER_NONE or PAGER_SHARED state and the lock held is less
** exclusive than a RESERVED lock, it is a no-op.
**
** Otherwise, any active savepoints are released.
**
** If the journal file is open, then it is "finalized". Once a journal 
** file has been finalized it is not possible to use it to roll back a 
** transaction. Nor will it be considered to be a hot-journal by this
** or any other database connection. Exactly how a journal is finalized

**     The journal file is closed and deleted using sqlite3OsDelete().
**
**     If the pager is running in exclusive mode, this method of finalizing
**     the journal file is never used. Instead, if the journalMode is
**     DELETE and the pager is in exclusive mode, the method described under
**     journalMode==PERSIST is used instead.
**
** After the journal is finalized, the pager moves to PAGER_READER state.
** If running in non-exclusive rollback mode, the lock on the file is 



** downgraded to a SHARED_LOCK.

**
** SQLITE_OK is returned if no error occurs. If an error occurs during
** any of the IO operations to finalize the journal file or unlock the
** database then the IO error code is returned to the user. If the 
** operation to finalize the journal file fails, then the code still
** tries to unlock the database file if not in exclusive mode. If the
** unlock operation fails as well, then the first error code related
** to the first error encountered (the journal finalization one) is
** returned.
*/
static int pager_end_transaction(Pager *pPager, int hasMaster){
  int rc = SQLITE_OK;      /* Error code from journal finalization operation */
  int rc2 = SQLITE_OK;     /* Error code from db file unlock operation */

  /* Do nothing if the pager does not have an open write transaction
  ** or at least a RESERVED lock. This function may be called when there
  ** is no write-transaction active but a RESERVED or greater lock is
  ** held under two circumstances:
  **
  **   1. After a successful hot-journal rollback, it is called with
  **      eState==PAGER_NONE and eLock==EXCLUSIVE_LOCK.
  **
  **   2. If a connection with locking_mode=exclusive holding an EXCLUSIVE 
  **      lock switches back to locking_mode=normal and then executes a
  **      read-transaction, this function is called with eState==PAGER_READER 
  **      and eLock==EXCLUSIVE_LOCK when the read-transaction is closed.
  */
  assert( assert_pager_state(pPager) );
  assert( pPager->eState!=PAGER_ERROR );
  if( pPager->eState<PAGER_WRITER_LOCKED && pPager->eLock<RESERVED_LOCK ){
    return SQLITE_OK;
  }

  releaseAllSavepoints(pPager);

  return (rc==SQLITE_OK?rc2:rc);
}

/*
** Execute a rollback if a transaction is active and unlock the 
** database file. 
**
** If the pager has already entered the ERROR state, do not attempt 
** the rollback at this time. Instead, pager_unlock() is called. The
** call to pager_unlock() will discard all in-memory pages, unlock
** the database file and move the pager back to OPEN state. If this 
** means that there is a hot-journal left in the file-system, the next 
** connection to obtain a shared lock on the pager (which may be this one) 
** will roll it back.
**
** If the pager has not already entered the ERROR state, but an IO or
** malloc error occurs during a rollback, then this will itself cause 
** the pager to enter the ERROR state. Which will be cleared by the
** call to pager_unlock(), as described above.
*/
static void pagerUnlockAndRollback(Pager *pPager){
  if( pPager->eState!=PAGER_ERROR && pPager->eState!=PAGER_OPEN ){
    assert( assert_pager_state(pPager) );
    if( pPager->eState>=PAGER_WRITER_LOCKED ){
      sqlite3BeginBenignMalloc();
      sqlite3PagerRollback(pPager);
      sqlite3EndBenignMalloc();
    }else if( !pPager->exclusiveMode ){
      assert( pPager->eState==PAGER_READER );
      pager_end_transaction(pPager, 0);
    }
  }
  pager_unlock(pPager);
}

/*
** Read a single page from either the journal file (if isMainJrnl==1) or
** from the sub-journal (if isMainJrnl==0) and playback that page.
** The page begins at offset *pOffset into the file. The *pOffset
** value is increased to the start of the next page in the journal.
**

** The main rollback journal uses checksums - the statement journal does 
** not.
**
** If the page number of the page record read from the (sub-)journal file
** is greater than the current value of Pager.dbSize, then playback is
** skipped and SQLITE_OK is returned.
**
** If pDone is not NULL, then it is a record of pages that have already
** been played back.  If the page at *pOffset has already been played back

  assert( (isSavepnt&~1)==0 );       /* isSavepnt is 0 or 1 */
  assert( isMainJrnl || pDone );     /* pDone always used on sub-journals */
  assert( isSavepnt || pDone==0 );   /* pDone never used on non-savepoint */

  aData = pPager->pTmpSpace;
  assert( aData );         /* Temp storage must have already been allocated */
  assert( pagerUseWal(pPager)==0 || (!isMainJrnl && isSavepnt) );

  /* Either the state is greater than PAGER_WRITER_CACHEMOD (a transaction 
  ** or savepoint rollback done at the request of the caller) or this is
  ** a hot-journal rollback. If it is a hot-journal rollback, the pager
  ** is in state OPEN and holds an EXCLUSIVE lock. Hot-journal rollback
  ** only reads from the main journal, not the sub-journal.
  */
  assert( pPager->eState>=PAGER_WRITER_CACHEMOD
       || (pPager->eState==PAGER_OPEN && pPager->eLock==EXCLUSIVE_LOCK)
  );
  assert( pPager->eState>=PAGER_WRITER_CACHEMOD || isMainJrnl );

  /* Read the page number and page data from the journal or sub-journal
  ** file. Return an error code to the caller if an IO error occurs.
  */
  jfd = isMainJrnl ? pPager->jfd : pPager->sjfd;
  rc = read32bits(jfd, *pOffset, &pgno);
  if( rc!=SQLITE_OK ) return rc;

