On Thu, 25 Jul 2024, Patrick Palka wrote:
> On Mon, 22 Jul 2024, Jason Merrill wrote:
>
> > On 7/19/24 10:30 AM, Patrick Palka wrote:
> > > On Thu, 18 Jul 2024, Jason Merrill wrote:
> > >
> > > > On 7/18/24 12:45 PM, Patrick Palka wrote:
> > > > > Bootstrapped and regtested on x86_64-pc-linux-gnu, does thi look
> > > > > OK for trunk/14?
> > > > >
> > > > > -- >8 --
> > > > >
> > > > > As a followup of r15-2047-g7954bb4fcb6fa8, we also need to consider
> > > > > dependent attributes when recursing into a non-template alias that
> > > > > names
> > > > > a dependent alias template specialization (and so STF_STRIP_DEPENDENT
> > > > > is set), otherwise in the first testcase below we undesirably strip B
> > > > > all the way to T instead of to A<T>.
> > > > >
> > > > > We also need to move the typedef recursion case of strip_typedefs up
> > > > > to
> > > > > get checked before the compound type recursion cases. Otherwise for C
> > > > > below (which ultimately aliases T*) we end up stripping it to T*
> > > > > instead
> > > > > of to A<T*> because the POINTER_TYPE recursion dominates the typedef
> > > > > recursion. It also means we issue an unexpected extra error in the
> > > > > third testcase below.
> > > > >
> > > > > Ideally we would also want to consider dependent attributes on
> > > > > non-template aliases, so that we accept the second testcase below, but
> > > > > making that work correctly would require broader changes to e.g.
> > > > > spec_hasher which currently assumes all non-template aliases are
> > > > > stripped and hence it'd conflate the dependent specializations A<T>
> > > > > and A<B> even if we didn't strip B.
> > > >
> > > > Wouldn't that just be a matter of changing structural_comptypes to
> > > > consider
> > > > dependent attributes as well as dependent specializations?
> > >
> > > Pretty much, it seems. ISTM we should check dependent attributes even
> > > when !comparing_dependent_aliases since they affect type identity rather
> > > than just SFINAE behavior.
> > >
> > > >
> > > > Or better, adding attributes to dependent_alias_template_spec_p (and
> > > > changing
> > > > its name)? It seems like other callers would also benefit from that
> > > > change.
> > >
> > > I ended up adding a new predicate opaque_alias_p separate from
> > > dependent_alias_template_spec_p since ISTM we need to call it from
> > > there and from alias_template_specialization_p to avoid looking through
> > > such aliases.
> >
> > Sounds good, but I think let's add the word "dependent" to the name of the
> > new
> > function.
>
> Done.
>
> >
> > > So opaque_alias_p checks for type identity of an alias, whereas
> > > dependent_alias_template_spec_p more broadly checks for SFINAE identity.
> > >
> > > Something like the following (as an incremental patch on top of the
> > > previous one, to consider separately for backportings since it's riskier):
> > >
> > > diff --git a/gcc/cp/pt.cc b/gcc/cp/pt.cc
> > > index 0620c8c023a..4d4a5cef92c 100644
> > > --- a/gcc/cp/pt.cc
> > > +++ b/gcc/cp/pt.cc
> > > @@ -6508,6 +6508,19 @@ alias_type_or_template_p (tree t)
> > > || DECL_ALIAS_TEMPLATE_P (t));
> > > }
> > > +/* Return true if substituting into T would yield a different type than
> > > + substituting into its expansion. */
> >
> > Please discuss when to use this vs dependent_alias_template_spec_p (both
> > here
> > and there). Maybe just to say that any place that checks one probably also
> > wants to check the other.
> >
> > Other places that use d_a_t_s_p and seem to need adjusting:
> > dependent_type_p_r, any_dependent_arguments_need_structural_equality_p,
> > alias_ctad_tweaks.
>
> For dependent_type_p_r, I think we can get rid of the existing d_a_t_s_p
> check if we manually set TYPE_DEPENDENT_P/_VALID to true at parse time
> like we do at instantiation time from instantiate_alias_template.
>
> For any_dependent_arguments_need_structural_equality_p, those checks
> can't be removed but we can relax nt_transparent to nt_opaque since we
> know the template arguments have already gone through strip_typedefs.
>
> For alias_ctad_tweaks I think we can replace the d_a_t_s_p with
> template_args_equal which has the desired comparing_dependent_aliases
> behavior.
