Re: About the is_gimple_min_invariant predicate

[Daniel Berlin]

On 7/5/07, Eric Botcazou <[EMAIL PROTECTED]> wrote:

> We never insert expressions for FRE, and the replacement we do will
> only replace the entire RHS with an SSA_NAME or a constant.

The problem is precisely that the expression is deemed a constant:

          /* Setting value numbers to constants will occasionally
             screw up phi congruence because constants are not
             uniquely associated with a single ssa name that can be
             looked up.  */
          if (simplified && is_gimple_min_invariant (simplified)
              && TREE_CODE (lhs) == SSA_NAME
              && simplified != rhs)
            {
              VN_INFO (lhs)->expr = simplified;
              VN_INFO (lhs)->has_constants = true;
              changed = set_ssa_val_to (lhs, simplified);
              goto done;
            }

Whee.
Okay, yes, looks like time for is_gimple_const to me :)