Mostly flatten the folder structure of libcu++ (#1630)

* Rename internal only header files to `__meow_` We want to name the folder cotnaining the subheaders as `__meow` so that we then can easily expose the internal header without any further code movement * Move all internal subfolders up to `cuda/std` We do not need the deep nested structure anymore, as a first step move all internal subfolders of `cuda/std/detail/libcxx/include` to `cuda/std` * Drop `__cccl_config` file in libcxx * Move `ptx` files from `cuda/std/__cuda` into `cuda/__ptx` * Adopt internal test files to new folder structure
NVIDIA · Apr 18, 2024 · 29a24f4 · 29a24f4
1 parent 05fadcf
commit 29a24f4
Show file tree

Hide file tree

Showing 681 changed files with 2,719 additions and 2,750 deletions.
diff --git a/ci/update_version.sh b/ci/update_version.sh
@@ -26,7 +26,7 @@ if [ -z "$major" ] || [ -z "$minor" ] || [ -z "$patch" ]; then
 fi
 
 # Version file paths
-CCCL_VERSION_FILE="libcudacxx/include/cuda/std/detail/libcxx/include/__cccl/version.h"
+CCCL_VERSION_FILE="libcudacxx/include/cuda/std/__cccl/version.h"
 THRUST_VERSION_FILE="thrust/thrust/version.h"
 CUB_VERSION_FILE="cub/cub/version.cuh"
 CCCL_CMAKE_VERSION_FILE="lib/cmake/cccl/cccl-config-version.cmake"

diff --git a/cub/test/catch2_test_helper.h b/cub/test/catch2_test_helper.h
@@ -33,7 +33,7 @@
 #include <tuple>
 #include <type_traits>
 
-#include "cuda/std/detail/libcxx/include/__cccl/diagnostic.h"
+#include "cuda/std/__cccl/diagnostic.h"
 #include "test_util_vec.h"
 #include <metal.hpp>
 

diff --git a/libcudacxx/include/cuda/__cccl_config b/libcudacxx/include/cuda/__cccl_config
@@ -4,13 +4,21 @@
 // under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-// SPDX-FileCopyrightText: Copyright (c) 2023 NVIDIA CORPORATION & AFFILIATES.
+// SPDX-FileCopyrightText: Copyright (c) 2023-24 NVIDIA CORPORATION & AFFILIATES.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef _CUDA__CCCL_CONFIG
 #define _CUDA__CCCL_CONFIG
 
-#include <cuda/std/detail/libcxx/include/__cccl_config>
+#include <cuda/std/__cccl/attributes.h>
+#include <cuda/std/__cccl/compiler.h>
+#include <cuda/std/__cccl/diagnostic.h>
+#include <cuda/std/__cccl/dialect.h>
+#include <cuda/std/__cccl/execution_space.h>
+#include <cuda/std/__cccl/ptx_isa.h>
+#include <cuda/std/__cccl/system_header.h>
+#include <cuda/std/__cccl/version.h>
+#include <cuda/std/__cccl/visibility.h>
 
 #endif // _CUDA__CCCL_CONFIG
diff --git a/libcudacxx/include/cuda/__memory_resource/cuda_managed_memory_resource.h b/libcudacxx/include/cuda/__memory_resource/cuda_managed_memory_resource.h
@@ -32,7 +32,7 @@
 #  include <cuda/__memory_resource/resource.h>
 #  include <cuda/__memory_resource/resource_ref.h>
 #  include <cuda/std/__cuda/api_wrapper.h>
-#  include <cuda/std/detail/libcxx/include/__new/bad_alloc.h>
+#  include <cuda/std/__new/bad_alloc.h>
 
 #  if _CCCL_STD_VER >= 2014
 

diff --git a/libcudacxx/include/cuda/__memory_resource/cuda_memory_resource.h b/libcudacxx/include/cuda/__memory_resource/cuda_memory_resource.h
@@ -32,7 +32,7 @@
 #  include <cuda/__memory_resource/resource.h>
 #  include <cuda/__memory_resource/resource_ref.h>
 #  include <cuda/std/__cuda/api_wrapper.h>
-#  include <cuda/std/detail/libcxx/include/__new/bad_alloc.h>
+#  include <cuda/std/__new/bad_alloc.h>
 
