SQLite: Check-in [09be42d5]

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

Overview

Comment:	Begin making experimental changes to use mmap() for reading content from a database. The code compiles, but crashes on the test suite.
Downloads:	Tarball \| ZIP archive \| SQL archive
Timelines:	family \| ancestors \| descendants \| both \| mmap-experimental
Files:	files \| file ages \| folders
SHA1:	09be42d5fa7ef692428579d4d8d48b3316190945
User & Date:	drh 2011-11-14 01:55:02

Context

2011-11-14
12:34		Further work on mmap(). Still does not work right - autovacuum tests are the first to fail. Closed-Leaf check-in: 638a39bb user: drh tags: mmap-experimental
01:55		Begin making experimental changes to use mmap() for reading content from a database. The code compiles, but crashes on the test suite. check-in: 09be42d5 user: drh tags: mmap-experimental
2011-11-13
21:44		Add a version number to the sqlite3_pcache_methods2 object. Other PCACHE2 documentation improvements. Closed-Leaf check-in: 9f839ac0 user: drh tags: experimental-pcache

Changes

Hide Diffs Unified Diffs Ignore Whitespace Patch

Changes to src/btree.c.

  if( pCur->eState!=CURSOR_VALID || pCur->skipNext!=0 ){
    *pHasMoved = 1;
  }else{
    *pHasMoved = 0;
  }
  return SQLITE_OK;
}














#ifndef SQLITE_OMIT_AUTOVACUUM
/*
** Given a page number of a regular database page, return the page
** number for the pointer-map page that contains the entry for the
** input page number.
**
................................................................................
/*
** Get a page from the pager.  Initialize the MemPage.pBt and
** MemPage.aData elements if needed.
**
** If the noContent flag is set, it means that we do not care about
** the content of the page at this time.  So do not go to the disk
** to fetch the content.  Just fill in the content with zeros for now.
** If in the future we call sqlite3PagerWrite() on this page, that
** means we have started to be concerned about content and the disk
** read should occur at that point.
*/
static int btreeGetPage(
  BtShared *pBt,       /* The btree */
  Pgno pgno,           /* Number of the page to fetch */
  MemPage **ppPage,    /* Return the page in this parameter */
................................................................................

  assert( sqlite3_mutex_held(pBt->mutex) );
  if( pBt->nPage>0 ){
    return SQLITE_OK;
  }
  pP1 = pBt->pPage1;
  assert( pP1!=0 );
  rc = sqlite3PagerWrite(pP1->pDbPage);
  if( rc ) return rc;
  data = pP1->aData;
  memcpy(data, zMagicHeader, sizeof(zMagicHeader));
  assert( sizeof(zMagicHeader)==16 );
  data[16] = (u8)((pBt->pageSize>>8)&0xff);
  data[17] = (u8)((pBt->pageSize>>16)&0xff);
  data[18] = 1;
................................................................................
      /* If the db-size header field is incorrect (as it may be if an old
      ** client has been writing the database file), update it now. Doing
      ** this sooner rather than later means the database size can safely 
      ** re-read the database size from page 1 if a savepoint or transaction
      ** rollback occurs within the transaction.
      */
      if( pBt->nPage!=get4byte(&pPage1->aData[28]) ){
        rc = sqlite3PagerWrite(pPage1->pDbPage);
        if( rc==SQLITE_OK ){
          put4byte(&pPage1->aData[28], pBt->nPage);
        }
      }
    }
  }

................................................................................
  ** iPtrPage.
  */
  if( eType!=PTRMAP_ROOTPAGE ){
    rc = btreeGetPage(pBt, iPtrPage, &pPtrPage, 0);
    if( rc!=SQLITE_OK ){
      return rc;
    }
    rc = sqlite3PagerWrite(pPtrPage->pDbPage);
    if( rc!=SQLITE_OK ){
      releasePage(pPtrPage);
      return rc;
    }
    rc = modifyPagePointer(pPtrPage, iDbPage, iFreePage, eType);
    releasePage(pPtrPage);
    if( rc==SQLITE_OK ){
................................................................................
          releasePage(pLastPg);
          return rc;
        }
        releasePage(pFreePg);
      }while( nFin!=0 && iFreePg>nFin );
      assert( iFreePg<iLastPg );
      
      rc = sqlite3PagerWrite(pLastPg->pDbPage);
      if( rc==SQLITE_OK ){
        rc = relocatePage(pBt, pLastPg, eType, iPtrPage, iFreePg, nFin!=0);
      }
      releasePage(pLastPg);
      if( rc!=SQLITE_OK ){
        return rc;
      }
................................................................................
    while( iLastPg==PENDING_BYTE_PAGE(pBt)||PTRMAP_ISPAGE(pBt, iLastPg) ){
      if( PTRMAP_ISPAGE(pBt, iLastPg) ){
        MemPage *pPg;
        rc = btreeGetPage(pBt, iLastPg, &pPg, 0);
        if( rc!=SQLITE_OK ){
          return rc;
        }
        rc = sqlite3PagerWrite(pPg->pDbPage);
        releasePage(pPg);
        if( rc!=SQLITE_OK ){
          return rc;
        }
      }
      iLastPg--;
    }
................................................................................
  assert( pBt->inTransaction==TRANS_WRITE && p->inTrans==TRANS_WRITE );
  if( !pBt->autoVacuum ){
    rc = SQLITE_DONE;
  }else{
    invalidateAllOverflowCache(pBt);
    rc = incrVacuumStep(pBt, 0, btreePagecount(pBt));
    if( rc==SQLITE_OK ){
      rc = sqlite3PagerWrite(pBt->pPage1->pDbPage);
      put4byte(&pBt->pPage1->aData[28], pBt->nPage);
    }
  }
  sqlite3BtreeLeave(p);
  return rc;
}

................................................................................
    }
    if( nFin>nOrig ) return SQLITE_CORRUPT_BKPT;

    for(iFree=nOrig; iFree>nFin && rc==SQLITE_OK; iFree--){
      rc = incrVacuumStep(pBt, nFin, iFree);
    }
    if( (rc==SQLITE_DONE || rc==SQLITE_OK) && nFree>0 ){
      rc = sqlite3PagerWrite(pBt->pPage1->pDbPage);
      put4byte(&pBt->pPage1->aData[32], 0);
      put4byte(&pBt->pPage1->aData[36], 0);
      put4byte(&pBt->pPage1->aData[28], nFin);
      sqlite3PagerTruncateImage(pBt->pPager, nFin);
      pBt->nPage = nFin;
    }
    if( rc!=SQLITE_OK ){
................................................................................
  /* Check if data must be read/written to/from the btree page itself. */
  if( offset<pCur->info.nLocal ){
    int a = amt;
    if( a+offset>pCur->info.nLocal ){
      a = pCur->info.nLocal - offset;
    }
    if( eOp ){
      if( (rc = sqlite3PagerWrite(pPage->pDbPage))!=SQLITE_OK ) return rc;
      getCellInfo(pCur);
      aPayload = pCur->info.pCell + pCur->info.nHeader;
      memcpy(aPayload+offset, pBuf, a);
    }else{
      memcpy(pBuf, aPayload+offset, a);
    }
    offset = 0;
................................................................................
  *pRes = 0;
  return rc;
}

/*
** Allocate a new page from the database file.
**
** The new page is marked as dirty.  (In other words, sqlite3PagerWrite()
** has already been called on the new page.)  The new page has also
** been referenced and the calling routine is responsible for calling
** sqlite3PagerUnref() on the new page when it is done.
**
** SQLITE_OK is returned on success.  Any other return value indicates
** an error.  *ppPage and *pPgno are undefined in the event of an error.
** Do not invoke sqlite3PagerUnref() on *ppPage if an error is returned.
................................................................................
      *pPgno = nearby;
    }
#endif

    /* Decrement the free-list count by 1. Set iTrunk to the index of the
    ** first free-list trunk page. iPrevTrunk is initially 1.
    */
    rc = sqlite3PagerWrite(pPage1->pDbPage);
    if( rc ) return rc;
    put4byte(&pPage1->aData[36], n-1);

    /* The code within this loop is run only once if the 'searchList' variable
    ** is not true. Otherwise, it runs once for each trunk-page on the
    ** free-list until the page 'nearby' is located.
    */
................................................................................

      k = get4byte(&pTrunk->aData[4]); /* # of leaves on this trunk page */
      if( k==0 && !searchList ){
        /* The trunk has no leaves and the list is not being searched. 
        ** So extract the trunk page itself and use it as the newly 
        ** allocated page */
        assert( pPrevTrunk==0 );
        rc = sqlite3PagerWrite(pTrunk->pDbPage);
        if( rc ){
          goto end_allocate_page;
        }
        *pPgno = iTrunk;
        memcpy(&pPage1->aData[32], &pTrunk->aData[0], 4);
        *ppPage = pTrunk;
        pTrunk = 0;
................................................................................
      }else if( searchList && nearby==iTrunk ){
        /* The list is being searched and this trunk page is the page
        ** to allocate, regardless of whether it has leaves.
        */
        assert( *pPgno==iTrunk );
        *ppPage = pTrunk;
        searchList = 0;
        rc = sqlite3PagerWrite(pTrunk->pDbPage);
        if( rc ){
          goto end_allocate_page;
        }
        if( k==0 ){
          if( !pPrevTrunk ){
            memcpy(&pPage1->aData[32], &pTrunk->aData[0], 4);
          }else{
            rc = sqlite3PagerWrite(pPrevTrunk->pDbPage);
            if( rc!=SQLITE_OK ){
              goto end_allocate_page;
            }
            memcpy(&pPrevTrunk->aData[0], &pTrunk->aData[0], 4);
          }
        }else{
          /* The trunk page is required by the caller but it contains 
................................................................................
            goto end_allocate_page;
          }
          testcase( iNewTrunk==mxPage );
          rc = btreeGetPage(pBt, iNewTrunk, &pNewTrunk, 0);
          if( rc!=SQLITE_OK ){
            goto end_allocate_page;
          }
          rc = sqlite3PagerWrite(pNewTrunk->pDbPage);
          if( rc!=SQLITE_OK ){
            releasePage(pNewTrunk);
            goto end_allocate_page;
          }
          memcpy(&pNewTrunk->aData[0], &pTrunk->aData[0], 4);
          put4byte(&pNewTrunk->aData[4], k-1);
          memcpy(&pNewTrunk->aData[8], &pTrunk->aData[12], (k-1)*4);
          releasePage(pNewTrunk);
          if( !pPrevTrunk ){
            assert( sqlite3PagerIswriteable(pPage1->pDbPage) );
            put4byte(&pPage1->aData[32], iNewTrunk);
          }else{
            rc = sqlite3PagerWrite(pPrevTrunk->pDbPage);
            if( rc ){
              goto end_allocate_page;
            }
            put4byte(&pPrevTrunk->aData[0], iNewTrunk);
          }
        }
        pTrunk = 0;
................................................................................
        testcase( iPage==mxPage );
        if( !searchList || iPage==nearby ){
          int noContent;
          *pPgno = iPage;
          TRACE(("ALLOCATE: %d was leaf %d of %d on trunk %d"
                 ": %d more free pages\n",
                 *pPgno, closest+1, k, pTrunk->pgno, n-1));
          rc = sqlite3PagerWrite(pTrunk->pDbPage);
          if( rc ) goto end_allocate_page;
          if( closest<k-1 ){
            memcpy(&aData[8+closest*4], &aData[4+k*4], 4);
          }
          put4byte(&aData[4], k-1);
          noContent = !btreeGetHasContent(pBt, *pPgno);
          rc = btreeGetPage(pBt, *pPgno, ppPage, noContent);
          if( rc==SQLITE_OK ){
            rc = sqlite3PagerWrite((*ppPage)->pDbPage);
            if( rc!=SQLITE_OK ){
              releasePage(*ppPage);
            }
          }
          searchList = 0;
        }
      }
      releasePage(pPrevTrunk);
      pPrevTrunk = 0;
    }while( searchList );
  }else{
    /* There are no pages on the freelist, so create a new page at the
    ** end of the file */
    rc = sqlite3PagerWrite(pBt->pPage1->pDbPage);
    if( rc ) return rc;
    pBt->nPage++;
    if( pBt->nPage==PENDING_BYTE_PAGE(pBt) ) pBt->nPage++;

