[Bug c/71598] Wrong optimization with aliasing enums

[amonakov at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71598

--- Comment #6 from Alexander Monakov <amonakov at gcc dot gnu.org> ---
... and even considering that the standard never actually says that "compatible
type" relation is transitive, and so two enums technically need not be
compatible with each other, the following should follow the standard to the
letter, but is still optimized to 'return 1':

enum e1 { c1=2 };

int f(unsigned *p, enum e1 *q)
{
  *p = 1;
  *q = c1;
  return *p;
}