[Bug middle-end/46909] [4.6 Regression] Logical OR expressions are miscompiled

["jakub at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46909

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
         AssignedTo|unassigned at gcc dot       |jakub at gcc dot gnu.org
                   |gnu.org                     |

--- Comment #7 from Jakub Jelinek <jakub at gcc dot gnu.org> 2010-12-13 13:02:22 UTC ---
The fix is easy:
@@ -2441,7 +2441,7 @@ or_var_with_comparison_1 (gimple stmt,
          => (t OR inner2)
          If the partial result t is a constant, we win.  Otherwise
          continue on to try reassociating with the other inner test.  */
-      if (innercode == TRUTH_OR_EXPR)
+      if (is_or)
         {
           if (integer_onep (t))
         return boolean_true_node;
@@ -2474,7 +2474,7 @@ or_var_with_comparison_1 (gimple stmt,
       /* Handle the OR case, where we are reassociating:
          (inner1 OR inner2) OR (op2a code2 op2b)
          => (inner1 OR t)  */
-      if (innercode == TRUTH_OR_EXPR)
+      if (is_or)
         {
           if (integer_zerop (t))
         return inner1;

(the and folder does it that way correctly).  Still looking why it isn't
optimized now into x != 4.