SQLite

/*
** 2004 April 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.
**
*************************************************************************
** $Id: btree.c,v 1.353 2007/04/07 15:03:17 danielk1977 Exp $
**
** This file implements a external (disk-based) database using BTrees.
** For a detailed discussion of BTrees, refer to
**
**     Donald E. Knuth, THE ART OF COMPUTER PROGRAMMING, Volume 3:
**     "Sorting And Searching", pages 473-480. Addison-Wesley
**     Publishing Company, Reading, Massachusetts.

    */
    rc = allocateBtreePage(pBt, &pPageMove, &pgnoMove, pgnoRoot, 1);
    if( rc!=SQLITE_OK ){
      return rc;
    }

    if( pgnoMove!=pgnoRoot ){
      /* pgnoRoot is the page that will be used for the root-page of
      ** the new table (assuming an error did not occur). But we were
      ** allocated pgnoMove. If required (i.e. if it was not allocated
      ** by extending the file), the current page at position pgnoMove
      ** is already journaled.
      */
      u8 eType;
      Pgno iPtrPage;

      releasePage(pPageMove);

      /* Move the page currently at pgnoRoot to pgnoMove. */
      rc = getPage(pBt, pgnoRoot, &pRoot, 0);
      if( rc!=SQLITE_OK ){
        return rc;
      }
      rc = ptrmapGet(pBt, pgnoRoot, &eType, &iPtrPage);
      if( rc!=SQLITE_OK || eType==PTRMAP_ROOTPAGE || eType==PTRMAP_FREEPAGE ){
        releasePage(pRoot);
        return rc;
      }
      assert( eType!=PTRMAP_ROOTPAGE );
      assert( eType!=PTRMAP_FREEPAGE );
      rc = sqlite3PagerWrite(pRoot->pDbPage);
      if( rc!=SQLITE_OK ){
        releasePage(pRoot);
        return rc;
      }
      rc = relocatePage(pBt, pRoot, eType, iPtrPage, pgnoMove);
      releasePage(pRoot);

      /* Obtain the page at pgnoRoot */
      if( rc!=SQLITE_OK ){
        return rc;
      }
      rc = getPage(pBt, pgnoRoot, &pRoot, 0);
      if( rc!=SQLITE_OK ){
        return rc;
      }

** The pager is used to access a database disk file.  It implements
** atomic commit and rollback through the use of a journal file that
** is separate from the database file.  The pager also implements file
** locking to prevent two processes from writing the same database
** file simultaneously, or one process from reading the database while
** another is writing.
**
** @(#) $Id: pager.c,v 1.325 2007/04/07 15:03:17 danielk1977 Exp $
*/
#ifndef SQLITE_OMIT_DISKIO
#include "sqliteInt.h"
#include "os.h"
#include "pager.h"
#include <assert.h>
#include <string.h>

typedef struct PgHdr PgHdr;
struct PgHdr {
  Pager *pPager;                 /* The pager to which this page belongs */
  Pgno pgno;                     /* The page number for this page */
  PgHdr *pNextHash, *pPrevHash;  /* Hash collision chain for PgHdr.pgno */
  PgHdr *pNextFree, *pPrevFree;  /* Freelist of pages where nRef==0 */
  PgHdr *pNextAll;               /* A list of all pages */

  u8 inJournal;                  /* TRUE if has been written to journal */

  u8 dirty;                      /* TRUE if we need to write back changes */
  u8 needSync;                   /* Sync journal before writing this page */
  u8 alwaysRollback;             /* Disable DontRollback() for this page */
  short int nRef;                /* Number of users of this page */
  PgHdr *pDirty, *pPrevDirty;    /* Dirty pages */
  u32 notUsed;                   /* Buffer space */
#ifdef SQLITE_CHECK_PAGES

** even for disk-based databases.  Splitting it out saves 8 bytes.  This
** is only a savings of 0.8% but those percentages add up.
*/
typedef struct PgHistory PgHistory;
struct PgHistory {
  u8 *pOrig;     /* Original page text.  Restore to this on a full rollback */
  u8 *pStmt;     /* Text as it was at the beginning of the current statement */
  PgHdr *pNextStmt, *pPrevStmt;  /* List of pages in the statement journal */
  u8 inStmt;                     /* TRUE if in the statement subjournal */
};