>
> After the above, it seems no relevant user of d_a_t_s_p survives that
> passes nt_transparent instead of nt_opaque, and so we can avoid adding
> a transparent_typedefs flag to the new predicate.
>
> >
> > > +bool
> > > +opaque_alias_p (const_tree t)
> > > +{
> > > + return (TYPE_P (t)
> > > + && typedef_variant_p (t)
> > > + && uses_template_parms (DECL_ORIGINAL_TYPE (TYPE_NAME (t)))
> >
> > Checking this seems wrong; a vector of dependent size seems opaque even if
> > it's a vector of int.
>
> Makes sense, fixed. I added a testcase variant too.
I forgot to mention, the related PRs PR83997 etc seem to be about
propagating/respecting attributes attached to the aliased type rather
than to the alias declaration.
I tried making dependent_opaque_alias_p also consider TYPE_ATTRIBUTES
alongside DECL_ATTRIBUTES, but that wasn't enough becuase we don't
actually call apply_late_template_attributes on the TYPE_ATTRIBUTES at
alias instantiation time and we end up ignoring them. So fixing that
seems to be a mostly orthogonal issue (and perhaps relatively low
priority since it works to just move the attribute to the decl).
>
> -- >8 --
>
> Subject: [PATCH 2/1] c++: non-template alias with dependent attributes
> [PR115897]
>
> PR c++/115897
>
> gcc/cp/ChangeLog:
>
> * cp-tree.h (dependent_opaque_alias_p): Declare.
> * pt.cc (push_template_decl): Manually mark a dependent opaque
> alias or dependent alias template specialization as dependent,
> and use structural equality for them.
> * pt.cc (dependent_opaque_alias_p): Define.
> (alias_template_specialization_p): Don't look through an
> opaque alias.
> (complex_alias_template_p): Use dependent_opaque_alias_p instead of
> any_dependent_template_arguments_p directly.
> (dependent_alias_template_spec_p): Don't look through an
> opaque alias.
> (get_underlying_template): Use dependent_opaque_alias_p instead of
> any_dependent_template_arguments_p.
> (instantiate_alias_template): Add comment mentioning similar
> behavior in push_template_decl above.
> (dependent_type_p_r): Remove dependent_alias_template_spec_p check.
> (any_template_arguments_need_structural_equality_p): Return true
> for a dependent opaque alias argument.
> (alias_ctad_tweaks): Use template_args_equal instead of same_type_p
> followed by dependent_alias_template_spec_p.
> * tree.cc (strip_typedefs): Don't strip an opaque alias.
> * typeck.cc (structural_comptypes): Compare declaration attributes
> for an opaque alias.
>
> gcc/testsuite/ChangeLog:
>
> * g++.dg/cpp0x/alias-decl-79.C: Remove xfails.
> * g++.dg/cpp0x/alias-decl-79a.C: New test.
> ---
> gcc/cp/cp-tree.h | 1 +
> gcc/cp/pt.cc | 55 +++++++++++++++------
> gcc/cp/tree.cc | 7 +--
> gcc/cp/typeck.cc | 17 +++++--
> gcc/testsuite/g++.dg/cpp0x/alias-decl-79.C | 16 +++---
> gcc/testsuite/g++.dg/cpp0x/alias-decl-79a.C | 41 +++++++++++++++
> 6 files changed, 106 insertions(+), 31 deletions(-)
> create mode 100644 gcc/testsuite/g++.dg/cpp0x/alias-decl-79a.C
>
> diff --git a/gcc/cp/cp-tree.h b/gcc/cp/cp-tree.h
> index 7d50aac4b6b..0de181ae84d 100644
> --- a/gcc/cp/cp-tree.h
> +++ b/gcc/cp/cp-tree.h
> @@ -7619,6 +7619,7 @@ extern tree instantiate_non_dependent_or_null (tree);
> extern bool variable_template_specialization_p (tree);
> extern bool alias_type_or_template_p (tree);
> enum { nt_opaque = false, nt_transparent = true };
> +extern bool dependent_opaque_alias_p (const_tree);
> extern tree alias_template_specialization_p (const_tree, bool);
> extern tree dependent_alias_template_spec_p (const_tree, bool);
> extern tree get_template_parm_object (tree expr, tree mangle);
> diff --git a/gcc/cp/pt.cc b/gcc/cp/pt.cc
> index e102e3ea490..4da3bb49ccf 100644
> --- a/gcc/cp/pt.cc
> +++ b/gcc/cp/pt.cc
> @@ -6271,6 +6271,18 @@ push_template_decl (tree decl, bool is_friend)
> }
> }
>
> + if (is_typedef_decl (decl)
> + && (dependent_opaque_alias_p (TREE_TYPE (decl))
> + || dependent_alias_template_spec_p (TREE_TYPE (decl), nt_opaque)))
