SQLite

/*
** 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.
**
*************************************************************************
** $Id: btree.c,v 1.79 2003/01/04 19:44:08 drh 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.

      *((uptr*)&pTo->apCell[i]) = x + to - from;
    }else{
      pTo->apCell[i] = pFrom->apCell[i];
    }
  }
}

/*
** The following parameters determine how many adjacent pages get involved
** in a balancing operation.  NN is the number of neighbors on either side
** of the page that participate in the balancing operation.  NB is the
** total number of pages that participate, including the target page and
** NN neighbors on either side.
**
** The minimum value of NN is 1 (of course).  Increasing NN above 1
** (to 2 or 3) gives a modest improvement in SELECT and DELETE performance
** in exchange for a larger degradation in INSERT and UPDATE performance.
** The value of NN appears to give the best results overall.
*/
#define NN 1             /* Number of neighbors on either side of pPage */
#define NB (NN*2+1)      /* Total pages involved in the balance */

/*
** This routine redistributes Cells on pPage and up to two siblings
** of pPage so that all pages have about the same amount of free space.
** Usually one sibling on either side of pPage is used in the balancing,
** though both siblings might come from one side if pPage is the first
** or last child of its parent.  If pPage has fewer than two siblings
** (something which can only happen if pPage is the root page or a 

**
** If this routine fails for any reason, it might leave the database
** in a corrupted state.  So if this routine fails, the database should
** be rolled back.
*/
static int balance(Btree *pBt, MemPage *pPage, BtCursor *pCur){
  MemPage *pParent;            /* The parent of pPage */






  int nCell;                   /* Number of cells in apCell[] */
  int nOld;                    /* Number of pages in apOld[] */
  int nNew;                    /* Number of pages in apNew[] */
  int nDiv;                    /* Number of cells in apDiv[] */
  int i, j, k;                 /* Loop counters */
  int idx;                     /* Index of pPage in pParent->apCell[] */
  int nxDiv;                   /* Next divider slot in pParent->apCell[] */
  int rc;                      /* The return code */
  int iCur;                    /* apCell[iCur] is the cell of the cursor */
  MemPage *pOldCurPage;        /* The cursor originally points to this page */
  int subtotal;                /* Subtotal of bytes in cells on one page */


  MemPage *extraUnref = 0;     /* A page that needs to be unref-ed */
  MemPage *apOld[NB];          /* pPage and up to two siblings */
  Pgno pgnoOld[NB];            /* Page numbers for each page in apOld[] */
  MemPage *apNew[NB+1];        /* pPage and up to NB siblings after balancing */
  Pgno pgnoNew[NB+1];          /* Page numbers for each page in apNew[] */
  int idxDiv[NB];              /* Indices of divider cells in pParent */
  Cell *apDiv[NB];             /* Divider cells in pParent */
  Cell aTemp[NB];              /* Temporary holding area for apDiv[] */
  int cntNew[NB+1];            /* Index in apCell[] of cell after i-th page */
  int szNew[NB+1];             /* Combined size of cells place on i-th page */
  MemPage aOld[NB];            /* Temporary copies of pPage and its siblings */
  Cell *apCell[(MX_CELL+2)*NB]; /* All cells from pages being balanced */
  int szCell[(MX_CELL+2)*NB];  /* Local size of all cells */

  /* 
  ** Return without doing any work if pPage is neither overfull nor
  ** underfull.
  */
  assert( sqlitepager_iswriteable(pPage) );
  if( !pPage->isOverfull && pPage->nFree<SQLITE_PAGE_SIZE/2 

  ** directly to balance_cleanup at any moment.
  */
  nOld = nNew = 0;
  sqlitepager_ref(pParent);

  /*
  ** Find sibling pages to pPage and the Cells in pParent that divide
  ** the siblings.  An attempt is made to find NN siblings on either
  ** side of pPage.  More siblings are taken from one side, however, if
  ** pPage there are fewer than NN siblings on the other side.  If pParent
  ** has NB or fewer children then all children of pParent are taken.
  */

  nxDiv = idx - NN;
  if( nxDiv + NB > pParent->nCell ){
    nxDiv = pParent->nCell - NB + 1;
  }
  if( nxDiv<0 ){
    nxDiv = 0;
  }
  nDiv = 0;
  for(i=0, k=nxDiv; i<NB; i++, k++){
    if( k<pParent->nCell ){
      idxDiv[i] = k;
      apDiv[i] = pParent->apCell[k];
      nDiv++;
      pgnoOld[i] = SWAB32(pBt, apDiv[i]->h.leftChild);
    }else if( k==pParent->nCell ){
      pgnoOld[i] = SWAB32(pBt, pParent->u.hdr.rightChild);

  ** Put the new pages in accending order.  This helps to
  ** keep entries in the disk file in order so that a scan
  ** of the table is a linear scan through the file.  That
  ** in turn helps the operating system to deliver pages
  ** from the disk more rapidly.
  **
  ** An O(n^2) insertion sort algorithm is used, but since
  ** n is never more than NB (a small constant), that should
  ** not be a problem.
  **
  ** When NB==3, this one optimization makes the database
  ** about 25% faster for large insertions and deletions.
  */
  for(i=0; i<k-1; i++){
    int minV = pgnoNew[i];
    int minI = i;
    for(j=i+1; j<k; j++){
      if( pgnoNew[j]<minV ){
        minI = j;