SQLite

typedef struct PagerSavepoint PagerSavepoint;
struct PagerSavepoint {
  i64 iOffset;                 /* Starting offset in main journal */
  i64 iHdrOffset;              /* See above */
  Bitvec *pInSavepoint;        /* Set of pages in this savepoint */
  Pgno nOrig;                  /* Original number of pages in file */
  Pgno iSubRec;                /* Index of first record in sub-journal */
  u32 iFrame;                  /* Last frame in WAL when savepoint opened */
  u32 aWalData[WAL_SAVEPOINT_NDATA];        /* WAL savepoint context */
};

/*
** A open page cache is an instance of the following structure.
**
** errCode
**

  ** will be skipped.  Out-of-range pages are also skipped.
  */
  if( pSavepoint ){
    u32 ii;            /* Loop counter */
    i64 offset = pSavepoint->iSubRec*(4+pPager->pageSize);

    if( pagerUseWal(pPager) ){
      rc = sqlite3WalSavepointUndo(pPager->pWal, pSavepoint->iFrame);
      rc = sqlite3WalSavepointUndo(pPager->pWal, pSavepoint->aWalData);
    }
    for(ii=pSavepoint->iSubRec; rc==SQLITE_OK && ii<pPager->nSubRec; ii++){
      assert( offset==ii*(4+pPager->pageSize) );
      rc = pager_playback_one_page(pPager, &offset, pDone, 0, 1);
    }
    assert( rc!=SQLITE_DONE );
  }

      }
      aNew[ii].iSubRec = pPager->nSubRec;
      aNew[ii].pInSavepoint = sqlite3BitvecCreate(nPage);
      if( !aNew[ii].pInSavepoint ){
        return SQLITE_NOMEM;
      }
      if( pagerUseWal(pPager) ){
        aNew[ii].iFrame = sqlite3WalSavepoint(pPager->pWal);
        sqlite3WalSavepoint(pPager->pWal, aNew[ii].aWalData);
      }
    }

    /* Open the sub-journal, if it is not already opened. */
    rc = openSubJournal(pPager);
    assertTruncateConstraint(pPager);
  }

/*
** 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.
*/
struct WalIndexHdr {
  u32 iChange;      /* Counter incremented each transaction */
  u16 bigEndCksum;  /* True if checksums in WAL are big-endian */
  u16 szPage;       /* Database page size in bytes */
  u32 mxFrame;      /* Index of last valid frame in the WAL */
  u32 nPage;        /* Size of database in pages */
  u32 aSalt[2];     /* Salt-1 and salt-2 values copied from WAL header */
  u32 aCksum[2];    /* Checksum over all prior fields */
  u32 iChange;                    /* Counter incremented each transaction */
  u16 bigEndCksum;                /* True if checksums in WAL are big-endian */
  u16 szPage;                     /* Database page size in bytes */
  u32 mxFrame;                    /* Index of last valid frame in the 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 */
};

/* A block of WALINDEX_LOCK_RESERVED bytes beginning at
** WALINDEX_LOCK_OFFSET is reserved for locks. Since some systems
** only support mandatory file-locks, we do not read or write data
** from the region of the file on which locks are applied.
*/

  Wal *pWal,                      /* The write-ahead log */
  u32 iPage,                      /* Database page number for frame */
  u32 nTruncate,                  /* New db size (or 0 for non-commit frames) */
  u8 *aData,                      /* Pointer to page data */
  u8 *aFrame                      /* OUT: Write encoded frame here */
){
  int nativeCksum;                /* True for native byte-order checksums */
  u32 aCksum[2];
  u32 *aCksum = pWal->hdr.aFrameCksum;
  assert( WAL_FRAME_HDRSIZE==24 );
  sqlite3Put4byte(&aFrame[0], iPage);
  sqlite3Put4byte(&aFrame[4], nTruncate);
  memcpy(&aFrame[8], pWal->hdr.aSalt, 8);

  nativeCksum = (pWal->hdr.bigEndCksum==SQLITE_BIGENDIAN);
  walChecksumBytes(nativeCksum, aFrame, 16, 0, aCksum);
  walChecksumBytes(nativeCksum, aFrame, 8, aCksum, aCksum);
  walChecksumBytes(nativeCksum, aData, pWal->szPage, aCksum, aCksum);

  sqlite3Put4byte(&aFrame[16], aCksum[0]);
  sqlite3Put4byte(&aFrame[20], aCksum[1]);
}

