NVIDIA · wmaxey · Jun 28, 2023 · Jun 23, 2023 · Jun 23, 2023 · Jun 23, 2023
diff --git a/ci/build_libcudacxx.sh b/ci/build_libcudacxx.sh
@@ -17,5 +17,5 @@ cmake -S .. -B ../build \
     -DTHRUST_ENABLE_INSTALL_RULES=ON \
     -G Ninja
 
-LIBCUDACXX_SITE_CONFIG="../build/libcudacxx/test/lit.site.cfg" lit -v --no-progress-bar --no-execute -Dcompute_archs=${GPU_ARCHS} -Dstd="c++${CXX_STANDARD}" ../libcudacxx/.upstream-tests/test
+LIBCUDACXX_SITE_CONFIG="../build/libcudacxx/test/lit.site.cfg" lit -v --no-progress-bar -Dexecutor="NoopExecutor()" -Dcompute_archs=${GPU_ARCHS} -Dstd="c++${CXX_STANDARD}" ../libcudacxx/.upstream-tests/test
 
diff --git a/libcudacxx/.upstream-tests/test/cuda/barrier_native_handle.pass.cpp b/libcudacxx/.upstream-tests/test/cuda/barrier_native_handle.pass.cpp
@@ -10,12 +10,13 @@
 
 // UNSUPPORTED: pre-sm-80
 
-#pragma nv_diag_suppress static_var_with_dynamic_init
-#pragma nv_diag_suppress set_but_not_used
-
 #include <cuda/barrier>
 
 #include "cuda_space_selector.h"
+#include "test_macros.h"
+
+TEST_NV_DIAG_SUPPRESS(static_var_with_dynamic_init)
+TEST_NV_DIAG_SUPPRESS(set_but_not_used)
 
 __device__
 void test()

diff --git a/libcudacxx/.upstream-tests/test/cuda/pipeline_arrive_on.pass.cpp b/libcudacxx/.upstream-tests/test/cuda/pipeline_arrive_on.pass.cpp
@@ -12,16 +12,18 @@
 
 // Remove after bump to version 4
 #define _LIBCUDACXX_CUDA_ABI_VERSION 3
-// TODO: Remove pointless comparison suppression when compiler fixes short-circuiting
-#pragma nv_diag_suppress 186
-
-#pragma nv_diag_suppress static_var_with_dynamic_init
-#pragma nv_diag_suppress declared_but_not_referenced
 
 #include <cuda_pipeline.h>
 #include <cuda/barrier>
 
 #include "cuda_space_selector.h"
+#include "test_macros.h"
+
+// TODO: Remove pointless comparison suppression when compiler fixes short-circuiting
+TEST_NV_DIAG_SUPPRESS(186) // pointless comparison of unsigned integer with zero
+
+TEST_NV_DIAG_SUPPRESS(static_var_with_dynamic_init)
+TEST_NV_DIAG_SUPPRESS(declared_but_not_referenced)
 
 using nvcuda::experimental::pipeline;
 

diff --git a/libcudacxx/.upstream-tests/test/cuda/pipeline_arrive_on_abi_v2.pass.cpp b/libcudacxx/.upstream-tests/test/cuda/pipeline_arrive_on_abi_v2.pass.cpp
@@ -12,16 +12,17 @@
 
 #define _LIBCUDACXX_CUDA_ABI_VERSION 2
 
-// TODO: Remove pointless comparison suppression when compiler fixes short-circuiting
-#pragma nv_diag_suppress 186
-
-#pragma nv_diag_suppress static_var_with_dynamic_init
-#pragma nv_diag_suppress declared_but_not_referenced
-
 #include <cuda_pipeline.h>
 #include <cuda/barrier>
 
