Wal2 Mode Notes
===============

## Activating/Deactivating Wal2 Mode

"Wal2" mode is very similar to "wal" mode. To change a database to wal2 mode,
use the command:

>
     PRAGMA journal_mode = wal2;

It is not possible to change a database directly from "wal" mode to "wal2"
mode. Instead, it must first be changed to rollback mode. So, to change a wal
mode database to wal2 mode, the following two commands may be used:

>
     PRAGMA journal_mode = delete;
     PRAGMA journal_mode = wal2;

A database in wal2 mode may only be accessed by versions of SQLite compiled
from this branch. Attempting to use any other version of SQLite results in an
SQLITE_NOTADB error. A wal2 mode database may be changed back to rollback mode
(making it accessible by all versions of SQLite) using:

>
     PRAGMA journal_mode = delete;

## The Advantage of Wal2 Mode

In legacy wal mode, when a writer writes data to the database, it doesn't
modify the database file directly. Instead, it appends new data to the
"<database>-wal" file. Readers read data from both the original database
file and the "<database>-wal" file. At some point, data is copied from the
"<database>-wal" file into the database file, after which the wal file can
be deleted or overwritten. Copying data from the wal file into the database
file is called a "checkpoint", and may be done explictly (either by "PRAGMA
wal_checkpoint" or sqlite3_wal_checkpoint_v2()), or
automatically (by configuring "PRAGMA wal_autocheckpoint" - this is the
default).

Checkpointers do not block writers, and writers do not block checkpointers.
However, if a writer writes to the database while a checkpoint is ongoing,
then the new data is appended to the end of the wal file. This means that,
even following the checkpoint, the wal file cannot be overwritten or deleted,
and so all subsequent transactions must also be appended to the wal file. The
work of the checkpointer is not wasted - SQLite remembers which parts of the
wal file have already been copied into the db file so that the next checkpoint
does not have to do so again - but it does mean that the wal file may grow
indefinitely if the checkpointer never gets a chance to finish without a
writer appending to the wal file. There are also circumstances in which
long-running readers may prevent a checkpointer from checkpointing the entire
wal file - also causing the wal file to grow indefinitely in a busy system.

Wal2 mode does not have this problem. In wal2 mode, wal files do not grow
indefinitely even if the checkpointer never has a chance to finish
uninterrupted.

In wal2 mode, the system uses two wal files instead of one. The files are named
"<database>-wal" and "<database>-wal2", where "<database>" is of
course the name of the database file. When data is written to the database, the
writer begins by appending the new data to the first wal file. Once the first
wal file has grown large enough, writers switch to appending data to the second
wal file. At this point the first wal file can be checkpointed (after which it
can be overwritten). Then, once the second wal file has grown large enough and
the first wal file has been checkpointed, writers switch back to the first wal
file. And so on.

## Application Programming

From the point of view of the user, the main differences between wal and 
wal2 mode are to do with checkpointing:

  * In wal mode, a checkpoint may be attempted at any time. In wal2 
    mode, the checkpointer has to wait until writers have switched 
    to the "other" wal file before a checkpoint can take place.

  * In wal mode, the wal-hook (callback registered using
    sqlite3_wal_hook()) is invoked after a transaction is committed
    with the total number of pages in the wal file as an argument. In wal2
    mode, the argument is either the total number of uncheckpointed pages in
    both wal files, or - if the "other" wal file is empty or already
    checkpointed - 0.

Clients are recommended to use the same strategies for checkpointing wal2 mode
databases as for wal databases - by registering a wal-hook using
sqlite3_wal_hook() and attempting a checkpoint when the parameter
exceeds a certain threshold.

However, it should be noted that although the wal-hook is invoked after each
transaction is committed to disk and database locks released, it is still
invoked from within the sqlite3_step() call used to execute the "COMMIT"
command. In BEGIN CONCURRENT systems, where the "COMMIT" is often protected by
an application mutex, this may reduce concurrency. In such systems, instead of
executing a checkpoint from within the wal-hook, a thread might defer this
action until after the application mutex has been released.

/*
** 2017-10-11
**
** 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.
**
******************************************************************************
**
*/

#if !defined(SQLITE_TEST) || defined(SQLITE_OS_UNIX)

#include "sqlite3.h"
#include <string.h>
#include <pthread.h>

/*
** API declarations.
*/
typedef struct Checkpointer Checkpointer;
int sqlite3_bgckpt_create(const char *zFilename, Checkpointer **pp);
int sqlite3_bgckpt_checkpoint(Checkpointer *p, int bBlock);
void sqlite3_bgckpt_destroy(Checkpointer *p);


struct Checkpointer {
  sqlite3 *db;                    /* Database handle */

  pthread_t thread;               /* Background thread */
  pthread_mutex_t mutex;
  pthread_cond_t cond;

  int rc;                         /* Error from "PRAGMA wal_checkpoint" */
  int bCkpt;                      /* True if checkpoint requested */
  int bExit;                      /* True if exit requested */
};

static void *bgckptThreadMain(void *pCtx){
  int rc = SQLITE_OK;
  Checkpointer *p = (Checkpointer*)pCtx;

  while( rc==SQLITE_OK ){
    int bExit;

    pthread_mutex_lock(&p->mutex);
    if( p->bCkpt==0 && p->bExit==0 ){
      pthread_cond_wait(&p->cond, &p->mutex);
    }
    p->bCkpt = 0;
    bExit = p->bExit;
    pthread_mutex_unlock(&p->mutex);

    if( bExit ) break;
    rc = sqlite3_exec(p->db, "PRAGMA wal_checkpoint", 0, 0, 0);
    if( rc==SQLITE_BUSY ){
      rc = SQLITE_OK;
    }
  }

  pthread_mutex_lock(&p->mutex);
  p->rc = rc;
  pthread_mutex_unlock(&p->mutex);
  return 0;
}

void sqlite3_bgckpt_destroy(Checkpointer *p){
  if( p ){
    void *ret = 0;

    /* Signal the background thread to exit */
    pthread_mutex_lock(&p->mutex);
    p->bExit = 1;
    pthread_cond_broadcast(&p->cond);
    pthread_mutex_unlock(&p->mutex);

    pthread_join(p->thread, &ret);
    sqlite3_close(p->db);
    sqlite3_free(p);
  }
}


int sqlite3_bgckpt_create(const char *zFilename, Checkpointer **pp){
  Checkpointer *pNew = 0;
  int rc;

  pNew = (Checkpointer*)sqlite3_malloc(sizeof(Checkpointer));
  if( pNew==0 ){
    rc = SQLITE_NOMEM;
  }else{
    memset(pNew, 0, sizeof(Checkpointer));
    rc = sqlite3_open(zFilename, &pNew->db);
  }

  if( rc==SQLITE_OK ){
    pthread_mutex_init(&pNew->mutex, 0);
    pthread_cond_init(&pNew->cond, 0);
    pthread_create(&pNew->thread, 0, bgckptThreadMain, (void*)pNew);
  }

  if( rc!=SQLITE_OK ){
    sqlite3_bgckpt_destroy(pNew);
    pNew = 0;
  }
  *pp = pNew;
  return rc;
}

int sqlite3_bgckpt_checkpoint(Checkpointer *p, int bBlock){
  int rc;
  pthread_mutex_lock(&p->mutex);
  rc = p->rc;
  if( rc==SQLITE_OK ){
    p->bCkpt = 1;
    pthread_cond_broadcast(&p->cond);
  }
  pthread_mutex_unlock(&p->mutex);
  return rc;
}

#ifdef SQLITE_TEST

#if defined(INCLUDE_SQLITE_TCL_H)
#  include "sqlite_tcl.h"
#else
#  include "tcl.h"
#  ifndef SQLITE_TCLAPI
#    define SQLITE_TCLAPI
#  endif
#endif

const char *sqlite3ErrName(int rc);

static void SQLITE_TCLAPI bgckpt_del(void * clientData){
  Checkpointer *pCkpt = (Checkpointer*)clientData;
  sqlite3_bgckpt_destroy(pCkpt);
}

/*
** Tclcmd: $ckpt SUBCMD ...
*/
static int SQLITE_TCLAPI bgckpt_obj_cmd(
  void * clientData,
  Tcl_Interp *interp,
  int objc,
  Tcl_Obj *CONST objv[]
){
  Checkpointer *pCkpt = (Checkpointer*)clientData;
  const char *aCmd[] = { "checkpoint", "destroy", 0 };
  int iCmd;

  if( objc<2 ){
    Tcl_WrongNumArgs(interp, 1, objv, "SUBCMD ...");
    return TCL_ERROR;
  }

  if( Tcl_GetIndexFromObj(interp, objv[1], aCmd, "sub-command", 0, &iCmd) ){
    return TCL_ERROR;
  }

  switch( iCmd ){
    case 0: {
      int rc;
      int bBlock = 0;

      if( objc>3 ){
        Tcl_WrongNumArgs(interp, 2, objv, "?BLOCKING?");
        return TCL_ERROR;
      }
      if( objc==3 && Tcl_GetBooleanFromObj(interp, objv[2], &bBlock) ){
        return TCL_ERROR;
      }

      rc = sqlite3_bgckpt_checkpoint(pCkpt, bBlock);
      if( rc!=SQLITE_OK ){
        Tcl_SetObjResult(interp, Tcl_NewStringObj(sqlite3ErrName(rc), -1));
        return TCL_ERROR;
      }
      break;
    }

    case 1: {
      Tcl_DeleteCommand(interp, Tcl_GetString(objv[0]));
      break;
    }
  }

  return TCL_OK;
}

/*
** Tclcmd: bgckpt CMDNAME FILENAME
*/
static int SQLITE_TCLAPI bgckpt_cmd(
  void * clientData,
  Tcl_Interp *interp,
  int objc,
  Tcl_Obj *CONST objv[]
){
  const char *zCmd;
  const char *zFilename;
  int rc;
  Checkpointer *pCkpt;

  if( objc!=3 ){
    Tcl_WrongNumArgs(interp, 1, objv, "CMDNAME FILENAME");
    return TCL_ERROR;
  }
  zCmd = Tcl_GetString(objv[1]);
  zFilename = Tcl_GetString(objv[2]);

  rc = sqlite3_bgckpt_create(zFilename, &pCkpt);
  if( rc!=SQLITE_OK ){
    Tcl_SetObjResult(interp, Tcl_NewStringObj(sqlite3ErrName(rc), -1));
    return TCL_ERROR;
  }

  Tcl_CreateObjCommand(interp, zCmd, bgckpt_obj_cmd, (void*)pCkpt, bgckpt_del);
  Tcl_SetObjResult(interp, objv[1]);
  return TCL_OK;
}

int Bgckpt_Init(Tcl_Interp *interp){
  Tcl_CreateObjCommand(interp, "bgckpt", bgckpt_cmd, 0, 0);
  return TCL_OK;
}
#endif   /* SQLITE_TEST */

#else
#if defined(INCLUDE_SQLITE_TCL_H)
#  include "sqlite_tcl.h"
#else
#  include "tcl.h"
#  ifndef SQLITE_TCLAPI
#    define SQLITE_TCLAPI
#  endif
#endif
int Bgckpt_Init(Tcl_Interp *interp){ return TCL_OK; }
#endif

** the same session object is disabled, no INSERT record will appear in the
** changeset, even though the delete took place while the session was disabled.
** Or, if one field of a row is updated while a session is disabled, and 
** another field of the same row is updated while the session is enabled, the
** resulting changeset will contain an UPDATE change that updates both fields.
*/
int sqlite3session_changeset(
  sqlite3_session *pSession,      /* Session object */
  int *pnChangeset,               /* OUT: Size of buffer at *ppChangeset */
  void **ppChangeset              /* OUT: Buffer containing changeset */
);

/*
** CAPI3REF: Generate A Full Changeset From A Session Object
**
** This function is similar to sqlite3session_changeset(), except that for
** each row affected by an UPDATE statement, all old.* values are recorded
** as part of the changeset, not just those modified.
*/
int sqlite3session_fullchangeset(
  sqlite3_session *pSession,      /* Session object */
  int *pnChangeset,               /* OUT: Size of buffer at *ppChangeset */
  void **ppChangeset              /* OUT: Buffer containing changeset */
);

/*
** CAPI3REF: Generate A Full Changeset From A Session Object

  $(TOP)/src/test_window.c \
  $(TOP)/src/test_wsd.c

# Extensions to be statically loaded.
#
TESTSRC += \
  $(TOP)/ext/misc/amatch.c \
  $(TOP)/ext/misc/bgckpt.c \
  $(TOP)/ext/misc/carray.c \
  $(TOP)/ext/misc/closure.c \
  $(TOP)/ext/misc/csv.c \
  $(TOP)/ext/misc/eval.c \
  $(TOP)/ext/misc/explain.c \
  $(TOP)/ext/misc/fileio.c \
  $(TOP)/ext/misc/fuzzer.c \

    if( page1[18]>1 ){
      pBt->btsFlags |= BTS_READ_ONLY;
    }
    if( page1[19]>1 ){
      goto page1_init_failed;
    }
#else
    if( page1[18]>2 ){
    if( page1[18]>3 ){
      pBt->btsFlags |= BTS_READ_ONLY;
    }
    if( page1[19]>2 ){
    if( page1[19]>3 ){
      goto page1_init_failed;
    }

    /* If the write version is set to 2, this database should be accessed
    ** in WAL mode. If the log is not already open, open it now. Then 
    ** return SQLITE_OK and return without populating BtShared.pPage1.
    ** The caller detects this and calls this function again. This is
    ** required as the version of page 1 currently in the page1 buffer
    ** may not be the latest version - there may be a newer one in the log
    ** file.
    */
    if( page1[19]==2 && (pBt->btsFlags & BTS_NO_WAL)==0 ){
    if( page1[19]>=2 && (pBt->btsFlags & BTS_NO_WAL)==0 ){
      int isOpen = 0;
      rc = sqlite3PagerOpenWal(pBt->pPager, &isOpen);
      rc = sqlite3PagerOpenWal(pBt->pPager, (page1[19]==3), &isOpen);
      if( rc!=SQLITE_OK ){
        goto page1_init_failed;
      }else{
        setDefaultSyncFlag(pBt, SQLITE_DEFAULT_WAL_SYNCHRONOUS+1);
        if( isOpen==0 ){
          releasePageOne(pPage1);
          return SQLITE_OK;

** "write version" (single byte at byte offset 19) fields in the database
** header to iVersion.
*/
int sqlite3BtreeSetVersion(Btree *pBtree, int iVersion){
  BtShared *pBt = pBtree->pBt;
  int rc;                         /* Return code */
 
  assert( iVersion==1 || iVersion==2 );
  assert( iVersion==1 || iVersion==2 || iVersion==3 );

  /* If setting the version fields to 1, do not automatically open the
  ** WAL connection, even if the version fields are currently set to 2.
  */
  pBt->btsFlags &= ~BTS_NO_WAL;
  if( iVersion==1 ) pBt->btsFlags |= BTS_NO_WAL;

#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!=0)

#ifdef SQLITE_DIRECT_OVERFLOW_READ
/*
** Return true if page pgno can be read directly from the database file
** by the b-tree layer. This is the case if:
**
**   * the database file is open,
**   * there are no dirty pages in the cache, and

        ** a rollback transaction that switches from journal_mode=off
        ** to journal_mode=wal.
        */
        assert( p->eLock>=RESERVED_LOCK );
        assert( isOpen(p->jfd) 
             || p->journalMode==PAGER_JOURNALMODE_OFF 
             || p->journalMode==PAGER_JOURNALMODE_WAL 
             || p->journalMode==PAGER_JOURNALMODE_WAL2
        );
      }
      assert( pPager->dbOrigSize==pPager->dbFileSize );
      assert( pPager->dbOrigSize==pPager->dbHintSize );
      break;

    case PAGER_WRITER_DBMOD:
      assert( p->eLock==EXCLUSIVE_LOCK );
      assert( pPager->errCode==SQLITE_OK );
      assert( !pagerUseWal(pPager) );
      assert( p->eLock>=EXCLUSIVE_LOCK );
      assert( isOpen(p->jfd) 
           || p->journalMode==PAGER_JOURNALMODE_OFF 
           || p->journalMode==PAGER_JOURNALMODE_WAL 
           || p->journalMode==PAGER_JOURNALMODE_WAL2
           || (sqlite3OsDeviceCharacteristics(p->fd)&SQLITE_IOCAP_BATCH_ATOMIC)
      );
      assert( pPager->dbOrigSize<=pPager->dbHintSize );
      break;

    case PAGER_WRITER_FINISHED:
      assert( p->eLock==EXCLUSIVE_LOCK );
      assert( pPager->errCode==SQLITE_OK );
      assert( !pagerUseWal(pPager) );
      assert( isOpen(p->jfd) 
           || p->journalMode==PAGER_JOURNALMODE_OFF 
           || p->journalMode==PAGER_JOURNALMODE_WAL 
           || p->journalMode==PAGER_JOURNALMODE_WAL2
           || (sqlite3OsDeviceCharacteristics(p->fd)&SQLITE_IOCAP_BATCH_ATOMIC)
      );
      break;

    case PAGER_ERROR:
      /* There must be at least one outstanding reference to the pager if
      ** in ERROR state. Otherwise the pager should have already dropped

          ** 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)
      || (pPager->exclusiveMode && pPager->journalMode<PAGER_JOURNALMODE_WAL)
    ){
      rc = zeroJournalHdr(pPager, hasMaster||pPager->tempFile);
      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;
      assert( sqlite3JournalIsInMemory(pPager->jfd)==0 );
      assert( pPager->journalMode==PAGER_JOURNALMODE_DELETE 
           || pPager->journalMode==PAGER_JOURNALMODE_MEMORY 
           || pPager->journalMode==PAGER_JOURNALMODE_WAL 
           || pPager->journalMode==PAGER_JOURNALMODE_WAL
           || pPager->journalMode==PAGER_JOURNALMODE_WAL2
      );
      sqlite3OsClose(pPager->jfd);
      if( bDelete ){
        rc = sqlite3OsDelete(pPager->pVfs, pPager->zJournal, pPager->extraSync);
      }
    }
  }

  */
  sqlite3WalEndReadTransaction(pPager->pWal);

  rc = sqlite3WalBeginReadTransaction(pPager->pWal, &changed);
  if( rc!=SQLITE_OK || changed ){
    pager_reset(pPager);
    if( USEFETCH(pPager) ) sqlite3OsUnfetch(pPager->fd, 0, 0);
    assert( pPager->journalMode==PAGER_JOURNALMODE_WAL
         || pPager->journalMode==PAGER_JOURNALMODE_WAL2
    );
    pPager->journalMode = sqlite3WalJournalMode(pPager->pWal);
  }

  return rc;
}
#endif

/*

        rc = pagerPagecount(pPager, &nPage);
        if( rc ) return rc;
        if( nPage==0 ){
          rc = sqlite3OsDelete(pPager->pVfs, pPager->zWal, 0);
        }else{
          testcase( sqlite3PcachePagecount(pPager->pPCache)==0 );
          rc = sqlite3PagerOpenWal(pPager, 0);
          rc = sqlite3PagerOpenWal(pPager, 0, 0);
        }
      }else if( pPager->journalMode==PAGER_JOURNALMODE_WAL ){
      }else if( pPager->journalMode>=PAGER_JOURNALMODE_WAL ){
        pPager->journalMode = PAGER_JOURNALMODE_DELETE;
      }
    }
  }
  return rc;
}
#endif

  /* The eMode parameter is always valid */
  assert(      eMode==PAGER_JOURNALMODE_DELETE
            || eMode==PAGER_JOURNALMODE_TRUNCATE
            || eMode==PAGER_JOURNALMODE_PERSIST
            || eMode==PAGER_JOURNALMODE_OFF 
            || eMode==PAGER_JOURNALMODE_WAL 
            || eMode==PAGER_JOURNALMODE_WAL2
            || eMode==PAGER_JOURNALMODE_MEMORY );

  /* This routine is only called from the OP_JournalMode opcode, and
  ** the logic there will never allow a temporary file to be changed
  ** to WAL mode.
  */
  assert( pPager->tempFile==0 || eMode!=PAGER_JOURNALMODE_WAL );

    */
    assert( (PAGER_JOURNALMODE_TRUNCATE & 5)==1 );
    assert( (PAGER_JOURNALMODE_PERSIST & 5)==1 );
    assert( (PAGER_JOURNALMODE_DELETE & 5)==0 );
    assert( (PAGER_JOURNALMODE_MEMORY & 5)==4 );
    assert( (PAGER_JOURNALMODE_OFF & 5)==0 );
    assert( (PAGER_JOURNALMODE_WAL & 5)==5 );
    assert( (PAGER_JOURNALMODE_WAL2 & 5)==4 );

    assert( isOpen(pPager->fd) || pPager->exclusiveMode );
    if( !pPager->exclusiveMode && (eOld & 5)==1 && (eMode & 1)==0 ){
    if( !pPager->exclusiveMode && (eOld & 5)==1 && (eMode & 1)==0 
     && eMode!=PAGER_JOURNALMODE_WAL2       /* TODO: fix this if possible */
    ){

      /* In this case we would like to delete the journal file. If it is
      ** not possible, then that is not a problem. Deleting the journal file
      ** here is an optimization only.
      **
      ** Before deleting the journal file, obtain a RESERVED lock on the
      ** database file. This ensures that the journal file is not deleted

/*
** 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){
static int pagerOpenWal(Pager *pPager, int bWal2){
  int rc = SQLITE_OK;

  assert( pPager->pWal==0 && pPager->tempFile==0 );
  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->journalSizeLimit, &pPager->pWal
        pPager->journalSizeLimit, bWal2, &pPager->pWal
    );
  }
  pagerFixMaplimit(pPager);

  return rc;
}

**
** If the pager is open on a temp-file (or in-memory database), or if
** the WAL file is already open, set *pbOpen to 1 and return SQLITE_OK
** without doing anything.
*/
int sqlite3PagerOpenWal(
  Pager *pPager,                  /* Pager object */
  int bWal2,                      /* Open in wal2 mode if not already open */
  int *pbOpen                     /* OUT: Set to true if call is a no-op */
){
  int rc = SQLITE_OK;             /* Return code */

  assert( assert_pager_state(pPager) );
  assert( pPager->eState==PAGER_OPEN   || pbOpen );
  assert( pPager->eState==PAGER_READER || !pbOpen );
  assert( pbOpen==0 || *pbOpen==0 );
  assert( pbOpen!=0 || (!pPager->tempFile && !pPager->pWal) );

  if( !pPager->tempFile && !pPager->pWal ){
    if( !sqlite3PagerWalSupported(pPager) ) return SQLITE_CANTOPEN;

    /* Close any rollback journal previously open */
    sqlite3OsClose(pPager->jfd);

    rc = pagerOpenWal(pPager);
    rc = pagerOpenWal(pPager, bWal2);
    if( rc==SQLITE_OK ){
      pPager->journalMode = PAGER_JOURNALMODE_WAL;
      pPager->journalMode = bWal2?PAGER_JOURNALMODE_WAL2:PAGER_JOURNALMODE_WAL;
      pPager->eState = PAGER_OPEN;
    }
  }else{
    *pbOpen = 1;
  }

  return rc;

** EXCLUSIVE lock on the database file. If this cannot be obtained, an
** error (SQLITE_BUSY) is returned and the log connection is not closed.
** If successful, the EXCLUSIVE lock is not released before returning.
*/
int sqlite3PagerCloseWal(Pager *pPager, sqlite3 *db){
  int rc = SQLITE_OK;

  assert( pPager->journalMode==PAGER_JOURNALMODE_WAL );
  assert( pPager->journalMode==PAGER_JOURNALMODE_WAL 
       || pPager->journalMode==PAGER_JOURNALMODE_WAL2
  );

  /* If the log file is not already open, but does exist in the file-system,
  ** it may need to be checkpointed before the connection can switch to
  ** rollback mode. Open it now so this can happen.
  */
  if( !pPager->pWal ){
    int logexists = 0;
    rc = pagerLockDb(pPager, SHARED_LOCK);
    if( rc==SQLITE_OK ){
      rc = sqlite3OsAccess(
          pPager->pVfs, pPager->zWal, SQLITE_ACCESS_EXISTS, &logexists
      );
    }
    if( rc==SQLITE_OK && logexists ){
      rc = pagerOpenWal(pPager);
      rc = pagerOpenWal(pPager, 0);
    }
  }
    
  /* 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 ){

#define PAGER_JOURNALMODE_QUERY     (-1)  /* Query the value of journalmode */
#define PAGER_JOURNALMODE_DELETE      0   /* Commit by deleting journal file */
#define PAGER_JOURNALMODE_PERSIST     1   /* Commit by zeroing journal header */
#define PAGER_JOURNALMODE_OFF         2   /* Journal omitted.  */
#define PAGER_JOURNALMODE_TRUNCATE    3   /* Commit by truncating journal */
#define PAGER_JOURNALMODE_MEMORY      4   /* In-memory journal file */
#define PAGER_JOURNALMODE_WAL         5   /* Use write-ahead logging */
#define PAGER_JOURNALMODE_WAL2        6   /* Use write-ahead logging mode 2 */

#define isWalMode(x) ((x)==PAGER_JOURNALMODE_WAL || (x)==PAGER_JOURNALMODE_WAL2)

/*
** 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!=0)

/*
** Flags that make up the mask passed to sqlite3PagerGet().
*/
#define PAGER_GET_NOCONTENT     0x01  /* Do not load data from disk */
#define PAGER_GET_READONLY      0x02  /* Read-only page is acceptable */

int sqlite3PagerSharedLock(Pager *pPager);


#ifndef SQLITE_OMIT_WAL
  int sqlite3PagerCheckpoint(Pager *pPager, sqlite3*, int, int*, int*);
  int sqlite3PagerWalSupported(Pager *pPager);
  int sqlite3PagerWalCallback(Pager *pPager);
  int sqlite3PagerOpenWal(Pager *pPager, int *pisOpen);
  int sqlite3PagerOpenWal(Pager *pPager, int, int *pisOpen);
  int sqlite3PagerCloseWal(Pager *pPager, sqlite3*);
# ifdef SQLITE_ENABLE_SNAPSHOT
  int sqlite3PagerSnapshotGet(Pager *pPager, sqlite3_snapshot **ppSnapshot);
  int sqlite3PagerSnapshotOpen(Pager *pPager, sqlite3_snapshot *pSnapshot);
  int sqlite3PagerSnapshotRecover(Pager *pPager);
  int sqlite3PagerSnapshotCheck(Pager *pPager, sqlite3_snapshot *pSnapshot);
  void sqlite3PagerSnapshotUnlock(Pager *pPager);

** defined in pager.h. This function returns the associated lowercase
** journal-mode name.
*/
const char *sqlite3JournalModename(int eMode){
  static char * const azModeName[] = {
    "delete", "persist", "off", "truncate", "memory"
#ifndef SQLITE_OMIT_WAL
     , "wal"
     , "wal", "wal2"
#endif
  };
  assert( PAGER_JOURNALMODE_DELETE==0 );
  assert( PAGER_JOURNALMODE_PERSIST==1 );
  assert( PAGER_JOURNALMODE_OFF==2 );
  assert( PAGER_JOURNALMODE_TRUNCATE==3 );
  assert( PAGER_JOURNALMODE_MEMORY==4 );
  assert( PAGER_JOURNALMODE_WAL==5 );
  assert( PAGER_JOURNALMODE_WAL2==6 );
  assert( eMode>=0 && eMode<=ArraySize(azModeName) );

  if( eMode==ArraySize(azModeName) ) return 0;
  return azModeName[eMode];
}

/*

  sqlite3_free(aOut);
  fclose(out);
  Tcl_SetObjResult(interp, Tcl_NewIntObj(written));
  return TCL_OK;
}

