[PATCH] c++: Pointer-to-member fn conversion with noexcept [PR99374]

[PATCH] c++: Pointer-to-member fn conversion with noexcept [PR99374]

[Marek Polacek]

The issue in this PR is that we wrongly reject converting pointers to
member function of incomplete types, one of which has noexcept.  Recall
that pointers (including pointers to member functions) to non-throwing
functions can be implicitly converted to potentially-throwing functions
(but not vice versa).

We reject the conversion when called from can_convert_arg_bad because
standard_conversion can't create such a conversion.  It comes down to
the DERIVED_FROM_P check in the TYPE_PTRMEMFUNC_P block.  It considers
every class derived from itself, but not when the class is incomplete.
But surely we want to reach fnptr_conv_p when tbase is fbase (one of
them could be an alias to the other so use same_type_p instead of ==).

Another approach would be to not perform DERIVED_FROM_P at all when
either tbase or fbase are incomplete (so perhaps something like at the
end of ptr_reasonably_similar).

Bootstrapped/regtested on x86_64-pc-linux-gnu, ok for trunk/10?

gcc/cp/ChangeLog:

	PR c++/99374
	* call.c (standard_conversion): When converting pointers to
	member, don't return NULL when the bases are equivalent but
	incomplete.

gcc/testsuite/ChangeLog:

	PR c++/99374
	* g++.dg/cpp1z/noexcept-type23.C: New test.
---
 gcc/cp/call.c                                |  4 +++-
 gcc/testsuite/g++.dg/cpp1z/noexcept-type23.C | 14 ++++++++++++++
 2 files changed, 17 insertions(+), 1 deletion(-)
 create mode 100644 gcc/testsuite/g++.dg/cpp1z/noexcept-type23.C

diff --git a/gcc/cp/call.c b/gcc/cp/call.c
index 123f06b1f2b..605cbe15f4e 100644
--- a/gcc/cp/call.c
+++ b/gcc/cp/call.c
@@ -1449,7 +1449,9 @@ standard_conversion (tree to, tree from, tree expr, bool c_cast_p,
       tree fbase = class_of_this_parm (fromfn);
       tree tbase = class_of_this_parm (tofn);
 
-      if (!DERIVED_FROM_P (fbase, tbase))
+      /* If FBASE and TBASE are equivalent but incomplete, DERIVED_FROM_P
+	 yields false.  But a pointer to member of incomplete class is OK.  */
+      if (!DERIVED_FROM_P (fbase, tbase) && !same_type_p (fbase, tbase))
 	return NULL;
 
       tree fstat = static_fn_type (fromfn);
diff --git a/gcc/testsuite/g++.dg/cpp1z/noexcept-type23.C b/gcc/testsuite/g++.dg/cpp1z/noexcept-type23.C
new file mode 100644
index 00000000000..612dd6ceb5e
--- /dev/null
+++ b/gcc/testsuite/g++.dg/cpp1z/noexcept-type23.C
@@ -0,0 +1,14 @@
+// PR c++/99374
+// { dg-do compile { target c++17 } }
+
+struct S;
+struct R;
+using F1 = int (S::*)();
+using F2 = int (S::*)() noexcept;
+using F3 = int (R::*)() noexcept;
+using T = S;
+using F4 = int (T::*)() noexcept;
+F1 f21 = F2();
+F1 f41 = F4();
+F2 f12 = F1(); // { dg-error "cannot convert" }
+F1 f31 = F3(); // { dg-error "cannot convert" }

base-commit: 0d737ed2171165ba39ab5647f8a94c588fc9a898
-- 
2.29.2

Re: [PATCH] c++: Pointer-to-member fn conversion with noexcept [PR99374]

[Jason Merrill]

On 3/4/21 6:19 PM, Marek Polacek wrote:
> The issue in this PR is that we wrongly reject converting pointers to
> member function of incomplete types, one of which has noexcept.  Recall
> that pointers (including pointers to member functions) to non-throwing
> functions can be implicitly converted to potentially-throwing functions
> (but not vice versa).
> 
> We reject the conversion when called from can_convert_arg_bad because
> standard_conversion can't create such a conversion.  It comes down to
> the DERIVED_FROM_P check in the TYPE_PTRMEMFUNC_P block.  It considers
> every class derived from itself, but not when the class is incomplete.
> But surely we want to reach fnptr_conv_p when tbase is fbase (one of
> them could be an alias to the other so use same_type_p instead of ==).
> 
> Another approach would be to not perform DERIVED_FROM_P at all when
> either tbase or fbase are incomplete (so perhaps something like at the
> end of ptr_reasonably_similar).
> 
> Bootstrapped/regtested on x86_64-pc-linux-gnu, ok for trunk/10?
> 
> gcc/cp/ChangeLog:
> 
> 	PR c++/99374
> 	* call.c (standard_conversion): When converting pointers to
> 	member, don't return NULL when the bases are equivalent but
> 	incomplete.
> 
> gcc/testsuite/ChangeLog:
> 
> 	PR c++/99374
> 	* g++.dg/cpp1z/noexcept-type23.C: New test.
> ---
>   gcc/cp/call.c                                |  4 +++-
>   gcc/testsuite/g++.dg/cpp1z/noexcept-type23.C | 14 ++++++++++++++
>   2 files changed, 17 insertions(+), 1 deletion(-)
>   create mode 100644 gcc/testsuite/g++.dg/cpp1z/noexcept-type23.C
> 
> diff --git a/gcc/cp/call.c b/gcc/cp/call.c
> index 123f06b1f2b..605cbe15f4e 100644
> --- a/gcc/cp/call.c
> +++ b/gcc/cp/call.c
> @@ -1449,7 +1449,9 @@ standard_conversion (tree to, tree from, tree expr, bool c_cast_p,
>         tree fbase = class_of_this_parm (fromfn);
>         tree tbase = class_of_this_parm (tofn);
>   
> -      if (!DERIVED_FROM_P (fbase, tbase))
> +      /* If FBASE and TBASE are equivalent but incomplete, DERIVED_FROM_P
> +	 yields false.  But a pointer to member of incomplete class is OK.  */
> +      if (!DERIVED_FROM_P (fbase, tbase) && !same_type_p (fbase, tbase))

Let's check same_type_p first.  OK with that change.

>   	return NULL;
>   
>         tree fstat = static_fn_type (fromfn);
> diff --git a/gcc/testsuite/g++.dg/cpp1z/noexcept-type23.C b/gcc/testsuite/g++.dg/cpp1z/noexcept-type23.C
> new file mode 100644
> index 00000000000..612dd6ceb5e
> --- /dev/null
> +++ b/gcc/testsuite/g++.dg/cpp1z/noexcept-type23.C
> @@ -0,0 +1,14 @@
> +// PR c++/99374
> +// { dg-do compile { target c++17 } }
> +
> +struct S;
> +struct R;
> +using F1 = int (S::*)();
> +using F2 = int (S::*)() noexcept;
> +using F3 = int (R::*)() noexcept;
> +using T = S;
> +using F4 = int (T::*)() noexcept;
> +F1 f21 = F2();
> +F1 f41 = F4();
> +F2 f12 = F1(); // { dg-error "cannot convert" }
> +F1 f31 = F3(); // { dg-error "cannot convert" }
> 
> base-commit: 0d737ed2171165ba39ab5647f8a94c588fc9a898
>