On Mon, Oct 15, 2012 at 10:37 AM, Steven Bosscher <stevenb....@gmail.com> wrote: > On Mon, Oct 15, 2012 at 10:19 AM, Paolo Bonzini wrote: >> I prefer to declare the notes invalid and drop the notes. > > I strongly disagree with this approach though. It destroys information > that is correct, that we had before DF_RD_PRUNE_DEAD_DEFS, that we can > update, and that helps with optimization.
PR54916 is a case where dropping the notes would cause a missed optimization in cse-after-loop. With my patch to update the notes, the optimization (CSE of a load-immediate) is retained. Ciao! Steven