/*
** Check to see if the frame with header in aFrame[] and content
** in aData[] is valid.  If it is a valid frame, fill *piPage and
** *pnTruncate and return true.  Return if the frame is not valid.
*/
static int walDecodeFrame(
  Wal *pWal,                      /* The write-ahead log */
  u32 *piPage,                    /* OUT: Database page number for frame */
  u32 *pnTruncate,                /* OUT: New db size (or 0 if not commit) */
  u8 *aData,                      /* Pointer to page data (for checksum) */
  u8 *aFrame                      /* Frame data */
){
  int nativeCksum;                /* True for native byte-order checksums */
  u32 *aCksum = pWal->hdr.aFrameCksum;
  u32 pgno;                       /* Page number of the frame */
  u32 aCksum[2];
  assert( WAL_FRAME_HDRSIZE==24 );

  /* A frame is only valid if the salt values in the frame-header
  ** match the salt values in the wal-header. 
  */
  if( memcmp(&pWal->hdr.aSalt, &aFrame[8], 8)!=0 ){
    return 0;

  }

  /* A frame is only valid if a checksum of the first 16 bytes
  ** of the frame-header, and the frame-data matches
  ** the checksum in the last 8 bytes of the frame-header.
  */
  nativeCksum = (pWal->hdr.bigEndCksum==SQLITE_BIGENDIAN);
  walChecksumBytes(nativeCksum, aFrame, 16, 0, aCksum);
  walChecksumBytes(nativeCksum, aFrame, 8, aCksum, aCksum);
  walChecksumBytes(nativeCksum, aData, pWal->szPage, aCksum, aCksum);
  if( aCksum[0]!=sqlite3Get4byte(&aFrame[16]) 
   || aCksum[1]!=sqlite3Get4byte(&aFrame[20]) 
  ){
    /* Checksum failed. */
    return 0;
  }

/*
** Recover the wal-index by reading the write-ahead log file. 
** The caller must hold RECOVER lock on the wal-index file.
*/
static int walIndexRecover(Wal *pWal){
  int rc;                         /* Return Code */
  i64 nSize;                      /* Size of log file */
  WalIndexHdr hdr;                /* Recovered wal-index header */
  u32 aFrameCksum[2] = {0, 0};

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

  rc = sqlite3OsFileSize(pWal->pWalFd, &nSize);
  if( rc!=SQLITE_OK ){
    return rc;
  }

  if( nSize>WAL_HDRSIZE ){

    if( (magic&0xFFFFFFFE)!=WAL_MAGIC 
     || szPage&(szPage-1) 
     || szPage>SQLITE_MAX_PAGE_SIZE 
     || szPage<512 
    ){
      goto finished;
    }
    hdr.bigEndCksum = pWal->hdr.bigEndCksum = (magic&0x00000001);
    pWal->hdr.bigEndCksum = (magic&0x00000001);
    pWal->szPage = szPage;
    pWal->nCkpt = sqlite3Get4byte(&aBuf[12]);
    memcpy(&pWal->hdr.aSalt, &aBuf[16], 8);
    walChecksumBytes(pWal->hdr.bigEndCksum==SQLITE_BIGENDIAN, 
        aBuf, WAL_HDRSIZE, 0, pWal->hdr.aFrameCksum
    );

    /* Malloc a buffer to read frames into. */
    szFrame = szPage + WAL_FRAME_HDRSIZE;
    aFrame = (u8 *)sqlite3_malloc(szFrame);
    if( !aFrame ){
      return SQLITE_NOMEM;
    }

      isValid = walDecodeFrame(pWal, &pgno, &nTruncate, aData, aFrame);
      if( !isValid ) break;
      rc = walIndexAppend(pWal, ++iFrame, pgno);
      if( rc!=SQLITE_OK ) break;