/*
** USAGE:   hexio_get_int   HEXDATA
** USAGE:   hexio_get_int [-littleendian] HEXDATA
**
** Interpret the HEXDATA argument as a big-endian integer.  Return
** the value of that integer.  HEXDATA can contain between 2 and 8
** hexadecimal digits.
*/
static int SQLITE_TCLAPI hexio_get_int(
  void * clientData,
  Tcl_Interp *interp,
  int objc,
  Tcl_Obj *CONST objv[]
){
  int val;
  int nIn, nOut;
  const unsigned char *zIn;
  unsigned char *aOut;
  unsigned char aNum[4];
  int bLittle = 0;

  if( objc!=2 ){
    Tcl_WrongNumArgs(interp, 1, objv, "HEXDATA");
  if( objc==3 ){
    int n;
    char *z = Tcl_GetStringFromObj(objv[1], &n);
    if( n>=2 && n<=13 && memcmp(z, "-littleendian", n)==0 ){
      bLittle = 1;
    }
  }
  if( (objc-bLittle)!=2 ){
    Tcl_WrongNumArgs(interp, 1, objv, "[-littleendian] HEXDATA");
    return TCL_ERROR;
  }
  zIn = (const unsigned char *)Tcl_GetStringFromObj(objv[1], &nIn);
  zIn = (const unsigned char *)Tcl_GetStringFromObj(objv[1+bLittle], &nIn);
  aOut = sqlite3_malloc( nIn/2 );
  if( aOut==0 ){
    return TCL_ERROR;
  }
  nOut = sqlite3TestHexToBin(zIn, nIn, aOut);
  if( nOut>=4 ){
    memcpy(aNum, aOut, 4);
  }else{
    memset(aNum, 0, sizeof(aNum));
    memcpy(&aNum[4-nOut], aOut, nOut);
  }
  sqlite3_free(aOut);
  if( bLittle ){
    val = (aNum[3]<<24) | (aNum[2]<<16) | (aNum[1]<<8) | aNum[0];
  }else{
  val = (aNum[0]<<24) | (aNum[1]<<16) | (aNum[2]<<8) | aNum[3];
    val = (aNum[0]<<24) | (aNum[1]<<16) | (aNum[2]<<8) | aNum[3];
  }
  Tcl_SetObjResult(interp, Tcl_NewIntObj(val));
  return TCL_OK;
}


/*
** USAGE:   hexio_render_int16   INTEGER

#if defined(SQLITE_ENABLE_SESSION) && defined(SQLITE_ENABLE_PREUPDATE_HOOK)
  extern int TestSession_Init(Tcl_Interp*);
#endif
  extern int Md5_Init(Tcl_Interp*);
  extern int Fts5tcl_Init(Tcl_Interp *);
  extern int SqliteRbu_Init(Tcl_Interp*);
  extern int Sqlitetesttcl_Init(Tcl_Interp*);
  extern int Bgckpt_Init(Tcl_Interp*);
#if defined(SQLITE_ENABLE_FTS3) || defined(SQLITE_ENABLE_FTS4)
  extern int Sqlitetestfts3_Init(Tcl_Interp *interp);
#endif
#ifdef SQLITE_ENABLE_ZIPVFS
  extern int Zipvfs_Init(Tcl_Interp*);
#endif
  extern int TestExpert_Init(Tcl_Interp*);

  SqlitetestSyscall_Init(interp);
#if defined(SQLITE_ENABLE_SESSION) && defined(SQLITE_ENABLE_PREUPDATE_HOOK)
  TestSession_Init(interp);
#endif
  Fts5tcl_Init(interp);
  SqliteRbu_Init(interp);
  Sqlitetesttcl_Init(interp);
  Bgckpt_Init(interp);


#if defined(SQLITE_ENABLE_FTS3) || defined(SQLITE_ENABLE_FTS4)
  Sqlitetestfts3_Init(interp);
#endif
  TestExpert_Init(interp);
  Sqlitetest_window_Init(interp);

  eNew = pOp->p3;
  assert( eNew==PAGER_JOURNALMODE_DELETE 
       || eNew==PAGER_JOURNALMODE_TRUNCATE 
       || eNew==PAGER_JOURNALMODE_PERSIST 
       || eNew==PAGER_JOURNALMODE_OFF
       || eNew==PAGER_JOURNALMODE_MEMORY
       || eNew==PAGER_JOURNALMODE_WAL
       || eNew==PAGER_JOURNALMODE_WAL2
       || eNew==PAGER_JOURNALMODE_QUERY
  );
  assert( pOp->p1>=0 && pOp->p1<db->nDb );
  assert( p->readOnly==0 );

  pBt = db->aDb[pOp->p1].pBt;
  pPager = sqlite3BtreePager(pBt);
  eOld = sqlite3PagerGetJournalMode(pPager);
  if( eNew==PAGER_JOURNALMODE_QUERY ) eNew = eOld;
  if( !sqlite3PagerOkToChangeJournalMode(pPager) ) eNew = eOld;

#ifndef SQLITE_OMIT_WAL
  zFilename = sqlite3PagerFilename(pPager, 1);

  /* Do not allow a transition to journal_mode=WAL for a database
  ** in temporary storage or if the VFS does not support shared memory 
  */
  if( eNew==PAGER_JOURNALMODE_WAL
  if( isWalMode(eNew)
   && (sqlite3Strlen30(zFilename)==0           /* Temp file */
       || !sqlite3PagerWalSupported(pPager))   /* No shared-memory support */
  ){
    eNew = eOld;
  }

  if( (eNew!=eOld)
   && (eOld==PAGER_JOURNALMODE_WAL || eNew==PAGER_JOURNALMODE_WAL)
  ){
  if( eNew!=eOld && (isWalMode(eNew) || isWalMode(eOld)) ){

    /* Prevent changing directly to wal2 from wal mode. And vice versa. */
    if( isWalMode(eNew) && isWalMode(eOld) ){
      rc = SQLITE_ERROR;
      sqlite3VdbeError(p, "cannot change from %s to %s mode",
          sqlite3JournalModename(eOld), sqlite3JournalModename(eNew)
      );
      goto abort_due_to_error;
    }

    /* Prevent switching into or out of wal/wal2 mode mid-transaction */
    if( !db->autoCommit || db->nVdbeRead>1 ){
      rc = SQLITE_ERROR;
      sqlite3VdbeError(p,
          "cannot change %s wal mode from within a transaction",
          (eNew==PAGER_JOURNALMODE_WAL ? "into" : "out of")
      );
      goto abort_due_to_error;
    }else{
 
      if( eOld==PAGER_JOURNALMODE_WAL ){
        /* If leaving WAL mode, close the log file. If successful, the call
        ** to PagerCloseWal() checkpoints and deletes the write-ahead-log 
        ** file. An EXCLUSIVE lock may still be held on the database file 
        ** after a successful return. 
        */
        rc = sqlite3PagerCloseWal(pPager, db);
        if( rc==SQLITE_OK ){
          sqlite3PagerSetJournalMode(pPager, eNew);
        }
      }else if( eOld==PAGER_JOURNALMODE_MEMORY ){
        /* Cannot transition directly from MEMORY to WAL.  Use mode OFF
        ** as an intermediate */
        sqlite3PagerSetJournalMode(pPager, PAGER_JOURNALMODE_OFF);
      }
  
      /* Open a transaction on the database file. Regardless of the journal
      ** mode, this transaction always uses a rollback journal.
      */
      assert( sqlite3BtreeIsInTrans(pBt)==0 );
      if( rc==SQLITE_OK ){
        rc = sqlite3BtreeSetVersion(pBt, (eNew==PAGER_JOURNALMODE_WAL ? 2 : 1));
    }
 
    if( isWalMode(eOld) ){
      /* If leaving WAL mode, close the log file. If successful, the call
      ** to PagerCloseWal() checkpoints and deletes the write-ahead-log 
      ** file. An EXCLUSIVE lock may still be held on the database file 
      ** after a successful return. 
      */
      rc = sqlite3PagerCloseWal(pPager, db);
      if( rc==SQLITE_OK ){
        sqlite3PagerSetJournalMode(pPager, eNew);
      }
    }else if( eOld==PAGER_JOURNALMODE_MEMORY ){
      /* Cannot transition directly from MEMORY to WAL.  Use mode OFF
      ** as an intermediate */
      sqlite3PagerSetJournalMode(pPager, PAGER_JOURNALMODE_OFF);
    }

    /* Open a transaction on the database file. Regardless of the journal
    ** mode, this transaction always uses a rollback journal.
    */
    assert( sqlite3BtreeIsInTrans(pBt)==0 );
    if( rc==SQLITE_OK ){
      /* 1==rollback, 2==wal, 3==wal2 */
      rc = sqlite3BtreeSetVersion(pBt, 
      }
          1 + isWalMode(eNew) + (eNew==PAGER_JOURNALMODE_WAL2)
      );
    }
  }
#endif /* ifndef SQLITE_OMIT_WAL */

  if( rc ) eNew = eOld;
  eNew = sqlite3PagerSetJournalMode(pPager, eNew);

** being considered valid at the same time and being checkpointing together
** following a crash.
**
** READER ALGORITHM
**
** To read a page from the database (call it page number P), a reader
** first checks the WAL to see if it contains page P.  If so, then the
** last valid instance of page P that is a followed by a commit frame
** last valid instance of page P that is followed by a commit frame
** or is a commit frame itself becomes the value read.  If the WAL
** contains no copies of page P that are valid and which are a commit
** frame or are followed by a commit frame, then page P is read from
** the database file.
**
** To start a read transaction, the reader records the index of the last
** valid frame in the WAL.  The reader uses this recorded "mxFrame" value

**
** Note that entries are added in order of increasing K.  Hence, one
** reader might be using some value K0 and a second reader that started
** at a later time (after additional transactions were added to the WAL
** and to the wal-index) might be using a different value K1, where K1>K0.
** Both readers can use the same hash table and mapping section to get
** the correct result.  There may be entries in the hash table with
** K>K0 but to the first reader, those entries will appear to be unused
** K>K0, but to the first reader those entries will appear to be unused
** slots in the hash table and so the first reader will get an answer as
** if no values greater than K0 had ever been inserted into the hash table
** in the first place - which is what reader one wants.  Meanwhile, the
** second reader using K1 will see additional values that were inserted
** later, which is exactly what reader two wants.  
**
** When a rollback occurs, the value of K is decreased. Hash table entries
** that correspond to frames greater than the new K value are removed
** from the hash table at this point.
*/

/*
** WAL2 NOTES
**
** This file also contains the implementation of "wal2" mode - activated
** using "PRAGMA journal_mode = wal2". Wal2 mode is very similar to wal
** mode, except that it uses two wal files instead of one. Under some
** circumstances, wal2 mode provides more concurrency than legacy wal 
** mode.
**
** THE PROBLEM WAL2 SOLVES:
**
** In legacy wal mode, if a writer wishes to write to the database while
** a checkpoint is ongoing, it may append frames to the existing wal file.
** This means that after the checkpoint has finished, the wal file consists
** of a large block of checkpointed frames, followed by a block of
** uncheckpointed frames. In a deployment that features a high volume of
** write traffic, this may mean that the wal file is never completely
** checkpointed. And so grows indefinitely.
**
** An alternative is to use "PRAGMA wal_checkpoint=RESTART" or similar to
** force a complete checkpoint of the wal file. But this must:
**
**   1) Wait on all existing readers to finish,
**   2) Wait on any existing writer, and then block all new writers,
**   3) Do the checkpoint,
**   4) Wait on any new readers that started during steps 2 and 3. Writers
**      are still blocked during this step.
**
** This means that in order to avoid the wal file growing indefinitely 
** in a busy system, writers must periodically pause to allow a checkpoint
** to complete. In a system with long running readers, such pauses may be
** for a non-trivial amount of time.
**
** OVERVIEW OF SOLUTION
**
** Wal2 mode uses two wal files. After writers have grown the first wal 
** file to a pre-configured size, they begin appending transactions to 
** the second wal file. Once all existing readers are reading snapshots
** new enough to include the entire first wal file, a checkpointer can
** checkpoint it.
**
** Meanwhile, writers are writing transactions to the second wal file.
** Once that wal file has grown larger than the pre-configured size, each
** new writer checks if:
**
**    * the first wal file has been checkpointed, and if so, if
**    * there are no readers still reading from the first wal file (once
**      it has been checkpointed, new readers read only from the second
**      wal file).
**
** If both these conditions are true, the writer may switch back to the
** first wal file. Eventually, a checkpointer can checkpoint the second
** wal file, and so on.
**
** The wal file that writers are currently appending to (the one they
** don't have to check the above two criteria before writing to) is called
** the "current" wal file.
**
** The first wal file takes the same name as the wal file in legacy wal
** mode systems - "<db>-wal". The second is named "<db>-wal2".

**
** CHECKPOINTS
**
** The "pre-configured size" mentioned above is the value set by 
** "PRAGMA journal_size_limit". Or, if journal_size_limit is not set, 
** 1000 pages.
**
** There is only a single type of checkpoint in wal2 mode (no "truncate",
** "restart" etc.), and it always checkpoints the entire contents of a single
** wal file. A wal file cannot be checkpointed until after a writer has written
** the first transaction into the other wal file and all readers are reading a
** snapshot that includes at least one transaction from the other wal file.
**
** The wal-hook, if one is registered, is invoked after a write-transaction
** is committed, just as it is in legacy wal mode. The integer parameter
** passed to the wal-hook is the total number of uncheckpointed frames in both
** wal files. Except, the parameter is set to zero if there is no frames 
** that may be checkpointed. This happens in two scenarios:
**
**   1. The "other" wal file (the one that the writer did not just append to)
**      is completely empty, or
**
**   2. The "other" wal file (the one that the writer did not just append to)
**      has already been checkpointed.
**
**
** WAL FILE FORMAT
**
** The file format used for each wal file in wal2 mode is the same as for
** legacy wal mode.  Except, the file format field is set to 3021000 
** instead of 3007000.
**
** WAL-INDEX FORMAT
**
** The wal-index format is also very similar. Even though there are two
** wal files, there is still a single wal-index shared-memory area (*-shm
** file with the default unix or win32 VFS). The wal-index header is the
** same size, with the following exceptions it has the same format:
**
**   * The version field is set to 3021000 instead of 3007000.
**
**   * An unused 32-bit field in the legacy wal-index header is
**     now used to store (a) a single bit indicating which of the
**     two wal files writers should append to and (b) the number
**     of frames in the second wal file (31 bits).
**
** The first hash table in the wal-index contains entries corresponding
** to the first HASHTABLE_NPAGE_ONE frames stored in the first wal file.
** The second hash table in the wal-index contains entries indexing the
** first HASHTABLE_NPAGE frames in the second wal file. The third hash
** table contains the next HASHTABLE_NPAGE frames in the first wal file,
** and so on.
**
** LOCKS
**
** Read-locks are simpler than for legacy wal mode. There are no locking
** slots that contain frame numbers. Instead, there are four distinct
** combinations of read locks a reader may hold:
**
**   WAL_LOCK_PART1:       "part" lock on first wal, none of second.
**   WAL_LOCK_PART1_FULL2: "part" lock on first wal, "full" of second.
**   WAL_LOCK_PART2: no lock on first wal, "part" lock on second.
**   WAL_LOCK_PART2_FULL1: "full" lock on first wal, "part" lock on second.
**
** When a reader reads the wal-index header as part of opening a read
** transaction, it takes a "part" lock on the current wal file. "Part" 
** because the wal file may grow while the read transaction is active, in 
** which case the reader would be reading only part of the wal file. 
** A part lock prevents a checkpointer from checkpointing the wal file 
** on which it is held.
**
** If there is data in the non-current wal file that has not been 
** checkpointed, the reader takes a "full" lock on that wal file. A 
** "full" lock indicates that the reader is using the entire wal file.
** A full lock prevents a writer from overwriting the wal file on which
** it is held, but does not prevent a checkpointer from checkpointing 
** it.
**
** There is still a single WRITER and a single CHECKPOINTER lock. The
** recovery procedure still takes the same exclusive lock on the entire
** range of SQLITE_SHM_NLOCK shm-locks. This works because the read-locks
** above use four of the six read-locking slots used by legacy wal mode.
**
** STARTUP/RECOVERY
**
** The read and write version fields of the database header in a wal2
** database are set to 0x03, instead of 0x02 as in legacy wal mode.
**
** The wal file format used in wal2 mode is the same as the format used
** in legacy wal mode. However, in order to support recovery, there are two
** differences in the way wal file header fields are populated, as follows:
**
**   * When the first wal file is first created, the "nCkpt" field in
**     the wal file header is set to 0. Thereafter, each time the writer
**     switches wal file, it sets the nCkpt field in the new wal file
**     header to ((nCkpt0 + 1) & 0x0F), where nCkpt0 is the value in
**     the previous wal file header. This means that the first wal file
**     always has an even value in the nCkpt field, and the second wal
**     file always has an odd value.
**
**   * When a writer switches wal file, it sets the salt values in the
**     new wal file to a copy of the checksum for the final frame in
**     the previous wal file.
**
** Recovery proceeds as follows:
**
** 1. Each wal file is recovered separately. Except, if the first wal 
**    file does not exist or is zero bytes in size, the second wal file
**    is truncated to zero bytes before it is "recovered".
**
** 2. If both wal files contain valid headers, then the nCkpt fields
**    are compared to see which of the two wal files is older. If the
**    salt keys in the second wal file match the final frame checksum 
**    in the older wal file, then both wal files are used. Otherwise,
**    the newer wal file is ignored.
**
** 3. Or, if only one or neither of the wal files has a valid header, 
**    then only a single or no wal files are recovered into the 
**    reconstructed wal-index.
**
** Refer to header comments for walIndexRecover() for further details.
*/

#ifndef SQLITE_OMIT_WAL

#include "wal.h"

/*
** Trace output macros
*/

# define AtomicStore(PTR,VAL)  __atomic_store_n((PTR),(VAL),__ATOMIC_RELAXED)
#else
# define AtomicLoad(PTR)       (*(PTR))
# define AtomicStore(PTR,VAL)  (*(PTR) = (VAL))
#endif

/*
** The maximum (and only) versions of the wal and wal-index formats
** that may be interpreted by this version of SQLite.
**
** If a client begins recovering a WAL file and finds that (a) the checksum
** values in the wal-header are correct and (b) the version field is not
** WAL_MAX_VERSION, recovery fails and SQLite returns SQLITE_CANTOPEN.
** Both the wal-file and the wal-index contain version fields 
** indicating the current version of the system. If a client
** reads the header of a wal file (as part of recovery), or the
** wal-index (as part of opening a read transaction) and (a) the
** header checksum is correct but (b) the version field is not
** recognized, the operation fails with SQLITE_CANTOPEN.
**
** Similarly, if a client successfully reads a wal-index header (i.e. the 
** checksum test is successful) and finds that the version field is not
** Currently, clients support both version-1 ("journal_mode=wal") and
** version-2 ("journal_mode=wal2"). Legacy clients may support version-1
** WALINDEX_MAX_VERSION, then no read-transaction is opened and SQLite
** returns SQLITE_CANTOPEN.
** only.
*/
#define WAL_MAX_VERSION      3007000
#define WALINDEX_MAX_VERSION 3007000
#define WAL_VERSION1 3007000      /* For "journal_mode=wal" */
#define WAL_VERSION2 3021000      /* For "journal_mode=wal2" */

/*
** Index numbers for various locking bytes.   WAL_NREADER is the number
** of available reader locks and should be at least 3.  The default
** is SQLITE_SHM_NLOCK==8 and  WAL_NREADER==5.
**
** Technically, the various VFSes are free to implement these locks however

#define WAL_WRITE_LOCK         0
#define WAL_ALL_BUT_WRITE      1
#define WAL_CKPT_LOCK          1
#define WAL_RECOVER_LOCK       2
#define WAL_READ_LOCK(I)       (3+(I))
#define WAL_NREADER            (SQLITE_SHM_NLOCK-3)

/*
** Values that may be stored in Wal.readLock in wal2 mode.
**
** In wal mode, the Wal.readLock member is set to -1 when no read-lock
** is held, or else is the index of the read-mark on which a lock is
** held.
**
** In wal2 mode, a value of -1 still indicates that no read-lock is held.
** And a non-zero value still represents the index of the read-mark on
** which a lock is held. There are two differences:
**
**   1. wal2 mode never uses read-mark 0.
**
**   2. locks on each read-mark have a different interpretation, as 
**      indicated by the symbolic names below.
*/
#define WAL_LOCK_NONE        -1
#define WAL_LOCK_PART1        1
#define WAL_LOCK_PART1_FULL2  2
#define WAL_LOCK_PART2_FULL1  3
#define WAL_LOCK_PART2        4

/* 
** This constant is used in wal2 mode only.
**
** In wal2 mode, when committing a transaction, if the current wal file 
** is sufficiently large and there are no conflicting locks held, the
** writer writes the new transaction into the start of the other wal
** file. Usually, "sufficiently large" is defined by the value configured
** using "PRAGMA journal_size_limit". However, if no such value has been
** configured, sufficiently large defaults to WAL_DEFAULT_WALSIZE frames.
*/
#define WAL_DEFAULT_WALSIZE 1000

/* Object declarations */
typedef struct WalIndexHdr WalIndexHdr;
typedef struct WalIterator WalIterator;
typedef struct WalCkptInfo WalCkptInfo;


/*
** The following object holds a copy of the wal-index header content.
**
** The actual header in the wal-index consists of two copies of this
** object followed by one instance of the WalCkptInfo object.
** For all versions of SQLite through 3.10.0 and probably beyond,
** the locking bytes (WalCkptInfo.aLock) start at offset 120 and
** the total header size is 136 bytes.
**
** The szPage value can be any power of 2 between 512 and 32768, inclusive.
** Or it can be 1 to represent a 65536-byte page.  The latter case was
** added in 3.7.1 when support for 64K pages was added.  
**
** WAL2 mode notes: Member variable mxFrame2 is only used in wal2 mode
** (when iVersion is set to WAL_VERSION2). The lower 31 bits store
** the maximum frame number in file *-wal2. The most significant bit
** is a flag - set if clients are currently appending to *-wal2, clear
** otherwise.
*/
struct WalIndexHdr {
  u32 iVersion;                   /* Wal-index version */
  u32 unused;                     /* Unused (padding) field */
  u32 mxFrame2;                   /* See "WAL2 mode notes" above */
  u32 iChange;                    /* Counter incremented each transaction */
  u8 isInit;                      /* 1 when initialized */
  u8 bigEndCksum;                 /* True if checksums in WAL are big-endian */
  u16 szPage;                     /* Database page size in bytes. 1==64K */
  u32 mxFrame;                    /* Index of last valid frame in the WAL */
  u32 mxFrame;                    /* Index of last valid frame in each WAL */
  u32 nPage;                      /* Size of database in pages */
  u32 aFrameCksum[2];             /* Checksum of last frame in log */
  u32 aSalt[2];                   /* Two salt values copied from WAL header */
  u32 aCksum[2];                  /* Checksum over all prior fields */
};

/*
** The following macros and functions are get/set methods for the maximum
** frame numbers and current wal file values stored in the WalIndexHdr
** structure. These are helpful because of the unorthodox way in which
** the values are stored in wal2 mode (see above). They are equivalent
** to functions with the following signatures.
**
**   u32  walidxGetMxFrame(WalIndexHdr*, int iWal);          // get mxFrame
**   void walidxSetMxFrame(WalIndexHdr*, int iWal, u32 val); // set mxFrame
**   int  walidxGetFile(WalIndexHdr*)                        // get file
**   void walidxSetFile(WalIndexHdr*, int val);              // set file
*/
#define walidxGetMxFrame(pHdr, iWal) \
  ((iWal) ? ((pHdr)->mxFrame2 & 0x7FFFFFFF) : (pHdr)->mxFrame)

static void walidxSetMxFrame(WalIndexHdr *pHdr, int iWal, u32 mxFrame){
  if( iWal ){
    pHdr->mxFrame2 = (pHdr->mxFrame2 & 0x80000000) | mxFrame;
  }else{
    pHdr->mxFrame = mxFrame;
  }
  assert( walidxGetMxFrame(pHdr, iWal)==mxFrame );
}

#define walidxGetFile(pHdr) ((pHdr)->mxFrame2 >> 31)

#define walidxSetFile(pHdr, iWal) (                                   \
    (pHdr)->mxFrame2 = ((pHdr)->mxFrame2 & 0x7FFFFFFF) | ((iWal)<<31) \
)

/*
** Argument is a pointer to a Wal structure. Return true if the current
** cache of the wal-index header indicates "journal_mode=wal2" mode, or
** false otherwise.
*/
#define isWalMode2(pWal) ((pWal)->hdr.iVersion==WAL_VERSION2)

/*
** A copy of the following object occurs in the wal-index immediately
** following the second copy of the WalIndexHdr.  This object stores
** information used by checkpoint.
**
** nBackfill is the number of frames in the WAL that have been written
** back into the database. (We call the act of moving content from WAL to

/*
** An open write-ahead log file is represented by an instance of the
** following object.
*/
struct Wal {
  sqlite3_vfs *pVfs;         /* The VFS used to create pDbFd */
  sqlite3_file *pDbFd;       /* File handle for the database file */
  sqlite3_file *pWalFd;      /* File handle for WAL file */
  sqlite3_file *apWalFd[2];  /* File handle for "*-wal" and "*-wal2" */
  u32 iCallback;             /* Value to pass to log callback (or 0) */
  i64 mxWalSize;             /* Truncate WAL to this size upon reset */
  int nWiData;               /* Size of array apWiData */
  int szFirstBlock;          /* Size of first block written to WAL file */
  volatile u32 **apWiData;   /* Pointer to wal-index content in memory */
  u32 szPage;                /* Database page size */
  i16 readLock;              /* Which read lock is being held.  -1 for none */

  u8 padToSectorBoundary;    /* Pad transactions out to the next sector */
  u8 bShmUnreliable;         /* SHM content is read-only and unreliable */
  WalIndexHdr hdr;           /* Wal-index header for current transaction */
  u32 minFrame;              /* Ignore wal frames before this one */
  u32 iReCksum;              /* On commit, recalculate checksums from here */
  u32 nPriorFrame;           /* For sqlite3WalInfo() */
  const char *zWalName;      /* Name of WAL file */
  char *zWalName2;           /* Name of second WAL file */
  u32 nCkpt;                 /* Checkpoint sequence counter in the wal-header */
#ifdef SQLITE_DEBUG
  u8 lockError;              /* True if a locking error has occurred */
#endif
#ifdef SQLITE_ENABLE_SNAPSHOT
  WalIndexHdr *pSnapshot;    /* Start transaction here if not NULL */
#endif
  int bWal2;                 /* bWal2 flag passed to WalOpen() */
};

/*
** Candidate values for Wal.exclusiveMode.
*/
#define WAL_NORMAL_MODE     0
#define WAL_EXCLUSIVE_MODE  1     

