SQLite

** 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 1.48 2002/07/06 16:28:48 drh Exp $
*/
#include "sqliteInt.h"
#include "pager.h"
#include "os.h"
#include <assert.h>
#include <string.h>

  PgHdr *pNextHash, *pPrevHash;  /* Hash collision chain for PgHdr.pgno */
  int nRef;                      /* Number of users of this page */
  PgHdr *pNextFree, *pPrevFree;  /* Freelist of pages where nRef==0 */
  PgHdr *pNextAll, *pPrevAll;    /* A list of all pages */
  char inJournal;                /* TRUE if has been written to journal */
  char inCkpt;                   /* TRUE if written to the checkpoint journal */
  char dirty;                    /* TRUE if we need to write back changes */
  char alwaysRollback;           /* Ignore dont_rollback() calls if true */
  /* SQLITE_PAGE_SIZE bytes of page data follow this header */
  /* Pager.nExtra bytes of local data follow the page data */
};

/*
** Convert a pointer to a PgHdr into a pointer to its data
** and back again.

    sqliteOsClose(&pPager->cpfd);
    pPager->ckptOpen = 0;
  }
  sqliteOsClose(&pPager->jfd);
  pPager->journalOpen = 0;
  sqliteOsDelete(pPager->zJournal);
  rc = sqliteOsReadLock(&pPager->fd);

  sqliteFree( pPager->aInJournal );
  pPager->aInJournal = 0;
  for(pPg=pPager->pAll; pPg; pPg=pPg->pNextAll){
    pPg->inJournal = 0;
    pPg->dirty = 0;
  }
  if( rc==SQLITE_OK ){
    pPager->state = SQLITE_READLOCK;
  }else{
    /* This can only happen if a process does a BEGIN, then forks and the
    ** child process does the COMMIT.  Because of the semantics of unix
    ** file locking, the unlock will fail.
    */
    pPager->state = SQLITE_UNLOCK;
  }
  return rc;
}

/*
** Read a single page from the journal file opened on file descriptor
** jfd.  Playback this one page.
*/

      return rc;
    }
    pPager->aInJournal = sqliteMalloc( pPager->dbSize/8 + 1 );
    if( pPager->aInJournal==0 ){
      sqliteOsReadLock(&pPager->fd);
      return SQLITE_NOMEM;
    }
    rc = sqliteOsOpenExclusive(pPager->zJournal, &pPager->jfd,pPager->tempFile);
    if( rc!=SQLITE_OK ){
      sqliteFree(pPager->aInJournal);
      pPager->aInJournal = 0;
      sqliteOsReadLock(&pPager->fd);
      return SQLITE_CANTOPEN;
    }
    pPager->journalOpen = 1;

** The overlying software layer calls this routine when all of the data
** on the given page is unused.  The pager marks the page as clean so
** that it does not get written to disk.
**
** Tests show that this optimization, together with the
** sqlitepager_dont_rollback() below, more than double the speed
** of large INSERT operations and quadruple the speed of large DELETEs.
**
** When this routine is called, set the alwaysRollback flag to true.
** Subsequent calls to sqlitepager_dont_rollback() for the same page
** will thereafter be ignored.  This is necessary to avoid a problem
** where a page with data is added to the freelist during one part of
** a transaction then removed from the freelist during a later part
** of the same transaction and reused for some other purpose.  When it
** is first added to the freelist, this routine is called.  When reused,
** the dont_rollback() routine is called.  But because the page contains
** critical data, we still need to be sure it gets rolled back in spite
** of the dont_rollback() call.
*/
void sqlitepager_dont_write(Pager *pPager, Pgno pgno){
  PgHdr *pPg;

  pPg = pager_lookup(pPager, pgno);
  pPg->alwaysRollback = 1;
  if( pPg && pPg->dirty ){
    if( pPager->dbSize==(int)pPg->pgno && pPager->origDbSize<pPager->dbSize ){
      /* If this pages is the last page in the file and the file has grown
      ** during the current transaction, then do NOT mark the page as clean.
      ** When the database file grows, we must make sure that the last page
      ** gets written at least once so that the disk file will be the correct
      ** size. If you do not write this page and the size of the file

** rollback journal.
*/
void sqlitepager_dont_rollback(void *pData){
  PgHdr *pPg = DATA_TO_PGHDR(pData);
  Pager *pPager = pPg->pPager;

  if( pPager->state!=SQLITE_WRITELOCK || pPager->journalOpen==0 ) return;
  if( pPg->alwaysRollback ) 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->ckptInUse ){
      pPager->aInCkpt[pPg->pgno/8] |= 1<<(pPg->pgno&7);
      pPg->inCkpt = 1;