[Bug rtl-optimization/43503] field in-place decrement isn't optimized

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43503

Richard Guenther <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
          Component|c++                         |rtl-optimization
         Resolution|                            |FIXED
   Target Milestone|---                         |4.7.0

--- Comment #4 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-11-06 11:37:44 UTC ---
As Andrew said - we keep the instructions so they can set the flag register,
but then figure out the conditional jump is unnecessary.  On the trunk at -O2
we figure that out via tail-merging, so it's optimized there.

Pretty artificial testcase.  Fixed.