*/
static void walIndexWriteHdr(Wal *pWal){
  volatile WalIndexHdr *aHdr = walIndexHdr(pWal);
  const int nCksum = offsetof(WalIndexHdr, aCksum);

  assert( pWal->writeLock );
  pWal->hdr.isInit = 1;
  pWal->hdr.iVersion = WALINDEX_MAX_VERSION;
  assert( pWal->hdr.iVersion==WAL_VERSION1||pWal->hdr.iVersion==WAL_VERSION2 );
  walChecksumBytes(1, (u8*)&pWal->hdr, nCksum, 0, pWal->hdr.aCksum);
  memcpy((void*)&aHdr[1], (const void*)&pWal->hdr, sizeof(WalIndexHdr));
  walShmBarrier(pWal);
  memcpy((void*)&aHdr[0], (const void*)&pWal->hdr, sizeof(WalIndexHdr));
}

/*

  */
  pgno = sqlite3Get4byte(&aFrame[0]);
  if( pgno==0 ){
    return 0;
  }

  /* A frame is only valid if a checksum of the WAL header,
  ** all prior frams, the first 16 bytes of this frame-header, 
  ** all prior frames, the first 16 bytes of this frame-header, 
  ** and the frame-data matches the checksum in the last 8 
  ** bytes of this frame-header.
  */
  nativeCksum = (pWal->hdr.bigEndCksum==SQLITE_BIGENDIAN);
  walChecksumBytes(nativeCksum, aFrame, 8, aCksum, aCksum);
  walChecksumBytes(nativeCksum, aData, pWal->szPage, aCksum, aCksum);
  if( aCksum[0]!=sqlite3Get4byte(&aFrame[16]) 

    }else{
      pLoc->iZero = HASHTABLE_NPAGE_ONE + (iHash-1)*HASHTABLE_NPAGE;
    }
    pLoc->aPgno = &pLoc->aPgno[-1];
  }
  return rc;
}

static u32 walExternalEncode(int iWal, u32 iFrame){
  u32 iRet;
  if( iWal ){
    iRet = HASHTABLE_NPAGE_ONE + iFrame;
    iRet += ((iFrame-1) / HASHTABLE_NPAGE) * HASHTABLE_NPAGE;
  }else{
    iRet = iFrame;
    iFrame += HASHTABLE_NPAGE - HASHTABLE_NPAGE_ONE;
    iRet += ((iFrame-1) / HASHTABLE_NPAGE) * HASHTABLE_NPAGE;
  }
  return iRet;
}

/*
** Parameter iExternal is an external frame identifier. This function
** transforms it to a wal file number (0 or 1) and frame number within
** this wal file (reported via output parameter *piRead).
*/
static int walExternalDecode(u32 iExternal, u32 *piRead){
  int iHash = (iExternal+HASHTABLE_NPAGE-HASHTABLE_NPAGE_ONE-1)/HASHTABLE_NPAGE;

  if( 0==(iHash & 0x01) ){
    /* A frame in wal file 0 */
    *piRead = (iExternal <= HASHTABLE_NPAGE_ONE) ? iExternal :
      iExternal - (iHash/2) * HASHTABLE_NPAGE;
    return 0;
  }

  *piRead = iExternal - HASHTABLE_NPAGE_ONE - ((iHash-1)/2) * HASHTABLE_NPAGE;
  return 1;
}

/*
** Return the number of the wal-index page that contains the hash-table
** and page-number array that contain entries corresponding to WAL frame
** iFrame. The wal-index is broken up into 32KB pages. Wal-index pages 
** are numbered starting from 0.
*/
static int walFramePage(u32 iFrame){
  int iHash = (iFrame+HASHTABLE_NPAGE-HASHTABLE_NPAGE_ONE-1) / HASHTABLE_NPAGE;
  assert( (iHash==0 || iFrame>HASHTABLE_NPAGE_ONE)
       && (iHash>=1 || iFrame<=HASHTABLE_NPAGE_ONE)
       && (iHash<=1 || iFrame>(HASHTABLE_NPAGE_ONE+HASHTABLE_NPAGE))
       && (iHash>=2 || iFrame<=HASHTABLE_NPAGE_ONE+HASHTABLE_NPAGE)
       && (iHash<=2 || iFrame>(HASHTABLE_NPAGE_ONE+2*HASHTABLE_NPAGE))
  );
  return iHash;
}

/*
** Return the index of the hash-table corresponding to frame iFrame of wal
** file iWal.
*/
static int walFramePage2(int iWal, u32 iFrame){
  int iRet;
  assert( iWal==0 || iWal==1 );
  assert( iFrame>0 );
  if( iWal==0 ){
    iRet = 2*((iFrame+HASHTABLE_NPAGE-HASHTABLE_NPAGE_ONE-1)/HASHTABLE_NPAGE);
  }else{
    iRet = 1 + 2 * ((iFrame-1) / HASHTABLE_NPAGE);
  }
  return iRet;
}

/*
** Return the page number associated with frame iFrame in this WAL.
*/
static u32 walFramePgno(Wal *pWal, u32 iFrame){
  int iHash = walFramePage(iFrame);
  if( iHash==0 ){
    return pWal->apWiData[0][WALINDEX_HDR_SIZE/sizeof(u32) + iFrame - 1];
  }
  return pWal->apWiData[iHash][(iFrame-1-HASHTABLE_NPAGE_ONE)%HASHTABLE_NPAGE];
}

static u32 walFramePgno2(Wal *pWal, int iWal, u32 iFrame){
  return walFramePgno(pWal, walExternalEncode(iWal, iFrame));
}

/*
** Remove entries from the hash table that point to WAL slots greater
** than pWal->hdr.mxFrame.
**
** This function is called whenever pWal->hdr.mxFrame is decreased due
** to a rollback or savepoint.
**
** At most only the hash table containing pWal->hdr.mxFrame needs to be
** updated.  Any later hash tables will be automatically cleared when
** pWal->hdr.mxFrame advances to the point where those hash tables are
** actually needed.
*/
static void walCleanupHash(Wal *pWal){
  WalHashLoc sLoc;                /* Hash table location */
  int iLimit = 0;                 /* Zero values greater than this */
  int nByte;                      /* Number of bytes to zero in aPgno[] */
  int i;                          /* Used to iterate through aHash[] */
  int iWal = walidxGetFile(&pWal->hdr);
  u32 mxFrame = walidxGetMxFrame(&pWal->hdr, iWal);

  u32 iExternal;
  if( isWalMode2(pWal) ){
    iExternal = walExternalEncode(iWal, mxFrame);
  }else{
    assert( iWal==0 );
    iExternal = mxFrame;
  }

  assert( pWal->writeLock );
  testcase( pWal->hdr.mxFrame==HASHTABLE_NPAGE_ONE-1 );
  testcase( pWal->hdr.mxFrame==HASHTABLE_NPAGE_ONE );
  testcase( pWal->hdr.mxFrame==HASHTABLE_NPAGE_ONE+1 );
  testcase( mxFrame==HASHTABLE_NPAGE_ONE-1 );
  testcase( mxFrame==HASHTABLE_NPAGE_ONE );
  testcase( mxFrame==HASHTABLE_NPAGE_ONE+1 );

  if( pWal->hdr.mxFrame==0 ) return;
  if( mxFrame==0 ) return;

  /* Obtain pointers to the hash-table and page-number array containing 
  ** the entry that corresponds to frame pWal->hdr.mxFrame. It is guaranteed
  ** the entry that corresponds to frame pWal->hdr.mxFrame.  */
  ** that the page said hash-table and array reside on is already mapped.
  */
  assert( pWal->nWiData>walFramePage(pWal->hdr.mxFrame) );
  assert( pWal->apWiData[walFramePage(pWal->hdr.mxFrame)] );
  walHashGet(pWal, walFramePage(pWal->hdr.mxFrame), &sLoc);
  assert( pWal->nWiData>walFramePage(iExternal) );
  assert( pWal->apWiData[walFramePage(iExternal)] );
  walHashGet(pWal, walFramePage(iExternal), &sLoc);

  /* Zero all hash-table entries that correspond to frame numbers greater
  ** than pWal->hdr.mxFrame.
  */
  iLimit = pWal->hdr.mxFrame - sLoc.iZero;
  iLimit = iExternal - sLoc.iZero;
  assert( iLimit>0 );
  for(i=0; i<HASHTABLE_NSLOT; i++){
    if( sLoc.aHash[i]>iLimit ){
      sLoc.aHash[i] = 0;
    }
  }
  
  /* Zero the entries in the aPgno array that correspond to frames with
  ** frame numbers greater than pWal->hdr.mxFrame. 
  ** frame numbers greater than pWal->hdr.mxFrame.  */
  */
  nByte = (int)((char *)sLoc.aHash - (char *)&sLoc.aPgno[iLimit+1]);
  memset((void *)&sLoc.aPgno[iLimit+1], 0, nByte);

#ifdef SQLITE_ENABLE_EXPENSIVE_ASSERT
  /* Verify that the every entry in the mapping region is still reachable
  ** via the hash table even after the cleanup.
  */
  if( iLimit ){
    int j;           /* Loop counter */
    int iKey;        /* Hash key */
    for(j=1; j<=iLimit; j++){
      for(iKey=walHash(sLoc.aPgno[j]);sLoc.aHash[iKey];iKey=walNextHash(iKey)){
        if( sLoc.aHash[iKey]==j ) break;
      }
      assert( sLoc.aHash[iKey]==j );
    }
  }
#endif /* SQLITE_ENABLE_EXPENSIVE_ASSERT */
}


/*
** Set an entry in the wal-index that will map database page number
** pPage into WAL frame iFrame.
*/
static int walIndexAppend(Wal *pWal, u32 iFrame, u32 iPage){
static int walIndexAppend(Wal *pWal, int iWal, u32 iFrame, u32 iPage){
  int rc;                         /* Return code */
  WalHashLoc sLoc;                /* Wal-index hash table location */
  u32 iExternal;

  rc = walHashGet(pWal, walFramePage(iFrame), &sLoc);
  
  if( isWalMode2(pWal) ){
    iExternal = walExternalEncode(iWal, iFrame);
  }else{
    assert( iWal==0 );
    iExternal = iFrame;
  }

  rc = walHashGet(pWal, walFramePage(iExternal), &sLoc);

  /* Assuming the wal-index file was successfully mapped, populate the
  ** page number array and hash table entry.
  */
  if( rc==SQLITE_OK ){
    int iKey;                     /* Hash table key */
    int idx;                      /* Value to write to hash-table slot */
    int nCollide;                 /* Number of hash collisions */

    idx = iFrame - sLoc.iZero;
    idx = iExternal - sLoc.iZero;
    assert( idx <= HASHTABLE_NSLOT/2 + 1 );
    
    /* If this is the first entry to be added to this hash-table, zero the
    ** entire hash table and aPgno[] array before proceeding. 
    */
    if( idx==1 ){
      int nByte = (int)((u8 *)&sLoc.aHash[HASHTABLE_NSLOT]

#endif /* SQLITE_ENABLE_EXPENSIVE_ASSERT */
  }


  return rc;
}

/*
** Recover a single wal file - *-wal if iWal==0, or *-wal2 if iWal==1.
*/
static int walIndexRecoverOne(Wal *pWal, int iWal, u32 *pnCkpt, int *pbZero){
  i64 nSize;                      /* Size of log file */
  u32 aFrameCksum[2] = {0, 0};
  int rc;
  sqlite3_file *pWalFd = pWal->apWalFd[iWal];

  assert( iWal==0 || iWal==1 );

  memset(&pWal->hdr, 0, sizeof(WalIndexHdr));
  sqlite3_randomness(8, pWal->hdr.aSalt);

  rc = sqlite3OsFileSize(pWalFd, &nSize);
  if( rc==SQLITE_OK ){
    if( nSize>WAL_HDRSIZE ){
      u8 aBuf[WAL_HDRSIZE];         /* Buffer to load WAL header into */
      u8 *aFrame = 0;               /* Malloc'd buffer to load entire frame */
      int szFrame;                  /* Number of bytes in buffer aFrame[] */
      u8 *aData;                    /* Pointer to data part of aFrame buffer */
      int iFrame;                   /* Index of last frame read */
      i64 iOffset;                  /* Next offset to read from log file */
      int szPage;                   /* Page size according to the log */
      u32 magic;                    /* Magic value read from WAL header */
      u32 version;                  /* Magic value read from WAL header */
      int isValid;                  /* True if this frame is valid */
  
      /* Read in the WAL header. */
      rc = sqlite3OsRead(pWalFd, aBuf, WAL_HDRSIZE, 0);
      if( rc!=SQLITE_OK ){
        return rc;
      }
  
      /* If the database page size is not a power of two, or is greater than
      ** SQLITE_MAX_PAGE_SIZE, conclude that the WAL file contains no valid 
      ** data. Similarly, if the 'magic' value is invalid, ignore the whole
      ** WAL file.
      */
      magic = sqlite3Get4byte(&aBuf[0]);
      szPage = sqlite3Get4byte(&aBuf[8]);
      if( (magic&0xFFFFFFFE)!=WAL_MAGIC 
       || szPage&(szPage-1) 
       || szPage>SQLITE_MAX_PAGE_SIZE 
       || szPage<512 
      ){
        return SQLITE_OK;
      }
      pWal->hdr.bigEndCksum = (u8)(magic&0x00000001);
      pWal->szPage = szPage;
  
      /* Verify that the WAL header checksum is correct */
      walChecksumBytes(pWal->hdr.bigEndCksum==SQLITE_BIGENDIAN, 
          aBuf, WAL_HDRSIZE-2*4, 0, pWal->hdr.aFrameCksum
      );
      if( pWal->hdr.aFrameCksum[0]!=sqlite3Get4byte(&aBuf[24])
       || pWal->hdr.aFrameCksum[1]!=sqlite3Get4byte(&aBuf[28])
      ){
        return SQLITE_OK;
      }
  
      memcpy(&pWal->hdr.aSalt, &aBuf[16], 8);
      *pnCkpt = sqlite3Get4byte(&aBuf[12]);
  
      /* Verify that the version number on the WAL format is one that
      ** are able to understand */
      version = sqlite3Get4byte(&aBuf[4]);
      if( version!=WAL_VERSION1 && version!=WAL_VERSION2 ){
        return SQLITE_CANTOPEN_BKPT;
      }
      pWal->hdr.iVersion = version;
  
      /* Malloc a buffer to read frames into. */
      szFrame = szPage + WAL_FRAME_HDRSIZE;
      aFrame = (u8 *)sqlite3_malloc64(szFrame);
      if( !aFrame ){
        return SQLITE_NOMEM_BKPT;
      }
      aData = &aFrame[WAL_FRAME_HDRSIZE];
  
      /* Read all frames from the log file. */
      iFrame = 0;
      for(iOffset=WAL_HDRSIZE; (iOffset+szFrame)<=nSize; iOffset+=szFrame){
        u32 pgno;                   /* Database page number for frame */
        u32 nTruncate;              /* dbsize field from frame header */
  
        /* Read and decode the next log frame. */
        iFrame++;
        rc = sqlite3OsRead(pWalFd, aFrame, szFrame, iOffset);
        if( rc!=SQLITE_OK ) break;
        isValid = walDecodeFrame(pWal, &pgno, &nTruncate, aData, aFrame);
        if( !isValid ) break;
        rc = walIndexAppend(pWal, iWal, iFrame, pgno);
        if( rc!=SQLITE_OK ) break;
  
        /* If nTruncate is non-zero, this is a commit record. */
        if( nTruncate ){
          pWal->hdr.mxFrame = iFrame;
          pWal->hdr.nPage = nTruncate;
          pWal->hdr.szPage = (u16)((szPage&0xff00) | (szPage>>16));
          testcase( szPage<=32768 );
          testcase( szPage>=65536 );
          aFrameCksum[0] = pWal->hdr.aFrameCksum[0];
          aFrameCksum[1] = pWal->hdr.aFrameCksum[1];
        }
      }
  
      sqlite3_free(aFrame);
    }else if( pbZero ){
      *pbZero = 1;
    }
  }

  pWal->hdr.aFrameCksum[0] = aFrameCksum[0];
  pWal->hdr.aFrameCksum[1] = aFrameCksum[1];

  return rc;
}

static int walOpenWal2(Wal *pWal){
  int rc = SQLITE_OK;
  if( !isOpen(pWal->apWalFd[1]) ){
    int f = (SQLITE_OPEN_READWRITE|SQLITE_OPEN_CREATE|SQLITE_OPEN_WAL);
    rc = sqlite3OsOpen(pWal->pVfs, pWal->zWalName2, pWal->apWalFd[1], f, &f);
  }
  return rc;
}

static int walTruncateWal2(Wal *pWal){
  int bIs;
  int rc;
  assert( !isOpen(pWal->apWalFd[1]) );
  rc = sqlite3OsAccess(pWal->pVfs, pWal->zWalName2, SQLITE_ACCESS_EXISTS, &bIs);
  if( rc==SQLITE_OK && bIs ){
    rc = walOpenWal2(pWal);
    if( rc==SQLITE_OK ){
      rc = sqlite3OsTruncate(pWal->apWalFd[1], 0);
      sqlite3OsClose(pWal->apWalFd[1]);
    }
  }
  return rc;
}

/*
** Recover the wal-index by reading the write-ahead log file. 
**
** This routine first tries to establish an exclusive lock on the
** wal-index to prevent other threads/processes from doing anything
** with the WAL or wal-index while recovery is running.  The
** WAL_RECOVER_LOCK is also held so that other threads will know
** that this thread is running recovery.  If unable to establish
** the necessary locks, this routine returns SQLITE_BUSY.
*/
static int walIndexRecover(Wal *pWal){
  int rc;                         /* Return Code */
  i64 nSize;                      /* Size of log file */
  u32 aFrameCksum[2] = {0, 0};
  int iLock;                      /* Lock offset to lock for checkpoint */
  u32 nCkpt1 = 0xFFFFFFFF;
  u32 nCkpt2 = 0xFFFFFFFF;
  int bZero = 0;
  WalIndexHdr hdr;

  /* Obtain an exclusive lock on all byte in the locking range not already
  ** locked by the caller. The caller is guaranteed to have locked the
  ** WAL_WRITE_LOCK byte, and may have also locked the WAL_CKPT_LOCK byte.
  ** If successful, the same bytes that are locked here are concurrent before
  ** this function returns.
  */

  }
  if( rc ){
    return rc;
  }

  WALTRACE(("WAL%p: recovery begin...\n", pWal));

  memset(&pWal->hdr, 0, sizeof(WalIndexHdr));

  /* Recover the *-wal file. If a valid version-1 header is recovered
  ** from it, do not open the *-wal2 file. Even if it exists.
  rc = sqlite3OsFileSize(pWal->pWalFd, &nSize);
  if( rc!=SQLITE_OK ){
    goto recovery_error;
  **
  ** Otherwise, if the *-wal2 file exists or if the "wal2" flag was 
  ** specified when sqlite3WalOpen() was called, open and recover
  }

  ** the *-wal2 file. Except, if the *-wal file was zero bytes in size,
  if( nSize>WAL_HDRSIZE ){
    u8 aBuf[WAL_HDRSIZE];         /* Buffer to load WAL header into */
    u8 *aFrame = 0;               /* Malloc'd buffer to load entire frame */
    int szFrame;                  /* Number of bytes in buffer aFrame[] */
  ** truncate the *-wal2 to zero bytes in size.
    u8 *aData;                    /* Pointer to data part of aFrame buffer */
    int iFrame;                   /* Index of last frame read */
    i64 iOffset;                  /* Next offset to read from log file */
    int szPage;                   /* Page size according to the log */
    u32 magic;                    /* Magic value read from WAL header */
    u32 version;                  /* Magic value read from WAL header */
  **
  ** After this block has run, if the *-wal2 file is open the system
  ** starts up in VERSION2 mode. In this case pWal->hdr contains the 
  ** wal-index header considering only *-wal2. Stack variable hdr
  ** contains the wal-index header considering only *-wal. The hash 
    int isValid;                  /* True if this frame is valid */

  ** tables are populated for both.  
    /* Read in the WAL header. */
    rc = sqlite3OsRead(pWal->pWalFd, aBuf, WAL_HDRSIZE, 0);
    if( rc!=SQLITE_OK ){
      goto recovery_error;
    }

  **
    /* If the database page size is not a power of two, or is greater than
    ** SQLITE_MAX_PAGE_SIZE, conclude that the WAL file contains no valid 
    ** data. Similarly, if the 'magic' value is invalid, ignore the whole
    ** WAL file.
    */
  ** Or, if the *-wal2 file is not open, start up in VERSION1 mode.
  ** pWal->hdr is already populated.
  */
    magic = sqlite3Get4byte(&aBuf[0]);
    szPage = sqlite3Get4byte(&aBuf[8]);
    if( (magic&0xFFFFFFFE)!=WAL_MAGIC 
     || szPage&(szPage-1) 
     || szPage>SQLITE_MAX_PAGE_SIZE 
     || szPage<512 
    ){
      goto finished;
    }
    pWal->hdr.bigEndCksum = (u8)(magic&0x00000001);
    pWal->szPage = szPage;
  rc = walIndexRecoverOne(pWal, 0, &nCkpt1, &bZero);
    pWal->nCkpt = sqlite3Get4byte(&aBuf[12]);
    memcpy(&pWal->hdr.aSalt, &aBuf[16], 8);

  assert( pWal->hdr.iVersion==0 
      || pWal->hdr.iVersion==WAL_VERSION1 
    /* Verify that the WAL header checksum is correct */
    walChecksumBytes(pWal->hdr.bigEndCksum==SQLITE_BIGENDIAN, 
        aBuf, WAL_HDRSIZE-2*4, 0, pWal->hdr.aFrameCksum
    );
      || pWal->hdr.iVersion==WAL_VERSION2 
  );
    if( pWal->hdr.aFrameCksum[0]!=sqlite3Get4byte(&aBuf[24])
     || pWal->hdr.aFrameCksum[1]!=sqlite3Get4byte(&aBuf[28])
    ){
      goto finished;
    }

  if( rc==SQLITE_OK && bZero ){
    /* Verify that the version number on the WAL format is one that
    ** are able to understand */
    version = sqlite3Get4byte(&aBuf[4]);
    if( version!=WAL_MAX_VERSION ){
      rc = SQLITE_CANTOPEN_BKPT;
    rc = walTruncateWal2(pWal);
      goto finished;
    }

  }
  if( rc==SQLITE_OK && pWal->hdr.iVersion!=WAL_VERSION1 ){
    /* Malloc a buffer to read frames into. */
    szFrame = szPage + WAL_FRAME_HDRSIZE;
    aFrame = (u8 *)sqlite3_malloc64(szFrame);
    if( !aFrame ){
    int bOpen = 1;
    sqlite3_vfs *pVfs = pWal->pVfs;
    if( pWal->hdr.iVersion==0 && pWal->bWal2==0 ){
      rc = SQLITE_NOMEM_BKPT;
      goto recovery_error;
    }
    aData = &aFrame[WAL_FRAME_HDRSIZE];

      rc = sqlite3OsAccess(pVfs, pWal->zWalName2, SQLITE_ACCESS_EXISTS, &bOpen);
    /* Read all frames from the log file. */
    iFrame = 0;
    for(iOffset=WAL_HDRSIZE; (iOffset+szFrame)<=nSize; iOffset+=szFrame){
      u32 pgno;                   /* Database page number for frame */
      u32 nTruncate;              /* dbsize field from frame header */

    }
      /* Read and decode the next log frame. */
      iFrame++;
      rc = sqlite3OsRead(pWal->pWalFd, aFrame, szFrame, iOffset);
      if( rc!=SQLITE_OK ) break;
    if( rc==SQLITE_OK && bOpen ){
      isValid = walDecodeFrame(pWal, &pgno, &nTruncate, aData, aFrame);
      if( !isValid ) break;
      rc = walIndexAppend(pWal, iFrame, pgno);
      if( rc!=SQLITE_OK ) break;
      rc = walOpenWal2(pWal);
      if( rc==SQLITE_OK ){

      /* If nTruncate is non-zero, this is a commit record. */
      if( nTruncate ){
        pWal->hdr.mxFrame = iFrame;
        pWal->hdr.nPage = nTruncate;
        hdr = pWal->hdr;
        rc = walIndexRecoverOne(pWal, 1, &nCkpt2, 0);
        pWal->hdr.szPage = (u16)((szPage&0xff00) | (szPage>>16));
        testcase( szPage<=32768 );
        testcase( szPage>=65536 );
        aFrameCksum[0] = pWal->hdr.aFrameCksum[0];
        aFrameCksum[1] = pWal->hdr.aFrameCksum[1];
      }
    }

  }
    sqlite3_free(aFrame);
  }


finished:
  if( rc==SQLITE_OK ){
    volatile WalCkptInfo *pInfo;
    int i;
    pWal->hdr.aFrameCksum[0] = aFrameCksum[0];
    pWal->hdr.aFrameCksum[1] = aFrameCksum[1];

    if( isOpen(pWal->apWalFd[1]) ){
      /* The case where *-wal2 may follow *-wal */
      if( nCkpt2<=0x0F && nCkpt2==nCkpt1+1 ){
        if( sqlite3Get4byte((u8*)(&pWal->hdr.aSalt[0]))==hdr.aFrameCksum[0]
         && sqlite3Get4byte((u8*)(&pWal->hdr.aSalt[1]))==hdr.aFrameCksum[1]
        ){
          walidxSetFile(&pWal->hdr, 1);
          walidxSetMxFrame(&pWal->hdr, 1, pWal->hdr.mxFrame);
          walidxSetMxFrame(&pWal->hdr, 0, hdr.mxFrame);
        }else{
          pWal->hdr = hdr;
        }
      }else

      /* When *-wal may follow *-wal2 */
      if( (nCkpt2==0x0F && nCkpt1==0) || (nCkpt2<0x0F && nCkpt2==nCkpt1-1) ){
        if( sqlite3Get4byte((u8*)(&hdr.aSalt[0]))==pWal->hdr.aFrameCksum[0]
         && sqlite3Get4byte((u8*)(&hdr.aSalt[1]))==pWal->hdr.aFrameCksum[1]
        ){
          SWAP(WalIndexHdr, pWal->hdr, hdr);
          walidxSetMxFrame(&pWal->hdr, 1, hdr.mxFrame);
        }else{
          walidxSetFile(&pWal->hdr, 1);
          walidxSetMxFrame(&pWal->hdr, 1, pWal->hdr.mxFrame);
          walidxSetMxFrame(&pWal->hdr, 0, 0);
        }
      }else

      /* Fallback */
      if( nCkpt1<=nCkpt2 ){
        pWal->hdr = hdr;
      }else{
        walidxSetFile(&pWal->hdr, 1);
        walidxSetMxFrame(&pWal->hdr, 1, pWal->hdr.mxFrame);
        walidxSetMxFrame(&pWal->hdr, 0, 0);
      }
      pWal->hdr.iVersion = WAL_VERSION2;
    }else{
      pWal->hdr.iVersion = WAL_VERSION1;
    }

    walIndexWriteHdr(pWal);

