* [committed] libstdc++: Fix common_reference for non-reference results [PR100894]
@ 2021-06-14 20:18 Jonathan Wakely
0 siblings, 0 replies; only message in thread
From: Jonathan Wakely @ 2021-06-14 20:18 UTC (permalink / raw)
To: libstdc++, gcc-patches
[-- Attachment #1: Type: text/plain, Size: 787 bytes --]
The result of COMMON-REF(A&, B&&) where they have no common reference
type should not be a reference. The implementation of COMMON-REF fails
to check that the result is a reference, so is well-formed when it
shouldn't be. This means that common_reference uses that result when it
shouldn't.
The fix is to reject the result of COMMON-REF(A, B) if it's not a
reference, so that common_reference falls through to the next case,
which uses COND-RES, which yields a non-reference result.
Signed-off-by: Jonathan Wakely <jwakely@redhat.com>
libstdc++-v3/ChangeLog:
PR libstdc++/100894
* include/std/type_traits (__common_ref_impl<X&, Y&>): Only
use the type if it's a reference.
* testsuite/20_util/common_reference/100894.cc: New test.
Tested powerpc64le-linux. Committed to trunk.
[-- Attachment #2: patch.txt --]
[-- Type: text/plain, Size: 2740 bytes --]
commit c37b5ddcc88e0cc0f6a4ad609eda51021df0f6bb
Author: Jonathan Wakely <jwakely@redhat.com>
Date: Mon Jun 14 20:31:00 2021
libstdc++: Fix common_reference for non-reference results [PR100894]
The result of COMMON-REF(A&, B&&) where they have no common reference
type should not be a reference. The implementation of COMMON-REF fails
to check that the result is a reference, so is well-formed when it
shouldn't be. This means that common_reference uses that result when it
shouldn't.
The fix is to reject the result of COMMON-REF(A, B) if it's not a
reference, so that common_reference falls through to the next case,
which uses COND-RES, which yields a non-reference result.
Signed-off-by: Jonathan Wakely <jwakely@redhat.com>
libstdc++-v3/ChangeLog:
PR libstdc++/100894
* include/std/type_traits (__common_ref_impl<X&, Y&>): Only
use the type if it's a reference.
* testsuite/20_util/common_reference/100894.cc: New test.
diff --git a/libstdc++-v3/include/std/type_traits b/libstdc++-v3/include/std/type_traits
index eaf06fcb036..d9068a06f08 100644
--- a/libstdc++-v3/include/std/type_traits
+++ b/libstdc++-v3/include/std/type_traits
@@ -3340,11 +3340,17 @@ template <typename _From, typename _To>
template<typename _Ap, typename _Bp>
using __common_ref = typename __common_ref_impl<_Ap, _Bp>::type;
+ // COND-RES(COPYCV(X, Y) &, COPYCV(Y, X) &)
+ template<typename _Xp, typename _Yp>
+ using __condres_cvref
+ = __cond_res<__copy_cv<_Xp, _Yp>&, __copy_cv<_Yp, _Xp>&>;
+
// If A and B are both lvalue reference types, ...
template<typename _Xp, typename _Yp>
- struct __common_ref_impl<_Xp&, _Yp&,
- __void_t<__cond_res<__copy_cv<_Xp, _Yp>&, __copy_cv<_Yp, _Xp>&>>>
- { using type = __cond_res<__copy_cv<_Xp, _Yp>&, __copy_cv<_Yp, _Xp>&>; };
+ struct __common_ref_impl<_Xp&, _Yp&, __void_t<__condres_cvref<_Xp, _Yp>>>
+ : enable_if<is_reference_v<__condres_cvref<_Xp, _Yp>>,
+ __condres_cvref<_Xp, _Yp>>
+ { };
// let C be remove_reference_t<COMMON-REF(X&, Y&)>&&
template<typename _Xp, typename _Yp>
diff --git a/libstdc++-v3/testsuite/20_util/common_reference/100894.cc b/libstdc++-v3/testsuite/20_util/common_reference/100894.cc
new file mode 100644
index 00000000000..5e144768002
--- /dev/null
+++ b/libstdc++-v3/testsuite/20_util/common_reference/100894.cc
@@ -0,0 +1,9 @@
+// { dg-options "-std=gnu++20" }
+// { dg-do compile { target c++20 } }
+// PR libstdc++/100894 - common_reference implementation seems to be wrong
+
+#include <type_traits>
+
+struct A {};
+struct B { B(A); };
+static_assert( std::is_same_v<std::common_reference_t<A&, B&&>, B> );
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2021-06-14 20:18 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-06-14 20:18 [committed] libstdc++: Fix common_reference for non-reference results [PR100894] Jonathan Wakely
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).