[Bug tree-optimization/71264] [4.9/5/6/7 Regression] ICE in convert_move

[ikonomisma at googlemail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71264

--- Comment #3 from ikonomisma at googlemail dot com ---
I've just found a small change to the testcase that causes a *different* ICE to
appear:

void test(uint8_t *ptr, uint8_t *mask)
{
  footype mv = {0,1,2,3};
  for (size_t i = 0; i < 16; i += 4)
    {
      footype temp;
      memcpy(&temp, &ptr[i], sizeof(temp));
      temp ^= mv;
      memcpy(&ptr[i], &temp, sizeof(temp));
    }
}

Maybe this can help track down the cause.