 #  if _CCCL_STD_VER >= 2014
 

diff --git a/libcudacxx/include/cuda/__memory_resource/cuda_pinned_memory_resource.h b/libcudacxx/include/cuda/__memory_resource/cuda_pinned_memory_resource.h
@@ -24,15 +24,15 @@
 #if !defined(_CCCL_COMPILER_MSVC_2017)
 
 #  if !defined(_CCCL_CUDA_COMPILER_NVCC) && !defined(_CCCL_CUDA_COMPILER_NVHPC)
-#    include <cuda_runtime_api.h>
+#    include <cuda_runtime.h>
 #  endif // !_CCCL_CUDA_COMPILER_NVCC && !_CCCL_CUDA_COMPILER_NVHPC
 
 #  include <cuda/__memory_resource/get_property.h>
 #  include <cuda/__memory_resource/properties.h>
 #  include <cuda/__memory_resource/resource.h>
 #  include <cuda/__memory_resource/resource_ref.h>
 #  include <cuda/std/__cuda/api_wrapper.h>
-#  include <cuda/std/detail/libcxx/include/__new/bad_alloc.h>
+#  include <cuda/std/__new/bad_alloc.h>
 
 #  if _CCCL_STD_VER >= 2014
 

diff --git a/libcudacxx/include/cuda/__memory_resource/get_property.h b/libcudacxx/include/cuda/__memory_resource/get_property.h
@@ -23,10 +23,10 @@
 
 #if !defined(_CCCL_COMPILER_MSVC_2017)
 
-#  include <cuda/std/detail/libcxx/include/__concepts/same_as.h>
-#  include <cuda/std/detail/libcxx/include/__type_traits/remove_const_ref.h>
-#  include <cuda/std/detail/libcxx/include/__type_traits/void_t.h>
-#  include <cuda/std/detail/libcxx/include/__utility/declval.h>
+#  include <cuda/std/__concepts/same_as.h>
+#  include <cuda/std/__type_traits/remove_const_ref.h>
+#  include <cuda/std/__type_traits/void_t.h>
+#  include <cuda/std/__utility/declval.h>
 
 #  if _CCCL_STD_VER >= 2014
 

diff --git a/libcudacxx/include/cuda/__memory_resource/resource.h b/libcudacxx/include/cuda/__memory_resource/resource.h
@@ -24,10 +24,10 @@
 #if !defined(_CCCL_COMPILER_MSVC_2017)
 
 #  include <cuda/__memory_resource/get_property.h>
-#  include <cuda/std/detail/libcxx/include/__concepts/all_of.h>
-#  include <cuda/std/detail/libcxx/include/__concepts/equality_comparable.h>
-#  include <cuda/std/detail/libcxx/include/__concepts/same_as.h>
-#  include <cuda/std/detail/libcxx/include/__type_traits/decay.h>
+#  include <cuda/std/__concepts/all_of.h>
+#  include <cuda/std/__concepts/equality_comparable.h>
+#  include <cuda/std/__concepts/same_as.h>
+#  include <cuda/std/__type_traits/decay.h>
 #  include <cuda/stream_ref>
 
 #  if _CCCL_STD_VER >= 2014

diff --git a/libcudacxx/include/cuda/__memory_resource/resource_ref.h b/libcudacxx/include/cuda/__memory_resource/resource_ref.h
@@ -25,11 +25,11 @@
 
 #  include <cuda/__memory_resource/get_property.h>
 #  include <cuda/__memory_resource/resource.h>
+#  include <cuda/std/__concepts/_One_of.h>
+#  include <cuda/std/__concepts/all_of.h>
+#  include <cuda/std/__memory/addressof.h>
+#  include <cuda/std/__type_traits/is_base_of.h>
 #  include <cuda/std/cstddef>
-#  include <cuda/std/detail/libcxx/include/__concepts/_One_of.h>
-#  include <cuda/std/detail/libcxx/include/__concepts/all_of.h>
-#  include <cuda/std/detail/libcxx/include/__memory/addressof.h>
-#  include <cuda/std/detail/libcxx/include/__type_traits/is_base_of.h>
 #  include <cuda/stream_ref>
 