 #include "cuda_space_selector.h"
+#include "test_macros.h"
+
+// TODO: Remove pointless comparison suppression when compiler fixes short-circuiting
+TEST_NV_DIAG_SUPPRESS(186) // pointless comparison of unsigned integer with zero
+
+TEST_NV_DIAG_SUPPRESS(static_var_with_dynamic_init)
+TEST_NV_DIAG_SUPPRESS(declared_but_not_referenced)
 
 using nvcuda::experimental::pipeline;
 

diff --git a/libcudacxx/.upstream-tests/test/cuda/pipeline_divergent_threads.pass.cpp b/libcudacxx/.upstream-tests/test/cuda/pipeline_divergent_threads.pass.cpp
@@ -11,13 +11,19 @@
 // UNSUPPORTED: pre-sm-70
 // UNSUPPORTED: nvrtc
 
+#include <cuda/std/type_traits>
 #include <cooperative_groups/memcpy_async.h>
 #include <cuda/pipeline>
 #include <cuda_pipeline.h>
 #include <math.h>
 #include <stdio.h>
 #include <stdlib.h>
 
+#include "test_macros.h"
+
+TEST_NV_DIAG_SUPPRESS(static_var_with_dynamic_init)
+TEST_NV_DIAG_SUPPRESS(186) // pointless comparison of unsigned integer with zero
+
 constexpr int nthreads = 256;
 constexpr size_t stages_count = 2; // Pipeline with two stages
 
@@ -40,7 +46,6 @@ __global__ void with_staging(int* global_out, int const* global_in, size_t size,
     size_t shared_offset[stages_count] = { 0, block.size() }; // Offsets to each batch
 
     // Allocate shared storage for a two-stage cuda::pipeline:
-    #pragma nv_diag_suppress static_var_with_dynamic_init
     __shared__ cuda::pipeline_shared_state<cuda::thread_scope::thread_scope_block, stages_count> shared_state;
     auto pipeline = cuda::make_pipeline(block, &shared_state);
 

diff --git a/libcudacxx/.upstream-tests/test/cuda/pipeline_group_concept.h b/libcudacxx/.upstream-tests/test/cuda/pipeline_group_concept.h
@@ -11,9 +11,11 @@
 // UNSUPPORTED: pre-sm-70
 
 // TODO: Remove pointless comparison suppression when compiler fixes short-circuiting
-#pragma nv_diag_suppress 186
 
 #include <cuda/pipeline>
+#include "test_macros.h"
+
+TEST_NV_DIAG_SUPPRESS(186) // pointless comparison of unsigned integer with zero
 
 template <typename T_size, typename T_thread_rank>
 struct single_thread_test_group {

diff --git a/libcudacxx/.upstream-tests/test/libcxx/libcpp_alignof.pass.cpp b/libcudacxx/.upstream-tests/test/libcxx/libcpp_alignof.pass.cpp
@@ -16,7 +16,7 @@
 #include <cuda/std/type_traits>
 #include "test_macros.h"
 
-#pragma nv_diag_suppress cuda_demote_unsupported_floating_point
+TEST_NV_DIAG_SUPPRESS(cuda_demote_unsupported_floating_point)
 
 template <class T>
 __host__ __device__

diff --git a/libcudacxx/.upstream-tests/test/std/atomics/atomics.lockfree/isalwayslockfree.pass.cpp b/libcudacxx/.upstream-tests/test/std/atomics/atomics.lockfree/isalwayslockfree.pass.cpp
@@ -13,11 +13,13 @@
 
 // static constexpr bool is_always_lock_free;
 
-#include <cuda/std/atomic>
 #include <cuda/std/cassert>
+#include <cuda/std/atomic>
 
 #include "test_macros.h"
 
