SQLite

Check-in [eb3ac895bd]
Login

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

Overview
Comment:Add variable pager.dbHintSize, used to limit the number of calls made to the xFileControl(FCNTL_SIZE_HINT) method.
Downloads: Tarball | ZIP archive
Timelines: family | ancestors | descendants | both | experimental
Files: files | file ages | folders
SHA1: eb3ac895bd92c880ac6acca58de8cf9643c2b5e4
User & Date: dan 2010-08-10 09:58:57.000
Context
2010-08-10
15:46
Fix a bug in pager.c causing it to omit the xSync() call required following a hot-journal rollback. (Closed-Leaf check-in: 42ba43ac69 user: dan tags: experimental)
09:58
Add variable pager.dbHintSize, used to limit the number of calls made to the xFileControl(FCNTL_SIZE_HINT) method. (check-in: eb3ac895bd user: dan tags: experimental)
2010-08-09
19:17
Modify some comments in pager.c. (check-in: 5662da6d4f user: dan tags: experimental)
Changes
Unified Diff Ignore Whitespace Patch
Changes to src/pager.c.
561
562
563
564
565
566
567
568























569
570
571
572
573
574
575
**   variable at the start of the transaction. It is used during rollback,
**   and to determine whether or not pages need to be journalled before
**   being modified.
**
**   Throughout a write-transaction, dbFileSize contains the size of
**   the file on disk in pages. It is set to a copy of dbSize when the
**   write-transaction is first opened, and updated when VFS calls are made
**   to write or truncate the database file on disk.























**
** errCode
**
**   The Pager.errCode variable is only ever used in PAGER_ERROR state. It
**   is set to zero in all other states. In PAGER_ERROR state, Pager.errCode 
**   is always set to SQLITE_FULL, SQLITE_IOERR or one of the SQLITE_IOERR_XXX 
**   sub-codes.







|
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>







561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
**   variable at the start of the transaction. It is used during rollback,
**   and to determine whether or not pages need to be journalled before
**   being modified.
**
**   Throughout a write-transaction, dbFileSize contains the size of
**   the file on disk in pages. It is set to a copy of dbSize when the
**   write-transaction is first opened, and updated when VFS calls are made
**   to write or truncate the database file on disk. 
**
**   The only reason the dbFileSize variable is required is to suppress 
**   unnecessary calls to xTruncate() after committing a transaction. If, 
**   when a transaction is committed, the dbFileSize variable indicates 
**   that the database file is larger than the database image (Pager.dbSize), 
**   pager_truncate() is called. The pager_truncate() call uses xFilesize()
**   to measure the database file on disk, and then truncates it if required.
**   dbFileSize is not used when rolling back a transaction. In this case
**   pager_truncate() is called unconditionally (which means there may be
**   a call to xFilesize() that is not strictly required). In either case,
**   pager_truncate() may cause the file to become smaller or larger.
**
** dbHintSize
**
**   The dbHintSize variable is used to limit the number of calls made to
**   the VFS xFileControl(FCNTL_SIZE_HINT) method. 
**
**   dbHintSize is set to a copy of the dbSize variable when a
**   write-transaction is opened (at the same time as dbFileSize and
**   dbOrigSize). If the xFileControl(FCNTL_SIZE_HINT) method is called,
**   dbHintSize is increased to the number of pages that correspond to the
**   size-hint passed to the method call. See pager_write_pagelist() for 
**   details.
**
** errCode
**
**   The Pager.errCode variable is only ever used in PAGER_ERROR state. It
**   is set to zero in all other states. In PAGER_ERROR state, Pager.errCode 
**   is always set to SQLITE_FULL, SQLITE_IOERR or one of the SQLITE_IOERR_XXX 
**   sub-codes.
615
616
617
618
619
620
621

622
623
624
625
626
627
628
  u8 setMaster;               /* True if a m-j name has been written to jrnl */
  u8 doNotSpill;              /* Do not spill the cache when non-zero */
  u8 doNotSyncSpill;          /* Do not do a spill that requires jrnl sync */
  u8 subjInMemory;            /* True to use in-memory sub-journals */
  Pgno dbSize;                /* Number of pages in the database */
  Pgno dbOrigSize;            /* dbSize before the current transaction */
  Pgno dbFileSize;            /* Number of pages in the database file */

  int errCode;                /* One of several kinds of errors */
  int nRec;                   /* Pages journalled since last j-header written */
  u32 cksumInit;              /* Quasi-random value added to every checksum */
  u32 nSubRec;                /* Number of records written to sub-journal */
  Bitvec *pInJournal;         /* One bit for each page in the database file */
  sqlite3_file *fd;           /* File descriptor for database */
  sqlite3_file *jfd;          /* File descriptor for main journal */







