On 3/2/21 4:51 PM, Marek Polacek wrote:
On Mon, Mar 01, 2021 at 09:29:19PM -0500, Jason Merrill via Gcc-patches wrote:
On 3/1/21 7:59 PM, Marek Polacek wrote:
On Mon, Mar 01, 2021 at 03:08:58PM -0500, Jason Merrill wrote:
On 3/1/21 2:54 PM, Marek Polacek wrote:
On Thu, Feb 25, 2021 at 10:45:29PM -0500, Jason Merrill via Gcc-patches wrote:
On 2/25/21 5:41 PM, Marek Polacek wrote:
On Thu, Feb 25, 2021 at 10:59:49AM -0500, Jason Merrill wrote:
On 2/12/21 6:12 PM, Marek Polacek wrote:
We represent deduction guides with FUNCTION_DECLs, but they are built
without DECL_CONTEXT

Hmm, that seems wrong: "A deduction-guide shall be declared in the
same scope as the corresponding class template and, for a member class
template, with the same access."  But it probably isn't necessary to change
this:

leading to an ICE in type_dependent_expression_p
on the assert that the type of a function template with no dependent
(innermost!) template arguments must be non-dependent.  Consider the
attached class-deduction79.C: we create a deduction guide:

       template<class T> G(T)-> E<Z>::G<T>

we deduce T and create a partial instantiation:

       G(T) -> E<Z>::G<T> [with T = int]

And then do_class_deduction wants to create a CALL_EXPR from the above
using build_new_function_call -> build_over_call which calls mark_used
-> maybe_instantiate_noexcept -> type_dependent_expression_p.

There, the innermost template arguments are non-dependent (<int>), but
the fntype is dependent -- the return type is a TYPENAME_TYPE, and
since we have no DECL_CONTEXT, this check holds:

       /* Otherwise, if the function decl isn't from a dependent scope, it 
can't be
          type-dependent.  Checking this is important for functions with auto 
return
          type, which looks like a dependent type.  */
       if (TREE_CODE (expression) == FUNCTION_DECL
           && !(DECL_CLASS_SCOPE_P (expression)
                && dependent_type_p (DECL_CONTEXT (expression)))

whereupon we ICE.

Experiments with setting DECL_CONTEXT didn't pan out.

In c8 of the PR it looks like you were using the class itself as
DECL_CONTEXT; the quote above says that the right context is the enclosing
scope of the class.

Sadly, using CP_TYPE_CONTEXT (type) would result in a crash in
retrieve_specialization:

      /* There should be as many levels of arguments as there are
         levels of parameters.  */
      gcc_assert (TMPL_ARGS_DEPTH (args)
                  == (TREE_CODE (tmpl) == TEMPLATE_DECL
                      ? TMPL_PARMS_DEPTH (DECL_TEMPLATE_PARMS (tmpl))
                      : template_class_depth (DECL_CONTEXT (tmpl))));

Yeah, probably simpler not to bother.

So perhaps we
just want to skip the assert for deduction guides, because they are
a little special.  Better ideas solicited.

In c3 you mention that one of the variants broke with r269093; this is
because my change to check CLASSTYPE_TEMPLATE_INSTANTIATION is false for the
template pattern itself (E<Z>).

And the original test started with my r11-1713 because using TREE_TYPE
directly instead of decltype (which is a non-deduced context) means we
can deduced from the argument.
But I think probably the right answer is to defer this deduction until the
enclosing scope is non-dependent, i.e. (untested)

Thanks.  That mostly works, except the new class-deduction-aggr[89].C
tests.  Consider 8:

namespace N {
template <typename, typename> struct S {
      template <typename T, typename U> S(T, U);
};
} // namespace N
template <int> struct E {
      template <typename T> struct G { T t; };
      void fn() { G{N::S<char, int>{'a', 1}}; }
};

void
g ()
{
      E<1> e;
      e.fn ();
}

With your patch, when in do_class_deduction when processing_template_decl,
we just return.  When we call do_class_deduction again when p_t_d is 0,
maybe_aggr_guide returns early here:

      if (!CP_AGGREGATE_TYPE_P (type))
        return NULL_TREE

because G is not complete (and rightly so, we didn't instantiate it).  So
we aren't able to deduce the template parameters.  I'm not sure if I should
pursue this direction further.  :(

I think so; we just need to test CP_AGGREGATE_TYPE_P on the original
template pattern instead of the instantiation E<1>::G.

I'm sorry, I've got stuck again.

Yes, using the original template pattern helps us get past the
CP_AGGREGATE_TYPE_P check.