#ifndef SQLITE_OMIT_AUTOVACUUM
    if( pBt->autoVacuum && PTRMAP_ISPAGE(pBt, pBt->nPage) ){
      /* If *pPgno refers to a pointer-map page, allocate two new pages
................................................................................
      ** becomes a new pointer-map page, the second is used by the caller.
      */
      MemPage *pPg = 0;
      TRACE(("ALLOCATE: %d from end of file (pointer-map page)\n", pBt->nPage));
      assert( pBt->nPage!=PENDING_BYTE_PAGE(pBt) );
      rc = btreeGetPage(pBt, pBt->nPage, &pPg, 1);
      if( rc==SQLITE_OK ){
        rc = sqlite3PagerWrite(pPg->pDbPage);
        releasePage(pPg);
      }
      if( rc ) return rc;
      pBt->nPage++;
      if( pBt->nPage==PENDING_BYTE_PAGE(pBt) ){ pBt->nPage++; }
    }
#endif
    put4byte(28 + (u8*)pBt->pPage1->aData, pBt->nPage);
    *pPgno = pBt->nPage;

    assert( *pPgno!=PENDING_BYTE_PAGE(pBt) );
    rc = btreeGetPage(pBt, *pPgno, ppPage, 1);
    if( rc ) return rc;
    rc = sqlite3PagerWrite((*ppPage)->pDbPage);
    if( rc!=SQLITE_OK ){
      releasePage(*ppPage);
    }
    TRACE(("ALLOCATE: %d from end of file\n", *pPgno));
  }

  assert( *pPgno!=PENDING_BYTE_PAGE(pBt) );
................................................................................
    pPage = pMemPage;
    sqlite3PagerRef(pPage->pDbPage);
  }else{
    pPage = btreePageLookup(pBt, iPage);
  }

  /* Increment the free page count on pPage1 */
  rc = sqlite3PagerWrite(pPage1->pDbPage);
  if( rc ) goto freepage_out;
  nFree = get4byte(&pPage1->aData[36]);
  put4byte(&pPage1->aData[36], nFree+1);

  if( pBt->secureDelete ){
    /* If the secure_delete option is enabled, then
    ** always fully overwrite deleted information with zeros.
    */
    if( (!pPage && ((rc = btreeGetPage(pBt, iPage, &pPage, 0))!=0) )
     ||            ((rc = sqlite3PagerWrite(pPage->pDbPage))!=0)
    ){
      goto freepage_out;
    }
    memset(pPage->aData, 0, pPage->pBt->pageSize);
  }

  /* If the database supports auto-vacuum, write an entry in the pointer-map
................................................................................
      ** usableSize/4 - 8 entries will be reported as corrupt.  In order
      ** to maintain backwards compatibility with older versions of SQLite,
      ** we will continue to restrict the number of entries to usableSize/4 - 8
      ** for now.  At some point in the future (once everyone has upgraded
      ** to 3.6.0 or later) we should consider fixing the conditional above
      ** to read "usableSize/4-2" instead of "usableSize/4-8".
      */
      rc = sqlite3PagerWrite(pTrunk->pDbPage);
      if( rc==SQLITE_OK ){
        put4byte(&pTrunk->aData[4], nLeaf+1);
        put4byte(&pTrunk->aData[8+nLeaf*4], iPage);
        if( pPage && !pBt->secureDelete ){
          sqlite3PagerDontWrite(pPage->pDbPage);
        }
        rc = btreeSetHasContent(pBt, iPage);
................................................................................
  ** Possibly because the free-list is empty, or possibly because the 
  ** first trunk in the free-list is full. Either way, the page being freed
  ** will become the new first trunk page in the free-list.
  */
  if( pPage==0 && SQLITE_OK!=(rc = btreeGetPage(pBt, iPage, &pPage, 0)) ){
    goto freepage_out;
  }
  rc = sqlite3PagerWrite(pPage->pDbPage);
  if( rc!=SQLITE_OK ){
    goto freepage_out;
  }
  put4byte(pPage->aData, iTrunk);
  put4byte(&pPage->aData[4], 0);
  put4byte(&pPage1->aData[32], iPage);
  TRACE(("FREE-PAGE: %d new trunk page replacing %d\n", pPage->pgno, iTrunk));
................................................................................
      put4byte(pCell, iChild);
    }
    j = pPage->nOverflow++;
    assert( j<(int)(sizeof(pPage->aOvfl)/sizeof(pPage->aOvfl[0])) );
    pPage->aOvfl[j].pCell = pCell;
    pPage->aOvfl[j].idx = (u16)i;
  }else{
    int rc = sqlite3PagerWrite(pPage->pDbPage);
    if( rc!=SQLITE_OK ){
      *pRC = rc;
      return;
    }
    assert( sqlite3PagerIswriteable(pPage->pDbPage) );
    data = pPage->aData;
    cellOffset = pPage->cellOffset;
................................................................................
  }
  pageFlags = apOld[0]->aData[0];
  for(i=0; i<k; i++){
    MemPage *pNew;
    if( i<nOld ){
      pNew = apNew[i] = apOld[i];
      apOld[i] = 0;
      rc = sqlite3PagerWrite(pNew->pDbPage);
      nNew++;
      if( rc ) goto balance_cleanup;
    }else{
      assert( i>0 );
      rc = allocateBtreePage(pBt, &pNew, &pgno, pgno, 0);
      if( rc ) goto balance_cleanup;
      apNew[i] = pNew;
................................................................................
  assert( pRoot->nOverflow>0 );
  assert( sqlite3_mutex_held(pBt->mutex) );

  /* Make pRoot, the root page of the b-tree, writable. Allocate a new 
  ** page that will become the new right-child of pPage. Copy the contents
  ** of the node stored on pRoot into the new child page.
  */
  rc = sqlite3PagerWrite(pRoot->pDbPage);
  if( rc==SQLITE_OK ){
    rc = allocateBtreePage(pBt,&pChild,&pgnoChild,pRoot->pgno,0);
    copyNodeContent(pRoot, pChild, &rc);
    if( ISAUTOVACUUM ){
      ptrmapPut(pBt, pgnoChild, PTRMAP_BTREE, pRoot->pgno, &rc);
    }
  }
................................................................................
      }
    }else if( pPage->nOverflow==0 && pPage->nFree<=nMin ){
      break;
    }else{
      MemPage * const pParent = pCur->apPage[iPage-1];
      int const iIdx = pCur->aiIdx[iPage-1];

      rc = sqlite3PagerWrite(pParent->pDbPage);
      if( rc==SQLITE_OK ){
#ifndef SQLITE_OMIT_QUICKBALANCE
        if( pPage->hasData
         && pPage->nOverflow==1
         && pPage->aOvfl[0].idx==pPage->nCell
         && pParent->pgno!=1
         && pParent->nCell==iIdx
................................................................................
  if( rc ) goto end_insert;
  assert( szNew==cellSizePtr(pPage, newCell) );
  assert( szNew <= MX_CELL_SIZE(pBt) );
  idx = pCur->aiIdx[pCur->iPage];
  if( loc==0 ){
    u16 szOld;
    assert( idx<pPage->nCell );
    rc = sqlite3PagerWrite(pPage->pDbPage);
    if( rc ){
      goto end_insert;
    }
    oldCell = findCell(pPage, idx);
    if( !pPage->leaf ){
      memcpy(newCell, oldCell, 4);
    }
................................................................................
  /* Save the positions of any other cursors open on this table before
  ** making any modifications. Make the page containing the entry to be 
  ** deleted writable. Then free any overflow pages associated with the 
  ** entry and finally remove the cell itself from within the page.  
  */
  rc = saveAllCursors(pBt, pCur->pgnoRoot, pCur);
  if( rc ) return rc;
  rc = sqlite3PagerWrite(pPage->pDbPage);
  if( rc ) return rc;
  rc = clearCell(pPage, pCell);
  dropCell(pPage, iCellIdx, cellSizePtr(pPage, pCell), &rc);
  if( rc ) return rc;

  /* If the cell deleted was not located on a leaf page, then the cursor
  ** is currently pointing to the largest entry in the sub-tree headed
................................................................................
    pCell = findCell(pLeaf, pLeaf->nCell-1);
    nCell = cellSizePtr(pLeaf, pCell);
    assert( MX_CELL_SIZE(pBt) >= nCell );

    allocateTempSpace(pBt);
    pTmp = pBt->pTmpSpace;

    rc = sqlite3PagerWrite(pLeaf->pDbPage);
    insertCell(pPage, iCellIdx, pCell-4, nCell+4, pTmp, n, &rc);
    dropCell(pLeaf, pLeaf->nCell-1, nCell, &rc);
    if( rc ) return rc;
  }

