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 1.96 2004/02/09 01:20:37 drh Exp $
*/
#include "os.h"         /* Must be first to enable large file support */
#include "sqliteInt.h"
#include "pager.h"
#include <assert.h>
#include <string.h>

#define SET_PAGER(X)
#define CLR_PAGER(X)
#define TRACE1(X)
#define TRACE2(X,Y)
#define TRACE3(X,Y,Z)
#endif

/*
** Number of extra bytes of data allocated at the end of each page and
** stored on disk but not used by the higher level btree layer.
*/
#ifndef SQLITE_PAGE_RESERVE
#define SQLITE_PAGE_RESERVE 0
#endif

/*
** The total number of bytes stored on disk for each page.
*/
#define SQLITE_BLOCK_SIZE (SQLITE_PAGE_SIZE+SQLITE_PAGE_RESERVE)


/*
** The page cache as a whole is always in one of the following
** states:
**
**   SQLITE_UNLOCK       The page cache is not currently reading or 
**                       writing the database file.  There is no

  PgHdr *pNextCkpt, *pPrevCkpt;  /* List of pages in the checkpoint journal */
  u8 inJournal;                  /* TRUE if has been written to journal */
  u8 inCkpt;                     /* TRUE if written to the checkpoint journal */
  u8 dirty;                      /* TRUE if we need to write back changes */
  u8 needSync;                   /* Sync journal before writing this page */
  u8 alwaysRollback;             /* Disable dont_rollback() for this page */
  PgHdr *pDirty;                 /* Dirty pages sorted by PgHdr.pgno */
  /* SQLITE_BLOCK_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.
*/
#define PGHDR_TO_DATA(P)  ((void*)(&(P)[1]))
#define DATA_TO_PGHDR(D)  (&((PgHdr*)(D))[-1])
#define PGHDR_TO_EXTRA(P) ((void*)&((char*)(&(P)[1]))[SQLITE_BLOCK_SIZE])

/*
** How big to make the hash table used for locating in-memory pages
** by page number.
*/
#define N_PG_HASH 2048

  int ckptNRec;               /* Number of records in the checkpoint journal */
  int nExtra;                 /* Add this many bytes to each in-memory page */
  void (*xDestructor)(void*); /* Call this routine when freeing pages */
  int nPage;                  /* Total number of in-memory pages */
  int nRef;                   /* Number of in-memory pages with PgHdr.nRef>0 */
  int mxPage;                 /* Maximum number of pages to hold in cache */
  int nHit, nMiss, nOvfl;     /* Cache hits, missing, and LRU overflows */
  void (*xCodec)(void*,void*,int); /* Routine for en/decoding on-disk data */
  void *pCodecArg;            /* First argument to xCodec() */
  u8 journalOpen;             /* True if journal file descriptors is valid */
  u8 journalStarted;          /* True if header of journal is synced */
  u8 useJournal;              /* Use a rollback journal on this file */
  u8 ckptOpen;                /* True if the checkpoint journal is open */
  u8 ckptInUse;               /* True we are in a checkpoint */
  u8 ckptAutoopen;            /* Open ckpt journal when main journal is opened*/
  u8 noSync;                  /* Do not sync the journal if true */

**
** Actually, this structure is the complete page record for pager
** formats less than 3.  Beginning with format 3, this record is surrounded
** by two checksums.
*/
typedef struct PageRecord PageRecord;
struct PageRecord {
  Pgno pgno;                      /* The page number */
  char aData[SQLITE_BLOCK_SIZE];  /* Original data for page pgno */
};

/*
** Journal files begin with the following magic string.  The data
** was obtained from /dev/random.  It is used only as a sanity check.
**
** There are three journal formats (so far). The 1st journal format writes

** the journal file after power is restored.  If an attempt is then made
** to roll the journal back, the database could be corrupted.  The additional
** sanity checking data is an attempt to discover the garbage in the
** journal and ignore it.
**
** The sanity checking information for the 3rd journal format consists
** of a 32-bit checksum on each page of data.  The checksum covers both
** the page number and the SQLITE_BLOCK_SIZE bytes of data for the page.
** This cksum is initialized to a 32-bit random value that appears in the
** journal file right after the header.  The random initializer is important,
** because garbage data that appears at the end of a journal is likely
** data that was once in other files that have now been deleted.  If the
** garbage data came from an obsolete journal file, the checksums might
** be correct.  But by initializing the checksum to random value which
** is different for every journal, we minimize that risk.

