SQLite

** The pager is used to access a database disk file.  It implements
** atomic commit and rollback through the use of a journal file that
** is separate from the database file.  The pager also implements file
** locking to prevent two processes from writing the same database
** file simultaneously, or one process from reading the database while
** another is writing.
**
** @(#) $Id: pager.c,v 1.533 2009/01/06 15:28:35 danielk1977 Exp $
*/
#ifndef SQLITE_OMIT_DISKIO
#include "sqliteInt.h"

/*
** Macros for troubleshooting.  Normally turned off
*/
#if 0
int sqlite3PagerTrace=1;  /* True to enable tracing */
#define sqlite3DebugPrintf printf
#define PAGERTRACE1(X)       if( sqlite3PagerTrace ) sqlite3DebugPrintf(X)
#define PAGERTRACE2(X,Y)     if( sqlite3PagerTrace ) sqlite3DebugPrintf(X,Y)
#define PAGERTRACE3(X,Y,Z)   if( sqlite3PagerTrace ) sqlite3DebugPrintf(X,Y,Z)
#define PAGERTRACE4(X,Y,Z,W) if( sqlite3PagerTrace ) sqlite3DebugPrintf(X,Y,Z,W)
#define PAGERTRACE5(X,Y,Z,W,V) if( sqlite3PagerTrace ) sqlite3DebugPrintf(X,Y,Z,W,V)
#else
#define PAGERTRACE1(X)
#define PAGERTRACE2(X,Y)
#define PAGERTRACE3(X,Y,Z)
#define PAGERTRACE4(X,Y,Z,W)
#define PAGERTRACE5(X,Y,Z,W,V)
#endif

/*
** The following two macros are used within the PAGERTRACEX() macros above
** to print out file-descriptors. 
**
** PAGERID() takes a pointer to a Pager struct as its argument. The
** associated file-descriptor is returned. FILEHANDLEID() takes an sqlite3_file
** struct as its argument.
*/
#define PAGERID(p) ((int)(p->fd))

*/
#define PAGER_UNLOCK      0
#define PAGER_SHARED      1   /* same as SHARED_LOCK */
#define PAGER_RESERVED    2   /* same as RESERVED_LOCK */
#define PAGER_EXCLUSIVE   4   /* same as EXCLUSIVE_LOCK */
#define PAGER_SYNCED      5

/*
** If the SQLITE_BUSY_RESERVED_LOCK macro is set to true at compile-time,
** then failed attempts to get a reserved lock will invoke the busy callback.
** This is off by default.  To see why, consider the following scenario:
** 
** Suppose thread A already has a shared lock and wants a reserved lock.
** Thread B already has a reserved lock and wants an exclusive lock.  If
** both threads are using their busy callbacks, it might be a long time
** be for one of the threads give up and allows the other to proceed.
** But if the thread trying to get the reserved lock gives up quickly
** (if it never invokes its busy callback) then the contention will be
** resolved quickly.
*/
#ifndef SQLITE_BUSY_RESERVED_LOCK
# define SQLITE_BUSY_RESERVED_LOCK 0
#endif

/*
** This macro rounds values up so that if the value is an address it
** is guaranteed to be an address that is aligned to an 8-byte boundary.
*/
#define FORCE_ALIGNMENT(X)   (((X)+7)&~7)

/*

/*
** Journal files begin with the following magic string.  The data
** was obtained from /dev/random.  It is used only as a sanity check.
**
** Since version 2.8.0, the journal format contains additional sanity
** checking information.  If the power fails while the journal is begin
** written, semi-random garbage data might appear in the journal
** file after power is restored.  If an attempt is then made
** to roll the journal back, the database could be corrupted.  The additional
** sanity checking data is an attempt to discover the garbage in the
** journal and ignore it.
**
** The sanity checking information for the new journal format consists

  ** the page is marked as needSync==0.
  **
  ** 2008-04-14:  When attempting to vacuum a corrupt database file, it
  ** is possible to fail a statement on a database that does not yet exist.
  ** Do not attempt to write if database file has never been opened.
  */
  pPg = pager_lookup(pPager, pgno);
  PAGERTRACE5("PLAYBACK %d page %d hash(%08x) %s\n",
               PAGERID(pPager), pgno, pager_datahash(pPager->pageSize, aData),
               (isMainJrnl?"main-journal":"sub-journal")
  );
  if( (pPager->state>=PAGER_EXCLUSIVE)
   && (pPg==0 || 0==(pPg->flags&PGHDR_NEED_SYNC))
   && (pPager->fd->pMethods)
  ){
    i64 ofst = (pgno-1)*(i64)pPager->pageSize;
    rc = sqlite3OsWrite(pPager->fd, aData, pPager->pageSize, ofst);
    if( pgno>pPager->dbFileSize ){

    sqlite3_free(pPager);
    return ((rc==SQLITE_OK)?SQLITE_NOMEM:rc);
  }
  nExtra = FORCE_ALIGNMENT(nExtra);
  sqlite3PcacheOpen(szPageDflt, nExtra, !memDb,
                    !memDb?pagerStress:0, (void *)pPager, pPager->pPCache);

