[libc++] Update the CI to Clang-20 and drop Clang-17 support

Author: philnik777

.github/workflows/libcxx-build-and-test.yaml

@@ -48,8 +48,8 @@ jobs:
           'generic-cxx26',
           'generic-modules'
         ]
-        cc: [  'clang-19' ]
-        cxx: [ 'clang++-19' ]
+        cc: [  'clang-20' ]
+        cxx: [ 'clang++-20' ]
         include:
           - config: 'generic-gcc'
             cc: 'gcc-14'
@@ -88,18 +88,18 @@ jobs:
           'generic-cxx20',
           'generic-cxx23'
         ]
-        cc: [ 'clang-19' ]
-        cxx: [ 'clang++-19' ]
+        cc: [ 'clang-20' ]
+        cxx: [ 'clang++-20' ]
         include:
           - config: 'generic-gcc-cxx11'
             cc: 'gcc-14'
             cxx: 'g++-14'
           - config: 'generic-cxx23'
-            cc: 'clang-17'
-            cxx: 'clang++-17'
-          - config: 'generic-cxx26'
             cc: 'clang-18'
             cxx: 'clang++-18'
+          - config: 'generic-cxx26'
+            cc: 'clang-19'
+            cxx: 'clang++-19'
     steps:
       - uses: actions/checkout@v4
       - name: ${{ matrix.config }}
@@ -169,8 +169,8 @@ jobs:
       - name: ${{ matrix.config }}
         run: libcxx/utils/ci/run-buildbot ${{ matrix.config }}
         env:
-          CC: clang-19
-          CXX: clang++-19
+          CC: clang-20
+          CXX: clang++-20
       - uses: actions/upload-artifact@26f96dfa697d77e81fd5907df203aa23a56210a8 # v4.3.0
         if: always()
         with:

libcxx/include/__configuration/compiler.h

@@ -33,8 +33,8 @@
 // Warn if a compiler version is used that is not supported anymore
 // LLVM RELEASE Update the minimum compiler versions
 #  if defined(_LIBCPP_CLANG_VER)
-#    if _LIBCPP_CLANG_VER < 1700
-#      warning "Libc++ only supports Clang 17 and later"
+#    if _LIBCPP_CLANG_VER < 1800
+#      warning "Libc++ only supports Clang 18 and later"
 #    endif
 #  elif defined(_LIBCPP_APPLE_CLANG_VER)
 #    if _LIBCPP_APPLE_CLANG_VER < 1500

libcxx/include/__cxx03/__memory/uninitialized_algorithms.h

@@ -642,7 +642,8 @@ __uninitialized_allocator_relocate(_Alloc& __alloc, _Tp* __first, _Tp* __last, _
     __guard.__complete();
     std::__allocator_destroy(__alloc, __first, __last);
   } else {
-    __builtin_memcpy(const_cast<__remove_const_t<_Tp>*>(__result), __first, sizeof(_Tp) * (__last - __first));
+    // Casting to void* to suppress clang complaining that this is technically UB.
+    __builtin_memcpy(static_cast<void*>(__result), __first, sizeof(_Tp) * (__last - __first));
   }
 }
 

libcxx/include/__type_traits/promote.h

@@ -13,20 +13,12 @@
 #include <__type_traits/integral_constant.h>
 #include <__type_traits/is_arithmetic.h>
 
-#if defined(_LIBCPP_CLANG_VER) && _LIBCPP_CLANG_VER == 1700
-#  include <__type_traits/is_same.h>
-#  include <__utility/declval.h>
-#endif
-
 #if !defined(_LIBCPP_HAS_NO_PRAGMA_SYSTEM_HEADER)
 #  pragma GCC system_header
 #endif
 
 _LIBCPP_BEGIN_NAMESPACE_STD
 
-// TODO(LLVM-20): Remove this workaround
-#if !defined(_LIBCPP_CLANG_VER) || _LIBCPP_CLANG_VER != 1700
-
 template <class... _Args>
 class __promote {
   static_assert((is_arithmetic<_Args>::value && ...));
@@ -39,90 +31,17 @@ class __promote {
   static double __test(unsigned long);
   static double __test(long long);
   static double __test(unsigned long long);
-#  if _LIBCPP_HAS_INT128
+#if _LIBCPP_HAS_INT128
   static double __test(__int128_t);
   static double __test(__uint128_t);
-#  endif
+#endif
   static double __test(double);
   static long double __test(long double);
 
 public:
   using type = decltype((__test(_Args()) + ...));
 };
 