  /* Balance the tree. If the entry deleted was located on a leaf page,
  ** then the cursor still points to that page. In this case the first
................................................................................
      if( rc!=SQLITE_OK ){
        return rc;
      }
      rc = btreeGetPage(pBt, pgnoRoot, &pRoot, 0);
      if( rc!=SQLITE_OK ){
        return rc;
      }
      rc = sqlite3PagerWrite(pRoot->pDbPage);
      if( rc!=SQLITE_OK ){
        releasePage(pRoot);
        return rc;
      }
    }else{
      pRoot = pPageMove;
    } 
................................................................................
    if( rc ) goto cleardatabasepage_out;
  }else if( pnChange ){
    assert( pPage->intKey );
    *pnChange += pPage->nCell;
  }
  if( freePageFlag ){
    freePage(pPage, &rc);
  }else if( (rc = sqlite3PagerWrite(pPage->pDbPage))==0 ){
    zeroPage(pPage, pPage->aData[0] | PTF_LEAF);
  }

cleardatabasepage_out:
  releasePage(pPage);
  return rc;
}
................................................................................
int sqlite3BtreeUpdateMeta(Btree *p, int idx, u32 iMeta){
  BtShared *pBt = p->pBt;
  int rc;
  assert( idx>=1 && idx<=15 );
  sqlite3BtreeEnter(p);
  assert( p->inTrans==TRANS_WRITE );
  assert( pBt->pPage1!=0 );
  rc = sqlite3PagerWrite(pBt->pPage1->pDbPage);
  if( rc==SQLITE_OK ){
    put4byte(&pBt->pPage1->aData[36 + idx*4], iMeta);
#ifndef SQLITE_OMIT_AUTOVACUUM
    if( idx==BTREE_INCR_VACUUM ){
      assert( pBt->autoVacuum || iMeta==0 );
      assert( iMeta==0 || iMeta==1 );
      pBt->incrVacuum = (u8)iMeta;
................................................................................

  rc = sqlite3BtreeBeginTrans(pBtree, 0);
  if( rc==SQLITE_OK ){
    u8 *aData = pBt->pPage1->aData;
    if( aData[18]!=(u8)iVersion || aData[19]!=(u8)iVersion ){
      rc = sqlite3BtreeBeginTrans(pBtree, 2);
      if( rc==SQLITE_OK ){
        rc = sqlite3PagerWrite(pBt->pPage1->pDbPage);
        if( rc==SQLITE_OK ){
          aData = pBt->pPage1->aData;
          aData[18] = (u8)iVersion;
          aData[19] = (u8)iVersion;
        }
      }
    }
  }

  pBt->doNotUseWAL = 0;
  return rc;
}

  if( pCur->eState!=CURSOR_VALID || pCur->skipNext!=0 ){
    *pHasMoved = 1;
  }else{
    *pHasMoved = 0;
  }
  return SQLITE_OK;
}

/*
** Make a btree page is writable.
*/
static int btreeMakePageWriteable(MemPage *pPage){
  int rc;
  if( sqlite3PagerIswriteable(pPage->pDbPage) ) return SQLITE_OK;
  rc = sqlite3PagerWrite(pPage->pDbPage);
  pPage->aData = sqlite3PagerGetData(pPage->pDbPage);
  pPage->aDataEnd = &pPage->aData[pPage->pBt->usableSize];
  pPage->aCellIdx = &pPage->aData[pPage->cellOffset];
  return rc;
}

#ifndef SQLITE_OMIT_AUTOVACUUM
/*
** Given a page number of a regular database page, return the page
** number for the pointer-map page that contains the entry for the
** input page number.
**
................................................................................
/*
** Get a page from the pager.  Initialize the MemPage.pBt and
** MemPage.aData elements if needed.
**
** If the noContent flag is set, it means that we do not care about
** the content of the page at this time.  So do not go to the disk
** to fetch the content.  Just fill in the content with zeros for now.
** If in the future we call btreeMakePageWriteable() on this page, that
** means we have started to be concerned about content and the disk
** read should occur at that point.
*/
static int btreeGetPage(
  BtShared *pBt,       /* The btree */
  Pgno pgno,           /* Number of the page to fetch */
  MemPage **ppPage,    /* Return the page in this parameter */
................................................................................

  assert( sqlite3_mutex_held(pBt->mutex) );
  if( pBt->nPage>0 ){
    return SQLITE_OK;
  }
  pP1 = pBt->pPage1;
  assert( pP1!=0 );
  rc = btreeMakePageWriteable(pP1);
  if( rc ) return rc;
  data = pP1->aData;
  memcpy(data, zMagicHeader, sizeof(zMagicHeader));
  assert( sizeof(zMagicHeader)==16 );
  data[16] = (u8)((pBt->pageSize>>8)&0xff);
  data[17] = (u8)((pBt->pageSize>>16)&0xff);
  data[18] = 1;
................................................................................
      /* If the db-size header field is incorrect (as it may be if an old
      ** client has been writing the database file), update it now. Doing
      ** this sooner rather than later means the database size can safely 
      ** re-read the database size from page 1 if a savepoint or transaction
      ** rollback occurs within the transaction.
      */
      if( pBt->nPage!=get4byte(&pPage1->aData[28]) ){
        rc = btreeMakePageWriteable(pPage1);
        if( rc==SQLITE_OK ){
          put4byte(&pPage1->aData[28], pBt->nPage);
        }
      }
    }
  }

................................................................................
  ** iPtrPage.
  */
  if( eType!=PTRMAP_ROOTPAGE ){
    rc = btreeGetPage(pBt, iPtrPage, &pPtrPage, 0);
    if( rc!=SQLITE_OK ){
      return rc;
    }
    rc = btreeMakePageWriteable(pPtrPage);
    if( rc!=SQLITE_OK ){
      releasePage(pPtrPage);
      return rc;
    }
    rc = modifyPagePointer(pPtrPage, iDbPage, iFreePage, eType);
    releasePage(pPtrPage);
    if( rc==SQLITE_OK ){
................................................................................
          releasePage(pLastPg);
          return rc;
        }
        releasePage(pFreePg);
      }while( nFin!=0 && iFreePg>nFin );
      assert( iFreePg<iLastPg );
      
      rc = btreeMakePageWriteable(pLastPg);
      if( rc==SQLITE_OK ){
        rc = relocatePage(pBt, pLastPg, eType, iPtrPage, iFreePg, nFin!=0);
      }
      releasePage(pLastPg);
      if( rc!=SQLITE_OK ){
        return rc;
      }
................................................................................
    while( iLastPg==PENDING_BYTE_PAGE(pBt)||PTRMAP_ISPAGE(pBt, iLastPg) ){
      if( PTRMAP_ISPAGE(pBt, iLastPg) ){
        MemPage *pPg;
        rc = btreeGetPage(pBt, iLastPg, &pPg, 0);
        if( rc!=SQLITE_OK ){
          return rc;
        }
        rc = btreeMakePageWriteable(pPg);
        releasePage(pPg);
        if( rc!=SQLITE_OK ){
          return rc;
        }
      }
      iLastPg--;
    }
................................................................................
  assert( pBt->inTransaction==TRANS_WRITE && p->inTrans==TRANS_WRITE );
  if( !pBt->autoVacuum ){
    rc = SQLITE_DONE;
  }else{
    invalidateAllOverflowCache(pBt);
    rc = incrVacuumStep(pBt, 0, btreePagecount(pBt));
    if( rc==SQLITE_OK ){
      rc = btreeMakePageWriteable(pBt->pPage1);
      put4byte(&pBt->pPage1->aData[28], pBt->nPage);
    }
  }
  sqlite3BtreeLeave(p);
  return rc;
}

................................................................................
    }
    if( nFin>nOrig ) return SQLITE_CORRUPT_BKPT;

    for(iFree=nOrig; iFree>nFin && rc==SQLITE_OK; iFree--){
      rc = incrVacuumStep(pBt, nFin, iFree);
    }
    if( (rc==SQLITE_DONE || rc==SQLITE_OK) && nFree>0 ){
      rc = btreeMakePageWriteable(pBt->pPage1);
      put4byte(&pBt->pPage1->aData[32], 0);
      put4byte(&pBt->pPage1->aData[36], 0);
      put4byte(&pBt->pPage1->aData[28], nFin);
      sqlite3PagerTruncateImage(pBt->pPager, nFin);
      pBt->nPage = nFin;
    }
    if( rc!=SQLITE_OK ){
................................................................................
  /* Check if data must be read/written to/from the btree page itself. */
  if( offset<pCur->info.nLocal ){
    int a = amt;
    if( a+offset>pCur->info.nLocal ){
      a = pCur->info.nLocal - offset;
    }
    if( eOp ){
      if( (rc = btreeMakePageWriteable(pPage))!=SQLITE_OK ) return rc;
      getCellInfo(pCur);
      aPayload = pCur->info.pCell + pCur->info.nHeader;
      memcpy(aPayload+offset, pBuf, a);
    }else{
      memcpy(pBuf, aPayload+offset, a);
    }
    offset = 0;
................................................................................
  *pRes = 0;
  return rc;
}

/*
** Allocate a new page from the database file.
**
** The new page is marked as dirty.  (In other words, btreeMakePageWriteable()
** has already been called on the new page.)  The new page has also
** been referenced and the calling routine is responsible for calling
** sqlite3PagerUnref() on the new page when it is done.
**
** SQLITE_OK is returned on success.  Any other return value indicates
** an error.  *ppPage and *pPgno are undefined in the event of an error.
** Do not invoke sqlite3PagerUnref() on *ppPage if an error is returned.
................................................................................
      *pPgno = nearby;
    }
#endif

    /* Decrement the free-list count by 1. Set iTrunk to the index of the
    ** first free-list trunk page. iPrevTrunk is initially 1.
    */
    rc = btreeMakePageWriteable(pPage1);
    if( rc ) return rc;
    put4byte(&pPage1->aData[36], n-1);

    /* The code within this loop is run only once if the 'searchList' variable
    ** is not true. Otherwise, it runs once for each trunk-page on the
    ** free-list until the page 'nearby' is located.
    */
................................................................................

