Re: [PATCH v2] c++: wrong ambiguity in accessing static field [PR112744]

[Marek Polacek <polacek@redhat.com>]

On Thu, Nov 30, 2023 at 11:42:36AM -0500, Jason Merrill wrote:
> On 11/29/23 17:01, Marek Polacek wrote:
> > On Wed, Nov 29, 2023 at 03:28:44PM -0500, Jason Merrill wrote:
> > > On 11/29/23 12:43, Marek Polacek wrote:
> > > > On Wed, Nov 29, 2023 at 12:23:46PM -0500, Patrick Palka wrote:
> > > > > On Wed, 29 Nov 2023, Marek Polacek wrote:
> > > > > 
> > > > > > Bootstrapped/regtested on x86_64-pc-linux-gnu, ok for trunk?
> > > > > > 
> > > > > > Now that I'm posting this patch, I think you'll probably want me to use
> > > > > > ba_any unconditionally.  That works too; g++.dg/tc1/dr52.C just needs
> > > > > > a trivial testsuite tweak:
> > > > > >     'C' is not an accessible base of 'X'
> > > > > > v.
> > > > > >     'C' is an inaccessible base of 'X'
> > > > > > We should probably unify those messages...
> > > > > > 
> > > > > > -- >8 --
> > > > > > Given
> > > > > > 
> > > > > >     struct A { constexpr static int a = 0; };
> > > > > >     struct B : A {};
> > > > > >     struct C : A {};
> > > > > >     struct D : B, C {};
> > > > > > 
> > > > > > we give the "'A' is an ambiguous base of 'D'" error for
> > > > > > 
> > > > > >     D{}.A::a;
> > > > > > 
> > > > > > which seems wrong: 'a' is a static data member so there is only one copy
> > > > > > so it can be unambiguously referred to even if there are multiple A
> > > > > > objects.  clang++/MSVC/icx agree.
> > > > > > 
> > > > > > 	PR c++/112744
> > > > > > 
> > > > > > gcc/cp/ChangeLog:
> > > > > > 
> > > > > > 	* typeck.cc (finish_class_member_access_expr): When accessing
> > > > > > 	a static data member, use ba_any for lookup_base.
> > > > > > 
> > > > > > gcc/testsuite/ChangeLog:
> > > > > > 
> > > > > > 	* g++.dg/lookup/scoped11.C: New test.
> > > > > > 	* g++.dg/lookup/scoped12.C: New test.
> > > > > > 	* g++.dg/lookup/scoped13.C: New test.
> > > > > > ---
> > > > > >    gcc/cp/typeck.cc                       | 21 ++++++++++++++++++---
> > > > > >    gcc/testsuite/g++.dg/lookup/scoped11.C | 14 ++++++++++++++
> > > > > >    gcc/testsuite/g++.dg/lookup/scoped12.C | 14 ++++++++++++++
> > > > > >    gcc/testsuite/g++.dg/lookup/scoped13.C | 14 ++++++++++++++
> > > > > >    4 files changed, 60 insertions(+), 3 deletions(-)
> > > > > >    create mode 100644 gcc/testsuite/g++.dg/lookup/scoped11.C
> > > > > >    create mode 100644 gcc/testsuite/g++.dg/lookup/scoped12.C
> > > > > >    create mode 100644 gcc/testsuite/g++.dg/lookup/scoped13.C
> > > > > > 
> > > > > > diff --git a/gcc/cp/typeck.cc b/gcc/cp/typeck.cc
> > > > > > index e995fb6ddd7..c4de8bb2616 100644
> > > > > > --- a/gcc/cp/typeck.cc
> > > > > > +++ b/gcc/cp/typeck.cc
> > > > > > @@ -3476,7 +3476,7 @@ finish_class_member_access_expr (cp_expr object, tree name, bool template_p,
> > > > > >    			   name, scope);
> > > > > >    		  return error_mark_node;
> > > > > >    		}
> > > > > > -	
> > > > > > +
> > > > > >    	      if (TREE_SIDE_EFFECTS (object))
> > > > > >    		val = build2 (COMPOUND_EXPR, TREE_TYPE (val), object, val);
> > > > > >    	      return val;
> > > > > > @@ -3493,9 +3493,24 @@ finish_class_member_access_expr (cp_expr object, tree name, bool template_p,
> > > > > >    	      return error_mark_node;
> > > > > >    	    }