  /* If this page has already been played by before during the current
  ** rollback, then don't bother to play it back again.
  */
  if( pDone && (rc = sqlite3BitvecSet(pDone, pgno))!=SQLITE_OK ){
    return rc;
  }




  /* When playing back page 1, restore the nReserve setting
  */
  if( pgno==1 && pPager->nReserve!=((u8*)aData)[20] ){
    pPager->nReserve = ((u8*)aData)[20];
    pagerReportSize(pPager);
  }

  if( rc==SQLITE_OK && changed ){
    pager_reset(pPager);
  }

  return rc;
}


/*
** This function is called as part of the transition from PAGER_OPEN
** to PAGER_READER state to determine the size of the database file
** in pages (assuming the page size currently stored in Pager.pageSize).
**
** If no error occurs, SQLITE_OK is returned and the size of the database
** in pages is stored in *pnPage. Otherwise, an error code (perhaps
** SQLITE_IOERR_FSTAT) is returned and *pnPage is left unmodified.
*/
static int pagerPagecount(Pager *pPager, Pgno *pnPage){
  Pgno nPage;                     /* Value to return via *pnPage */

  /* Query the WAL sub-system for the database size. The WalDbsize()
  ** function returns zero if the WAL is not open (i.e. Pager.pWal==0), or
  ** if the database size is not available. The database size is not
  ** available from the WAL sub-system if the log file is empty or
  ** contains no valid committed transactions.
  */
  assert( pPager->eState==PAGER_OPEN );
  assert( pPager->eLock>=SHARED_LOCK || pPager->noReadlock );
  nPage = sqlite3WalDbsize(pPager->pWal);

  /* If the database size was not available from the WAL sub-system,
  ** determine it based on the size of the database file. If the size
  ** of the database file is not an integer multiple of the page-size,
  ** round down to the nearest page. Except, any file larger than 0
  ** bytes in size is considered to contain at least one page.
  */
  if( nPage==0 ){
    i64 n = 0;                    /* Size of db file in bytes */
    assert( isOpen(pPager->fd) || pPager->tempFile );
    if( isOpen(pPager->fd) ){
      int rc = sqlite3OsFileSize(pPager->fd, &n);
      if( rc!=SQLITE_OK ){
        return rc;
      }
    }
    nPage = (Pgno)(n / pPager->pageSize);
    if( nPage==0 && n>0 ){
      nPage = 1;
    }
  }

  /* If the current number of pages in the file is greater than the
  ** configured maximum pager number, increase the allowed limit so
  ** that the file can be read.
  */
  if( nPage>pPager->mxPgno ){
    pPager->mxPgno = (Pgno)nPage;
  }

** If the database is not empty and the *-wal file exists, open the pager
** in WAL mode.  If the database is empty or if no *-wal file exists and
** if no error occurs, make sure Pager.journalMode is not set to
** PAGER_JOURNALMODE_WAL.
**
** Return SQLITE_OK or an error code.
**


** The caller must hold a SHARED lock on the database file to call this
** function. Because an EXCLUSIVE lock on the db file is required to delete 
** a WAL on a none-empty database, this ensures there is no race condition 
** between the xAccess() below and an xDelete() being executed by some 
** other connection.
*/
static int pagerOpenWalIfPresent(Pager *pPager){
  int rc = SQLITE_OK;
  assert( pPager->eState==PAGER_OPEN );
  assert( pPager->eLock>=SHARED_LOCK || pPager->noReadlock );

  if( !pPager->tempFile ){
    int isWal;                    /* True if WAL file exists */
    Pgno nPage;                   /* Size of the database file */



    rc = pagerPagecount(pPager, &nPage);
    if( rc ) return rc;
    if( nPage==0 ){
      rc = sqlite3OsDelete(pPager->pVfs, pPager->zWal, 0);
      isWal = 0;
    }else{
      rc = sqlite3OsAccess(
          pPager->pVfs, pPager->zWal, SQLITE_ACCESS_EXISTS, &isWal
      );
    }
    if( rc==SQLITE_OK ){
      if( isWal ){
        testcase( sqlite3PcachePagecount(pPager->pPCache)==0 );
        rc = sqlite3PagerOpenWal(pPager, 0);



      }else if( pPager->journalMode==PAGER_JOURNALMODE_WAL ){
        pPager->journalMode = PAGER_JOURNALMODE_DELETE;
      }
    }
  }
  return rc;
}

** write-transaction has already been opened, this function is a no-op.
**
** If the exFlag argument is false, then acquire at least a RESERVED
** lock on the database file. If exFlag is true, then acquire at least
** an EXCLUSIVE lock. If such a lock is already held, no locking 
** functions need be called.
**








** If the subjInMemory argument is non-zero, then any sub-journal opened
** within this transaction will be opened as an in-memory file. This
** has no effect if the sub-journal is already opened (as it may be when
** running in exclusive mode) or if the transaction does not require a
** sub-journal. If the subjInMemory argument is zero, then any required
** sub-journal is implemented in-memory if pPager is an in-memory database, 
** or using a temporary file otherwise.

      ** lock, but not when obtaining the RESERVED lock.
      */
      rc = pagerLockDb(pPager, RESERVED_LOCK);
      if( rc==SQLITE_OK && exFlag ){
        rc = pager_wait_on_lock(pPager, EXCLUSIVE_LOCK);
      }
    }


    if( rc==SQLITE_OK ){
      /* Change to WRITER_LOCKED state.
      **
      ** WAL mode sets Pager.eState to PAGER_WRITER_LOCKED or CACHEMOD
      ** when it has an open transaction, but never to DBMOD or FINISHED.
      ** This is because in those states the code to roll back savepoint