[libcxx-commits] [libcxx] [libcxxabi] [libunwind] [llvm] [libc++] Unify the benchmarks with the test suite (PR #101399)
Louis Dionne via libcxx-commits
libcxx-commits at lists.llvm.org
Tue Nov 5 06:16:04 PST 2024
https://github.com/ldionne updated https://github.com/llvm/llvm-project/pull/101399
>From dd2d2ee9ff54b5263d0f65bcac64407ffb0c9e12 Mon Sep 17 00:00:00 2001
From: Louis Dionne <ldionne.2 at gmail.com>
Date: Thu, 25 Jul 2024 10:41:10 -0400
Subject: [PATCH] [libc++] Unify the benchmarks with the test suite
Instead of building the benchmarks separately via CMake and running them
separately from the test suite, this patch merges the benchmarks into
the test suite and handles both uniformly.
As a result:
- It is now possible to run individual benchmarks like we run tests
(e.g. using libcxx-lit), which is a huge quality-of-life improvement.
- The benchmarks will be run under exactly the same configuration as
the rest of the tests, which is a nice simplification. This does
mean that one has to be careful to enable the desired optimization
flags when running benchmarks, but that is easy with e.g.
`libcxx-lit <...> --param optimization=speed`.
- Benchmarks can use the same annotations as the rest of the test
suite, such as `// UNSUPPORTED` & friends.
When running the tests via `check-cxx`, we only compile the benchmarks
because running them would be too time consuming. This introduces a bit
of complexity in the testing setup, and instead it would be better to
allow passing a --dry-run flag to GoogleBenchmark executables, which is
the topic of a GoogleBenchmark issue.
I am not really satisfied with the layering violation of adding the
%{benchmark_flags} substitution to cmake-bridge, however I believe
this can be improved in the future.
---
.github/workflows/libcxx-build-and-test.yaml | 3 -
libcxx/CMakeLists.txt | 13 +-
libcxx/docs/TestingLibcxx.rst | 49 ++---
libcxx/docs/VendorDocumentation.rst | 12 +-
libcxx/test/CMakeLists.txt | 15 +-
libcxx/test/benchmarks/CMakeLists.txt | 185 +-----------------
.../test/benchmarks/algorithms/min.bench.cpp | 2 +
.../atomic_wait_vs_mutex_lock.bench.cpp | 4 -
.../test/benchmarks/formatter_int.bench.cpp | 2 +-
libcxx/test/benchmarks/lit.cfg.py.in | 23 ---
libcxx/test/benchmarks/lit.site.cfg.py.in | 10 -
libcxx/test/configs/cmake-bridge.cfg.in | 1 +
libcxx/utils/ci/run-buildbot | 10 -
libcxx/utils/libcxx/test/config.py | 2 +-
libcxx/utils/libcxx/test/format.py | 31 ++-
libcxx/utils/libcxx/test/googlebenchmark.py | 125 ------------
libcxx/utils/libcxx/test/params.py | 10 +
libcxxabi/test/configs/cmake-bridge.cfg.in | 1 +
libunwind/test/configs/cmake-bridge.cfg.in | 1 +
19 files changed, 91 insertions(+), 408 deletions(-)
delete mode 100644 libcxx/test/benchmarks/lit.cfg.py.in
delete mode 100644 libcxx/test/benchmarks/lit.site.cfg.py.in
delete mode 100644 libcxx/utils/libcxx/test/googlebenchmark.py
diff --git a/.github/workflows/libcxx-build-and-test.yaml b/.github/workflows/libcxx-build-and-test.yaml
index 657e7f1e7f0f7e..32e9350d7353da 100644
--- a/.github/workflows/libcxx-build-and-test.yaml
+++ b/.github/workflows/libcxx-build-and-test.yaml
@@ -157,9 +157,6 @@ jobs:
'generic-no-rtti',
'generic-optimized-speed',
'generic-static',
- # TODO Find a better place for the benchmark and bootstrapping builds to live. They're either very expensive
- # or don't provide much value since the benchmark run results are too noise on the bots.
- 'benchmarks',
'bootstrapping-build'
]
machine: [ 'libcxx-runners-8-set' ]
diff --git a/libcxx/CMakeLists.txt b/libcxx/CMakeLists.txt
index 95a7d10f055ea7..4993e73e7841ce 100644
--- a/libcxx/CMakeLists.txt
+++ b/libcxx/CMakeLists.txt
@@ -150,12 +150,13 @@ message(STATUS "Using libc++ testing configuration: ${LIBCXX_TEST_CONFIG}")
set(LIBCXX_TEST_PARAMS "" CACHE STRING
"A list of parameters to run the Lit test suite with.")