** The size of the header and of each page in the journal varies according
** to which journal format is being used.  The following macros figure out
** the sizes based on format numbers.
*/
#define JOURNAL_HDR_SZ(X) \
   (sizeof(aJournalMagic1) + sizeof(Pgno) + ((X)>=3)*2*sizeof(u32))
#define JOURNAL_PG_SZ(X) \
   (SQLITE_BLOCK_SIZE + sizeof(Pgno) + ((X)>=3)*sizeof(u32))

/*
** Enable reference count tracking here:
*/
#ifdef SQLITE_TEST
  int pager_refinfo_enable = 0;
  static void pager_refinfo(PgHdr *p){

  }

  /* Playback the page.  Update the in-memory copy of the page
  ** at the same time, if there is one.
  */
  pPg = pager_lookup(pPager, pgRec.pgno);
  TRACE2("PLAYBACK %d\n", pgRec.pgno);
  sqliteOsSeek(&pPager->fd, (pgRec.pgno-1)*(off_t)SQLITE_BLOCK_SIZE);
  rc = sqliteOsWrite(&pPager->fd, pgRec.aData, SQLITE_BLOCK_SIZE);
  if( pPg ){
    /* No page should ever be rolled back that is in use, except for page
    ** 1 which is held in use in order to keep the lock on the database
    ** active.
    */
    assert( pPg->nRef==0 || pPg->pgno==1 );
    memcpy(PGHDR_TO_DATA(pPg), pgRec.aData, SQLITE_BLOCK_SIZE);
    memset(PGHDR_TO_EXTRA(pPg), 0, pPager->nExtra);
    pPg->dirty = 0;
    pPg->needSync = 0;
    if( pPager->xCodec ){
      pPager->xCodec(pPager->pCodecArg, PGHDR_TO_DATA(pPg), 0);
    }
  }
  return rc;
}

/*
** Playback the journal and thus restore the database file to
** the state it was in before we started making changes.  

**       in format 3 only.
**    *  4 byte big-endian integer which is the initial value for the 
**       sanity checksum.  This field appears in format 3 only.
**    *  4 byte integer which is the number of pages to truncate the
**       database to during a rollback.
**    *  Zero or more pages instances, each as follows:
**        +  4 byte page number.
**        +  SQLITE_BLOCK_SIZE bytes of data.
**        +  4 byte checksum (format 3 only)
**
** When we speak of the journal header, we mean the first 4 bullets above.
** Each entry in the journal is an instance of the 5th bullet.  Note that
** bullets 2 and 3 only appear in format-3 journals.
**
** Call the value from the second bullet "nRec".  nRec is the number of

    assert( nRec*JOURNAL_PG_SZ(2)+JOURNAL_HDR_SZ(2)==szJ );
  }
  rc = read32bits(format, &pPager->jfd, &mxPg);
  if( rc!=SQLITE_OK ){
    goto end_playback;
  }
  assert( pPager->origDbSize==0 || pPager->origDbSize==mxPg );
  rc = sqliteOsTruncate(&pPager->fd, SQLITE_BLOCK_SIZE*(off_t)mxPg);
  if( rc!=SQLITE_OK ){
    goto end_playback;
  }
  pPager->dbSize = mxPg;
  
  /* Copy original pages out of the journal and back into the database file.
  */

  /* Pages that have been written to the journal but never synced
  ** where not restored by the loop above.  We have to restore those
  ** pages by reading them back from the original database.
  */
  if( rc==SQLITE_OK ){
    PgHdr *pPg;
    for(pPg=pPager->pAll; pPg; pPg=pPg->pNextAll){
      char zBuf[SQLITE_BLOCK_SIZE];
      if( !pPg->dirty ) continue;
      if( (int)pPg->pgno <= pPager->origDbSize ){
        sqliteOsSeek(&pPager->fd, SQLITE_BLOCK_SIZE*(off_t)(pPg->pgno-1));
        rc = sqliteOsRead(&pPager->fd, zBuf, SQLITE_BLOCK_SIZE);
        if( rc ) break;
        if( pPager->xCodec ){
          pPager->xCodec(pPager->pCodecArg, zBuf, 0);
        }
      }else{
        memset(zBuf, 0, SQLITE_BLOCK_SIZE);
      }
      if( pPg->nRef==0 || memcmp(zBuf, PGHDR_TO_DATA(pPg), SQLITE_BLOCK_SIZE) ){
        memcpy(PGHDR_TO_DATA(pPg), zBuf, SQLITE_BLOCK_SIZE);
        memset(PGHDR_TO_EXTRA(pPg), 0, pPager->nExtra);
      }
      pPg->needSync = 0;
      pPg->dirty = 0;
    }
  }

  off_t szJ;               /* Size of the full journal */
  int nRec;                /* Number of Records */
  int i;                   /* Loop counter */
  int rc;