    /* Reset the checkpoint-header. This is safe because this thread is 
    ** currently holding locks that exclude all other readers, writers and
    ** checkpointers.
    ** checkpointers.  */
    */
    pInfo = walCkptInfo(pWal);
    pInfo->nBackfill = 0;
    pInfo->nBackfillAttempted = pWal->hdr.mxFrame;
    pInfo->aReadMark[0] = 0;
    for(i=1; i<WAL_NREADER; i++) pInfo->aReadMark[i] = READMARK_NOT_USED;
    if( pWal->hdr.mxFrame ) pInfo->aReadMark[1] = pWal->hdr.mxFrame;
    memset((void*)pInfo, 0, sizeof(WalCkptInfo));
    if( 0==isWalMode2(pWal) ){
      int i;
      pInfo->nBackfillAttempted = pWal->hdr.mxFrame;
      pInfo->aReadMark[0] = 0;
      for(i=1; i<WAL_NREADER; i++) pInfo->aReadMark[i] = READMARK_NOT_USED;
      if( pWal->hdr.mxFrame ) pInfo->aReadMark[1] = pWal->hdr.mxFrame;
    }

    /* If more than one frame was recovered from the log file, report an
    ** event via sqlite3_log(). This is to help with identifying performance
    ** problems caused by applications routinely shutting down without
    ** checkpointing the log file.
    ** checkpointing the log file.  */
    */
    if( pWal->hdr.nPage ){
      if( isWalMode2(pWal) ){
      sqlite3_log(SQLITE_NOTICE_RECOVER_WAL,
          "recovered %d frames from WAL file %s",
          pWal->hdr.mxFrame, pWal->zWalName
      );
        sqlite3_log(SQLITE_NOTICE_RECOVER_WAL,
            "recovered (%d,%d) frames from WAL files %s[2] (wal2 mode)",
            walidxGetMxFrame(&pWal->hdr, 0), walidxGetMxFrame(&pWal->hdr, 1), 
            pWal->zWalName
        );
      }else{
        sqlite3_log(SQLITE_NOTICE_RECOVER_WAL,
            "recovered %d frames from WAL file %s",
            pWal->hdr.mxFrame, pWal->zWalName
        );
      }
    }
  }

recovery_error:
  WALTRACE(("WAL%p: recovery %s\n", pWal, rc ? "failed" : "ok"));
  walUnlockExclusive(pWal, iLock, WAL_READ_LOCK(0)-iLock);
  walUnlockExclusive(pWal, WAL_READ_LOCK(1), WAL_NREADER-1);
  return rc;
}

/*
** Close an open wal-index.
** Close an open wal-index and wal files.
*/
static void walIndexClose(Wal *pWal, int isDelete){
  if( pWal->exclusiveMode==WAL_HEAPMEMORY_MODE || pWal->bShmUnreliable ){
    int i;
    for(i=0; i<pWal->nWiData; i++){
      sqlite3_free((void *)pWal->apWiData[i]);
      pWal->apWiData[i] = 0;
    }
  }
  if( pWal->exclusiveMode!=WAL_HEAPMEMORY_MODE ){
    sqlite3OsShmUnmap(pWal->pDbFd, isDelete);
  }
  sqlite3OsClose(pWal->apWalFd[0]);
  sqlite3OsClose(pWal->apWalFd[1]);
}

/* 
** Open a connection to the WAL file zWalName. The database file must 
** already be opened on connection pDbFd. The buffer that zWalName points
** to must remain valid for the lifetime of the returned Wal* handle.
**

*/
int sqlite3WalOpen(
  sqlite3_vfs *pVfs,              /* vfs module to open wal and wal-index */
  sqlite3_file *pDbFd,            /* The open database file */
  const char *zWalName,           /* Name of the WAL file */
  int bNoShm,                     /* True to run in heap-memory mode */
  i64 mxWalSize,                  /* Truncate WAL to this size on reset */
  int bWal2,                      /* True to open in wal2 mode */
  Wal **ppWal                     /* OUT: Allocated Wal handle */
){
  int rc;                         /* Return Code */
  Wal *pRet;                      /* Object to allocate and return */
  int flags;                      /* Flags passed to OsOpen() */
  int nWalName;                   /* Length of zWalName in bytes */
  int nByte;                      /* Bytes of space to allocate */

  assert( zWalName && zWalName[0] );
  assert( pDbFd );

  /* In the amalgamation, the os_unix.c and os_win.c source files come before
  ** this source file.  Verify that the #defines of the locking byte offsets
  ** in os_unix.c and os_win.c agree with the WALINDEX_LOCK_OFFSET value.
  ** For that matter, if the lock offset ever changes from its initial design
  ** value of 120, we need to know that so there is an assert() to check it.
  */
  assert( 120==WALINDEX_LOCK_OFFSET );
  assert( 136==WALINDEX_HDR_SIZE );
#ifdef WIN_SHM_BASE
  assert( WIN_SHM_BASE==WALINDEX_LOCK_OFFSET );
#endif
#ifdef UNIX_SHM_BASE
  assert( UNIX_SHM_BASE==WALINDEX_LOCK_OFFSET );
#endif

  nWalName = sqlite3Strlen30(zWalName);
  nByte = sizeof(Wal) + pVfs->szOsFile*2 + nWalName+2;

  /* Allocate an instance of struct Wal to return. */
  *ppWal = 0;
  pRet = (Wal*)sqlite3MallocZero(sizeof(Wal) + pVfs->szOsFile);
  pRet = (Wal*)sqlite3MallocZero(nByte);
  if( !pRet ){
    return SQLITE_NOMEM_BKPT;
  }

  pRet->pVfs = pVfs;
  pRet->pWalFd = (sqlite3_file *)&pRet[1];
  pRet->apWalFd[0] = (sqlite3_file*)((char*)pRet+sizeof(Wal));
  pRet->apWalFd[1] = (sqlite3_file*)((char*)pRet+sizeof(Wal)+pVfs->szOsFile);
  pRet->pDbFd = pDbFd;
  pRet->readLock = -1;
  pRet->readLock = WAL_LOCK_NONE;
  pRet->mxWalSize = mxWalSize;
  pRet->zWalName = zWalName;
  pRet->syncHeader = 1;
  pRet->padToSectorBoundary = 1;
  pRet->exclusiveMode = (bNoShm ? WAL_HEAPMEMORY_MODE: WAL_NORMAL_MODE);
  pRet->bWal2 = bWal2;

  pRet->zWalName2 = (char*)pRet + sizeof(Wal) + 2*pVfs->szOsFile;
  memcpy(pRet->zWalName2, zWalName, nWalName);
  pRet->zWalName2[nWalName] = '2';
  pRet->zWalName2[nWalName+1] = '\0';

  /* Open file handle on the write-ahead log file. */
  /* Open a file handle on the first write-ahead log file. */
  flags = (SQLITE_OPEN_READWRITE|SQLITE_OPEN_CREATE|SQLITE_OPEN_WAL);
  rc = sqlite3OsOpen(pVfs, zWalName, pRet->pWalFd, flags, &flags);
  rc = sqlite3OsOpen(pVfs, zWalName, pRet->apWalFd[0], flags, &flags);
  if( rc==SQLITE_OK && flags&SQLITE_OPEN_READONLY ){
    pRet->readOnly = WAL_RDONLY;
  }

  if( rc!=SQLITE_OK ){
    walIndexClose(pRet, 0);
    sqlite3OsClose(pRet->pWalFd);
    sqlite3_free(pRet);
  }else{
    int iDC = sqlite3OsDeviceCharacteristics(pDbFd);
    if( iDC & SQLITE_IOCAP_SEQUENTIAL ){ pRet->syncHeader = 0; }
    if( iDC & SQLITE_IOCAP_POWERSAFE_OVERWRITE ){
      pRet->padToSectorBoundary = 0;
    }

*/
static void walIteratorFree(WalIterator *p){
  sqlite3_free(p);
}

/*
** Construct a WalInterator object that can be used to loop over all 
** pages in the WAL following frame nBackfill in ascending order. Frames
** pages in wal file iWal following frame nBackfill in ascending order. Frames
** nBackfill or earlier may be included - excluding them is an optimization
** only. The caller must hold the checkpoint lock.
**
** On success, make *pp point to the newly allocated WalInterator object
** return SQLITE_OK. Otherwise, return an error code. If this routine
** returns an error, the value of *pp is undefined.
** On success, make *pp point to the newly allocated WalIterator object
** and return SQLITE_OK. Otherwise, return an error code. If this routine
** returns an error, the final value of *pp is undefined.
**
** The calling routine should invoke walIteratorFree() to destroy the
** WalIterator object when it has finished with it.
*/
static int walIteratorInit(Wal *pWal, u32 nBackfill, WalIterator **pp){
static int walIteratorInit(
  Wal *pWal, 
  int iWal, 
  u32 nBackfill, 
  WalIterator **pp
){
  WalIterator *p;                 /* Return value */
  int nSegment;                   /* Number of segments to merge */
  u32 iLast;                      /* Last frame in log */
  int nByte;                      /* Number of bytes to allocate */
  int i;                          /* Iterator variable */
  int iLastSeg;                   /* Last hash table to iterate though */
  ht_slot *aTmp;                  /* Temp space used by merge-sort */
  int rc = SQLITE_OK;             /* Return Code */
  int iMode = isWalMode2(pWal) ? 2 : 1;

  assert( isWalMode2(pWal) || iWal==0 );
  assert( 0==isWalMode2(pWal) || nBackfill==0 );

  /* This routine only runs while holding the checkpoint lock. And
  ** it only runs if there is actually content in the log (mxFrame>0).
  */
  iLast = walidxGetMxFrame(&pWal->hdr, iWal);
  assert( pWal->ckptLock && pWal->hdr.mxFrame>0 );
  iLast = pWal->hdr.mxFrame;

  assert( pWal->ckptLock && iLast>0 );

  if( iMode==2 ){
    iLastSeg = walFramePage2(iWal, iLast);
  }else{
    iLastSeg = walFramePage(iLast);
  }
  nSegment = 1 + (iLastSeg/iMode);

  /* Allocate space for the WalIterator object. */
  nSegment = walFramePage(iLast) + 1;
  nByte = sizeof(WalIterator) 
        + (nSegment-1)*sizeof(struct WalSegment)
        + iLast*sizeof(ht_slot);
  p = (WalIterator *)sqlite3_malloc64(nByte);
  if( !p ){
    return SQLITE_NOMEM_BKPT;
  }
  memset(p, 0, nByte);
  p->nSegment = nSegment;

  /* Allocate temporary space used by the merge-sort routine. This block
  ** of memory will be freed before this function returns.
  */
  aTmp = (ht_slot *)sqlite3_malloc64(
      sizeof(ht_slot) * (iLast>HASHTABLE_NPAGE?HASHTABLE_NPAGE:iLast)
  );
  if( !aTmp ){
    rc = SQLITE_NOMEM_BKPT;
  }

  for(i=walFramePage(nBackfill+1); rc==SQLITE_OK && i<nSegment; i++){
  i = iMode==2 ? iWal : walFramePage(nBackfill+1);
  for(; rc==SQLITE_OK && i<=iLastSeg; i+=iMode){
    WalHashLoc sLoc;

    rc = walHashGet(pWal, i, &sLoc);
    if( rc==SQLITE_OK ){
      int j;                      /* Counter variable */
      int nEntry;                 /* Number of entries in this segment */
      ht_slot *aIndex;            /* Sorted index for this segment */
      u32 iZero;

      if( iMode==2 ){
        walExternalDecode(sLoc.iZero+1, &iZero);
        iZero--;
        assert( iZero==0 || i>=2 );
      }else{
        iZero = sLoc.iZero;
      }

      sLoc.aPgno++;
      if( (i+1)==nSegment ){
        nEntry = (int)(iLast - sLoc.iZero);
      if( i==iLastSeg ){
        nEntry = (int)(iLast - iZero);
      }else{
        nEntry = (int)((u32*)sLoc.aHash - (u32*)sLoc.aPgno);
      }
      aIndex = &((ht_slot *)&p->aSegment[p->nSegment])[sLoc.iZero];
      sLoc.iZero++;
      aIndex = &((ht_slot *)&p->aSegment[p->nSegment])[iZero];
      iZero++;
  
      for(j=0; j<nEntry; j++){
        aIndex[j] = (ht_slot)j;
      }
      walMergesort((u32 *)sLoc.aPgno, aTmp, aIndex, &nEntry);
      p->aSegment[i].iZero = sLoc.iZero;
      p->aSegment[i].nEntry = nEntry;
      p->aSegment[i].aIndex = aIndex;
      p->aSegment[i].aPgno = (u32 *)sLoc.aPgno;
      walMergesort((u32*)sLoc.aPgno, aTmp, aIndex, &nEntry);
      p->aSegment[i/iMode].iZero = iZero;
      p->aSegment[i/iMode].nEntry = nEntry;
      p->aSegment[i/iMode].aIndex = aIndex;
      p->aSegment[i/iMode].aPgno = (u32*)sLoc.aPgno;
    }
  }
  sqlite3_free(aTmp);

  if( rc!=SQLITE_OK ){
    walIteratorFree(p);
    p = 0;

** new wal-index header. It should be passed a pseudo-random value (i.e. 
** one obtained from sqlite3_randomness()).
*/
static void walRestartHdr(Wal *pWal, u32 salt1){
  volatile WalCkptInfo *pInfo = walCkptInfo(pWal);
  int i;                          /* Loop counter */
  u32 *aSalt = pWal->hdr.aSalt;   /* Big-endian salt values */
  assert( isWalMode2(pWal)==0 );
  pWal->nCkpt++;
  pWal->hdr.mxFrame = 0;
  sqlite3Put4byte((u8*)&aSalt[0], 1 + sqlite3Get4byte((u8*)&aSalt[0]));
  memcpy(&pWal->hdr.aSalt[1], &salt1, 4);
  walIndexWriteHdr(pWal);
  pInfo->nBackfill = 0;
  pInfo->nBackfillAttempted = 0;
  pInfo->aReadMark[1] = 0;
  for(i=2; i<WAL_NREADER; i++) pInfo->aReadMark[i] = READMARK_NOT_USED;
  assert( pInfo->aReadMark[0]==0 );
}

/*
** This function is used in wal2 mode.
**
** This function is called when writer pWal is just about to start 
** writing out frames. Parameter iApp is the current wal file. The "other" wal
** file (wal file !iApp) has been fully checkpointed. This function returns
** SQLITE_OK if there are no readers preventing the writer from switching to
** the other wal file. Or SQLITE_BUSY if there are.
*/
static int wal2RestartOk(Wal *pWal, int iApp){
  /* The other wal file (wal file !iApp) can be overwritten if there
  ** are no readers reading from it - no "full" or "partial" locks.
  ** Technically speaking it is not possible for any reader to hold
  ** a "part" lock, as this would have prevented the file from being
  ** checkpointed. But checking anyway doesn't hurt. The following
  ** is equivalent to:
  **
  **   if( iApp==0 ) eLock = WAL_LOCK_PART1_FULL2;
  **   if( iApp==1 ) eLock = WAL_LOCK_PART1;
  */
  int eLock = 1 + (iApp==0);

  assert( WAL_LOCK_PART1==1 );
  assert( WAL_LOCK_PART1_FULL2==2 );
  assert( WAL_LOCK_PART2_FULL1==3 );
  assert( WAL_LOCK_PART2==4 );

  assert( iApp!=0 || eLock==WAL_LOCK_PART1_FULL2 );
  assert( iApp!=1 || eLock==WAL_LOCK_PART1 );

  return walLockExclusive(pWal, WAL_READ_LOCK(eLock), 3);
}
static void wal2RestartFinished(Wal *pWal, int iApp){
  walUnlockExclusive(pWal, WAL_READ_LOCK(1 + (iApp==0)), 3);
}

/*
** This function is used in wal2 mode.
**
** This function is called when a checkpointer wishes to checkpoint wal
** file iCkpt. It takes the required lock and, if successful, returns
** SQLITE_OK. Otherwise, an SQLite error code (e.g. SQLITE_BUSY). If this
** function returns SQLITE_OK, it is the responsibility of the caller
** to invoke wal2CheckpointFinished() to release the lock.
*/
static int wal2CheckpointOk(Wal *pWal, int iCkpt){
  int eLock = 1 + (iCkpt*2);

  assert( WAL_LOCK_PART1==1 );
  assert( WAL_LOCK_PART1_FULL2==2 );
  assert( WAL_LOCK_PART2_FULL1==3 );
  assert( WAL_LOCK_PART2==4 );

  assert( iCkpt!=0 || eLock==WAL_LOCK_PART1 );
  assert( iCkpt!=1 || eLock==WAL_LOCK_PART2_FULL1 );

  return walLockExclusive(pWal, WAL_READ_LOCK(eLock), 2);
}
static void wal2CheckpointFinished(Wal *pWal, int iCkpt){
  walUnlockExclusive(pWal, WAL_READ_LOCK(1 + (iCkpt*2)), 2);
}

/*
** Copy as much content as we can from the WAL back into the database file
** in response to an sqlite3_wal_checkpoint() request or the equivalent.
**
** The amount of information copies from WAL to database might be limited
** by active readers.  This routine will never overwrite a database page

  WalIterator *pIter = 0;         /* Wal iterator context */
  u32 iDbpage = 0;                /* Next database page to write */
  u32 iFrame = 0;                 /* Wal frame containing data for iDbpage */
  u32 mxSafeFrame;                /* Max frame that can be backfilled */
  u32 mxPage;                     /* Max database page to write */
  int i;                          /* Loop counter */
  volatile WalCkptInfo *pInfo;    /* The checkpoint status information */
  int bWal2 = isWalMode2(pWal);   /* True for wal2 connections */
  int iCkpt = bWal2 ? !walidxGetFile(&pWal->hdr) : 0;

  mxSafeFrame = walidxGetMxFrame(&pWal->hdr, iCkpt);
  szPage = walPagesize(pWal);
  testcase( szPage<=32768 );
  testcase( szPage>=65536 );
  pInfo = walCkptInfo(pWal);
  if( (bWal2==1 && pInfo->nBackfill==0 && mxSafeFrame) 
  if( pInfo->nBackfill<pWal->hdr.mxFrame ){
   || (bWal2==0 && pInfo->nBackfill<mxSafeFrame)
  ){
    sqlite3_file *pWalFd = pWal->apWalFd[iCkpt];
    mxPage = pWal->hdr.nPage;

    /* If this is a wal2 system, check for a reader holding a lock 
    ** preventing this checkpoint operation. If one is found, return
    ** early.  */
    if( bWal2 ){
      rc = wal2CheckpointOk(pWal, iCkpt);
      if( rc!=SQLITE_OK ) return rc;
    }

    /* EVIDENCE-OF: R-62920-47450 The busy-handler callback is never invoked
    ** in the SQLITE_CHECKPOINT_PASSIVE mode. */
    assert( eMode!=SQLITE_CHECKPOINT_PASSIVE || xBusy==0 );

    /* Compute in mxSafeFrame the index of the last frame of the WAL that is
    ** safe to write into the database.  Frames beyond mxSafeFrame might
    ** overwrite database pages that are in use by active readers and thus
    ** cannot be backfilled from the WAL.
    /* If this is a wal system (not wal2), compute in mxSafeFrame the index 
    ** of the last frame of the WAL that is safe to write into the database.
    ** Frames beyond mxSafeFrame might overwrite database pages that are in 
    ** use by active readers and thus cannot be backfilled from the WAL.
    */
    if( bWal2==0 ){
    mxSafeFrame = pWal->hdr.mxFrame;
    mxPage = pWal->hdr.nPage;
    for(i=1; i<WAL_NREADER; i++){
      /* Thread-sanitizer reports that the following is an unsafe read,
      ** as some other thread may be in the process of updating the value
      ** of the aReadMark[] slot. The assumption here is that if that is
      ** happening, the other client may only be increasing the value,
      ** not decreasing it. So assuming either that either the "old" or
      ** "new" version of the value is read, and not some arbitrary value
      ** that would never be written by a real client, things are still 
      ** safe.  */
      u32 y = pInfo->aReadMark[i];
      if( mxSafeFrame>y ){
        assert( y<=pWal->hdr.mxFrame );
        rc = walBusyLock(pWal, xBusy, pBusyArg, WAL_READ_LOCK(i), 1);
        if( rc==SQLITE_OK ){
          pInfo->aReadMark[i] = (i==1 ? mxSafeFrame : READMARK_NOT_USED);
          walUnlockExclusive(pWal, WAL_READ_LOCK(i), 1);
        }else if( rc==SQLITE_BUSY ){
          mxSafeFrame = y;
          xBusy = 0;
        }else{
          goto walcheckpoint_out;
      mxSafeFrame = pWal->hdr.mxFrame;
      mxPage = pWal->hdr.nPage;
      for(i=1; i<WAL_NREADER; i++){
        /* Thread-sanitizer reports that the following is an unsafe read,
        ** as some other thread may be in the process of updating the value
        ** of the aReadMark[] slot. The assumption here is that if that is
        ** happening, the other client may only be increasing the value,
        ** not decreasing it. So assuming either that either the "old" or
        ** "new" version of the value is read, and not some arbitrary value
        ** that would never be written by a real client, things are still 
        ** safe.  */
        u32 y = pInfo->aReadMark[i];
        if( mxSafeFrame>y ){
          assert( y<=pWal->hdr.mxFrame );
          rc = walBusyLock(pWal, xBusy, pBusyArg, WAL_READ_LOCK(i), 1);
          if( rc==SQLITE_OK ){
            pInfo->aReadMark[i] = (i==1 ? mxSafeFrame : READMARK_NOT_USED);
            walUnlockExclusive(pWal, WAL_READ_LOCK(i), 1);
          }else if( rc==SQLITE_BUSY ){
            mxSafeFrame = y;
            xBusy = 0;
          }else{
            goto walcheckpoint_out;
          }
        }
      }
    }

    /* Allocate the iterator */
    if( pInfo->nBackfill<mxSafeFrame ){
      rc = walIteratorInit(pWal, pInfo->nBackfill, &pIter);
    if( bWal2 || pInfo->nBackfill<mxSafeFrame ){
      assert( bWal2==0 || pInfo->nBackfill==0 );
      rc = walIteratorInit(pWal, iCkpt, pInfo->nBackfill, &pIter);
      assert( rc==SQLITE_OK || pIter==0 );
    }

    if( pIter
     && (rc = walBusyLock(pWal, xBusy, pBusyArg, WAL_READ_LOCK(0),1))==SQLITE_OK
    ){
    if( pIter && (bWal2 
     || (rc = walBusyLock(pWal, xBusy, pBusyArg,WAL_READ_LOCK(0),1))==SQLITE_OK
    )){
      u32 nBackfill = pInfo->nBackfill;

      assert( bWal2==0 || nBackfill==0 );
      pInfo->nBackfillAttempted = mxSafeFrame;

      /* Sync the WAL to disk */
      rc = sqlite3OsSync(pWal->pWalFd, CKPT_SYNC_FLAGS(sync_flags));
      /* Sync the wal file being checkpointed to disk */
      rc = sqlite3OsSync(pWalFd, CKPT_SYNC_FLAGS(sync_flags));

      /* If the database may grow as a result of this checkpoint, hint
      ** about the eventual size of the db file to the VFS layer.
      ** about the eventual size of the db file to the VFS layer.  */
      */
      if( rc==SQLITE_OK ){
        i64 nReq = ((i64)mxPage * szPage);
        i64 nSize;                    /* Current size of database file */
        rc = sqlite3OsFileSize(pWal->pDbFd, &nSize);
        if( rc==SQLITE_OK && nSize<nReq ){
          sqlite3OsFileControlHint(pWal->pDbFd, SQLITE_FCNTL_SIZE_HINT, &nReq);
        }
      }


      /* Iterate through the contents of the WAL, copying data to the db file */
      while( rc==SQLITE_OK && 0==walIteratorNext(pIter, &iDbpage, &iFrame) ){
        i64 iOffset;

        assert( walFramePgno(pWal, iFrame)==iDbpage );
        assert( bWal2==1 || walFramePgno(pWal, iFrame)==iDbpage );
        assert( bWal2==0 || walFramePgno2(pWal, iCkpt, iFrame)==iDbpage );

        if( db->u1.isInterrupted ){
          rc = db->mallocFailed ? SQLITE_NOMEM_BKPT : SQLITE_INTERRUPT;
          break;
        }
        if( iFrame<=nBackfill || iFrame>mxSafeFrame || iDbpage>mxPage ){
          assert( bWal2==0 || iDbpage>mxPage );
          continue;
        }
        iOffset = walFrameOffset(iFrame, szPage) + WAL_FRAME_HDRSIZE;
        WALTRACE(("WAL%p: checkpoint frame %d of wal %d to db page %d\n",
              pWal, (int)iFrame, iCkpt, (int)iDbpage
        ));
        /* testcase( IS_BIG_INT(iOffset) ); // requires a 4GiB WAL file */
        rc = sqlite3OsRead(pWal->pWalFd, zBuf, szPage, iOffset);
        rc = sqlite3OsRead(pWalFd, zBuf, szPage, iOffset);
        if( rc!=SQLITE_OK ) break;
        iOffset = (iDbpage-1)*(i64)szPage;
        testcase( IS_BIG_INT(iOffset) );
        rc = sqlite3OsWrite(pWal->pDbFd, zBuf, szPage, iOffset);
        if( rc!=SQLITE_OK ) break;
      }

      /* If work was actually accomplished... */
      if( rc==SQLITE_OK ){
        if( mxSafeFrame==walIndexHdr(pWal)->mxFrame ){
      /* If work was actually accomplished, truncate the db file, sync the wal
      ** file and set WalCkptInfo.nBackfill to indicate so. */
      if( rc==SQLITE_OK && (bWal2 || mxSafeFrame==walIndexHdr(pWal)->mxFrame) ){
        if( !bWal2 ){
          i64 szDb = pWal->hdr.nPage*(i64)szPage;
          testcase( IS_BIG_INT(szDb) );
          rc = sqlite3OsTruncate(pWal->pDbFd, szDb);
        }
          if( rc==SQLITE_OK ){
            rc = sqlite3OsSync(pWal->pDbFd, CKPT_SYNC_FLAGS(sync_flags));
          }
        }
        if( rc==SQLITE_OK ){
          pInfo->nBackfill = mxSafeFrame;
        }
      }
        if( rc==SQLITE_OK ){
          rc = sqlite3OsSync(pWal->pDbFd, CKPT_SYNC_FLAGS(sync_flags));
        }
      }
      if( rc==SQLITE_OK ){
        pInfo->nBackfill = bWal2 ? 1 : mxSafeFrame;
      }


      /* Release the reader lock held while backfilling */
      if( bWal2==0 ){
      walUnlockExclusive(pWal, WAL_READ_LOCK(0), 1);
        walUnlockExclusive(pWal, WAL_READ_LOCK(0), 1);
      }
    }

    if( rc==SQLITE_BUSY ){
      /* Reset the return code so as not to report a checkpoint failure
      ** just because there are active readers.  */
      rc = SQLITE_OK;
    }
    if( bWal2 ) wal2CheckpointFinished(pWal, iCkpt);
  }

  /* If this is an SQLITE_CHECKPOINT_RESTART or TRUNCATE operation, and the
  ** entire wal file has been copied into the database file, then block 
  ** until all readers have finished using the wal file. This ensures that 
  ** the next process to write to the database restarts the wal file.
  */
  if( rc==SQLITE_OK && eMode!=SQLITE_CHECKPOINT_PASSIVE ){
  if( bWal2==0 && rc==SQLITE_OK && eMode!=SQLITE_CHECKPOINT_PASSIVE ){
    assert( pWal->writeLock );
    if( pInfo->nBackfill<pWal->hdr.mxFrame ){
      rc = SQLITE_BUSY;
    }else if( eMode>=SQLITE_CHECKPOINT_RESTART ){
      u32 salt1;
      sqlite3_randomness(4, &salt1);
      assert( pInfo->nBackfill==pWal->hdr.mxFrame );

