On Fri, Oct 23, 2015 at 04:08:14PM +0200, Petr Uzel wrote:
> When creating a new partition, parted tries to snap its boundaries.
> If the new partition lies within free space, it only makes sense
> to snap it so that is it bigger, not smaller.
>
> * parted/parted.c (snap_to_boundaries): Limit direction in what snapping
> is done.
> ---
> parted/parted.c | 8 ++++++++
> 1 file changed, 8 insertions(+)
>
> diff --git a/parted/parted.c b/parted/parted.c
> index a9426c4..840e9e3 100644
> --- a/parted/parted.c
> +++ b/parted/parted.c
> @@ -379,6 +379,14 @@ snap_to_boundaries (PedGeometry* new_geom, PedGeometry*
> old_geom,
> end_part = ped_disk_get_partition_by_sector (disk, end);
> adjacent = (start_part->geom.end + 1 == end_part->geom.start);
>
> + /* If start lies within free space, then don't allow it to move up.
> + * If end lies within free space, then don't allow it to move down.
> + */
> + if (start_part->type == PED_PARTITION_FREESPACE)
> + start_allow &= ~MOVE_UP;
> + if (end_part->type == PED_PARTITION_FREESPACE)
> + end_allow &= ~MOVE_DOWN;
> +
> /* If we can snap to old_geom, then we will... */
> /* and this will enforce the snapped positions */
> if (old_geom) {
> --
> 1.8.4.5
>
>
Looks ok to me, but we could use a test to go along with it.
--
Brian C. Lane | Anaconda Team | IRC: bcl #anaconda | Port Orchard, WA (PST8PDT)