-#else
-
-template <class _Tp>
-struct __numeric_type {
-  static void __test(...);
-  static float __test(float);
-  static double __test(char);
-  static double __test(int);
-  static double __test(unsigned);
-  static double __test(long);
-  static double __test(unsigned long);
-  static double __test(long long);
-  static double __test(unsigned long long);
-#  if _LIBCPP_HAS_INT128
-  static double __test(__int128_t);
-  static double __test(__uint128_t);
-#  endif
-  static double __test(double);
-  static long double __test(long double);
-
-  typedef decltype(__test(std::declval<_Tp>())) type;
-  static const bool value = _IsNotSame<type, void>::value;
-};
-
-template <>
-struct __numeric_type<void> {
-  static const bool value = true;
-};
-
-template <class _A1,
-          class _A2 = void,
-          class _A3 = void,
-          bool      = __numeric_type<_A1>::value && __numeric_type<_A2>::value && __numeric_type<_A3>::value>
-class __promote_imp {
-public:
-  static const bool value = false;
-};
-
-template <class _A1, class _A2, class _A3>
-class __promote_imp<_A1, _A2, _A3, true> {
-private:
-  typedef typename __promote_imp<_A1>::type __type1;
-  typedef typename __promote_imp<_A2>::type __type2;
-  typedef typename __promote_imp<_A3>::type __type3;
-
-public:
-  typedef decltype(__type1() + __type2() + __type3()) type;
-  static const bool value = true;
-};
-
-template <class _A1, class _A2>
-class __promote_imp<_A1, _A2, void, true> {
-private:
-  typedef typename __promote_imp<_A1>::type __type1;
-  typedef typename __promote_imp<_A2>::type __type2;
-
-public:
-  typedef decltype(__type1() + __type2()) type;
-  static const bool value = true;
-};
-
-template <class _A1>
-class __promote_imp<_A1, void, void, true> {
-public:
-  typedef typename __numeric_type<_A1>::type type;
-  static const bool value = true;
-};
-
-template <class _A1, class _A2 = void, class _A3 = void>
-class __promote : public __promote_imp<_A1, _A2, _A3> {};
-
-#endif // !defined(_LIBCPP_CLANG_VER) || _LIBCPP_CLANG_VER >= 1700
-
 _LIBCPP_END_NAMESPACE_STD
 
 #endif // _LIBCPP___TYPE_TRAITS_PROMOTE_H

libcxx/src/experimental/time_zone.cpp

@@ -199,7 +199,7 @@ __format(const __tz::__continuation& __continuation, const string& __letters, se
                    // active at the end. This should be determined separately.
                    return chrono::seconds{0};
                  else
-                   static_assert(sizeof(_Tp) == 0); // TODO TZDB static_assert(false); after droping clang-16 support
+                   static_assert(false);
 
                  std::__libcpp_unreachable();
                },
@@ -225,7 +225,7 @@ __format(const __tz::__continuation& __continuation, const string& __letters, se
         else if constexpr (same_as<_Tp, __tz::__constrained_weekday>)
           return __value(__year, __month);
         else
-          static_assert(sizeof(_Tp) == 0); // TODO TZDB static_assert(false); after droping clang-16 support
+          static_assert(false);
 
         std::__libcpp_unreachable();
       },
@@ -688,7 +688,7 @@ __get_sys_info(sys_seconds __time,
         else if constexpr (same_as<_Tp, __tz::__save>)
           return chrono::__get_sys_info_basic(__time, __continuation_begin, __continuation, __value.__time);
         else
-          static_assert(sizeof(_Tp) == 0); // TODO TZDB static_assert(false); after droping clang-16 support
+          static_assert(false);
 
         std::__libcpp_unreachable();
       },

libcxx/test/libcxx/atomics/diagnose_invalid_memory_order.verify.cpp

@@ -6,11 +6,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-// This test fails with Clang <18 because diagnose_if doesn't emit all of the
-// diagnostics when -fdelayed-template-parsing is enabled, like it is in MSVC
-// mode.
-// XFAIL: msvc && clang-17
-
 // REQUIRES: diagnose-if-support
 
 // <atomic>

libcxx/test/libcxx/clang_tidy.gen.py

@@ -26,9 +26,6 @@
 // The GCC compiler flags are not always compatible with clang-tidy.
 // UNSUPPORTED: gcc
 
-// Clang 17 has false positives.
-// UNSUPPORTED: clang-17
-
 {lit_header_restrictions.get(header, '')}
 {lit_header_undeprecations.get(header, '')}
 

libcxx/test/libcxx/gdb/gdb_pretty_printer_test.sh.cpp

@@ -12,7 +12,7 @@
 // UNSUPPORTED: c++03
 
 // TODO: Investigate these failures which break the CI.
-// UNSUPPORTED: clang-17, clang-18, clang-19
+// UNSUPPORTED: clang-18, clang-19, clang-20
 