      k = get4byte(&pTrunk->aData[4]); /* # of leaves on this trunk page */
      if( k==0 && !searchList ){
        /* The trunk has no leaves and the list is not being searched. 
        ** So extract the trunk page itself and use it as the newly 
        ** allocated page */
        assert( pPrevTrunk==0 );
        rc = btreeMakePageWriteable(pTrunk);
        if( rc ){
          goto end_allocate_page;
        }
        *pPgno = iTrunk;
        memcpy(&pPage1->aData[32], &pTrunk->aData[0], 4);
        *ppPage = pTrunk;
        pTrunk = 0;
................................................................................
      }else if( searchList && nearby==iTrunk ){
        /* The list is being searched and this trunk page is the page
        ** to allocate, regardless of whether it has leaves.
        */
        assert( *pPgno==iTrunk );
        *ppPage = pTrunk;
        searchList = 0;
        rc = btreeMakePageWriteable(pTrunk);
        if( rc ){
          goto end_allocate_page;
        }
        if( k==0 ){
          if( !pPrevTrunk ){
            memcpy(&pPage1->aData[32], &pTrunk->aData[0], 4);
          }else{
            rc = btreeMakePageWriteable(pPrevTrunk);
            if( rc!=SQLITE_OK ){
              goto end_allocate_page;
            }
            memcpy(&pPrevTrunk->aData[0], &pTrunk->aData[0], 4);
          }
        }else{
          /* The trunk page is required by the caller but it contains 
................................................................................
            goto end_allocate_page;
          }
          testcase( iNewTrunk==mxPage );
          rc = btreeGetPage(pBt, iNewTrunk, &pNewTrunk, 0);
          if( rc!=SQLITE_OK ){
            goto end_allocate_page;
          }
          rc = btreeMakePageWriteable(pNewTrunk);
          if( rc!=SQLITE_OK ){
            releasePage(pNewTrunk);
            goto end_allocate_page;
          }
          memcpy(&pNewTrunk->aData[0], &pTrunk->aData[0], 4);
          put4byte(&pNewTrunk->aData[4], k-1);
          memcpy(&pNewTrunk->aData[8], &pTrunk->aData[12], (k-1)*4);
          releasePage(pNewTrunk);
          if( !pPrevTrunk ){
            assert( sqlite3PagerIswriteable(pPage1->pDbPage) );
            put4byte(&pPage1->aData[32], iNewTrunk);
          }else{
            rc = btreeMakePageWriteable(pPrevTrunk);
            if( rc ){
              goto end_allocate_page;
            }
            put4byte(&pPrevTrunk->aData[0], iNewTrunk);
          }
        }
        pTrunk = 0;
................................................................................
        testcase( iPage==mxPage );
        if( !searchList || iPage==nearby ){
          int noContent;
          *pPgno = iPage;
          TRACE(("ALLOCATE: %d was leaf %d of %d on trunk %d"
                 ": %d more free pages\n",
                 *pPgno, closest+1, k, pTrunk->pgno, n-1));
          rc = btreeMakePageWriteable(pTrunk);
          if( rc ) goto end_allocate_page;
          if( closest<k-1 ){
            memcpy(&aData[8+closest*4], &aData[4+k*4], 4);
          }
          put4byte(&aData[4], k-1);
          noContent = !btreeGetHasContent(pBt, *pPgno);
          rc = btreeGetPage(pBt, *pPgno, ppPage, noContent);
          if( rc==SQLITE_OK ){
            rc = btreeMakePageWriteable(*ppPage);
            if( rc!=SQLITE_OK ){
              releasePage(*ppPage);
            }
          }
          searchList = 0;
        }
      }
      releasePage(pPrevTrunk);
      pPrevTrunk = 0;
    }while( searchList );
  }else{
    /* There are no pages on the freelist, so create a new page at the
    ** end of the file */
    rc = btreeMakePageWriteable(pBt->pPage1);
    if( rc ) return rc;
    pBt->nPage++;
    if( pBt->nPage==PENDING_BYTE_PAGE(pBt) ) pBt->nPage++;

#ifndef SQLITE_OMIT_AUTOVACUUM
    if( pBt->autoVacuum && PTRMAP_ISPAGE(pBt, pBt->nPage) ){
      /* If *pPgno refers to a pointer-map page, allocate two new pages
................................................................................
      ** becomes a new pointer-map page, the second is used by the caller.
      */
      MemPage *pPg = 0;
      TRACE(("ALLOCATE: %d from end of file (pointer-map page)\n", pBt->nPage));
      assert( pBt->nPage!=PENDING_BYTE_PAGE(pBt) );
      rc = btreeGetPage(pBt, pBt->nPage, &pPg, 1);
      if( rc==SQLITE_OK ){
        rc = btreeMakePageWriteable(pPg);
        releasePage(pPg);
      }
      if( rc ) return rc;
      pBt->nPage++;
      if( pBt->nPage==PENDING_BYTE_PAGE(pBt) ){ pBt->nPage++; }
    }
#endif
    put4byte(28 + (u8*)pBt->pPage1->aData, pBt->nPage);
    *pPgno = pBt->nPage;

    assert( *pPgno!=PENDING_BYTE_PAGE(pBt) );
    rc = btreeGetPage(pBt, *pPgno, ppPage, 1);
    if( rc ) return rc;
    rc = btreeMakePageWriteable(*ppPage);
    if( rc!=SQLITE_OK ){
      releasePage(*ppPage);
    }
    TRACE(("ALLOCATE: %d from end of file\n", *pPgno));
  }

  assert( *pPgno!=PENDING_BYTE_PAGE(pBt) );
................................................................................
    pPage = pMemPage;
    sqlite3PagerRef(pPage->pDbPage);
  }else{
    pPage = btreePageLookup(pBt, iPage);
  }

  /* Increment the free page count on pPage1 */
  rc = btreeMakePageWriteable(pPage1);
  if( rc ) goto freepage_out;
  nFree = get4byte(&pPage1->aData[36]);
  put4byte(&pPage1->aData[36], nFree+1);

  if( pBt->secureDelete ){
    /* If the secure_delete option is enabled, then
    ** always fully overwrite deleted information with zeros.
    */
    if( (!pPage && ((rc = btreeGetPage(pBt, iPage, &pPage, 0))!=0) )
     ||            ((rc = btreeMakePageWriteable(pPage))!=0)
    ){
      goto freepage_out;
    }
    memset(pPage->aData, 0, pPage->pBt->pageSize);
  }

  /* If the database supports auto-vacuum, write an entry in the pointer-map
................................................................................
      ** usableSize/4 - 8 entries will be reported as corrupt.  In order
      ** to maintain backwards compatibility with older versions of SQLite,
      ** we will continue to restrict the number of entries to usableSize/4 - 8
      ** for now.  At some point in the future (once everyone has upgraded
      ** to 3.6.0 or later) we should consider fixing the conditional above
      ** to read "usableSize/4-2" instead of "usableSize/4-8".
      */
      rc = btreeMakePageWriteable(pTrunk);
      if( rc==SQLITE_OK ){
        put4byte(&pTrunk->aData[4], nLeaf+1);
        put4byte(&pTrunk->aData[8+nLeaf*4], iPage);
        if( pPage && !pBt->secureDelete ){
          sqlite3PagerDontWrite(pPage->pDbPage);
        }
        rc = btreeSetHasContent(pBt, iPage);
................................................................................
  ** Possibly because the free-list is empty, or possibly because the 
  ** first trunk in the free-list is full. Either way, the page being freed
  ** will become the new first trunk page in the free-list.
  */
  if( pPage==0 && SQLITE_OK!=(rc = btreeGetPage(pBt, iPage, &pPage, 0)) ){
    goto freepage_out;
  }
  rc = btreeMakePageWriteable(pPage);
  if( rc!=SQLITE_OK ){
    goto freepage_out;
  }
  put4byte(pPage->aData, iTrunk);
  put4byte(&pPage->aData[4], 0);
  put4byte(&pPage1->aData[32], iPage);
  TRACE(("FREE-PAGE: %d new trunk page replacing %d\n", pPage->pgno, iTrunk));
................................................................................
      put4byte(pCell, iChild);
    }
    j = pPage->nOverflow++;
    assert( j<(int)(sizeof(pPage->aOvfl)/sizeof(pPage->aOvfl[0])) );
    pPage->aOvfl[j].pCell = pCell;
    pPage->aOvfl[j].idx = (u16)i;
  }else{
    int rc = btreeMakePageWriteable(pPage);
    if( rc!=SQLITE_OK ){
      *pRC = rc;
      return;
    }
    assert( sqlite3PagerIswriteable(pPage->pDbPage) );
    data = pPage->aData;
    cellOffset = pPage->cellOffset;
................................................................................
  }
  pageFlags = apOld[0]->aData[0];
  for(i=0; i<k; i++){
    MemPage *pNew;
    if( i<nOld ){
      pNew = apNew[i] = apOld[i];
      apOld[i] = 0;
      rc = btreeMakePageWriteable(pNew);
      nNew++;
      if( rc ) goto balance_cleanup;
    }else{
      assert( i>0 );
      rc = allocateBtreePage(pBt, &pNew, &pgno, pgno, 0);
      if( rc ) goto balance_cleanup;
      apNew[i] = pNew;
................................................................................
  assert( pRoot->nOverflow>0 );
  assert( sqlite3_mutex_held(pBt->mutex) );

  /* Make pRoot, the root page of the b-tree, writable. Allocate a new 
  ** page that will become the new right-child of pPage. Copy the contents
  ** of the node stored on pRoot into the new child page.
  */
  rc = btreeMakePageWriteable(pRoot);
  if( rc==SQLITE_OK ){
    rc = allocateBtreePage(pBt,&pChild,&pgnoChild,pRoot->pgno,0);
    copyNodeContent(pRoot, pChild, &rc);
    if( ISAUTOVACUUM ){
      ptrmapPut(pBt, pgnoChild, PTRMAP_BTREE, pRoot->pgno, &rc);
    }
  }
................................................................................
      }
    }else if( pPage->nOverflow==0 && pPage->nFree<=nMin ){
      break;
    }else{
      MemPage * const pParent = pCur->apPage[iPage-1];
      int const iIdx = pCur->aiIdx[iPage-1];

      rc = btreeMakePageWriteable(pParent);
      if( rc==SQLITE_OK ){
#ifndef SQLITE_OMIT_QUICKBALANCE
        if( pPage->hasData
         && pPage->nOverflow==1
         && pPage->aOvfl[0].idx==pPage->nCell
         && pParent->pgno!=1
         && pParent->nCell==iIdx
................................................................................
  if( rc ) goto end_insert;
  assert( szNew==cellSizePtr(pPage, newCell) );
  assert( szNew <= MX_CELL_SIZE(pBt) );
  idx = pCur->aiIdx[pCur->iPage];
  if( loc==0 ){
    u16 szOld;
    assert( idx<pPage->nCell );
    rc = btreeMakePageWriteable(pPage);
    if( rc ){
      goto end_insert;
    }
    oldCell = findCell(pPage, idx);
    if( !pPage->leaf ){
      memcpy(newCell, oldCell, 4);
    }
................................................................................
  /* Save the positions of any other cursors open on this table before
  ** making any modifications. Make the page containing the entry to be 
  ** deleted writable. Then free any overflow pages associated with the 
  ** entry and finally remove the cell itself from within the page.  
  */
  rc = saveAllCursors(pBt, pCur->pgnoRoot, pCur);
  if( rc ) return rc;
  rc = btreeMakePageWriteable(pPage);
  if( rc ) return rc;
  rc = clearCell(pPage, pCell);
  dropCell(pPage, iCellIdx, cellSizePtr(pPage, pCell), &rc);
  if( rc ) return rc;

  /* If the cell deleted was not located on a leaf page, then the cursor
  ** is currently pointing to the largest entry in the sub-tree headed
................................................................................
    pCell = findCell(pLeaf, pLeaf->nCell-1);
    nCell = cellSizePtr(pLeaf, pCell);
    assert( MX_CELL_SIZE(pBt) >= nCell );

    allocateTempSpace(pBt);
    pTmp = pBt->pTmpSpace;

    rc = btreeMakePageWriteable(pLeaf);
    insertCell(pPage, iCellIdx, pCell-4, nCell+4, pTmp, n, &rc);
    dropCell(pLeaf, pLeaf->nCell-1, nCell, &rc);
    if( rc ) return rc;
  }