 #  if _CCCL_STD_VER >= 2014

diff --git a/...__cuda/ptx/instructions/barrier_cluster.h → ...cuda/__ptx/instructions/barrier_cluster.h b/...__cuda/ptx/instructions/barrier_cluster.h → ...cuda/__ptx/instructions/barrier_cluster.h
@@ -22,9 +22,9 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__ptx/ptx_dot_variants.h>
+#include <cuda/__ptx/ptx_helper_functions.h>
 #include <cuda/std/cstdint>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_dot_variants.h>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_helper_functions.h>
 
 #include <nv/target> // __CUDA_MINIMUM_ARCH__ and friends
 

diff --git a/...e/__cuda/ptx/instructions/cp_async_bulk.h → ...e/cuda/__ptx/instructions/cp_async_bulk.h b/...e/__cuda/ptx/instructions/cp_async_bulk.h → ...e/cuda/__ptx/instructions/cp_async_bulk.h
@@ -22,9 +22,9 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__ptx/ptx_dot_variants.h>
+#include <cuda/__ptx/ptx_helper_functions.h>
 #include <cuda/std/cstdint>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_dot_variants.h>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_helper_functions.h>
 
 #include <nv/target> // __CUDA_MINIMUM_ARCH__ and friends
 

diff --git a/...instructions/cp_async_bulk_commit_group.h → ...instructions/cp_async_bulk_commit_group.h b/...instructions/cp_async_bulk_commit_group.h → ...instructions/cp_async_bulk_commit_group.h
@@ -22,9 +22,9 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__ptx/ptx_dot_variants.h>
+#include <cuda/__ptx/ptx_helper_functions.h>
 #include <cuda/std/cstdint>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_dot_variants.h>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_helper_functions.h>
 
 #include <nv/target> // __CUDA_MINIMUM_ARCH__ and friends
 

diff --git a/...a/ptx/instructions/cp_async_bulk_tensor.h → ...__ptx/instructions/cp_async_bulk_tensor.h b/...a/ptx/instructions/cp_async_bulk_tensor.h → ...__ptx/instructions/cp_async_bulk_tensor.h
@@ -22,9 +22,9 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__ptx/ptx_dot_variants.h>
+#include <cuda/__ptx/ptx_helper_functions.h>
 #include <cuda/std/cstdint>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_dot_variants.h>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_helper_functions.h>
 
 #include <nv/target> // __CUDA_MINIMUM_ARCH__ and friends
 

diff --git a/...x/instructions/cp_async_bulk_wait_group.h → ...x/instructions/cp_async_bulk_wait_group.h b/...x/instructions/cp_async_bulk_wait_group.h → ...x/instructions/cp_async_bulk_wait_group.h
@@ -22,9 +22,9 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__ptx/ptx_dot_variants.h>
+#include <cuda/__ptx/ptx_helper_functions.h>
 #include <cuda/std/cstdint>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_dot_variants.h>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_helper_functions.h>
 
 #include <nv/target> // __CUDA_MINIMUM_ARCH__ and friends
 

diff --git a/...a/ptx/instructions/cp_reduce_async_bulk.h → ...__ptx/instructions/cp_reduce_async_bulk.h b/...a/ptx/instructions/cp_reduce_async_bulk.h → ...__ptx/instructions/cp_reduce_async_bulk.h
@@ -22,9 +22,9 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__ptx/ptx_dot_variants.h>
+#include <cuda/__ptx/ptx_helper_functions.h>
 #include <cuda/std/cstdint>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_dot_variants.h>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_helper_functions.h>
 
 #include <nv/target> // __CUDA_MINIMUM_ARCH__ and friends
 

diff --git a/...nstructions/cp_reduce_async_bulk_tensor.h → ...nstructions/cp_reduce_async_bulk_tensor.h b/...nstructions/cp_reduce_async_bulk_tensor.h → ...nstructions/cp_reduce_async_bulk_tensor.h
@@ -22,9 +22,9 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__ptx/ptx_dot_variants.h>
+#include <cuda/__ptx/ptx_helper_functions.h>
 #include <cuda/std/cstdint>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_dot_variants.h>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_helper_functions.h>
 