  PAGERTRACE3("OPEN %d %s\n", FILEHANDLEID(pPager->fd), pPager->zFilename);
  IOTRACE(("OPEN %p %s\n", pPager, pPager->zFilename))

  /* Fill in Pager.zDirectory[] */
  memcpy(pPager->zDirectory, pPager->zFilename, nPathname+1);
  for(i=sqlite3Strlen30(pPager->zDirectory); 
      i>0 && pPager->zDirectory[i-1]!='/'; i--){}
  if( i>0 ) pPager->zDirectory[i-1] = 0;

    ** this is happening, the database may become corrupt.
    */
    pPager->journalHdr = -1;
    pagerUnlockAndRollback(pPager);
  }
  enable_simulated_io_errors();
  sqlite3EndBenignMalloc();
  PAGERTRACE2("CLOSE %d\n", PAGERID(pPager));
  IOTRACE(("CLOSE %p\n", pPager))
  if( pPager->journalOpen ){
    sqlite3OsClose(pPager->jfd);
  }
  sqlite3BitvecDestroy(pPager->pInJournal);
  sqlite3BitvecDestroy(pPager->pAlwaysRollback);
  releaseAllSavepoint(pPager);

        ** SAFE_APPEND property. Because in this case it is not possible 
        ** for garbage data to be appended to the file, the nRec field
        ** is populated with 0xFFFFFFFF when the journal header is written
        ** and never needs to be updated.
        */
        i64 jrnlOff;
        if( pPager->fullSync && 0==(iDc&SQLITE_IOCAP_SEQUENTIAL) ){
          PAGERTRACE2("SYNC journal of %d\n", PAGERID(pPager));
          IOTRACE(("JSYNC %p\n", pPager))
          rc = sqlite3OsSync(pPager->jfd, pPager->sync_flags);
          if( rc!=0 ) return rc;
        }

        jrnlOff = pPager->journalHdr + sizeof(aJournalMagic);
        IOTRACE(("JHDR %p %lld %d\n", pPager, jrnlOff, 4));
        rc = write32bits(pPager->jfd, jrnlOff, pPager->nRec);
        if( rc ) return rc;
      }
      if( 0==(iDc&SQLITE_IOCAP_SEQUENTIAL) ){
        PAGERTRACE2("SYNC journal of %d\n", PAGERID(pPager));
        IOTRACE(("JSYNC %p\n", pPager))
        rc = sqlite3OsSync(pPager->jfd, pPager->sync_flags| 
          (pPager->sync_flags==SQLITE_SYNC_FULL?SQLITE_SYNC_DATAONLY:0)
        );
        if( rc!=0 ) return rc;
      }
      pPager->journalStarted = 1;

    ** make the file smaller (presumably by auto-vacuum code). Do not write
    ** any such pages to the file.
    */
    if( pList->pgno<=pPager->dbSize && 0==(pList->flags&PGHDR_DONT_WRITE) ){
      i64 offset = (pList->pgno-1)*(i64)pPager->pageSize;
      char *pData = CODEC2(pPager, pList->pData, pList->pgno, 6);

      PAGERTRACE4("STORE %d page %d hash(%08x)\n",
                   PAGERID(pPager), pList->pgno, pager_pagehash(pList));
      IOTRACE(("PGOUT %p %d\n", pPager, pList->pgno));
      rc = sqlite3OsWrite(pPager->fd, pData, pPager->pageSize, offset);
      PAGER_INCR(sqlite3_pager_writedb_count);
      PAGER_INCR(pPager->nWrite);
      if( pList->pgno==1 ){
        memcpy(&pPager->dbFileVers, &pData[24], sizeof(pPager->dbFileVers));
      }
      if( pList->pgno>pPager->dbFileSize ){
        pPager->dbFileSize = pList->pgno;
      }
    }
#ifndef NDEBUG
    else{
      PAGERTRACE3("NOSTORE %d page %d\n", PAGERID(pPager), pList->pgno);
    }
#endif
    if( rc ) return rc;
#ifdef SQLITE_CHECK_PAGES
    pList->pageHash = pager_pagehash(pList);
#endif
    pList = pList->pDirty;

static int subjournalPage(PgHdr *pPg){
  int rc;
  void *pData = pPg->pData;
  Pager *pPager = pPg->pPager;
  i64 offset = pPager->stmtNRec*(4+pPager->pageSize);
  char *pData2 = CODEC2(pPager, pData, pPg->pgno, 7);