  /* Balance the tree. If the entry deleted was located on a leaf page,
  ** then the cursor still points to that page. In this case the first
................................................................................
      if( rc!=SQLITE_OK ){
        return rc;
      }
      rc = btreeGetPage(pBt, pgnoRoot, &pRoot, 0);
      if( rc!=SQLITE_OK ){
        return rc;
      }
      rc = btreeMakePageWriteable(pRoot);
      if( rc!=SQLITE_OK ){
        releasePage(pRoot);
        return rc;
      }
    }else{
      pRoot = pPageMove;
    } 
................................................................................
    if( rc ) goto cleardatabasepage_out;
  }else if( pnChange ){
    assert( pPage->intKey );
    *pnChange += pPage->nCell;
  }
  if( freePageFlag ){
    freePage(pPage, &rc);
  }else if( (rc = btreeMakePageWriteable(pPage))==0 ){
    zeroPage(pPage, pPage->aData[0] | PTF_LEAF);
  }

cleardatabasepage_out:
  releasePage(pPage);
  return rc;
}
................................................................................
int sqlite3BtreeUpdateMeta(Btree *p, int idx, u32 iMeta){
  BtShared *pBt = p->pBt;
  int rc;
  assert( idx>=1 && idx<=15 );
  sqlite3BtreeEnter(p);
  assert( p->inTrans==TRANS_WRITE );
  assert( pBt->pPage1!=0 );
  rc = btreeMakePageWriteable(pBt->pPage1);
  if( rc==SQLITE_OK ){
    put4byte(&pBt->pPage1->aData[36 + idx*4], iMeta);
#ifndef SQLITE_OMIT_AUTOVACUUM
    if( idx==BTREE_INCR_VACUUM ){
      assert( pBt->autoVacuum || iMeta==0 );
      assert( iMeta==0 || iMeta==1 );
      pBt->incrVacuum = (u8)iMeta;
................................................................................

  rc = sqlite3BtreeBeginTrans(pBtree, 0);
  if( rc==SQLITE_OK ){
    u8 *aData = pBt->pPage1->aData;
    if( aData[18]!=(u8)iVersion || aData[19]!=(u8)iVersion ){
      rc = sqlite3BtreeBeginTrans(pBtree, 2);
      if( rc==SQLITE_OK ){
        rc = btreeMakePageWriteable(pBt->pPage1);
        if( rc==SQLITE_OK ){
          aData = pBt->pPage1->aData;
          aData[18] = (u8)iVersion;
          aData[19] = (u8)iVersion;
        }
      }
    }
  }

  pBt->doNotUseWAL = 0;
  return rc;
}

Changes to src/os.c.

  }else{
    double r;
    rc = pVfs->xCurrentTime(pVfs, &r);
    *pTimeOut = (sqlite3_int64)(r*86400000.0);
  }
  return rc;
}















































int sqlite3OsOpenMalloc(
  sqlite3_vfs *pVfs, 
  const char *zFile, 
  sqlite3_file **ppFile, 
  int flags,
  int *pOutFlags








>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>

  }else{
    double r;
    rc = pVfs->xCurrentTime(pVfs, &r);
    *pTimeOut = (sqlite3_int64)(r*86400000.0);
  }
  return rc;
}

/* Attempt to map all or part of a file into memory.  VFSes are not
** required to implement this.  The VFS might be an older version (less then
** 3) that does not have an xMap pointer.  Or the xMap pointer might be NULL.
*/
int sqlite3OsMap(
  sqlite3_file *pFile,    /* The file to be mapped into memory */
  sqlite3_int64 ofst,     /* Index of the first byte to map */
  sqlite3_int64 len,      /* Number of bytes to be mapped */
  int mmapFlags,          /* Map control flags */
  void **ppMemObj,        /* Write a mapping object here */
  void **ppMem            /* Write the start of the mapped file here */
){
  int rc;

  /* The current implementation only does read-only mmap.  This could change
  ** in the future. */
  assert( mmapFlags==SQLITE_OPEN_READONLY );

  /* The current implementation currently only maps the whole file.  This
  ** could change in the future. */
  assert( ofst==0 );

  if( pFile->pMethods==0 || pFile->pMethods->iVersion<3
     || pFile->pMethods->xMap==0 ){
    *ppMemObj = 0;
    *ppMem = 0;
    rc = SQLITE_CANTOPEN;
  }else{
    rc = pFile->pMethods->xMap(pFile, ofst, len, mmapFlags, ppMemObj, ppMem);
  }
  return rc;
}

/* Undo a mapping.
**
** The pMemObj parameter will have been obtained by a prior call to
** sqlite3OsMap().  So if pMemObj is not NULL, we know that the current
** VFS does support xMap and xUnmap.
*/
int sqlite3OsUnmap(sqlite3_file *pFile, void *pMemObj){
  int rc = SQLITE_OK;
  if( pMemObj ) rc = pFile->pMethods->xUnmap(pFile, pMemObj);
  return rc;
}


int sqlite3OsOpenMalloc(
  sqlite3_vfs *pVfs, 
  const char *zFile, 
  sqlite3_file **ppFile, 
  int flags,
  int *pOutFlags

Changes to src/os.h.

#define SQLITE_FCNTL_DB_UNCHANGED 0xca093fa0
int sqlite3OsSectorSize(sqlite3_file *id);
int sqlite3OsDeviceCharacteristics(sqlite3_file *id);
int sqlite3OsShmMap(sqlite3_file *,int,int,int,void volatile **);
int sqlite3OsShmLock(sqlite3_file *id, int, int, int);
void sqlite3OsShmBarrier(sqlite3_file *id);
int sqlite3OsShmUnmap(sqlite3_file *id, int);



/* 
** Functions for accessing sqlite3_vfs methods 
*/
int sqlite3OsOpen(sqlite3_vfs *, const char *, sqlite3_file*, int, int *);
int sqlite3OsDelete(sqlite3_vfs *, const char *, int);
int sqlite3OsAccess(sqlite3_vfs *, const char *, int, int *pResOut);

>
>

#define SQLITE_FCNTL_DB_UNCHANGED 0xca093fa0
int sqlite3OsSectorSize(sqlite3_file *id);
int sqlite3OsDeviceCharacteristics(sqlite3_file *id);
int sqlite3OsShmMap(sqlite3_file *,int,int,int,void volatile **);
int sqlite3OsShmLock(sqlite3_file *id, int, int, int);
void sqlite3OsShmBarrier(sqlite3_file *id);
int sqlite3OsShmUnmap(sqlite3_file *id, int);
int sqlite3OsMap(sqlite3_file*,sqlite3_int64,sqlite3_int64,int,void**,void**);
int sqlite3OsUnmap(sqlite3_file*,void*);

/* 
** Functions for accessing sqlite3_vfs methods 
*/
int sqlite3OsOpen(sqlite3_vfs *, const char *, sqlite3_file*, int, int *);
int sqlite3OsDelete(sqlite3_vfs *, const char *, int);
int sqlite3OsAccess(sqlite3_vfs *, const char *, int, int *pResOut);

Changes to src/os_unix.c.

#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <unistd.h>
#include <time.h>
#include <sys/time.h>
#include <errno.h>
#ifndef SQLITE_OMIT_WAL
#include <sys/mman.h>
#endif

#if SQLITE_ENABLE_LOCKING_STYLE
# include <sys/ioctl.h>
# if OS_VXWORKS
#  include <semaphore.h>
#  include <limits.h>
# else
................................................................................
#define osOpenDirectory ((int(*)(const char*,int*))aSyscall[17].pCurrent)

  { "mkdir",        (sqlite3_syscall_ptr)mkdir,           0 },
#define osMkdir     ((int(*)(const char*,mode_t))aSyscall[18].pCurrent)

  { "rmdir",        (sqlite3_syscall_ptr)rmdir,           0 },
#define osRmdir     ((int(*)(const char*))aSyscall[19].pCurrent)









}; /* End of the overrideable system calls */

/*
** This is the xSetSystemCall() method of sqlite3_vfs for all of the
** "unix" VFSes.  Return SQLITE_OK opon successfully updating the
** system call pointer, or SQLITE_NOTFOUND if there is no configurable
................................................................................

#else
# define unixShmMap     0
# define unixShmLock    0
# define unixShmBarrier 0
# define unixShmUnmap   0
#endif /* #ifndef SQLITE_OMIT_WAL */


























































/*
** Here ends the implementation of all sqlite3_file methods.
**
********************** End sqlite3_file Methods *******************************
******************************************************************************/

................................................................................
**
**   *  A constant sqlite3_io_methods object call METHOD that has locking
**      methods CLOSE, LOCK, UNLOCK, CKRESLOCK.
**
**   *  An I/O method finder function called FINDER that returns a pointer
**      to the METHOD object in the previous bullet.
*/
#define IOMETHODS(FINDER, METHOD, VERSION, CLOSE, LOCK, UNLOCK, CKLOCK)      \
static const sqlite3_io_methods METHOD = {                                   \
   VERSION,                    /* iVersion */                                \
   CLOSE,                      /* xClose */                                  \
   unixRead,                   /* xRead */                                   \
   unixWrite,                  /* xWrite */                                  \
   unixTruncate,               /* xTruncate */                               \
   unixSync,                   /* xSync */                                   \
   unixFileSize,               /* xFileSize */                               \
   LOCK,                       /* xLock */                                   \
   UNLOCK,                     /* xUnlock */                                 \
   CKLOCK,                     /* xCheckReservedLock */                      \
   unixFileControl,            /* xFileControl */                            \
   unixSectorSize,             /* xSectorSize */                             \
   unixDeviceCharacteristics,  /* xDeviceCapabilities */                     \
   unixShmMap,                 /* xShmMap */                                 \
   unixShmLock,                /* xShmLock */                                \
   unixShmBarrier,             /* xShmBarrier */                             \
   unixShmUnmap                /* xShmUnmap */                               \


};                                                                           \
static const sqlite3_io_methods *FINDER##Impl(const char *z, unixFile *p){   \
  UNUSED_PARAMETER(z); UNUSED_PARAMETER(p);                                  \
  return &METHOD;                                                            \
}                                                                            \
static const sqlite3_io_methods *(*const FINDER)(const char*,unixFile *p)    \
    = FINDER##Impl;
................................................................................
** Here are all of the sqlite3_io_methods objects for each of the
** locking strategies.  Functions that return pointers to these methods
** are also created.
*/
IOMETHODS(
  posixIoFinder,            /* Finder function name */
  posixIoMethods,           /* sqlite3_io_methods object name */
  2,                        /* shared memory is enabled */
  unixClose,                /* xClose method */
  unixLock,                 /* xLock method */
  unixUnlock,               /* xUnlock method */
  unixCheckReservedLock     /* xCheckReservedLock method */