 // The Android libc++ tests are run on a non-Android host, connected to an
 // Android device over adb. gdb needs special support to make this work (e.g.

libcxx/test/libcxx/ranges/range.adaptors/range.lazy.split/no_unique_address.compile.pass.cpp

@@ -7,7 +7,6 @@
 //===----------------------------------------------------------------------===//
 
 // UNSUPPORTED: c++03, c++11, c++14, c++17
-// XFAIL: msvc && clang-17
 
 // class lazy_split_view {
 //   _LIBCPP_NO_UNIQUE_ADDRESS _View __base_ = _View();

libcxx/test/libcxx/ranges/range.adaptors/range.split/no_unique_address.compile.pass.cpp

@@ -7,7 +7,6 @@
 //===----------------------------------------------------------------------===//
 
 // UNSUPPORTED: c++03, c++11, c++14, c++17
-// XFAIL: msvc && clang-17
 
 // class split_view {
 //   _LIBCPP_NO_UNIQUE_ADDRESS _View __base_ = _View();

libcxx/test/libcxx/ranges/range.factories/range.istream.view/no_unique_address.compile.pass.cpp

@@ -8,7 +8,6 @@
 
 // UNSUPPORTED: no-localization
 // UNSUPPORTED: c++03, c++11, c++14, c++17
-// XFAIL: msvc && clang-17
 
 // Test the libc++ extension that the value stored in `std::ranges::istream_view` has been marked
 // as _LIBCPP_NO_UNIQUE_ADDRESS
@@ -21,4 +20,3 @@ struct Empty {
 };
 
 static_assert(sizeof(std::ranges::istream_view<Empty>) == sizeof(void*));
-

libcxx/test/std/atomics/atomics.types.generic/atomics.types.float/fetch_add.pass.cpp

@@ -10,7 +10,7 @@
 
 // Older versions of clang have a bug with atomic builtins affecting double and long double.
 // Fixed by 5fdd0948.
-// XFAIL: target=powerpc-ibm-{{.*}} && (clang-17 || clang-18)
+// XFAIL: target=powerpc-ibm-{{.*}} && clang-18
 
 // https://github.com/llvm/llvm-project/issues/72893
 // XFAIL: target={{x86_64-.*}} && tsan

libcxx/test/std/atomics/atomics.types.generic/atomics.types.float/fetch_sub.pass.cpp

@@ -10,7 +10,7 @@
 
 // Older versions of clang have a bug with atomic builtins affecting double and long double.
 // Fixed by 5fdd0948.
-// XFAIL: target=powerpc-ibm-{{.*}} && (clang-17 || clang-18)
+// XFAIL: target=powerpc-ibm-{{.*}} && clang-18
 
 // https://github.com/llvm/llvm-project/issues/72893
 // XFAIL: target={{x86_64-.*}} && tsan

libcxx/test/std/atomics/atomics.types.generic/atomics.types.float/operator.minus_equals.pass.cpp

@@ -10,7 +10,7 @@
 
 // Older versions of clang have a bug with atomic builtins affecting double and long double.
 // Fixed by 5fdd0948.
-// XFAIL: target=powerpc-ibm-{{.*}} && (clang-17 || clang-18)
+// XFAIL: target=powerpc-ibm-{{.*}} && clang-18
 
 // floating-point-type operator-=(floating-point-type) volatile noexcept;
 // floating-point-type operator-=(floating-point-type) noexcept;

libcxx/test/std/atomics/atomics.types.generic/atomics.types.float/operator.plus_equals.pass.cpp

@@ -10,7 +10,7 @@
 
 // Older versions of clang have a bug with atomic builtins affecting double and long double.
 // Fixed by 5fdd0948.
-// XFAIL: target=powerpc-ibm-{{.*}} && (clang-17 || clang-18)
+// XFAIL: target=powerpc-ibm-{{.*}} && clang-18
 
 // floating-point-type operator+=(floating-point-type) volatile noexcept;
 // floating-point-type operator+=(floating-point-type) noexcept;

libcxx/test/std/containers/views/mdspan/mdspan/index_operator.pass.cpp

@@ -39,8 +39,8 @@
 #include "../ConvertibleToIntegral.h"
 #include "../CustomTestLayouts.h"
 
-// Clang 16 does not support argument packs as input to operator []
-#if defined(__clang_major__) && __clang_major__ < 17
+// Apple Clang does not support argument packs as input to operator []
+#ifdef TEST_COMPILER_APPLE_CLANG
 template <class MDS>
 constexpr auto& access(MDS mds) {
   return mds[];
@@ -84,7 +84,7 @@ template <class MDS, class... Args>
 constexpr void iterate(MDS mds, Args... args) {
   constexpr int r = static_cast<int>(MDS::extents_type::rank()) - 1 - static_cast<int>(sizeof...(Args));
   if constexpr (-1 == r) {
-#if defined(__clang_major__) && __clang_major__ < 17
+#ifdef TEST_COMPILER_APPLE_CLANG
     int* ptr1 = &access(mds, args...);
 #else
     int* ptr1 = &mds[args...];

libcxx/test/std/containers/views/views.span/span.cons/array.pass.cpp

@@ -93,14 +93,17 @@ constexpr bool testSpan()
     assert(s3.data() == val && s3.size() == 2);
     assert(s4.data() == val && s4.size() == 2);
 