      /* If nTruncate is non-zero, this is a commit record. */
      if( nTruncate ){
        hdr.mxFrame = iFrame;
        hdr.nPage = nTruncate;
        hdr.szPage = szPage;
        pWal->hdr.mxFrame = iFrame;
        pWal->hdr.nPage = nTruncate;
        pWal->hdr.szPage = szPage;
        aFrameCksum[0] = pWal->hdr.aFrameCksum[0];
        aFrameCksum[1] = pWal->hdr.aFrameCksum[1];
      }
    }

    sqlite3_free(aFrame);
  }else{
    memset(&hdr, 0, sizeof(hdr));
  }

finished:
  if( rc==SQLITE_OK && hdr.mxFrame==0 ){
  if( rc==SQLITE_OK && pWal->hdr.mxFrame==0 ){
    rc = walIndexRemap(pWal, WALINDEX_MMAP_INCREMENT);
  }
  if( rc==SQLITE_OK ){
    memcpy(&pWal->hdr, &hdr, sizeof(hdr));
    pWal->hdr.aFrameCksum[0] = aFrameCksum[0];
    pWal->hdr.aFrameCksum[1] = aFrameCksum[1];
    walIndexWriteHdr(pWal);
  }
  return rc;
}

/*
** Close an open wal-index.

      }
    }
    walIndexUnmap(pWal);
  }
  return rc;
}

/* Return an integer that records the current (uncommitted) write
** position in the WAL
/* 
** 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()).
*/
u32 sqlite3WalSavepoint(Wal *pWal){
void sqlite3WalSavepoint(Wal *pWal, u32 *aWalData){
  assert( pWal->lockState==SQLITE_SHM_WRITE );
  return pWal->hdr.mxFrame;
  aWalData[0] = pWal->hdr.mxFrame;
  aWalData[1] = pWal->hdr.aFrameCksum[0];
  aWalData[2] = pWal->hdr.aFrameCksum[1];
}

/* 
/* Move the write position of the WAL back to iFrame.  Called in
** Move the write position of the WAL back to the point identified by
** response to a ROLLBACK TO command.
** 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 iFrame){
int sqlite3WalSavepointUndo(Wal *pWal, u32 *aWalData){
  int rc = SQLITE_OK;
  assert( pWal->lockState==SQLITE_SHM_WRITE );

  assert( iFrame<=pWal->hdr.mxFrame );
  if( iFrame<pWal->hdr.mxFrame ){
  assert( aWalData[0]<=pWal->hdr.mxFrame );
  if( aWalData[0]<pWal->hdr.mxFrame ){
    rc = walIndexMap(pWal, walMappingSize(pWal->hdr.mxFrame));
    pWal->hdr.mxFrame = iFrame;
    pWal->hdr.mxFrame = aWalData[0];
    pWal->hdr.aFrameCksum[0] = aWalData[1];
    pWal->hdr.aFrameCksum[1] = aWalData[2];
    if( rc==SQLITE_OK ){
      walCleanupHash(pWal);
      walIndexUnmap(pWal);
    }
  }
  return rc;
}

    pWal->hdr.bigEndCksum = SQLITE_BIGENDIAN;
    sqlite3Put4byte(&aWalHdr[12], pWal->nCkpt);
    memcpy(&aWalHdr[16], pWal->hdr.aSalt, 8);
    rc = sqlite3OsWrite(pWal->pWalFd, aWalHdr, sizeof(aWalHdr), 0);
    if( rc!=SQLITE_OK ){
      return rc;
    }
    walChecksumBytes(1, aWalHdr, sizeof(aWalHdr), 0, pWal->hdr.aFrameCksum);
  }
  assert( pWal->szPage==szPage );

    /* Write the log file. */
  for(p=pList; p; p=p->pDirty){
    u32 nDbsize;                  /* Db-size field for frame header */
    i64 iOffset;                  /* Write offset in log file */

# define sqlite3WalClose(w,x,y,z)          0
# define sqlite3WalOpenSnapshot(y,z)       0
# define sqlite3WalCloseSnapshot(z) 
# define sqlite3WalRead(v,w,x,y,z)         0
# define sqlite3WalDbsize(y,z)
# define sqlite3WalWriteLock(y,z)          0
# define sqlite3WalUndo(x,y,z)             0
# define sqlite3WalSavepoint(z)            0
# define sqlite3WalSavepoint(y,z)
# define sqlite3WalSavepointUndo(y,z)      0
# define sqlite3WalFrames(u,v,w,x,y,z)     0
# define sqlite3WalCheckpoint(u,v,w,x,y,z) 0
# define sqlite3WalCallback(z)             0
#else

#define WAL_SAVEPOINT_NDATA 3

/* 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 sqlite3WalWriteLock(Wal *pWal, int op);

/* Undo any frames written (but not committed) to the log */
int sqlite3WalUndo(Wal *pWal, int (*xUndo)(void *, Pgno), void *pUndoCtx);

