https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117432
--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
func_checker::compare_gimple_call
Does:
/* Checking of argument. */
for (i = 0; i < gimple_call_num_args (s1); ++i)
{
t1 = gimple_call_arg (s1, i);
t2 = gimple_call_arg (s2, i);
if (!compare_operand (t1, t2, get_operand_access_type (&map, t1)))
return return_false_with_msg ("GIMPLE call operands are different");
}
But then compare_operand does:
if (operand_equal_p (t1, t2, OEP_MATCH_SIDE_EFFECTS))
return true;
BUT operand_equal_p will returns true for -1 with different types:
/* Check equality of integer constants before bailing out due to
precision differences. */
if (TREE_CODE (arg0) == INTEGER_CST && TREE_CODE (arg1) == INTEGER_CST)
{
/* Address of INTEGER_CST is not defined; check that we did not forget
to drop the OEP_ADDRESS_OF flags. */
gcc_checking_assert (!(flags & OEP_ADDRESS_OF));
return tree_int_cst_equal (arg0, arg1);
}
So maybe I suspect there needs to be a check for the type difference in
func_checker::compare_gimple_call. But I am not 100% sure if there are other
cases where the operand_equal_p will return true and the types mismatch matter
though; Maybe there is not one except for call arguments.
Maybe OEP_BITWISE should be passed to operand_equal_p here ...
