On Mon, 2 Jun 2014, Hugh Dickins wrote:> Why not just move that cond_resched() down below the if/else?> No need to test need_resched() separately, and this page is not busy.> Would you like to propose your version from our kernel instead?