> > > > > > +	  /* NAME may refer to a static data member, in which case there is
> > > > > > +	     one copy of the data member that is shared by all the objects of
> > > > > > +	     the class.  So NAME can be unambiguously referred to even if
> > > > > > +	     there are multiple indirect base classes containing NAME.  */
> > > > > > +	  const base_access ba = [scope, name] ()
> > > > > > +	    {
> > > > > > +	      if (identifier_p (name))
> > > > > > +		{
> > > > > > +		  tree m = lookup_member (scope, name, /*protect=*/0,
> > > > > > +					  /*want_type=*/false, tf_none);
> > > > > > +		  if (!m || VAR_P (m))
> > > > > > +		    return ba_any;
> > > > > 
> > > > > I wonder if we want to return ba_check_bit instead of ba_any so that we
> > > > > still check access of the selected base?
> > > > 
> > > > That would certainly make sense to me.  I didn't do that because
> > > > I'd not seen ba_check_bit being used except as part of ba_check,
> > > > but that may not mean much.
> > > > 
> > > > So either I can tweak the lambda to return ba_check_bit rather
> > > > than ba_any or use ba_check_bit unconditionally.  Any opinions on that?
> > > 
> > > The relevant passage seems to be
> > > https://eel.is/c++draft/class.access.base#6
> > > after DR 52, which seems to have clarified that the pointer conversion only
> > > applies to non-static members.
> > > 
> > > > >     struct A { constexpr static int a = 0; };
> > > > >     struct D : private A {};
> > > > > 
> > > > >     void f() {
> > > > >       D{}.A::a; // #1 GCC (and Clang) currently rejects
> > > > >     }
> > > 
> > > I see that MSVC also rejects it, while EDG accepts.
> > > 
> > > https://eel.is/c++draft/class.access.base#5.1 seems to say that a is
> > > accessible when named in A.
> > > 
> > > https://eel.is/c++draft/expr.ref#7 also only constrains references to
> > > non-static members.
> > > 
> > > But first we need to look up A in D, and A's injected-class-name looked up
> > > as a member of D is not accessible; it's private, and f() is not a friend,
> > > and we correctly complain about that.
> > > 
> > > If we avoid the lookup of A in D with
> > > 
> > > D{}.::A::a;
> > > 
> > > clang accepts it, which is consistent with accepting the template version,
> > > and seems correct.
> > > 
> > > So, I think ba_any is what we want here.
> > 
> > Wow, that is not intuitive (to me at least).  So I had it right but
> > only by accident.
> > 
> > Bootstrapped/regtested on x86_64-pc-linux-gnu, ok for trunk?
> > 
> > -- >8 --
> > Given
> > 
> >    struct A { constexpr static int a = 0; };
> >    struct B : A {};
> >    struct C : A {};
> >    struct D : B, C {};
> > 
> > we give the "'A' is an ambiguous base of 'D'" error for
> > 
> >    D{}.A::a;
> > 
> > which seems wrong: 'a' is a static data member so there is only one copy
> > so it can be unambiguously referred to even if there are multiple A
> > objects.  clang++/MSVC/icx agree.
> > 
> > The rationale for using ba_any is explained at
> > <https://gcc.gnu.org/pipermail/gcc-patches/2023-November/638676.html>.
> 
> I'd prefer not to cite the mailing list for rationales.
> 
> To summarize:
> [class.access.base] requires conversion to a unique base subobject for
> non-static data members, but it does not require that the base be unique or
> accessible for static data members.