 #include <nv/target> // __CUDA_MINIMUM_ARCH__ and friends
 

diff --git a/...x/include/__cuda/ptx/instructions/fence.h → ...x/include/cuda/__ptx/instructions/fence.h b/...x/include/__cuda/ptx/instructions/fence.h → ...x/include/cuda/__ptx/instructions/fence.h
@@ -22,9 +22,9 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__ptx/ptx_dot_variants.h>
+#include <cuda/__ptx/ptx_helper_functions.h>
 #include <cuda/std/cstdint>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_dot_variants.h>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_helper_functions.h>
 
 #include <nv/target> // __CUDA_MINIMUM_ARCH__ and friends
 

diff --git a/...nclude/__cuda/ptx/instructions/get_sreg.h → ...nclude/cuda/__ptx/instructions/get_sreg.h b/...nclude/__cuda/ptx/instructions/get_sreg.h → ...nclude/cuda/__ptx/instructions/get_sreg.h
@@ -22,9 +22,9 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__ptx/ptx_dot_variants.h>
+#include <cuda/__ptx/ptx_helper_functions.h>
 #include <cuda/std/cstdint>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_dot_variants.h>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_helper_functions.h>
 
 #include <nv/target> // __CUDA_MINIMUM_ARCH__ and friends
 

diff --git a/...lude/__cuda/ptx/instructions/getctarank.h → ...lude/cuda/__ptx/instructions/getctarank.h b/...lude/__cuda/ptx/instructions/getctarank.h → ...lude/cuda/__ptx/instructions/getctarank.h
@@ -22,9 +22,9 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__ptx/ptx_dot_variants.h>
+#include <cuda/__ptx/ptx_helper_functions.h>
 #include <cuda/std/cstdint>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_dot_variants.h>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_helper_functions.h>
 
 #include <nv/target> // __CUDA_MINIMUM_ARCH__ and friends
 

diff --git a/...__cuda/ptx/instructions/mbarrier_arrive.h → ...cuda/__ptx/instructions/mbarrier_arrive.h b/...__cuda/ptx/instructions/mbarrier_arrive.h → ...cuda/__ptx/instructions/mbarrier_arrive.h
@@ -22,9 +22,9 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__ptx/ptx_dot_variants.h>
+#include <cuda/__ptx/ptx_helper_functions.h>
 #include <cuda/std/cstdint>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_dot_variants.h>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_helper_functions.h>
 
 #include <nv/target> // __CUDA_MINIMUM_ARCH__ and friends
 

diff --git a/...e/__cuda/ptx/instructions/mbarrier_init.h → ...e/cuda/__ptx/instructions/mbarrier_init.h b/...e/__cuda/ptx/instructions/mbarrier_init.h → ...e/cuda/__ptx/instructions/mbarrier_init.h
@@ -22,9 +22,9 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__ptx/ptx_dot_variants.h>
+#include <cuda/__ptx/ptx_helper_functions.h>
 #include <cuda/std/cstdint>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_dot_variants.h>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_helper_functions.h>
 
 #include <nv/target> // __CUDA_MINIMUM_ARCH__ and friends
 

diff --git a/...e/__cuda/ptx/instructions/mbarrier_wait.h → ...e/cuda/__ptx/instructions/mbarrier_wait.h b/...e/__cuda/ptx/instructions/mbarrier_wait.h → ...e/cuda/__ptx/instructions/mbarrier_wait.h
@@ -22,9 +22,9 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__ptx/ptx_dot_variants.h>
+#include <cuda/__ptx/ptx_helper_functions.h>
 #include <cuda/std/cstdint>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_dot_variants.h>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_helper_functions.h>
 
 #include <nv/target> // __CUDA_MINIMUM_ARCH__ and friends
 

diff --git a/...clude/__cuda/ptx/instructions/red_async.h → ...clude/cuda/__ptx/instructions/red_async.h b/...clude/__cuda/ptx/instructions/red_async.h → ...clude/cuda/__ptx/instructions/red_async.h
@@ -22,9 +22,9 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__ptx/ptx_dot_variants.h>
+#include <cuda/__ptx/ptx_helper_functions.h>
 #include <cuda/std/cstdint>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_dot_variants.h>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_helper_functions.h>
 