  /* Truncate the database back to its original size.
  */
  rc = sqliteOsTruncate(&pPager->fd, SQLITE_BLOCK_SIZE*(off_t)pPager->ckptSize);
  pPager->dbSize = pPager->ckptSize;

  /* Figure out how many records are in the checkpoint journal.
  */
  assert( pPager->ckptInUse && pPager->journalOpen );
  sqliteOsSeek(&pPager->cpfd, 0);
  nRec = pPager->ckptNRec;

  if( pPager->dbSize>=0 ){
    return pPager->dbSize;
  }
  if( sqliteOsFileSize(&pPager->fd, &n)!=SQLITE_OK ){
    pPager->errMask |= PAGER_ERR_DISK;
    return 0;
  }
  n /= SQLITE_BLOCK_SIZE;
  if( pPager->state!=SQLITE_UNLOCK ){
    pPager->dbSize = n;
  }
  return n;
}

/*

    rc = pager_errcode(pPager);
    return rc;
  }
  if( nPage>=(unsigned)pPager->dbSize ){
    return SQLITE_OK;
  }
  syncJournal(pPager);
  rc = sqliteOsTruncate(&pPager->fd, SQLITE_BLOCK_SIZE*(off_t)nPage);
  if( rc==SQLITE_OK ){
    pPager->dbSize = nPage;
  }
  return rc;
}

/*

  Pager *pPager;
  int rc;

  if( pList==0 ) return SQLITE_OK;
  pPager = pList->pPager;
  while( pList ){
    assert( pList->dirty );
    sqliteOsSeek(&pPager->fd, (pList->pgno-1)*(off_t)SQLITE_BLOCK_SIZE);
    if( pPager->xCodec ){
      pPager->xCodec(pPager->pCodecArg, PGHDR_TO_DATA(pList), 1);
    }
    rc = sqliteOsWrite(&pPager->fd, PGHDR_TO_DATA(pList), SQLITE_BLOCK_SIZE);
    if( pPager->xCodec ){
      pPager->xCodec(pPager->pCodecArg, PGHDR_TO_DATA(pList), 0);
    }
    if( rc ) return rc;
    pList->dirty = 0;
    pList = pList->pDirty;
  }
  return SQLITE_OK;
}

  }
  if( pPg==0 ){
    /* The requested page is not in the page cache. */
    int h;
    pPager->nMiss++;
    if( pPager->nPage<pPager->mxPage || pPager->pFirst==0 ){
      /* Create a new page */
      pPg = sqliteMallocRaw( sizeof(*pPg) + SQLITE_BLOCK_SIZE 
                              + sizeof(u32) + pPager->nExtra );
      if( pPg==0 ){
        pager_unwritelock(pPager);
        pPager->errMask |= PAGER_ERR_MEM;
        return SQLITE_NOMEM;
      }
      memset(pPg, 0, sizeof(*pPg));

    if( pPager->dbSize<0 ) sqlitepager_pagecount(pPager);
    if( pPager->errMask!=0 ){
      sqlitepager_unref(PGHDR_TO_DATA(pPg));
      rc = pager_errcode(pPager);
      return rc;
    }
    if( pPager->dbSize<(int)pgno ){
      memset(PGHDR_TO_DATA(pPg), 0, SQLITE_BLOCK_SIZE);
    }else{
      int rc;
      sqliteOsSeek(&pPager->fd, (pgno-1)*(off_t)SQLITE_BLOCK_SIZE);
      rc = sqliteOsRead(&pPager->fd, PGHDR_TO_DATA(pPg), SQLITE_BLOCK_SIZE);
      if( rc!=SQLITE_OK ){
        off_t fileSize;
        if( sqliteOsFileSize(&pPager->fd,&fileSize)!=SQLITE_OK
               || fileSize>=pgno*SQLITE_BLOCK_SIZE ){
          sqlitepager_unref(PGHDR_TO_DATA(pPg));
          return rc;
        }else{
          memset(PGHDR_TO_DATA(pPg), 0, SQLITE_BLOCK_SIZE);
        }
      }else if( pPager->xCodec ){
        pPager->xCodec(pPager->pCodecArg, PGHDR_TO_DATA(pPg), 0);
      }
    }
  }else{
    /* The requested page is in the page cache. */
    pPager->nHit++;
    page_ref(pPg);
  }

