[Bug tree-optimization/19828] [4.0 Regression] LIM is pulling out a pure function even though there is something which can modify global memory

["pinskia at gcc dot gnu dot org" <gcc-bugzilla at gcc dot gnu dot org>]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-02-08 20:02 -------
(In reply to comment #2)
> Here's another one.  This may be a different bug.  
Yes that is a different (but related) bug.  The problem is now, what is definition of pure functions (for 
that testcase).
Take the following:
int *a;
int *func_pure () __attribute__((pure));
int func_pure_2 () __attribute__((pure));
int *func_pure() { return a; }
int func_pure_2() { return *a;}
int i;
int
func_loop_3 (int arg)
{
  int var = 0;
  i = 1;
  while (arg--)
    {
      if (func_pure ())
        var = func_pure_2 ();
    }
  return var;
}
void abort (void);

int main()
{
  int i;
  i = func_loop_3 (10);
  if (i)
   abort ();
}

Is func_pure_2 really pure and if it is, then we should change it so we can trap on it.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19828