> + {
> + /* Manually mark such aliases as dependent so that dependent_type_p_r
> + doesn't have to check these predicates. */
> + TYPE_DEPENDENT_P_VALID (TREE_TYPE (decl)) = true;
> + TYPE_DEPENDENT_P (TREE_TYPE (decl)) = true;
> + /* The identity of such aliases is hairy; see structural_comptypes. */
> + SET_TYPE_STRUCTURAL_EQUALITY (TREE_TYPE (decl));
> + }
> +
> if (flag_implicit_templates
> && !is_friend
> && TREE_PUBLIC (decl)
> @@ -6513,6 +6525,19 @@ alias_type_or_template_p (tree t)
> || DECL_ALIAS_TEMPLATE_P (t));
> }
>
> +/* Return true if substituting into T would yield a different type than
> + substituting into its expansion. Checking this predicate is usually
> + done in tandem with dependent_alias_template_spec_p. */
> +
> +bool
> +dependent_opaque_alias_p (const_tree t)
> +{
> + return (TYPE_P (t)
> + && typedef_variant_p (t)
> + && any_dependent_type_attributes_p (DECL_ATTRIBUTES
> + (TYPE_NAME (t))));
> +}
> +
> /* If T is a specialization of an alias template, return it; otherwise return
> NULL_TREE. If TRANSPARENT_TYPEDEFS is true, look through other aliases.
> */
>
> @@ -6530,7 +6555,7 @@ alias_template_specialization_p (const_tree t,
> if (tree tinfo = TYPE_ALIAS_TEMPLATE_INFO (t))
> if (PRIMARY_TEMPLATE_P (TI_TEMPLATE (tinfo)))
> return CONST_CAST_TREE (t);
> - if (transparent_typedefs)
> + if (transparent_typedefs && !dependent_opaque_alias_p (t))
> return alias_template_specialization_p (DECL_ORIGINAL_TYPE
> (TYPE_NAME (t)),
> transparent_typedefs);
> @@ -6635,8 +6660,7 @@ complex_alias_template_p (const_tree tmpl, tree
> *seen_out)
> return true;
>
> /* An alias with dependent type attributes is complex. */
> - if (any_dependent_type_attributes_p (DECL_ATTRIBUTES
> - (DECL_TEMPLATE_RESULT (tmpl))))
> + if (dependent_opaque_alias_p (TREE_TYPE (tmpl)))
> return true;
>
> if (!complex_alias_tmpl_info)
> @@ -6687,7 +6711,10 @@ complex_alias_template_p (const_tree tmpl, tree
> *seen_out)
> /* If T is a specialization of a complex alias template with a dependent
> argument for an unused template parameter, return it; otherwise return
> NULL_TREE. If T is a typedef to such a specialization, return the
> - specialization. */
> + specialization. Checking this predicate is usually done in tandem
> + with dependent_opaque_alias_p. Whereas dependent_opaque_alias_p checks
> + type equivalence of an alias vs its expansion, this predicate more
> + broadly checks SFINAE equivalence. */
>
> tree
> dependent_alias_template_spec_p (const_tree t, bool transparent_typedefs)
> @@ -6723,7 +6750,7 @@ dependent_alias_template_spec_p (const_tree t, bool
> transparent_typedefs)
> }
> }
>
> - if (transparent_typedefs)
> + if (transparent_typedefs && !dependent_opaque_alias_p (t))
> {
> tree utype = DECL_ORIGINAL_TYPE (TYPE_NAME (t));
> return dependent_alias_template_spec_p (utype, transparent_typedefs);
> @@ -6792,8 +6819,7 @@ get_underlying_template (tree tmpl)
> break;
>
> /* If TMPL adds dependent type attributes, it isn't equivalent. */
> - if (any_dependent_type_attributes_p (DECL_ATTRIBUTES
> - (DECL_TEMPLATE_RESULT (tmpl))))
> + if (dependent_opaque_alias_p (TREE_TYPE (tmpl)))
> break;
>
> /* Alias is equivalent. Strip it and repeat. */
> @@ -22278,6 +22304,7 @@ instantiate_alias_template (tree tmpl, tree args,
> tsubst_flags_t complain)
>
> if (tree d = dependent_alias_template_spec_p (TREE_TYPE (r), nt_opaque))
> {
> + /* Note this is also done at parse time from push_template_decl. */
> /* An alias template specialization can be dependent
> even if its underlying type is not. */
> TYPE_DEPENDENT_P (d) = true;
> @@ -27906,11 +27933,6 @@ dependent_type_p_r (tree type)
> if (TREE_CODE (type) == TYPENAME_TYPE)
> return true;
>
> - /* An alias template specialization can be dependent even if the
> - resulting type is not. */
> - if (dependent_alias_template_spec_p (type, nt_transparent))
> - return true;
> -
> /* -- a cv-qualified type where the cv-unqualified type is
> dependent.