+TEST_NV_DIAG_SUPPRESS(cuda_demote_unsupported_floating_point)
+
 // NVRTC doesn't include host atomic making this feature test invalid
 // TODO: Should we define __cpp_lib_atomic_is_always_lock_free for NVRTC?
 #if !defined(__CUDACC_RTC__)
@@ -97,7 +99,7 @@ void run()
     CHECK_ALWAYS_LOCK_FREE(void*);
     CHECK_ALWAYS_LOCK_FREE(float);
     CHECK_ALWAYS_LOCK_FREE(double);
-    CHECK_ALWAYS_LOCK_FREE(long double);
+    //CHECK_ALWAYS_LOCK_FREE(long double); // long double is not supported
 #if __has_attribute(vector_size) && defined(_LIBCUDACXX_VERSION) && !defined(__CUDACC__)
     // NOTE: NVCC doesn't support the vector_size attribute in device code.
     CHECK_ALWAYS_LOCK_FREE(int __attribute__((vector_size(1 * sizeof(int)))));

diff --git a/...pstream-tests/test/std/concepts/concepts.lang/concept.convertible/convertible_to.pass.cpp b/...pstream-tests/test/std/concepts/concepts.lang/concept.convertible/convertible_to.pass.cpp
@@ -11,7 +11,6 @@
 
 // template<class From, class To>
 // concept convertible_to;
-#pragma nv_diag_suppress 3013 // a volatile function parameter is deprecated
 
 #if defined(__clang__)
 #pragma clang diagnostic ignored "-Wc++17-extensions"

diff --git a/.../std/concepts/concepts.lang/concept.copyconstructible/copy_constructible.compile.pass.cpp b/.../std/concepts/concepts.lang/concept.copyconstructible/copy_constructible.compile.pass.cpp
@@ -11,7 +11,6 @@
 
 // template<class T>
 // concept copy_constructible;
-#pragma nv_diag_suppress 3013 // a volatile function parameter is deprecated
 
 #if defined(__clang__)
 #pragma clang diagnostic ignored "-Wc++17-extensions"

diff --git a/.../std/concepts/concepts.lang/concept.moveconstructible/move_constructible.compile.pass.cpp b/.../std/concepts/concepts.lang/concept.moveconstructible/move_constructible.compile.pass.cpp
@@ -11,7 +11,6 @@
 
 // template<class T>
 // concept move_constructible;
-#pragma nv_diag_suppress 3013 // a volatile function parameter is deprecated
 
 #if defined(__clang__)
 #pragma clang diagnostic ignored "-Wc++17-extensions"
@@ -21,6 +20,7 @@
 #include <cuda/std/type_traits>
 
 #include "type_classification/moveconstructible.h"
+#include "test_macros.h"
 
 using cuda::std::move_constructible;
 

diff --git a/...acxx/.upstream-tests/test/std/concepts/concepts.lang/concept.swappable/swappable.pass.cpp b/...acxx/.upstream-tests/test/std/concepts/concepts.lang/concept.swappable/swappable.pass.cpp
@@ -11,7 +11,6 @@
 
 // template<class T>
 // concept swappable = // see below
-#pragma nv_diag_suppress 3013 // a volatile function parameter is deprecated
 
 #if defined(__clang__)
 #pragma clang diagnostic ignored "-Wc++17-extensions"

diff --git a/...m-tests/test/std/concepts/concepts.lang/concept.swappable/swappable_with.compile.pass.cpp b/...m-tests/test/std/concepts/concepts.lang/concept.swappable/swappable_with.compile.pass.cpp
@@ -11,7 +11,6 @@
 
 // template<class T, class U>
 // concept swappable_with = // see below
-#pragma nv_diag_suppress 3013 // a volatile function parameter is deprecated
 
 #if defined(__clang__)
 #pragma clang diagnostic ignored "-Wc++17-extensions"

diff --git a/libcudacxx/.upstream-tests/test/std/concepts/concepts.object/copyable.compile.pass.cpp b/libcudacxx/.upstream-tests/test/std/concepts/concepts.object/copyable.compile.pass.cpp
@@ -11,7 +11,6 @@
 
 // template<class T>
 // concept copyable = see below;
-#pragma nv_diag_suppress 3013 // a volatile function parameter is deprecated
 
 #if defined(__clang__)
 #pragma clang diagnostic ignored "-Wc++17-extensions"
@@ -21,6 +20,7 @@
 #include <cuda/std/type_traits>
 