-# Benchmark options -----------------------------------------------------------
-option(LIBCXX_INCLUDE_BENCHMARKS "Build the libc++ benchmarks and their dependencies" ON)
-
-set(LIBCXX_BENCHMARK_TEST_ARGS_DEFAULT --benchmark_min_time=0.01)
-set(LIBCXX_BENCHMARK_TEST_ARGS "${LIBCXX_BENCHMARK_TEST_ARGS_DEFAULT}" CACHE STRING
- "Arguments to pass when running the benchmarks using check-cxx-benchmarks")
+# TODO: On Windows and MinGW, disable the benchmarks since we don't know how to build GoogleBenchmark yet
+if (WIN32 OR MINGW)
+ set(_include_benchmarks OFF)
+else()
+ set(_include_benchmarks ON)
+endif()
+option(LIBCXX_INCLUDE_BENCHMARKS "Build the libc++ benchmarks and their dependencies" ${_include_benchmarks})
option(LIBCXX_INCLUDE_DOCS "Build the libc++ documentation." ${LLVM_INCLUDE_DOCS})
set(LIBCXX_LIBDIR_SUFFIX "${LLVM_LIBDIR_SUFFIX}" CACHE STRING
diff --git a/libcxx/docs/TestingLibcxx.rst b/libcxx/docs/TestingLibcxx.rst
index 3d43a66745bedf..6d2c954489f62e 100644
--- a/libcxx/docs/TestingLibcxx.rst
+++ b/libcxx/docs/TestingLibcxx.rst
@@ -392,6 +392,10 @@ Test Filenames`_ when determining the names for new test files.
of Lit test to be executed. This can be used to generate multiple Lit tests from a single source file, which is useful for testing repetitive properties
in the library. Be careful not to abuse this since this is not a replacement for usual code reuse techniques.
+ * - ``FOO.bench.cpp``
+ - A benchmark test. These tests are linked against the GoogleBenchmark library and generally consist of micro-benchmarks of individual
+ components of the library.
+
libc++-Specific Lit Features
----------------------------
@@ -438,43 +442,29 @@ Libc++ contains benchmark tests separately from the test of the test suite.
The benchmarks are written using the `Google Benchmark`_ library, a copy of which
is stored in the libc++ repository.
-For more information about using the Google Benchmark library see the
+For more information about using the Google Benchmark library, see the
`official documentation <https://github.com/google/benchmark>`_.
-.. _`Google Benchmark`: https://github.com/google/benchmark
-
-Building Benchmarks
--------------------
-
-The benchmark tests are not built by default. The benchmarks can be built using
-the ``cxx-benchmarks`` target.
-
-An example build would look like:
+The benchmarks are located under ``libcxx/test/benchmarks``. Running a benchmark
+works in the same way as running a test. Both the benchmarks and the tests share
+the same configuration, so make sure to enable the relevant optimization level
+when running the benchmarks. For example,
.. code-block:: bash
- $ ninja -C build cxx-benchmarks
-
-This will build all of the benchmarks under ``<libcxx>/test/benchmarks`` to be
-built against the just-built libc++. The compiled tests are output into
-``build/libcxx/test/benchmarks``.
+ $ libcxx/utils/libcxx-lit <build> -sv libcxx/test/benchmarks/string.bench.cpp --param optimization=speed
-Running Benchmarks
-------------------
+If you want to see where a benchmark is located (e.g. you want to store the executable
+for subsequent analysis), you can print that information by passing ``--show-all`` to
+``lit``. That will print the command-lines being executed, which includes the location
+of the executable created for that benchmark.
-The benchmarks must be run manually by the user. Currently there is no way
-to run them as part of the build.
-
-For example:
-
-.. code-block:: bash
-
- $ cd build/libcxx/test/benchmarks
- $ ./find.bench.out # Runs all the benchmarks
- $ ./find.bench.out --benchmark_filter="bm_ranges_find<std::vector<char>>" # Only runs that specific benchmark
-
-For more information about running benchmarks see `Google Benchmark`_.
+Note that benchmarks are only dry-run when run via the ``check-cxx`` target since
+we only want to make sure they don't rot. Do not rely on the results of benchmarks
+run through ``check-cxx`` for anything, instead run the benchmarks manually using
+the instructions for running individual tests.
+.. _`Google Benchmark`: https://github.com/google/benchmark
.. _testing-hardening-assertions:
@@ -518,4 +508,3 @@ A toy example:
Note that error messages are only tested (matched) if the ``debug``
hardening mode is used.
-
diff --git a/libcxx/docs/VendorDocumentation.rst b/libcxx/docs/VendorDocumentation.rst
index 3795381264c93b..959a28607d75dd 100644
--- a/libcxx/docs/VendorDocumentation.rst
+++ b/libcxx/docs/VendorDocumentation.rst
@@ -244,7 +244,8 @@ General purpose options
**Default**: ``ON`` (or value of ``LLVM_INCLUDE_TESTS``)
- Build the libc++ tests.