/* Return an integer that records the current (uncommitted) write
** position in the WAL */
u32 sqlite3WalSavepoint(Wal *pWal);
void sqlite3WalSavepoint(Wal *pWal, u32 *aWalData);

/* Move the write position of the WAL back to iFrame.  Called in
** response to a ROLLBACK TO command. */
int sqlite3WalSavepointUndo(Wal *pWal, u32 iFrame);
int sqlite3WalSavepointUndo(Wal *pWal, u32 *aWalData);

/* Write a frame or frames to the log. */
int sqlite3WalFrames(Wal *pWal, int, PgHdr *, Pgno, int, int);

/* Copy pages from the log to the database file */ 
int sqlite3WalCheckpoint(
  Wal *pWal,                      /* Write-ahead log connection */

      # be ignored.
      #
      set walhdr [binary format IIIIII 931071618 3007000 $pgsz 1234 22 23]
      set framebody [randomblob $pgsz]
      set framehdr  [binary format IIII $pg 5 22 23]
      set c1 0
      set c2 0
      logcksum c1 c2 $framehdr
      logcksum c1 c2 $walhdr
      logcksum c1 c2 [string range $framehdr 0 7]
      logcksum c1 c2 $framebody
      set framehdr [binary format IIIIII $pg 5 22 23 $c1 $c2]

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

set testdir [file dirname $argv0]
source $testdir/tester.tcl
source $testdir/lock_common.tcl
ifcapable !wal {finish_test ; return }

proc set_tvfs_hdr {file args} {

  # Set $nHdr to the number of bytes in the wal-index header:
  set nHdr 80
  set nInt [expr {$nHdr/4}]

  if {[llength $args]>1} {
    return -code error {wrong # args: should be "set_tvfs_hdr fileName ?val?"}
  }

  set blob [tvfs shm $file]
  if {[llength $args]} {
    set blob [
    set blob [binary format i16a* [lindex $args 0] [string range $blob 64 end]]
      binary format i${nInt}a* [lindex $args 0] [string range $blob $nHdr end]
    ]
    tvfs shm $file $blob
  }

  binary scan $blob i16 ints
  binary scan $blob i${nInt} ints
  return $ints
}

proc incr_tvfs_hdr {file idx incrval} {
  set ints [set_tvfs_hdr $file]
  set v [lindex $ints $idx]
  incr v $incrval

# File $filename must be a WAL file on disk. Check that the checksum of frame
# $iFrame in the file is correct when interpreting data as $endian-endian
# integers ($endian must be either "big" or "little"). If the checksum looks
# correct, return 1. Otherwise 0.
#
proc log_checksum_verify {filename iFrame endian} {
  set data [readfile $filename]
  set c1 0
  set c2 0
  
  binary scan [string range $data 8 11] I pgsz

  foreach {offset c1 c2} [log_checksum_calc $data $iFrame $endian] {}

  set n [log_file_size [expr $iFrame-1] $pgsz]
  binary scan [string range $data [expr $n+16] [expr $n+23]] II expect1 expect2
  binary scan [string range $data $offset [expr $offset+7]] II expect1 expect2
  log_cksum $endian c1 c2 [string range $data $n [expr $n+15]]
  log_cksum $endian c1 c2 [string range $data [expr $n+24] [expr $n+24+$pgsz-1]]

  set expect1 [expr $expect1&0xFFFFFFFF]
  set expect2 [expr $expect2&0xFFFFFFFF]

  expr {$c1==$expect1 && $c2==$expect2}
}

#
# File $filename must be a WAL file on disk. Compute the checksum for frame
# $iFrame in the file by interpreting data as $endian-endian integers 
# ($endian must be either "big" or "little"). Then write the computed 
# checksum into the file.
#
proc log_checksum_write {filename iFrame endian} {
  set data [readfile $filename]
  set c1 0
  set c2 0
  

  binary scan [string range $data 8 11] I pgsz

  foreach {offset c1 c2} [log_checksum_calc $data $iFrame $endian] {}
  set n [log_file_size [expr $iFrame-1] $pgsz]
  log_cksum $endian c1 c2 [string range $data $n [expr $n+15]]
  log_cksum $endian c1 c2 [string range $data [expr $n+24] [expr $n+24+$pgsz-1]]

  set bin [binary format II $c1 $c2]
  set fd [open $filename r+]
  fconfigure $fd -encoding binary
  fconfigure $fd -translation binary
  seek $fd [expr $n+16]
  seek $fd $offset
  puts -nonewline $fd $bin
  close $fd
}