          ** writer clients should see that the entire log file has been
          ** checkpointed and behave accordingly. This seems unsafe though,
          ** as it would leave the system in a state where the contents of
          ** the wal-index header do not match the contents of the 
          ** file-system. To avoid this, update the wal-index header to
          ** indicate that the log file contains zero valid frames.  */
          walRestartHdr(pWal, salt1);
          rc = sqlite3OsTruncate(pWal->pWalFd, 0);
          rc = sqlite3OsTruncate(pWal->apWalFd[0], 0);
        }
        walUnlockExclusive(pWal, WAL_READ_LOCK(1), WAL_NREADER-1);
      }
    }
  }

 walcheckpoint_out:
  walIteratorFree(pIter);
  return rc;
}

/*
** If the WAL file is currently larger than nMax bytes in size, truncate
** it to exactly nMax bytes. If an error occurs while doing so, ignore it.
*/
static void walLimitSize(Wal *pWal, i64 nMax){
  if( isWalMode2(pWal)==0 ){
  i64 sz;
  int rx;
  sqlite3BeginBenignMalloc();
  rx = sqlite3OsFileSize(pWal->pWalFd, &sz);
  if( rx==SQLITE_OK && (sz > nMax ) ){
    rx = sqlite3OsTruncate(pWal->pWalFd, nMax);
  }
  sqlite3EndBenignMalloc();
  if( rx ){
    sqlite3_log(rx, "cannot limit WAL size: %s", pWal->zWalName);
    i64 sz;
    int rx;
    sqlite3BeginBenignMalloc();
    rx = sqlite3OsFileSize(pWal->apWalFd[0], &sz);
    if( rx==SQLITE_OK && (sz > nMax ) ){
      rx = sqlite3OsTruncate(pWal->apWalFd[0], nMax);
    }
    sqlite3EndBenignMalloc();
    if( rx ){
      sqlite3_log(rx, "cannot limit WAL size: %s", pWal->zWalName);
    }
  }
}

/*
** Close a connection to a log file.
*/
int sqlite3WalClose(

    ** the wal and wal-index files.
    **
    ** The EXCLUSIVE lock is not released before returning.
    */
    if( zBuf!=0
     && SQLITE_OK==(rc = sqlite3OsLock(pWal->pDbFd, SQLITE_LOCK_EXCLUSIVE))
    ){
      int i;
      if( pWal->exclusiveMode==WAL_NORMAL_MODE ){
        pWal->exclusiveMode = WAL_EXCLUSIVE_MODE;
      }
      for(i=0; rc==SQLITE_OK && i<2; i++){
      rc = sqlite3WalCheckpoint(pWal, db, 
          SQLITE_CHECKPOINT_PASSIVE, 0, 0, sync_flags, nBuf, zBuf, 0, 0
      );
      if( rc==SQLITE_OK ){
        int bPersist = -1;
        sqlite3OsFileControlHint(
            pWal->pDbFd, SQLITE_FCNTL_PERSIST_WAL, &bPersist
        );
        if( bPersist!=1 ){
          /* Try to delete the WAL file if the checkpoint completed and
          ** fsyned (rc==SQLITE_OK) and if we are not in persistent-wal
          ** mode (!bPersist) */
          isDelete = 1;
        }else if( pWal->mxWalSize>=0 ){
          /* Try to truncate the WAL file to zero bytes if the checkpoint
          ** completed and fsynced (rc==SQLITE_OK) and we are in persistent
          ** WAL mode (bPersist) and if the PRAGMA journal_size_limit is a
          ** non-negative value (pWal->mxWalSize>=0).  Note that we truncate
          ** to zero bytes as truncating to the journal_size_limit might
          ** leave a corrupt WAL file on disk. */
          walLimitSize(pWal, 0);
        }
      }
    }
        rc = sqlite3WalCheckpoint(pWal, db, 
            SQLITE_CHECKPOINT_PASSIVE, 0, 0, sync_flags, nBuf, zBuf, 0, 0
        );
        if( rc==SQLITE_OK ){
          int bPersist = -1;
          sqlite3OsFileControlHint(
              pWal->pDbFd, SQLITE_FCNTL_PERSIST_WAL, &bPersist
          );
          if( bPersist!=1 ){
            /* Try to delete the WAL file if the checkpoint completed and
            ** fsyned (rc==SQLITE_OK) and if we are not in persistent-wal
            ** mode (!bPersist) */
            isDelete = 1;
          }else if( pWal->mxWalSize>=0 ){
            /* Try to truncate the WAL file to zero bytes if the checkpoint
            ** completed and fsynced (rc==SQLITE_OK) and we are in persistent
            ** WAL mode (bPersist) and if the PRAGMA journal_size_limit is a
            ** non-negative value (pWal->mxWalSize>=0).  Note that we truncate
            ** to zero bytes as truncating to the journal_size_limit might
            ** leave a corrupt WAL file on disk. */
            walLimitSize(pWal, 0);
          }
        }

        if( isWalMode2(pWal)==0 ) break;

        walCkptInfo(pWal)->nBackfill = 0;
        walidxSetFile(&pWal->hdr, !walidxGetFile(&pWal->hdr));
        pWal->writeLock = 1;
        walIndexWriteHdr(pWal);
        pWal->writeLock = 0;
      }
    }

    walIndexClose(pWal, isDelete);
    sqlite3OsClose(pWal->pWalFd);
    if( isDelete ){
      sqlite3BeginBenignMalloc();
      sqlite3OsDelete(pWal->pVfs, pWal->zWalName, 0);
      sqlite3OsDelete(pWal->pVfs, pWal->zWalName2, 0);
      sqlite3EndBenignMalloc();
    }
    WALTRACE(("WAL%p: closed\n", pWal));
    sqlite3_free((void *)pWal->apWiData);
    sqlite3_free(pWal);
  }
  return rc;

    }
  }

  /* If the header is read successfully, check the version number to make
  ** sure the wal-index was not constructed with some future format that
  ** this version of SQLite cannot understand.
  */
  if( badHdr==0 && pWal->hdr.iVersion!=WALINDEX_MAX_VERSION ){
  if( badHdr==0 
   && pWal->hdr.iVersion!=WAL_VERSION1 && pWal->hdr.iVersion!=WAL_VERSION2
  ){
    rc = SQLITE_CANTOPEN_BKPT;
  }
  if( pWal->bShmUnreliable ){
    if( rc!=SQLITE_OK ){
      walIndexClose(pWal, 0);
      pWal->bShmUnreliable = 0;
      assert( pWal->nWiData>0 && pWal->apWiData[0]==0 );

  **
  ** Once sqlite3OsShmMap() has been called for an sqlite3_file and has
  ** returned any SQLITE_READONLY value, it must return only SQLITE_READONLY
  ** or SQLITE_READONLY_CANTINIT or some error for all subsequent invocations,
  ** even if some external agent does a "chmod" to make the shared-memory
  ** writable by us, until sqlite3OsShmUnmap() has been called.
  ** This is a requirement on the VFS implementation.
   */
  */
  rc = sqlite3OsShmMap(pWal->pDbFd, 0, WALINDEX_PGSZ, 0, &pDummy);
  assert( rc!=SQLITE_OK ); /* SQLITE_OK not possible for read-only connection */
  if( rc!=SQLITE_READONLY_CANTINIT ){
    rc = (rc==SQLITE_READONLY ? WAL_RETRY : rc);
    goto begin_unreliable_shm_out;
  }

  /* We reach this point only if the real shared-memory is still unreliable.
  ** Assume the in-memory WAL-index substitute is correct and load it
  ** into pWal->hdr.
  */
  memcpy(&pWal->hdr, (void*)walIndexHdr(pWal), sizeof(WalIndexHdr));

  /* Make sure some writer hasn't come in and changed the WAL file out
  ** from under us, then disconnected, while we were not looking.
  */
  rc = sqlite3OsFileSize(pWal->pWalFd, &szWal);
  rc = sqlite3OsFileSize(pWal->apWalFd[0], &szWal);
  if( rc!=SQLITE_OK ){
    goto begin_unreliable_shm_out;
  }
  if( szWal<WAL_HDRSIZE ){
    /* If the wal file is too small to contain a wal-header and the
    ** wal-index header has mxFrame==0, then it must be safe to proceed
    ** reading the database file only. However, the page cache cannot
    ** be trusted, as a read/write connection may have connected, written
    ** the db, run a checkpoint, truncated the wal file and disconnected
    ** since this client's last read transaction.  */
    *pChanged = 1;
    rc = (pWal->hdr.mxFrame==0 ? SQLITE_OK : WAL_RETRY);
    goto begin_unreliable_shm_out;
  }

  /* Check the salt keys at the start of the wal file still match. */
  rc = sqlite3OsRead(pWal->pWalFd, aBuf, WAL_HDRSIZE, 0);
  rc = sqlite3OsRead(pWal->apWalFd[0], aBuf, WAL_HDRSIZE, 0);
  if( rc!=SQLITE_OK ){
    goto begin_unreliable_shm_out;
  }
  if( memcmp(&pWal->hdr.aSalt, &aBuf[16], 8) ){
    /* Some writer has wrapped the WAL file while we were not looking.
    ** Return WAL_RETRY which will cause the in-memory WAL-index to be
    ** rebuilt. */

      iOffset+szFrame<=szWal; 
      iOffset+=szFrame
  ){
    u32 pgno;                   /* Database page number for frame */
    u32 nTruncate;              /* dbsize field from frame header */

    /* Read and decode the next log frame. */
    rc = sqlite3OsRead(pWal->pWalFd, aFrame, szFrame, iOffset);
    rc = sqlite3OsRead(pWal->apWalFd[0], aFrame, szFrame, iOffset);
    if( rc!=SQLITE_OK ) break;
    if( !walDecodeFrame(pWal, &pgno, &nTruncate, aData, aFrame) ) break;

    /* If nTruncate is non-zero, then a complete transaction has been
    ** appended to this wal file. Set rc to WAL_RETRY and break out of
    ** the loop.  */
    if( nTruncate ){

** checkpoint process do as much work as possible.  This routine might
** update values of the aReadMark[] array in the header, but if it does
** so it takes care to hold an exclusive lock on the corresponding
** WAL_READ_LOCK() while changing values.
*/
static int walTryBeginRead(Wal *pWal, int *pChanged, int useWal, int cnt){
  volatile WalCkptInfo *pInfo;    /* Checkpoint information in wal-index */
  u32 mxReadMark;                 /* Largest aReadMark[] value */
  int mxI;                        /* Index of largest aReadMark[] value */
  int i;                          /* Loop counter */
  int rc = SQLITE_OK;             /* Return code  */
  u32 mxFrame;                    /* Wal frame to lock to */

  assert( pWal->readLock<0 );     /* Not currently locked */
  assert( pWal->readLock==WAL_LOCK_NONE );     /* Not currently locked */

  /* useWal may only be set for read/write connections */
  assert( (pWal->readOnly & WAL_SHM_RDONLY)==0 || useWal==0 );

  /* Take steps to avoid spinning forever if there is a protocol error.
  **
  ** Circumstances that cause a RETRY should only last for the briefest

      return walBeginShmUnreliable(pWal, pChanged);
    }
  }

  assert( pWal->nWiData>0 );
  assert( pWal->apWiData[0]!=0 );
  pInfo = walCkptInfo(pWal);
  if( isWalMode2(pWal) ){
    /* This connection needs a "part" lock on the current wal file and, 
    ** unless pInfo->nBackfill is set to indicate that it has already been
    ** checkpointed, a "full" lock on the other wal file.  */
    int iWal = walidxGetFile(&pWal->hdr);
    int nBackfill = pInfo->nBackfill || walidxGetMxFrame(&pWal->hdr, !iWal)==0;
    int eLock = 1 + (iWal*2) + (nBackfill==iWal);

    assert( nBackfill==0 || nBackfill==1 );
    assert( iWal==0 || iWal==1 );
    assert( iWal!=0 || nBackfill!=1 || eLock==WAL_LOCK_PART1 );
    assert( iWal!=0 || nBackfill!=0 || eLock==WAL_LOCK_PART1_FULL2 );
    assert( iWal!=1 || nBackfill!=1 || eLock==WAL_LOCK_PART2 );
    assert( iWal!=1 || nBackfill!=0 || eLock==WAL_LOCK_PART2_FULL1 );

    rc = walLockShared(pWal, WAL_READ_LOCK(eLock));
    if( rc!=SQLITE_OK ){
      return (rc==SQLITE_BUSY ? WAL_RETRY : rc);
    }
    walShmBarrier(pWal);
    if( memcmp((void *)walIndexHdr(pWal), &pWal->hdr, sizeof(WalIndexHdr)) ){
      walUnlockShared(pWal, WAL_READ_LOCK(eLock));
      return WAL_RETRY;
    }else{
      pWal->readLock = eLock;
    }
    assert( pWal->minFrame==0 && walFramePage(pWal->minFrame)==0 );
  }else{
    u32 mxReadMark;               /* Largest aReadMark[] value */
    int mxI;                      /* Index of largest aReadMark[] value */
    int i;                        /* Loop counter */
    u32 mxFrame;                  /* Wal frame to lock to */
  if( !useWal && pInfo->nBackfill==pWal->hdr.mxFrame
#ifdef SQLITE_ENABLE_SNAPSHOT
   && (pWal->pSnapshot==0 || pWal->hdr.mxFrame==0)
#endif
  ){
    /* The WAL has been completely backfilled (or it is empty).
    ** and can be safely ignored.
    */
    rc = walLockShared(pWal, WAL_READ_LOCK(0));
    walShmBarrier(pWal);
    if( rc==SQLITE_OK ){
      if( memcmp((void *)walIndexHdr(pWal), &pWal->hdr, sizeof(WalIndexHdr)) ){
        /* It is not safe to allow the reader to continue here if frames
        ** may have been appended to the log before READ_LOCK(0) was obtained.
        ** When holding READ_LOCK(0), the reader ignores the entire log file,
        ** which implies that the database file contains a trustworthy
        ** snapshot. Since holding READ_LOCK(0) prevents a checkpoint from
        ** happening, this is usually correct.
        **
        ** However, if frames have been appended to the log (or if the log 
        ** is wrapped and written for that matter) before the READ_LOCK(0)
        ** is obtained, that is not necessarily true. A checkpointer may
        ** have started to backfill the appended frames but crashed before
        ** it finished. Leaving a corrupt image in the database file.
        */
        walUnlockShared(pWal, WAL_READ_LOCK(0));
        return WAL_RETRY;
      }
      pWal->readLock = 0;
      return SQLITE_OK;
    }else if( rc!=SQLITE_BUSY ){
      return rc;
    }
  }

  /* If we get this far, it means that the reader will want to use
  ** the WAL to get at content from recent commits.  The job now is
  ** to select one of the aReadMark[] entries that is closest to
  ** but not exceeding pWal->hdr.mxFrame and lock that entry.
  */
  mxReadMark = 0;
  mxI = 0;
  mxFrame = pWal->hdr.mxFrame;
#ifdef SQLITE_ENABLE_SNAPSHOT
  if( pWal->pSnapshot && pWal->pSnapshot->mxFrame<mxFrame ){
    mxFrame = pWal->pSnapshot->mxFrame;
  }
#endif
  for(i=1; i<WAL_NREADER; i++){
    u32 thisMark = AtomicLoad(pInfo->aReadMark+i);
    if( mxReadMark<=thisMark && thisMark<=mxFrame ){
      assert( thisMark!=READMARK_NOT_USED );
      mxReadMark = thisMark;
      mxI = i;
    }
  }
  if( (pWal->readOnly & WAL_SHM_RDONLY)==0
   && (mxReadMark<mxFrame || mxI==0)
  ){
    for(i=1; i<WAL_NREADER; i++){
      rc = walLockExclusive(pWal, WAL_READ_LOCK(i), 1);
      if( rc==SQLITE_OK ){
        mxReadMark = AtomicStore(pInfo->aReadMark+i,mxFrame);
        mxI = i;
        walUnlockExclusive(pWal, WAL_READ_LOCK(i), 1);
        break;
      }else if( rc!=SQLITE_BUSY ){
        return rc;
      }
    }
  }
  if( mxI==0 ){
    assert( rc==SQLITE_BUSY || (pWal->readOnly & WAL_SHM_RDONLY)!=0 );
    return rc==SQLITE_BUSY ? WAL_RETRY : SQLITE_READONLY_CANTINIT;
  }

  rc = walLockShared(pWal, WAL_READ_LOCK(mxI));
  if( rc ){
    return rc==SQLITE_BUSY ? WAL_RETRY : rc;
  }
  /* Now that the read-lock has been obtained, check that neither the
  ** value in the aReadMark[] array or the contents of the wal-index
  ** header have changed.
  **
  ** It is necessary to check that the wal-index header did not change
  ** between the time it was read and when the shared-lock was obtained
  ** on WAL_READ_LOCK(mxI) was obtained to account for the possibility
  ** that the log file may have been wrapped by a writer, or that frames
  ** that occur later in the log than pWal->hdr.mxFrame may have been
  ** copied into the database by a checkpointer. If either of these things
  ** happened, then reading the database with the current value of
  ** pWal->hdr.mxFrame risks reading a corrupted snapshot. So, retry
  ** instead.
  **
  ** Before checking that the live wal-index header has not changed
  ** since it was read, set Wal.minFrame to the first frame in the wal
  ** file that has not yet been checkpointed. This client will not need
  ** to read any frames earlier than minFrame from the wal file - they
  ** can be safely read directly from the database file.
  **
  ** Because a ShmBarrier() call is made between taking the copy of 
  ** nBackfill and checking that the wal-header in shared-memory still
  ** matches the one cached in pWal->hdr, it is guaranteed that the 
  ** checkpointer that set nBackfill was not working with a wal-index
  ** header newer than that cached in pWal->hdr. If it were, that could
  ** cause a problem. The checkpointer could omit to checkpoint
  ** a version of page X that lies before pWal->minFrame (call that version
  ** A) on the basis that there is a newer version (version B) of the same
  ** page later in the wal file. But if version B happens to like past
  ** frame pWal->hdr.mxFrame - then the client would incorrectly assume
  ** that it can read version A from the database file. However, since
  ** we can guarantee that the checkpointer that set nBackfill could not
  ** see any pages past pWal->hdr.mxFrame, this problem does not come up.
  */
  pWal->minFrame = AtomicLoad(&pInfo->nBackfill)+1;
  walShmBarrier(pWal);
  if( AtomicLoad(pInfo->aReadMark+mxI)!=mxReadMark
   || memcmp((void *)walIndexHdr(pWal), &pWal->hdr, sizeof(WalIndexHdr))
  ){
    walUnlockShared(pWal, WAL_READ_LOCK(mxI));
    return WAL_RETRY;
  }else{
    assert( mxReadMark<=pWal->hdr.mxFrame );
    pWal->readLock = (i16)mxI;
    if( !useWal && pInfo->nBackfill==pWal->hdr.mxFrame
  #ifdef SQLITE_ENABLE_SNAPSHOT
     && (pWal->pSnapshot==0 || pWal->hdr.mxFrame==0)
  #endif
    ){
      /* The WAL has been completely backfilled (or it is empty).
      ** and can be safely ignored.
      */
      rc = walLockShared(pWal, WAL_READ_LOCK(0));
      walShmBarrier(pWal);
      if( rc==SQLITE_OK ){
        if( memcmp((void *)walIndexHdr(pWal), &pWal->hdr,sizeof(WalIndexHdr)) ){
          /* It is not safe to allow the reader to continue here if frames
          ** may have been appended to the log before READ_LOCK(0) was obtained.
          ** When holding READ_LOCK(0), the reader ignores the entire log file,
          ** which implies that the database file contains a trustworthy
          ** snapshot. Since holding READ_LOCK(0) prevents a checkpoint from
          ** happening, this is usually correct.
          **
          ** However, if frames have been appended to the log (or if the log 
          ** is wrapped and written for that matter) before the READ_LOCK(0)
          ** is obtained, that is not necessarily true. A checkpointer may
          ** have started to backfill the appended frames but crashed before
          ** it finished. Leaving a corrupt image in the database file.
          */
          walUnlockShared(pWal, WAL_READ_LOCK(0));
          return WAL_RETRY;
        }
        pWal->readLock = 0;
        return SQLITE_OK;
      }else if( rc!=SQLITE_BUSY ){
        return rc;
      }
    }
  
    /* If we get this far, it means that the reader will want to use
    ** the WAL to get at content from recent commits.  The job now is
    ** to select one of the aReadMark[] entries that is closest to
    ** but not exceeding pWal->hdr.mxFrame and lock that entry.
    */
    mxReadMark = 0;
    mxI = 0;
    mxFrame = pWal->hdr.mxFrame;
  #ifdef SQLITE_ENABLE_SNAPSHOT
    if( pWal->pSnapshot && pWal->pSnapshot->mxFrame<mxFrame ){
      mxFrame = pWal->pSnapshot->mxFrame;
    }
  #endif
    for(i=1; i<WAL_NREADER; i++){
      u32 thisMark = AtomicLoad(pInfo->aReadMark+i);
      if( mxReadMark<=thisMark && thisMark<=mxFrame ){
        assert( thisMark!=READMARK_NOT_USED );
        mxReadMark = thisMark;
        mxI = i;
      }
    }
    if( (pWal->readOnly & WAL_SHM_RDONLY)==0
     && (mxReadMark<mxFrame || mxI==0)
    ){
      for(i=1; i<WAL_NREADER; i++){
        rc = walLockExclusive(pWal, WAL_READ_LOCK(i), 1);
        if( rc==SQLITE_OK ){
          mxReadMark = AtomicStore(pInfo->aReadMark+i,mxFrame);
          mxI = i;
          walUnlockExclusive(pWal, WAL_READ_LOCK(i), 1);
          break;
        }else if( rc!=SQLITE_BUSY ){
          return rc;
        }
      }
    }
    if( mxI==0 ){
      assert( rc==SQLITE_BUSY || (pWal->readOnly & WAL_SHM_RDONLY)!=0 );
      return rc==SQLITE_BUSY ? WAL_RETRY : SQLITE_READONLY_CANTINIT;
    }
  
    rc = walLockShared(pWal, WAL_READ_LOCK(mxI));
    if( rc ){
      return rc==SQLITE_BUSY ? WAL_RETRY : rc;
    }
    /* Now that the read-lock has been obtained, check that neither the
    ** value in the aReadMark[] array or the contents of the wal-index
    ** header have changed.
    **
    ** It is necessary to check that the wal-index header did not change
    ** between the time it was read and when the shared-lock was obtained
    ** on WAL_READ_LOCK(mxI) was obtained to account for the possibility
    ** that the log file may have been wrapped by a writer, or that frames
    ** that occur later in the log than pWal->hdr.mxFrame may have been
    ** copied into the database by a checkpointer. If either of these things
    ** happened, then reading the database with the current value of
    ** pWal->hdr.mxFrame risks reading a corrupted snapshot. So, retry
    ** instead.
    **
    ** Before checking that the live wal-index header has not changed
    ** since it was read, set Wal.minFrame to the first frame in the wal
    ** file that has not yet been checkpointed. This client will not need
    ** to read any frames earlier than minFrame from the wal file - they
    ** can be safely read directly from the database file.
    **
    ** Because a ShmBarrier() call is made between taking the copy of 
    ** nBackfill and checking that the wal-header in shared-memory still
    ** matches the one cached in pWal->hdr, it is guaranteed that the 
    ** checkpointer that set nBackfill was not working with a wal-index
    ** header newer than that cached in pWal->hdr. If it were, that could
    ** cause a problem. The checkpointer could omit to checkpoint
    ** a version of page X that lies before pWal->minFrame (call that version
    ** A) on the basis that there is a newer version (version B) of the same
    ** page later in the wal file. But if version B happens to like past
    ** frame pWal->hdr.mxFrame - then the client would incorrectly assume
    ** that it can read version A from the database file. However, since
    ** we can guarantee that the checkpointer that set nBackfill could not
    ** see any pages past pWal->hdr.mxFrame, this problem does not come up.
    */
    pWal->minFrame = AtomicLoad(&pInfo->nBackfill)+1;
    walShmBarrier(pWal);
    if( AtomicLoad(pInfo->aReadMark+mxI)!=mxReadMark
     || memcmp((void *)walIndexHdr(pWal), &pWal->hdr, sizeof(WalIndexHdr))
    ){
      walUnlockShared(pWal, WAL_READ_LOCK(mxI));
      return WAL_RETRY;
    }else{
      assert( mxReadMark<=pWal->hdr.mxFrame );
      pWal->readLock = (i16)mxI;
    }
  }
  return rc;
}

#ifdef SQLITE_ENABLE_SNAPSHOT
/*
** Attempt to reduce the value of the WalCkptInfo.nBackfillAttempted 

**
** SQLITE_OK is returned if successful, or an SQLite error code if an
** error occurs. It is not an error if nBackfillAttempted cannot be
** decreased at all.
*/
int sqlite3WalSnapshotRecover(Wal *pWal){
  int rc;

  /* Snapshots may not be used with wal2 mode databases. */
  if( isWalMode2(pWal) ) return SQLITE_ERROR;

  assert( pWal->readLock>=0 );
  rc = walLockExclusive(pWal, WAL_CKPT_LOCK, 1);
  if( rc==SQLITE_OK ){
    volatile WalCkptInfo *pInfo = walCkptInfo(pWal);
    int szPage = (int)pWal->szPage;
    i64 szDb;                   /* Size of db file in bytes */

          rc = walHashGet(pWal, walFramePage(i), &sLoc);
          if( rc!=SQLITE_OK ) break;
          pgno = sLoc.aPgno[i-sLoc.iZero];
          iDbOff = (i64)(pgno-1) * szPage;

          if( iDbOff+szPage<=szDb ){
            iWalOff = walFrameOffset(i, szPage) + WAL_FRAME_HDRSIZE;
            rc = sqlite3OsRead(pWal->pWalFd, pBuf1, szPage, iWalOff);
            rc = sqlite3OsRead(pWal->apWalFd[0], pBuf1, szPage, iWalOff);

            if( rc==SQLITE_OK ){
              rc = sqlite3OsRead(pWal->pDbFd, pBuf2, szPage, iDbOff);
            }

            if( rc!=SQLITE_OK || 0==memcmp(pBuf1, pBuf2, szPage) ){
              break;

int sqlite3WalBeginReadTransaction(Wal *pWal, int *pChanged){
  int rc;                         /* Return code */
  int cnt = 0;                    /* Number of TryBeginRead attempts */

#ifdef SQLITE_ENABLE_SNAPSHOT
  int bChanged = 0;
  WalIndexHdr *pSnapshot = pWal->pSnapshot;
  if( pSnapshot && isWalMode2(pWal) ) return SQLITE_ERROR;
  if( pSnapshot && memcmp(pSnapshot, &pWal->hdr, sizeof(WalIndexHdr))!=0 ){
    bChanged = 1;
  }
#endif

  do{
    rc = walTryBeginRead(pWal, pChanged, 0, ++cnt);
  }while( rc==WAL_RETRY );
  testcase( (rc&0xff)==SQLITE_BUSY );
  testcase( (rc&0xff)==SQLITE_IOERR );
  testcase( rc==SQLITE_PROTOCOL );
  testcase( rc==SQLITE_OK );
  
  if( rc==SQLITE_OK && pWal->hdr.iVersion==WAL_VERSION2 ){
    rc = walOpenWal2(pWal);
  }

  pWal->nPriorFrame = pWal->hdr.mxFrame;
#ifdef SQLITE_ENABLE_SNAPSHOT
  if( rc==SQLITE_OK ){
    if( pSnapshot && memcmp(pSnapshot, &pWal->hdr, sizeof(WalIndexHdr))!=0 ){
      /* At this point the client has a lock on an aReadMark[] slot holding
      ** a value equal to or smaller than pSnapshot->mxFrame, but pWal->hdr