> 
> > 	PR c++/112744
> > 
> > gcc/cp/ChangeLog:
> > 
> > 	* typeck.cc (finish_class_member_access_expr): When accessing
> > 	a static data member, use ba_any for lookup_base.
> > 
> > gcc/testsuite/ChangeLog:
> > 
> > 	* g++.dg/lookup/scoped11.C: New test.
> > 	* g++.dg/lookup/scoped12.C: New test.
> > 	* g++.dg/lookup/scoped13.C: New test.
> > 	* g++.dg/lookup/scoped14.C: New test.
> > 	* g++.dg/lookup/scoped15.C: New test.
> > ---
> >   gcc/cp/typeck.cc                       | 21 ++++++++++++++++++---
> >   gcc/testsuite/g++.dg/lookup/scoped11.C | 14 ++++++++++++++
> >   gcc/testsuite/g++.dg/lookup/scoped12.C | 14 ++++++++++++++
> >   gcc/testsuite/g++.dg/lookup/scoped13.C | 14 ++++++++++++++
> >   gcc/testsuite/g++.dg/lookup/scoped14.C | 14 ++++++++++++++
> >   gcc/testsuite/g++.dg/lookup/scoped15.C | 20 ++++++++++++++++++++
> >   6 files changed, 94 insertions(+), 3 deletions(-)
> >   create mode 100644 gcc/testsuite/g++.dg/lookup/scoped11.C
> >   create mode 100644 gcc/testsuite/g++.dg/lookup/scoped12.C
> >   create mode 100644 gcc/testsuite/g++.dg/lookup/scoped13.C
> >   create mode 100644 gcc/testsuite/g++.dg/lookup/scoped14.C
> >   create mode 100644 gcc/testsuite/g++.dg/lookup/scoped15.C
> > 
> > diff --git a/gcc/cp/typeck.cc b/gcc/cp/typeck.cc
> > index 0839d0a4167..bf8ffaa7e75 100644
> > --- a/gcc/cp/typeck.cc
> > +++ b/gcc/cp/typeck.cc
> > @@ -3467,7 +3467,7 @@ finish_class_member_access_expr (cp_expr object, tree name, bool template_p,
> >   			   name, scope);
> >   		  return error_mark_node;
> >   		}
> > -	
> > +
> >   	      if (TREE_SIDE_EFFECTS (object))
> >   		val = build2 (COMPOUND_EXPR, TREE_TYPE (val), object, val);
> >   	      return val;
> > @@ -3484,9 +3484,24 @@ finish_class_member_access_expr (cp_expr object, tree name, bool template_p,
> >   	      return error_mark_node;
> >   	    }
> > +	  /* NAME may refer to a static data member, in which case there is
> > +	     one copy of the data member that is shared by all the objects of
> > +	     the class.  So NAME can be unambiguously referred to even if
> > +	     there are multiple indirect base classes containing NAME.  */
> > +	  const base_access ba = [scope, name] ()
> > +	    {
> > +	      if (identifier_p (name)) > +		{
> > +		  tree m = lookup_member (scope, name, /*protect=*/0,
> > +					  /*want_type=*/false, tf_none);
> > +		  if (!m || shared_member_p (m))
> > +		    return ba_any;
> > +		}
> > +	      return ba_check;
> > +	    } ();
> > +
> >   	  /* Find the base of OBJECT_TYPE corresponding to SCOPE.  */
> > -	  access_path = lookup_base (object_type, scope, ba_check,
> > -				     NULL, complain);
> > +	  access_path = lookup_base (object_type, scope, ba, NULL, complain);
> >   	  if (access_path == error_mark_node)
> >   	    return error_mark_node;
> >   	  if (!access_path)
> > diff --git a/gcc/testsuite/g++.dg/lookup/scoped11.C b/gcc/testsuite/g++.dg/lookup/scoped11.C
> > new file mode 100644
