On Wed, Mar 01, 2023 at 04:53:23PM -0500, Jason Merrill wrote:
> > @@ -13791,12 +13830,39 @@ std_pair_ref_ref_p (tree t)
> > const int& y = (f(1), 42); // NULL_TREE
> > const int& z = f(f(1)); // f(f(1))
> > - EXPR is the initializer. */
> > + EXPR is the initializer. If ARG_P is true, we're processing an argument
> > + to a function; the point is to distinguish between, for example,
> > +
> > + Ref::inner (&TARGET_EXPR <D.2839, F::foo (fm)>)
> > +
> > + where we shouldn't warn, and
> > +
> > + Ref::inner (&TARGET_EXPR <D.2908, F::foo (&TARGET_EXPR <...>)>)
> > +
> > + where we should warn (Ref is a reference_like_class_p so we see through
> > + it. */
> > static tree
> > -do_warn_dangling_reference (tree expr)
> > +do_warn_dangling_reference (tree expr, bool arg_p)
> > {
> > STRIP_NOPS (expr);
> > + if (TREE_CODE (expr) == ADDR_EXPR)
> > + expr = TREE_OPERAND (expr, 0);
>
> I think if we move this here, we also need to check that expr before
> STRIP_NOPS had REFERENCE_TYPE. OK with that change.
Sorry but I don't think I can do that. There can be CONVERT_EXPRs
that need to be stripped, whether arg_p or !arg_p. For example, we can get
(const int *) f ((const int &) &TARGET_EXPR <D.2765, NON_LVALUE_EXPR <10>>)
for
const int& r5 = (42, f(10));
Is the patch OK as-is then?
Marek