 #include "type_classification/copyable.h"
+#include "test_macros.h"
 
 using cuda::std::copyable;
 

diff --git a/libcudacxx/.upstream-tests/test/std/concepts/concepts.object/movable.compile.pass.cpp b/libcudacxx/.upstream-tests/test/std/concepts/concepts.object/movable.compile.pass.cpp
@@ -11,7 +11,6 @@
 
 // template<class T>
 // concept movable = see below;
-#pragma nv_diag_suppress 3013 // a volatile function parameter is deprecated
 
 #if defined(__clang__)
 #pragma clang diagnostic ignored "-Wc++17-extensions"

diff --git a/libcudacxx/.upstream-tests/test/std/concepts/concepts.object/regular.compile.pass.cpp b/libcudacxx/.upstream-tests/test/std/concepts/concepts.object/regular.compile.pass.cpp
@@ -11,7 +11,6 @@
 
 // template<class T>
 // concept regular = see below;
-#pragma nv_diag_suppress 3013 // a volatile function parameter is deprecated
 
 #if defined(__clang__)
 #pragma clang diagnostic ignored "-Wc++17-extensions"
@@ -22,6 +21,7 @@
 
 #include "type_classification/moveconstructible.h"
 #include "type_classification/semiregular.h"
+#include "test_macros.h"
 
 using cuda::std::regular;
 

diff --git a/libcudacxx/.upstream-tests/test/std/concepts/concepts.object/semiregular.compile.pass.cpp b/libcudacxx/.upstream-tests/test/std/concepts/concepts.object/semiregular.compile.pass.cpp
@@ -11,7 +11,6 @@
 
 // template<class T>
 // concept semiregular = see below;
-#pragma nv_diag_suppress 3013 // a volatile function parameter is deprecated
 
 #if defined(__clang__)
 #pragma clang diagnostic ignored "-Wc++17-extensions"
@@ -21,6 +20,7 @@
 #include <cuda/std/type_traits>
 
 #include "type_classification/semiregular.h"
+#include "test_macros.h"
 
 using cuda::std::semiregular;
 

diff --git a/libcudacxx/.upstream-tests/test/std/containers/sequences/array/array.cons/deduct.pass.cpp b/libcudacxx/.upstream-tests/test/std/containers/sequences/array/array.cons/deduct.pass.cpp
@@ -12,6 +12,7 @@
 // UNSUPPORTED: libcpp-no-deduction-guides
 // Clang 5 will generate bad implicit deduction guides
 //	Specifically, for the copy constructor.
+// UNSUPPORTED: clang-9 && nvcc-11.1
 
 
 // template <class T, class... U>

diff --git a/...dacxx/.upstream-tests/test/std/containers/views/mdspan/mdspan.extents.cons/array.pass.cpp b/...dacxx/.upstream-tests/test/std/containers/views/mdspan/mdspan.extents.cons/array.pass.cpp
@@ -55,6 +55,7 @@ int main(int, char**)
 
     static_assert( is_array_cons_avail_v< cuda::std::dextents<   int,2>, my_int, 2 > == true , "" );
 
+#if !defined(TEST_COMPILER_NVCC_BELOW_11_3)
     // Constraint: rank consistency
     static_assert( is_array_cons_avail_v< cuda::std::dextents<   int,1>, int   , 2 > == false, "" );
 
@@ -65,6 +66,7 @@ int main(int, char**)
 #ifndef TEST_COMPILER_NVHPC
     static_assert( is_array_cons_avail_v< cuda::std::dextents<   int,1>, my_int_non_nothrow_constructible, 1 > == false, "" );
 #endif // TEST_COMPILER_NVHPC
+#endif // !defined(TEST_COMPILER_NVCC_BELOW_11_3)
 
     return 0;
 }
diff --git a/...udacxx/.upstream-tests/test/std/containers/views/mdspan/mdspan.extents.cons/span.pass.cpp b/...udacxx/.upstream-tests/test/std/containers/views/mdspan/mdspan.extents.cons/span.pass.cpp
@@ -56,6 +56,7 @@ int main(int, char**)
 
     static_assert( is_span_cons_avail_v< cuda::std::dextents<int,2>, my_int, 2 > == true , "" );
 