/*
** Finish with a read transaction.  All this does is release the
** read-lock.
*/
void sqlite3WalEndReadTransaction(Wal *pWal){
  sqlite3WalEndWriteTransaction(pWal);
  if( pWal->readLock>=0 ){
  if( pWal->readLock!=WAL_LOCK_NONE ){
    walUnlockShared(pWal, WAL_READ_LOCK(pWal->readLock));
    pWal->readLock = -1;
    pWal->readLock = WAL_LOCK_NONE;
  }
}

/* Search hash table iHash for an entry matching page number
** pgno. Each call to this function searches a single hash table
** (each hash table indexes up to HASHTABLE_NPAGE frames).
**
** This code might run concurrently to the code in walIndexAppend()
** that adds entries to the wal-index (and possibly to this hash 
** table). This means the value just read from the hash 
** slot (aHash[iKey]) may have been added before or after the 
** current read transaction was opened. Values added after the
** read transaction was opened may have been written incorrectly -
** i.e. these slots may contain garbage data. However, we assume
** that any slots written before the current read transaction was
** opened remain unmodified.
**
** For the reasons above, the if(...) condition featured in the inner
** loop of the following block is more stringent that would be required 
** if we had exclusive access to the hash-table:
**
**   (aPgno[iFrame]==pgno): 
**     This condition filters out normal hash-table collisions.
**
**   (iFrame<=iLast): 
**     This condition filters out entries that were added to the hash
**     table after the current read-transaction had started.
*/
static int walSearchHash(
  Wal *pWal, 
  u32 iLast,
  int iHash, 
  Pgno pgno, 
  u32 *piRead
){
  WalHashLoc sLoc;                /* Hash table location */
  int iKey;                       /* Hash slot index */
  int nCollide;                   /* Number of hash collisions remaining */
  int rc;                         /* Error code */

  rc = walHashGet(pWal, iHash, &sLoc);
  if( rc!=SQLITE_OK ){
    return rc;
  }
  nCollide = HASHTABLE_NSLOT;
  for(iKey=walHash(pgno); sLoc.aHash[iKey]; iKey=walNextHash(iKey)){
    u32 iFrame = sLoc.aHash[iKey] + sLoc.iZero;
    if( iFrame<=iLast 
     && iFrame>=pWal->minFrame 
     && sLoc.aPgno[sLoc.aHash[iKey]]==pgno 
    ){
      assert( iFrame>*piRead || CORRUPT_DB );
      *piRead = iFrame;
    }
    if( (nCollide--)==0 ){
      return SQLITE_CORRUPT_BKPT;
    }
  }

  return SQLITE_OK;
}

static int walSearchWal(
  Wal *pWal, 
  int iWal, 
  Pgno pgno, 
  u32 *piRead
){
  int rc = SQLITE_OK;
  int bWal2 = isWalMode2(pWal);
  u32 iLast = walidxGetMxFrame(&pWal->hdr, iWal);
  if( iLast ){
    int iHash;
    int iMinHash = walFramePage(pWal->minFrame);
    u32 iExternal = bWal2 ? walExternalEncode(iWal, iLast) : iLast;
    assert( bWal2==0 || pWal->minFrame==0 );
    for(iHash=walFramePage(iExternal); 
        iHash>=iMinHash && *piRead==0; 
        iHash-=(1+bWal2)
    ){
      rc = walSearchHash(pWal, iExternal, iHash, pgno, piRead);
      if( rc!=SQLITE_OK ) break;
    }
  }
  return rc;
}

/*
** Search the wal file for page pgno. If found, set *piRead to the frame that
** contains the page. Otherwise, if pgno is not in the wal file, set *piRead
** to zero.
**
** Return SQLITE_OK if successful, or an error code if an error occurs. If an
** error does occur, the final value of *piRead is undefined.
*/
int sqlite3WalFindFrame(
  Wal *pWal,                      /* WAL handle */
  Pgno pgno,                      /* Database page number to read data for */
  u32 *piRead                     /* OUT: Frame number (or zero) */
){
  int bWal2 = isWalMode2(pWal);
  int iApp = walidxGetFile(&pWal->hdr);
  int rc = SQLITE_OK;
  u32 iRead = 0;                  /* If !=0, WAL frame to return data from */
  u32 iLast = pWal->hdr.mxFrame;  /* Last page in WAL for this reader */
  int iHash;                      /* Used to loop through N hash tables */
  int iMinHash;

  /* This routine is only be called from within a read transaction. */
  assert( pWal->readLock>=0 || pWal->lockError );
  /* This routine is only be called from within a read transaction. Or,
  ** sometimes, as part of a rollback that occurs after an error reaquiring
  ** a read-lock in walRestartLog().  */
  assert( pWal->readLock!=WAL_LOCK_NONE || pWal->writeLock );

  /* If the "last page" field of the wal-index header snapshot is 0, then
  /* If this is a regular wal system, then iApp must be set to 0 (there is
  ** no data will be read from the wal under any circumstances. Return early
  ** in this case as an optimization.  Likewise, if pWal->readLock==0, 
  ** only one wal file, after all). Or, if this is a wal2 system and the
  ** write-lock is not held, the client must have a partial-wal lock on wal 
  ** file iApp. This is not always true if the write-lock is held and this
  ** function is being called after WalLockForCommit() as part of committing
  ** a CONCURRENT transaction.  */
#ifdef SQLITE_DEBUG
  if( bWal2 ){
    if( pWal->writeLock==0 ){
      int l = pWal->readLock;
  ** then the WAL is ignored by the reader so return early, as if the 
  ** WAL were empty.
  */
  if( iLast==0 || (pWal->readLock==0 && pWal->bShmUnreliable==0) ){
      assert( iApp==1 || l==WAL_LOCK_PART1 || l==WAL_LOCK_PART1_FULL2 );
      assert( iApp==0 || l==WAL_LOCK_PART2 || l==WAL_LOCK_PART2_FULL1 );
    }
  }else{
    assert( iApp==0 );
  }
#endif

  /* Return early if read-lock 0 is held. */
  if( (pWal->readLock==0 && pWal->bShmUnreliable==0) ){
    assert( !bWal2 );
    *piRead = 0;
    return SQLITE_OK;
  }

  /* Search the wal file that the client holds a partial lock on first. */
  /* Each iteration of the following for() loop searches one
  ** hash table (each hash table indexes up to HASHTABLE_NPAGE frames).
  **
  rc = walSearchWal(pWal, iApp, pgno, &iRead);

  /* If the requested page was not found, no error has occured, and 
  ** the client holds a full-wal lock on the other wal file, search it
  ** too.  */
  ** This code might run concurrently to the code in walIndexAppend()
  ** that adds entries to the wal-index (and possibly to this hash 
  ** table). This means the value just read from the hash 
  ** slot (aHash[iKey]) may have been added before or after the 
  ** current read transaction was opened. Values added after the
  ** read transaction was opened may have been written incorrectly -
  ** i.e. these slots may contain garbage data. However, we assume
  ** that any slots written before the current read transaction was
  ** opened remain unmodified.
  if( rc==SQLITE_OK && bWal2 && iRead==0 && (
        pWal->readLock==WAL_LOCK_PART1_FULL2 
     || pWal->readLock==WAL_LOCK_PART2_FULL1
#ifndef SQLITE_OMIT_CONCURRENT
     || (pWal->readLock==WAL_LOCK_PART1 && iApp==1)
     || (pWal->readLock==WAL_LOCK_PART2 && iApp==0)
#endif
  **
  ** For the reasons above, the if(...) condition featured in the inner
  ** loop of the following block is more stringent that would be required 
  )){
    rc = walSearchWal(pWal, !iApp, pgno, &iRead);
  ** if we had exclusive access to the hash-table:
  **
  **   (aPgno[iFrame]==pgno): 
  **     This condition filters out normal hash-table collisions.
  **
  **   (iFrame<=iLast): 
  **     This condition filters out entries that were added to the hash
  **     table after the current read-transaction had started.
  */
  }
  iMinHash = walFramePage(pWal->minFrame);
  for(iHash=walFramePage(iLast); iHash>=iMinHash; iHash--){
    WalHashLoc sLoc;              /* Hash table location */
    int iKey;                     /* Hash slot index */
    int nCollide;                 /* Number of hash collisions remaining */
    int rc;                       /* Error code */

    rc = walHashGet(pWal, iHash, &sLoc);
    if( rc!=SQLITE_OK ){
#if defined(SQLITE_TEST) && defined(SQLITE_DEBUG)
  if( iRead ){ 
      return rc;
    }
    nCollide = HASHTABLE_NSLOT;
    for(iKey=walHash(pgno); sLoc.aHash[iKey]; iKey=walNextHash(iKey)){
      u32 iFrame = sLoc.aHash[iKey] + sLoc.iZero;
    u32 iFrame;
      if( iFrame<=iLast && iFrame>=pWal->minFrame
       && sLoc.aPgno[sLoc.aHash[iKey]]==pgno ){
        assert( iFrame>iRead || CORRUPT_DB );
        iRead = iFrame;
      }
    int iWal = walExternalDecode(iRead, &iFrame);
    WALTRACE(("WAL%p: page %d @ frame %d wal %d\n",pWal,(int)pgno,iFrame,iWal));
  }else{
      if( (nCollide--)==0 ){
        return SQLITE_CORRUPT_BKPT;
      }
    WALTRACE(("WAL%p: page %d not found\n", pWal, (int)pgno));
  }
    }
    if( iRead ) break;
#endif
  }

#ifdef SQLITE_ENABLE_EXPENSIVE_ASSERT
  /* If expensive assert() statements are available, do a linear search
  ** of the wal-index file content. Make sure the results agree with the
  ** result obtained using the hash indexes above.  */
  ** result obtained using the hash indexes above.  
  {
  **
  ** TODO: This is broken for wal2.
  */
  if( rc==SQLITE_OK ){
    u32 iRead2 = 0;
    u32 iTest;
    assert( pWal->bShmUnreliable || pWal->minFrame>0 );
    for(iTest=iLast; iTest>=pWal->minFrame && iTest>0; iTest--){
      if( walFramePgno(pWal, iTest)==pgno ){
        iRead2 = iTest;
        break;

/*
** Read the contents of frame iRead from the wal file into buffer pOut
** (which is nOut bytes in size). Return SQLITE_OK if successful, or an
** error code otherwise.
*/
int sqlite3WalReadFrame(
  Wal *pWal,                      /* WAL handle */
  u32 iRead,                      /* Frame to read */
  u32 iExternal,                  /* Frame to read */
  int nOut,                       /* Size of buffer pOut in bytes */
  u8 *pOut                        /* Buffer to write page data to */
){
  int sz;
  int iWal = 0;
  u32 iRead;
  i64 iOffset;

  /* Figure out the page size */
  sz = pWal->hdr.szPage;
  sz = (sz&0xfe00) + ((sz&0x0001)<<16);
  testcase( sz<=32768 );
  testcase( sz>=65536 );

  if( isWalMode2(pWal) ){
    /* Figure out which of the two wal files, and the frame within, that 
    ** iExternal refers to.  */
    iWal = walExternalDecode(iExternal, &iRead);
  }else{
    iRead = iExternal;
  }

  WALTRACE(("WAL%p: reading frame %d wal %d\n", pWal, iRead, iWal));
  iOffset = walFrameOffset(iRead, sz) + WAL_FRAME_HDRSIZE;
  /* testcase( IS_BIG_INT(iOffset) ); // requires a 4GiB WAL */
  return sqlite3OsRead(pWal->pWalFd, pOut, (nOut>sz ? sz : nOut), iOffset);
  return sqlite3OsRead(pWal->apWalFd[iWal], pOut, (nOut>sz?sz:nOut), iOffset);
}

/* 
** Return the size of the database in pages (or zero, if unknown).
*/
Pgno sqlite3WalDbsize(Wal *pWal){
  if( pWal && ALWAYS(pWal->readLock>=0) ){
  if( pWal && ALWAYS(pWal->readLock!=WAL_LOCK_NONE) ){
    return pWal->hdr.nPage;
  }
  return 0;
}

/*
** Take the WRITER lock on the WAL file. Return SQLITE_OK if successful,

**
** SQLITE_OK is returned if successful, or an SQLite error code otherwise.
*/
static int walUpgradeReadlock(Wal *pWal){
  int cnt;
  int rc;
  assert( pWal->writeLock && pWal->readLock==0 );
  assert( isWalMode2(pWal)==0 );
  walUnlockShared(pWal, WAL_READ_LOCK(0));
  pWal->readLock = -1;
  cnt = 0;
  do{
    int notUsed;
    rc = walTryBeginRead(pWal, &notUsed, 1, ++cnt);
  }while( rc==WAL_RETRY );

    if( walIndexLoadHdr(pWal, &head) ){
      /* This branch is taken if the wal-index header is corrupted. This 
      ** occurs if some other writer has crashed while committing a 
      ** transaction to this database since the current concurrent transaction
      ** was opened.  */
      rc = SQLITE_BUSY_SNAPSHOT;
    }else if( memcmp(&pWal->hdr, (void*)&head, sizeof(WalIndexHdr))!=0 ){
      int bWal2 = isWalMode2(pWal);
      int iHash;
      int nLoop = 1+(bWal2 && walidxGetFile(&head)!=walidxGetFile(&pWal->hdr));
      int iLastHash = walFramePage(head.mxFrame);
      u32 iFirst = pWal->hdr.mxFrame+1;     /* First wal frame to check */
      if( memcmp(pWal->hdr.aSalt, (u32*)head.aSalt, sizeof(u32)*2) ){
        assert( pWal->readLock==0 );
        iFirst = 1;
      }
      for(iHash=walFramePage(iFirst); iHash<=iLastHash; iHash++){
        WalHashLoc sLoc;
      int iLoop;
      

      assert( nLoop==1 || nLoop==2 );
      for(iLoop=0; rc==SQLITE_OK && iLoop<nLoop; iLoop++){
        u32 iFirst;               /* First (external) wal frame to check */
        u32 iLastHash;            /* Last hash to check this loop */
        u32 mxFrame;              /* Last (external) wal frame to check */

        if( bWal2==0 ){
          assert( iLoop==0 );
          /* Special case for wal mode. If this concurrent transaction was
          ** opened after the entire wal file had been checkpointed, and
          ** another connection has since wrapped the wal file, then we wish to
          ** iterate through every frame in the new wal file - not just those
          ** that follow the current value of pWal->hdr.mxFrame (which will be
          ** set to the size of the old, now overwritten, wal file). This
          ** doesn't come up in wal2 mode, as in wal2 mode the client always
          ** has a PART lock on one of the wal files, preventing it from being
          ** checkpointed or overwritten. */
          iFirst = pWal->hdr.mxFrame+1;
          if( memcmp(pWal->hdr.aSalt, (u32*)head.aSalt, sizeof(u32)*2) ){
            assert( pWal->readLock==0 );
            iFirst = 1;
          }
          mxFrame = head.mxFrame;
        }else{
          int iA = walidxGetFile(&pWal->hdr);
          if( iLoop==0 ){
            iFirst = walExternalEncode(iA, 1+walidxGetMxFrame(&pWal->hdr, iA));
            mxFrame = walExternalEncode(iA, walidxGetMxFrame(&head, iA));
          }else{
            iFirst = walExternalEncode(!iA, 1);
            mxFrame = walExternalEncode(!iA, walidxGetMxFrame(&head, !iA));
          }
        }
        iLastHash = walFramePage(mxFrame);

        for(iHash=walFramePage(iFirst); iHash<=iLastHash; iHash += (1+bWal2)){
          WalHashLoc sLoc;

        rc = walHashGet(pWal, iHash, &sLoc);
        if( rc==SQLITE_OK ){
          u32 i, iMin, iMax;
          assert( head.mxFrame>=sLoc.iZero );
          iMin = (sLoc.iZero >= iFirst) ? 1 : (iFirst - sLoc.iZero);
          iMax = (iHash==0) ? HASHTABLE_NPAGE_ONE : HASHTABLE_NPAGE;
          if( iMax>(head.mxFrame-sLoc.iZero) ) iMax = (head.mxFrame-sLoc.iZero);
          for(i=iMin; rc==SQLITE_OK && i<=iMax; i++){
            PgHdr *pPg;
            if( sLoc.aPgno[i]==1 ){
              /* Check that the schema cookie has not been modified. If
              ** it has not, the commit can proceed. */
              u8 aNew[4];
              u8 *aOld = &((u8*)pPage1->pData)[40];
              int sz;
              i64 iOffset;
              sz = pWal->hdr.szPage;
              sz = (sz&0xfe00) + ((sz&0x0001)<<16);
              iOffset = walFrameOffset(i+sLoc.iZero, sz) + WAL_FRAME_HDRSIZE+40;
              rc = sqlite3OsRead(pWal->pWalFd, aNew, sizeof(aNew), iOffset);
              if( rc==SQLITE_OK && memcmp(aOld, aNew, sizeof(aNew)) ){
                rc = SQLITE_BUSY_SNAPSHOT;
              }
            }else if( sqlite3BitvecTestNotNull(pAllRead, sLoc.aPgno[i]) ){
              *piConflict = sLoc.aPgno[i];
              rc = SQLITE_BUSY_SNAPSHOT;
            }else if( (pPg = sqlite3PagerLookup(pPager, sLoc.aPgno[i])) ){
              /* Page aPgno[i], which is present in the pager cache, has been
              ** modified since the current CONCURRENT transaction was started.
              ** However it was not read by the current transaction, so is not
              ** a conflict. There are two possibilities: (a) the page was
              ** allocated at the of the file by the current transaction or 
              ** (b) was present in the cache at the start of the transaction.
              **
              ** For case (a), do nothing. This page will be moved within the
              ** database file by the commit code to avoid the conflict. The
              ** call to PagerUnref() is to release the reference grabbed by
              ** the sqlite3PagerLookup() above.  
              **
              ** In case (b), drop the page from the cache - otherwise
              ** following the snapshot upgrade the cache would be inconsistent
              ** with the database as stored on disk. */
              if( sqlite3PagerIswriteable(pPg) ){
                sqlite3PagerUnref(pPg);
              }else{
                sqlite3PcacheDrop(pPg);
              }
            }
          }
        }
        if( rc!=SQLITE_OK ) break;
          rc = walHashGet(pWal, iHash, &sLoc);
          if( rc==SQLITE_OK ){
            u32 i, iMin, iMax;
            assert( mxFrame>=sLoc.iZero );
            iMin = (sLoc.iZero >= iFirst) ? 1 : (iFirst - sLoc.iZero);
            iMax = (iHash==0) ? HASHTABLE_NPAGE_ONE : HASHTABLE_NPAGE;
            if( iMax>(mxFrame-sLoc.iZero) ) iMax = (mxFrame-sLoc.iZero);
            for(i=iMin; rc==SQLITE_OK && i<=iMax; i++){
              PgHdr *pPg;
              if( sLoc.aPgno[i]==1 ){
                /* Check that the schema cookie has not been modified. If
                ** it has not, the commit can proceed. */
                u8 aNew[4];
                u8 *aOld = &((u8*)pPage1->pData)[40];
                int sz;
                i64 iOff;
                u32 iFrame = sLoc.iZero + i;
                int iWal = 0;
                if( bWal2 ){
                  iWal = walExternalDecode(iFrame, &iFrame);
                }
                sz = pWal->hdr.szPage;
                sz = (sz&0xfe00) + ((sz&0x0001)<<16);
                iOff = walFrameOffset(iFrame, sz) + WAL_FRAME_HDRSIZE + 40;
                rc = sqlite3OsRead(pWal->apWalFd[iWal],aNew,sizeof(aNew),iOff);
                if( rc==SQLITE_OK && memcmp(aOld, aNew, sizeof(aNew)) ){
                  rc = SQLITE_BUSY_SNAPSHOT;
                }
              }else if( sqlite3BitvecTestNotNull(pAllRead, sLoc.aPgno[i]) ){
                *piConflict = sLoc.aPgno[i];
                rc = SQLITE_BUSY_SNAPSHOT;
              }else if( (pPg = sqlite3PagerLookup(pPager, sLoc.aPgno[i])) ){
                /* Page aPgno[i], which is present in the pager cache, has been
                ** modified since the current CONCURRENT transaction was
                ** started.  However it was not read by the current
                ** transaction, so is not a conflict. There are two
                ** possibilities: (a) the page was allocated at the of the file
                ** by the current transaction or (b) was present in the cache
                ** at the start of the transaction.
                **
                ** For case (a), do nothing. This page will be moved within the
                ** database file by the commit code to avoid the conflict. The
                ** call to PagerUnref() is to release the reference grabbed by
                ** the sqlite3PagerLookup() above.  
                **
                ** In case (b), drop the page from the cache - otherwise
                ** following the snapshot upgrade the cache would be
                ** inconsistent with the database as stored on disk. */
                if( sqlite3PagerIswriteable(pPg) ){
                  sqlite3PagerUnref(pPg);
                }else{
                  sqlite3PcacheDrop(pPg);
                }
              }
            }
          }
          if( rc!=SQLITE_OK ) break;
        }
      }
    }
  }

  pWal->nPriorFrame = pWal->hdr.mxFrame;
  return rc;
}

  memcpy(&pWal->hdr, (void*)walIndexHdr(pWal), sizeof(WalIndexHdr));

  /* If this client has its read-lock on slot aReadmark[0] and the entire
  ** wal has not been checkpointed, switch it to a different slot. Otherwise
  ** any reads performed between now and committing the transaction will
  ** read from the old snapshot - not the one just upgraded to.  */
  if( pWal->readLock==0 && pWal->hdr.mxFrame!=walCkptInfo(pWal)->nBackfill ){
    assert( isWalMode2(pWal)==0 );
    rc = walUpgradeReadlock(pWal);
  }
  return rc;
}
#endif   /* SQLITE_OMIT_CONCURRENT */

/*

  Wal *pWal, 
  int (*xUndo)(void *, Pgno), 
  void *pUndoCtx,
  int bConcurrent                 /* True if this is a CONCURRENT transaction */
){
  int rc = SQLITE_OK;
  if( pWal->writeLock ){
    int iWal = walidxGetFile(&pWal->hdr);
    Pgno iMax = walidxGetMxFrame(&pWal->hdr, iWal);
    Pgno iMax = pWal->hdr.mxFrame;
    Pgno iNew;
    Pgno iFrame;
  

    assert( isWalMode2(pWal) || iWal==0 );

    /* Restore the clients cache of the wal-index header to the state it
    ** was in before the client began writing to the database. 
    */
    memcpy(&pWal->hdr, (void *)walIndexHdr(pWal), sizeof(WalIndexHdr));
    iNew = walidxGetMxFrame(&pWal->hdr, walidxGetFile(&pWal->hdr));

    /* BEGIN CONCURRENT transactions are different, as the header just
    ** memcpy()d into pWal->hdr may not be the same as the current header 
    ** when the transaction was started. Instead, pWal->hdr now contains
    ** the header written by the most recent successful COMMIT. Because
    ** Wal.writeLock is set, if this is a BEGIN CONCURRENT transaction,
    ** the rollback must be taking place because an error occurred during
    ** a COMMIT.
    **
    ** The code below is still valid. All frames between (iNew+1) and iMax 
    ** must have been written by this transaction before the error occurred.
    ** The exception is in wal2 mode - if the current wal file at the time
    ** of the last COMMIT is not wal file iWal, then the error must have
    ** occurred in WalLockForCommit(), before any pages were written
    ** to the database file. In this case return early.  */
#ifndef SQLITE_OMIT_CONCURRENT
    if( bConcurrent ){
      pWal->hdr.aCksum[0]++;
    }
#else
    UNUSED_PARAMETER(bConcurrent);
    if( walidxGetFile(&pWal->hdr)!=iWal ){
      assert( bConcurrent && isWalMode2(pWal) );
      return SQLITE_OK;
    }
#endif
    assert( walidxGetFile(&pWal->hdr)==iWal );

    for(iFrame=pWal->hdr.mxFrame+1; 
        ALWAYS(rc==SQLITE_OK) && iFrame<=iMax; 
    for(iFrame=iNew+1; ALWAYS(rc==SQLITE_OK) && iFrame<=iMax; iFrame++){
        iFrame++
    ){
      /* This call cannot fail. Unless the page for which the page number
      ** is passed as the second argument is (a) in the cache and 
      ** (b) has an outstanding reference, then xUndo is either a no-op
      ** (if (a) is false) or simply expels the page from the cache (if (b)
      ** is false).
      **
      ** If the upper layer is doing a rollback, it is guaranteed that there
      ** are no outstanding references to any page other than page 1. And
      ** page 1 is never written to the log until the transaction is
      ** committed. As a result, the call to xUndo may not fail.
      */
      Pgno pgno;
      if( isWalMode2(pWal) ){
        pgno = walFramePgno2(pWal, iWal, iFrame);
      }else{
      assert( walFramePgno(pWal, iFrame)!=1 );
      rc = xUndo(pUndoCtx, walFramePgno(pWal, iFrame));
        pgno = walFramePgno(pWal, iFrame);
      }
      assert( pgno!=1 );
      rc = xUndo(pUndoCtx, pgno);
    }
    if( iMax!=pWal->hdr.mxFrame ) walCleanupHash(pWal);
    if( iMax!=iNew ) walCleanupHash(pWal);
  }
  return rc;
}

/* 
** Argument aWalData must point to an array of WAL_SAVEPOINT_NDATA u32 
** values. This function populates the array with values required to 
** "rollback" the write position of the WAL handle back to the current 
** point in the event of a savepoint rollback (via WalSavepointUndo()).
*/
void sqlite3WalSavepoint(Wal *pWal, u32 *aWalData){
  int iWal = walidxGetFile(&pWal->hdr);
  assert( isWalMode2(pWal) || iWal==0 );
  aWalData[0] = pWal->hdr.mxFrame;
  aWalData[0] = walidxGetMxFrame(&pWal->hdr, iWal);
  aWalData[1] = pWal->hdr.aFrameCksum[0];
  aWalData[2] = pWal->hdr.aFrameCksum[1];
  aWalData[3] = pWal->nCkpt;
  aWalData[3] = isWalMode2(pWal) ? iWal : pWal->nCkpt;
}

