[sql] Import reference version of SQLite 3.8.7.4.

third_party/sqlite/sqlite-src-3080704/src/pager.c

 /*
 ** 2001 September 15
 **
 ** The author disclaims copyright to this source code.  In place of
 ** a legal notice, here is a blessing:
 **
 **    May you do good and not evil.
 **    May you find forgiveness for yourself and forgive others.
 **    May you share freely, never taking more than you give.
 **
@@ skipping 57 matching lines @@
 **     first 100 bytes of the database file.
 ** 
 ** (5) All writes to the database file are synced prior to the rollback journal
 **     being deleted, truncated, or zeroed.
 ** 
 ** (6) If a master journal file is used, then all writes to the database file
 **     are synced prior to the master journal being deleted.
 ** 
 ** Definition: Two databases (or the same database at two points it time)
 ** are said to be "logically equivalent" if they give the same answer to
-** all queries.  Note in particular the the content of freelist leaf
-** pages can be changed arbitarily without effecting the logical equivalence
+** all queries.  Note in particular the content of freelist leaf
+** pages can be changed arbitrarily without affecting the logical equivalence
 ** of the database.
 ** 
 ** (7) At any time, if any subset, including the empty set and the total set,
 **     of the unsynced changes to a rollback journal are removed and the 
-**     journal is rolled back, the resulting database file will be logical
+**     journal is rolled back, the resulting database file will be logically
 **     equivalent to the database file at the beginning of the transaction.
 ** 
 ** (8) When a transaction is rolled back, the xTruncate method of the VFS
 **     is called to restore the database file to the same size it was at
 **     the beginning of the transaction.  (In some VFSes, the xTruncate
 **     method is a no-op, but that does not change the fact the SQLite will
 **     invoke it.)
 ** 
 ** (9) Whenever the database file is modified, at least one bit in the range
 **     of bytes from 24 through 39 inclusive will be changed prior to releasing
@@ skipping 171 matching lines @@
 **    A rollback-mode pager changes to WRITER_FINISHED state from WRITER_DBMOD
 **    state after the entire transaction has been successfully written into the
 **    database file. In this state the transaction may be committed simply
 **    by finalizing the journal file. Once in WRITER_FINISHED state, it is 
 **    not possible to modify the database further. At this point, the upper 
 **    layer must either commit or rollback the transaction.
 **
 **    * A write transaction is active.
 **    * An EXCLUSIVE or greater lock is held on the database file.
 **    * All writing and syncing of journal and database data has finished.
-**      If no error occured, all that remains is to finalize the journal to
+**      If no error occurred, all that remains is to finalize the journal to
 **      commit the transaction. If an error did occur, the caller will need
 **      to rollback the transaction. 
 **
 **  ERROR:
 **
 **    The ERROR state is entered when an IO or disk-full error (including
 **    SQLITE_IOERR_NOMEM) occurs at a point in the code that makes it 
 **    difficult to be sure that the in-memory pager state (cache contents, 
 **    db size etc.) are consistent with the contents of the file-system.
 **
@@ skipping 84 matching lines @@
 ** VFS call is successful. This way, the Pager.eLock variable may be set
 ** to a less exclusive (lower) value than the lock that is actually held
 ** at the system level, but it is never set to a more exclusive value.
 **
 ** This is usually safe. If an xUnlock fails or appears to fail, there may 
 ** be a few redundant xLock() calls or a lock may be held for longer than
 ** required, but nothing really goes wrong.
 **
 ** The exception is when the database file is unlocked as the pager moves
 ** from ERROR to OPEN state. At this point there may be a hot-journal file 
-** in the file-system that needs to be rolled back (as part of a OPEN->SHARED
+** in the file-system that needs to be rolled back (as part of an OPEN->SHARED
 ** transition, by the same pager or any other). If the call to xUnlock()
 ** fails at this point and the pager is left holding an EXCLUSIVE lock, this
 ** can confuse the call to xCheckReservedLock() call made later as part
 ** of hot-journal detection.
 **
 ** xCheckReservedLock() is defined as returning true "if there is a RESERVED 
 ** lock held by this process or any others". So xCheckReservedLock may 
 ** return true because the caller itself is holding an EXCLUSIVE lock (but
 ** doesn't know it because of a previous error in xUnlock). If this happens
 ** a hot-journal may be mistaken for a journal being created by an active
@@ skipping 55 matching lines @@
   i64 iHdrOffset;              /* See above */
   Bitvec *pInSavepoint;        /* Set of pages in this savepoint */
   Pgno nOrig;                  /* Original number of pages in file */
   Pgno iSubRec;                /* Index of first record in sub-journal */
 #ifndef SQLITE_OMIT_WAL
   u32 aWalData[WAL_SAVEPOINT_NDATA];        /* WAL savepoint context */
 #endif
 };
 
 /*
-** A open page cache is an instance of struct Pager. A description of
+** Bits of the Pager.doNotSpill flag.  See further description below.
+*/
+#define SPILLFLAG_OFF         0x01      /* Never spill cache.  Set via pragma */
+#define SPILLFLAG_ROLLBACK    0x02      /* Current rolling back, so do not spill */
+#define SPILLFLAG_NOSYNC      0x04      /* Spill is ok, but do not sync */
+
+/*
+** An open page cache is an instance of struct Pager. A description of
 ** some of the more important member variables follows:
 **
 ** eState
 **
 **   The current 'state' of the pager object. See the comment and state
 **   diagram above for a description of the pager state.
 **
 ** eLock
 **
 **   For a real on-disk database, the current lock held on the database file -
@@ skipping 44 matching lines @@
 **   Journal files that contain master journal pointers cannot be finalized
 **   simply by overwriting the first journal-header with zeroes, as the
 **   master journal pointer could interfere with hot-journal rollback of any
 **   subsequently interrupted transaction that reuses the journal file.
 **
 **   The flag is cleared as soon as the journal file is finalized (either
 **   by PagerCommitPhaseTwo or PagerRollback). If an IO error prevents the
 **   journal file from being successfully finalized, the setMaster flag
 **   is cleared anyway (and the pager will move to ERROR state).
 **
-** doNotSpill, doNotSyncSpill
+** doNotSpill
 **
-**   These two boolean variables control the behaviour of cache-spills
-**   (calls made by the pcache module to the pagerStress() routine to
-**   write cached data to the file-system in order to free up memory).
+**   This variables control the behavior of cache-spills  (calls made by
+**   the pcache module to the pagerStress() routine to write cached data
+**   to the file-system in order to free up memory).
 **
-**   When doNotSpill is non-zero, writing to the database from pagerStress()
-**   is disabled altogether. This is done in a very obscure case that
+**   When bits SPILLFLAG_OFF or SPILLFLAG_ROLLBACK of doNotSpill are set,
+**   writing to the database from pagerStress() is disabled altogether.
+**   The SPILLFLAG_ROLLBACK case is done in a very obscure case that
 **   comes up during savepoint rollback that requires the pcache module
 **   to allocate a new page to prevent the journal file from being written
-**   while it is being traversed by code in pager_playback().
+**   while it is being traversed by code in pager_playback().  The SPILLFLAG_OFF
+**   case is a user preference.
 ** 
-**   If doNotSyncSpill is non-zero, writing to the database from pagerStress()
+**   If the SPILLFLAG_NOSYNC bit is set, writing to the database from pagerStress()
 **   is permitted, but syncing the journal file is not. This flag is set
 **   by sqlite3PagerWrite() when the file-system sector-size is larger than
 **   the database page-size in order to prevent a journal sync from happening 
 **   in between the journalling of two pages on the same sector. 
 **
 ** subjInMemory
 **
 **   This is a boolean variable. If true, then any required sub-journal
 **   is opened as an in-memory journal file. If false, then in-memory
 **   sub-journals are only used for in-memory pager files.
@@ skipping 60 matching lines @@
 **   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.
 */
 struct Pager {
   sqlite3_vfs *pVfs;          /* OS functions to use for IO */
   u8 exclusiveMode;           /* Boolean. True if locking_mode==EXCLUSIVE */
   u8 journalMode;             /* One of the PAGER_JOURNALMODE_* values */
   u8 useJournal;              /* Use a rollback journal on this file */
-  u8 noReadlock;              /* Do not bother to obtain readlocks */
   u8 noSync;                  /* Do not sync the journal if true */
   u8 fullSync;                /* Do extra syncs of the journal for robustness */
   u8 ckptSyncFlags;           /* SYNC_NORMAL or SYNC_FULL for checkpoint */
+  u8 walSyncFlags;            /* SYNC_NORMAL or SYNC_FULL for wal writes */
   u8 syncFlags;               /* SYNC_NORMAL or SYNC_FULL otherwise */
-  u8 tempFile;                /* zFilename is a temporary file */
+  u8 tempFile;                /* zFilename is a temporary or immutable file */
+  u8 noLock;                  /* Do not lock (except in WAL mode) */
   u8 readOnly;                /* True for a read-only database */
   u8 memDb;                   /* True to inhibit all file I/O */
 
   /**************************************************************************
   ** The following block contains those class members that change during
-  ** routine opertion.  Class members not in this block are either fixed
+  ** routine operation.  Class members not in this block are either fixed
   ** when the pager is first created or else only change when there is a
   ** significant mode change (such as changing the page_size, locking_mode,
   ** or the journal_mode).  From another view, these class members describe
   ** the "state" of the pager, while other class members describe the
   ** "configuration" of the pager.
   */
   u8 eState;                  /* Pager state (OPEN, READER, WRITER_LOCKED..) */
   u8 eLock;                   /* Current lock held on database file */
   u8 changeCountDone;         /* Set after incrementing the change-counter */
   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 */
   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 bUseFetch;               /* True to use xFetch() */
+  int nMmapOut;               /* Number of mmap pages currently outstanding */
+  sqlite3_int64 szMmap;       /* Desired maximum mmap size */
+  PgHdr *pMmapFreelist;       /* List of free mmap page headers (pDirty) */
   /*
   ** 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() */
   u32 sectorSize;             /* Assumed sector size during rollback */
   int pageSize;               /* Number of bytes in a page */
   Pgno mxPgno;                /* Maximum allowed size of the database */
   i64 journalSizeLimit;       /* Size limit for persistent journal files */
   char *zFilename;            /* Name of the database file */
   char *zJournal;             /* Name of the journal file */
   int (*xBusyHandler)(void*); /* Function to call when busy */
   void *pBusyHandlerArg;      /* Context argument for xBusyHandler */
+  int aStat[3];               /* Total cache hits, misses and writes */
 #ifdef SQLITE_TEST
-  int nHit, nMiss;            /* Cache hits and missing */
-  int nRead, nWrite;          /* Database pages read/written */
+  int nRead;                  /* Database pages read */
 #endif
   void (*xReiniter)(DbPage*); /* Call this routine when reloading pages */
 #ifdef SQLITE_HAS_CODEC
   void *(*xCodec)(void*,void*,Pgno,int); /* Routine for en/decoding data */
   void (*xCodecSizeChng)(void*,int,int); /* Notify of page size changes */
   void (*xCodecFree)(void*);             /* Destructor for the codec */
   void *pCodec;               /* First argument to xCodec... methods */
 #endif
   char *pTmpSpace;            /* Pager.pageSize bytes of space for tmp use */
   PCache *pPCache;            /* Pointer to page cache object */
 #ifndef SQLITE_OMIT_WAL
   Wal *pWal;                  /* Write-ahead log used by "journal_mode=wal" */
   char *zWal;                 /* File name for write-ahead log */
 #endif
 };
 
 /*
+** Indexes for use with Pager.aStat[]. The Pager.aStat[] array contains
+** the values accessed by passing SQLITE_DBSTATUS_CACHE_HIT, CACHE_MISS 
+** or CACHE_WRITE to sqlite3_db_status().
+*/
+#define PAGER_STAT_HIT   0
+#define PAGER_STAT_MISS  1
+#define PAGER_STAT_WRITE 2
+
+/*
 ** The following global variables hold counters used for
 ** testing purposes only.  These variables do not exist in
 ** a non-testing build.  These variables are not thread-safe.
 */
 #ifdef SQLITE_TEST
 int sqlite3_pager_readdb_count = 0;    /* Number of full pages read from DB */
 int sqlite3_pager_writedb_count = 0;   /* Number of full pages written to DB */
 int sqlite3_pager_writej_count = 0;    /* Number of pages written to journal */
 # define PAGER_INCR(v)  v++
 #else
@@ skipping 47 matching lines @@
 ** the value of MEMDB will be a constant and the compiler will optimize
 ** out code that would never execute.
 */
 #ifdef SQLITE_OMIT_MEMORYDB
 # define MEMDB 0
 #else
 # define MEMDB pPager->memDb
 #endif
 
 /*
+** The macro USEFETCH is true if we are allowed to use the xFetch and xUnfetch
+** interfaces to access the database using memory-mapped I/O.
+*/
+#if SQLITE_MAX_MMAP_SIZE>0
+# define USEFETCH(x) ((x)->bUseFetch)
+#else
+# define USEFETCH(x) 0
+#endif
+
+/*
 ** The maximum legal page number is (2^31 - 1).
 */
 #define PAGER_MAX_PGNO 2147483647
 
 /*
 ** The argument to this macro is a file descriptor (type sqlite3_file*).
 ** Return 0 if it is not open, or non-zero (but not 1) if it is.
 **
 ** This is so that expressions can be written as:
 **
 **   if( isOpen(pPager->jfd) ){ ...
 **
 ** instead of
 **
 **   if( pPager->jfd->pMethods ){ ...
 */
 #define isOpen(pFd) ((pFd)->pMethods)
 
 /*
 ** Return true if this pager uses a write-ahead log instead of the usual
 ** rollback journal. Otherwise false.
 */
 #ifndef SQLITE_OMIT_WAL
 static int pagerUseWal(Pager *pPager){
   return (pPager->pWal!=0);
 }
 #else
 # define pagerUseWal(x) 0
 # define pagerRollbackWal(x) 0
-# define pagerWalFrames(v,w,x,y,z) 0
+# define pagerWalFrames(v,w,x,y) 0
 # define pagerOpenWalIfPresent(z) SQLITE_OK
 # define pagerBeginReadTransaction(z) SQLITE_OK
 #endif
 
 #ifndef NDEBUG 
 /*
 ** Usage:
 **
 **   assert( assert_pager_state(pPager) );
 **
@@ skipping 52 matching lines @@
   switch( p->eState ){
     case PAGER_OPEN:
       assert( !MEMDB );
       assert( pPager->errCode==SQLITE_OK );
       assert( sqlite3PcacheRefCount(pPager->pPCache)==0 || pPager->tempFile );
       break;
 
     case PAGER_READER:
       assert( pPager->errCode==SQLITE_OK );
       assert( p->eLock!=UNKNOWN_LOCK );
-      assert( p->eLock>=SHARED_LOCK || p->noReadlock );
+      assert( p->eLock>=SHARED_LOCK );
       break;
 
     case PAGER_WRITER_LOCKED:
       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 );
@@ skipping 110 matching lines @@
 /*
 ** Return true if it is necessary to write page *pPg into the sub-journal.
 ** A page needs to be written into the sub-journal if there exists one
 ** or more open savepoints for which:
 **
 **   * The page-number is less than or equal to PagerSavepoint.nOrig, and
 **   * The bit corresponding to the page-number is not set in
 **     PagerSavepoint.pInSavepoint.
 */
 static int subjRequiresPage(PgHdr *pPg){
+  Pager *pPager = pPg->pPager;
+  PagerSavepoint *p;
   Pgno pgno = pPg->pgno;
-  Pager *pPager = pPg->pPager;
   int i;
   for(i=0; i<pPager->nSavepoint; i++){
-    PagerSavepoint *p = &pPager->aSavepoint[i];
+    p = &pPager->aSavepoint[i];
     if( p->nOrig>=pgno && 0==sqlite3BitvecTest(p->pInSavepoint, pgno) ){
       return 1;
     }
   }
   return 0;
 }
 
 /*
 ** Return true if the page is already in the journal file.
 */
-static int pageInJournal(PgHdr *pPg){
-  return sqlite3BitvecTest(pPg->pPager->pInJournal, pPg->pgno);
+static int pageInJournal(Pager *pPager, PgHdr *pPg){
+  return sqlite3BitvecTest(pPager->pInJournal, pPg->pgno);
 }
 
 /*
 ** Read a 32-bit integer from the given file descriptor.  Store the integer
 ** that is read in *pRes.  Return SQLITE_OK if everything worked, or an
 ** error code is something goes wrong.
 **
 ** All values are stored on disk as big-endian.
 */
 static int read32bits(sqlite3_file *fd, i64 offset, u32 *pRes){
@@ skipping 31 matching lines @@
 ** UNKNOWN_LOCK for an explanation of this.
 */
 static int pagerUnlockDb(Pager *pPager, int eLock){
   int rc = SQLITE_OK;
 
   assert( !pPager->exclusiveMode || pPager->eLock==eLock );
   assert( eLock==NO_LOCK || eLock==SHARED_LOCK );
   assert( eLock!=NO_LOCK || pagerUseWal(pPager)==0 );
   if( isOpen(pPager->fd) ){
     assert( pPager->eLock>=eLock );
-    rc = sqlite3OsUnlock(pPager->fd, eLock);
+    rc = pPager->noLock ? SQLITE_OK : sqlite3OsUnlock(pPager->fd, eLock);
     if( pPager->eLock!=UNKNOWN_LOCK ){
       pPager->eLock = (u8)eLock;
     }
     IOTRACE(("UNLOCK %p %d\n", pPager, eLock))
   }
   return rc;
 }
 
 /*
 ** Lock the database file to level eLock, which must be either SHARED_LOCK,
 ** RESERVED_LOCK or EXCLUSIVE_LOCK. If the caller is successful, set the
 ** Pager.eLock variable to the new locking state. 
 **
 ** Except, if Pager.eLock is set to UNKNOWN_LOCK when this function is 
 ** called, do not modify it unless the new locking state is EXCLUSIVE_LOCK. 
 ** See the comment above the #define of UNKNOWN_LOCK for an explanation 
 ** of this.
 */
 static int pagerLockDb(Pager *pPager, int eLock){
   int rc = SQLITE_OK;
 
   assert( eLock==SHARED_LOCK || eLock==RESERVED_LOCK || eLock==EXCLUSIVE_LOCK );
   if( pPager->eLock<eLock || pPager->eLock==UNKNOWN_LOCK ){
-    rc = sqlite3OsLock(pPager->fd, eLock);
+    rc = pPager->noLock ? SQLITE_OK : sqlite3OsLock(pPager->fd, eLock);
     if( rc==SQLITE_OK && (pPager->eLock!=UNKNOWN_LOCK||eLock==EXCLUSIVE_LOCK) ){
       pPager->eLock = (u8)eLock;
       IOTRACE(("LOCK %p %d\n", pPager, eLock))
     }
   }
   return rc;
 }
 
 /*
 ** This function determines whether or not the atomic-write optimization
@@ skipping 110 matching lines @@
   i64 szJ;                   /* Total size in bytes of journal file pJrnl */
   u32 cksum;                 /* MJ checksum value read from journal */
   u32 u;                     /* Unsigned loop counter */
   unsigned char aMagic[8];   /* A buffer to hold the magic header */
   zMaster[0] = '\0';
 
   if( SQLITE_OK!=(rc = sqlite3OsFileSize(pJrnl, &szJ))
    || szJ<16
    || SQLITE_OK!=(rc = read32bits(pJrnl, szJ-16, &len))
    || len>=nMaster 
+   || len==0 
    || SQLITE_OK!=(rc = read32bits(pJrnl, szJ-12, &cksum))
    || SQLITE_OK!=(rc = sqlite3OsRead(pJrnl, aMagic, 8, szJ-8))
    || memcmp(aMagic, aJournalMagic, 8)
    || SQLITE_OK!=(rc = sqlite3OsRead(pJrnl, zMaster, len, szJ-16-len))
   ){
     return rc;
   }
 
   /* See if the checksum matches the master journal name */
   for(u=0; u<len; u++){
@@ skipping 157 matching lines @@
   assert( isOpen(pPager->fd) || pPager->noSync );
   if( pPager->noSync || (pPager->journalMode==PAGER_JOURNALMODE_MEMORY)
    || (sqlite3OsDeviceCharacteristics(pPager->fd)&SQLITE_IOCAP_SAFE_APPEND) 
   ){
     memcpy(zHeader, aJournalMagic, sizeof(aJournalMagic));
     put32bits(&zHeader[sizeof(aJournalMagic)], 0xffffffff);
   }else{
     memset(zHeader, 0, sizeof(aJournalMagic)+4);
   }
 