+#if !defined(TEST_COMPILER_NVCC_BELOW_11_3)
     // Constraint: rank consistency
     static_assert( is_span_cons_avail_v< cuda::std::dextents<int,1>, int   , 2 > == false, "" );
 
@@ -66,6 +67,7 @@ int main(int, char**)
 #ifndef TEST_COMPILER_NVHPC
     static_assert( is_span_cons_avail_v< cuda::std::dextents<int,1>, my_int_non_nothrow_constructible, 1 > == false, "" );
 #endif // TEST_COMPILER_NVHPC
+#endif // !defined(TEST_COMPILER_NVCC_BELOW_11_3)
 
     return 0;
 }
diff --git a/libcudacxx/.upstream-tests/test/std/iterators/iterator.container/data.pass.cpp b/libcudacxx/.upstream-tests/test/std/iterators/iterator.container/data.pass.cpp
@@ -71,7 +71,7 @@ void test_const_array( const T (&array)[Sz] )
     assert ( cuda::std::data(array) == &array[0]);
 }
 
-STATIC_TEST_GLOBAL_VAR constexpr int arrA [] { 1, 2, 3 };
+STATIC_TEST_GLOBAL_VAR TEST_CONSTEXPR_GLOBAL int arrA [] { 1, 2, 3 };
 
 int main(int, char**)
 {

diff --git a/libcudacxx/.upstream-tests/test/std/iterators/iterator.container/empty.pass.cpp b/libcudacxx/.upstream-tests/test/std/iterators/iterator.container/empty.pass.cpp
@@ -72,7 +72,7 @@ void test_const_array( const T (&array)[Sz] )
     assert (!cuda::std::empty(array));
 }
 
-STATIC_TEST_GLOBAL_VAR constexpr int arrA [] { 1, 2, 3 };
+STATIC_TEST_GLOBAL_VAR TEST_CONSTEXPR_GLOBAL int arrA [] { 1, 2, 3 };
 
 int main(int, char**)
 {

diff --git a/libcudacxx/.upstream-tests/test/std/iterators/iterator.container/size.pass.cpp b/libcudacxx/.upstream-tests/test/std/iterators/iterator.container/size.pass.cpp
@@ -76,7 +76,7 @@ void test_const_array( const T (&array)[Sz] )
     assert ( cuda::std::size(array) == Sz );
 }
 
-STATIC_TEST_GLOBAL_VAR constexpr int arrA [] { 1, 2, 3 };
+STATIC_TEST_GLOBAL_VAR TEST_CONSTEXPR_GLOBAL int arrA [] { 1, 2, 3 };
 
 int main(int, char**)
 {

diff --git a/libcudacxx/.upstream-tests/test/std/iterators/iterator.container/ssize.pass.cpp b/libcudacxx/.upstream-tests/test/std/iterators/iterator.container/ssize.pass.cpp
@@ -85,7 +85,7 @@ void test_const_array(const T (&array)[Sz])
     assert ( cuda::std::ssize(array) == Sz );
 }
 
-STATIC_TEST_GLOBAL_VAR constexpr int arrA [] { 1, 2, 3 };
+STATIC_TEST_GLOBAL_VAR TEST_CONSTEXPR_GLOBAL int arrA [] { 1, 2, 3 };
 
 int main(int, char**)
 {

diff --git a/libcudacxx/.upstream-tests/test/std/iterators/iterator.range/begin-end.pass.cpp b/libcudacxx/.upstream-tests/test/std/iterators/iterator.range/begin-end.pass.cpp
@@ -145,10 +145,12 @@ void test_const_array( const T (&array)[Sz] ) {
 #endif
     }
 