/* 
** Move the write position of the WAL back to the point identified by
** the values in the aWalData[] array. aWalData must point to an array
** of WAL_SAVEPOINT_NDATA u32 values that has been previously populated
** by a call to WalSavepoint().
*/
int sqlite3WalSavepointUndo(Wal *pWal, u32 *aWalData){
  int rc = SQLITE_OK;
  int iWal = walidxGetFile(&pWal->hdr);
  int iCmp = isWalMode2(pWal) ? iWal : pWal->nCkpt;

  assert( pWal->writeLock || aWalData[0]==pWal->hdr.mxFrame );
  assert( isWalMode2(pWal) || iWal==0 );
  assert( aWalData[3]!=pWal->nCkpt || aWalData[0]<=pWal->hdr.mxFrame );
  assert( aWalData[3]!=iCmp || aWalData[0]<=walidxGetMxFrame(&pWal->hdr,iWal) );

  if( aWalData[3]!=pWal->nCkpt ){
  if( aWalData[3]!=iCmp ){
    /* This savepoint was opened immediately after the write-transaction
    ** was started. Right after that, the writer decided to wrap around
    ** to the start of the log. Update the savepoint values to match.
    */
    aWalData[0] = 0;
    aWalData[3] = pWal->nCkpt;
    aWalData[3] = iCmp;
  }

  if( aWalData[0]<pWal->hdr.mxFrame ){
    pWal->hdr.mxFrame = aWalData[0];
  if( aWalData[0]<walidxGetMxFrame(&pWal->hdr, iWal) ){
    walidxSetMxFrame(&pWal->hdr, iWal, aWalData[0]);
    pWal->hdr.aFrameCksum[0] = aWalData[1];
    pWal->hdr.aFrameCksum[1] = aWalData[2];
    walCleanupHash(pWal);
  }

  return rc;
}

/*
** This function is called just before writing a set of frames to the log
** file (see sqlite3WalFrames()). It checks to see if, instead of appending
** to the current log file, it is possible to overwrite the start of the
** existing log file with the new frames (i.e. "reset" the log). If so,
** it sets pWal->hdr.mxFrame to 0. Otherwise, pWal->hdr.mxFrame is left
** unchanged.
** to the current log file, it is possible and desirable to switch to the
** other log file and write the new transaction to the start of it.
** If so, the wal-index header is updated accordingly - both in heap memory
** and in the *-shm file.
**
** SQLITE_OK is returned if no error is encountered (regardless of whether
** or not pWal->hdr.mxFrame is modified). An SQLite error code is returned
** or not the wal-index header is modified). An SQLite error code is returned
** if an error occurs.
*/
static int walRestartLog(Wal *pWal){
  int rc = SQLITE_OK;

  if( isWalMode2(pWal) ){
    int iApp = walidxGetFile(&pWal->hdr);
    int nWalSize = WAL_DEFAULT_WALSIZE;
    if( pWal->mxWalSize>0 ){
      nWalSize = (pWal->mxWalSize-WAL_HDRSIZE+pWal->szPage+WAL_FRAME_HDRSIZE-1) 
        / (pWal->szPage+WAL_FRAME_HDRSIZE);
      nWalSize = MAX(nWalSize, 1);
    }

    if( walidxGetMxFrame(&pWal->hdr, iApp)>=nWalSize ){
      volatile WalCkptInfo *pInfo = walCkptInfo(pWal);
      u32 mxFrame = walidxGetMxFrame(&pWal->hdr, !iApp);
      if( mxFrame==0 || pInfo->nBackfill ){
        rc = wal2RestartOk(pWal, iApp);
        if( rc==SQLITE_OK ){
          int iNew = !iApp;
          pWal->nCkpt++;
          walidxSetFile(&pWal->hdr, iNew);
          walidxSetMxFrame(&pWal->hdr, iNew, 0);
          sqlite3Put4byte((u8*)&pWal->hdr.aSalt[0], pWal->hdr.aFrameCksum[0]);
          sqlite3Put4byte((u8*)&pWal->hdr.aSalt[1], pWal->hdr.aFrameCksum[1]);
          walIndexWriteHdr(pWal);
          pInfo->nBackfill = 0;
          wal2RestartFinished(pWal, iApp);
          walUnlockShared(pWal, WAL_READ_LOCK(pWal->readLock));
          pWal->readLock = iNew ? WAL_LOCK_PART2_FULL1 : WAL_LOCK_PART1_FULL2;
          rc = walLockShared(pWal, WAL_READ_LOCK(pWal->readLock));
        }else if( rc==SQLITE_BUSY ){
          rc = SQLITE_OK;
        }
      }
    }
  if( pWal->readLock==0 ){
  }else if( pWal->readLock==0 ){
    volatile WalCkptInfo *pInfo = walCkptInfo(pWal);
    assert( pInfo->nBackfill==pWal->hdr.mxFrame );
    if( pInfo->nBackfill>0 ){
      u32 salt1;
      sqlite3_randomness(4, &salt1);
      rc = walLockExclusive(pWal, WAL_READ_LOCK(1), WAL_NREADER-1);
      if( rc==SQLITE_OK ){

    ** only - never from the wal file. This means that if a writer holding
    ** a lock on aReadmark[0] were to commit a transaction but not close the
    ** read-transaction, subsequent read operations would read directly from
    ** the database file - ignoring the new pages just appended
    ** to the wal file. */
    rc = walUpgradeReadlock(pWal);
  }

  return rc;
}

/*
** Information about the current state of the WAL file and where
** the next fsync should occur - passed from sqlite3WalFrames() into
** walWriteToLog().

  PgHdr *pPage,               /* The page of the frame to be written */
  int nTruncate,              /* The commit flag.  Usually 0.  >0 for commit */
  sqlite3_int64 iOffset       /* Byte offset at which to write */
){
  int rc;                         /* Result code from subfunctions */
  void *pData;                    /* Data actually written */
  u8 aFrame[WAL_FRAME_HDRSIZE];   /* Buffer to assemble frame-header in */

#if defined(SQLITE_TEST) && defined(SQLITE_DEBUG)
  { 
    int iWal = walidxGetFile(&p->pWal->hdr);
    int iFrame = 1 + (iOffset / (WAL_FRAME_HDRSIZE + p->pWal->szPage));
    assert( p->pWal->apWalFd[iWal]==p->pFd );
    WALTRACE(("WAL%p: page %d written to frame %d of wal %d\n",
          p->pWal, (int)pPage->pgno, iFrame, iWal
    ));
  }
#endif

#if defined(SQLITE_HAS_CODEC)
  if( (pData = sqlite3PagerCodec(pPage))==0 ) return SQLITE_NOMEM_BKPT;
#else
  pData = pPage->pData;
#endif
  walEncodeFrame(p->pWal, pPage->pgno, nTruncate, pData, aFrame);
  rc = walWriteToLog(p, aFrame, sizeof(aFrame), iOffset);

** one or more frames have been overwritten. It updates the checksums for
** all frames written to the wal file by the current transaction starting
** with the earliest to have been overwritten.
**
** SQLITE_OK is returned if successful, or an SQLite error code otherwise.
*/
static int walRewriteChecksums(Wal *pWal, u32 iLast){
  const int szPage = pWal->szPage;/* Database page size */
  int rc = SQLITE_OK;             /* Return code */
  const int szPage = pWal->szPage;/* Database page size */
  u8 *aBuf;                       /* Buffer to load data from wal file into */
  u8 aFrame[WAL_FRAME_HDRSIZE];   /* Buffer to assemble frame-headers in */
  u32 iRead;                      /* Next frame to read from wal file */
  i64 iCksumOff;
  sqlite3_file *pWalFd = pWal->apWalFd[walidxGetFile(&pWal->hdr)];

  aBuf = sqlite3_malloc(szPage + WAL_FRAME_HDRSIZE);
  if( aBuf==0 ) return SQLITE_NOMEM_BKPT;

  /* Find the checksum values to use as input for the recalculating the
  ** first checksum. If the first frame is frame 1 (implying that the current
  ** transaction restarted the wal file), these values must be read from the
  ** wal-file header. Otherwise, read them from the frame header of the
  ** previous frame.  */
  assert( pWal->iReCksum>0 );
  if( pWal->iReCksum==1 ){
    iCksumOff = 24;
  }else{
    iCksumOff = walFrameOffset(pWal->iReCksum-1, szPage) + 16;
  }
  rc = sqlite3OsRead(pWal->pWalFd, aBuf, sizeof(u32)*2, iCksumOff);
  rc = sqlite3OsRead(pWalFd, aBuf, sizeof(u32)*2, iCksumOff);
  pWal->hdr.aFrameCksum[0] = sqlite3Get4byte(aBuf);
  pWal->hdr.aFrameCksum[1] = sqlite3Get4byte(&aBuf[sizeof(u32)]);

  iRead = pWal->iReCksum;
  pWal->iReCksum = 0;
  for(; rc==SQLITE_OK && iRead<=iLast; iRead++){
    i64 iOff = walFrameOffset(iRead, szPage);
    rc = sqlite3OsRead(pWal->pWalFd, aBuf, szPage+WAL_FRAME_HDRSIZE, iOff);
    rc = sqlite3OsRead(pWalFd, aBuf, szPage+WAL_FRAME_HDRSIZE, iOff);
    if( rc==SQLITE_OK ){
      u32 iPgno, nDbSize;
      iPgno = sqlite3Get4byte(aBuf);
      nDbSize = sqlite3Get4byte(&aBuf[4]);

      walEncodeFrame(pWal, iPgno, nDbSize, &aBuf[WAL_FRAME_HDRSIZE], aFrame);
      rc = sqlite3OsWrite(pWal->pWalFd, aFrame, sizeof(aFrame), iOff);
      rc = sqlite3OsWrite(pWalFd, aFrame, sizeof(aFrame), iOff);
    }
  }

  sqlite3_free(aBuf);
  return rc;
}

  PgHdr *pLast = 0;               /* Last frame in list */
  int nExtra = 0;                 /* Number of extra copies of last page */
  int szFrame;                    /* The size of a single frame */
  i64 iOffset;                    /* Next byte to write in WAL file */
  WalWriter w;                    /* The writer */
  u32 iFirst = 0;                 /* First frame that may be overwritten */
  WalIndexHdr *pLive;             /* Pointer to shared header */
  int iApp;
  int bWal2 = isWalMode2(pWal);

  assert( pList );
  assert( pWal->writeLock );

  /* If this frame set completes a transaction, then nTruncate>0.  If
  ** nTruncate==0 then this frame set does not complete the transaction. */
  assert( (isCommit!=0)==(nTruncate!=0) );

#if defined(SQLITE_TEST) && defined(SQLITE_DEBUG)
  { int cnt; for(cnt=0, p=pList; p; p=p->pDirty, cnt++){}
    WALTRACE(("WAL%p: frame write begin. %d frames. mxFrame=%d. %s\n",
              pWal, cnt, pWal->hdr.mxFrame, isCommit ? "Commit" : "Spill"));
  }
#endif

  pLive = (WalIndexHdr*)walIndexHdr(pWal);
  if( memcmp(&pWal->hdr, (void *)pLive, sizeof(WalIndexHdr))!=0 ){
    /* if( isWalMode2(pWal)==0 ) */
    iFirst = pLive->mxFrame+1;
    iFirst = walidxGetMxFrame(pLive, walidxGetFile(pLive))+1;
  }

  /* See if it is possible to write these frames into the start of the
  ** log file, instead of appending to it at pWal->hdr.mxFrame.
  */
  if( SQLITE_OK!=(rc = walRestartLog(pWal)) ){
  else if( SQLITE_OK!=(rc = walRestartLog(pWal)) ){
    return rc;
  }

  /* If this is the first frame written into the log, write the WAL
  ** header to the start of the WAL file. See comments at the top of
  ** this source file for a description of the WAL header format.
  */
  iApp = walidxGetFile(&pWal->hdr);
  iFrame = walidxGetMxFrame(&pWal->hdr, iApp);
  assert( iApp==0 || bWal2 );
  iFrame = pWal->hdr.mxFrame;

#if defined(SQLITE_TEST) && defined(SQLITE_DEBUG)
  { int cnt; for(cnt=0, p=pList; p; p=p->pDirty, cnt++){}
    WALTRACE(("WAL%p: frame write begin. %d frames. iWal=%d. mxFrame=%d. %s\n",
              pWal, cnt, iApp, iFrame, isCommit ? "Commit" : "Spill"));
  }
#endif

  if( iFrame==0 ){
    u32 iCkpt = 0;
    u8 aWalHdr[WAL_HDRSIZE];      /* Buffer to assemble wal-header in */
    u32 aCksum[2];                /* Checksum for wal-header */

    sqlite3Put4byte(&aWalHdr[0], (WAL_MAGIC | SQLITE_BIGENDIAN));
    sqlite3Put4byte(&aWalHdr[4], WAL_MAX_VERSION);
    sqlite3Put4byte(&aWalHdr[4], pWal->hdr.iVersion);
    sqlite3Put4byte(&aWalHdr[8], szPage);
    if( bWal2 ){
      if( walidxGetMxFrame(&pWal->hdr, !iApp)>0 ){
        u8 aPrev[4];
        rc = sqlite3OsRead(pWal->apWalFd[!iApp], aPrev, 4, 12);
        if( rc!=SQLITE_OK ){
          return rc;
        }
        iCkpt = (sqlite3Get4byte(aPrev) + 1) & 0x0F;
      }
    }else{
      iCkpt = pWal->nCkpt;
    }
    sqlite3Put4byte(&aWalHdr[12], pWal->nCkpt);
    sqlite3Put4byte(&aWalHdr[12], iCkpt);
    if( pWal->nCkpt==0 ) sqlite3_randomness(8, pWal->hdr.aSalt);
    memcpy(&aWalHdr[16], pWal->hdr.aSalt, 8);
    walChecksumBytes(1, aWalHdr, WAL_HDRSIZE-2*4, 0, aCksum);
    sqlite3Put4byte(&aWalHdr[24], aCksum[0]);
    sqlite3Put4byte(&aWalHdr[28], aCksum[1]);
    

    pWal->szPage = szPage;
    pWal->hdr.bigEndCksum = SQLITE_BIGENDIAN;
    pWal->hdr.aFrameCksum[0] = aCksum[0];
    pWal->hdr.aFrameCksum[1] = aCksum[1];
    pWal->truncateOnCommit = 1;

    rc = sqlite3OsWrite(pWal->pWalFd, aWalHdr, sizeof(aWalHdr), 0);
    rc = sqlite3OsWrite(pWal->apWalFd[iApp], aWalHdr, sizeof(aWalHdr), 0);
    WALTRACE(("WAL%p: wal-header write %s\n", pWal, rc ? "failed" : "ok"));
    if( rc!=SQLITE_OK ){
      return rc;
    }

    /* Sync the header (unless SQLITE_IOCAP_SEQUENTIAL is true or unless
    ** all syncing is turned off by PRAGMA synchronous=OFF).  Otherwise
    ** an out-of-order write following a WAL restart could result in
    ** database corruption.  See the ticket:
    **
    **     https://sqlite.org/src/info/ff5be73dee
    */
    if( pWal->syncHeader ){
      rc = sqlite3OsSync(pWal->pWalFd, CKPT_SYNC_FLAGS(sync_flags));
      rc = sqlite3OsSync(pWal->apWalFd[iApp], CKPT_SYNC_FLAGS(sync_flags));
      if( rc ) return rc;
    }
  }
  assert( (int)pWal->szPage==szPage );

  /* Setup information needed to write frames into the WAL */
  w.pWal = pWal;
  w.pFd = pWal->pWalFd;
  w.pFd = pWal->apWalFd[iApp];
  w.iSyncPoint = 0;
  w.syncFlags = sync_flags;
  w.szPage = szPage;
  iOffset = walFrameOffset(iFrame+1, szPage);
  szFrame = szPage + WAL_FRAME_HDRSIZE;

  /* Write all frames into the log file exactly once */
  for(p=pList; p; p=p->pDirty){
    int nDbSize;   /* 0 normally.  Positive == commit flag */

    /* Check if this page has already been written into the wal file by
    ** the current transaction. If so, overwrite the existing frame and
    ** set Wal.writeLock to WAL_WRITELOCK_RECKSUM - indicating that 
    ** checksums must be recomputed when the transaction is committed.  */
    if( iFirst && (p->pDirty || isCommit==0) ){
      u32 iWrite = 0;
      VVA_ONLY(rc =) sqlite3WalFindFrame(pWal, p->pgno, &iWrite);
      VVA_ONLY(rc =) walSearchWal(pWal, iApp, p->pgno, &iWrite);
      assert( rc==SQLITE_OK || iWrite==0 );
      if( iWrite && bWal2 ){
        walExternalDecode(iWrite, &iWrite);
      }
      if( iWrite>=iFirst ){
        i64 iOff = walFrameOffset(iWrite, szPage) + WAL_FRAME_HDRSIZE;
        void *pData;
        if( pWal->iReCksum==0 || iWrite<pWal->iReCksum ){
          pWal->iReCksum = iWrite;
        }
#if defined(SQLITE_HAS_CODEC)
        if( (pData = sqlite3PagerCodec(p))==0 ) return SQLITE_NOMEM;
#else
        pData = p->pData;
#endif
        rc = sqlite3OsWrite(pWal->pWalFd, pData, szPage, iOff);
        rc = sqlite3OsWrite(pWal->apWalFd[iApp], pData, szPage, iOff);
        if( rc ) return rc;
        p->flags &= ~PGHDR_WAL_APPEND;
        continue;
      }
    }

    iFrame++;

  ** boundary is crossed.  Only the part of the WAL prior to the last
  ** sector boundary is synced; the part of the last frame that extends
  ** past the sector boundary is written after the sync.
  */
  if( isCommit && WAL_SYNC_FLAGS(sync_flags)!=0 ){
    int bSync = 1;
    if( pWal->padToSectorBoundary ){
      int sectorSize = sqlite3SectorSize(pWal->pWalFd);
      int sectorSize = sqlite3SectorSize(w.pFd);
      w.iSyncPoint = ((iOffset+sectorSize-1)/sectorSize)*sectorSize;
      bSync = (w.iSyncPoint==iOffset);
      testcase( bSync );
      while( iOffset<w.iSyncPoint ){
        rc = walWriteOneFrame(&w, pLast, nTruncate, iOffset);
        if( rc ) return rc;
        iOffset += szFrame;

  }

  /* Append data to the wal-index. It is not necessary to lock the 
  ** wal-index to do this as the SQLITE_SHM_WRITE lock held on the wal-index
  ** guarantees that there are no other writers, and no data that may
  ** be in use by existing readers is being overwritten.
  */
  iFrame = pWal->hdr.mxFrame;
  iFrame = walidxGetMxFrame(&pWal->hdr, iApp);
  for(p=pList; p && rc==SQLITE_OK; p=p->pDirty){
    if( (p->flags & PGHDR_WAL_APPEND)==0 ) continue;
    iFrame++;
    rc = walIndexAppend(pWal, iFrame, p->pgno);
    rc = walIndexAppend(pWal, iApp, iFrame, p->pgno);
  }
  while( rc==SQLITE_OK && nExtra>0 ){
    iFrame++;
    nExtra--;
    rc = walIndexAppend(pWal, iFrame, pLast->pgno);
    rc = walIndexAppend(pWal, iApp, iFrame, pLast->pgno);
  }

  if( rc==SQLITE_OK ){
    /* Update the private copy of the header. */
    pWal->hdr.szPage = (u16)((szPage&0xff00) | (szPage>>16));
    testcase( szPage<=32768 );
    testcase( szPage>=65536 );
    pWal->hdr.mxFrame = iFrame;
    walidxSetMxFrame(&pWal->hdr, iApp, iFrame);
    if( isCommit ){
      pWal->hdr.iChange++;
      pWal->hdr.nPage = nTruncate;
    }
    /* If this is a commit, update the wal-index header too. */
    if( isCommit ){
      walIndexWriteHdr(pWal);
      if( bWal2 ){
        int iOther = !walidxGetFile(&pWal->hdr);
        if( walidxGetMxFrame(&pWal->hdr, iOther) 
            && !walCkptInfo(pWal)->nBackfill 
        ){
          pWal->iCallback = walidxGetMxFrame(&pWal->hdr, 0);
          pWal->iCallback += walidxGetMxFrame(&pWal->hdr, 1);
        }
      }else{
      pWal->iCallback = iFrame;
        pWal->iCallback = iFrame;
      }
    }
  }

  WALTRACE(("WAL%p: frame write %s\n", pWal, rc ? "failed" : "ok"));
  return rc;
}

    if( isChanged && pWal->pDbFd->pMethods->iVersion>=3 ){
      sqlite3OsUnfetch(pWal->pDbFd, 0, 0);
    }
  }

  /* Copy data from the log to the database file. */
  if( rc==SQLITE_OK ){
    if( pWal->hdr.mxFrame && walPagesize(pWal)!=nBuf ){
    if( (walPagesize(pWal)!=nBuf) 
     && ((pWal->hdr.mxFrame2 & 0x7FFFFFFF) || pWal->hdr.mxFrame)
    ){
      rc = SQLITE_CORRUPT_BKPT;
    }else{
      rc = walCheckpoint(pWal, db, eMode2, xBusy2, pBusyArg, sync_flags, zBuf);
    }

    /* If no error occurred, set the output variables. */
    if( rc==SQLITE_OK || rc==SQLITE_BUSY ){
      if( pnLog ) *pnLog = (int)pWal->hdr.mxFrame;
      if( pnCkpt ) *pnCkpt = (int)(walCkptInfo(pWal)->nBackfill);
      if( pnLog ){
        *pnLog = walidxGetMxFrame(&pWal->hdr,0)+walidxGetMxFrame(&pWal->hdr,1);
      }
      if( pnCkpt ){
        if( isWalMode2(pWal) ){
          if( (int)(walCkptInfo(pWal)->nBackfill) ){
            *pnCkpt = walidxGetMxFrame(&pWal->hdr, !walidxGetFile(&pWal->hdr));
          }else{
            *pnCkpt = 0;
          }
        }else{
          *pnCkpt = walCkptInfo(pWal)->nBackfill;
        }
      }
    }
  }

  if( isChanged ){
    /* If a new wal-index header was loaded before the checkpoint was 
    ** performed, then the pager-cache associated with pWal is now
    ** out of date. So zero the cached wal-index header to ensure that

** If op is negative, then do a dry-run of the op==1 case but do
** not actually change anything. The pager uses this to see if it
** should acquire the database exclusive lock prior to invoking
** the op==1 case.
*/
int sqlite3WalExclusiveMode(Wal *pWal, int op){
  int rc;

  assert( pWal->writeLock==0 );
  assert( pWal->exclusiveMode!=WAL_HEAPMEMORY_MODE || op==-1 );

  /* pWal->readLock is usually set, but might be -1 if there was a 
  ** prior error while attempting to acquire are read-lock. This cannot 
  ** happen if the connection is actually in exclusive mode (as no xShmLock
  ** locks are taken in this case). Nor should the pager attempt to
  ** upgrade to exclusive-mode following such an error.
  */
  assert( pWal->readLock>=0 || pWal->lockError );
  assert( pWal->readLock>=0 || (op<=0 && pWal->exclusiveMode==0) );
  assert( pWal->readLock!=WAL_LOCK_NONE || pWal->lockError );
  assert( pWal->readLock!=WAL_LOCK_NONE || (op<=0 && pWal->exclusiveMode==0) );

  if( op==0 ){
    if( pWal->exclusiveMode!=WAL_NORMAL_MODE ){
    if( pWal->exclusiveMode ){
      pWal->exclusiveMode = WAL_NORMAL_MODE;
      if( walLockShared(pWal, WAL_READ_LOCK(pWal->readLock))!=SQLITE_OK ){
      rc = walLockShared(pWal, WAL_READ_LOCK(pWal->readLock));
      if( rc!=SQLITE_OK ){
        pWal->exclusiveMode = WAL_EXCLUSIVE_MODE;
      }
      rc = pWal->exclusiveMode==WAL_NORMAL_MODE;
    }else{
      /* Already in locking_mode=NORMAL */
      rc = 0;
    }

** every other subsystem, so the WAL module can put whatever it needs
** in the object.
*/
int sqlite3WalSnapshotGet(Wal *pWal, sqlite3_snapshot **ppSnapshot){
  int rc = SQLITE_OK;
  WalIndexHdr *pRet;
  static const u32 aZero[4] = { 0, 0, 0, 0 };

  /* Snapshots may not be used with wal2 mode databases. */
  if( isWalMode2(pWal) ) return SQLITE_ERROR;

  assert( pWal->readLock>=0 && pWal->writeLock==0 );

  if( memcmp(&pWal->hdr.aFrameCksum[0],aZero,16)==0 ){
  if( memcmp(&pWal->hdr.aFrameCksum[0],aZero,8)==0 ){
    *ppSnapshot = 0;
    return SQLITE_ERROR;
  }
  pRet = (WalIndexHdr*)sqlite3_malloc(sizeof(WalIndexHdr));
  if( pRet==0 ){
    rc = SQLITE_NOMEM_BKPT;
  }else{

** If the snapshot is not available, SQLITE_ERROR is returned. Or, if
** the CHECKPOINTER lock cannot be obtained, SQLITE_BUSY. If any error
** occurs (any value other than SQLITE_OK is returned), the CHECKPOINTER
** lock is released before returning.
*/
int sqlite3WalSnapshotCheck(Wal *pWal, sqlite3_snapshot *pSnapshot){
  int rc;

  /* Snapshots may not be used with wal2 mode databases. */
  if( isWalMode2(pWal) ) return SQLITE_ERROR;

  rc = walLockShared(pWal, WAL_CKPT_LOCK);
  if( rc==SQLITE_OK ){
    WalIndexHdr *pNew = (WalIndexHdr*)pSnapshot;
    if( memcmp(pNew->aSalt, pWal->hdr.aSalt, sizeof(pWal->hdr.aSalt))
     || pNew->mxFrame<walCkptInfo(pWal)->nBackfillAttempted
    ){
      rc = SQLITE_ERROR_SNAPSHOT;

  return (pWal ? pWal->szPage : 0);
}
#endif

/* Return the sqlite3_file object for the WAL file
*/
sqlite3_file *sqlite3WalFile(Wal *pWal){
  return pWal->pWalFd;
  return pWal->apWalFd[0];
}

/* 
** Return the values required by sqlite3_wal_info().
*/
int sqlite3WalInfo(Wal *pWal, u32 *pnPrior, u32 *pnFrame){
  int rc = SQLITE_OK;
  if( pWal ){
    *pnFrame = pWal->hdr.mxFrame;
    *pnPrior = pWal->nPriorFrame;
  }
  return rc;
}

/* 
** Return the journal mode used by this Wal object.
*/
int sqlite3WalJournalMode(Wal *pWal){
  assert( pWal );
  return (isWalMode2(pWal) ? PAGER_JOURNALMODE_WAL2 : PAGER_JOURNALMODE_WAL);
}

#endif /* #ifndef SQLITE_OMIT_WAL */

/* Macros for extracting appropriate sync flags for either transaction
** commits (WAL_SYNC_FLAGS(X)) or for checkpoint ops (CKPT_SYNC_FLAGS(X)):
*/
#define WAL_SYNC_FLAGS(X)   ((X)&0x03)
#define CKPT_SYNC_FLAGS(X)  (((X)>>2)&0x03)

#ifdef SQLITE_OMIT_WAL
# define sqlite3WalOpen(x,y,z)                   0
# define sqlite3WalOpen(w,x,y,z)                 0
# define sqlite3WalLimit(x,y)
# define sqlite3WalClose(v,w,x,y,z)              0
# define sqlite3WalBeginReadTransaction(y,z)     0
# define sqlite3WalEndReadTransaction(z)
# define sqlite3WalDbsize(y)                     0
# define sqlite3WalBeginWriteTransaction(y)      0
# define sqlite3WalEndWriteTransaction(x)        0
# define sqlite3WalUndo(w,x,y,z)                 0
# define sqlite3WalSavepoint(y,z)
# define sqlite3WalSavepointUndo(y,z)            0
# define sqlite3WalFrames(u,v,w,x,y,z)           0
# define sqlite3WalCheckpoint(q,r,s,t,u,v,w,x,y,z) 0
# define sqlite3WalCallback(z)                   0
# define sqlite3WalExclusiveMode(y,z)            0
# define sqlite3WalHeapMemory(z)                 0
# define sqlite3WalFramesize(z)                  0
# define sqlite3WalFindFrame(x,y,z)              0
# define sqlite3WalFile(x)                       0
# define sqlite3WalJournalMode(x)                0
#else