>







638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
  u8 setMaster;               /* True if a m-j name has been written to jrnl */
  u8 doNotSpill;              /* Do not spill the cache when non-zero */
  u8 doNotSyncSpill;          /* Do not do a spill that requires jrnl sync */
  u8 subjInMemory;            /* True to use in-memory sub-journals */
  Pgno dbSize;                /* Number of pages in the database */
  Pgno dbOrigSize;            /* dbSize before the current transaction */
  Pgno dbFileSize;            /* Number of pages in the database file */
  Pgno dbHintSize;            /* Value passed to FCNTL_SIZE_HINT call */
  int errCode;                /* One of several kinds of errors */
  int nRec;                   /* Pages journalled since last j-header written */
  u32 cksumInit;              /* Quasi-random value added to every checksum */
  u32 nSubRec;                /* Number of records written to sub-journal */
  Bitvec *pInJournal;         /* One bit for each page in the database file */
  sqlite3_file *fd;           /* File descriptor for database */
  sqlite3_file *jfd;          /* File descriptor for main journal */
827
828
829
830
831
832
833

834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852

853
854
855
856
857
858
859
860
861
862
863

864
865
866
867
868
869
870
      assert( p->eLock!=UNKNOWN_LOCK );
      assert( pPager->errCode==SQLITE_OK );
      if( !pagerUseWal(pPager) ){
        assert( p->eLock>=RESERVED_LOCK );
      }
      assert( pPager->dbSize==pPager->dbOrigSize );
      assert( pPager->dbOrigSize==pPager->dbFileSize );

      assert( pPager->setMaster==0 );
      break;

    case PAGER_WRITER_CACHEMOD:
      assert( p->eLock!=UNKNOWN_LOCK );
      assert( pPager->errCode==SQLITE_OK );
      if( !pagerUseWal(pPager) ){
        /* It is possible that if journal_mode=wal here that neither the
        ** journal file nor the WAL file are open. This happens during
        ** a rollback transaction that switches from journal_mode=off
        ** to journal_mode=wal.
        */
        assert( p->eLock>=RESERVED_LOCK );
        assert( isOpen(p->jfd) 
             || p->journalMode==PAGER_JOURNALMODE_OFF 
             || p->journalMode==PAGER_JOURNALMODE_WAL 
        );
      }
      assert( pPager->dbOrigSize==pPager->dbFileSize );

      break;

    case PAGER_WRITER_DBMOD:
      assert( p->eLock==EXCLUSIVE_LOCK );
      assert( pPager->errCode==SQLITE_OK );
      assert( !pagerUseWal(pPager) );
      assert( p->eLock>=EXCLUSIVE_LOCK );
      assert( isOpen(p->jfd) 
           || p->journalMode==PAGER_JOURNALMODE_OFF 
           || p->journalMode==PAGER_JOURNALMODE_WAL 
      );

      break;

    case PAGER_WRITER_FINISHED:
      assert( p->eLock==EXCLUSIVE_LOCK );
      assert( pPager->errCode==SQLITE_OK );
      assert( !pagerUseWal(pPager) );
      assert( isOpen(p->jfd) 







>



















>











>







851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
      assert( p->eLock!=UNKNOWN_LOCK );
      assert( pPager->errCode==SQLITE_OK );
      if( !pagerUseWal(pPager) ){
        assert( p->eLock>=RESERVED_LOCK );
      }
      assert( pPager->dbSize==pPager->dbOrigSize );
      assert( pPager->dbOrigSize==pPager->dbFileSize );
      assert( pPager->dbOrigSize==pPager->dbHintSize );
      assert( pPager->setMaster==0 );
      break;

    case PAGER_WRITER_CACHEMOD:
      assert( p->eLock!=UNKNOWN_LOCK );
      assert( pPager->errCode==SQLITE_OK );
      if( !pagerUseWal(pPager) ){
        /* It is possible that if journal_mode=wal here that neither the
        ** journal file nor the WAL file are open. This happens during
        ** a rollback transaction that switches from journal_mode=off
        ** to journal_mode=wal.
        */
        assert( p->eLock>=RESERVED_LOCK );
        assert( isOpen(p->jfd) 
             || p->journalMode==PAGER_JOURNALMODE_OFF 
             || p->journalMode==PAGER_JOURNALMODE_WAL 
        );
      }
      assert( pPager->dbOrigSize==pPager->dbFileSize );
      assert( pPager->dbOrigSize==pPager->dbHintSize );
      break;