)
IOMETHODS(
  nolockIoFinder,           /* Finder function name */
  nolockIoMethods,          /* sqlite3_io_methods object name */
  1,                        /* shared memory is disabled */
  nolockClose,              /* xClose method */
  nolockLock,               /* xLock method */
  nolockUnlock,             /* xUnlock method */
  nolockCheckReservedLock   /* xCheckReservedLock method */

)
IOMETHODS(
  dotlockIoFinder,          /* Finder function name */
  dotlockIoMethods,         /* sqlite3_io_methods object name */
  1,                        /* shared memory is disabled */
  dotlockClose,             /* xClose method */
  dotlockLock,              /* xLock method */
  dotlockUnlock,            /* xUnlock method */
  dotlockCheckReservedLock  /* xCheckReservedLock method */

)

#if SQLITE_ENABLE_LOCKING_STYLE && !OS_VXWORKS
IOMETHODS(
  flockIoFinder,            /* Finder function name */
  flockIoMethods,           /* sqlite3_io_methods object name */
  1,                        /* shared memory is disabled */
  flockClose,               /* xClose method */
  flockLock,                /* xLock method */
  flockUnlock,              /* xUnlock method */
  flockCheckReservedLock    /* xCheckReservedLock method */

)
#endif

#if OS_VXWORKS
IOMETHODS(
  semIoFinder,              /* Finder function name */
  semIoMethods,             /* sqlite3_io_methods object name */
  1,                        /* shared memory is disabled */
  semClose,                 /* xClose method */
  semLock,                  /* xLock method */
  semUnlock,                /* xUnlock method */
  semCheckReservedLock      /* xCheckReservedLock method */

)
#endif

#if defined(__APPLE__) && SQLITE_ENABLE_LOCKING_STYLE
IOMETHODS(
  afpIoFinder,              /* Finder function name */
  afpIoMethods,             /* sqlite3_io_methods object name */
  1,                        /* shared memory is disabled */
  afpClose,                 /* xClose method */
  afpLock,                  /* xLock method */
  afpUnlock,                /* xUnlock method */
  afpCheckReservedLock      /* xCheckReservedLock method */

)
#endif

/*
** The proxy locking method is a "super-method" in the sense that it
** opens secondary file descriptors for the conch and lock files and
** it uses proxy, dot-file, AFP, and flock() locking methods on those
................................................................................
static int proxyClose(sqlite3_file*);
static int proxyLock(sqlite3_file*, int);
static int proxyUnlock(sqlite3_file*, int);
static int proxyCheckReservedLock(sqlite3_file*, int*);
IOMETHODS(
  proxyIoFinder,            /* Finder function name */
  proxyIoMethods,           /* sqlite3_io_methods object name */
  1,                        /* shared memory is disabled */
  proxyClose,               /* xClose method */
  proxyLock,                /* xLock method */
  proxyUnlock,              /* xUnlock method */
  proxyCheckReservedLock    /* xCheckReservedLock method */

)
#endif

/* nfs lockd on OSX 10.3+ doesn't clear write locks when a read lock is set */
#if defined(__APPLE__) && SQLITE_ENABLE_LOCKING_STYLE
IOMETHODS(
  nfsIoFinder,               /* Finder function name */
  nfsIoMethods,              /* sqlite3_io_methods object name */
  1,                         /* shared memory is disabled */
  unixClose,                 /* xClose method */
  unixLock,                  /* xLock method */
  nfsUnlock,                 /* xUnlock method */
  unixCheckReservedLock      /* xCheckReservedLock method */

)
#endif

#if defined(__APPLE__) && SQLITE_ENABLE_LOCKING_STYLE
/* 
** This "finder" function attempts to determine the best locking strategy 
** for the database file "filePath".  It then returns the sqlite3_io_methods
................................................................................
    UNIXVFS("unix-proxy",    proxyIoFinder ),
#endif
  };
  unsigned int i;          /* Loop counter */

  /* Double-check that the aSyscall[] array has been constructed
  ** correctly.  See ticket [bb3a86e890c8e96ab] */
  assert( ArraySize(aSyscall)==20 );

  /* Register all VFSes defined in the aVfs[] array */
  for(i=0; i<(sizeof(aVfs)/sizeof(sqlite3_vfs)); i++){
    sqlite3_vfs_register(&aVfs[i], i==0);
  }
  return SQLITE_OK; 
}








<

<







 







>
>
>
>
>
>
>
>







 







>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>







 







|

|












|


|
>
>







 







<



|
>




<



|
>




<



|
>






<



|
>







<



|
>







<



|
>







 







<



|
>








<



|
>







 







|

#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <unistd.h>
#include <time.h>
#include <sys/time.h>
#include <errno.h>

#include <sys/mman.h>


#if SQLITE_ENABLE_LOCKING_STYLE
# include <sys/ioctl.h>
# if OS_VXWORKS
#  include <semaphore.h>
#  include <limits.h>
# else
................................................................................
#define osOpenDirectory ((int(*)(const char*,int*))aSyscall[17].pCurrent)

  { "mkdir",        (sqlite3_syscall_ptr)mkdir,           0 },
#define osMkdir     ((int(*)(const char*,mode_t))aSyscall[18].pCurrent)

  { "rmdir",        (sqlite3_syscall_ptr)rmdir,           0 },
#define osRmdir     ((int(*)(const char*))aSyscall[19].pCurrent)

  { "mmap",         (sqlite3_syscall_ptr)mmap,            0 },
#define osMmap  ((int(*)(void*,size_t,int,int,int,off_t))aSyscall[20].pCurrent)

  { "munmap",       (sqlite3_syscall_ptr)munmap,          0 },
#define osMunmap    ((int(*)(void*,size_t))aSyscall[21].pCurrent)



}; /* End of the overrideable system calls */

/*
** This is the xSetSystemCall() method of sqlite3_vfs for all of the
** "unix" VFSes.  Return SQLITE_OK opon successfully updating the
** system call pointer, or SQLITE_NOTFOUND if there is no configurable
................................................................................

#else
# define unixShmMap     0
# define unixShmLock    0
# define unixShmBarrier 0
# define unixShmUnmap   0
#endif /* #ifndef SQLITE_OMIT_WAL */

/*
** An object used to record enough information about a file mapping to
** undo that mapping.
*/
struct unixMapping {
  sqlite3_int64 len;
  void *p;
};

/*
** Try to map some or all of a file into memory
*/
static int unixMap(
  sqlite3_file *pFile,   /* File to be mapped */
  sqlite3_int64 ofst,    /* Offset of start of section to be mapped */
  sqlite3_int64 len,     /* Length of the section to be mapped */
  int mmapFlags,         /* Flags controlling the mapping */
  void **ppMapObj,       /* Write here an object to undo the mapping */
  void **ppMem           /* Write here a pointer to the mapped file */
){
  struct unixMapping *pNew;
  unixFile *pUFile = (unixFile*)pFile;

  assert( mmapFlags==SQLITE_OPEN_READONLY );
  sqlite3BeginBenignMalloc();
  pNew = sqlite3_malloc( sizeof(*pNew) );
  sqlite3EndBenignMalloc();
  if( pNew==0 ){
    *ppMapObj = 0;
    *ppMem = 0;
    return SQLITE_CANTOPEN;
  }
  pNew->len = len;
  pNew->p = *ppMem = mmap(0, len, PROT_READ, MAP_SHARED, pUFile->h, 0);
  if( pNew->p==0 ){
    sqlite3_free(pNew);
    return SQLITE_CANTOPEN;
  }else{
    *ppMapObj = pNew;
    return SQLITE_OK;
  }
}

/*
** Undo a prior memory mapping.
*/
static int unixUnmap(
  sqlite3_file *pFile,
  void *pMapObj
){
  struct unixMapping *pMap = (struct unixMapping*)pMapObj;
  assert( pMap!=0 );
  munmap(pMap->p, pMap->len);
  sqlite3_free(pMap);
  return SQLITE_OK;
}

/*
** Here ends the implementation of all sqlite3_file methods.
**
********************** End sqlite3_file Methods *******************************
******************************************************************************/

................................................................................
**
**   *  A constant sqlite3_io_methods object call METHOD that has locking
**      methods CLOSE, LOCK, UNLOCK, CKRESLOCK.
**
**   *  An I/O method finder function called FINDER that returns a pointer
**      to the METHOD object in the previous bullet.
*/
#define IOMETHODS(FINDER, METHOD, CLOSE, LOCK, UNLOCK, CKLOCK, SHMMAP)       \
static const sqlite3_io_methods METHOD = {                                   \
   3,                          /* iVersion */                                \
   CLOSE,                      /* xClose */                                  \
   unixRead,                   /* xRead */                                   \
   unixWrite,                  /* xWrite */                                  \
   unixTruncate,               /* xTruncate */                               \
   unixSync,                   /* xSync */                                   \
   unixFileSize,               /* xFileSize */                               \
   LOCK,                       /* xLock */                                   \
   UNLOCK,                     /* xUnlock */                                 \
   CKLOCK,                     /* xCheckReservedLock */                      \
   unixFileControl,            /* xFileControl */                            \
   unixSectorSize,             /* xSectorSize */                             \
   unixDeviceCharacteristics,  /* xDeviceCapabilities */                     \
   SHMMAP,                     /* xShmMap */                                 \
   unixShmLock,                /* xShmLock */                                \
   unixShmBarrier,             /* xShmBarrier */                             \
   unixShmUnmap,               /* xShmUnmap */                               \
   unixMap,                    /* xMap */                                    \
   unixUnmap                   /* xUnmap */                                  \
};                                                                           \
static const sqlite3_io_methods *FINDER##Impl(const char *z, unixFile *p){   \
  UNUSED_PARAMETER(z); UNUSED_PARAMETER(p);                                  \
  return &METHOD;                                                            \
}                                                                            \
static const sqlite3_io_methods *(*const FINDER)(const char*,unixFile *p)    \
    = FINDER##Impl;
................................................................................
** Here are all of the sqlite3_io_methods objects for each of the
** locking strategies.  Functions that return pointers to these methods
** are also created.
*/
IOMETHODS(
  posixIoFinder,            /* Finder function name */
  posixIoMethods,           /* sqlite3_io_methods object name */

  unixClose,                /* xClose method */
  unixLock,                 /* xLock method */
  unixUnlock,               /* xUnlock method */
  unixCheckReservedLock,    /* xCheckReservedLock method */
  unixShmMap                /* Shared memory enabled */
)
IOMETHODS(
  nolockIoFinder,           /* Finder function name */
  nolockIoMethods,          /* sqlite3_io_methods object name */

  nolockClose,              /* xClose method */
  nolockLock,               /* xLock method */
  nolockUnlock,             /* xUnlock method */
  nolockCheckReservedLock,  /* xCheckReservedLock method */
  0                         /* Shared memory disabled */
)
IOMETHODS(
  dotlockIoFinder,          /* Finder function name */
  dotlockIoMethods,         /* sqlite3_io_methods object name */

  dotlockClose,             /* xClose method */
  dotlockLock,              /* xLock method */
  dotlockUnlock,            /* xUnlock method */
  dotlockCheckReservedLock, /* xCheckReservedLock method */
  0                         /* Shared memory disabled */
)