However, using TREE_TYPE (DECL_TI_TEMPLATE (tmpl)) as the type of the deduction 
guide
means the guide will be "template<class T> G(T)-> E<<anonymous> >::G<T>" which
results in

     class-deduction-aggr8.C:11:15: error: invalid use of dependent type 'typename 
E<<anonymous> >::G<N::S<char, int> >'

which makes sense I guess: when we defer building up the guide until
we've instantiated E<1>, finding the dependent type E<> is not expected.

Yeah, I was only thinking to use the pattern for the aggregate check.

Ack.  Though I think I also have to use the pattern here:

    init = reshape_init (type, init, complain);

otherwise reshape_init returns a TARGET_EXPR and we immediately
crash in collect_ctor_idx_types because that only expects a CONSTRUCTOR.
And what we need to get is the type T -- of the constructor's index.
But creating the guide with "struct E<1>::G<T>" as its type seems
wrong; I'm not even sure if a guide like

     template<class T> G(T)-> E<1>::G<T>

makes sense.

It looks fine to me.

In any case the deduction fails (when we call
build_new_function_call in do_class_deduction), because we've got
a mismatch: the template parameter T has level 1, but the template
function parameter has level 2.

Sure, because E<1>::G<T> has been partially instantiated, so the T has been
reduced from level 2 to 1.

Right.
You'll need to do a similar partial instantiation for building the deduction
guide, either as part of the deduction guide rewriting or on the constructor
before rewriting.

So I've tried.  I can't actually tsubst the constructor itself, because it
at this point contains an AGGR_INIT_EXPR, which tsubst* can't handle.  But
what I could do is

    parms = tsubst (parms, DECL_TI_ARGS (tmpl), complain, init);

just after the call to collect_ctor_idx_types.  After all, this is what
we care about and create the function template parameters from.  So now
T's level is reduced to 1, and the guide we create is

    template<class T> G(T)-> E<1>::G<T>

This guide is then chosen in do_class_deduction -> build_new_function_call,
but we crash in fn_type_unification -> instantiate_template (after we've
deduced T to N::S<char, int>) in retrieve_specialization:

    gcc_assert (TMPL_ARGS_DEPTH (args)
                == (TREE_CODE (tmpl) == TEMPLATE_DECL
                    ? TMPL_PARMS_DEPTH (DECL_TEMPLATE_PARMS (tmpl))
                    : template_class_depth (DECL_CONTEXT (tmpl))));

args is <1, S> (depth 2), tmpl is our deduction guide, and
DECL_TEMPLATE_PARMS (tmpl) is [1 T] (depth 1).

I'd think args should only be S (depth 1).

Ah!  I think I finally got a working patch.  Thanks for all the help!

Added class-deduction-aggr10.C to test the deduction in more nested classes.

Bootstrapped/regtested on x86_64-pc-linux-gnu, ok for trunk?
Additionally, I've built cmcstl2 and Boost 1.75.

OK, thanks.

-- >8 --
We represent deduction guides with FUNCTION_DECLs, but they are built
without DECL_CONTEXT, leading to an ICE in type_dependent_expression_p
on the assert that the type of a function template with no dependent
(innermost!) template arguments must be non-dependent.  Consider the
attached class-deduction79.C: we create a deduction guide:

   template<class T> G(T)-> E<Z>::G<T>

we deduce T and create a partial instantiation:

   G(T) -> E<Z>::G<T> [with T = int]

And then do_class_deduction wants to create a CALL_EXPR from the above
using build_new_function_call -> build_over_call which calls mark_used
-> maybe_instantiate_noexcept -> type_dependent_expression_p.

There, the innermost template arguments are non-dependent (<int>), but
the fntype is dependent -- the return type is a TYPENAME_TYPE, and
since we have no DECL_CONTEXT, this check holds:

   /* Otherwise, if the function decl isn't from a dependent scope, it can't be
      type-dependent.  Checking this is important for functions with auto return
      type, which looks like a dependent type.  */
   if (TREE_CODE (expression) == FUNCTION_DECL
       && !(DECL_CLASS_SCOPE_P (expression)
            && dependent_type_p (DECL_CONTEXT (expression)))

whereupon we ICE.

This patch fixes it by deferring the class deduction until the
enclosing scope is non-dependent.  build_deduction_guide and maybe_aggr_guide
needed a little tweaking to make the deduction work in a member
template.

