fix the penalty calculation update

    ELSE

    {

#ifdef FIX_QMETADATA_PENALTY    

        Word16 m, sorted, index1, index2;

        FOR( j = 0; j < coding_subbands; j++ )

        {

            penalty[j] = 1;

            move32();

            FOR( k = 0; k < coding_subbands; k++ )

            ind_order[j] = j;

            move16();

        }

        sorted = 0;

        move16();

        FOR( m = coding_subbands - 1; m && !sorted; m-- )

        {

                IF( NE_16(k, j) )

            sorted = 1;

            move16();

            FOR( j = 0; j < m; j++ )

            {

                index1 = ind_order[j];

                index2 = ind_order[j + 1];

                tmp = 0;

                move16();

                FOR( k = 0; k < no_subframes; k++ )

                {

                    penalty[j] = L_shr( W_extract_l( W_mult_32_16( penalty[j], bits_sph_idx_orig[k][0] ) ), 1 );

                    tmp = add( tmp, sub( bits_sph_idx_orig[index1][k], q_direction->band_data[index1].bits_sph_idx[k] ) );

                }

                penalty[0] = L_shr( W_extract_l( W_mult_32_16( tmp, bits_sph_idx_orig[index2][0] ) ), 1 );

                tmp = 0;

                move16();

                FOR( k = 0; k < no_subframes; k++ )

                {

                    tmp = add( tmp, sub( bits_sph_idx_orig[index2][k], q_direction->band_data[index2].bits_sph_idx[k] ) );

            }

            tmp = sub( bits_sph_idx_orig[j][0], q_direction->band_data[j].bits_sph_idx[0] ); 

            FOR( k = 1; k < no_subframes; k++ )

                penalty[1] = L_shr( W_extract_l( W_mult_32_16( tmp, bits_sph_idx_orig[index1][0] ) ), 1 );

                IF( LT_32( penalty[0], penalty[1] ) )

            {

                tmp = add( tmp, sub( bits_sph_idx_orig[j][k], q_direction->band_data[j].bits_sph_idx[k] ) ) ;

                    sorted = 0;

                    move16();

                    ind_order[j] = index2;

                    move16();

                    ind_order[j + 1] = index1;

                    move16();

            }

            penalty[j] =  L_shr(W_extract_l( W_mult_32_16( penalty[j], tmp )), 1);

        }

        sort_desc_ind_32_fx( penalty, coding_subbands, ind_order );

        for ( j = 0; j < coding_subbands; j++ )

        {

            printf( "%d ", ind_order[j] );

        }

        printf( "\n" );

#else

        FOR( j = 0; j < coding_subbands; j++ )

        {