-  /* The random check-hash initialiser */ 
+  /* The random check-hash initializer */ 
   sqlite3_randomness(sizeof(pPager->cksumInit), &pPager->cksumInit);
   put32bits(&zHeader[sizeof(aJournalMagic)+4], pPager->cksumInit);
   /* The initial database size */
   put32bits(&zHeader[sizeof(aJournalMagic)+8], pPager->dbOrigSize);
   /* The assumed sector size for this process */
   put32bits(&zHeader[sizeof(aJournalMagic)+12], pPager->sectorSize);
 
   /* The page size */
   put32bits(&zHeader[sizeof(aJournalMagic)+16], pPager->pageSize);
 
@@ skipping 179 matching lines @@
   int nMaster;                     /* Length of string zMaster */
   i64 iHdrOff;                     /* Offset of header in journal file */
   i64 jrnlSize;                    /* Size of journal file on disk */
   u32 cksum = 0;                   /* Checksum of string zMaster */
 
   assert( pPager->setMaster==0 );
   assert( !pagerUseWal(pPager) );
 
   if( !zMaster 
    || pPager->journalMode==PAGER_JOURNALMODE_MEMORY 
-   || pPager->journalMode==PAGER_JOURNALMODE_OFF 
+   || !isOpen(pPager->jfd)
   ){
     return SQLITE_OK;
   }
   pPager->setMaster = 1;
-  assert( isOpen(pPager->jfd) );
   assert( pPager->journalHdr <= pPager->journalOff );
 
   /* Calculate the length in bytes and the checksum of zMaster */
   for(nMaster=0; zMaster[nMaster]; nMaster++){
     cksum += zMaster[nMaster];
   }
 
   /* If in full-sync mode, advance to the next disk sector before writing
   ** the master journal name. This is in case the previous page written to
   ** the journal has already been synced.
@@ skipping 28 matching lines @@
   */ 
   if( SQLITE_OK==(rc = sqlite3OsFileSize(pPager->jfd, &jrnlSize))
    && jrnlSize>pPager->journalOff
   ){
     rc = sqlite3OsTruncate(pPager->jfd, pPager->journalOff);
   }
   return rc;
 }
 
 /*
-** Find a page in the hash table given its page number. Return
-** a pointer to the page or NULL if the requested page is not 
-** already in memory.
-*/
-static PgHdr *pager_lookup(Pager *pPager, Pgno pgno){
-  PgHdr *p;                         /* Return value */
-
-  /* It is not possible for a call to PcacheFetch() with createFlag==0 to
-  ** fail, since no attempt to allocate dynamic memory will be made.
-  */
-  (void)sqlite3PcacheFetch(pPager->pPCache, pgno, 0, &p);
-  return p;
-}
-
-/*
 ** Discard the entire contents of the in-memory page-cache.
 */
 static void pager_reset(Pager *pPager){
   sqlite3BackupRestart(pPager->pBackup);
   sqlite3PcacheClear(pPager->pPCache);
 }
 
 /*
 ** Free all structures in the Pager.aSavepoint[] array and set both
 ** Pager.aSavepoint and Pager.nSavepoint to zero. Close the sub-journal
@@ skipping 109 matching lines @@
   ** 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;
+    if( USEFETCH(pPager) ) sqlite3OsUnfetch(pPager->fd, 0, 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.
@@ skipping 21 matching lines @@
        pPager->errCode==SQLITE_OK ||
        (pPager->errCode & 0xff)==SQLITE_IOERR
   );
   if( rc2==SQLITE_FULL || rc2==SQLITE_IOERR ){
     pPager->errCode = rc;
     pPager->eState = PAGER_ERROR;
   }
   return rc;
 }
 
+static int pager_truncate(Pager *pPager, Pgno nPage);
+
 /*
 ** This routine ends a transaction. A transaction is usually ended by 
 ** either a COMMIT or a ROLLBACK operation. This routine may be called 
 ** after rollback of a hot-journal, or if an error occurs while opening
 ** the journal file or writing the very first journal-header of a
 ** database transaction.
 ** 
 ** This routine is never called in PAGER_ERROR state. If it is called
 ** in PAGER_NONE or PAGER_SHARED state and the lock held is less
 ** exclusive than a RESERVED lock, it is a no-op.
@@ skipping 33 matching lines @@
 **
 ** SQLITE_OK is returned if no error occurs. If an error occurs during
 ** any of the IO operations to finalize the journal file or unlock the
 ** database then the IO error code is returned to the user. If the 
 ** operation to finalize the journal file fails, then the code still
 ** tries to unlock the database file if not in exclusive mode. If the
 ** unlock operation fails as well, then the first error code related
 ** to the first error encountered (the journal finalization one) is
 ** returned.
 */
-static int pager_end_transaction(Pager *pPager, int hasMaster){
+static int pager_end_transaction(Pager *pPager, int hasMaster, int bCommit){
   int rc = SQLITE_OK;      /* Error code from journal finalization operation */
   int rc2 = SQLITE_OK;     /* Error code from db file unlock operation */
 
   /* Do nothing if the pager does not have an open write transaction
   ** or at least a RESERVED lock. This function may be called when there
   ** is no write-transaction active but a RESERVED or greater lock is
   ** held under two circumstances:
   **
   **   1. After a successful hot-journal rollback, it is called with
   **      eState==PAGER_NONE and eLock==EXCLUSIVE_LOCK.
@@ skipping 16 matching lines @@
 
     /* Finalize the journal file. */
     if( sqlite3IsMemJournal(pPager->jfd) ){
       assert( pPager->journalMode==PAGER_JOURNALMODE_MEMORY );
       sqlite3OsClose(pPager->jfd);
     }else if( pPager->journalMode==PAGER_JOURNALMODE_TRUNCATE ){
       if( pPager->journalOff==0 ){
         rc = SQLITE_OK;
       }else{
         rc = sqlite3OsTruncate(pPager->jfd, 0);
+        if( rc==SQLITE_OK && pPager->fullSync ){
+          /* Make sure the new file size is written into the inode right away.
+          ** Otherwise the journal might resurrect following a power loss and
+          ** cause the last transaction to roll back.  See
+          ** https://bugzilla.mozilla.org/show_bug.cgi?id=1072773
+          */
+          rc = sqlite3OsSync(pPager->jfd, pPager->syncFlags);
+        }
       }
       pPager->journalOff = 0;
     }else if( pPager->journalMode==PAGER_JOURNALMODE_PERSIST
       || (pPager->exclusiveMode && pPager->journalMode!=PAGER_JOURNALMODE_WAL)
     ){
       rc = zeroJournalHdr(pPager, hasMaster);
       pPager->journalOff = 0;
     }else{
       /* This branch may be executed with Pager.journalMode==MEMORY if
       ** a hot-journal was just rolled back. In this case the journal
       ** file should be closed and deleted. If this connection writes to
       ** the database file, it will do so using an in-memory journal. 
       */
+      int bDelete = (!pPager->tempFile && sqlite3JournalExists(pPager->jfd));
       assert( pPager->journalMode==PAGER_JOURNALMODE_DELETE 
            || pPager->journalMode==PAGER_JOURNALMODE_MEMORY 
            || pPager->journalMode==PAGER_JOURNALMODE_WAL 
       );
       sqlite3OsClose(pPager->jfd);
-      if( !pPager->tempFile ){
+      if( bDelete ){
         rc = sqlite3OsDelete(pPager->pVfs, pPager->zJournal, 0);
       }
     }
   }
 
 #ifdef SQLITE_CHECK_PAGES
   sqlite3PcacheIterateDirty(pPager->pPCache, pager_set_pagehash);
   if( pPager->dbSize==0 && sqlite3PcacheRefCount(pPager->pPCache)>0 ){
-    PgHdr *p = pager_lookup(pPager, 1);
+    PgHdr *p = sqlite3PagerLookup(pPager, 1);
     if( p ){
       p->pageHash = 0;
-      sqlite3PagerUnref(p);
+      sqlite3PagerUnrefNotNull(p);
     }
   }
 #endif
 
   sqlite3BitvecDestroy(pPager->pInJournal);
   pPager->pInJournal = 0;
   pPager->nRec = 0;
   sqlite3PcacheCleanAll(pPager->pPCache);
   sqlite3PcacheTruncate(pPager->pPCache, pPager->dbSize);
 
   if( pagerUseWal(pPager) ){
     /* Drop the WAL write-lock, if any. Also, if the connection was in 
     ** locking_mode=exclusive mode but is no longer, drop the EXCLUSIVE 
     ** lock held on the database file.
     */
     rc2 = sqlite3WalEndWriteTransaction(pPager->pWal);
     assert( rc2==SQLITE_OK );
+  }else if( rc==SQLITE_OK && bCommit && pPager->dbFileSize>pPager->dbSize ){
+    /* This branch is taken when committing a transaction in rollback-journal
+    ** mode if the database file on disk is larger than the database image.
+    ** At this point the journal has been finalized and the transaction 
+    ** successfully committed, but the EXCLUSIVE lock is still held on the
+    ** file. So it is safe to truncate the database file to its minimum
+    ** required size.  */
+    assert( pPager->eLock==EXCLUSIVE_LOCK );
+    rc = pager_truncate(pPager, pPager->dbSize);
   }
+
+  if( rc==SQLITE_OK && bCommit && isOpen(pPager->fd) ){
+    rc = sqlite3OsFileControl(pPager->fd, SQLITE_FCNTL_COMMIT_PHASETWO, 0);
+    if( rc==SQLITE_NOTFOUND ) rc = SQLITE_OK;
+  }
+
   if( !pPager->exclusiveMode 
    && (!pagerUseWal(pPager) || sqlite3WalExclusiveMode(pPager->pWal, 0))
   ){
     rc2 = pagerUnlockDb(pPager, SHARED_LOCK);
     pPager->changeCountDone = 0;
   }
   pPager->eState = PAGER_READER;
   pPager->setMaster = 0;
 
   return (rc==SQLITE_OK?rc2:rc);
@@ skipping 18 matching lines @@
 */
 static void pagerUnlockAndRollback(Pager *pPager){
   if( pPager->eState!=PAGER_ERROR && pPager->eState!=PAGER_OPEN ){
     assert( assert_pager_state(pPager) );
     if( pPager->eState>=PAGER_WRITER_LOCKED ){
       sqlite3BeginBenignMalloc();
       sqlite3PagerRollback(pPager);
       sqlite3EndBenignMalloc();
     }else if( !pPager->exclusiveMode ){
       assert( pPager->eState==PAGER_READER );
-      pager_end_transaction(pPager, 0);
+      pager_end_transaction(pPager, 0, 0);
     }
   }
   pager_unlock(pPager);
 }
 
 /*
 ** Parameter aData must point to a buffer of pPager->pageSize bytes
 ** of data. Compute and return a checksum based ont the contents of the 
 ** page of data and the current value of pPager->cksumInit.
 **
@@ skipping 183 matching lines @@
   ** in the main journal either because the page is not in cache or else
   ** the page is marked as needSync==0.
   **
   ** 2008-04-14:  When attempting to vacuum a corrupt database file, it
   ** is possible to fail a statement on a database that does not yet exist.
   ** Do not attempt to write if database file has never been opened.
   */
   if( pagerUseWal(pPager) ){
     pPg = 0;
   }else{
-    pPg = pager_lookup(pPager, pgno);
+    pPg = sqlite3PagerLookup(pPager, pgno);
   }
   assert( pPg || !MEMDB );
   assert( pPager->eState!=PAGER_OPEN || pPg==0 );
   PAGERTRACE(("PLAYBACK %d page %d hash(%08x) %s\n",
            PAGERID(pPager), pgno, pager_datahash(pPager->pageSize, (u8*)aData),
            (isMainJrnl?"main-journal":"sub-journal")
   ));
   if( isMainJrnl ){
     isSynced = pPager->noSync || (*pOffset <= pPager->journalHdr);
   }else{
     isSynced = (pPg==0 || 0==(pPg->flags & PGHDR_NEED_SYNC));
   }
   if( isOpen(pPager->fd)
    && (pPager->eState>=PAGER_WRITER_DBMOD || pPager->eState==PAGER_OPEN)
    && isSynced
   ){
     i64 ofst = (pgno-1)*(i64)pPager->pageSize;
     testcase( !isSavepnt && pPg!=0 && (pPg->flags&PGHDR_NEED_SYNC)!=0 );
     assert( !pagerUseWal(pPager) );
-    rc = sqlite3OsWrite(pPager->fd, (u8*)aData, pPager->pageSize, ofst);
+    rc = sqlite3OsWrite(pPager->fd, (u8 *)aData, pPager->pageSize, ofst);
     if( pgno>pPager->dbFileSize ){
       pPager->dbFileSize = pgno;
     }
     if( pPager->pBackup ){
       CODEC1(pPager, aData, pgno, 3, rc=SQLITE_NOMEM);
       sqlite3BackupUpdate(pPager->pBackup, pgno, (u8*)aData);
       CODEC2(pPager, aData, pgno, 7, rc=SQLITE_NOMEM, aData);
     }
   }else if( !isMainJrnl && pPg==0 ){
     /* If this is a rollback of a savepoint and data was not written to
     ** the database and the page is not in-memory, there is a potential
     ** problem. When the page is next fetched by the b-tree layer, it 
     ** will be read from the database file, which may or may not be 
     ** current. 
     **
     ** There are a couple of different ways this can happen. All are quite
     ** obscure. When running in synchronous mode, this can only happen 
     ** if the page is on the free-list at the start of the transaction, then
     ** populated, then moved using sqlite3PagerMovepage().
     **
     ** The solution is to add an in-memory page to the cache containing
     ** the data just read from the sub-journal. Mark the page as dirty 
     ** and if the pager requires a journal-sync, then mark the page as 
     ** requiring a journal-sync before it is written.
     */
     assert( isSavepnt );
-    assert( pPager->doNotSpill==0 );
-    pPager->doNotSpill++;
+    assert( (pPager->doNotSpill & SPILLFLAG_ROLLBACK)==0 );
+    pPager->doNotSpill |= SPILLFLAG_ROLLBACK;
     rc = sqlite3PagerAcquire(pPager, pgno, &pPg, 1);
-    assert( pPager->doNotSpill==1 );
-    pPager->doNotSpill--;
+    assert( (pPager->doNotSpill & SPILLFLAG_ROLLBACK)!=0 );
+    pPager->doNotSpill &= ~SPILLFLAG_ROLLBACK;
     if( rc!=SQLITE_OK ) return rc;
     pPg->flags &= ~PGHDR_NEED_READ;
     sqlite3PcacheMakeDirty(pPg);
   }
   if( pPg ){
     /* No page should ever be explicitly rolled back that is in use, except
     ** for page 1 which is held in use in order to keep the lock on the
     ** database active. However such a page may be rolled back as a result
     ** of an internal error resulting in an automatic call to
     ** sqlite3PagerRollback().
@@ skipping 108 matching lines @@
   if( rc!=SQLITE_OK ) goto delmaster_out;
 
   /* Load the entire master journal file into space obtained from
   ** sqlite3_malloc() and pointed to by zMasterJournal.   Also obtain
   ** sufficient space (in zMasterPtr) to hold the names of master
   ** journal files extracted from regular rollback-journals.
   */
   rc = sqlite3OsFileSize(pMaster, &nMasterJournal);
   if( rc!=SQLITE_OK ) goto delmaster_out;
   nMasterPtr = pVfs->mxPathname+1;
-  zMasterJournal = sqlite3Malloc((int)nMasterJournal + nMasterPtr + 1);
+  zMasterJournal = sqlite3Malloc(nMasterJournal + nMasterPtr + 1);
   if( !zMasterJournal ){
     rc = SQLITE_NOMEM;
     goto delmaster_out;
   }
   zMasterPtr = &zMasterJournal[nMasterJournal+1];
   rc = sqlite3OsRead(pMaster, zMasterJournal, (int)nMasterJournal, 0);
   if( rc!=SQLITE_OK ) goto delmaster_out;
   zMasterJournal[nMasterJournal] = 0;
 
   zJournal = zMasterJournal;
@@ skipping 48 matching lines @@
 ** This function is used to change the actual size of the database 
 ** file in the file-system. This only happens when committing a transaction,
 ** or rolling back a transaction (including rolling back a hot-journal).
 **
 ** If the main database file is not open, or the pager is not in either
 ** DBMOD or OPEN state, this function is a no-op. Otherwise, the size 
 ** of the file is changed to nPage pages (nPage*pPager->pageSize bytes). 
 ** If the file on disk is currently larger than nPage pages, then use the VFS
 ** xTruncate() method to truncate it.
 **
-** Or, it might might be the case that the file on disk is smaller than 
+** Or, it might be the case that the file on disk is smaller than 
 ** nPage pages. Some operating system implementations can get confused if 
 ** you try to truncate a file to some size that is larger than it 
 ** currently is, so detect this case and write a single zero byte to 
 ** the end of the new file instead.
 **
 ** If successful, return SQLITE_OK. If an IO error occurs while modifying
 ** the database file, return the error code to the caller.
 */
 static int pager_truncate(Pager *pPager, Pgno nPage){
   int rc = SQLITE_OK;
   assert( pPager->eState!=PAGER_ERROR );
   assert( pPager->eState!=PAGER_READER );
   
   if( isOpen(pPager->fd) 
    && (pPager->eState>=PAGER_WRITER_DBMOD || pPager->eState==PAGER_OPEN) 
   ){
     i64 currentSize, newSize;
     int szPage = pPager->pageSize;
     assert( pPager->eLock==EXCLUSIVE_LOCK );
     /* TODO: Is it safe to use Pager.dbFileSize here? */
     rc = sqlite3OsFileSize(pPager->fd, &currentSize);
     newSize = szPage*(i64)nPage;
     if( rc==SQLITE_OK && currentSize!=newSize ){
       if( currentSize>newSize ){
         rc = sqlite3OsTruncate(pPager->fd, newSize);
-      }else{
+      }else if( (currentSize+szPage)<=newSize ){
         char *pTmp = pPager->pTmpSpace;
         memset(pTmp, 0, szPage);
-        testcase( (newSize-szPage) <  currentSize );
         testcase( (newSize-szPage) == currentSize );
         testcase( (newSize-szPage) >  currentSize );
         rc = sqlite3OsWrite(pPager->fd, pTmp, szPage, newSize-szPage);
       }
       if( rc==SQLITE_OK ){
         pPager->dbFileSize = nPage;
       }
     }
   }
   return rc;
 }
 
 /*
+** Return a sanitized version of the sector-size of OS file pFile. The
+** return value is guaranteed to lie between 32 and MAX_SECTOR_SIZE.
+*/
+int sqlite3SectorSize(sqlite3_file *pFile){
+  int iRet = sqlite3OsSectorSize(pFile);
+  if( iRet<32 ){
+    iRet = 512;
+  }else if( iRet>MAX_SECTOR_SIZE ){
+    assert( MAX_SECTOR_SIZE>=512 );
+    iRet = MAX_SECTOR_SIZE;
+  }
+  return iRet;
+}
+
+/*
 ** Set the value of the Pager.sectorSize variable for the given
 ** pager based on the value returned by the xSectorSize method
-** of the open database file. The sector size will be used used 
+** of the open database file. The sector size will be used 
 ** to determine the size and alignment of journal header and 
 ** master journal pointers within created journal files.
 **
 ** For temporary files the effective sector size is always 512 bytes.
 **
 ** Otherwise, for non-temporary files, the effective sector size is
 ** the value returned by the xSectorSize() method rounded up to 32 if
 ** it is less than 32, or rounded down to MAX_SECTOR_SIZE if it
 ** is greater than MAX_SECTOR_SIZE.
+**
+** If the file has the SQLITE_IOCAP_POWERSAFE_OVERWRITE property, then set
+** the effective sector size to its minimum value (512).  The purpose of
+** pPager->sectorSize is to define the "blast radius" of bytes that
+** might change if a crash occurs while writing to a single byte in
+** that range.  But with POWERSAFE_OVERWRITE, the blast radius is zero
+** (that is what POWERSAFE_OVERWRITE means), so we minimize the sector
+** size.  For backwards compatibility of the rollback journal file format,
+** we cannot reduce the effective sector size below 512.
 */
 static void setSectorSize(Pager *pPager){
   assert( isOpen(pPager->fd) || pPager->tempFile );
 
