[Bug tree-optimization/96921] Failure to optimize combined boolean not patterns

[pinskia at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96921

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
           Assignee|unassigned at gcc dot gnu.org      |pinskia at gcc dot 
gnu.org

--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
I was thinking if we see:
  _2 = 1 - _1;

and _1 has a range of [0,1] aka boolean
turn it into _1 ^ 1

There is already a pattern which turns ((int)a)^1 into (int)(~a).

You can see the other patterns in action if you write the code as:
int
foo1 (_Bool a, _Bool b)
{
  int c = a;
  c ^= 1;
  int d = b;
  d = d ^ 1;
  int e = c & d;
  return e ^ 1;
}

So something like:
(simplify
 (minus integer_one@0 SSA_NAME@1)
 (if (TREE_CODE (@0) == SSA_NAME
      && ssa_name_has_boolean_range (@0))
  (bit_xor @1 @0)))

That is for boolean types/ranges, we always use a ^ 1 (or ~a) instead of 1 - a
Take:
int
fooneg (_Bool a)
{
  return 1 - a;
}
int
fooxor (_Bool a)
{
  return a^1;
}