#if SQLITE_ENABLE_LOCKING_STYLE && !OS_VXWORKS
IOMETHODS(
  flockIoFinder,            /* Finder function name */
  flockIoMethods,           /* sqlite3_io_methods object name */

  flockClose,               /* xClose method */
  flockLock,                /* xLock method */
  flockUnlock,              /* xUnlock method */
  flockCheckReservedLock,   /* xCheckReservedLock method */
  0                         /* Shared memory disabled */
)
#endif

#if OS_VXWORKS
IOMETHODS(
  semIoFinder,              /* Finder function name */
  semIoMethods,             /* sqlite3_io_methods object name */

  semClose,                 /* xClose method */
  semLock,                  /* xLock method */
  semUnlock,                /* xUnlock method */
  semCheckReservedLock,     /* xCheckReservedLock method */
  0                         /* Shared memory disabled */
)
#endif

#if defined(__APPLE__) && SQLITE_ENABLE_LOCKING_STYLE
IOMETHODS(
  afpIoFinder,              /* Finder function name */
  afpIoMethods,             /* sqlite3_io_methods object name */

  afpClose,                 /* xClose method */
  afpLock,                  /* xLock method */
  afpUnlock,                /* xUnlock method */
  afpCheckReservedLock,     /* xCheckReservedLock method */
  0                         /* Shared memory disabled */
)
#endif

/*
** The proxy locking method is a "super-method" in the sense that it
** opens secondary file descriptors for the conch and lock files and
** it uses proxy, dot-file, AFP, and flock() locking methods on those
................................................................................
static int proxyClose(sqlite3_file*);
static int proxyLock(sqlite3_file*, int);
static int proxyUnlock(sqlite3_file*, int);
static int proxyCheckReservedLock(sqlite3_file*, int*);
IOMETHODS(
  proxyIoFinder,            /* Finder function name */
  proxyIoMethods,           /* sqlite3_io_methods object name */

  proxyClose,               /* xClose method */
  proxyLock,                /* xLock method */
  proxyUnlock,              /* xUnlock method */
  proxyCheckReservedLock,   /* xCheckReservedLock method */
  0                         /* Shared memory disabled */
)
#endif

/* nfs lockd on OSX 10.3+ doesn't clear write locks when a read lock is set */
#if defined(__APPLE__) && SQLITE_ENABLE_LOCKING_STYLE
IOMETHODS(
  nfsIoFinder,               /* Finder function name */
  nfsIoMethods,              /* sqlite3_io_methods object name */

  unixClose,                 /* xClose method */
  unixLock,                  /* xLock method */
  nfsUnlock,                 /* xUnlock method */
  unixCheckReservedLock,     /* xCheckReservedLock method */
  0                          /* Shared memory disabled */
)
#endif

#if defined(__APPLE__) && SQLITE_ENABLE_LOCKING_STYLE
/* 
** This "finder" function attempts to determine the best locking strategy 
** for the database file "filePath".  It then returns the sqlite3_io_methods
................................................................................
    UNIXVFS("unix-proxy",    proxyIoFinder ),
#endif
  };
  unsigned int i;          /* Loop counter */

  /* Double-check that the aSyscall[] array has been constructed
  ** correctly.  See ticket [bb3a86e890c8e96ab] */
  assert( ArraySize(aSyscall)==22 );

  /* Register all VFSes defined in the aVfs[] array */
  for(i=0; i<(sizeof(aVfs)/sizeof(sqlite3_vfs)); i++){
    sqlite3_vfs_register(&aVfs[i], i==0);
  }
  return SQLITE_OK; 
}

Changes to src/pager.c.

  sqlite3_file *sjfd;         /* File descriptor for sub-journal */
  i64 journalOff;             /* Current write offset in the journal file */
  i64 journalHdr;             /* Byte offset to previous journal header */
  sqlite3_backup *pBackup;    /* Pointer to list of ongoing backup processes */
  PagerSavepoint *aSavepoint; /* Array of active savepoints */
  int nSavepoint;             /* Number of elements in aSavepoint[] */
  char dbFileVers[16];        /* Changes whenever database file changes */



  /*
  ** End of the routinely-changing class members
  ***************************************************************************/

  u16 nExtra;                 /* Add this many bytes to each in-memory page */
  i16 nReserve;               /* Number of unused bytes at end of each page */
  u32 vfsFlags;               /* Flags for sqlite3_vfs.xOpen() */
................................................................................
  }

  /* If Pager.errCode is set, the contents of the pager cache cannot be
  ** trusted. Now that there are no outstanding references to the pager,
  ** it can safely move back to PAGER_OPEN state. This happens in both
  ** normal and exclusive-locking mode.
  */
  if( pPager->errCode ){
    assert( !MEMDB );
    pager_reset(pPager);
    pPager->changeCountDone = pPager->tempFile;
    pPager->eState = PAGER_OPEN;
    pPager->errCode = SQLITE_OK;




  }

  pPager->journalOff = 0;
  pPager->journalHdr = 0;
  pPager->setMaster = 0;

}

/*
** This function is called whenever an IOERR or FULL error that requires
** the pager to transition into the ERROR state may ahve occurred.
** The first argument is a pointer to the pager structure, the second 
** the error-code about to be returned by a pager API function. The 
................................................................................
  Pgno pgno = pPg->pgno;       /* Page number to read */
  int rc = SQLITE_OK;          /* Return code */
  int isInWal = 0;             /* True if page is in log file */
  int pgsz = pPager->pageSize; /* Number of bytes to read */

  assert( pPager->eState>=PAGER_READER && !MEMDB );
  assert( isOpen(pPager->fd) );

  if( NEVER(!isOpen(pPager->fd)) ){
    assert( pPager->tempFile );
    memset(pPg->pData, 0, pPager->pageSize);
    return SQLITE_OK;
  }


  if( pagerUseWal(pPager) ){
    /* Try to pull the page from the write-ahead log. */
    rc = sqlite3WalRead(pPager->pWal, pgno, &isInWal, pgsz, pPg->pData);
  }
  if( rc==SQLITE_OK && !isInWal ){
    i64 iOffset = (pgno-1)*(i64)pPager->pageSize;



    rc = sqlite3OsRead(pPager->fd, pPg->pData, pgsz, iOffset);
    if( rc==SQLITE_IOERR_SHORT_READ ){
      rc = SQLITE_OK;

    }
  }

  if( pgno==1 ){
    if( rc ){
      /* If the read is unsuccessful, set the dbFileVers[] to something
      ** that will never be a valid file version.  dbFileVers[] is a copy
................................................................................
    rc = pagerBeginReadTransaction(pPager);
  }

  if( pPager->eState==PAGER_OPEN && rc==SQLITE_OK ){
    rc = pagerPagecount(pPager, &pPager->dbSize);
  }








 failed:
  if( rc!=SQLITE_OK ){
    assert( !MEMDB );
    pager_unlock(pPager);
    assert( pPager->eState==PAGER_OPEN );
  }else{
    pPager->eState = PAGER_READER;
................................................................................
  */
  if( pPager->eState==PAGER_WRITER_LOCKED ){
    rc = pager_open_journal(pPager);
    if( rc!=SQLITE_OK ) return rc;
  }
  assert( pPager->eState>=PAGER_WRITER_CACHEMOD );
  assert( assert_pager_state(pPager) );







  /* Mark the page as dirty.  If the page has already been written
  ** to the journal then we can return right away.
  */
  sqlite3PcacheMakeDirty(pPg);
  if( pageInJournal(pPg) && !subjRequiresPage(pPg) ){
    assert( !pagerUseWal(pPager) );
................................................................................
}

/*
** Return TRUE if the page given in the argument was previously passed
** to sqlite3PagerWrite().  In other words, return TRUE if it is ok
** to change the content of the page.
*/
#ifndef NDEBUG
int sqlite3PagerIswriteable(DbPage *pPg){
  return pPg->flags&PGHDR_DIRTY;
}
#endif

/*
** A call to this routine tells the pager that it is not necessary to
** write the information on page pPg back to the disk, even though
** that page might be marked as dirty.  This happens, for example, when
** the page has been added as a leaf of the freelist and so its
** content no longer matters.








>
>
>







 







|





>
>
>
>





>







 







<
<
<
<
<
<
>






|
>
>
>
|
|
|
>







 







>
>
>
>
>
>
>







 







>
>
>
>
>
>







 







<



<

  sqlite3_file *sjfd;         /* File descriptor for sub-journal */
  i64 journalOff;             /* Current write offset in the journal file */
  i64 journalHdr;             /* Byte offset to previous journal header */
  sqlite3_backup *pBackup;    /* Pointer to list of ongoing backup processes */
  PagerSavepoint *aSavepoint; /* Array of active savepoints */
  int nSavepoint;             /* Number of elements in aSavepoint[] */
  char dbFileVers[16];        /* Changes whenever database file changes */
  u8 *aFileContent;           /* File mapped into memory */
  sqlite3_int64 nFileContent; /* Bytes of memory mapped into aFileContent */
  void *pMapObject;           /* Used to unmap the file */
  /*
  ** End of the routinely-changing class members
  ***************************************************************************/

  u16 nExtra;                 /* Add this many bytes to each in-memory page */
  i16 nReserve;               /* Number of unused bytes at end of each page */
  u32 vfsFlags;               /* Flags for sqlite3_vfs.xOpen() */
................................................................................
  }

  /* If Pager.errCode is set, the contents of the pager cache cannot be
  ** trusted. Now that there are no outstanding references to the pager,
  ** it can safely move back to PAGER_OPEN state. This happens in both
  ** normal and exclusive-locking mode.
  */
  if( pPager->errCode || pPager->pMapObject ){
    assert( !MEMDB );
    pager_reset(pPager);
    pPager->changeCountDone = pPager->tempFile;
    pPager->eState = PAGER_OPEN;
    pPager->errCode = SQLITE_OK;
    sqlite3OsUnmap(pPager->fd, pPager->pMapObject);
    pPager->pMapObject = 0;
    pPager->aFileContent = 0;
    pPager->nFileContent = 0;
  }

  pPager->journalOff = 0;
  pPager->journalHdr = 0;
  pPager->setMaster = 0;

