https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117317
--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Slightly cleaned up:
struct C {
constexpr bool operator== (const C &b) const { return foo (); }
constexpr virtual bool foo () const = 0;
};
class A : public C {};
class B : public C {};
template <int>
struct D : A, B
{
constexpr bool operator== (const D &) const = default;
constexpr bool foo () const override { return true; }
};
struct E : D<1> {};
constexpr E e;
constexpr E f;
static_assert (e == f, "");
Seems we call expand_or_defer_fn on the D<1>::foo, and given that it is
constexpr and thus inline, we just call tentative_decl_linkage on it until we
figure out if we really need to emit it or not.
But expand_or_defer_fn calls emit_associated_thunks, which calls use_thunk and
that asserts DECL_INTERFACE_KNOWN, which is not the case for the tentative
linkage decls.