-  if( !pPager->tempFile ){
+  if( pPager->tempFile
+   || (sqlite3OsDeviceCharacteristics(pPager->fd) & 
+              SQLITE_IOCAP_POWERSAFE_OVERWRITE)!=0
+  ){
     /* Sector size doesn't matter for temporary files. Also, the file
     ** may not have been opened yet, in which case the OsSectorSize()
-    ** call will segfault.
-    */
-    pPager->sectorSize = sqlite3OsSectorSize(pPager->fd);
-  }
-  if( pPager->sectorSize<32 ){
+    ** call will segfault. */
     pPager->sectorSize = 512;
-  }
-  if( pPager->sectorSize>MAX_SECTOR_SIZE ){
-    assert( MAX_SECTOR_SIZE>=512 );
-    pPager->sectorSize = MAX_SECTOR_SIZE;
+  }else{
+    pPager->sectorSize = sqlite3SectorSize(pPager->fd);
   }
 }
 
 /*
 ** Playback the journal and thus restore the database file to
 ** the state it was in before we started making changes.  
 **
 ** The journal file format is as follows: 
 **
 **  (1)  8 byte prefix.  A copy of aJournalMagic[].
@@ skipping 50 matching lines @@
 static int pager_playback(Pager *pPager, int isHot){
   sqlite3_vfs *pVfs = pPager->pVfs;
   i64 szJ;                 /* Size of the journal file in bytes */
   u32 nRec;                /* Number of Records in the journal */
   u32 u;                   /* Unsigned loop counter */
   Pgno mxPg = 0;           /* Size of the original file in pages */
   int rc;                  /* Result code of a subroutine */
   int res = 1;             /* Value returned by sqlite3OsAccess() */
   char *zMaster = 0;       /* Name of master journal file if any */
   int needPagerReset;      /* True to reset page prior to first page rollback */
+  int nPlayback = 0;       /* Total number of pages restored from journal */
 
   /* Figure out how many records are in the journal.  Abort early if
   ** the journal is empty.
   */
   assert( isOpen(pPager->jfd) );
   rc = sqlite3OsFileSize(pPager->jfd, &szJ);
   if( rc!=SQLITE_OK ){
     goto end_playback;
   }
 
@@ skipping 80 matching lines @@
 
     /* Copy original pages out of the journal and back into the 
     ** database file and/or page cache.
     */
     for(u=0; u<nRec; u++){
       if( needPagerReset ){
         pager_reset(pPager);
         needPagerReset = 0;
       }
       rc = pager_playback_one_page(pPager,&pPager->journalOff,0,1,0);
-      if( rc!=SQLITE_OK ){
+      if( rc==SQLITE_OK ){
+        nPlayback++;
+      }else{
         if( rc==SQLITE_DONE ){
-          rc = SQLITE_OK;
           pPager->journalOff = szJ;
           break;
         }else if( rc==SQLITE_IOERR_SHORT_READ ){
           /* If the journal has been truncated, simply stop reading and
           ** processing the journal. This might happen if the journal was
           ** not completely written and synced prior to a crash.  In that
           ** case, the database should have never been written in the
           ** first place so it is OK to simply abandon the rollback. */
           rc = SQLITE_OK;
           goto end_playback;
@@ skipping 10 matching lines @@
   }
   /*NOTREACHED*/
   assert( 0 );
 
 end_playback:
   /* Following a rollback, the database file should be back in its original
   ** state prior to the start of the transaction, so invoke the
   ** SQLITE_FCNTL_DB_UNCHANGED file-control method to disable the
   ** assertion that the transaction counter was modified.
   */
-  assert(
-    pPager->fd->pMethods==0 ||
-    sqlite3OsFileControl(pPager->fd,SQLITE_FCNTL_DB_UNCHANGED,0)>=SQLITE_OK
-  );
+#ifdef SQLITE_DEBUG
+  if( pPager->fd->pMethods ){
+    sqlite3OsFileControlHint(pPager->fd,SQLITE_FCNTL_DB_UNCHANGED,0);
+  }
+#endif
 
   /* If this playback is happening automatically as a result of an IO or 
   ** malloc error that occurred after the change-counter was updated but 
   ** before the transaction was committed, then the change-counter 
   ** modification may just have been reverted. If this happens in exclusive 
   ** mode, then subsequent transactions performed by the connection will not
   ** update the change-counter at all. This may lead to cache inconsistency
   ** problems for other processes at some point in the future. So, just
   ** in case this has happened, clear the changeCountDone flag now.
   */
   pPager->changeCountDone = pPager->tempFile;
 
   if( rc==SQLITE_OK ){
     zMaster = pPager->pTmpSpace;
     rc = readMasterJournal(pPager->jfd, zMaster, pPager->pVfs->mxPathname+1);
     testcase( rc!=SQLITE_OK );
   }
   if( rc==SQLITE_OK
    && (pPager->eState>=PAGER_WRITER_DBMOD || pPager->eState==PAGER_OPEN)
   ){
-    rc = sqlite3PagerSync(pPager);
+    rc = sqlite3PagerSync(pPager, 0);
   }
   if( rc==SQLITE_OK ){
-    rc = pager_end_transaction(pPager, zMaster[0]!='\0');
+    rc = pager_end_transaction(pPager, zMaster[0]!='\0', 0);
     testcase( rc!=SQLITE_OK );
   }
   if( rc==SQLITE_OK && zMaster[0] && res ){
     /* If there was a master journal and this routine will return success,
     ** see if it is possible to delete the master journal.
     */
     rc = pager_delmaster(pPager, zMaster);
     testcase( rc!=SQLITE_OK );
   }
+  if( isHot && nPlayback ){
+    sqlite3_log(SQLITE_NOTICE_RECOVER_ROLLBACK, "recovered %d pages from %s",
+                nPlayback, pPager->zJournal);
+  }
 
   /* The Pager.sectorSize variable may have been updated while rolling
   ** back a journal created by a process with a different sector size
   ** value. Reset it to the correct value for this process.
   */
   setSectorSize(pPager);
   return rc;
 }
 
 
 /*
 ** Read the content for page pPg out of the database file and into 
 ** pPg->pData. A shared lock or greater must be held on the database
 ** file before this function is called.
 **
 ** If page 1 is read, then the value of Pager.dbFileVers[] is set to
 ** the value read from the database file.
 **
 ** If an IO error occurs, then the IO error is returned to the caller.
 ** Otherwise, SQLITE_OK is returned.
 */
-static int readDbPage(PgHdr *pPg){
+static int readDbPage(PgHdr *pPg, u32 iFrame){
   Pager *pPager = pPg->pPager; /* Pager object associated with page pPg */
   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) ){
+#ifndef SQLITE_OMIT_WAL
+  if( iFrame ){
     /* Try to pull the page from the write-ahead log. */
-    rc = sqlite3WalRead(pPager->pWal, pgno, &isInWal, pgsz, pPg->pData);
-  }
-  if( rc==SQLITE_OK && !isInWal ){
+    rc = sqlite3WalReadFrame(pPager->pWal, iFrame, pgsz, pPg->pData);
+  }else
+#endif
+  {
     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
@@ skipping 58 matching lines @@
 ** it is discarded. Otherwise, if there are one or more outstanding
 ** references, the page content is reloaded from the database. If the
 ** attempt to reload content from the database is required and fails, 
 ** return an SQLite error code. Otherwise, SQLITE_OK.
 */
 static int pagerUndoCallback(void *pCtx, Pgno iPg){
   int rc = SQLITE_OK;
   Pager *pPager = (Pager *)pCtx;
   PgHdr *pPg;
 
+  assert( pagerUseWal(pPager) );
   pPg = sqlite3PagerLookup(pPager, iPg);
   if( pPg ){
     if( sqlite3PcachePageRefcount(pPg)==1 ){
       sqlite3PcacheDrop(pPg);
     }else{
-      rc = readDbPage(pPg);
+      u32 iFrame = 0;
+      rc = sqlite3WalFindFrame(pPager->pWal, pPg->pgno, &iFrame);
+      if( rc==SQLITE_OK ){
+        rc = readDbPage(pPg, iFrame);
+      }
       if( rc==SQLITE_OK ){
         pPager->xReiniter(pPg);
       }
-      sqlite3PagerUnref(pPg);
+      sqlite3PagerUnrefNotNull(pPg);
     }
   }
 
   /* Normally, if a transaction is rolled back, any backup processes are
   ** updated as data is copied out of the rollback journal and into the
   ** database. This is not generally possible with a WAL database, as
   ** rollback involves simply truncating the log file. Therefore, if one
   ** or more frames have already been written to the log (and therefore 
   ** also copied into the backup databases) as part of this transaction,
   ** the backups must be restarted.
@@ skipping 35 matching lines @@
 ** this function notifies any active backup processes that the pages have
 ** changed. 
 **
 ** The list of pages passed into this routine is always sorted by page number.
 ** Hence, if page 1 appears anywhere on the list, it will be the first page.
 */ 
 static int pagerWalFrames(
   Pager *pPager,                  /* Pager object */
   PgHdr *pList,                   /* List of frames to log */
   Pgno nTruncate,                 /* Database size after this commit */
-  int isCommit,                   /* True if this is a commit */
-  int syncFlags                   /* Flags to pass to OsSync() (or 0) */
+  int isCommit                    /* True if this is a commit */
 ){
   int rc;                         /* Return code */
+  int nList;                      /* Number of pages in pList */
 #if defined(SQLITE_DEBUG) || defined(SQLITE_CHECK_PAGES)
   PgHdr *p;                       /* For looping over pages */
 #endif
 
   assert( pPager->pWal );
+  assert( pList );
 #ifdef SQLITE_DEBUG
   /* Verify that the page list is in accending order */
   for(p=pList; p && p->pDirty; p=p->pDirty){
     assert( p->pgno < p->pDirty->pgno );
   }
 #endif
 
+  assert( pList->pDirty==0 || isCommit );
   if( isCommit ){
     /* If a WAL transaction is being committed, there is no point in writing
     ** any pages with page numbers greater than nTruncate into the WAL file.
     ** They will never be read by any client. So remove them from the pDirty
     ** list here. */
     PgHdr *p;
     PgHdr **ppNext = &pList;
-    for(p=pList; (*ppNext = p); p=p->pDirty){
-      if( p->pgno<=nTruncate ) ppNext = &p->pDirty;
+    nList = 0;
+    for(p=pList; (*ppNext = p)!=0; p=p->pDirty){
+      if( p->pgno<=nTruncate ){
+        ppNext = &p->pDirty;
+        nList++;
+      }
     }
     assert( pList );
+  }else{
+    nList = 1;
   }
+  pPager->aStat[PAGER_STAT_WRITE] += nList;
 
   if( pList->pgno==1 ) pager_write_changecounter(pList);
   rc = sqlite3WalFrames(pPager->pWal, 
-      pPager->pageSize, pList, nTruncate, isCommit, syncFlags
+      pPager->pageSize, pList, nTruncate, isCommit, pPager->walSyncFlags
   );
   if( rc==SQLITE_OK && pPager->pBackup ){
     PgHdr *p;
     for(p=pList; p; p=p->pDirty){
       sqlite3BackupUpdate(pPager->pBackup, p->pgno, (u8 *)p->pData);
     }
   }
 
 #ifdef SQLITE_CHECK_PAGES
   pList = sqlite3PcacheDirtyList(pPager->pPCache);
@@ skipping 23 matching lines @@
   /* sqlite3WalEndReadTransaction() was not called for the previous
   ** transaction in locking_mode=EXCLUSIVE.  So call it now.  If we
   ** are in locking_mode=NORMAL and EndRead() was previously called,
   ** the duplicate call is harmless.
   */
   sqlite3WalEndReadTransaction(pPager->pWal);
 
   rc = sqlite3WalBeginReadTransaction(pPager->pWal, &changed);
   if( rc!=SQLITE_OK || changed ){
     pager_reset(pPager);
+    if( USEFETCH(pPager) ) sqlite3OsUnfetch(pPager->fd, 0, 0);
   }
 
   return rc;
 }
 #endif
 
 /*
 ** This function is called as part of the transition from PAGER_OPEN
 ** to PAGER_READER state to determine the size of the database file
 ** in pages (assuming the page size currently stored in Pager.pageSize).
 **
 ** If no error occurs, SQLITE_OK is returned and the size of the database
 ** in pages is stored in *pnPage. Otherwise, an error code (perhaps
 ** SQLITE_IOERR_FSTAT) is returned and *pnPage is left unmodified.
 */
 static int pagerPagecount(Pager *pPager, Pgno *pnPage){
   Pgno nPage;                     /* Value to return via *pnPage */
 
   /* Query the WAL sub-system for the database size. The WalDbsize()
   ** function returns zero if the WAL is not open (i.e. Pager.pWal==0), or
   ** if the database size is not available. The database size is not
   ** available from the WAL sub-system if the log file is empty or
   ** contains no valid committed transactions.
   */
   assert( pPager->eState==PAGER_OPEN );
-  assert( pPager->eLock>=SHARED_LOCK || pPager->noReadlock );
+  assert( pPager->eLock>=SHARED_LOCK );
   nPage = sqlite3WalDbsize(pPager->pWal);
 
   /* If the database size was not available from the WAL sub-system,
   ** determine it based on the size of the database file. If the size
   ** of the database file is not an integer multiple of the page-size,
   ** round down to the nearest page. Except, any file larger than 0
   ** bytes in size is considered to contain at least one page.
   */
   if( nPage==0 ){
     i64 n = 0;                    /* Size of db file in bytes */
     assert( isOpen(pPager->fd) || pPager->tempFile );
     if( isOpen(pPager->fd) ){
       int rc = sqlite3OsFileSize(pPager->fd, &n);
       if( rc!=SQLITE_OK ){
         return rc;
       }
     }
-    nPage = (Pgno)(n / pPager->pageSize);
-    if( nPage==0 && n>0 ){
-      nPage = 1;
-    }
+    nPage = (Pgno)((n+pPager->pageSize-1) / pPager->pageSize);
   }
 
   /* If the current number of pages in the file is greater than the
   ** configured maximum pager number, increase the allowed limit so
   ** that the file can be read.
   */
   if( nPage>pPager->mxPgno ){
     pPager->mxPgno = (Pgno)nPage;
   }
 
@@ skipping 16 matching lines @@
 **
 ** The caller must hold a SHARED lock on the database file to call this
 ** function. Because an EXCLUSIVE lock on the db file is required to delete 
 ** a WAL on a none-empty database, this ensures there is no race condition 
 ** between the xAccess() below and an xDelete() being executed by some 
 ** other connection.
 */
 static int pagerOpenWalIfPresent(Pager *pPager){
   int rc = SQLITE_OK;
   assert( pPager->eState==PAGER_OPEN );
-  assert( pPager->eLock>=SHARED_LOCK || pPager->noReadlock );
+  assert( pPager->eLock>=SHARED_LOCK );
 
   if( !pPager->tempFile ){
     int isWal;                    /* True if WAL file exists */
     Pgno nPage;                   /* Size of the database file */
 
     rc = pagerPagecount(pPager, &nPage);
     if( rc ) return rc;
     if( nPage==0 ){
       rc = sqlite3OsDelete(pPager->pVfs, pPager->zWal, 0);
+      if( rc==SQLITE_IOERR_DELETE_NOENT ) rc = SQLITE_OK;
       isWal = 0;
     }else{
       rc = sqlite3OsAccess(
           pPager->pVfs, pPager->zWal, SQLITE_ACCESS_EXISTS, &isWal
       );
     }
     if( rc==SQLITE_OK ){
       if( isWal ){
         testcase( sqlite3PcachePagecount(pPager->pPCache)==0 );
         rc = sqlite3PagerOpenWal(pPager, 0);
@@ skipping 123 matching lines @@
     assert( rc!=SQLITE_DONE );
   }
   assert( rc!=SQLITE_OK || pPager->journalOff>=szJ );
 
   /* Finally,  rollback pages from the sub-journal.  Page that were
   ** previously rolled back out of the main journal (and are hence in pDone)
   ** will be skipped.  Out-of-range pages are also skipped.
   */
   if( pSavepoint ){
     u32 ii;            /* Loop counter */
-    i64 offset = pSavepoint->iSubRec*(4+pPager->pageSize);
+    i64 offset = (i64)pSavepoint->iSubRec*(4+pPager->pageSize);
 
     if( pagerUseWal(pPager) ){
       rc = sqlite3WalSavepointUndo(pPager->pWal, pSavepoint->aWalData);
     }
     for(ii=pSavepoint->iSubRec; rc==SQLITE_OK && ii<pPager->nSubRec; ii++){
-      assert( offset==ii*(4+pPager->pageSize) );
+      assert( offset==(i64)ii*(4+pPager->pageSize) );
       rc = pager_playback_one_page(pPager, &offset, pDone, 0, 1);
     }
     assert( rc!=SQLITE_DONE );
   }
 
   sqlite3BitvecDestroy(pDone);
   if( rc==SQLITE_OK ){
     pPager->journalOff = szJ;
   }
 
   return rc;
 }
 
 /*
 ** Change the maximum number of in-memory pages that are allowed.
 */
 void sqlite3PagerSetCachesize(Pager *pPager, int mxPage){
   sqlite3PcacheSetCachesize(pPager->pPCache, mxPage);
 }
 
 /*
-** Adjust the robustness of the database to damage due to OS crashes
-** or power failures by changing the number of syncs()s when writing
-** the rollback journal.  There are three levels:
+** Invoke SQLITE_FCNTL_MMAP_SIZE based on the current value of szMmap.
+*/
+static void pagerFixMaplimit(Pager *pPager){
+#if SQLITE_MAX_MMAP_SIZE>0
+  sqlite3_file *fd = pPager->fd;
+  if( isOpen(fd) && fd->pMethods->iVersion>=3 ){
+    sqlite3_int64 sz;
+    sz = pPager->szMmap;
+    pPager->bUseFetch = (sz>0);
+    sqlite3OsFileControlHint(pPager->fd, SQLITE_FCNTL_MMAP_SIZE, &sz);
+  }
+#endif
+}
+
+/*
+** Change the maximum size of any memory mapping made of the database file.
+*/
+void sqlite3PagerSetMmapLimit(Pager *pPager, sqlite3_int64 szMmap){
+  pPager->szMmap = szMmap;
+  pagerFixMaplimit(pPager);
+}
+
+/*
+** Free as much memory as possible from the pager.
+*/
+void sqlite3PagerShrink(Pager *pPager){
+  sqlite3PcacheShrink(pPager->pPCache);
+}
+
+/*
+** Adjust settings of the pager to those specified in the pgFlags parameter.
+**
+** The "level" in pgFlags & PAGER_SYNCHRONOUS_MASK sets the robustness
+** of the database to damage due to OS crashes or power failures by
+** changing the number of syncs()s when writing the journals.
+** There are three levels:
 **
 **    OFF       sqlite3OsSync() is never called.  This is the default
 **              for temporary and transient files.
 **
 **    NORMAL    The journal is synced once before writes begin on the
 **              database.  This is normally adequate protection, but
 **              it is theoretically possible, though very unlikely,
 **              that an inopertune power failure could leave the journal
 **              in a state which would cause damage to the database
 **              when it is rolled back.
@@ skipping 20 matching lines @@
 ** using fcntl(F_FULLFSYNC).  SQLITE_SYNC_NORMAL means to do an
 ** ordinary fsync() call.  There is no difference between SQLITE_SYNC_FULL
 ** and SQLITE_SYNC_NORMAL on platforms other than MacOSX.  But the
 ** synchronous=FULL versus synchronous=NORMAL setting determines when
 ** the xSync primitive is called and is relevant to all platforms.
 **
 ** Numeric values associated with these states are OFF==1, NORMAL=2,
 ** and FULL=3.
 */
 #ifndef SQLITE_OMIT_PAGER_PRAGMAS
-void sqlite3PagerSetSafetyLevel(
+void sqlite3PagerSetFlags(
   Pager *pPager,        /* The pager to set safety level for */
-  int level,            /* PRAGMA synchronous.  1=OFF, 2=NORMAL, 3=FULL */  
-  int bFullFsync,       /* PRAGMA fullfsync */
-  int bCkptFullFsync    /* PRAGMA checkpoint_fullfsync */
+  unsigned pgFlags      /* Various flags */
 ){
+  unsigned level = pgFlags & PAGER_SYNCHRONOUS_MASK;
   assert( level>=1 && level<=3 );
   pPager->noSync =  (level==1 || pPager->tempFile) ?1:0;
   pPager->fullSync = (level==3 && !pPager->tempFile) ?1:0;
   if( pPager->noSync ){
     pPager->syncFlags = 0;
     pPager->ckptSyncFlags = 0;
-  }else if( bFullFsync ){
+  }else if( pgFlags & PAGER_FULLFSYNC ){
     pPager->syncFlags = SQLITE_SYNC_FULL;
     pPager->ckptSyncFlags = SQLITE_SYNC_FULL;
-  }else if( bCkptFullFsync ){
+  }else if( pgFlags & PAGER_CKPT_FULLFSYNC ){
     pPager->syncFlags = SQLITE_SYNC_NORMAL;
     pPager->ckptSyncFlags = SQLITE_SYNC_FULL;
   }else{
     pPager->syncFlags = SQLITE_SYNC_NORMAL;
     pPager->ckptSyncFlags = SQLITE_SYNC_NORMAL;
   }
+  pPager->walSyncFlags = pPager->syncFlags;
+  if( pPager->fullSync ){
+    pPager->walSyncFlags |= WAL_SYNC_TRANSACTIONS;
+  }
+  if( pgFlags & PAGER_CACHESPILL ){
+    pPager->doNotSpill &= ~SPILLFLAG_OFF;
+  }else{
+    pPager->doNotSpill |= SPILLFLAG_OFF;
+  }
 }
 #endif
 
