[Bug rtl-optimization/49330] Integer arithmetic on addresses optimised with pointer arithmetic rules

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49330

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |wrong-code
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2011.06.09 09:36:46
                 CC|                            |ebotcazou at gcc dot
                   |                            |gnu.org
          Component|c                           |rtl-optimization
     Ever Confirmed|0                           |1
      Known to fail|                            |4.3.5, 4.4.5, 4.5.3, 4.6.0,
                   |                            |4.7.0

--- Comment #2 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-06-09 
09:36:46 UTC ---
I can confirm the original testcase with r173419 (not the one from comment#1).
The tree level optimizers handle this ok and we expand from

<bb 2>:
  px_1 = (uintptr_t) &x;
  py_2 = (uintptr_t) &y;
  d.0_3 = px_1 - py_2;
  d ={v} d.0_3;
  d.1_4 ={v} d;
  p_5 = d.1_4 + py_2;
  x = 1;
  # PT = { x y } (glob)
  p.2_6 = (int *) p_5;
  *p.2_6 = 2;
  D.2722_7 = x;
  return D.2722_7;

where you can also see that we correctly assume that p.2_6 may point either
x or y.

CSE1 optimizes the load from x to 1.