Gcse replacement (Was: Re: Don't mark operands of jump instructions available in gcse)

[law@redhat.com]

  In message < 20010814150741.C8866@atrey.karlin.mff.cuni.cz >you write:
  > The lowering pass is likely going to show more architecture dependent
  > thinks and thus introduce more temporary results and more opurtunities
  > for CSE.
  > How we are going to cleanup these?
I suspect the best way to address this is going to be with a limited form
of local cse.  For exmaple we probably won't mess with all the path following
code that's in cse.c right now.


jeff