  PAGERTRACE3("STMT-JOURNAL %d page %d\n", PAGERID(pPager), pPg->pgno);

  assert( pageInJournal(pPg) || pPg->pgno>pPager->dbOrigSize );
  rc = write32bits(pPager->sjfd, offset, pPg->pgno);
  if( rc==SQLITE_OK ){
    rc = sqlite3OsWrite(pPager->sjfd, pData2, pPager->pageSize, offset+4);
  }
  if( rc==SQLITE_OK ){

    }
    if( rc!=SQLITE_OK ){
      pager_error(pPager, rc);
    }
  }

  if( rc==SQLITE_OK ){
    PAGERTRACE3("STRESS %d page %d\n", PAGERID(pPager), pPg->pgno);
    sqlite3PcacheMakeClean(pPg);
  }
  return rc;
}


/*

  PAGER_INCR(pPager->nRead);
  IOTRACE(("PGIN %p %d\n", pPager, pgno));
  if( pgno==1 ){
    memcpy(&pPager->dbFileVers, &((u8*)pPg->pData)[24],
                                              sizeof(pPager->dbFileVers));
  }
  CODEC1(pPager, pPg->pData, pPg->pgno, 3);
  PAGERTRACE4("FETCH %d page %d hash(%08x)\n",
               PAGERID(pPager), pPg->pgno, pager_pagehash(pPg));
  return rc;
}


/*
** This function is called to obtain the shared lock required before
** data may be read from the pager cache. If the shared lock has already

        rc = pager_wait_on_lock(pPager, EXCLUSIVE_LOCK);
      }
    }
    if( rc!=SQLITE_OK ){
      return rc;
    }
    pPager->dirtyCache = 0;
    PAGERTRACE2("TRANSACTION %d\n", PAGERID(pPager));
    if( pPager->useJournal && !pPager->tempFile
           && pPager->journalMode!=PAGER_JOURNALMODE_OFF ){
      rc = pager_open_journal(pPager);
    }
  }else if( pPager->journalOpen && pPager->journalOff==0 ){
    /* This happens when the pager was in exclusive-access mode the last
    ** time a (read or write) transaction was successfully concluded

        if( rc==SQLITE_OK ){
          rc = write32bits(pPager->jfd, pPager->journalOff, cksum);
          pPager->journalOff += 4;
        }
        IOTRACE(("JOUT %p %d %lld %d\n", pPager, pPg->pgno, 
                 pPager->journalOff, pPager->pageSize));
        PAGER_INCR(sqlite3_pager_writej_count);
        PAGERTRACE5("JOURNAL %d page %d needSync=%d hash(%08x)\n",
             PAGERID(pPager), pPg->pgno, 
             ((pPg->flags&PGHDR_NEED_SYNC)?1:0), pager_pagehash(pPg));

        /* Even if an IO or diskfull error occurred while journalling the
        ** page in the block above, set the need-sync flag for the page.
        ** Otherwise, when the transaction is rolled back, the logic in
        ** playback_one_page() will think that the page needs to be restored
        ** in the database file. And if an IO error occurs while doing so,
        ** then corruption may follow.

          return rc;
        }
      }else{
        if( !pPager->journalStarted && !pPager->noSync ){
          pPg->flags |= PGHDR_NEED_SYNC;
          pPager->needSync = 1;
        }
        PAGERTRACE4("APPEND %d page %d needSync=%d\n",
                PAGERID(pPager), pPg->pgno,
               ((pPg->flags&PGHDR_NEED_SYNC)?1:0));
      }
    }
  
    /* If the statement journal is open and the page is not in it,
    ** then write the current page to the statement journal.  Note that
    ** the statement journal format differs from the standard journal format
    ** in that it omits the checksums and the header.