#define WAL_SAVEPOINT_NDATA 4

/* Connection to a write-ahead log (WAL) file. 
** There is one object of this type for each pager. 
*/
typedef struct Wal Wal;

/* Open and close a connection to a write-ahead log. */
int sqlite3WalOpen(sqlite3_vfs*, sqlite3_file*, const char *, int, i64, Wal**);
int sqlite3WalOpen(sqlite3_vfs*, sqlite3_file*, const char *,int,i64,int,Wal**);
int sqlite3WalClose(Wal *pWal, sqlite3*, int sync_flags, int, u8 *);

/* Set the limiting size of a WAL file. */
void sqlite3WalLimit(Wal*, i64);

/* Used by readers to open (lock) and close (unlock) a snapshot.  A 
** snapshot is like a read-transaction.  It is the state of the database

** stored in each frame (i.e. the db page-size when the WAL was created).
*/
int sqlite3WalFramesize(Wal *pWal);
#endif

/* Return the sqlite3_file object for the WAL file */
sqlite3_file *sqlite3WalFile(Wal *pWal);

/* Return the journal mode (WAL or WAL2) used by this Wal object. */
int sqlite3WalJournalMode(Wal *pWal);

/* sqlite3_wal_info() data */
int sqlite3WalInfo(Wal *pWal, u32 *pnPrior, u32 *pnFrame);

/* sqlite3_wal_info() data */
int sqlite3WalInfo(Wal *pWal, u32 *pnPrior, u32 *pnFrame);

#endif /* ifndef SQLITE_OMIT_WAL */
#endif /* SQLITE_WAL_H */

  }
} -test {
  faultsim_test_result {0 {}} 
  catchsql { ROLLBACK }
  faultsim_integrity_check
}


#-------------------------------------------------------------------------
reset_db

do_execsql_test 2.0 {
  PRAGMA auto_vacuum = 0;
  PRAGMA journal_mode = wal;
  CREATE TABLE t1(a PRIMARY KEY, b);
  CREATE TABLE t2(a PRIMARY KEY, b);
  INSERT INTO t1 VALUES(randomblob(1000), randomblob(100));
  INSERT INTO t1 SELECT randomblob(1000), randomblob(1000) FROM t1;
  INSERT INTO t1 SELECT randomblob(1000), randomblob(1000) FROM t1;
  INSERT INTO t1 SELECT randomblob(1000), randomblob(1000) FROM t1;
  INSERT INTO t1 SELECT randomblob(1000), randomblob(1000) FROM t1;
  DELETE FROM t1 WHERE rowid%2;
} {wal}

faultsim_save_and_close
do_faultsim_test 1 -prep {
  faultsim_restore_and_reopen
  execsql {
    SELECT * FROM t1;
    BEGIN CONCURRENT;
      INSERT INTO t2 VALUES(1, 2);
  }
  sqlite3 db2 test.db
  execsql {
    PRAGMA journal_size_limit = 10000;
    INSERT INTO t1 VALUES(randomblob(1000), randomblob(1000));
  } db2
  db2 close
} -body {
  execsql { COMMIT }
} -test {
  faultsim_test_result {0 {}} 
  catchsql { ROLLBACK }
  set res [catchsql { SELECT count(*) FROM t1 }]
  if {$res!="0 9"} { error "expected {0 9} got {$res}" }
  faultsim_integrity_check
}

finish_test

# 2018 Dec 28
#
# 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.
#
#***********************************************************************
#
# This file contains fault injection tests designed to test the concurrent
# transactions feature.
#

set testdir [file dirname $argv0]
source $testdir/tester.tcl
source $testdir/malloc_common.tcl
set testprefix concfault2

ifcapable !concurrent {
  finish_test
  return
}

do_execsql_test 1.0 {
  PRAGMA auto_vacuum = 0;
  PRAGMA journal_mode = wal2;
  CREATE TABLE t1(a PRIMARY KEY, b);
  CREATE TABLE t2(a PRIMARY KEY, b);
  INSERT INTO t1 VALUES(randomblob(1000), randomblob(100));
  INSERT INTO t1 SELECT randomblob(1000), randomblob(1000) FROM t1;
  INSERT INTO t1 SELECT randomblob(1000), randomblob(1000) FROM t1;
  INSERT INTO t1 SELECT randomblob(1000), randomblob(1000) FROM t1;
  INSERT INTO t1 SELECT randomblob(1000), randomblob(1000) FROM t1;
  DELETE FROM t1 WHERE rowid%2;
} {wal2}

do_test 1.1 {
  list [expr [file size test.db-wal]>75000] [file size test.db-shm]
} {1 32768}

faultsim_save_and_close

do_faultsim_test 1 -prep {
  faultsim_restore_and_reopen
  execsql {
    SELECT * FROM t1;
    BEGIN CONCURRENT;
      INSERT INTO t2 VALUES(1, 2);
  }
  sqlite3 db2 test.db
  execsql {
    PRAGMA journal_size_limit = 10000;
    INSERT INTO t1 VALUES(randomblob(1000), randomblob(1000));
  } db2
  db2 close
} -body {
  execsql { COMMIT }
} -test {
  faultsim_test_result {0 {}} 
  catchsql { ROLLBACK }
  set res [catchsql { SELECT count(*) FROM t1 }]
  if {$res!="0 9"} { error "expected {0 9} got {$res}" }
  faultsim_integrity_check
}

finish_test

source $testdir/wal_common.tcl
set ::testprefix concurrent2

ifcapable !concurrent {
  finish_test
  return
}

do_test 0.1 {
  llength [sqlite3_wal_info db main]
} {2}

do_multiclient_test tn {

  do_test 1.$tn.1 {
    sql1 {
      PRAGMA journal_mode = wal;
      CREATE TABLE t1(x);

# Corrupt the file header in various ways and make sure the corruption
# is detected when opening the database file.
#
db close
forcecopy test.db test.db-template

set unreadable_version 02
ifcapable wal { set unreadable_version 03 }
ifcapable wal { set unreadable_version 04 }
do_test corruptA-2.1 {
  forcecopy test.db-template test.db
  hexio_write test.db 19 $unreadable_version   ;# the read format number
  sqlite3 db test.db
  catchsql {SELECT * FROM t1}  
} {1 {file is not a database}}
 

# Define the coverage related test suites:
#
#   coverage-wal
#
test_suite "coverage-wal" -description {
  Coverage tests for file wal.c.
} -files {
wal2big.test wal2recover.test wal2rewrite.test
wal2simple.test wal2snapshot.test wal2.test
  wal.test wal2.test wal3.test wal4.test wal5.test
  wal64k.test wal6.test wal7.test wal8.test wal9.test
  walbak.test walbig.test walblock.test walcksum.test walcrash2.test
  walcrash3.test walcrash4.test walcrash.test walfault.test walhook.test
  walmode.test walnoshm.test waloverwrite.test walpersist.test 
  walprotocol2.test walprotocol.test walro2.test walrofault.test 
  walro.test walshared.test walslow.test walvfs.test
  walfault2.test
  nockpt.test
wal3.test wal4.test wal5.test
wal64k.test wal6.test wal7.test wal8.test wal9.test
walbak.test walbig.test walblock.test walcksum.test 
walfault.test walhook.test walmode.test walnoshm.test
waloverwrite.test walpersist.test walprotocol2.test
walprotocol.test walro2.test walrofault.test walro.test
walshared.test walslow.test wal.test
wal2savepoint.test wal2lock.test wal2recover2.test

  wal2concurrent.test
  concurrent.test concurrent2.test concurrent3.test
  concurrent4.test concurrent5.test concurrent6.test
  concurrent7.test
  concfault.test concfault2.test

  walvfs.test walfault2.test nockpt.test
  snapshot2.test snapshot3.test snapshot4.test
  snapshot_fault.test snapshot.test snapshot_up.test
  walcrash2.test walcrash3.test walcrash4.test walcrash.test
  wal2fault.test
} 

test_suite "coverage-pager" -description {
  Coverage tests for file pager.c.
} -files {
  pager1.test    pager2.test  pagerfault.test  pagerfault2.test
  walfault.test  walbak.test  journal2.test    tkt-9d68c883.test

    insert.test   insert2.test  insert3.test rollback.test 
    select1.test  select2.test  select3.test
  }
}

test_suite "wal" -description {
  Run tests with journal_mode=WAL
} -initialize {
  set ::G(savepoint6_iterations) 100
} -shutdown {
  unset -nocomplain ::G(savepoint6_iterations)
} -files {
  savepoint.test     savepoint2.test     savepoint6.test
  trans.test         avtrans.test

  fts3aa.test  fts3ab.test  fts3ac.test  fts3ad.test
  fts3ae.test  fts3af.test  fts3ag.test  fts3ah.test
  fts3ai.test  fts3aj.test  fts3ak.test  fts3al.test
  fts3am.test  fts3an.test  fts3ao.test  fts3b.test
  fts3c.test   fts3d.test   fts3e.test   fts3query.test 
}

test_suite "wal2" -description {
  Run tests with journal_mode=WAL2
} -initialize {
  set ::G(savepoint6_iterations) 100
} -shutdown {
  unset -nocomplain ::G(savepoint6_iterations)
} -files {
  savepoint.test     savepoint2.test     savepoint6.test
  trans.test         avtrans.test

# returns 1 if the database N of connection D is read-only, 0 if it is
# read/write, or -1 if N is not the name of a database on connection D.
#
do_test rdonly-1.1.1 {
  sqlite3_db_readonly db main
} {0}

# Changes the write version from 1 to 3.  Verify that the database
# Changes the write version from 1 to 4.  Verify that the database
# can be read but not written.
#
do_test rdonly-1.2 {
  db close
  hexio_get_int [hexio_read test.db 18 1]
} 1
do_test rdonly-1.3 {
  hexio_write test.db 18 03
  hexio_write test.db 18 04
  sqlite3 db test.db
  execsql {
    SELECT * FROM t1;
  }
} {1}
do_test rdonly-1.3.1 {
  sqlite3_db_readonly db main

# Now, after connection [db] has loaded the database schema, modify the
# write-version of the file (and the change-counter, so that the 
# write-version is reloaded). This way, SQLite does not discover that
# the database is read-only until after it is locked.
#
set ro_version 02
ifcapable wal { set ro_version 03 }
ifcapable wal { set ro_version 04 }
do_test rdonly-1.6 {
  hexio_write test.db 18 $ro_version     ; # write-version
  hexio_write test.db 24 11223344        ; # change-counter
  catchsql {
    INSERT INTO t1 VALUES(2);
  }
} {1 {attempt to write a readonly database}}

finish_test

do_test savepoint-1.1 {
  wal_set_journal_mode
  execsql {
    SAVEPOINT sp1;
    RELEASE sp1;
  }
} {}
wal_check_journal_mode savepoint-1.1
do_test savepoint-1.2 {
  execsql {
    SAVEPOINT sp1;
    ROLLBACK TO sp1;
  }
} {}
do_test savepoint-1.3 {

    CREATE TABLE t3(a, b, UNIQUE(a, b));
    ROLLBACK TO one;
  }
} {}
integrity_check savepoint-11.7
do_test savepoint-11.8 {
  execsql { ROLLBACK }
  db close
  execsql { PRAGMA wal_checkpoint }
  sqlite3 db test.db
  file size test.db
} {8192}

do_test savepoint-11.9 {
  execsql {
    DROP TABLE IF EXISTS t1;
    DROP TABLE IF EXISTS t2;

#
# $Id: savepoint6.test,v 1.4 2009/06/05 17:09:12 drh Exp $

set testdir [file dirname $argv0]
source $testdir/tester.tcl

proc sql {zSql} {
  if {0 && $::debug_op} {
    puts stderr "$zSql ;"
    flush stderr
  }
  uplevel db eval [list $zSql]
  #puts stderr "$zSql ;"
}

set DATABASE_SCHEMA {
    PRAGMA auto_vacuum = incremental;
    CREATE TABLE t1(x, y);

#   rollback  NAME
#   release   NAME
#
#   insert_rows XVALUES
#   delete_rows XVALUES
#
proc savepoint {zName} {
  if {$::debug_op} { puts stderr "savepoint $zName" ; flush stderr }
  catch { sql "SAVEPOINT $zName" }
  lappend ::lSavepoint [list $zName [array get ::aEntry]]
}

proc rollback {zName} {
  if {$::debug_op} { puts stderr "rollback $zName" ; flush stderr }
  catch { sql "ROLLBACK TO $zName" }
  for {set i [expr {[llength $::lSavepoint]-1}]} {$i>=0} {incr i -1} {
    set zSavepoint [lindex $::lSavepoint $i 0]
    if {$zSavepoint eq $zName} {
      unset -nocomplain ::aEntry
      array set ::aEntry [lindex $::lSavepoint $i 1]


      if {$i+1 < [llength $::lSavepoint]} {
        set ::lSavepoint [lreplace $::lSavepoint [expr $i+1] end]
      }
      break
    }
  }
}

proc release {zName} {
  if {$::debug_op} { puts stderr "release $zName" ; flush stderr }
  catch { sql "RELEASE $zName" }
  for {set i [expr {[llength $::lSavepoint]-1}]} {$i>=0} {incr i -1} {
    set zSavepoint [lindex $::lSavepoint $i 0]
    if {$zSavepoint eq $zName} {
      set ::lSavepoint [lreplace $::lSavepoint $i end]
      break
    }
  }

  if {[llength $::lSavepoint] == 0} {
    #puts stderr "-- End of transaction!!!!!!!!!!!!!"
  }
}

proc insert_rows {lX} {
  if {$::debug_op} { puts stderr "insert_rows $lX" ; flush stderr }
  foreach x $lX {
    set y [x_to_y $x]

    # Update database [db]
    sql "INSERT OR REPLACE INTO t1 VALUES($x, '$y')"

    # Update the Tcl database.
    set ::aEntry($x) $y
  }
}

proc delete_rows {lX} {
  if {$::debug_op} { puts stderr "delete_rows $lX" ; flush stderr }
  foreach x $lX {
    # Update database [db]
    sql "DELETE FROM t1 WHERE x = $x"

    # Update the Tcl database.
    unset -nocomplain ::aEntry($x)
  }

  set ret [list]
  for {set i 0} {$i<$nRes} {incr i} {
    lappend ret [expr int(rand()*$nRange)]
  }
  return $ret
} 
#-------------------------------------------------------------------------

set ::debug_op 0
proc debug_ops {} {
  set ::debug_op 1
}

proc database_op {} {
  set i [expr int(rand()*2)] 
  if {$i==0} {
    insert_rows [random_integers 100 1000]
  }
  if {$i==1} {

proc savepoint_op {} {
  set names {one two three four five}
  set cmds  {savepoint savepoint savepoint savepoint release rollback}

  set C [lindex $cmds [expr int(rand()*6)]]
  set N [lindex $names [expr int(rand()*5)]]

  #puts stderr "   $C $N ;  "
  #flush stderr

  $C $N
  return ok
}

expr srand(0)

############################################################################

# Create a test database
#
proc reset_db {} {
  catch {db close}
  forcedelete test.db
  forcedelete test.db-journal
  forcedelete test.db-wal
  forcedelete test.db-wal2
  sqlite3 db ./test.db
  set ::DB [sqlite3_connection_pointer db]
  if {[info exists ::SETUP_SQL]} {
    db eval $::SETUP_SQL
  }
}
reset_db

#     Otherwise (if not running a WAL permutation) this is a no-op.
#
#   wal_is_wal_mode
#
#     Returns true if this test should be run in WAL mode. False otherwise.
#
proc wal_is_wal_mode {} {
  expr {[permutation] eq "wal"}
  if {[permutation] eq "wal"} { return 1 }
  if {[permutation] eq "wal2"} { return 2 }
  return 0
}
proc wal_set_journal_mode {{db db}} {
  if { [wal_is_wal_mode] } {
    $db eval "PRAGMA journal_mode = WAL"
  switch -- [wal_is_wal_mode] {
    0 {
    }

    1 {
      $db eval "PRAGMA journal_mode = WAL"
    }

    2 {
      $db eval "PRAGMA journal_mode = WAL2"
    }
  }
}
proc wal_check_journal_mode {testname {db db}} {
  if { [wal_is_wal_mode] } {
    $db eval { SELECT * FROM sqlite_master }
    set expected "wal"
    if {[wal_is_wal_mode]==2} {
      set expected "wal2"
    }
    do_test $testname [list $db eval "PRAGMA main.journal_mode"] {wal}
    do_test $testname [list $db eval "PRAGMA main.journal_mode"] $expected
  }
}

proc wal_is_capable {} {
  ifcapable !wal { return 0 }
  if {[permutation]=="journaltest"} { return 0 }
  return 1

      CREATE TABLE aux.t2(a, b);
      PRAGMA main.journal_mode = WAL;
      PRAGMA aux.journal_mode = WAL;
      INSERT INTO t1 VALUES('x', 'y');
      INSERT INTO t2 VALUES('x', 'y');
    }
    lsort [array names ::T1]
  } {test.db1 test.db1-journal test.db1-wal}
  } {test.db1 test.db1-journal test.db1-wal test.db1-wal2}
  
  do_test 5.1.2 {
    lsort [array names ::T2]
  } {test.db2 test.db2-journal test.db2-wal}
  } {test.db2 test.db2-journal test.db2-wal test.db2-wal2}
  db close
  
  tvfs1 delete
  tvfs2 delete
}

#-------------------------------------------------------------------------

  5   2048    1
  6   4096    1
  7   8192    1
  8  16384    1
  9  32768    1
 10  65536    1
 11 131072    0
 11   1016    0
 12   1016    0
} {

  if {$::SQLITE_MAX_PAGE_SIZE < $pgsz} {
    set works 0
  }

  for {set pg 1} {$pg <= 3} {incr pg} {
    forcecopy testX.db test.db
    forcedelete test.db-wal
  

    # Check that the database now exists and consists of three pages. And
    # that there is no associated wal file.
    #
    do_test wal-18.2.$tn.$pg.1 { file exists test.db-wal } 0
    do_test wal-18.2.$tn.$pg.2 { file exists test.db } 1
    do_test wal-18.2.$tn.$pg.3 { file size test.db } [expr 1024*3]
  

    do_test wal-18.2.$tn.$pg.4 {

      # Create a wal file that contains a single frame (database page
      # number $pg) with the commit flag set. The frame checksum is
      # correct, but the contents of the database page are corrupt.
      #
      # The page-size in the log file header is set to $pgsz. If the

      set fd [open test.db-wal w]
      fconfigure $fd -encoding binary -translation binary
      puts -nonewline $fd $walhdr
      puts -nonewline $fd $framehdr
      puts -nonewline $fd $framebody
      close $fd
  

      file size test.db-wal
    } [wal_file_size 1 $pgsz]
  

    do_test wal-18.2.$tn.$pg.5 {
      sqlite3 db test.db
      set rc [catch { db one {PRAGMA integrity_check} } msg]
      expr { $rc!=0 || $msg!="ok" }
    } $works
  
 
    db close
  }
}

#-------------------------------------------------------------------------
# The following test - wal-19.* - fixes a bug that was present during
# development.

# 2017 September 19
#
# 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.
#
#***********************************************************************
# This file implements regression tests for SQLite library.  The
# focus of this file is testing the operation of the library in
# "PRAGMA journal_mode=WAL2" mode.
#

set testdir [file dirname $argv0]
source $testdir/tester.tcl
source $testdir/lock_common.tcl
source $testdir/malloc_common.tcl
source $testdir/wal_common.tcl

set testprefix wal2big
ifcapable !wal {finish_test ; return }

do_execsql_test 1.0 {
  CREATE TABLE t1(a, b, c);
  CREATE INDEX t1a ON t1(a);
  CREATE INDEX t1b ON t1(b);
  CREATE INDEX t1c ON t1(c);
  PRAGMA journal_mode = wal2;
  PRAGMA journal_size_limit = 10000000;

  WITH s(i) AS (
    SELECT 1 UNION ALL SELECT i+1 FROM s WHERE i<200000
  )
  INSERT INTO t1 SELECT random(), random(), random() FROM s;
} {wal2 10000000}

do_execsql_test 1.1 {
  WITH s(i) AS (
    SELECT 1 UNION ALL SELECT i+1 FROM s WHERE i<200000
  )
  INSERT INTO t1 SELECT random(), random(), random() FROM s;
}

do_test 1.2 {
  list [expr [file size test.db-wal]>10000000] \
       [expr [file size test.db-wal2]>10000000]
} {1 1}

do_test 1.3 {
  sqlite3 db2 test.db
  execsql {
    SELECT count(*) FROM t1;
    PRAGMA integrity_check;
  } db2
} {400000 ok}

do_test 1.4 {
  db2 close
  forcecopy test.db test.db2
  forcecopy test.db-wal test.db2-wal
  forcecopy test.db-wal2 test.db2-wal2
  sqlite3 db2 test.db2
  execsql {
    SELECT count(*) FROM t1;
    PRAGMA integrity_check;
  }
} {400000 ok}

finish_test

# 2018 December 6
#
# 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.
#
#***********************************************************************
#

set testdir [file dirname $argv0]
source $testdir/tester.tcl
source $testdir/lock_common.tcl
set ::testprefix wal2concurrent

ifcapable !concurrent {
  finish_test
  return
}


#-------------------------------------------------------------------------
# Warm-body test.
#
foreach tn {1 2} {
  reset_db
  sqlite3 db2 test.db
  do_execsql_test 1.0 {
    PRAGMA page_size = 1024;
    CREATE TABLE t1(x);
    CREATE TABLE t2(y);
    PRAGMA journal_size_limit = 5000;
    PRAGMA journal_mode = wal2;
  } {5000 wal2}
  
  do_execsql_test 1.1 {
    INSERT INTO t1 VALUES(1);
    BEGIN CONCURRENT;
      INSERT INTO t1 VALUES(2);
  } {}
  
  do_test 1.2 {
    execsql { 
      PRAGMA journal_size_limit = 5000;
      INSERT INTO t1 VALUES(3) 
    } db2
    catchsql { COMMIT }
  } {1 {database is locked}}
  
  do_catchsql_test 1.3 { COMMIT   } {1 {database is locked}}
  do_catchsql_test 1.4 { ROLLBACK } {0 {}}
  
  do_test 1.5 {
    list [file size test.db-wal] [file size test.db-wal2]
  } {2128 0}
  
  do_execsql_test 1.6 {
    BEGIN CONCURRENT;
      INSERT INTO t1 VALUES(2);
  } {}
  
  do_test 1.7 {
    execsql { INSERT INTO t2 VALUES(randomblob(4000)) } db2
    list [file size test.db-wal] [file size test.db-wal2]
  } {7368 0}
  
  if {$tn==1} {
    do_test 1.8 {
      execsql { 
        INSERT INTO t2 VALUES(1);
        INSERT INTO t1 VALUES(5);
      } db2
      list [file size test.db-wal] [file size test.db-wal2]
    } {7368 2128}

    do_catchsql_test 1.9  { COMMIT   } {1 {database is locked}}
    do_catchsql_test 1.10 { ROLLBACK } {0 {}}
    db close
    sqlite3 db test.db
    do_execsql_test 1.11 { SELECT * FROM t1 } {1 3 5}
    do_execsql_test 1.12 { SELECT count(*) FROM t2 } {2}
  } else {
    do_test 1.8 {
      execsql { 
        INSERT INTO t2 VALUES(1);
      } db2
      list [file size test.db-wal] [file size test.db-wal2]
    } {7368 1080}

    do_catchsql_test 1.9  { COMMIT   } {0 {}}
    db close
    sqlite3 db test.db
    do_execsql_test 1.11 { SELECT * FROM t1 } {1 3 2}
    do_execsql_test 1.12 { SELECT count(*) FROM t2 } {2}

    do_test 1.13 {
      list [file size test.db-wal] [file size test.db-wal2]
    } {7368 2128}
  }
}

do_multiclient_test tn {
  do_test 2.$tn.1 {
    sql1 {
      PRAGMA auto_vacuum = OFF;
      CREATE TABLE t1(x UNIQUE);
      CREATE TABLE t2(x UNIQUE);
      PRAGMA journal_mode = wal2;
      PRAGMA journal_size_limit = 15000;
    }
  } {wal2 15000}

  do_test 2.$tn.2 {
    sql1 {
      WITH s(i) AS (
        SELECT 1 UNION ALL SELECT i+1 FROM s WHERE i<=10
      )
      INSERT INTO t1 SELECT randomblob(800) FROM s;
    }
  } {}

  do_test 2.$tn.3 {
    sql1 { DELETE FROM t1 WHERE (rowid%4)==0 }
    list [expr [file size test.db-wal]>15000] \
         [expr [file size test.db-wal2]>15000]
  } {1 0}

  do_test 2.$tn.4 {
    sql1 { PRAGMA wal_checkpoint; }
    sql1 { 
      BEGIN CONCURRENT;
        INSERT INTO t1 VALUES(randomblob(800));
    }
  } {}

  do_test 2.$tn.5 {
    sql2 { 
      PRAGMA journal_size_limit = 15000;
      INSERT INTO t2 VALUES(randomblob(800));
      INSERT INTO t2 VALUES(randomblob(800));
      INSERT INTO t2 VALUES(randomblob(800));
      INSERT INTO t2 VALUES(randomblob(800));
      INSERT INTO t2 VALUES(randomblob(800));
      DELETE FROM t2;
    }
    list [expr [file size test.db-wal]>15000] \
         [expr [file size test.db-wal2]>15000]
  } {1 1}

  do_test 2.$tn.6 {
    sql1 { 
        INSERT INTO t1 VALUES(randomblob(800));
      COMMIT;
      PRAGMA integrity_check;
    }
  } {ok}
}



finish_test

# 2010 May 03
#
# 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.
#
#***********************************************************************
# This file implements regression tests for SQLite library.  The
# focus of this file is testing the operation of the library in
# "PRAGMA journal_mode=WAL" mode.
#

set testdir [file dirname $argv0]
source $testdir/tester.tcl
source $testdir/malloc_common.tcl
source $testdir/lock_common.tcl

ifcapable !wal {finish_test ; return }
set testprefix wal2fault

do_execsql_test 1.0 {
  CREATE TABLE t1(x,y);
  PRAGMA journal_mode = wal2;
  WITH s(i) AS ( SELECT 100 UNION ALL SELECT i-1 FROM s WHERE (i-1)>0 )
  INSERT INTO t1 SELECT i, randomblob(i) FROM s;
  WITH s(i) AS ( SELECT 100 UNION ALL SELECT i-1 FROM s WHERE (i-1)>0 )
  INSERT INTO t1 SELECT i, randomblob(i) FROM s;
} {wal2}

do_test 1.1 {
  expr [file size test.db-wal]>10000
} {1}
faultsim_save_and_close

do_faultsim_test 1 -prep {
  faultsim_restore_and_reopen
  execsql {
    PRAGMA journal_size_limit = 10000;
    SELECT count(*) FROM sqlite_master;
  }
} -body {
  execsql {
    INSERT INTO t1 VALUES(1, 2);
  }
} -test {
  faultsim_test_result {0 {}}
}

finish_test