        saved = *(u32*)PGHDR_TO_EXTRA(pPg);
        store32bits(cksum, pPg, SQLITE_PAGE_SIZE);
        szPg = SQLITE_PAGE_SIZE+8;
      }else{
        szPg = SQLITE_PAGE_SIZE+4;
      }
      store32bits(pPg->pgno, pPg, -4);
      if( pPager->xCodec ){
        pPager->xCodec(pPager->pCodecArg, pData, 1);
      }
      rc = sqliteOsWrite(&pPager->jfd, &((char*)pData)[-4], szPg);
      if( pPager->xCodec ){
        pPager->xCodec(pPager->pCodecArg, pData, 0);
      }
      if( journal_format>=JOURNAL_FORMAT_3 ){
        *(u32*)PGHDR_TO_EXTRA(pPg) = saved;
      }
      if( rc!=SQLITE_OK ){
        sqlitepager_rollback(pPager);
        pPager->errMask |= PAGER_ERR_FULL;
        return rc;

  ** then write the current page to the checkpoint journal.  Note that
  ** the checkpoint journal always uses the simplier format 2 that lacks
  ** checksums.  The header is also omitted from the checkpoint journal.
  */
  if( pPager->ckptInUse && !pPg->inCkpt && (int)pPg->pgno<=pPager->ckptSize ){
    assert( pPg->inJournal || (int)pPg->pgno>pPager->origDbSize );
    store32bits(pPg->pgno, pPg, -4);
    rc = sqliteOsWrite(&pPager->cpfd, &((char*)pData)[-4], SQLITE_BLOCK_SIZE+4);
    if( rc!=SQLITE_OK ){
      sqlitepager_rollback(pPager);
      pPager->errMask |= PAGER_ERR_FULL;
      return rc;
    }
    pPager->ckptNRec++;
    assert( pPager->aInCkpt!=0 );

  void *pPage;
  int rc;

  rc = sqlitepager_get(pPager, pgno, &pPage);
  if( rc==SQLITE_OK ){
    rc = sqlitepager_write(pPage);
    if( rc==SQLITE_OK ){
      memcpy(pPage, pData, SQLITE_BLOCK_SIZE);
    }
    sqlitepager_unref(pPage);
  }
  return rc;
}

/*

/*
** Return the full pathname of the database file.
*/
const char *sqlitepager_filename(Pager *pPager){
  return pPager->zFilename;
}

/*
** Set the codec for this pager
*/
void sqlitepager_set_codec(
  Pager *pPager,
  void (*xCodec)(void*,void*,int),
  void *pCodecArg
){
  pPager->xCodec = xCodec;
  pPager->pCodecArg = pCodecArg;
}

#ifdef SQLITE_TEST
/*
** Print a listing of all referenced pages and their ref count.
*/
void sqlitepager_refdump(Pager *pPager){
  PgHdr *pPg;

**    May you share freely, never taking more than you give.
**
*************************************************************************
** This header file defines the interface that the sqlite page cache
** subsystem.  The page cache subsystem reads and writes a file a page
** at a time and provides a journal for rollback.
**
** @(#) $Id: pager.h,v 1.24 2004/02/09 01:20:37 drh Exp $
*/

/*
** The size of one page
**
** You can change this value to another (reasonable) power of two
** such as 512, 2048, 4096, or 8192 and things will still work.  But
** experiments show that a page size of 1024 gives the best speed.
** (The speed differences are minimal.)
*/
#ifndef SQLITE_PAGE_SIZE
#define SQLITE_PAGE_SIZE 1024
#endif

/*
** Maximum number of pages in one database.  (This is a limitation of
** imposed by 4GB files size limits.)
*/
#define SQLITE_MAX_PAGE 1073741823

int sqlitepager_ckpt_rollback(Pager*);
void sqlitepager_dont_rollback(void*);
void sqlitepager_dont_write(Pager*, Pgno);
int *sqlitepager_stats(Pager*);
void sqlitepager_set_safety_level(Pager*,int);
const char *sqlitepager_filename(Pager*);
int sqlitepager_rename(Pager*, const char *zNewName);
void sqlitepager_set_codec(Pager*,void(*)(void*,void*,int),void*);

#ifdef SQLITE_TEST
void sqlitepager_refdump(Pager*);
int pager_refinfo_enable;
int journal_format;
#endif