-    std::span<const int> s5 = {{1,2}};
+    TEST_DIAGNOSTIC_PUSH
+    TEST_CLANG_DIAGNOSTIC_IGNORED("-Wdangling")
+    std::span<const int> s5 = {{1, 2}};
 #if TEST_STD_VER >= 26
     std::span<const int, 2> s6({1, 2});
 #else
     std::span<const int, 2> s6 = {{1,2}};
 #endif
     assert(s5.size() == 2);  // and it dangles
     assert(s6.size() == 2);  // and it dangles
+    TEST_DIAGNOSTIC_POP
 
     return true;
 }

libcxx/test/std/experimental/simd/simd.class/simd_copy.pass.cpp

@@ -10,7 +10,7 @@
 
 // Older versions of clang may encounter a backend error (see 0295c2ad):
 //   Pass-by-value arguments with alignment greater than register width are not supported.
-// XFAIL: target=powerpc{{.*}}-ibm-{{.*}} && (clang-17 || clang-18)
+// XFAIL: target=powerpc{{.*}}-ibm-{{.*}} && clang-18
 
 // <experimental/simd>
 //

libcxx/test/std/experimental/simd/simd.class/simd_unary.pass.cpp

@@ -10,7 +10,7 @@
 
 // Older versions of clang may encounter a backend error (see 0295c2ad):
 //   Pass-by-value arguments with alignment greater than register width are not supported.
-// XFAIL: target=powerpc{{.*}}-ibm-{{.*}} && (clang-17 || clang-18)
+// XFAIL: target=powerpc{{.*}}-ibm-{{.*}} && clang-18
 
 // This test crashes AppleClang 15 but not later versions.
 // UNSUPPORTED: apple-clang-15

libcxx/test/std/language.support/support.dynamic/new.delete/new.delete.array/sized_delete_array.pass.cpp

@@ -11,7 +11,6 @@
 // UNSUPPORTED: c++03, c++11
 
 // These compiler versions and platforms don't enable sized deallocation by default.
-// ADDITIONAL_COMPILE_FLAGS(clang-17): -fsized-deallocation
 // ADDITIONAL_COMPILE_FLAGS(clang-18): -fsized-deallocation
 // ADDITIONAL_COMPILE_FLAGS(apple-clang-15): -fsized-deallocation
 // ADDITIONAL_COMPILE_FLAGS(apple-clang-16): -fsized-deallocation

libcxx/test/std/language.support/support.dynamic/new.delete/new.delete.single/sized_delete.pass.cpp

@@ -11,7 +11,6 @@
 // UNSUPPORTED: c++03, c++11
 
 // These compiler versions and platforms don't enable sized deallocation by default.
-// ADDITIONAL_COMPILE_FLAGS(clang-17): -fsized-deallocation
 // ADDITIONAL_COMPILE_FLAGS(clang-18): -fsized-deallocation
 // ADDITIONAL_COMPILE_FLAGS(apple-clang-15): -fsized-deallocation
 // ADDITIONAL_COMPILE_FLAGS(apple-clang-16): -fsized-deallocation

libcxx/test/std/numerics/c.math/signbit.pass.cpp

@@ -12,7 +12,7 @@
 // UNSUPPORTED: windows
 
 // These compilers don't support constexpr `__builtin_signbit` yet.
-// UNSUPPORTED: clang-17, clang-18, clang-19, apple-clang-15, apple-clang-16
+// UNSUPPORTED: clang-18, clang-19, apple-clang-15, apple-clang-16
 
 // XFAIL: FROZEN-CXX03-HEADERS-FIXME
 

libcxx/test/std/utilities/expected/expected.expected/ctor/ctor.copy.pass.cpp

@@ -121,8 +121,8 @@ constexpr bool test() {
   }
 
   {
-    // TODO(LLVM 20): Remove once we drop support for Clang 17
-#if defined(TEST_CLANG_VER) && TEST_CLANG_VER >= 1800
+    // TODO: Drop this once AppleClang is upgraded
+#ifndef TEST_COMPILER_APPLE_CLANG
     // https://github.com/llvm/llvm-project/issues/92676
     std::expected<Any, int> e1;
     auto e2 = e1;