 /*
 ** The following global variable is incremented whenever the library
 ** attempts to open a temporary file.  This information is used for
 ** testing and analysis only.  
 */
 #ifdef SQLITE_TEST
 int sqlite3_opentemp_count = 0;
@@ skipping 50 matching lines @@
 **   RESERVED_LOCK -> EXCLUSIVE_LOCK   | Yes
 **
 ** If the busy-handler callback returns non-zero, the lock is 
 ** retried. If it returns zero, then the SQLITE_BUSY error is
 ** returned to the caller of the pager API function.
 */
 void sqlite3PagerSetBusyhandler(
   Pager *pPager,                       /* Pager object */
   int (*xBusyHandler)(void *),         /* Pointer to busy-handler function */
   void *pBusyHandlerArg                /* Argument to pass to xBusyHandler */
-){  
+){
   pPager->xBusyHandler = xBusyHandler;
   pPager->pBusyHandlerArg = pBusyHandlerArg;
+
+  if( isOpen(pPager->fd) ){
+    void **ap = (void **)&pPager->xBusyHandler;
+    assert( ((int(*)(void *))(ap[0]))==xBusyHandler );
+    assert( ap[1]==pBusyHandlerArg );
+    sqlite3OsFileControlHint(pPager->fd, SQLITE_FCNTL_BUSYHANDLER, (void *)ap);
+  }
 }
 
 /*
 ** Change the page size used by the Pager object. The new page size 
 ** is passed in *pPageSize.
 **
 ** If the pager is in the error state when this function is called, it
 ** is a no-op. The value returned is the error state error code (i.e. 
 ** one of SQLITE_IOERR, an SQLITE_IOERR_xxx sub-code or SQLITE_FULL).
 **
@@ skipping 44 matching lines @@
     if( pPager->eState>PAGER_OPEN && isOpen(pPager->fd) ){
       rc = sqlite3OsFileSize(pPager->fd, &nByte);
     }
     if( rc==SQLITE_OK ){
       pNew = (char *)sqlite3PageMalloc(pageSize);
       if( !pNew ) rc = SQLITE_NOMEM;
     }
 
     if( rc==SQLITE_OK ){
       pager_reset(pPager);
-      pPager->dbSize = (Pgno)(nByte/pageSize);
-      pPager->pageSize = pageSize;
+      rc = sqlite3PcacheSetPageSize(pPager->pPCache, pageSize);
+    }
+    if( rc==SQLITE_OK ){
       sqlite3PageFree(pPager->pTmpSpace);
       pPager->pTmpSpace = pNew;
-      sqlite3PcacheSetPageSize(pPager->pPCache, pageSize);
+      pPager->dbSize = (Pgno)((nByte+pageSize-1)/pageSize);
+      pPager->pageSize = pageSize;
+    }else{
+      sqlite3PageFree(pNew);
     }
   }
 
   *pPageSize = pPager->pageSize;
   if( rc==SQLITE_OK ){
     if( nReserve<0 ) nReserve = pPager->nReserve;
     assert( nReserve>=0 && nReserve<1000 );
     pPager->nReserve = (i16)nReserve;
     pagerReportSize(pPager);
+    pagerFixMaplimit(pPager);
   }
   return rc;
 }
 
 /*
 ** Return a pointer to the "temporary page" buffer held internally
 ** by the pager.  This is a buffer that is big enough to hold the
 ** entire content of a database page.  This buffer is used internally
 ** during rollback and will be overwritten whenever a rollback
 ** occurs.  But other modules are free to use it too, as long as
@@ skipping 103 matching lines @@
 ** obtain the lock succeeds.
 **
 ** Return SQLITE_OK on success and an error code if we cannot obtain
 ** the lock. If the lock is obtained successfully, set the Pager.state 
 ** variable to locktype before returning.
 */
 static int pager_wait_on_lock(Pager *pPager, int locktype){
   int rc;                              /* Return code */
 
   /* Check that this is either a no-op (because the requested lock is 
-  ** already held, or one of the transistions that the busy-handler
+  ** already held), or one of the transitions that the busy-handler
   ** may be invoked during, according to the comment above
   ** sqlite3PagerSetBusyhandler().
   */
   assert( (pPager->eLock>=locktype)
        || (pPager->eLock==NO_LOCK && locktype==SHARED_LOCK)
        || (pPager->eLock==RESERVED_LOCK && locktype==EXCLUSIVE_LOCK)
   );
 
   do {
     rc = pagerLockDb(pPager, locktype);
   }while( rc==SQLITE_BUSY && pPager->xBusyHandler(pPager->pBusyHandlerArg) );
   return rc;
 }
 
 /*
 ** Function assertTruncateConstraint(pPager) checks that one of the 
 ** following is true for all dirty pages currently in the page-cache:
 **
 **   a) The page number is less than or equal to the size of the 
 **      current database image, in pages, OR
 **
 **   b) if the page content were written at this time, it would not
 **      be necessary to write the current content out to the sub-journal
 **      (as determined by function subjRequiresPage()).
 **
 ** If the condition asserted by this function were not true, and the
 ** dirty page were to be discarded from the cache via the pagerStress()
 ** routine, pagerStress() would not write the current page content to
 ** the database file. If a savepoint transaction were rolled back after
-** this happened, the correct behaviour would be to restore the current
+** this happened, the correct behavior would be to restore the current
 ** content of the page. However, since this content is not present in either
 ** the database file or the portion of the rollback journal and 
 ** sub-journal rolled back the content could not be restored and the
 ** database image would become corrupt. It is therefore fortunate that 
 ** this circumstance cannot arise.
 */
 #if defined(SQLITE_DEBUG)
 static void assertTruncateConstraintCb(PgHdr *pPg){
   assert( pPg->flags&PGHDR_DIRTY );
   assert( !subjRequiresPage(pPg) || pPg->pgno<=pPg->pPager->dbSize );
 }
 static void assertTruncateConstraint(Pager *pPager){
   sqlite3PcacheIterateDirty(pPager->pPCache, assertTruncateConstraintCb);
 }
 #else
 # define assertTruncateConstraint(pPager)
 #endif
 
 /*
 ** Truncate the in-memory database file image to nPage pages. This 
 ** function does not actually modify the database file on disk. It 
 ** just sets the internal state of the pager object so that the 
 ** truncation will be done when the current transaction is committed.
+**
+** This function is only called right before committing a transaction.
+** Once this function has been called, the transaction must either be
+** rolled back or committed. It is not safe to call this function and
+** then continue writing to the database.
 */
 void sqlite3PagerTruncateImage(Pager *pPager, Pgno nPage){
   assert( pPager->dbSize>=nPage );
   assert( pPager->eState>=PAGER_WRITER_CACHEMOD );
   pPager->dbSize = nPage;
-  assertTruncateConstraint(pPager);
+
+  /* At one point the code here called assertTruncateConstraint() to
+  ** ensure that all pages being truncated away by this operation are,
+  ** if one or more savepoints are open, present in the savepoint 
+  ** journal so that they can be restored if the savepoint is rolled
+  ** back. This is no longer necessary as this function is now only
+  ** called right before committing a transaction. So although the 
+  ** Pager object may still have open savepoints (Pager.nSavepoint!=0), 
+  ** they cannot be rolled back. So the assertTruncateConstraint() call
+  ** is no longer correct. */
 }
 
 
 /*
 ** This function is called before attempting a hot-journal rollback. It
 ** syncs the journal file to disk, then sets pPager->journalHdr to the
 ** size of the journal file so that the pager_playback() routine knows
 ** that the entire journal file has been synced.
 **
 ** Syncing a hot-journal to disk before attempting to roll it back ensures 
 ** that if a power-failure occurs during the rollback, the process that
 ** attempts rollback following system recovery sees the same journal
 ** content as this process.
 **
 ** If everything goes as planned, SQLITE_OK is returned. Otherwise, 
 ** an SQLite error code.
 */
 static int pagerSyncHotJournal(Pager *pPager){
   int rc = SQLITE_OK;
   if( !pPager->noSync ){
     rc = sqlite3OsSync(pPager->jfd, SQLITE_SYNC_NORMAL);
   }
   if( rc==SQLITE_OK ){
     rc = sqlite3OsFileSize(pPager->jfd, &pPager->journalHdr);
   }
   return rc;
 }
 
 /*
+** Obtain a reference to a memory mapped page object for page number pgno. 
+** The new object will use the pointer pData, obtained from xFetch().
+** If successful, set *ppPage to point to the new page reference
+** and return SQLITE_OK. Otherwise, return an SQLite error code and set
+** *ppPage to zero.
+**
+** Page references obtained by calling this function should be released
+** by calling pagerReleaseMapPage().
+*/
+static int pagerAcquireMapPage(
+  Pager *pPager,                  /* Pager object */
+  Pgno pgno,                      /* Page number */
+  void *pData,                    /* xFetch()'d data for this page */
+  PgHdr **ppPage                  /* OUT: Acquired page object */
+){
+  PgHdr *p;                       /* Memory mapped page to return */
+
+  if( pPager->pMmapFreelist ){
+    *ppPage = p = pPager->pMmapFreelist;
+    pPager->pMmapFreelist = p->pDirty;
+    p->pDirty = 0;
+    memset(p->pExtra, 0, pPager->nExtra);
+  }else{
+    *ppPage = p = (PgHdr *)sqlite3MallocZero(sizeof(PgHdr) + pPager->nExtra);
+    if( p==0 ){
+      sqlite3OsUnfetch(pPager->fd, (i64)(pgno-1) * pPager->pageSize, pData);
+      return SQLITE_NOMEM;
+    }
+    p->pExtra = (void *)&p[1];
+    p->flags = PGHDR_MMAP;
+    p->nRef = 1;
+    p->pPager = pPager;
+  }
+
+  assert( p->pExtra==(void *)&p[1] );
+  assert( p->pPage==0 );
+  assert( p->flags==PGHDR_MMAP );
+  assert( p->pPager==pPager );
+  assert( p->nRef==1 );
+
+  p->pgno = pgno;
+  p->pData = pData;
+  pPager->nMmapOut++;
+
+  return SQLITE_OK;
+}
+
+/*
+** Release a reference to page pPg. pPg must have been returned by an 
+** earlier call to pagerAcquireMapPage().
+*/
+static void pagerReleaseMapPage(PgHdr *pPg){
+  Pager *pPager = pPg->pPager;
+  pPager->nMmapOut--;
+  pPg->pDirty = pPager->pMmapFreelist;
+  pPager->pMmapFreelist = pPg;
+
+  assert( pPager->fd->pMethods->iVersion>=3 );
+  sqlite3OsUnfetch(pPager->fd, (i64)(pPg->pgno-1)*pPager->pageSize, pPg->pData);
+}
+
+/*
+** Free all PgHdr objects stored in the Pager.pMmapFreelist list.
+*/
+static void pagerFreeMapHdrs(Pager *pPager){
+  PgHdr *p;
+  PgHdr *pNext;
+  for(p=pPager->pMmapFreelist; p; p=pNext){
+    pNext = p->pDirty;
+    sqlite3_free(p);
+  }
+}
+
+
+/*
 ** Shutdown the page cache.  Free all memory and close all files.
 **
 ** If a transaction was in progress when this routine is called, that
 ** transaction is rolled back.  All outstanding pages are invalidated
 ** and their memory is freed.  Any attempt to use a page associated
 ** with this page cache after this function returns will likely
 ** result in a coredump.
 **
 ** This function always succeeds. If a transaction is active an attempt
 ** is made to roll it back. If an error occurs during the rollback 
 ** a hot journal may be left in the filesystem but no error is returned
 ** to the caller.
 */
 int sqlite3PagerClose(Pager *pPager){
   u8 *pTmp = (u8 *)pPager->pTmpSpace;
 
+  assert( assert_pager_state(pPager) );
   disable_simulated_io_errors();
   sqlite3BeginBenignMalloc();
+  pagerFreeMapHdrs(pPager);
   /* pPager->errCode = 0; */
   pPager->exclusiveMode = 0;
 #ifndef SQLITE_OMIT_WAL
   sqlite3WalClose(pPager->pWal, pPager->ckptSyncFlags, pPager->pageSize, pTmp);
   pPager->pWal = 0;
 #endif
   pager_reset(pPager);
   if( MEMDB ){
     pager_unlock(pPager);
   }else{
@@ skipping 49 matching lines @@
   sqlite3PcacheRef(pPg);
 }
 
 /*
 ** Sync the journal. In other words, make sure all the pages that have
 ** been written to the journal have actually reached the surface of the
 ** disk and can be restored in the event of a hot-journal rollback.
 **
 ** If the Pager.noSync flag is set, then this function is a no-op.
 ** Otherwise, the actions required depend on the journal-mode and the 
-** device characteristics of the the file-system, as follows:
+** device characteristics of the file-system, as follows:
 **
 **   * If the journal file is an in-memory journal file, no action need
 **     be taken.
 **
 **   * Otherwise, if the device does not support the SAFE_APPEND property,
 **     then the nRec field of the most recently written journal header
 **     is updated to contain the number of journal records that have
 **     been written following it. If the pager is operating in full-sync
 **     mode, then the journal file is synced before this field is updated.
 **
@@ skipping 171 matching lines @@
   */
   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 ){
+  if( rc==SQLITE_OK 
+   && pPager->dbHintSize<pPager->dbSize
+   && (pList->pDirty || pList->pgno>pPager->dbHintSize)
+  ){
     sqlite3_int64 szFile = pPager->pageSize * (sqlite3_int64)pPager->dbSize;
-    sqlite3OsFileControl(pPager->fd, SQLITE_FCNTL_SIZE_HINT, &szFile);
+    sqlite3OsFileControlHint(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
     ** make the file smaller (presumably by auto-vacuum code). Do not write
     ** any such pages to the file.
@@ skipping 17 matching lines @@
       /* If page 1 was just written, update Pager.dbFileVers to match
       ** the value now stored in the database file. If writing this 
       ** page caused the database file to grow, update dbFileSize. 
       */
       if( pgno==1 ){
         memcpy(&pPager->dbFileVers, &pData[24], sizeof(pPager->dbFileVers));
       }
       if( pgno>pPager->dbFileSize ){
         pPager->dbFileSize = pgno;
       }
+      pPager->aStat[PAGER_STAT_WRITE]++;
 
       /* Update any backup objects copying the contents of this pager. */
       sqlite3BackupUpdate(pPager->pBackup, pgno, (u8*)pList->pData);
 
       PAGERTRACE(("STORE %d page %d hash(%08x)\n",
                    PAGERID(pPager), pgno, pager_pagehash(pList)));
       IOTRACE(("PGOUT %p %d\n", pPager, pgno));
       PAGER_INCR(sqlite3_pager_writedb_count);
-      PAGER_INCR(pPager->nWrite);
     }else{
       PAGERTRACE(("NOSTORE %d page %d\n", PAGERID(pPager), pgno));
     }
     pager_set_pagehash(pList);
     pList = pList->pDirty;
   }
 
   return rc;
 }
 
@@ skipping 33 matching lines @@
 static int subjournalPage(PgHdr *pPg){
   int rc = SQLITE_OK;
   Pager *pPager = pPg->pPager;
   if( pPager->journalMode!=PAGER_JOURNALMODE_OFF ){
 
     /* Open the sub-journal, if it has not already been opened */
     assert( pPager->useJournal );
     assert( isOpen(pPager->jfd) || pagerUseWal(pPager) );
     assert( isOpen(pPager->sjfd) || pPager->nSubRec==0 );
     assert( pagerUseWal(pPager) 
-         || pageInJournal(pPg) 
+         || pageInJournal(pPager, pPg) 
          || pPg->pgno>pPager->dbOrigSize 
     );
     rc = openSubJournal(pPager);
 
     /* If the sub-journal was opened successfully (or was already open),
     ** write the journal record into the file.  */
     if( rc==SQLITE_OK ){
       void *pData = pPg->pData;
-      i64 offset = pPager->nSubRec*(4+pPager->pageSize);
+      i64 offset = (i64)pPager->nSubRec*(4+pPager->pageSize);
       char *pData2;
   
       CODEC2(pPager, pData, pPg->pgno, 7, return SQLITE_NOMEM, pData2);
       PAGERTRACE(("STMT-JOURNAL %d page %d\n", PAGERID(pPager), pPg->pgno));
       rc = write32bits(pPager->sjfd, offset, pPg->pgno);
       if( rc==SQLITE_OK ){
         rc = sqlite3OsWrite(pPager->sjfd, pData2, pPager->pageSize, offset+4);
       }
     }
   }
@@ skipping 24 matching lines @@
 ** made clean for some other reason, but no error occurs, then SQLITE_OK
 ** is returned by sqlite3PcacheMakeClean() is not called.
 */
 static int pagerStress(void *p, PgHdr *pPg){
   Pager *pPager = (Pager *)p;
   int rc = SQLITE_OK;
 
   assert( pPg->pPager==pPager );
   assert( pPg->flags&PGHDR_DIRTY );
 
-  /* The doNotSyncSpill flag is set during times when doing a sync of
+  /* The doNotSpill NOSYNC bit is set during times when doing a sync of
   ** journal (and adding a new header) is not allowed.  This occurs
   ** during calls to sqlite3PagerWrite() while trying to journal multiple
   ** pages belonging to the same sector.
   **
-  ** The doNotSpill flag inhibits all cache spilling regardless of whether
-  ** or not a sync is required.  This is set during a rollback.
+  ** The doNotSpill ROLLBACK and OFF bits inhibits all cache spilling
+  ** regardless of whether or not a sync is required.  This is set during
+  ** a rollback or by user request, respectively.
   **
   ** Spilling is also prohibited when in an error state since that could
-  ** lead to database corruption.   In the current implementaton it 
-  ** is impossible for sqlite3PCacheFetch() to be called with createFlag==1
+  ** lead to database corruption.   In the current implementation it 
+  ** is impossible for sqlite3PcacheFetch() to be called with createFlag==3
   ** while in the error state, hence it is impossible for this routine to
   ** be called in the error state.  Nevertheless, we include a NEVER()
   ** test for the error state as a safeguard against future changes.
   */
   if( NEVER(pPager->errCode) ) return SQLITE_OK;
-  if( pPager->doNotSpill ) return SQLITE_OK;
-  if( pPager->doNotSyncSpill && (pPg->flags & PGHDR_NEED_SYNC)!=0 ){
+  testcase( pPager->doNotSpill & SPILLFLAG_ROLLBACK );
+  testcase( pPager->doNotSpill & SPILLFLAG_OFF );
+  testcase( pPager->doNotSpill & SPILLFLAG_NOSYNC );
+  if( pPager->doNotSpill
+   && ((pPager->doNotSpill & (SPILLFLAG_ROLLBACK|SPILLFLAG_OFF))!=0
+      || (pPg->flags & PGHDR_NEED_SYNC)!=0)
+  ){
     return SQLITE_OK;
   }
 
   pPg->pDirty = 0;
   if( pagerUseWal(pPager) ){
     /* Write a single frame for this page to the log. */
     if( subjRequiresPage(pPg) ){ 
       rc = subjournalPage(pPg); 
     }
     if( rc==SQLITE_OK ){
-      rc = pagerWalFrames(pPager, pPg, 0, 0, 0);
+      rc = pagerWalFrames(pPager, pPg, 0, 0);
     }
   }else{
   
     /* Sync the journal file if required. */
     if( pPg->flags&PGHDR_NEED_SYNC 
      || pPager->eState==PAGER_WRITER_CACHEMOD
     ){
       rc = syncJournal(pPager, 1);
     }
   