    case PAGER_WRITER_DBMOD:
      assert( p->eLock==EXCLUSIVE_LOCK );
      assert( pPager->errCode==SQLITE_OK );
      assert( !pagerUseWal(pPager) );
      assert( p->eLock>=EXCLUSIVE_LOCK );
      assert( isOpen(p->jfd) 
           || p->journalMode==PAGER_JOURNALMODE_OFF 
           || p->journalMode==PAGER_JOURNALMODE_WAL 
      );
      assert( pPager->dbOrigSize<=pPager->dbHintSize );
      break;

    case PAGER_WRITER_FINISHED:
      assert( p->eLock==EXCLUSIVE_LOCK );
      assert( pPager->errCode==SQLITE_OK );
      assert( !pagerUseWal(pPager) );
      assert( isOpen(p->jfd) 
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
3841
3842
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862
3863
3864
3865
3866
3867
3868
3869
3870
3871
3872
3873
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883

3884
3885
3886
3887
3888
3889
3890
** the database file.
**
** If everything is successful, SQLITE_OK is returned. If an IO error 
** occurs, an IO error code is returned. Or, if the EXCLUSIVE lock cannot
** be obtained, SQLITE_BUSY is returned.
*/
static int pager_write_pagelist(Pager *pPager, PgHdr *pList){
  int rc;                              /* Return code */

  /* Normally, this function is called in WRITER_DBMOD state.
  **
  ** However it may be called in WRITER_CACHEMOD state if the page being
  ** written (and all other pages that reside on the same disk sector) was
  ** a free-list leaf page at the start of the transaction. In that case
  ** the database file is not really being modified, so it is Ok to write
  ** to it in CACHEMOD state.
  */
  assert( !pagerUseWal(pPager) );
  assert( pPager->eState==PAGER_WRITER_DBMOD
       || pPager->eState==PAGER_WRITER_CACHEMOD 
  );
  assert( pPager->eState==PAGER_WRITER_DBMOD 
       || (pList->pDirty==0 && pList->pgno<=pPager->dbFileSize)
  );

  /* At this point there may be either a RESERVED or EXCLUSIVE lock on the
  ** database file. If there is already an EXCLUSIVE lock, the following
  ** call is a no-op.
  **
  ** Moving the lock from RESERVED to EXCLUSIVE actually involves going
  ** through an intermediate state PENDING.   A PENDING lock prevents new
  ** readers from attaching to the database but is unsufficient for us to
  ** write.  The idea of a PENDING lock is to prevent new readers from
  ** coming in while we wait for existing readers to clear.
  **
  ** While the pager is in the RESERVED state, the original database file
  ** is unchanged and we can rollback without having to playback the
  ** journal into the original database file.  Once we transition to
  ** EXCLUSIVE, it means the database file has been changed and any rollback
  ** will require a journal playback.
  */
  rc = pager_wait_on_lock(pPager, EXCLUSIVE_LOCK);

  /* If the file is a temp-file has not yet been opened, open it now. It
  ** is not possible for rc to be other than SQLITE_OK if this branch
  ** is taken, as pager_wait_on_lock() is a no-op for temp-files.
  */
  if( !isOpen(pPager->fd) ){
    assert( pPager->tempFile && rc==SQLITE_OK );
    rc = pagerOpentemp(pPager, pPager->fd, pPager->vfsFlags);
  }

  /* Before the first write, give the VFS a hint of what the final
  ** file size will be.
  */
  assert( rc!=SQLITE_OK || isOpen(pPager->fd) );
  if( rc==SQLITE_OK && pPager->dbSize>pPager->dbOrigSize ){
    sqlite3_int64 szFile = pPager->pageSize * (sqlite3_int64)pPager->dbSize;
    sqlite3OsFileControl(pPager->fd, SQLITE_FCNTL_SIZE_HINT, &szFile);

  }

  while( rc==SQLITE_OK && pList ){
    Pgno pgno = pList->pgno;

    /* If there are dirty pages in the page cache with page numbers greater
    ** than Pager.dbSize, this means sqlite3PagerTruncateImage() was called to







|

|
<
<
<
<
<
<
<

|
<
<
|
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<














|


>







3852
3853
3854
3855
3856
3857
3858
3859
3860
3861







3862
3863


3864




















3865
3866
3867
3868
3869
3870
3871
3872
3873
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883
3884
3885
3886
3887
3888
3889
** the database file.
**
** If everything is successful, SQLITE_OK is returned. If an IO error 
** occurs, an IO error code is returned. Or, if the EXCLUSIVE lock cannot
** be obtained, SQLITE_BUSY is returned.
*/
static int pager_write_pagelist(Pager *pPager, PgHdr *pList){
  int rc = SQLITE_OK;                  /* Return code */