Co-Authored-By: Jason Merrill <ja...@redhat.com>

gcc/cp/ChangeLog:

        PR c++/97034
        PR c++/99009
        * pt.c (build_deduction_guide): Use INNERMOST_TEMPLATE_ARGS.
        (maybe_aggr_guide): Use the original template type where needed.  In
        a class member template, partially instantiate the result of
        collect_ctor_idx_types.
        (do_class_deduction): Defer the deduction until the enclosing
        scope is non-dependent.

gcc/testsuite/ChangeLog:

        PR c++/97034
        PR c++/99009
        * g++.dg/cpp1z/class-deduction81.C: New test.
        * g++.dg/cpp1z/class-deduction82.C: New test.
        * g++.dg/cpp2a/class-deduction-aggr8.C: New test.
        * g++.dg/cpp2a/class-deduction-aggr9.C: New test.
        * g++.dg/cpp2a/class-deduction-aggr10.C: New test.
---
  gcc/cp/pt.c                                   | 36 +++++++++++++++++--
  .../g++.dg/cpp1z/class-deduction81.C          | 20 +++++++++++
  .../g++.dg/cpp1z/class-deduction82.C          | 12 +++++++
  .../g++.dg/cpp2a/class-deduction-aggr10.C     | 21 +++++++++++
  .../g++.dg/cpp2a/class-deduction-aggr8.C      | 19 ++++++++++
  .../g++.dg/cpp2a/class-deduction-aggr9.C      | 18 ++++++++++
  6 files changed, 123 insertions(+), 3 deletions(-)
  create mode 100644 gcc/testsuite/g++.dg/cpp1z/class-deduction81.C
  create mode 100644 gcc/testsuite/g++.dg/cpp1z/class-deduction82.C
  create mode 100644 gcc/testsuite/g++.dg/cpp2a/class-deduction-aggr10.C
  create mode 100644 gcc/testsuite/g++.dg/cpp2a/class-deduction-aggr8.C
  create mode 100644 gcc/testsuite/g++.dg/cpp2a/class-deduction-aggr9.C