@@ skipping 58 matching lines @@
 ** automatically when they are closed. If zFilename is ":memory:" then 
 ** all information is held in cache. It is never written to disk. 
 ** This can be used to implement an in-memory database.
 **
 ** The nExtra parameter specifies the number of bytes of space allocated
 ** along with each page reference. This space is available to the user
 ** via the sqlite3PagerGetExtra() API.
 **
 ** The flags argument is used to specify properties that affect the
 ** operation of the pager. It should be passed some bitwise combination
-** of the PAGER_OMIT_JOURNAL and PAGER_NO_READLOCK flags.
+** of the PAGER_* flags.
 **
 ** The vfsFlags parameter is a bitmask to pass to the flags parameter
 ** of the xOpen() method of the supplied VFS when opening files. 
 **
 ** If the pager object is allocated and the specified file opened 
 ** successfully, SQLITE_OK is returned and *ppPager set to point to
 ** the new pager object. If an error occurs, *ppPager is set to NULL
 ** and error code returned. This function may return SQLITE_NOMEM
 ** (sqlite3Malloc() is used to allocate memory), SQLITE_CANTOPEN or 
 ** various SQLITE_IO_XXX errors.
@@ skipping 10 matching lines @@
   u8 *pPtr;
   Pager *pPager = 0;       /* Pager object to allocate and return */
   int rc = SQLITE_OK;      /* Return code */
   int tempFile = 0;        /* True for temp files (incl. in-memory files) */
   int memDb = 0;           /* True if this is an in-memory file */
   int readOnly = 0;        /* True if this is a read-only file */
   int journalFileSize;     /* Bytes to allocate for each journal fd */
   char *zPathname = 0;     /* Full path to database file */
   int nPathname = 0;       /* Number of bytes in zPathname */
   int useJournal = (flags & PAGER_OMIT_JOURNAL)==0; /* False to omit journal */
-  int noReadlock = (flags & PAGER_NO_READLOCK)!=0;  /* True to omit read-lock */
   int pcacheSize = sqlite3PcacheSize();       /* Bytes to allocate for PCache */
   u32 szPageDflt = SQLITE_DEFAULT_PAGE_SIZE;  /* Default page size */
+  const char *zUri = 0;    /* URI args to copy */
+  int nUri = 0;            /* Number of bytes of URI args at *zUri */
 
   /* Figure out how much space is required for each journal file-handle
   ** (there are two of them, the main journal and the sub-journal). This
   ** is the maximum space required for an in-memory journal file handle 
   ** and a regular journal file-handle. Note that a "regular journal-handle"
   ** may be a wrapper capable of caching the first portion of the journal
   ** file in memory to implement the atomic-write optimization (see 
   ** source file journal.c).
   */
   if( sqlite3JournalSize(pVfs)>sqlite3MemJournalSize() ){
     journalFileSize = ROUND8(sqlite3JournalSize(pVfs));
   }else{
     journalFileSize = ROUND8(sqlite3MemJournalSize());
   }
 
   /* Set the output variable to NULL in case an error occurs. */
   *ppPager = 0;
 
 #ifndef SQLITE_OMIT_MEMORYDB
   if( flags & PAGER_MEMORY ){
     memDb = 1;
-    zFilename = 0;
+    if( zFilename && zFilename[0] ){
+      zPathname = sqlite3DbStrDup(0, zFilename);
+      if( zPathname==0  ) return SQLITE_NOMEM;
+      nPathname = sqlite3Strlen30(zPathname);
+      zFilename = 0;
+    }
   }
 #endif
 
   /* Compute and store the full pathname in an allocated buffer pointed
   ** to by zPathname, length nPathname. Or, if this is a temporary file,
   ** leave both nPathname and zPathname set to 0.
   */
   if( zFilename && zFilename[0] ){
+    const char *z;
     nPathname = pVfs->mxPathname+1;
-    zPathname = sqlite3Malloc(nPathname*2);
+    zPathname = sqlite3DbMallocRaw(0, nPathname*2);
     if( zPathname==0 ){
       return SQLITE_NOMEM;
     }
     zPathname[0] = 0; /* Make sure initialized even if FullPathname() fails */
     rc = sqlite3OsFullPathname(pVfs, zFilename, nPathname, zPathname);
     nPathname = sqlite3Strlen30(zPathname);
+    z = zUri = &zFilename[sqlite3Strlen30(zFilename)+1];
+    while( *z ){
+      z += sqlite3Strlen30(z)+1;
+      z += sqlite3Strlen30(z)+1;
+    }
+    nUri = (int)(&z[1] - zUri);
+    assert( nUri>=0 );
     if( rc==SQLITE_OK && nPathname+8>pVfs->mxPathname ){
       /* This branch is taken when the journal path required by
       ** the database being opened will be more than pVfs->mxPathname
       ** bytes in length. This means the database cannot be opened,
       ** as it will not be possible to open the journal file or even
       ** check for a hot-journal before reading.
       */
       rc = SQLITE_CANTOPEN_BKPT;
     }
     if( rc!=SQLITE_OK ){
-      sqlite3_free(zPathname);
+      sqlite3DbFree(0, zPathname);
       return rc;
     }
   }
 
   /* Allocate memory for the Pager structure, PCache object, the
   ** three file descriptors, the database file name and the journal 
   ** file name. The layout in memory is as follows:
   **
   **     Pager object                    (sizeof(Pager) bytes)
   **     PCache object                   (sqlite3PcacheSize() bytes)
   **     Database file handle            (pVfs->szOsFile bytes)
   **     Sub-journal file handle         (journalFileSize bytes)
   **     Main journal file handle        (journalFileSize bytes)
   **     Database file name              (nPathname+1 bytes)
   **     Journal file name               (nPathname+8+1 bytes)
   */
   pPtr = (u8 *)sqlite3MallocZero(
     ROUND8(sizeof(*pPager)) +      /* Pager structure */
     ROUND8(pcacheSize) +           /* PCache object */
     ROUND8(pVfs->szOsFile) +       /* The main db file */
     journalFileSize * 2 +          /* The two journal files */ 
-    nPathname + 1 +                /* zFilename */
-    nPathname + 8 + 1              /* zJournal */
+    nPathname + 1 + nUri +         /* zFilename */
+    nPathname + 8 + 2              /* zJournal */
 #ifndef SQLITE_OMIT_WAL
-    + nPathname + 4 + 1              /* zWal */
+    + nPathname + 4 + 2            /* zWal */
 #endif
   );
   assert( EIGHT_BYTE_ALIGNMENT(SQLITE_INT_TO_PTR(journalFileSize)) );
   if( !pPtr ){
-    sqlite3_free(zPathname);
+    sqlite3DbFree(0, zPathname);
     return SQLITE_NOMEM;
   }
   pPager =              (Pager*)(pPtr);
   pPager->pPCache =    (PCache*)(pPtr += ROUND8(sizeof(*pPager)));
   pPager->fd =   (sqlite3_file*)(pPtr += ROUND8(pcacheSize));
   pPager->sjfd = (sqlite3_file*)(pPtr += ROUND8(pVfs->szOsFile));
   pPager->jfd =  (sqlite3_file*)(pPtr += journalFileSize);
   pPager->zFilename =    (char*)(pPtr += journalFileSize);
   assert( EIGHT_BYTE_ALIGNMENT(pPager->jfd) );
 
   /* Fill in the Pager.zFilename and Pager.zJournal buffers, if required. */
   if( zPathname ){
     assert( nPathname>0 );
-    pPager->zJournal =   (char*)(pPtr += nPathname + 1);
+    pPager->zJournal =   (char*)(pPtr += nPathname + 1 + nUri);
     memcpy(pPager->zFilename, zPathname, nPathname);
+    if( nUri ) memcpy(&pPager->zFilename[nPathname+1], zUri, nUri);
     memcpy(pPager->zJournal, zPathname, nPathname);
-    memcpy(&pPager->zJournal[nPathname], "-journal", 8);
+    memcpy(&pPager->zJournal[nPathname], "-journal\000", 8+2);
+    sqlite3FileSuffix3(pPager->zFilename, pPager->zJournal);
 #ifndef SQLITE_OMIT_WAL
     pPager->zWal = &pPager->zJournal[nPathname+8+1];
     memcpy(pPager->zWal, zPathname, nPathname);
-    memcpy(&pPager->zWal[nPathname], "-wal", 4);
+    memcpy(&pPager->zWal[nPathname], "-wal\000", 4+1);
+    sqlite3FileSuffix3(pPager->zFilename, pPager->zWal);
 #endif
-    sqlite3_free(zPathname);
+    sqlite3DbFree(0, zPathname);
   }
   pPager->pVfs = pVfs;
   pPager->vfsFlags = vfsFlags;
 
   /* Open the pager file.
   */
   if( zFilename && zFilename[0] ){
     int fout = 0;                    /* VFS flags returned by xOpen() */
     rc = sqlite3OsOpen(pVfs, pPager->zFilename, pPager->fd, vfsFlags, &fout);
     assert( !memDb );
     readOnly = (fout&SQLITE_OPEN_READONLY);
 
     /* If the file was successfully opened for read/write access,
     ** choose a default page size in case we have to create the
     ** database file. The default page size is the maximum of:
     **
     **    + SQLITE_DEFAULT_PAGE_SIZE,
     **    + The value returned by sqlite3OsSectorSize()
     **    + The largest page size that can be written atomically.
     */
-    if( rc==SQLITE_OK && !readOnly ){
-      setSectorSize(pPager);
-      assert(SQLITE_DEFAULT_PAGE_SIZE<=SQLITE_MAX_DEFAULT_PAGE_SIZE);
-      if( szPageDflt<pPager->sectorSize ){
-        if( pPager->sectorSize>SQLITE_MAX_DEFAULT_PAGE_SIZE ){
-          szPageDflt = SQLITE_MAX_DEFAULT_PAGE_SIZE;
-        }else{
-          szPageDflt = (u32)pPager->sectorSize;
-        }
-      }
-#ifdef SQLITE_ENABLE_ATOMIC_WRITE
-      {
-        int iDc = sqlite3OsDeviceCharacteristics(pPager->fd);
-        int ii;
-        assert(SQLITE_IOCAP_ATOMIC512==(512>>8));
-        assert(SQLITE_IOCAP_ATOMIC64K==(65536>>8));
-        assert(SQLITE_MAX_DEFAULT_PAGE_SIZE<=65536);
-        for(ii=szPageDflt; ii<=SQLITE_MAX_DEFAULT_PAGE_SIZE; ii=ii*2){
-          if( iDc&(SQLITE_IOCAP_ATOMIC|(ii>>8)) ){
-            szPageDflt = ii;
+    if( rc==SQLITE_OK ){
+      int iDc = sqlite3OsDeviceCharacteristics(pPager->fd);
+      if( !readOnly ){
+        setSectorSize(pPager);
+        assert(SQLITE_DEFAULT_PAGE_SIZE<=SQLITE_MAX_DEFAULT_PAGE_SIZE);
+        if( szPageDflt<pPager->sectorSize ){
+          if( pPager->sectorSize>SQLITE_MAX_DEFAULT_PAGE_SIZE ){
+            szPageDflt = SQLITE_MAX_DEFAULT_PAGE_SIZE;
+          }else{
+            szPageDflt = (u32)pPager->sectorSize;
           }
         }
+#ifdef SQLITE_ENABLE_ATOMIC_WRITE
+        {
+          int ii;
+          assert(SQLITE_IOCAP_ATOMIC512==(512>>8));
+          assert(SQLITE_IOCAP_ATOMIC64K==(65536>>8));
+          assert(SQLITE_MAX_DEFAULT_PAGE_SIZE<=65536);
+          for(ii=szPageDflt; ii<=SQLITE_MAX_DEFAULT_PAGE_SIZE; ii=ii*2){
+            if( iDc&(SQLITE_IOCAP_ATOMIC|(ii>>8)) ){
+              szPageDflt = ii;
+            }
+          }
+        }
+#endif
       }
-#endif
+      pPager->noLock = sqlite3_uri_boolean(zFilename, "nolock", 0);
+      if( (iDc & SQLITE_IOCAP_IMMUTABLE)!=0
+       || sqlite3_uri_boolean(zFilename, "immutable", 0) ){
+          vfsFlags |= SQLITE_OPEN_READONLY;
+          goto act_like_temp_file;
+      }
     }
   }else{
     /* If a temporary file is requested, it is not opened immediately.
     ** In this case we accept the default page size and delay actually
     ** opening the file until the first call to OsWrite().
     **
     ** This branch is also run for an in-memory database. An in-memory
     ** database is the same as a temp-file that is never written out to
     ** disk and uses an in-memory rollback journal.
+    **
+    ** This branch also runs for files marked as immutable.
     */ 
+act_like_temp_file:
     tempFile = 1;
-    pPager->eState = PAGER_READER;
-    pPager->eLock = EXCLUSIVE_LOCK;
+    pPager->eState = PAGER_READER;     /* Pretend we already have a lock */
+    pPager->eLock = EXCLUSIVE_LOCK;    /* Pretend we are in EXCLUSIVE locking mode */
+    pPager->noLock = 1;                /* Do no locking */
     readOnly = (vfsFlags&SQLITE_OPEN_READONLY);
   }
 
   /* The following call to PagerSetPagesize() serves to set the value of 
   ** Pager.pageSize and to allocate the Pager.pTmpSpace buffer.
   */
   if( rc==SQLITE_OK ){
     assert( pPager->memDb==0 );
     rc = sqlite3PagerSetPagesize(pPager, &szPageDflt, -1);
     testcase( rc!=SQLITE_OK );
   }
 
-  /* If an error occurred in either of the blocks above, free the 
-  ** Pager structure and close the file.
+  /* Initialize the PCache object. */
+  if( rc==SQLITE_OK ){
+    assert( nExtra<1000 );
+    nExtra = ROUND8(nExtra);
+    rc = sqlite3PcacheOpen(szPageDflt, nExtra, !memDb,
+                           !memDb?pagerStress:0, (void *)pPager, pPager->pPCache);
+  }
+
+  /* If an error occurred above, free the  Pager structure and close the file.
   */
   if( rc!=SQLITE_OK ){
-    assert( !pPager->pTmpSpace );
     sqlite3OsClose(pPager->fd);
+    sqlite3PageFree(pPager->pTmpSpace);
     sqlite3_free(pPager);
     return rc;
   }
 
-  /* Initialize the PCache object. */
-  assert( nExtra<1000 );
-  nExtra = ROUND8(nExtra);
-  sqlite3PcacheOpen(szPageDflt, nExtra, !memDb,
-                    !memDb?pagerStress:0, (void *)pPager, pPager->pPCache);
-
   PAGERTRACE(("OPEN %d %s\n", FILEHANDLEID(pPager->fd), pPager->zFilename));
   IOTRACE(("OPEN %p %s\n", pPager, pPager->zFilename))
 
   pPager->useJournal = (u8)useJournal;
-  pPager->noReadlock = (noReadlock && readOnly) ?1:0;
   /* pPager->stmtOpen = 0; */
   /* pPager->stmtInUse = 0; */
   /* pPager->nRef = 0; */
   /* pPager->stmtSize = 0; */
   /* pPager->stmtJSize = 0; */
   /* pPager->nPage = 0; */
   pPager->mxPgno = SQLITE_MAX_PAGE_COUNT;
   /* pPager->state = PAGER_UNLOCK; */
-#if 0
-  assert( pPager->state == (tempFile ? PAGER_EXCLUSIVE : PAGER_UNLOCK) );
-#endif
   /* pPager->errMask = 0; */
   pPager->tempFile = (u8)tempFile;
   assert( tempFile==PAGER_LOCKINGMODE_NORMAL 
           || tempFile==PAGER_LOCKINGMODE_EXCLUSIVE );
   assert( PAGER_LOCKINGMODE_EXCLUSIVE==1 );
   pPager->exclusiveMode = (u8)tempFile; 
   pPager->changeCountDone = pPager->tempFile;
   pPager->memDb = (u8)memDb;
   pPager->readOnly = (u8)readOnly;
   assert( useJournal || pPager->tempFile );
   pPager->noSync = pPager->tempFile;
-  pPager->fullSync = pPager->noSync ?0:1;
-  pPager->syncFlags = pPager->noSync ? 0 : SQLITE_SYNC_NORMAL;
-  pPager->ckptSyncFlags = pPager->syncFlags;
+  if( pPager->noSync ){
+    assert( pPager->fullSync==0 );
+    assert( pPager->syncFlags==0 );
+    assert( pPager->walSyncFlags==0 );
+    assert( pPager->ckptSyncFlags==0 );
+  }else{
+    pPager->fullSync = 1;
+    pPager->syncFlags = SQLITE_SYNC_NORMAL;
+    pPager->walSyncFlags = SQLITE_SYNC_NORMAL | WAL_SYNC_TRANSACTIONS;
+    pPager->ckptSyncFlags = SQLITE_SYNC_NORMAL;
+  }
   /* pPager->pFirst = 0; */
   /* pPager->pFirstSynced = 0; */
   /* pPager->pLast = 0; */
   pPager->nExtra = (u16)nExtra;
   pPager->journalSizeLimit = SQLITE_DEFAULT_JOURNAL_SIZE_LIMIT;
   assert( isOpen(pPager->fd) || tempFile );
   setSectorSize(pPager);
   if( !useJournal ){
     pPager->journalMode = PAGER_JOURNALMODE_OFF;
   }else if( memDb ){
     pPager->journalMode = PAGER_JOURNALMODE_MEMORY;
   }
   /* pPager->xBusyHandler = 0; */
   /* pPager->pBusyHandlerArg = 0; */
   pPager->xReiniter = xReinit;
   /* memset(pPager->aHash, 0, sizeof(pPager->aHash)); */
+  /* pPager->szMmap = SQLITE_DEFAULT_MMAP_SIZE // will be set by btree.c */
 
   *ppPager = pPager;
   return SQLITE_OK;
 }
 
 
+/* Verify that the database file has not be deleted or renamed out from
+** under the pager.  Return SQLITE_OK if the database is still were it ought
+** to be on disk.  Return non-zero (SQLITE_READONLY_DBMOVED or some other error
+** code from sqlite3OsAccess()) if the database has gone missing.
+*/
+static int databaseIsUnmoved(Pager *pPager){
+  int bHasMoved = 0;
+  int rc;
+
+  if( pPager->tempFile ) return SQLITE_OK;
+  if( pPager->dbSize==0 ) return SQLITE_OK;
+  assert( pPager->zFilename && pPager->zFilename[0] );
+  rc = sqlite3OsFileControl(pPager->fd, SQLITE_FCNTL_HAS_MOVED, &bHasMoved);
+  if( rc==SQLITE_NOTFOUND ){
+    /* If the HAS_MOVED file-control is unimplemented, assume that the file
+    ** has not been moved.  That is the historical behavior of SQLite: prior to
+    ** version 3.8.3, it never checked */
+    rc = SQLITE_OK;
+  }else if( rc==SQLITE_OK && bHasMoved ){
+    rc = SQLITE_READONLY_DBMOVED;
+  }
+  return rc;
+}
+
 
 /*
 ** This function is called after transitioning from PAGER_UNLOCK to
 ** PAGER_SHARED state. It tests if there is a hot journal present in
 ** the file-system for the given pager. A hot journal is one that 
 ** needs to be played back. According to this function, a hot-journal
 ** file exists if the following criteria are met:
 **
 **   * The journal file exists in the file system, and
 **   * No process holds a RESERVED or greater lock on the database file, and
@@ skipping 45 matching lines @@
     ** call above, but then delete the journal and drop the lock before
     ** we get to the following sqlite3OsCheckReservedLock() call.  If that
     ** is the case, this routine might think there is a hot journal when
     ** in fact there is none.  This results in a false-positive which will
     ** be dealt with by the playback routine.  Ticket #3883.
     */
     rc = sqlite3OsCheckReservedLock(pPager->fd, &locked);
     if( rc==SQLITE_OK && !locked ){
       Pgno nPage;                 /* Number of pages in database file */
 