> No code is necessary for this bullet; the code below handles
> @@ -29018,7 +29040,8 @@ any_template_arguments_need_structural_equality_p
> (tree args)
> return true;
> else if (TYPE_P (arg)
> && TYPE_STRUCTURAL_EQUALITY_P (arg)
> - && dependent_alias_template_spec_p (arg, nt_transparent))
> + && (dependent_alias_template_spec_p (arg, nt_opaque)
> + || dependent_opaque_alias_p (arg)))
> /* Require structural equality for specializations written
> in terms of a dependent alias template specialization. */
> return true;
> @@ -30418,9 +30441,9 @@ alias_ctad_tweaks (tree tmpl, tree uguides)
> A with the same template arguments. */
> ret = TREE_TYPE (TREE_TYPE (fprime));
> if (ctad_kind == alias
> - && (!same_type_p (atype, ret)
> - /* FIXME this should mean they don't compare as equivalent. */
> - || dependent_alias_template_spec_p (atype, nt_opaque)))
> + /* Use template_args_equal instead of same_type_p to get the
> + comparing_dependent_aliases behavior. */
> + && !template_args_equal (atype, ret))
> {
> tree same = finish_trait_expr (loc, CPTK_IS_DEDUCIBLE, tmpl, ret);
> ci = append_constraint (ci, same);
> diff --git a/gcc/cp/tree.cc b/gcc/cp/tree.cc
> index f2001ace6db..31ecbb1ac79 100644
> --- a/gcc/cp/tree.cc
> +++ b/gcc/cp/tree.cc
> @@ -1613,12 +1613,13 @@ strip_typedefs (tree t, bool *remove_attributes /* =
> NULL */,
> && !user_facing_original_type_p (t))
> return t;
>
> + if (dependent_opaque_alias_p (t))
> + return t;
> +
> if (alias_template_specialization_p (t, nt_opaque))
> {
> if (dependent_alias_template_spec_p (t, nt_opaque)
> - && (!(flags & STF_STRIP_DEPENDENT)
> - || any_dependent_type_attributes_p (DECL_ATTRIBUTES
> - (TYPE_NAME (t)))))
> + && !(flags & STF_STRIP_DEPENDENT))
> /* DR 1558: However, if the template-id is dependent, subsequent
> template argument substitution still applies to the template-id.
> */
> return t;
> diff --git a/gcc/cp/typeck.cc b/gcc/cp/typeck.cc
> index 8df8b871676..f26b5b2a1f4 100644
> --- a/gcc/cp/typeck.cc
> +++ b/gcc/cp/typeck.cc
> @@ -1658,8 +1658,17 @@ structural_comptypes (tree t1, tree t2, int strict)
> return false;
>
> check_alias:
> - if (comparing_dependent_aliases)
> - {
> + if (comparing_dependent_aliases
> + && (typedef_variant_p (t1) || typedef_variant_p (t2)))
> + {
> + tree dep1 = dependent_opaque_alias_p (t1) ? t1 : NULL_TREE;
> + tree dep2 = dependent_opaque_alias_p (t2) ? t2 : NULL_TREE;
> + if ((dep1 || dep2)
> + && (!(dep1 && dep2)
> + || !comp_type_attributes (DECL_ATTRIBUTES (TYPE_NAME (dep1)),
> + DECL_ATTRIBUTES (TYPE_NAME (dep2)))))
> + return false;
> +
> /* Don't treat an alias template specialization with dependent
> arguments as equivalent to its underlying type when used as a
> template argument; we need them to be distinct so that we
> @@ -1667,8 +1676,8 @@ structural_comptypes (tree t1, tree t2, int strict)
> time. And aliases can't be equivalent without being ==, so
> we don't need to look any deeper. */
> ++processing_template_decl;
> - tree dep1 = dependent_alias_template_spec_p (t1, nt_transparent);
> - tree dep2 = dependent_alias_template_spec_p (t2, nt_transparent);
> + dep1 = dependent_alias_template_spec_p (t1, nt_transparent);
> + dep2 = dependent_alias_template_spec_p (t2, nt_transparent);