diff --git a/gcc/cp/pt.c b/gcc/cp/pt.c
index 8d65a6e5bd2..a4686e0affb 100644
--- a/gcc/cp/pt.c
+++ b/gcc/cp/pt.c
@@ -28643,7 +28643,7 @@ build_deduction_guide (tree type, tree ctor, tree 
outer_args, tsubst_flags_t com
tree ctmpl = CLASSTYPE_TI_TEMPLATE (type);
        tparms = DECL_TEMPLATE_PARMS (ctmpl);
-      targs = CLASSTYPE_TI_ARGS (type);
+      targs = INNERMOST_TEMPLATE_ARGS (CLASSTYPE_TI_ARGS (type));
        ci = NULL_TREE;
        fargs = NULL_TREE;
        loc = DECL_SOURCE_LOCATION (ctmpl);
@@ -28866,8 +28866,22 @@ maybe_aggr_guide (tree tmpl, tree init, 
vec<tree,va_gc> *args)
    if (init == NULL_TREE)
      return NULL_TREE;
+ /* We might be creating a guide for a class member template, e.g.,
+
+       template<typename U> struct A {
+        template<typename T> struct B { T t; };
+       };
+
+     At this point, A will have been instantiated.  Below, we need to
+     use both A<U>::B<T> (TEMPLATE_TYPE) and A<int>::B<T> (TYPE) types.  */
+  const bool member_template_p
+    = (DECL_TEMPLATE_INFO (tmpl)
+       && DECL_MEMBER_TEMPLATE_P (DECL_TI_TEMPLATE (tmpl)));
    tree type = TREE_TYPE (tmpl);
-  if (!CP_AGGREGATE_TYPE_P (type))
+  tree template_type = (member_template_p
+                       ? TREE_TYPE (DECL_TI_TEMPLATE (tmpl))
+                       : type);
+  if (!CP_AGGREGATE_TYPE_P (template_type))
      return NULL_TREE;
/* No aggregate candidate for copy-initialization. */
@@ -28884,10 +28898,21 @@ maybe_aggr_guide (tree tmpl, tree init, 
vec<tree,va_gc> *args)
    tree parms = NULL_TREE;
    if (BRACE_ENCLOSED_INITIALIZER_P (init))
      {
-      init = reshape_init (type, init, complain);
+      init = reshape_init (template_type, init, complain);
        if (init == error_mark_node)
        return NULL_TREE;
        parms = collect_ctor_idx_types (init, parms);
+      /* If we're creating a deduction guide for a member class template,
+        we've used the original template pattern type for the reshape_init
+        above; this is done because we want PARMS to be a template parameter
+        type, something that can be deduced when used as a function template
+        parameter.  At this point the outer class template has already been
+        partially instantiated (we deferred the deduction until the enclosing
+        scope is non-dependent).  Therefore we have to partially instantiate
+        PARMS, so that its template level is properly reduced and we don't get
+        mismatches when deducing types using the guide with PARMS.  */
+      if (member_template_p)
+       parms = tsubst (parms, DECL_TI_ARGS (tmpl), complain, init);
      }
    else if (TREE_CODE (init) == TREE_LIST)
      {
@@ -29225,6 +29250,11 @@ do_class_deduction (tree ptype, tree tmpl, tree init,
    if (DECL_TEMPLATE_TEMPLATE_PARM_P (tmpl))
      return ptype;
+ /* Wait until the enclosing scope is non-dependent. */
+  if (DECL_CLASS_SCOPE_P (tmpl)
+      && dependent_type_p (DECL_CONTEXT (tmpl)))
+    return ptype;
+
    /* Initializing one placeholder from another.  */
    if (init && TREE_CODE (init) == TEMPLATE_PARM_INDEX
        && is_auto (TREE_TYPE (init))
diff --git a/gcc/testsuite/g++.dg/cpp1z/class-deduction81.C 
b/gcc/testsuite/g++.dg/cpp1z/class-deduction81.C
new file mode 100644
index 00000000000..86a68248157
--- /dev/null
+++ b/gcc/testsuite/g++.dg/cpp1z/class-deduction81.C
@@ -0,0 +1,20 @@
+// PR c++/97034
+// { dg-do compile { target c++17 } }
+
+template <typename Z>
+struct E {
+  template <typename T>
+  struct G {
+    T t;
+    G(T) { }
+  };
+
+  void fn() { G{1}; }
+};
+
+void
+g ()
+{
+  E<int> e;
+  e.fn ();
+}
diff --git a/gcc/testsuite/g++.dg/cpp1z/class-deduction82.C 
b/gcc/testsuite/g++.dg/cpp1z/class-deduction82.C
new file mode 100644
index 00000000000..238024c508f
--- /dev/null
+++ b/gcc/testsuite/g++.dg/cpp1z/class-deduction82.C
@@ -0,0 +1,12 @@
+// PR c++/99009
+// { dg-do compile { target c++17 } }
+
+template<typename> struct B {
+  B(int = A()) {}
+  template <typename ...> struct A;
+};
+
+template<typename T> struct X {
+  template <T...> struct Y;
+  X() { Y y; };
+};
diff --git a/gcc/testsuite/g++.dg/cpp2a/class-deduction-aggr10.C 
b/gcc/testsuite/g++.dg/cpp2a/class-deduction-aggr10.C
new file mode 100644
index 00000000000..be922bbfb73
--- /dev/null
+++ b/gcc/testsuite/g++.dg/cpp2a/class-deduction-aggr10.C
@@ -0,0 +1,21 @@
+// PR c++/97034
+// { dg-do compile { target c++20 } }
+
+namespace N {
+template <typename, typename> struct S {
+  template <typename T, typename U> S(T, U);
+};
+} // namespace N
+template <int I> struct E {
+  template<typename U> struct M {
+    template <typename T> struct G { T t; };
+    void fn() { G{N::S<char, int>{'a', 1}}; }
+  };
+};
+
+void
+g ()
+{
+  E<1>::M<int> m;
+  m.fn ();
+}
diff --git a/gcc/testsuite/g++.dg/cpp2a/class-deduction-aggr8.C 
b/gcc/testsuite/g++.dg/cpp2a/class-deduction-aggr8.C
new file mode 100644
index 00000000000..399061100ae
--- /dev/null
+++ b/gcc/testsuite/g++.dg/cpp2a/class-deduction-aggr8.C
@@ -0,0 +1,19 @@
+// PR c++/97034
+// { dg-do compile { target c++20 } }
+
+namespace N {
+template <typename, typename> struct S {
+  template <typename T, typename U> S(T, U);
+};
+} // namespace N
+template <int> struct E {
+  template <typename T> struct G { T t; };
+  void fn() { G{N::S<char, int>{'a', 1}}; }
+};
+
+void
+g ()
+{
+  E<1> e;
+  e.fn ();
+}
diff --git a/gcc/testsuite/g++.dg/cpp2a/class-deduction-aggr9.C 
b/gcc/testsuite/g++.dg/cpp2a/class-deduction-aggr9.C
new file mode 100644
index 00000000000..245a04cd5f9
--- /dev/null
+++ b/gcc/testsuite/g++.dg/cpp2a/class-deduction-aggr9.C
@@ -0,0 +1,18 @@
+// PR c++/97034
+// { dg-do compile { target c++20 } }
+
+template<typename>
+struct E {
+  template <typename T>
+  struct G {
+    T t;
+  };
+
+  void fn() { G{1}; }
+};
+
+void
+g () {
+  E<int> e;
+  e.fn ();
+}

base-commit: 41fbacdd10305654b1d10f887fa3f4677f9b8f34


Reply via email to