      ** When the database file grows, we must make sure that the last page
      ** gets written at least once so that the disk file will be the correct
      ** size. If you do not write this page and the size of the file
      ** on the disk ends up being too small, that can lead to database
      ** corruption during the next transaction.
      */
    }else{
      PAGERTRACE3("DONT_WRITE page %d of %d\n", pPg->pgno, PAGERID(pPager));
      IOTRACE(("CLEAN %p %d\n", pPager, pPg->pgno))
      pPg->flags |= PGHDR_DONT_WRITE;
#ifdef SQLITE_CHECK_PAGES
      pPg->pageHash = pager_pagehash(pPg);
#endif
    }
  }

  TESTONLY( rc = ) sqlite3BitvecSet(pPager->pInJournal, pPg->pgno);
  testcase( rc==SQLITE_NOMEM );
  TESTONLY( rc = ) addToSavepointBitvecs(pPager, pPg->pgno);
  testcase( rc==SQLITE_NOMEM );
  sqlite3EndBenignMalloc();


  PAGERTRACE3("DONT_ROLLBACK page %d of %d\n", pPg->pgno, PAGERID(pPager));
  IOTRACE(("GARBAGE %p %d\n", pPager, pPg->pgno))
}


/*
** This routine is called to increment the database file change-counter,
** stored at byte 24 of the pager file.

  if( pPager->dbModified==0 &&
        (pPager->journalMode!=PAGER_JOURNALMODE_DELETE ||
          pPager->exclusiveMode!=0) ){
    assert( pPager->dirtyCache==0 || pPager->journalOpen==0 );
    return SQLITE_OK;
  }

  PAGERTRACE4("DATABASE SYNC: File=%s zMaster=%s nSize=%d\n", 
      pPager->zFilename, zMaster, pPager->dbSize);

  /* If this is an in-memory db, or no pages have been written to, or this
  ** function has already been called, it is a no-op.
  */
  if( pPager->state!=PAGER_SYNCED && !MEMDB && pPager->dirtyCache ){
    PgHdr *pPg;

          ** being discarded by the truncation must be written to the journal
          ** file.
          */
          Pgno i;
          Pgno iSkip = PAGER_MJ_PGNO(pPager);
          Pgno dbSize = pPager->dbSize;
          pPager->dbSize = pPager->dbOrigSize;
          for( i=dbSize+1; i<=pPager->dbOrigSize; i++ ){
            if( !sqlite3BitvecTest(pPager->pInJournal, i) && i!=iSkip ){
              rc = sqlite3PagerGet(pPager, i, &pPg);
              if( rc!=SQLITE_OK ) goto sync_exit;
              rc = sqlite3PagerWrite(pPg);
              sqlite3PagerUnref(pPg);
              if( rc!=SQLITE_OK ) goto sync_exit;
            }

  }
  if( pPager->dbModified==0 &&
        (pPager->journalMode!=PAGER_JOURNALMODE_DELETE ||
          pPager->exclusiveMode!=0) ){
    assert( pPager->dirtyCache==0 || pPager->journalOpen==0 );
    return SQLITE_OK;
  }
  PAGERTRACE2("COMMIT %d\n", PAGERID(pPager));
  assert( pPager->state==PAGER_SYNCED || MEMDB || !pPager->dirtyCache );
  rc = pager_end_transaction(pPager, pPager->setMaster);
  rc = pager_error(pPager, rc);
  return rc;
}

/*
** Rollback all changes.  The database falls back to PAGER_SHARED mode.
** All in-memory cache pages revert to their original data contents.
** The journal is deleted.
**
** This routine cannot fail unless some other process is not following
** the correct locking protocol or unless some other
** process is writing trash into the journal file (SQLITE_CORRUPT) or
** unless a prior malloc() failed (SQLITE_NOMEM).  Appropriate error
** codes are returned for all these occasions.  Otherwise,
** SQLITE_OK is returned.
*/
int sqlite3PagerRollback(Pager *pPager){
  int rc = SQLITE_OK;
  PAGERTRACE2("ROLLBACK %d\n", PAGERID(pPager));
  if( !pPager->dirtyCache || !pPager->journalOpen ){
    rc = pager_end_transaction(pPager, pPager->setMaster);
  }else if( pPager->errCode && pPager->errCode!=SQLITE_FULL ){
    if( pPager->state>=PAGER_EXCLUSIVE ){
      pager_playback(pPager, 0);
    }
    rc = pPager->errCode;

*/
int sqlite3PagerMovepage(Pager *pPager, DbPage *pPg, Pgno pgno, int isCommit){
  PgHdr *pPgOld;  /* The page being overwritten. */
  Pgno needSyncPgno = 0;

  assert( pPg->nRef>0 );

  PAGERTRACE5("MOVE %d page %d (needSync=%d) moves to %d\n", 
      PAGERID(pPager), pPg->pgno, (pPg->flags&PGHDR_NEED_SYNC)?1:0, pgno);
  IOTRACE(("MOVE %p %d %d\n", pPager, pPg->pgno, pgno))

  pager_get_content(pPg);

  /* If the journal needs to be sync()ed before page pPg->pgno can
  ** be written to, store pPg->pgno in local variable needSyncPgno.
  **