We don't need to do that anymore, since blk_rq_map_sg() clears
individual entries.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>