> --processing_template_decl;
> if ((dep1 || dep2) && dep1 != dep2)
> return false;
> diff --git a/gcc/testsuite/g++.dg/cpp0x/alias-decl-79.C
> b/gcc/testsuite/g++.dg/cpp0x/alias-decl-79.C
> index e0f07475cc1..58436f907ef 100644
> --- a/gcc/testsuite/g++.dg/cpp0x/alias-decl-79.C
> +++ b/gcc/testsuite/g++.dg/cpp0x/alias-decl-79.C
> @@ -14,22 +14,22 @@ template<class T> struct A;
> template<class T>
> void f() {
> using B [[gnu::vector_size(16)]] = T;
> - static_assert(!is_same<T, B>::value, ""); // { dg-bogus "" "" {
> xfail *-*-* } }
> - static_assert(!is_same<A<T>, A<B>>::value, ""); // { dg-bogus "" "" {
> xfail *-*-* } }
> + static_assert(!is_same<T, B>::value, "");
> + static_assert(!is_same<A<T>, A<B>>::value, "");
> #if __cpp_variable_templates
> - static_assert(!is_same_v<T, B>, ""); // { dg-bogus "" "" {
> xfail c++14 } }
> - static_assert(!is_same_v<A<T>, A<B>>, ""); // { dg-bogus "" "" {
> xfail c++14 } }
> + static_assert(!is_same_v<T, B>, "");
> + static_assert(!is_same_v<A<T>, A<B>>, "");
> #endif
> };
>
> template<class T>
> void g() {
> using C [[gnu::vector_size(16)]] = T*;
> - static_assert(!is_same<T*, C>::value, ""); // { dg-bogus "" "" {
> xfail *-*-* } }
> - static_assert(!is_same<A<T*>, A<C>>::value, ""); // { dg-bogus "" "" {
> xfail *-*-* } }
> + static_assert(!is_same<T*, C>::value, "");
> + static_assert(!is_same<A<T*>, A<C>>::value, "");
> #if __cpp_variable_templates
> - static_assert(!is_same_v<T*, C>, ""); // { dg-bogus "" "" {
> xfail c++14 } }
> - static_assert(!is_same_v<A<T*>, A<C>>, ""); // { dg-bogus "" "" {
> xfail c++14 } }
> + static_assert(!is_same_v<T*, C>, "");
> + static_assert(!is_same_v<A<T*>, A<C>>, "");
> #endif
> };
>
> diff --git a/gcc/testsuite/g++.dg/cpp0x/alias-decl-79a.C
> b/gcc/testsuite/g++.dg/cpp0x/alias-decl-79a.C
> new file mode 100644
> index 00000000000..151b8487e1f
> --- /dev/null
> +++ b/gcc/testsuite/g++.dg/cpp0x/alias-decl-79a.C
> @@ -0,0 +1,41 @@
> +// A version of alias-decl-79.C where defining-type-id of B and C
> +// are not dependent and instead their vector_size attribute is.
> +// PR c++/115897
> +// { dg-do compile { target c++11 } }
> +
> +template<class T, class U>
> +struct is_same { static constexpr bool value = __is_same(T, U); };
> +
> +#if __cpp_variable_templates
> +template<class T, class U>
> +constexpr bool is_same_v = __is_same(T, U);
> +#endif
> +
> +template<class T> struct A;
> +
> +template<int N>
> +void f() {
> + using T = float;
> + using B [[gnu::vector_size(N * sizeof(float))]] = T;
> + static_assert(!is_same<T, B>::value, "");
> + static_assert(!is_same<A<T>, A<B>>::value, "");
> +#if __cpp_variable_templates
> + static_assert(!is_same_v<T, B>, "");
> + static_assert(!is_same_v<A<T>, A<B>>, "");
> +#endif
> +};
> +
> +template<int N>
> +void g() {
> + using T = float*;
> + using C [[gnu::vector_size(N * sizeof(float*))]] = T;
> + static_assert(!is_same<T*, C>::value, "");
> + static_assert(!is_same<A<T*>, A<C>>::value, "");
> +#if __cpp_variable_templates
> + static_assert(!is_same_v<T*, C>, "");
> + static_assert(!is_same_v<A<T*>, A<C>>, "");
> +#endif
> +};
> +
> +template void f<4>();
> +template void g<4>();
> --
> 2.46.0.rc2
>
>