-STATIC_TEST_GLOBAL_VAR constexpr int global_array [] { 1, 2, 3 };
+STATIC_TEST_GLOBAL_VAR TEST_CONSTEXPR_GLOBAL int global_array [] { 1, 2, 3 };
 #if TEST_STD_VER > 14
-STATIC_TEST_GLOBAL_VAR constexpr const int global_const_array[] = {0,1,2,3,4};
-#endif
+#if !defined(TEST_COMPILER_NVCC_BELOW_11_3)
+STATIC_TEST_GLOBAL_VAR TEST_CONSTEXPR_GLOBAL int global_const_array[] = {0,1,2,3,4};
+#endif // nvcc > 11.2
+#endif // TEST_STD_VER > 14
 
 int main(int, char**) {
 #if defined(_LIBCUDACXX_HAS_VECTOR)
@@ -214,14 +216,16 @@ int main(int, char**) {
         static_assert ( *cuda::std::crbegin(local_const_array) == 4, "" );
     }
 
+#if !defined(TEST_COMPILER_NVCC_BELOW_11_3)
     {
 
         static_assert ( *cuda::std::begin(global_const_array)   == 0, "" );
         static_assert ( *cuda::std::cbegin(global_const_array)  == 0, "" );
         static_assert ( *cuda::std::rbegin(global_const_array)  == 4, "" );
         static_assert ( *cuda::std::crbegin(global_const_array) == 4, "" );
     }
-#endif
+#endif // nvcc > 11.2
+#endif // TEST_STD_VER > 14
 
   return 0;
 }
diff --git a/libcudacxx/.upstream-tests/test/std/language.support/support.types/max_align_t.pass.cpp b/libcudacxx/.upstream-tests/test/std/language.support/support.types/max_align_t.pass.cpp
@@ -11,13 +11,13 @@
 
 // max_align_t is a trivial standard-layout type whose alignment requirement
 //   is at least as great as that of every scalar type
+#include "test_macros.h"
 
-#ifndef __CUDACC_RTC__
+#ifndef TEST_COMPILER_NVRTC
 #include <stdio.h>
-#endif // __CUDACC_RTC__
-#include "test_macros.h"
+#endif // TEST_COMPILER_NVRTC
 
-#pragma nv_diag_suppress cuda_demote_unsupported_floating_point
+TEST_NV_DIAG_SUPPRESS(cuda_demote_unsupported_floating_point)
 
 int main(int, char**)
 {

diff --git a/...-tests/test/std/utilities/expected/expected.expected/ctor/ctor.inplace_init_list.pass.cpp b/...-tests/test/std/utilities/expected/expected.expected/ctor/ctor.inplace_init_list.pass.cpp
@@ -146,7 +146,9 @@ __host__ __device__ void testException() {
 int main(int, char**) {
   test();
 #if defined(_LIBCUDACXX_ADDRESSOF)
+#if !(defined(TEST_COMPILER_NVCC_BELOW_11_3) && defined(TEST_COMPILER_CLANG))
   static_assert(test(), "");
+#endif // !(defined(TEST_COMPILER_NVCC_BELOW_11_3) && defined(TEST_COMPILER_CLANG))
 #endif // defined(_LIBCUDACXX_ADDRESSOF)
   testException();
   return 0;

diff --git a/...tests/test/std/utilities/expected/expected.expected/ctor/ctor.unexpect_init_list.pass.cpp b/...tests/test/std/utilities/expected/expected.expected/ctor/ctor.unexpect_init_list.pass.cpp
@@ -147,7 +147,9 @@ __host__ __device__ void testException() {
 int main(int, char**) {
   test();
 #if defined(_LIBCUDACXX_ADDRESSOF)
+#if !(defined(TEST_COMPILER_NVCC_BELOW_11_3) && defined(TEST_COMPILER_CLANG))
   static_assert(test(), "");
+#endif // !(defined(TEST_COMPILER_NVCC_BELOW_11_3) && defined(TEST_COMPILER_CLANG))
 #endif // defined(_LIBCUDACXX_ADDRESSOF)
   testException();
   return 0;