proc log_checksum_calc {data iFrame endian} {
  
  binary scan [string range $data 8 11] I pgsz
  if {$iFrame > 1} {
    set n [log_file_size [expr $iFrame-2] $pgsz]
    binary scan [string range $data [expr $n+16] [expr $n+23]] II c1 c2
  } else {
    set c1 0
    set c2 0
    log_cksum $endian c1 c2 [string range $data 0 23]
  }

  set n [log_file_size [expr $iFrame-1] $pgsz]
  log_cksum $endian c1 c2 [string range $data $n [expr $n+7]]
  log_cksum $endian c1 c2 [string range $data [expr $n+24] [expr $n+24+$pgsz-1]]

  list [expr $n+16] $c1 $c2
}

#
# File $filename must be a WAL file on disk. Set the 'magic' field of the
# WAL header to indicate that checksums are $endian-endian ($endian must be
# either "big" or "little").
#
proc log_checksum_writemagic {filename endian} {

      log_checksum_verify test2.db-wal $f $native
    } 1
  }

  # Replace all checksums in the current WAL file with $endian versions.
  # Then check that it is still possible to recover and read the database.
  #
  log_checksum_writemagic test2.db-wal $endian
  for {set f 1} {$f <= 6} {incr f} {
    do_test walcksum-1.$endian.3.$f {
      log_checksum_write test2.db-wal $f $endian
      log_checksum_verify test2.db-wal $f $endian
    } {1}
  }
  do_test walcksum-1.$endian.4.1 {
    log_checksum_writemagic test2.db-wal $endian
    file copy -force test2.db test.db
    file copy -force test2.db-wal test.db-wal
    sqlite3 db test.db
    execsql { SELECT a FROM t1 }
  } {1 2 3 5 8 13 21}

  # Following recovery, any frames written to the log should use the same 

    log_checksum_verify test.db-wal 1 $native
  } {1}
  do_test walcksum-1.$endian.8.2 {
    log_checksum_verify test.db-wal 2 $native
  } {1}
  do_test walcksum-1.$endian.8.3 {
    log_checksum_verify test.db-wal 3 $native
  } [expr {$native == $endian}]
  } {0}

  do_test walcksum-1.$endian.9 {
    execsql { 
      PRAGMA integrity_check;
      SELECT a FROM t1;
    } db2
  } {ok 1 2 3 5 8 13 21 34 55 89}

  catch { db close }
  catch { db2 close }
}

do_test walcksum-2.1 {
  file delete -force test.db test.db-wal test.db-journal
  sqlite3 db test.db
  execsql {
    PRAGMA synchronous = NORMAL;
    PRAGMA page_size = 1024;
    PRAGMA journal_mode = WAL;
    PRAGMA cache_size = 10;
    CREATE TABLE t1(x PRIMARY KEY);
    PRAGMA wal_checkpoint;
    INSERT INTO t1 VALUES(randomblob(800));
    BEGIN;
      INSERT INTO t1 SELECT randomblob(800) FROM t1;   /*   2 */
      INSERT INTO t1 SELECT randomblob(800) FROM t1;   /*   4 */
      INSERT INTO t1 SELECT randomblob(800) FROM t1;   /*   8 */
      INSERT INTO t1 SELECT randomblob(800) FROM t1;   /*  16 */
      SAVEPOINT one;
        INSERT INTO t1 SELECT randomblob(800) FROM t1;   /*  32 */
        INSERT INTO t1 SELECT randomblob(800) FROM t1;   /*  64 */
        INSERT INTO t1 SELECT randomblob(800) FROM t1;   /* 128 */
        INSERT INTO t1 SELECT randomblob(800) FROM t1;   /* 256 */
      ROLLBACK TO one;
      INSERT INTO t1 SELECT randomblob(800) FROM t1;   /*  32 */
      INSERT INTO t1 SELECT randomblob(800) FROM t1;   /*  64 */
      INSERT INTO t1 SELECT randomblob(800) FROM t1;   /* 128 */
      INSERT INTO t1 SELECT randomblob(800) FROM t1;   /* 256 */
    COMMIT;
  }

  file copy -force test.db test2.db
  file copy -force test.db-wal test2.db-wal

  sqlite3 db2 test2.db
  execsql {
    PRAGMA integrity_check;
    SELECT count(*) FROM t1;
  } db2
} {ok 256}
catch { db close }
catch { db2 close }

finish_test