/*
** A macro used for invoking the codec if there is one
*/
#ifdef SQLITE_HAS_CODEC
# define CODEC1(P,D,N,X) if( P->xCodec!=0 ){ P->xCodec(P->pCodecArg,D,N,X); }

    cnt++;   /* Something to set a breakpoint on */
  }
# define REFINFO(X)  pager_refinfo(X)
#else
# define REFINFO(X)
#endif

/*
** Return true if page *pPg has already been written to the statement
** journal (or statement snapshot has been created, if *pPg is part
** of an in-memory database).
*/
static int pageInStatement(PgHdr *pPg){
  Pager *pPager = pPg->pPager;
  if( MEMDB ){
    return PGHDR_TO_HIST(pPg, pPager)->inStmt;
  }else{
    Pgno pgno = pPg->pgno;
    u8 *a = pPager->aInStmt;
    return (a && (int)pgno<=pPager->stmtSize && (a[pgno/8] & (1<<(pgno&7))));
  }
}

/*
** Change the size of the pager hash table to N.  N must be a power
** of two.
*/
static void pager_resize_hash_table(Pager *pPager, int N){
  PgHdr **aHash, *pPg;

** The Pager keeps a separate list of pages that are currently in
** the statement journal.  This helps the sqlite3PagerStmtCommit()
** routine run MUCH faster for the common case where there are many
** pages in memory but only a few are in the statement journal.
*/
static void page_add_to_stmt_list(PgHdr *pPg){
  Pager *pPager = pPg->pPager;
  PgHistory *pHist = PGHDR_TO_HIST(pPg, pPager);
  assert( MEMDB );
  if( !pHist->inStmt ){
    assert( pHist->pPrevStmt==0 && pHist->pNextStmt==0 );

    if( pPager->pStmt ){
      PGHDR_TO_HIST(pPager->pStmt, pPager)->pPrevStmt = pPg;
    }
    pHist->pNextStmt = pPager->pStmt;
    pPager->pStmt = pPg;
    pHist->inStmt = 1;
  }
















}

/*
** Find a page in the hash table given its page number.  Return
** a pointer to the page or NULL if not found.
*/
static PgHdr *pager_lookup(Pager *pPager, Pgno pgno){

        ** Remove it from this list before freeing.
        **
        ** Todo: Check the Pager.pStmt list to make sure this is Ok. It 
        ** probably is though.
        */
        PgHdr *pTmp;
        assert( pPg );

        if( pPg==pPager->pAll ){
           pPager->pAll = pPg->pNextAll;
        }else{
          for( pTmp=pPager->pAll; pTmp->pNextAll!=pPg; pTmp=pTmp->pNextAll ){}
          pTmp->pNextAll = pPg->pNextAll;
        }
        nReleased += sqliteAllocSize(pPg);

    TEST_INCR(pPager->nMiss);
    rc = pagerAllocatePage(pPager, &pPg);
    if( rc!=SQLITE_OK ){
      return rc;
    }

    pPg->pgno = pgno;
    assert( !MEMDB || pgno>pPager->stmtSize );
    if( pPager->aInJournal && (int)pgno<=pPager->origDbSize ){
      sqlite3CheckMemory(pPager->aInJournal, pgno/8);
      assert( pPager->journalOpen );
      pPg->inJournal = (pPager->aInJournal[pgno/8] & (1<<(pgno&7)))!=0;
      pPg->needSync = 0;
    }else{
      pPg->inJournal = 0;
      pPg->needSync = 0;
    }






    makeClean(pPg);
    pPg->nRef = 1;
    REFINFO(pPg);

    pPager->nRef++;
    if( pPager->nExtra>0 ){
      memset(PGHDR_TO_EXTRA(pPg, pPager), 0, pPager->nExtra);

  CHECK_PAGE(pPg);

  /* Mark the page as dirty.  If the page has already been written
  ** to the journal then we can return right away.
  */
  makeDirty(pPg);
  if( pPg->inJournal && (pageInStatement(pPg) || pPager->stmtInUse==0) ){
    pPager->dirtyCache = 1;
  }else{