 #include <nv/target> // __CUDA_MINIMUM_ARCH__ and friends
 

diff --git a/...nclude/__cuda/ptx/instructions/st_async.h → ...nclude/cuda/__ptx/instructions/st_async.h b/...nclude/__cuda/ptx/instructions/st_async.h → ...nclude/cuda/__ptx/instructions/st_async.h
@@ -22,9 +22,9 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__ptx/ptx_dot_variants.h>
+#include <cuda/__ptx/ptx_helper_functions.h>
 #include <cuda/std/cstdint>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_dot_variants.h>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_helper_functions.h>
 
 #include <nv/target> // __CUDA_MINIMUM_ARCH__ and friends
 

diff --git a/...tx/instructions/tensormap_cp_fenceproxy.h → ...tx/instructions/tensormap_cp_fenceproxy.h b/...tx/instructions/tensormap_cp_fenceproxy.h → ...tx/instructions/tensormap_cp_fenceproxy.h
@@ -22,9 +22,9 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__ptx/ptx_dot_variants.h>
+#include <cuda/__ptx/ptx_helper_functions.h>
 #include <cuda/std/cstdint>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_dot_variants.h>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_helper_functions.h>
 
 #include <nv/target> // __CUDA_MINIMUM_ARCH__ and friends
 

diff --git a/...cuda/ptx/instructions/tensormap_replace.h → ...da/__ptx/instructions/tensormap_replace.h b/...cuda/ptx/instructions/tensormap_replace.h → ...da/__ptx/instructions/tensormap_replace.h
@@ -22,9 +22,9 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__ptx/ptx_dot_variants.h>
+#include <cuda/__ptx/ptx_helper_functions.h>
 #include <cuda/std/cstdint>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_dot_variants.h>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_helper_functions.h>
 
 #include <nv/target> // __CUDA_MINIMUM_ARCH__ and friends
 

diff --git a/...and_communication_instructions_mbarrier.h → ...and_communication_instructions_mbarrier.h b/...and_communication_instructions_mbarrier.h → ...and_communication_instructions_mbarrier.h
@@ -22,9 +22,9 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/__ptx/ptx_dot_variants.h>
+#include <cuda/__ptx/ptx_helper_functions.h>
 #include <cuda/std/cstdint>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_dot_variants.h>
-#include <cuda/std/detail/libcxx/include/__cuda/ptx/ptx_helper_functions.h>
 
 _LIBCUDACXX_BEGIN_NAMESPACE_CUDA_PTX
 

diff --git a/...cxx/include/__cuda/ptx/ptx_dot_variants.h → ...cxx/include/cuda/__ptx/ptx_dot_variants.h b/...cxx/include/__cuda/ptx/ptx_dot_variants.h → ...cxx/include/cuda/__ptx/ptx_dot_variants.h
@@ -12,7 +12,7 @@
 #ifndef _CUDA_PTX_DOT_VARIANTS_H_
 #define _CUDA_PTX_DOT_VARIANTS_H_
 
-#include <cuda/std/detail/libcxx/include/__type_traits/integral_constant.h> // std::integral_constant
+#include <cuda/std/__type_traits/integral_constant.h> // std::integral_constant
 
 #if defined(_CCCL_IMPLICIT_SYSTEM_HEADER_GCC)
 #  pragma GCC system_header

diff --git a/...include/__cuda/ptx/ptx_helper_functions.h → ...include/cuda/__ptx/ptx_helper_functions.h b/...include/__cuda/ptx/ptx_helper_functions.h → ...include/cuda/__ptx/ptx_helper_functions.h
@@ -12,9 +12,9 @@
 #ifndef _CUDA_PTX_HELPER_FUNCTIONS_H_
 #define _CUDA_PTX_HELPER_FUNCTIONS_H_
 
+#include <cuda/std/__type_traits/integral_constant.h>
 #include <cuda/std/cstddef> // size_t
 #include <cuda/std/cstdint> // uint32_t
-#include <cuda/std/detail/libcxx/include/__type_traits/integral_constant.h>
 
 #if defined(_CCCL_IMPLICIT_SYSTEM_HEADER_GCC)
 #  pragma GCC system_header