> > index 00000000000..be743522fce
> > --- /dev/null
> > +++ b/gcc/testsuite/g++.dg/lookup/scoped11.C
> > @@ -0,0 +1,14 @@
> > +// PR c++/112744
> > +// { dg-do compile }
> > +
> > +struct A { const static int a = 0; };
> > +struct B : A {};
> > +struct C : A {};
> > +struct D : B, C {};
> > +
> > +int main()
> > +{
> > +  D d;
> > +  (void) d.a;
> > +  (void) d.A::a;
> > +}
> > diff --git a/gcc/testsuite/g++.dg/lookup/scoped12.C b/gcc/testsuite/g++.dg/lookup/scoped12.C
> > new file mode 100644
> > index 00000000000..ffa145598fd
> > --- /dev/null
> > +++ b/gcc/testsuite/g++.dg/lookup/scoped12.C
> > @@ -0,0 +1,14 @@
> > +// PR c++/112744
> > +// { dg-do compile }
> > +
> > +class A { const static int a = 0; };
> > +struct B : A {};
> > +struct C : A {};
> > +struct D : B, C {};
> > +
> > +int main()
> > +{
> > +  D d;
> > +  (void) d.a;	  // { dg-error "private" }
> > +  (void) d.A::a;  // { dg-error "private" }
> > +}
> > diff --git a/gcc/testsuite/g++.dg/lookup/scoped13.C b/gcc/testsuite/g++.dg/lookup/scoped13.C
> > new file mode 100644
> > index 00000000000..970e1aa833e
> > --- /dev/null
> > +++ b/gcc/testsuite/g++.dg/lookup/scoped13.C
> > @@ -0,0 +1,14 @@
> > +// PR c++/112744
> > +// { dg-do compile }
> > +
> > +struct A { const static int a = 0; };
> > +struct B : A {};
> > +struct C : A {};
> > +struct D : B, C {};
> > +
> > +int main()
> > +{
> > +  D d;
> > +  (void) d.x;	  // { dg-error ".struct D. has no member named .x." }
> > +  (void) d.A::x;  // { dg-error ".struct A. has no member named .x." }
> > +}
> > diff --git a/gcc/testsuite/g++.dg/lookup/scoped14.C b/gcc/testsuite/g++.dg/lookup/scoped14.C
> > new file mode 100644
> > index 00000000000..141aa0d2b1a
> > --- /dev/null
> > +++ b/gcc/testsuite/g++.dg/lookup/scoped14.C
> > @@ -0,0 +1,14 @@
> > +// PR c++/112744
> > +// { dg-do compile { target c++11 } }
> > +
> > +struct A { int a = 0; };
> > +struct B : A {};
> > +struct C : A {};
> > +struct D : B, C {};
> > +
> > +int main()
> > +{
> > +  D d;
> > +  (void) d.a;	  // { dg-error "request for member .a. is ambiguous" }
> > +  (void) d.A::a;  // { dg-error ".A. is an ambiguous base of .D." }
> > +}
> > diff --git a/gcc/testsuite/g++.dg/lookup/scoped15.C b/gcc/testsuite/g++.dg/lookup/scoped15.C
> > new file mode 100644
> > index 00000000000..d450a41a617
> > --- /dev/null
> > +++ b/gcc/testsuite/g++.dg/lookup/scoped15.C
> > @@ -0,0 +1,20 @@
> > +// PR c++/112744
> > +// { dg-do compile { target c++11 } }
> > +
> > +struct A { constexpr static int a = 0; };
> > +struct D : private A {};
> > +
> > +// See <https://gcc.gnu.org/pipermail/gcc-patches/2023-November/638676.html>
> > +// for rationale.
> 
> The injected-class-name of A is private when named in D, but if A is named
> some other way, there is no requirement in [class.access.base] for static
> data members that it be an accessible base.
> 
> OK with those comment adjustments.

Thanks; pushed with that adjusted.

I'm not planning to backport the patch.

Marek