    /* If we get this far, it means that the page needs to be
    ** written to the transaction journal or the ckeckpoint journal
    ** or both.
    **

          pPager->nRec++;
          assert( pPager->aInJournal!=0 );
          pPager->aInJournal[pPg->pgno/8] |= 1<<(pPg->pgno&7);
          pPg->needSync = !pPager->noSync;
          if( pPager->stmtInUse ){
            pPager->aInStmt[pPg->pgno/8] |= 1<<(pPg->pgno&7);

          }
        }
      }else{
        pPg->needSync = !pPager->journalStarted && !pPager->noSync;
        PAGERTRACE4("APPEND %d page %d needSync=%d\n",
                PAGERID(pPager), pPg->pgno, pPg->needSync);
      }
      if( pPg->needSync ){
        pPager->needSync = 1;
      }
      pPg->inJournal = 1;
    }
  
    /* If the statement journal is open and the page is not in it,
    ** then write the current page to the statement journal.  Note that
    ** the statement journal format differs from the standard journal format
    ** in that it omits the checksums and the header.
    */
    if( pPager->stmtInUse 
     && !pageInStatement(pPg) 
     && (int)pPg->pgno<=pPager->stmtSize 
    ){
      assert( pPg->inJournal || (int)pPg->pgno>pPager->origDbSize );
      if( MEMDB ){
        PgHistory *pHist = PGHDR_TO_HIST(pPg, pPager);
        assert( pHist->pStmt==0 );
        pHist->pStmt = sqliteMallocRaw( pPager->pageSize );
        if( pHist->pStmt ){
          memcpy(pHist->pStmt, PGHDR_TO_DATA(pPg), pPager->pageSize);
        }
        PAGERTRACE3("STMT-JOURNAL %d page %d\n", PAGERID(pPager), pPg->pgno);
        page_add_to_stmt_list(pPg);
      }else{
        char *pData2 = CODEC2(pPager, pData, pPg->pgno, 7)-4;
        put32bits(pData2, pPg->pgno);
        rc = sqlite3OsWrite(pPager->stfd, pData2, pPager->pageSize+4);
        PAGERTRACE3("STMT-JOURNAL %d page %d\n", PAGERID(pPager), pPg->pgno);
        if( rc!=SQLITE_OK ){
          return rc;
        }
        pPager->stmtNRec++;
        assert( pPager->aInStmt!=0 );
        pPager->aInStmt[pPg->pgno/8] |= 1<<(pPg->pgno&7);
      }

    }
  }

  /* Update the database size and return.
  */
  assert( pPager->state>=PAGER_SHARED );
  if( pPager->dbSize<(int)pPg->pgno ){

  if( pPg->alwaysRollback || pPager->alwaysRollback || MEMDB ) return;
  if( !pPg->inJournal && (int)pPg->pgno <= pPager->origDbSize ){
    assert( pPager->aInJournal!=0 );
    pPager->aInJournal[pPg->pgno/8] |= 1<<(pPg->pgno&7);
    pPg->inJournal = 1;
    if( pPager->stmtInUse ){
      pPager->aInStmt[pPg->pgno/8] |= 1<<(pPg->pgno&7);

    }
    PAGERTRACE3("DONT_ROLLBACK page %d of %d\n", pPg->pgno, PAGERID(pPager));
    IOTRACE(("GARBAGE %p %d\n", pPager, pPg->pgno))
  }
  if( pPager->stmtInUse 
   && !pageInStatement(pPg) 
   && (int)pPg->pgno<=pPager->stmtSize 
  ){
    assert( pPg->inJournal || (int)pPg->pgno>pPager->origDbSize );
    assert( pPager->aInStmt!=0 );
    pPager->aInStmt[pPg->pgno/8] |= 1<<(pPg->pgno&7);

  }
}


/*
** This routine is called to increment the database file change-counter,
** stored at byte 24 of the pager file.

  if( pPager->state<PAGER_RESERVED ){
    return SQLITE_ERROR;
  }
  PAGERTRACE2("COMMIT %d\n", PAGERID(pPager));
  if( MEMDB ){
    pPg = pager_get_all_dirty_pages(pPager);
    while( pPg ){
      PgHistory *pHist = PGHDR_TO_HIST(pPg, pPager);
      clearHistory(pHist);
      pPg->dirty = 0;
      pPg->inJournal = 0;
      pHist->inStmt = 0;
      pPg->needSync = 0;
      pHist->pPrevStmt = pHist->pNextStmt = 0;
      pPg = pPg->pDirty;
    }
    pPager->pDirty = 0;
#ifndef NDEBUG
    for(pPg=pPager->pAll; pPg; pPg=pPg->pNextAll){
      PgHistory *pHist = PGHDR_TO_HIST(pPg, pPager);
      assert( !pPg->alwaysRollback );