}

/*
** This function is called whenever an IOERR or FULL error that requires
** the pager to transition into the ERROR state may ahve occurred.
** The first argument is a pointer to the pager structure, the second 
** the error-code about to be returned by a pager API function. The 
................................................................................
  Pgno pgno = pPg->pgno;       /* Page number to read */
  int rc = SQLITE_OK;          /* Return code */
  int isInWal = 0;             /* True if page is in log file */
  int pgsz = pPager->pageSize; /* Number of bytes to read */

  assert( pPager->eState>=PAGER_READER && !MEMDB );
  assert( isOpen(pPager->fd) );






  assert( pPg->pBuf==pPg->pData );

  if( pagerUseWal(pPager) ){
    /* Try to pull the page from the write-ahead log. */
    rc = sqlite3WalRead(pPager->pWal, pgno, &isInWal, pgsz, pPg->pData);
  }
  if( rc==SQLITE_OK && !isInWal ){
    i64 iOffset = (pgno-1)*(i64)pgsz;
    if( iOffset+pgsz <= pPager->nFileContent ){
      pPg->pData = &pPager->aFileContent[iOffset];
    }else{
      rc = sqlite3OsRead(pPager->fd, pPg->pData, pgsz, iOffset);
      if( rc==SQLITE_IOERR_SHORT_READ ){
        rc = SQLITE_OK;
      }
    }
  }

  if( pgno==1 ){
    if( rc ){
      /* If the read is unsuccessful, set the dbFileVers[] to something
      ** that will never be a valid file version.  dbFileVers[] is a copy
................................................................................
    rc = pagerBeginReadTransaction(pPager);
  }

  if( pPager->eState==PAGER_OPEN && rc==SQLITE_OK ){
    rc = pagerPagecount(pPager, &pPager->dbSize);
  }

  assert( pPager->aFileContent==0 );
  pPager->nFileContent = pPager->dbSize*(sqlite3_int64)pPager->pageSize;
  sqlite3OsMap(pPager->fd, 0, pPager->nFileContent,
               SQLITE_OPEN_READONLY, (void**)&pPager->pMapObject,
               (void**)&pPager->aFileContent);
  if( pPager->aFileContent==0 ) pPager->nFileContent = 0;

 failed:
  if( rc!=SQLITE_OK ){
    assert( !MEMDB );
    pager_unlock(pPager);
    assert( pPager->eState==PAGER_OPEN );
  }else{
    pPager->eState = PAGER_READER;
................................................................................
  */
  if( pPager->eState==PAGER_WRITER_LOCKED ){
    rc = pager_open_journal(pPager);
    if( rc!=SQLITE_OK ) return rc;
  }
  assert( pPager->eState>=PAGER_WRITER_CACHEMOD );
  assert( assert_pager_state(pPager) );

  /* Make sure page content is held in malloced memory */
  if( pPg->pData!=pPg->pBuf ){
    memcpy(pPg->pBuf, pPg->pData, pPager->pageSize);
    pData = pPg->pData = pPg->pBuf;
  }

  /* Mark the page as dirty.  If the page has already been written
  ** to the journal then we can return right away.
  */
  sqlite3PcacheMakeDirty(pPg);
  if( pageInJournal(pPg) && !subjRequiresPage(pPg) ){
    assert( !pagerUseWal(pPager) );
................................................................................
}

/*
** Return TRUE if the page given in the argument was previously passed
** to sqlite3PagerWrite().  In other words, return TRUE if it is ok
** to change the content of the page.
*/

int sqlite3PagerIswriteable(DbPage *pPg){
  return pPg->flags&PGHDR_DIRTY;
}


/*
** A call to this routine tells the pager that it is not necessary to
** write the information on page pPg back to the disk, even though
** that page might be marked as dirty.  This happens, for example, when
** the page has been added as a leaf of the freelist and so its
** content no longer matters.

Changes to src/pager.h.

#if defined(SQLITE_HAS_CODEC) && !defined(SQLITE_OMIT_WAL)
void *sqlite3PagerCodec(DbPage *);
#endif

/* Functions to support testing and debugging. */
#if !defined(NDEBUG) || defined(SQLITE_TEST)
  Pgno sqlite3PagerPagenumber(DbPage*);
  int sqlite3PagerIswriteable(DbPage*);
#endif

#ifdef SQLITE_TEST
  int *sqlite3PagerStats(Pager*);
  void sqlite3PagerRefdump(Pager*);
  void disable_simulated_io_errors(void);
  void enable_simulated_io_errors(void);
#else
# define disable_simulated_io_errors()
# define enable_simulated_io_errors()
#endif

#endif /* _PAGER_H_ */








<

>

#if defined(SQLITE_HAS_CODEC) && !defined(SQLITE_OMIT_WAL)
void *sqlite3PagerCodec(DbPage *);
#endif

/* Functions to support testing and debugging. */
#if !defined(NDEBUG) || defined(SQLITE_TEST)
  Pgno sqlite3PagerPagenumber(DbPage*);

#endif
int sqlite3PagerIswriteable(DbPage*);
#ifdef SQLITE_TEST
  int *sqlite3PagerStats(Pager*);
  void sqlite3PagerRefdump(Pager*);
  void disable_simulated_io_errors(void);
  void enable_simulated_io_errors(void);
#else
# define disable_simulated_io_errors()
# define enable_simulated_io_errors()
#endif

#endif /* _PAGER_H_ */

Changes to src/pcache.c.


  if( pPage ){
    pPgHdr = (PgHdr *)pPage->pExtra;

    if( !pPgHdr->pPage ){
      memset(pPgHdr, 0, sizeof(PgHdr));
      pPgHdr->pPage = pPage;
      pPgHdr->pData = pPage->pBuf;
      pPgHdr->pExtra = (void *)&pPgHdr[1];
      memset(pPgHdr->pExtra, 0, pCache->szExtra);
      pPgHdr->pCache = pCache;
      pPgHdr->pgno = pgno;
    }
    assert( pPgHdr->pCache==pCache );
    assert( pPgHdr->pgno==pgno );
    assert( pPgHdr->pData==pPage->pBuf );
    assert( pPgHdr->pExtra==(void *)&pPgHdr[1] );

    if( 0==pPgHdr->nRef ){
      pCache->nRef++;
    }
    pPgHdr->nRef++;
    if( pgno==1 ){


  if( pPage ){
    pPgHdr = (PgHdr *)pPage->pExtra;

    if( !pPgHdr->pPage ){
      memset(pPgHdr, 0, sizeof(PgHdr));
      pPgHdr->pPage = pPage;
      pPgHdr->pData = pPgHdr->pBuf = pPage->pBuf;
      pPgHdr->pExtra = (void *)&pPgHdr[1];
      memset(pPgHdr->pExtra, 0, pCache->szExtra);
      pPgHdr->pCache = pCache;
      pPgHdr->pgno = pgno;
    }
    assert( pPgHdr->pCache==pCache );
    assert( pPgHdr->pgno==pgno );
    assert( pPgHdr->pBuf==pPage->pBuf );
    assert( pPgHdr->pExtra==(void *)&pPgHdr[1] );

    if( 0==pPgHdr->nRef ){
      pCache->nRef++;
    }
    pPgHdr->nRef++;
    if( pgno==1 ){

Changes to src/pcache.h.


/*
** Every page in the cache is controlled by an instance of the following
** structure.
*/
struct PgHdr {
  sqlite3_pcache_page *pPage;    /* Pcache object page handle */
  void *pData;                   /* Page data */

  void *pExtra;                  /* Extra content */
  PgHdr *pDirty;                 /* Transient list of dirty pages */
  Pgno pgno;                     /* Page number for this page */
  Pager *pPager;                 /* The pager this page is part of */
#ifdef SQLITE_CHECK_PAGES
  u32 pageHash;                  /* Hash of page content */
#endif

|
>


/*
** Every page in the cache is controlled by an instance of the following
** structure.
*/
struct PgHdr {
  sqlite3_pcache_page *pPage;    /* Pcache object page handle */
  void *pData;                   /* Page data to actually use */
  void *pBuf;                    /* Malloced buffer to hold pData */
  void *pExtra;                  /* Extra content */
  PgHdr *pDirty;                 /* Transient list of dirty pages */
  Pgno pgno;                     /* Page number for this page */
  Pager *pPager;                 /* The pager this page is part of */
#ifdef SQLITE_CHECK_PAGES
  u32 pageHash;                  /* Hash of page content */
#endif

Changes to src/sqlite.h.in.

  int (*xDeviceCharacteristics)(sqlite3_file*);
  /* Methods above are valid for version 1 */
  int (*xShmMap)(sqlite3_file*, int iPg, int pgsz, int, void volatile**);
  int (*xShmLock)(sqlite3_file*, int offset, int n, int flags);
  void (*xShmBarrier)(sqlite3_file*);
  int (*xShmUnmap)(sqlite3_file*, int deleteFlag);
  /* Methods above are valid for version 2 */




  /* Additional methods may be added in future releases */
};

/*
** CAPI3REF: Standard File Control Opcodes
**
** These integer constants are opcodes for the xFileControl method








>
>
>
>

  int (*xDeviceCharacteristics)(sqlite3_file*);
  /* Methods above are valid for version 1 */
  int (*xShmMap)(sqlite3_file*, int iPg, int pgsz, int, void volatile**);
  int (*xShmLock)(sqlite3_file*, int offset, int n, int flags);
  void (*xShmBarrier)(sqlite3_file*);
  int (*xShmUnmap)(sqlite3_file*, int deleteFlag);
  /* Methods above are valid for version 2 */
  int (*xMap)(sqlite3_file*, sqlite3_int64 ofst, sqlite3_int64 len,
              int mmapFlags, void **ppMemObj, void **ppMem);
  int (*xUnmap)(sqlite3_file*, void *pMemObj);
  /* Methods above are valid for version 3 */
  /* Additional methods may be added in future releases */
};

/*
** CAPI3REF: Standard File Control Opcodes
**
** These integer constants are opcodes for the xFileControl method

Changes to test/syscall.test.

#-------------------------------------------------------------------------
# Tests for the xNextSystemCall method.
#
foreach s {
    open close access getcwd stat fstat ftruncate
    fcntl read pread write pwrite fchmod fallocate
    pread64 pwrite64 unlink openDirectory mkdir rmdir

} {
  if {[test_syscall exists $s]} {lappend syscall_list $s}
}
do_test 3.1 { lsort [test_syscall list] } [lsort $syscall_list]

#-------------------------------------------------------------------------
# This test verifies that if a call to open() fails and errno is set to

#-------------------------------------------------------------------------
# Tests for the xNextSystemCall method.
#
foreach s {
    open close access getcwd stat fstat ftruncate
    fcntl read pread write pwrite fchmod fallocate
    pread64 pwrite64 unlink openDirectory mkdir rmdir
    mmap munmap
} {
  if {[test_syscall exists $s]} {lappend syscall_list $s}
}
do_test 3.1 { lsort [test_syscall list] } [lsort $syscall_list]

#-------------------------------------------------------------------------
# This test verifies that if a call to open() fails and errno is set to