  /* This function is only called for rollback pagers in WRITER_DBMOD state. */







  assert( !pagerUseWal(pPager) );
  assert( pPager->eState==PAGER_WRITER_DBMOD );


  assert( pPager->eLock==EXCLUSIVE_LOCK );





















  /* If the file is a temp-file has not yet been opened, open it now. It
  ** is not possible for rc to be other than SQLITE_OK if this branch
  ** is taken, as pager_wait_on_lock() is a no-op for temp-files.
  */
  if( !isOpen(pPager->fd) ){
    assert( pPager->tempFile && rc==SQLITE_OK );
    rc = pagerOpentemp(pPager, pPager->fd, pPager->vfsFlags);
  }

  /* Before the first write, give the VFS a hint of what the final
  ** file size will be.
  */
  assert( rc!=SQLITE_OK || isOpen(pPager->fd) );
  if( rc==SQLITE_OK && pPager->dbSize>pPager->dbHintSize ){
    sqlite3_int64 szFile = pPager->pageSize * (sqlite3_int64)pPager->dbSize;
    sqlite3OsFileControl(pPager->fd, SQLITE_FCNTL_SIZE_HINT, &szFile);
    pPager->dbHintSize = pPager->dbSize;
  }

  while( rc==SQLITE_OK && pList ){
    Pgno pgno = pList->pgno;

    /* If there are dirty pages in the page cache with page numbers greater
    ** than Pager.dbSize, this means sqlite3PagerTruncateImage() was called to
4058
4059
4060
4061
4062
4063
4064
4065


4066
4067
4068
4069
4070
4071
4072
    }
    if( rc==SQLITE_OK ){
      rc = pagerWalFrames(pPager, pPg, 0, 0, 0);
    }
  }else{
  
    /* Sync the journal file if required. */
    if( pPg->flags&PGHDR_NEED_SYNC ){


      rc = syncJournal(pPager, 1);
    }
  
    /* If the page number of this page is larger than the current size of
    ** the database image, it may need to be written to the sub-journal.
    ** This is because the call to pager_write_pagelist() below will not
    ** actually write data to the file in this case.







|
>
>







4057
4058
4059
4060
4061
4062
4063
4064
4065
4066
4067
4068
4069
4070
4071
4072
4073
    }
    if( rc==SQLITE_OK ){
      rc = pagerWalFrames(pPager, pPg, 0, 0, 0);
    }
  }else{
  
    /* Sync the journal file if required. */
    if( pPg->flags&PGHDR_NEED_SYNC 
     || pPager->eState==PAGER_WRITER_CACHEMOD
    ){
      rc = syncJournal(pPager, 1);
    }
  
    /* If the page number of this page is larger than the current size of
    ** the database image, it may need to be written to the sub-journal.
    ** This is because the call to pager_write_pagelist() below will not
    ** actually write data to the file in this case.
5120
5121
5122
5123
5124
5125
5126

5127

5128
5129
5130
5131
5132
5133
5134
      ** when it has an open transaction, but never to DBMOD or FINISHED.
      ** This is because in those states the code to roll back savepoint 
      ** transactions may copy data from the sub-journal into the database 
      ** file as well as into the page cache. Which would be incorrect in 
      ** WAL mode.
      */
      pPager->eState = PAGER_WRITER_LOCKED;

      pPager->dbFileSize = pPager->dbOrigSize = pPager->dbSize;

      pPager->journalOff = 0;
    }

    assert( rc==SQLITE_OK || pPager->eState==PAGER_READER );
    assert( rc!=SQLITE_OK || pPager->eState==PAGER_WRITER_LOCKED );
    assert( assert_pager_state(pPager) );
  }







>
|
>







5121
5122
5123
5124
5125
5126
5127
5128
5129
5130
5131
5132
5133
5134
5135
5136
5137
      ** when it has an open transaction, but never to DBMOD or FINISHED.
      ** This is because in those states the code to roll back savepoint 
      ** transactions may copy data from the sub-journal into the database 
      ** file as well as into the page cache. Which would be incorrect in 
      ** WAL mode.
      */
      pPager->eState = PAGER_WRITER_LOCKED;
      pPager->dbHintSize = pPager->dbSize;
      pPager->dbFileSize = pPager->dbSize;
      pPager->dbOrigSize = pPager->dbSize;
      pPager->journalOff = 0;
    }

    assert( rc==SQLITE_OK || pPager->eState==PAGER_READER );
    assert( rc!=SQLITE_OK || pPager->eState==PAGER_WRITER_LOCKED );
    assert( assert_pager_state(pPager) );
  }