        PAGERTRACE3("ROLLBACK-PAGE %d of %d\n", p->pgno, PAGERID(pPager));
      }else{
        PAGERTRACE3("PAGE %d is clean on %d\n", p->pgno, PAGERID(pPager));
      }
      clearHistory(pHist);
      p->dirty = 0;
      p->inJournal = 0;
      pHist->inStmt = 0;
      pHist->pPrevStmt = pHist->pNextStmt = 0;
      if( pPager->xReiniter ){
        pPager->xReiniter(p, pPager->pageSize);
      }
    }
    pPager->pDirty = 0;
    pPager->pStmt = 0;
    pPager->dbSize = pPager->origDbSize;

    PgHdr *pPg, *pNext;
    PAGERTRACE2("STMT-COMMIT %d\n", PAGERID(pPager));
    if( !MEMDB ){
      sqlite3OsSeek(pPager->stfd, 0);
      /* sqlite3OsTruncate(pPager->stfd, 0); */
      sqliteFree( pPager->aInStmt );
      pPager->aInStmt = 0;
    }else{
      for(pPg=pPager->pStmt; pPg; pPg=pNext){
        PgHistory *pHist = PGHDR_TO_HIST(pPg, pPager);
        pNext = pHist->pNextStmt;
        assert( pHist->inStmt );
        pHist->inStmt = 0;
        pHist->pPrevStmt = pHist->pNextStmt = 0;


        sqliteFree(pHist->pStmt);
        pHist->pStmt = 0;
      }
    }
    pPager->stmtNRec = 0;
    pPager->stmtInUse = 0;
    pPager->pStmt = 0;

*/
int sqlite3PagerStmtRollback(Pager *pPager){
  int rc;
  if( pPager->stmtInUse ){
    PAGERTRACE2("STMT-ROLLBACK %d\n", PAGERID(pPager));
    if( MEMDB ){
      PgHdr *pPg;
      PgHistory *pHist;
      for(pPg=pPager->pStmt; pPg; pPg=pHist->pNextStmt){
        pHist = PGHDR_TO_HIST(pPg, pPager);
        if( pHist->pStmt ){
          memcpy(PGHDR_TO_DATA(pPg), pHist->pStmt, pPager->pageSize);
          sqliteFree(pHist->pStmt);
          pHist->pStmt = 0;
        }
      }
      pPager->dbSize = pPager->stmtSize;

# 2001 September 15
#
# The author disclaims copyright to this source code.  In place of
# a legal notice, here is a blessing:
#
#    May you do good and not evil.
#    May you find forgiveness for yourself and forgive others.
#    May you share freely, never taking more than you give.
#
#***********************************************************************
# This file implements regression tests for SQLite library.  The
# focus of this file is testing the SELECT statement.
#
# $Id: autovacuum.test,v 1.26 2007/04/07 15:03:17 danielk1977 Exp $

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

# If this build of the library does not support auto-vacuum, omit this
# whole file.
ifcapable {!autovacuum || !pragma} {

} $AUTOVACUUM
do_test autovacuum-3.7 {
  execsql {
    DROP TABLE av1;
  }
  file_pages
} [expr $AUTOVACUUM?1:2]


#-----------------------------------------------------------------------
# Test that if a statement transaction around a CREATE INDEX statement is
# rolled back no corruption occurs.
#
do_test autovacuum-4.0 {
  # The last round of tests may have left the db in non-autovacuum mode.
  # Reset everything just in case.
  #
  db close
  file delete -force test.db test.db-journal
  sqlite3 db test.db
  execsql {
    PRAGMA auto_vacuum = 1;
    PRAGMA auto_vacuum;
  }
} {1}
do_test autovacuum-4.1 {
  execsql {
    CREATE TABLE av1(a, b);
    BEGIN;
  }
  for {set i 0} {$i<100} {incr i} {
    execsql "INSERT INTO av1 VALUES($i, '[string repeat X 200]');"