-      /* Check the size of the database file. If it consists of 0 pages,
-      ** then delete the journal file. See the header comment above for 
-      ** the reasoning here.  Delete the obsolete journal file under
-      ** a RESERVED lock to avoid race conditions and to avoid violating
-      ** [H33020].
-      */
       rc = pagerPagecount(pPager, &nPage);
       if( rc==SQLITE_OK ){
-        if( nPage==0 ){
+        /* If the database is zero pages in size, that means that either (1) the
+        ** journal is a remnant from a prior database with the same name where
+        ** the database file but not the journal was deleted, or (2) the initial
+        ** transaction that populates a new database is being rolled back.
+        ** In either case, the journal file can be deleted.  However, take care
+        ** not to delete the journal file if it is already open due to
+        ** journal_mode=PERSIST.
+        */
+        if( nPage==0 && !jrnlOpen ){
           sqlite3BeginBenignMalloc();
           if( pagerLockDb(pPager, RESERVED_LOCK)==SQLITE_OK ){
             sqlite3OsDelete(pVfs, pPager->zJournal, 0);
             if( !pPager->exclusiveMode ) pagerUnlockDb(pPager, SHARED_LOCK);
           }
           sqlite3EndBenignMalloc();
         }else{
           /* The journal file exists and no other connection has a reserved
           ** or greater lock on the database file. Now check that there is
           ** at least one non-zero bytes at the start of the journal file.
           ** If there is, then we consider this journal to be hot. If not, 
           ** it can be ignored.
           */
           if( !jrnlOpen ){
             int f = SQLITE_OPEN_READONLY|SQLITE_OPEN_MAIN_JOURNAL;
             rc = sqlite3OsOpen(pVfs, pPager->zJournal, pPager->jfd, f, &f);
           }
           if( rc==SQLITE_OK ){
             u8 first = 0;
             rc = sqlite3OsRead(pPager->jfd, (void *)&first, 1, 0);
             if( rc==SQLITE_IOERR_SHORT_READ ){
               rc = SQLITE_OK;
             }
             if( !jrnlOpen ){
               sqlite3OsClose(pPager->jfd);
             }
             *pExists = (first!=0);
           }else if( rc==SQLITE_CANTOPEN ){
             /* If we cannot open the rollback journal file in order to see if
-            ** its has a zero header, that might be due to an I/O error, or
+            ** it has a zero header, that might be due to an I/O error, or
             ** it might be due to the race condition described above and in
             ** ticket #3883.  Either way, assume that the journal is hot.
             ** This might be a false positive.  But if it is, then the
             ** automatic journal playback and recovery mechanism will deal
             ** with it under an EXCLUSIVE lock where we do not need to
             ** worry so much with race conditions.
             */
             *pExists = 1;
             rc = SQLITE_OK;
           }
@@ skipping 42 matching lines @@
   */
   assert( sqlite3PcacheRefCount(pPager->pPCache)==0 );
   assert( assert_pager_state(pPager) );
   assert( pPager->eState==PAGER_OPEN || pPager->eState==PAGER_READER );
   if( NEVER(MEMDB && pPager->errCode) ){ return pPager->errCode; }
 
   if( !pagerUseWal(pPager) && pPager->eState==PAGER_OPEN ){
     int bHotJournal = 1;          /* True if there exists a hot journal-file */
 
     assert( !MEMDB );
-    assert( pPager->noReadlock==0 || pPager->readOnly );
 
-    if( pPager->noReadlock==0 ){
-      rc = pager_wait_on_lock(pPager, SHARED_LOCK);
-      if( rc!=SQLITE_OK ){
-        assert( pPager->eLock==NO_LOCK || pPager->eLock==UNKNOWN_LOCK );
-        goto failed;
-      }
+    rc = pager_wait_on_lock(pPager, SHARED_LOCK);
+    if( rc!=SQLITE_OK ){
+      assert( pPager->eLock==NO_LOCK || pPager->eLock==UNKNOWN_LOCK );
+      goto failed;
     }
 
     /* If a journal file exists, and there is no RESERVED lock on the
     ** database file, then it either needs to be played back or deleted.
     */
     if( pPager->eLock<=SHARED_LOCK ){
       rc = hasHotJournal(pPager, &bHotJournal);
     }
     if( rc!=SQLITE_OK ){
       goto failed;
     }
     if( bHotJournal ){
+      if( pPager->readOnly ){
+        rc = SQLITE_READONLY_ROLLBACK;
+        goto failed;
+      }
+
       /* Get an EXCLUSIVE lock on the database file. At this point it is
       ** important that a RESERVED lock is not obtained on the way to the
       ** EXCLUSIVE lock. If it were, another process might open the
       ** database file, detect the RESERVED lock, and conclude that the
       ** database is safe to read while this process is still rolling the 
       ** hot-journal back.
       ** 
       ** Because the intermediate RESERVED lock is not requested, any
       ** other process attempting to access the database file will get to 
       ** this point in the code and fail to obtain its own EXCLUSIVE lock 
@@ skipping 77 matching lines @@
         pager_error(pPager, rc);
         goto failed;
       }
 
       assert( pPager->eState==PAGER_OPEN );
       assert( (pPager->eLock==SHARED_LOCK)
            || (pPager->exclusiveMode && pPager->eLock>SHARED_LOCK)
       );
     }
 
-    if( !pPager->tempFile 
-     && (pPager->pBackup || sqlite3PcachePagecount(pPager->pPCache)>0) 
-    ){
+    if( !pPager->tempFile && (
+        pPager->pBackup 
+     || sqlite3PcachePagecount(pPager->pPCache)>0 
+     || USEFETCH(pPager)
+    )){
       /* The shared-lock has just been acquired on the database file
       ** and there are already pages in the cache (from a previous
       ** read or write transaction).  Check to see if the database
       ** has been modified.  If the database has changed, flush the
       ** cache.
       **
       ** Database changes is detected by looking at 15 bytes beginning
       ** at offset 24 into the file.  The first 4 of these 16 bytes are
       ** a 32-bit counter that is incremented with each change.  The
       ** other bytes change randomly with each file change when
       ** a codec is in use.
       ** 
       ** There is a vanishingly small chance that a change will not be 
       ** detected.  The chance of an undetected change is so small that
       ** it can be neglected.
       */
       Pgno nPage = 0;
       char dbFileVers[sizeof(pPager->dbFileVers)];
 
       rc = pagerPagecount(pPager, &nPage);
       if( rc ) goto failed;
 
       if( nPage>0 ){
         IOTRACE(("CKVERS %p %d\n", pPager, sizeof(dbFileVers)));
         rc = sqlite3OsRead(pPager->fd, &dbFileVers, sizeof(dbFileVers), 24);
-        if( rc!=SQLITE_OK ){
+        if( rc!=SQLITE_OK && rc!=SQLITE_IOERR_SHORT_READ ){
           goto failed;
         }
       }else{
         memset(dbFileVers, 0, sizeof(dbFileVers));
       }
 
       if( memcmp(pPager->dbFileVers, dbFileVers, sizeof(dbFileVers))!=0 ){
         pager_reset(pPager);
+
+        /* Unmap the database file. It is possible that external processes
+        ** may have truncated the database file and then extended it back
+        ** to its original size while this process was not holding a lock.
+        ** In this case there may exist a Pager.pMap mapping that appears
+        ** to be the right size but is not actually valid. Avoid this
+        ** possibility by unmapping the db here. */
+        if( USEFETCH(pPager) ){
+          sqlite3OsUnfetch(pPager->fd, 0, 0);
+        }
       }
     }
 
     /* If there is a WAL file in the file-system, open this database in WAL
     ** mode. Otherwise, the following function call is a no-op.
     */
     rc = pagerOpenWalIfPresent(pPager);
 #ifndef SQLITE_OMIT_WAL
     assert( pPager->pWal==0 || rc==SQLITE_OK );
 #endif
@@ skipping 21 matching lines @@
 
 /*
 ** If the reference count has reached zero, rollback any active
 ** transaction and unlock the pager.
 **
 ** Except, in locking_mode=EXCLUSIVE when there is nothing to in
 ** the rollback journal, the unlock is not performed and there is
 ** nothing to rollback, so this routine is a no-op.
 */ 
 static void pagerUnlockIfUnused(Pager *pPager){
-  if( (sqlite3PcacheRefCount(pPager->pPCache)==0) ){
+  if( pPager->nMmapOut==0 && (sqlite3PcacheRefCount(pPager->pPCache)==0) ){
     pagerUnlockAndRollback(pPager);
   }
 }
 
 /*
 ** Acquire a reference to page number pgno in pager pPager (a page
 ** reference has type DbPage*). If the requested reference is 
 ** successfully obtained, it is copied to *ppPage and SQLITE_OK returned.
 **
 ** If the requested page is already in the cache, it is returned. 
 ** Otherwise, a new page object is allocated and populated with data
 ** read from the database file. In some cases, the pcache module may
 ** choose not to allocate a new page object and may reuse an existing
 ** object with no outstanding references.
 **
 ** The extra data appended to a page is always initialized to zeros the 
 ** first time a page is loaded into memory. If the page requested is 
 ** already in the cache when this function is called, then the extra
 ** data is left as it was when the page object was last used.
 **
 ** If the database image is smaller than the requested page or if a 
 ** non-zero value is passed as the noContent parameter and the 
 ** requested page is not already stored in the cache, then no 
 ** actual disk read occurs. In this case the memory image of the 
 ** page is initialized to all zeros. 
 **
 ** If noContent is true, it means that we do not care about the contents
-** of the page. This occurs in two seperate scenarios:
+** of the page. This occurs in two scenarios:
 **
 **   a) When reading a free-list leaf page from the database, and
 **
 **   b) When a savepoint is being rolled back and we need to load
 **      a new page into the cache to be filled with the data read
 **      from the savepoint journal.
 **
 ** If noContent is true, then the data returned is zeroed instead of
 ** being read from the database. Additionally, the bits corresponding
 ** to pgno in Pager.pInJournal (bitvec of pages already written to the
@@ skipping 10 matching lines @@
 ** in memory, this routine goes to disk to read it in whereas Lookup()
 ** just returns 0.  This routine acquires a read-lock the first time it
 ** has to go to disk, and could also playback an old journal if necessary.
 ** Since Lookup() never goes to disk, it never has to deal with locks
 ** or journal files.
 */
 int sqlite3PagerAcquire(
   Pager *pPager,      /* The pager open on the database file */
   Pgno pgno,          /* Page number to fetch */
   DbPage **ppPage,    /* Write a pointer to the page here */
-  int noContent       /* Do not bother reading content from disk if true */
+  int flags           /* PAGER_GET_XXX flags */
 ){
-  int rc;
-  PgHdr *pPg;
+  int rc = SQLITE_OK;
+  PgHdr *pPg = 0;
+  u32 iFrame = 0;                 /* Frame to read from WAL file */
+  const int noContent = (flags & PAGER_GET_NOCONTENT);
+
+  /* It is acceptable to use a read-only (mmap) page for any page except
+  ** page 1 if there is no write-transaction open or the ACQUIRE_READONLY
+  ** flag was specified by the caller. And so long as the db is not a 
+  ** temporary or in-memory database.  */
+  const int bMmapOk = (pgno!=1 && USEFETCH(pPager)
+   && (pPager->eState==PAGER_READER || (flags & PAGER_GET_READONLY))
+#ifdef SQLITE_HAS_CODEC
+   && pPager->xCodec==0
+#endif
+  );
 
   assert( pPager->eState>=PAGER_READER );
   assert( assert_pager_state(pPager) );
+  assert( noContent==0 || bMmapOk==0 );
 
   if( pgno==0 ){
     return SQLITE_CORRUPT_BKPT;
   }
 
   /* If the pager is in the error state, return an error immediately. 
   ** Otherwise, request the page from the PCache layer. */
   if( pPager->errCode!=SQLITE_OK ){
     rc = pPager->errCode;
   }else{
-    rc = sqlite3PcacheFetch(pPager->pPCache, pgno, 1, ppPage);
+    if( bMmapOk && pagerUseWal(pPager) ){
+      rc = sqlite3WalFindFrame(pPager->pWal, pgno, &iFrame);
+      if( rc!=SQLITE_OK ) goto pager_acquire_err;
+    }
+
+    if( bMmapOk && iFrame==0 ){
+      void *pData = 0;
+
+      rc = sqlite3OsFetch(pPager->fd, 
+          (i64)(pgno-1) * pPager->pageSize, pPager->pageSize, &pData
+      );
+
+      if( rc==SQLITE_OK && pData ){
+        if( pPager->eState>PAGER_READER ){
+          pPg = sqlite3PagerLookup(pPager, pgno);
+        }
+        if( pPg==0 ){
+          rc = pagerAcquireMapPage(pPager, pgno, pData, &pPg);
+        }else{
+          sqlite3OsUnfetch(pPager->fd, (i64)(pgno-1)*pPager->pageSize, pData);
+        }
+        if( pPg ){
+          assert( rc==SQLITE_OK );
+          *ppPage = pPg;
+          return SQLITE_OK;
+        }
+      }
+      if( rc!=SQLITE_OK ){
+        goto pager_acquire_err;
+      }
+    }
+
+    {
+      sqlite3_pcache_page *pBase;
+      pBase = sqlite3PcacheFetch(pPager->pPCache, pgno, 3);
+      if( pBase==0 ){
+        rc = sqlite3PcacheFetchStress(pPager->pPCache, pgno, &pBase);
+        if( rc!=SQLITE_OK ) goto pager_acquire_err;
+      }
+      pPg = *ppPage = sqlite3PcacheFetchFinish(pPager->pPCache, pgno, pBase);
+      if( pPg==0 ) rc = SQLITE_NOMEM;
+    }
   }
 
   if( rc!=SQLITE_OK ){
     /* Either the call to sqlite3PcacheFetch() returned an error or the
     ** pager was already in the error-state when this function was called.
     ** Set pPg to 0 and jump to the exception handler.  */
     pPg = 0;
     goto pager_acquire_err;
   }
   assert( (*ppPage)->pgno==pgno );
   assert( (*ppPage)->pPager==pPager || (*ppPage)->pPager==0 );
 
   if( (*ppPage)->pPager && !noContent ){
     /* In this case the pcache already contains an initialized copy of
     ** the page. Return without further ado.  */
     assert( pgno<=PAGER_MAX_PGNO && pgno!=PAGER_MJ_PGNO(pPager) );
-    PAGER_INCR(pPager->nHit);
+    pPager->aStat[PAGER_STAT_HIT]++;
     return SQLITE_OK;
 
   }else{
     /* The pager cache has created a new page. Its content needs to 
     ** be initialized.  */
 
-    PAGER_INCR(pPager->nMiss);
     pPg = *ppPage;
     pPg->pPager = pPager;
 
     /* The maximum page number is 2^31. Return SQLITE_CORRUPT if a page
     ** number greater than this, or the unused locking-page, is requested. */
     if( pgno>PAGER_MAX_PGNO || pgno==PAGER_MJ_PGNO(pPager) ){
       rc = SQLITE_CORRUPT_BKPT;
       goto pager_acquire_err;
     }
 
@@ skipping 14 matching lines @@
           TESTONLY( rc = ) sqlite3BitvecSet(pPager->pInJournal, pgno);
           testcase( rc==SQLITE_NOMEM );
         }
         TESTONLY( rc = ) addToSavepointBitvecs(pPager, pgno);
         testcase( rc==SQLITE_NOMEM );
         sqlite3EndBenignMalloc();
       }
       memset(pPg->pData, 0, pPager->pageSize);
       IOTRACE(("ZERO %p %d\n", pPager, pgno));
     }else{
+      if( pagerUseWal(pPager) && bMmapOk==0 ){
+        rc = sqlite3WalFindFrame(pPager->pWal, pgno, &iFrame);
+        if( rc!=SQLITE_OK ) goto pager_acquire_err;
+      }
       assert( pPg->pPager==pPager );
-      rc = readDbPage(pPg);
+      pPager->aStat[PAGER_STAT_MISS]++;
+      rc = readDbPage(pPg, iFrame);
       if( rc!=SQLITE_OK ){
         goto pager_acquire_err;
       }
     }
     pager_set_pagehash(pPg);
   }
 
   return SQLITE_OK;
 
 pager_acquire_err:
@@ skipping 12 matching lines @@
 ** not read the page from disk.  Return a pointer to the page,
 ** or 0 if the page is not in cache. 
 **
 ** See also sqlite3PagerGet().  The difference between this routine
 ** and sqlite3PagerGet() is that _get() will go to the disk and read
 ** in the page if the page is not already in cache.  This routine
 ** returns NULL if the page is not in cache or if a disk I/O error 
 ** has ever happened.
 */
 DbPage *sqlite3PagerLookup(Pager *pPager, Pgno pgno){
-  PgHdr *pPg = 0;
+  sqlite3_pcache_page *pPage;
   assert( pPager!=0 );
   assert( pgno!=0 );
   assert( pPager->pPCache!=0 );
-  assert( pPager->eState>=PAGER_READER && pPager->eState!=PAGER_ERROR );
-  sqlite3PcacheFetch(pPager->pPCache, pgno, 0, &pPg);
-  return pPg;
+  pPage = sqlite3PcacheFetch(pPager->pPCache, pgno, 0);
+  return sqlite3PcacheFetchFinish(pPager->pPCache, pgno, pPage);
 }
 
 /*
 ** Release a page reference.
 **
 ** If the number of references to the page drop to zero, then the
 ** page is added to the LRU list.  When all references to all pages
 ** are released, a rollback occurs and the lock on the database is
 ** removed.
 */
+void sqlite3PagerUnrefNotNull(DbPage *pPg){
+  Pager *pPager;
+  assert( pPg!=0 );
+  pPager = pPg->pPager;
+  if( pPg->flags & PGHDR_MMAP ){
+    pagerReleaseMapPage(pPg);
+  }else{
+    sqlite3PcacheRelease(pPg);
+  }
+  pagerUnlockIfUnused(pPager);
+}
 void sqlite3PagerUnref(DbPage *pPg){
-  if( pPg ){
-    Pager *pPager = pPg->pPager;
-    sqlite3PcacheRelease(pPg);
-    pagerUnlockIfUnused(pPager);
-  }
+  if( pPg ) sqlite3PagerUnrefNotNull(pPg);
 }
 
-#if defined(__APPLE__)
-/*
-** Create and return a CFURLRef given a cstring containing the path to a file.
-*/
-static CFURLRef create_cfurl_from_cstring(const char* filePath){
-  CFStringRef urlString = CFStringCreateWithFileSystemRepresentation(
-      kCFAllocatorDefault, filePath);
-  CFURLRef urlRef = CFURLCreateWithFileSystemPath(kCFAllocatorDefault,
-      urlString, kCFURLPOSIXPathStyle, FALSE);
-  CFRelease(urlString);
-  return urlRef;
-}
-#endif
-
 /*
 ** This function is called at the start of every write transaction.
 ** There must already be a RESERVED or EXCLUSIVE lock on the database 
 ** file when this routine is called.
 **
 ** Open the journal file for pager pPager and write a journal header
 ** to the start of it. If there are active savepoints, open the sub-journal
 ** as well. This function is only used when the journal file is being 
 ** opened to write a rollback log for a transaction. It is not used 
 ** when opening a hot journal file to roll it back.
@@ skipping 32 matching lines @@
     if( !isOpen(pPager->jfd) ){
       if( pPager->journalMode==PAGER_JOURNALMODE_MEMORY ){
         sqlite3MemJournalOpen(pPager->jfd);
       }else{
         const int flags =                   /* VFS flags to open journal file */
           SQLITE_OPEN_READWRITE|SQLITE_OPEN_CREATE|
           (pPager->tempFile ? 
             (SQLITE_OPEN_DELETEONCLOSE|SQLITE_OPEN_TEMP_JOURNAL):
             (SQLITE_OPEN_MAIN_JOURNAL)
           );
-  #ifdef SQLITE_ENABLE_ATOMIC_WRITE
-        rc = sqlite3JournalOpen(
-            pVfs, pPager->zJournal, pPager->jfd, flags, jrnlBufferSize(pPager)
-        );
-  #else
-        rc = sqlite3OsOpen(pVfs, pPager->zJournal, pPager->jfd, flags, 0);
-  #endif
-#if defined(__APPLE__)
-        /* Set the TimeMachine exclusion metadata for the journal if it has
-        ** been set for the database. Only do this for unix-type vfs
-        ** implementations. */
-        if( rc==SQLITE_OK && pPager->zFilename!=NULL
-         && strlen(pPager->zFilename)>0
-         && strncmp(pVfs->zName, "unix", 4)==0
-         && ( pVfs->zName[4]=='-' || pVfs->zName[4]=='\0' ) ){
-          CFURLRef database = create_cfurl_from_cstring(pPager->zFilename);
-          if( CSBackupIsItemExcluded(database, NULL) ){
-            CFURLRef journal = create_cfurl_from_cstring(pPager->zJournal);
-            /* Ignore errors from the following exclusion call. */
-            CSBackupSetItemExcluded(journal, TRUE, FALSE);
-            CFRelease(journal);
-          }
-          CFRelease(database);
+
+        /* Verify that the database still has the same name as it did when
+        ** it was originally opened. */
+        rc = databaseIsUnmoved(pPager);
+        if( rc==SQLITE_OK ){
+#ifdef SQLITE_ENABLE_ATOMIC_WRITE
+          rc = sqlite3JournalOpen(
+              pVfs, pPager->zJournal, pPager->jfd, flags, jrnlBufferSize(pPager)
+          );
+#else
+          rc = sqlite3OsOpen(pVfs, pPager->zJournal, pPager->jfd, flags, 0);
+#endif
         }
-#endif
       }
       assert( rc!=SQLITE_OK || isOpen(pPager->jfd) );
     }
   
