[Bug tree-optimization/117764] New: cddce should handle __builtin_unreachable guards

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

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117764

            Bug ID: 117764
           Summary: cddce should handle __builtin_unreachable guards
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: hubicka at gcc dot gnu.org
  Target Milestone: ---

In the following testcase:

#include <vector>

void test (std::vector<int> v)
{
        for (size_t i = 0; i < v.size (); i++)
                ;
}

we used to be able to optimize out the empty loop pre-inlining, however now
after cddce2 we get:

void test (struct vector & v)
{
  ptrdiff_t __dif;
  size_t i;
  int * _4;
  int * _6;
  long int _7;
  long unsigned int _9;

  <bb 2> [local count: 118111600]:
  goto <bb 4>; [100.00%]

  <bb 3> [local count: 955630224]:
  i_5 = i_1 + 1;

  <bb 4> [local count: 1073741824]:
  # i_1 = PHI <0(2), i_5(3)>
  _4 = MEM[(const struct vector *)v_3(D)].D.25542._M_impl.D.24853._M_finish;
  _6 = MEM[(const struct vector *)v_3(D)].D.25542._M_impl.D.24853._M_start;
  _7 = _4 - _6;
  __dif_8 = _7 /[ex] 4;
  if (_7 < 0)
    goto <bb 5>; [0.00%]
  else
    goto <bb 6>; [100.00%]

  <bb 5> [count: 0]:
  __builtin_unreachable ();

  <bb 6> [local count: 1073741824]:
  _9 = (long unsigned int) __dif_8;
  if (i_1 < _9)
    goto <bb 3>; [89.00%]
  else
    goto <bb 7>; [11.00%]

  <bb 7> [local count: 118111600]:
  return;

}

The mini-dce of fnsummary is able to figure out that if (_7 < 0)
is unnecesary, but rest is held together by the empty loop.

I think cddce needs to understand __builtin_unreachable guards, too. However
what is not clear to me is how to decide what guards are potentially useful for
later optimization...