   
     /* Write the first journal header to the journal file and open 
     ** the sub-journal if necessary.
     */
     if( rc==SQLITE_OK ){
       /* TODO: Check if all of these are really required. */
@@ skipping 100 matching lines @@
 }
 
 /*
 ** Mark a single data page as writeable. The page is written into the 
 ** main journal or sub-journal as required. If the page is written into
 ** one of the journals, the corresponding bit is set in the 
 ** Pager.pInJournal bitvec and the PagerSavepoint.pInSavepoint bitvecs
 ** of any open savepoints as appropriate.
 */
 static int pager_write(PgHdr *pPg){
-  void *pData = pPg->pData;
   Pager *pPager = pPg->pPager;
   int rc = SQLITE_OK;
+  int inJournal;
 
   /* This routine is not called unless a write-transaction has already 
   ** been started. The journal file may or may not be open at this point.
   ** It is never called in the ERROR state.
   */
   assert( pPager->eState==PAGER_WRITER_LOCKED
        || pPager->eState==PAGER_WRITER_CACHEMOD
        || pPager->eState==PAGER_WRITER_DBMOD
   );
   assert( assert_pager_state(pPager) );
-
-  /* If an error has been previously detected, report the same error
-  ** again. This should not happen, but the check provides robustness. */
-  if( NEVER(pPager->errCode) )  return pPager->errCode;
-
-  /* Higher-level routines never call this function if database is not
-  ** writable.  But check anyway, just for robustness. */
-  if( NEVER(pPager->readOnly) ) return SQLITE_PERM;
+  assert( pPager->errCode==0 );
+  assert( pPager->readOnly==0 );
 
   CHECK_PAGE(pPg);
 
   /* The journal file needs to be opened. Higher level routines have already
   ** obtained the necessary locks to begin the write-transaction, but the
   ** rollback journal might not yet be open. Open it now if this is the case.
   **
   ** This is done before calling sqlite3PcacheMakeDirty() on the page. 
   ** Otherwise, if it were done after calling sqlite3PcacheMakeDirty(), then
   ** an error might occur and the pager would end up in WRITER_LOCKED state
   ** with pages marked as dirty in the cache.
   */
   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) ){
+  inJournal = pageInJournal(pPager, pPg);
+  if( inJournal && (pPager->nSavepoint==0 || !subjRequiresPage(pPg)) ){
     assert( !pagerUseWal(pPager) );
   }else{
   
     /* The transaction journal now exists and we have a RESERVED or an
     ** EXCLUSIVE lock on the main database file.  Write the current page to
     ** the transaction journal if it is not there already.
     */
-    if( !pageInJournal(pPg) && !pagerUseWal(pPager) ){
+    if( !inJournal && !pagerUseWal(pPager) ){
       assert( pagerUseWal(pPager)==0 );
       if( pPg->pgno<=pPager->dbOrigSize && isOpen(pPager->jfd) ){
         u32 cksum;
         char *pData2;
         i64 iOff = pPager->journalOff;
 
         /* We should never write to the journal file the page that
         ** contains the database locks.  The following assert verifies
         ** that we do not. */
         assert( pPg->pgno!=PAGER_MJ_PGNO(pPager) );
 
         assert( pPager->journalHdr<=pPager->journalOff );
-        CODEC2(pPager, pData, pPg->pgno, 7, return SQLITE_NOMEM, pData2);
+        CODEC2(pPager, pPg->pData, pPg->pgno, 7, return SQLITE_NOMEM, pData2);
         cksum = pager_cksum(pPager, (u8*)pData2);
 
         /* Even if an IO or diskfull error occurs while journalling the
         ** page in the block above, set the need-sync flag for the page.
         ** Otherwise, when the transaction is rolled back, the logic in
         ** playback_one_page() will think that the page needs to be restored
         ** in the database file. And if an IO error occurs while doing so,
         ** then corruption may follow.
         */
         pPg->flags |= PGHDR_NEED_SYNC;
@@ skipping 31 matching lines @@
                 PAGERID(pPager), pPg->pgno,
                ((pPg->flags&PGHDR_NEED_SYNC)?1:0)));
       }
     }
   
     /* If the statement journal is open and the page is not in it,
     ** then write the current page to the statement journal.  Note that
     ** the statement journal format differs from the standard journal format
     ** in that it omits the checksums and the header.
     */
-    if( subjRequiresPage(pPg) ){
+    if( pPager->nSavepoint>0 && subjRequiresPage(pPg) ){
       rc = subjournalPage(pPg);
     }
   }
 
   /* Update the database size and return.
   */
   if( pPager->dbSize<pPg->pgno ){
     pPager->dbSize = pPg->pgno;
   }
   return rc;
 }
 
 /*
+** This is a variant of sqlite3PagerWrite() that runs when the sector size
+** is larger than the page size.  SQLite makes the (reasonable) assumption that
+** all bytes of a sector are written together by hardware.  Hence, all bytes of
+** a sector need to be journalled in case of a power loss in the middle of
+** a write.
+**
+** Usually, the sector size is less than or equal to the page size, in which
+** case pages can be individually written.  This routine only runs in the exceptional
+** case where the page size is smaller than the sector size.
+*/
+static SQLITE_NOINLINE int pagerWriteLargeSector(PgHdr *pPg){
+  int rc = SQLITE_OK;            /* Return code */
+  Pgno nPageCount;               /* Total number of pages in database file */
+  Pgno pg1;                      /* First page of the sector pPg is located on. */
+  int nPage = 0;                 /* Number of pages starting at pg1 to journal */
+  int ii;                        /* Loop counter */
+  int needSync = 0;              /* True if any page has PGHDR_NEED_SYNC */
+  Pager *pPager = pPg->pPager;   /* The pager that owns pPg */
+  Pgno nPagePerSector = (pPager->sectorSize/pPager->pageSize);
+
+  /* Set the doNotSpill NOSYNC bit to 1. This is because we cannot allow
+  ** a journal header to be written between the pages journaled by
+  ** this function.
+  */
+  assert( !MEMDB );
+  assert( (pPager->doNotSpill & SPILLFLAG_NOSYNC)==0 );
+  pPager->doNotSpill |= SPILLFLAG_NOSYNC;
+
+  /* This trick assumes that both the page-size and sector-size are
+  ** an integer power of 2. It sets variable pg1 to the identifier
+  ** of the first page of the sector pPg is located on.
+  */
+  pg1 = ((pPg->pgno-1) & ~(nPagePerSector-1)) + 1;
+
+  nPageCount = pPager->dbSize;
+  if( pPg->pgno>nPageCount ){
+    nPage = (pPg->pgno - pg1)+1;
+  }else if( (pg1+nPagePerSector-1)>nPageCount ){
+    nPage = nPageCount+1-pg1;
+  }else{
+    nPage = nPagePerSector;
+  }
+  assert(nPage>0);
+  assert(pg1<=pPg->pgno);
+  assert((pg1+nPage)>pPg->pgno);
+
+  for(ii=0; ii<nPage && rc==SQLITE_OK; ii++){
+    Pgno pg = pg1+ii;
+    PgHdr *pPage;
+    if( pg==pPg->pgno || !sqlite3BitvecTest(pPager->pInJournal, pg) ){
+      if( pg!=PAGER_MJ_PGNO(pPager) ){
+        rc = sqlite3PagerGet(pPager, pg, &pPage);
+        if( rc==SQLITE_OK ){
+          rc = pager_write(pPage);
+          if( pPage->flags&PGHDR_NEED_SYNC ){
+            needSync = 1;
+          }
+          sqlite3PagerUnrefNotNull(pPage);
+        }
+      }
+    }else if( (pPage = sqlite3PagerLookup(pPager, pg))!=0 ){
+      if( pPage->flags&PGHDR_NEED_SYNC ){
+        needSync = 1;
+      }
+      sqlite3PagerUnrefNotNull(pPage);
+    }
+  }
+
+  /* If the PGHDR_NEED_SYNC flag is set for any of the nPage pages 
+  ** starting at pg1, then it needs to be set for all of them. Because
+  ** writing to any of these nPage pages may damage the others, the
+  ** journal file must contain sync()ed copies of all of them
+  ** before any of them can be written out to the database file.
+  */
+  if( rc==SQLITE_OK && needSync ){
+    assert( !MEMDB );
+    for(ii=0; ii<nPage; ii++){
+      PgHdr *pPage = sqlite3PagerLookup(pPager, pg1+ii);
+      if( pPage ){
+        pPage->flags |= PGHDR_NEED_SYNC;
+        sqlite3PagerUnrefNotNull(pPage);
+      }
+    }
+  }
+
+  assert( (pPager->doNotSpill & SPILLFLAG_NOSYNC)!=0 );
+  pPager->doNotSpill &= ~SPILLFLAG_NOSYNC;
+  return rc;
+}
+
+/*
 ** Mark a data page as writeable. This routine must be called before 
 ** making changes to a page. The caller must check the return value 
 ** of this function and be careful not to change any page data unless 
 ** this routine returns SQLITE_OK.
 **
 ** The difference between this function and pager_write() is that this
 ** function also deals with the special case where 2 or more pages
 ** fit on a single disk sector. In this case all co-resident pages
 ** must have been written to the journal file before returning.
 **
 ** If an error occurs, SQLITE_NOMEM or an IO error code is returned
 ** as appropriate. Otherwise, SQLITE_OK.
 */
-int sqlite3PagerWrite(DbPage *pDbPage){
-  int rc = SQLITE_OK;
-
-  PgHdr *pPg = pDbPage;
-  Pager *pPager = pPg->pPager;
-  Pgno nPagePerSector = (pPager->sectorSize/pPager->pageSize);
-
-  assert( pPager->eState>=PAGER_WRITER_LOCKED );
-  assert( pPager->eState!=PAGER_ERROR );
-  assert( assert_pager_state(pPager) );
-
-  if( nPagePerSector>1 ){
-    Pgno nPageCount;          /* Total number of pages in database file */
-    Pgno pg1;                 /* First page of the sector pPg is located on. */
-    int nPage = 0;            /* Number of pages starting at pg1 to journal */
-    int ii;                   /* Loop counter */
-    int needSync = 0;         /* True if any page has PGHDR_NEED_SYNC */
-
-    /* Set the doNotSyncSpill flag to 1. This is because we cannot allow
-    ** a journal header to be written between the pages journaled by
-    ** this function.
-    */
-    assert( !MEMDB );
-    assert( pPager->doNotSyncSpill==0 );
-    pPager->doNotSyncSpill++;
-
-    /* This trick assumes that both the page-size and sector-size are
-    ** an integer power of 2. It sets variable pg1 to the identifier
-    ** of the first page of the sector pPg is located on.
-    */
-    pg1 = ((pPg->pgno-1) & ~(nPagePerSector-1)) + 1;
-
-    nPageCount = pPager->dbSize;
-    if( pPg->pgno>nPageCount ){
-      nPage = (pPg->pgno - pg1)+1;
-    }else if( (pg1+nPagePerSector-1)>nPageCount ){
-      nPage = nPageCount+1-pg1;
-    }else{
-      nPage = nPagePerSector;
-    }
-    assert(nPage>0);
-    assert(pg1<=pPg->pgno);
-    assert((pg1+nPage)>pPg->pgno);
-
-    for(ii=0; ii<nPage && rc==SQLITE_OK; ii++){
-      Pgno pg = pg1+ii;
-      PgHdr *pPage;
-      if( pg==pPg->pgno || !sqlite3BitvecTest(pPager->pInJournal, pg) ){
-        if( pg!=PAGER_MJ_PGNO(pPager) ){
-          rc = sqlite3PagerGet(pPager, pg, &pPage);
-          if( rc==SQLITE_OK ){
-            rc = pager_write(pPage);
-            if( pPage->flags&PGHDR_NEED_SYNC ){
-              needSync = 1;
-            }
-            sqlite3PagerUnref(pPage);
-          }
-        }
-      }else if( (pPage = pager_lookup(pPager, pg))!=0 ){
-        if( pPage->flags&PGHDR_NEED_SYNC ){
-          needSync = 1;
-        }
-        sqlite3PagerUnref(pPage);
-      }
-    }
-
-    /* If the PGHDR_NEED_SYNC flag is set for any of the nPage pages 
-    ** starting at pg1, then it needs to be set for all of them. Because
-    ** writing to any of these nPage pages may damage the others, the
-    ** journal file must contain sync()ed copies of all of them
-    ** before any of them can be written out to the database file.
-    */
-    if( rc==SQLITE_OK && needSync ){
-      assert( !MEMDB );
-      for(ii=0; ii<nPage; ii++){
-        PgHdr *pPage = pager_lookup(pPager, pg1+ii);
-        if( pPage ){
-          pPage->flags |= PGHDR_NEED_SYNC;
-          sqlite3PagerUnref(pPage);
-        }
-      }
-    }
-
-    assert( pPager->doNotSyncSpill==1 );
-    pPager->doNotSyncSpill--;
+int sqlite3PagerWrite(PgHdr *pPg){
+  assert( (pPg->flags & PGHDR_MMAP)==0 );
+  assert( pPg->pPager->eState>=PAGER_WRITER_LOCKED );
+  assert( pPg->pPager->eState!=PAGER_ERROR );
+  assert( assert_pager_state(pPg->pPager) );
+  if( pPg->pPager->sectorSize > (u32)pPg->pPager->pageSize ){
+    return pagerWriteLargeSector(pPg);
   }else{
-    rc = pager_write(pDbPage);
+    return pager_write(pPg);
   }
-  return rc;
 }
 
 /*
 ** 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;
@@ skipping 65 matching lines @@
   ** "if( isDirect )" condition.
   */
 #ifndef SQLITE_ENABLE_ATOMIC_WRITE
 # define DIRECT_MODE 0
   assert( isDirectMode==0 );
   UNUSED_PARAMETER(isDirectMode);
 #else
 # define DIRECT_MODE isDirectMode
 #endif
 
-  if( !pPager->changeCountDone && pPager->dbSize>0 ){
+  if( !pPager->changeCountDone && ALWAYS(pPager->dbSize>0) ){
     PgHdr *pPgHdr;                /* Reference to page 1 */
 
     assert( !pPager->tempFile && isOpen(pPager->fd) );
 
     /* Open page 1 of the file for writing. */
     rc = sqlite3PagerGet(pPager, 1, &pPgHdr);
     assert( pPgHdr==0 || rc==SQLITE_OK );
 
     /* If page one was fetched successfully, and this function is not
     ** operating in direct-mode, make page 1 writable.  When not in 
     ** direct mode, page 1 is always held in cache and hence the PagerGet()
     ** above is always successful - hence the ALWAYS on rc==SQLITE_OK.
     */
     if( !DIRECT_MODE && ALWAYS(rc==SQLITE_OK) ){
       rc = sqlite3PagerWrite(pPgHdr);
     }
 
     if( rc==SQLITE_OK ){
       /* Actually do the update of the change counter */
       pager_write_changecounter(pPgHdr);
 
       /* If running in direct mode, write the contents of page 1 to the file. */
       if( DIRECT_MODE ){
         const void *zBuf;
         assert( pPager->dbFileSize>0 );
         CODEC2(pPager, pPgHdr->pData, 1, 6, rc=SQLITE_NOMEM, zBuf);
         if( rc==SQLITE_OK ){
           rc = sqlite3OsWrite(pPager->fd, zBuf, pPager->pageSize, 0);
+          pPager->aStat[PAGER_STAT_WRITE]++;
         }
         if( rc==SQLITE_OK ){
+          /* Update the pager's copy of the change-counter. Otherwise, the
+          ** next time a read transaction is opened the cache will be
+          ** flushed (as the change-counter values will not match).  */
+          const void *pCopy = (const void *)&((const char *)zBuf)[24];
+          memcpy(&pPager->dbFileVers, pCopy, sizeof(pPager->dbFileVers));
           pPager->changeCountDone = 1;
         }
       }else{
         pPager->changeCountDone = 1;
       }
     }
 
     /* Release the page reference. */
     sqlite3PagerUnref(pPgHdr);
   }
   return rc;
 }
 
 /*
 ** Sync the database file to disk. This is a no-op for in-memory databases
 ** or pages with the Pager.noSync flag set.
 **
 ** If successful, or if called on a pager for which it is a no-op, this
 ** function returns SQLITE_OK. Otherwise, an IO error code is returned.
 */
-int sqlite3PagerSync(Pager *pPager){
+int sqlite3PagerSync(Pager *pPager, const char *zMaster){
   int rc = SQLITE_OK;
-  if( !pPager->noSync ){
+
+  if( isOpen(pPager->fd) ){
+    void *pArg = (void*)zMaster;
+    rc = sqlite3OsFileControl(pPager->fd, SQLITE_FCNTL_SYNC, pArg);
+    if( rc==SQLITE_NOTFOUND ) rc = SQLITE_OK;
+  }
+  if( rc==SQLITE_OK && !pPager->noSync ){
     assert( !MEMDB );
     rc = sqlite3OsSync(pPager->fd, pPager->syncFlags);
-  }else if( isOpen(pPager->fd) ){
-    assert( !MEMDB );
-    sqlite3OsFileControl(pPager->fd, SQLITE_FCNTL_SYNC_OMITTED, (void *)&rc);
   }
   return rc;
 }
 
 /*
 ** This function may only be called while a write-transaction is active in
 ** rollback. If the connection is in WAL mode, this call is a no-op. 
 ** Otherwise, if the connection does not already have an EXCLUSIVE lock on 
 ** the database file, an attempt is made to obtain one.
 **
@@ skipping 74 matching lines @@
     if( pagerUseWal(pPager) ){
       PgHdr *pList = sqlite3PcacheDirtyList(pPager->pPCache);
       PgHdr *pPageOne = 0;
       if( pList==0 ){
         /* Must have at least one page for the WAL commit flag.
         ** Ticket [2d1a5c67dfc2363e44f29d9bbd57f] 2011-05-18 */
         rc = sqlite3PagerGet(pPager, 1, &pPageOne);
         pList = pPageOne;
         pList->pDirty = 0;
       }
-      assert( pList!=0 || rc!=SQLITE_OK );
-      if( pList ){
-        rc = pagerWalFrames(pPager, pList, pPager->dbSize, 1, 
-            (pPager->fullSync ? pPager->syncFlags : 0)
-        );
+      assert( rc==SQLITE_OK );
+      if( ALWAYS(pList) ){
+        rc = pagerWalFrames(pPager, pList, pPager->dbSize, 1);
       }
       sqlite3PagerUnref(pPageOne);
       if( rc==SQLITE_OK ){
         sqlite3PcacheCleanAll(pPager->pPCache);
       }
     }else{
       /* The following block updates the change-counter. Exactly how it
       ** does this depends on whether or not the atomic-update optimization
       ** was enabled at compile time, and if this transaction meets the 
       ** runtime criteria to use the operation: 
@@ skipping 38 matching lines @@
         rc = sqlite3JournalCreate(pPager->jfd);
         if( rc==SQLITE_OK ){
           rc = pager_incr_changecounter(pPager, 0);
         }
       }
   #else
       rc = pager_incr_changecounter(pPager, 0);
   #endif
       if( rc!=SQLITE_OK ) goto commit_phase_one_exit;
   
-      /* If this transaction has made the database smaller, then all pages
-      ** being discarded by the truncation must be written to the journal
-      ** file. This can only happen in auto-vacuum mode.
-      **
-      ** Before reading the pages with page numbers larger than the 
-      ** current value of Pager.dbSize, set dbSize back to the value
-      ** that it took at the start of the transaction. Otherwise, the
-      ** calls to sqlite3PagerGet() return zeroed pages instead of 
-      ** reading data from the database file.
-      */
-  #ifndef SQLITE_OMIT_AUTOVACUUM
-      if( pPager->dbSize<pPager->dbOrigSize 
-       && pPager->journalMode!=PAGER_JOURNALMODE_OFF
-      ){
-        Pgno i;                                   /* Iterator variable */
-        const Pgno iSkip = PAGER_MJ_PGNO(pPager); /* Pending lock page */
-        const Pgno dbSize = pPager->dbSize;       /* Database image size */ 
-        pPager->dbSize = pPager->dbOrigSize;
-        for( i=dbSize+1; i<=pPager->dbOrigSize; i++ ){
-          if( !sqlite3BitvecTest(pPager->pInJournal, i) && i!=iSkip ){
-            PgHdr *pPage;             /* Page to journal */
-            rc = sqlite3PagerGet(pPager, i, &pPage);
-            if( rc!=SQLITE_OK ) goto commit_phase_one_exit;
-            rc = sqlite3PagerWrite(pPage);
-            sqlite3PagerUnref(pPage);
-            if( rc!=SQLITE_OK ) goto commit_phase_one_exit;
-          }
-        }
-        pPager->dbSize = dbSize;
-      } 
-  #endif
-  
       /* Write the master journal name into the journal file. If a master 
       ** journal file name has already been written to the journal file, 
       ** or if zMaster is NULL (no master journal), then this call is a no-op.
       */
       rc = writeMasterJournal(pPager, zMaster);
       if( rc!=SQLITE_OK ) goto commit_phase_one_exit;
   
       /* Sync the journal file and write all dirty pages to the database.
       ** If the atomic-update optimization is being used, this sync will not 
       ** create the journal file or perform any real IO.
       **
       ** Because the change-counter page was just modified, unless the
       ** atomic-update optimization is used it is almost certain that the
       ** journal requires a sync here. However, in locking_mode=exclusive
       ** on a system under memory pressure it is just possible that this is 
       ** not the case. In this case it is likely enough that the redundant
       ** xSync() call will be changed to a no-op by the OS anyhow. 
       */
       rc = syncJournal(pPager, 0);
       if( rc!=SQLITE_OK ) goto commit_phase_one_exit;
   
       rc = pager_write_pagelist(pPager,sqlite3PcacheDirtyList(pPager->pPCache));
       if( rc!=SQLITE_OK ){
         assert( rc!=SQLITE_IOERR_BLOCKED );
         goto commit_phase_one_exit;
       }
       sqlite3PcacheCleanAll(pPager->pPCache);
-  
-      /* If the file on disk is not the same size as the database image,
-      ** then use pager_truncate to grow or shrink the file here.
-      */
-      if( pPager->dbSize!=pPager->dbFileSize ){
+
+      /* If the file on disk is smaller than the database image, use 
+      ** pager_truncate to grow the file here. This can happen if the database
+      ** image was extended as part of the current transaction and then the
+      ** last page in the db image moved to the free-list. In this case the
+      ** last page is never written out to disk, leaving the database file
+      ** undersized. Fix this now if it is the case.  */
+      if( pPager->dbSize>pPager->dbFileSize ){
         Pgno nNew = pPager->dbSize - (pPager->dbSize==PAGER_MJ_PGNO(pPager));
         assert( pPager->eState==PAGER_WRITER_DBMOD );
         rc = pager_truncate(pPager, nNew);
         if( rc!=SQLITE_OK ) goto commit_phase_one_exit;
       }
   
       /* Finally, sync the database file. */
       if( !noSync ){
-        rc = sqlite3PagerSync(pPager);
+        rc = sqlite3PagerSync(pPager, zMaster);
       }
       IOTRACE(("DBSYNC %p\n", pPager))
     }
   }
 
 commit_phase_one_exit:
   if( rc==SQLITE_OK && !pagerUseWal(pPager) ){
     pPager->eState = PAGER_WRITER_FINISHED;
   }
   return rc;
@@ skipping 43 matching lines @@
   if( pPager->eState==PAGER_WRITER_LOCKED 
    && pPager->exclusiveMode 
    && pPager->journalMode==PAGER_JOURNALMODE_PERSIST
   ){
     assert( pPager->journalOff==JOURNAL_HDR_SZ(pPager) || !pPager->journalOff );
     pPager->eState = PAGER_READER;
     return SQLITE_OK;
   }
 
   PAGERTRACE(("COMMIT %d\n", PAGERID(pPager)));
-  rc = pager_end_transaction(pPager, pPager->setMaster);
+  rc = pager_end_transaction(pPager, pPager->setMaster, 1);
   return pager_error(pPager, rc);
 }
 
 /*
 ** If a write transaction is open, then all changes made within the 
 ** transaction are reverted and the current write-transaction is closed.
 ** The pager falls back to PAGER_READER state if successful, or PAGER_ERROR
 ** state if an error occurs.
 **
 ** If the pager is already in PAGER_ERROR state when this function is called,
@@ skipping 24 matching lines @@
   ** the pager is already in the ERROR state, the rollback is not 
   ** attempted here. Instead, the error code is returned to the caller.
   */
   assert( assert_pager_state(pPager) );
   if( pPager->eState==PAGER_ERROR ) return pPager->errCode;
   if( pPager->eState<=PAGER_READER ) return SQLITE_OK;
 
   if( pagerUseWal(pPager) ){
     int rc2;
     rc = sqlite3PagerSavepoint(pPager, SAVEPOINT_ROLLBACK, -1);
-    rc2 = pager_end_transaction(pPager, pPager->setMaster);
+    rc2 = pager_end_transaction(pPager, pPager->setMaster, 0);
     if( rc==SQLITE_OK ) rc = rc2;
   }else if( !isOpen(pPager->jfd) || pPager->eState==PAGER_WRITER_LOCKED ){
     int eState = pPager->eState;
-    rc = pager_end_transaction(pPager, 0);
+    rc = pager_end_transaction(pPager, 0, 0);
     if( !MEMDB && eState>PAGER_WRITER_LOCKED ){
       /* This can happen using journal_mode=off. Move the pager to the error 
       ** state to indicate that the contents of the cache may not be trusted.
       ** Any active readers will get SQLITE_ABORT.
       */
       pPager->errCode = SQLITE_ABORT;
       pPager->eState = PAGER_ERROR;
       return rc;
     }
   }else{
     rc = pager_playback(pPager, 0);
   }
 
   assert( pPager->eState==PAGER_READER || rc!=SQLITE_OK );
-  assert( rc==SQLITE_OK || rc==SQLITE_FULL || (rc&0xFF)==SQLITE_IOERR );
+  assert( rc==SQLITE_OK || rc==SQLITE_FULL || rc==SQLITE_CORRUPT
+          || rc==SQLITE_NOMEM || (rc&0xFF)==SQLITE_IOERR 
+          || rc==SQLITE_CANTOPEN
+  );
 
   /* If an error occurs during a ROLLBACK, we can no longer trust the pager
   ** cache. So call pager_error() on the way out to make any error persistent.
   */
   return pager_error(pPager, rc);
 }
 
 /*
 ** Return TRUE if the database file is opened read-only.  Return FALSE
 ** if the database is (in theory) writable.
@@ skipping 33 matching lines @@
 ** This routine is used for testing and analysis only.
 */
 int *sqlite3PagerStats(Pager *pPager){
   static int a[11];
   a[0] = sqlite3PcacheRefCount(pPager->pPCache);
   a[1] = sqlite3PcachePagecount(pPager->pPCache);
   a[2] = sqlite3PcacheGetCachesize(pPager->pPCache);
   a[3] = pPager->eState==PAGER_OPEN ? -1 : (int) pPager->dbSize;
   a[4] = pPager->eState;
   a[5] = pPager->errCode;
-  a[6] = pPager->nHit;
-  a[7] = pPager->nMiss;
+  a[6] = pPager->aStat[PAGER_STAT_HIT];
+  a[7] = pPager->aStat[PAGER_STAT_MISS];
   a[8] = 0;  /* Used to be pPager->nOvfl */
   a[9] = pPager->nRead;
-  a[10] = pPager->nWrite;
+  a[10] = pPager->aStat[PAGER_STAT_WRITE];
   return a;
 }
 #endif
 
 /*
+** Parameter eStat must be either SQLITE_DBSTATUS_CACHE_HIT or
+** SQLITE_DBSTATUS_CACHE_MISS. Before returning, *pnVal is incremented by the
+** current cache hit or miss count, according to the value of eStat. If the 
+** reset parameter is non-zero, the cache hit or miss count is zeroed before 
+** returning.
+*/
+void sqlite3PagerCacheStat(Pager *pPager, int eStat, int reset, int *pnVal){
+
+  assert( eStat==SQLITE_DBSTATUS_CACHE_HIT
+       || eStat==SQLITE_DBSTATUS_CACHE_MISS
+       || eStat==SQLITE_DBSTATUS_CACHE_WRITE
+  );
+
+  assert( SQLITE_DBSTATUS_CACHE_HIT+1==SQLITE_DBSTATUS_CACHE_MISS );
+  assert( SQLITE_DBSTATUS_CACHE_HIT+2==SQLITE_DBSTATUS_CACHE_WRITE );
+  assert( PAGER_STAT_HIT==0 && PAGER_STAT_MISS==1 && PAGER_STAT_WRITE==2 );
+
+  *pnVal += pPager->aStat[eStat - SQLITE_DBSTATUS_CACHE_HIT];
+  if( reset ){
+    pPager->aStat[eStat - SQLITE_DBSTATUS_CACHE_HIT] = 0;
+  }
+}
+
+/*
 ** Return true if this is an in-memory pager.
 */
 int sqlite3PagerIsMemdb(Pager *pPager){
   return MEMDB;
 }
 
 /*
 ** Check that there are at least nSavepoint savepoints open. If there are
 ** currently less than nSavepoints open, then open one or more savepoints
 ** to make up the difference. If the number of savepoints is already
@@ skipping 124 matching lines @@
       rc = pagerPlaybackSavepoint(pPager, pSavepoint);
       assert(rc!=SQLITE_DONE);
     }
   }
 
   return rc;
 }
 
 /*
 ** Return the full pathname of the database file.
+**
+** Except, if the pager is in-memory only, then return an empty string if
+** nullIfMemDb is true.  This routine is called with nullIfMemDb==1 when
+** used to report the filename to the user, for compatibility with legacy
+** behavior.  But when the Btree needs to know the filename for matching to
+** shared cache, it uses nullIfMemDb==0 so that in-memory databases can
+** participate in shared-cache.
 */
-const char *sqlite3PagerFilename(Pager *pPager){
-  return pPager->zFilename;
+const char *sqlite3PagerFilename(Pager *pPager, int nullIfMemDb){
+  return (nullIfMemDb && pPager->memDb) ? "" : pPager->zFilename;
 }
 
 /*
 ** Return the VFS structure for the pager.
 */
 const sqlite3_vfs *sqlite3PagerVfs(Pager *pPager){
   return pPager->pVfs;
 }
 
 /*
@@ skipping 34 matching lines @@
   if( pPager->xCodecFree ) pPager->xCodecFree(pPager->pCodec);
   pPager->xCodec = pPager->memDb ? 0 : xCodec;
   pPager->xCodecSizeChng = xCodecSizeChng;
   pPager->xCodecFree = xCodecFree;
   pPager->pCodec = pCodec;
   pagerReportSize(pPager);
 }
 void *sqlite3PagerGetCodec(Pager *pPager){
   return pPager->pCodec;
 }
-#endif
+
+/*
+** This function is called by the wal module when writing page content
+** into the log file.
+**
+** This function returns a pointer to a buffer containing the encrypted
+** page content. If a malloc fails, this function may return NULL.
+*/
+void *sqlite3PagerCodec(PgHdr *pPg){
+  void *aData = 0;
+  CODEC2(pPg->pPager, pPg->pData, pPg->pgno, 6, return 0, aData);
+  return aData;
+}
+
+/*
+** Return the current pager state
+*/
+int sqlite3PagerState(Pager *pPager){
+  return pPager->eState;
+}
+#endif /* SQLITE_HAS_CODEC */
 
 #ifndef SQLITE_OMIT_AUTOVACUUM
 /*
 ** Move the page pPg to location pgno in the file.
 **
 ** There must be no references to the page previously located at
 ** pgno (which we call pPgOld) though that page is allowed to be
 ** in cache.  If the page previously located at pgno is not already
 ** in the rollback journal, it is not put there by by this routine.
 **
@@ skipping 65 matching lines @@
 
   /* If the journal needs to be sync()ed before page pPg->pgno can
   ** be written to, store pPg->pgno in local variable needSyncPgno.
   **
   ** If the isCommit flag is set, there is no need to remember that
   ** the journal needs to be sync()ed before database page pPg->pgno 
   ** can be written to. The caller has already promised not to write to it.
   */
   if( (pPg->flags&PGHDR_NEED_SYNC) && !isCommit ){
     needSyncPgno = pPg->pgno;
-    assert( pageInJournal(pPg) || pPg->pgno>pPager->dbOrigSize );
+    assert( pPager->journalMode==PAGER_JOURNALMODE_OFF ||
+            pageInJournal(pPager, pPg) || pPg->pgno>pPager->dbOrigSize );
     assert( pPg->flags&PGHDR_DIRTY );
   }
 
   /* If the cache contains a page with page-number pgno, remove it
   ** from its hash chain. Also, if the PGHDR_NEED_SYNC flag was set for 
   ** page pgno before the 'move' operation, it needs to be retained 
   ** for the page moved there.
   */
   pPg->flags &= ~PGHDR_NEED_SYNC;
-  pPgOld = pager_lookup(pPager, pgno);
+  pPgOld = sqlite3PagerLookup(pPager, pgno);
   assert( !pPgOld || pPgOld->nRef==1 );
   if( pPgOld ){
     pPg->flags |= (pPgOld->flags&PGHDR_NEED_SYNC);
     if( MEMDB ){
       /* Do not discard pages from an in-memory database since we might
       ** need to rollback later.  Just move the page out of the way. */
       sqlite3PcacheMove(pPgOld, pPager->dbSize+1);
     }else{
       sqlite3PcacheDrop(pPgOld);
     }
   }
 
   origPgno = pPg->pgno;
   sqlite3PcacheMove(pPg, pgno);
   sqlite3PcacheMakeDirty(pPg);
 
   /* For an in-memory database, make sure the original page continues
   ** to exist, in case the transaction needs to roll back.  Use pPgOld
   ** as the original page since it has already been allocated.
   */
   if( MEMDB ){
     assert( pPgOld );
     sqlite3PcacheMove(pPgOld, origPgno);
-    sqlite3PagerUnref(pPgOld);
+    sqlite3PagerUnrefNotNull(pPgOld);
   }
 
   if( needSyncPgno ){
     /* If needSyncPgno is non-zero, then the journal file needs to be 
     ** sync()ed before any data is written to database file page needSyncPgno.
     ** Currently, no such page exists in the page-cache and the 
     ** "is journaled" bitvec flag has been set. This needs to be remedied by
     ** loading the page into the pager-cache and setting the PGHDR_NEED_SYNC
     ** flag.
     **
     ** If the attempt to load the page into the page-cache fails, (due
     ** to a malloc() or IO failure), clear the bit in the pInJournal[]
     ** array. Otherwise, if the page is loaded and written again in
     ** this transaction, it may be written to the database file before
     ** it is synced into the journal file. This way, it may end up in
     ** the journal file twice, but that is not a problem.
     */
     PgHdr *pPgHdr;
     rc = sqlite3PagerGet(pPager, needSyncPgno, &pPgHdr);
     if( rc!=SQLITE_OK ){
       if( needSyncPgno<=pPager->dbOrigSize ){
         assert( pPager->pTmpSpace!=0 );
         sqlite3BitvecClear(pPager->pInJournal, needSyncPgno, pPager->pTmpSpace);
       }
       return rc;
     }
     pPgHdr->flags |= PGHDR_NEED_SYNC;
     sqlite3PcacheMakeDirty(pPgHdr);
-    sqlite3PagerUnref(pPgHdr);
+    sqlite3PagerUnrefNotNull(pPgHdr);
   }
 
   return SQLITE_OK;
 }
 #endif
 
 /*
 ** Return a pointer to the data for the specified page.
 */
 void *sqlite3PagerGetData(DbPage *pPg){
@@ skipping 166 matching lines @@
 
 /*
 ** Get/set the size-limit used for persistent journal files.
 **
 ** Setting the size limit to -1 means no limit is enforced.
 ** An attempt to set a limit smaller than -1 is a no-op.
 */
 i64 sqlite3PagerJournalSizeLimit(Pager *pPager, i64 iLimit){
   if( iLimit>=-1 ){
     pPager->journalSizeLimit = iLimit;
+    sqlite3WalLimit(pPager->pWal, iLimit);
   }
   return pPager->journalSizeLimit;
 }
 
 /*
 ** Return a pointer to the pPager->pBackup variable. The backup module
 ** in backup.c maintains the content of this variable. This module
 ** uses it opaquely as an argument to sqlite3BackupRestart() and
 ** sqlite3BackupUpdate() only.
 */
 sqlite3_backup **sqlite3PagerBackupPtr(Pager *pPager){
   return &pPager->pBackup;
 }
 
+#ifndef SQLITE_OMIT_VACUUM
+/*
+** Unless this is an in-memory or temporary database, clear the pager cache.
+*/
+void sqlite3PagerClearCache(Pager *pPager){
+  if( !MEMDB && pPager->tempFile==0 ) pager_reset(pPager);
+}
+#endif
+
 #ifndef SQLITE_OMIT_WAL
 /*
 ** This function is called when the user invokes "PRAGMA wal_checkpoint",
 ** "PRAGMA wal_blocking_checkpoint" or calls the sqlite3_wal_checkpoint()
 ** or wal_blocking_checkpoint() API functions.
 **
 ** Parameter eMode is one of SQLITE_CHECKPOINT_PASSIVE, FULL or RESTART.
 */
 int sqlite3PagerCheckpoint(Pager *pPager, int eMode, int *pnLog, int *pnCkpt){
   int rc = SQLITE_OK;
@@ skipping 41 matching lines @@
 /*
 ** Call sqlite3WalOpen() to open the WAL handle. If the pager is in 
 ** exclusive-locking mode when this function is called, take an EXCLUSIVE
 ** lock on the database file and use heap-memory to store the wal-index
 ** in. Otherwise, use the normal shared-memory.
 */
 static int pagerOpenWal(Pager *pPager){
   int rc = SQLITE_OK;
 
   assert( pPager->pWal==0 && pPager->tempFile==0 );
-  assert( pPager->eLock==SHARED_LOCK || pPager->eLock==EXCLUSIVE_LOCK || pPager->noReadlock);
+  assert( pPager->eLock==SHARED_LOCK || pPager->eLock==EXCLUSIVE_LOCK );
 
   /* If the pager is already in exclusive-mode, the WAL module will use 
   ** heap-memory for the wal-index instead of the VFS shared-memory 
   ** implementation. Take the exclusive lock now, before opening the WAL
   ** file, to make sure this is safe.
   */
   if( pPager->exclusiveMode ){
     rc = pagerExclusiveLock(pPager);
   }
 
   /* Open the connection to the log file. If this operation fails, 
   ** (e.g. due to malloc() failure), return an error code.
   */
   if( rc==SQLITE_OK ){
-    rc = sqlite3WalOpen(pPager->pVfs, 
-        pPager->fd, pPager->zWal, pPager->exclusiveMode, &pPager->pWal
+    rc = sqlite3WalOpen(pPager->pVfs,
+        pPager->fd, pPager->zWal, pPager->exclusiveMode,
+        pPager->journalSizeLimit, &pPager->pWal
     );
   }
+  pagerFixMaplimit(pPager);
 
   return rc;
 }
 
 
 /*
 ** The caller must be holding a SHARED lock on the database file to call
 ** this function.
 **
 ** If the pager passed as the first argument is open on a real database
@@ skipping 70 matching lines @@
     
   /* Checkpoint and close the log. Because an EXCLUSIVE lock is held on
   ** the database file, the log and log-summary files will be deleted.
   */
   if( rc==SQLITE_OK && pPager->pWal ){
     rc = pagerExclusiveLock(pPager);
     if( rc==SQLITE_OK ){
       rc = sqlite3WalClose(pPager->pWal, pPager->ckptSyncFlags,
                            pPager->pageSize, (u8*)pPager->pTmpSpace);
       pPager->pWal = 0;
+      pagerFixMaplimit(pPager);
     }
   }
   return rc;
 }
 
-#ifdef SQLITE_HAS_CODEC
-/*
-** This function is called by the wal module when writing page content
-** into the log file.
-**
-** This function returns a pointer to a buffer containing the encrypted
-** page content. If a malloc fails, this function may return NULL.
-*/
-void *sqlite3PagerCodec(PgHdr *pPg){
-  void *aData = 0;
-  CODEC2(pPg->pPager, pPg->pData, pPg->pgno, 6, return 0, aData);
-  return aData;
-}
-#endif /* SQLITE_HAS_CODEC */
-
 #endif /* !SQLITE_OMIT_WAL */
 
+#ifdef SQLITE_ENABLE_ZIPVFS
+/*
+** A read-lock must be held on the pager when this function is called. If
+** the pager is in WAL mode and the WAL file currently contains one or more
+** frames, return the size in bytes of the page images stored within the
+** WAL frames. Otherwise, if this is not a WAL database or the WAL file
+** is empty, return 0.
+*/
+int sqlite3PagerWalFramesize(Pager *pPager){
+  assert( pPager->eState>=PAGER_READER );
+  return sqlite3WalFramesize(pPager->pWal);
+}
+#endif
+
 #endif /* SQLITE_OMIT_DISKIO */