+ Build the libc++ test suite, which includes various types of tests like conformance
+ tests, vendor-specific tests and benchmarks.
.. option:: LIBCXX_INCLUDE_BENCHMARKS:BOOL
@@ -253,15 +254,6 @@ General purpose options
Build the libc++ benchmark tests and the Google Benchmark library needed
to support them.
-.. option:: LIBCXX_BENCHMARK_TEST_ARGS:STRING
-
- **Default**: ``--benchmark_min_time=0.01``
-
- A semicolon list of arguments to pass when running the libc++ benchmarks using the
- ``check-cxx-benchmarks`` rule. By default we run the benchmarks for a very short amount of time,
- since the primary use of ``check-cxx-benchmarks`` is to get test and sanitizer coverage, not to
- get accurate measurements.
-
.. option:: LIBCXX_ASSERTION_HANDLER_FILE:PATH
**Default**:: ``"${CMAKE_CURRENT_SOURCE_DIR}/vendor/llvm/default_assertion_handler.in"``
diff --git a/libcxx/test/CMakeLists.txt b/libcxx/test/CMakeLists.txt
index aa49979affedfa..aedf0406dd99d2 100644
--- a/libcxx/test/CMakeLists.txt
+++ b/libcxx/test/CMakeLists.txt
@@ -1,10 +1,6 @@
include(HandleLitArguments)
add_subdirectory(tools)
-if (LIBCXX_INCLUDE_BENCHMARKS)
- add_subdirectory(benchmarks)
-endif()
-
# Install the library at a fake location so we can run the test suite against it.
# This ensures that we run the test suite against a setup that matches what we ship
# in production as closely as possible (in terms of file paths, rpaths, etc).
@@ -66,6 +62,16 @@ set(SERIALIZED_LIT_PARAMS "# Lit parameters serialized here for llvm-lit to pick
serialize_lit_string_param(SERIALIZED_LIT_PARAMS compiler "${CMAKE_CXX_COMPILER}")
+if (LIBCXX_INCLUDE_BENCHMARKS
+ AND LIBCXX_ENABLE_LOCALIZATION AND LIBCXX_ENABLE_THREADS AND LIBCXX_ENABLE_FILESYSTEM AND LIBCXX_ENABLE_RANDOM_DEVICE
+ AND LIBCXX_ENABLE_EXCEPTIONS AND LIBCXX_ENABLE_RTTI) # TODO: The benchmarks should work with exceptions/RTTI disabled
+ add_subdirectory(benchmarks)
+ set(_libcxx_benchmark_mode "dry-run")
+else()
+ serialize_lit_string_param(SERIALIZED_LIT_PARAMS enable_benchmarks "no")
+ set(_libcxx_benchmark_mode "no")
+endif()
+
if (NOT LIBCXX_ENABLE_EXCEPTIONS)
serialize_lit_param(SERIALIZED_LIT_PARAMS enable_exceptions False)
endif()
@@ -102,4 +108,5 @@ configure_lit_site_cfg(
add_lit_testsuite(check-cxx
"Running libcxx tests"
${CMAKE_CURRENT_BINARY_DIR}
+ PARAMS enable_benchmarks="${_libcxx_benchmark_mode}"
DEPENDS cxx-test-depends)
diff --git a/libcxx/test/benchmarks/CMakeLists.txt b/libcxx/test/benchmarks/CMakeLists.txt
index 5e829a639d39a5..b5a4aae82c06ab 100644
--- a/libcxx/test/benchmarks/CMakeLists.txt
+++ b/libcxx/test/benchmarks/CMakeLists.txt
@@ -1,10 +1,8 @@
-include(ExternalProject)
-include(CheckCXXCompilerFlag)
-
#==============================================================================
# Build Google Benchmark
#==============================================================================
+include(ExternalProject)
set(BENCHMARK_COMPILE_FLAGS
-Wno-unused-command-line-argument
-nostdinc++
@@ -24,6 +22,12 @@ if (DEFINED LIBCXX_CXX_ABI_LIBRARY_PATH)
endif()
split_list(BENCHMARK_COMPILE_FLAGS)
+set(BENCHMARK_CXX_LIBRARIES)
+list(APPEND BENCHMARK_CXX_LIBRARIES c++)
+if (NOT LIBCXX_ENABLE_SHARED)
+ list(APPEND BENCHMARK_CXX_LIBRARIES c++abi)
+endif()
+
ExternalProject_Add(google-benchmark
EXCLUDE_FROM_ALL ON
DEPENDS cxx cxx-headers
@@ -37,176 +41,7 @@ ExternalProject_Add(google-benchmark
-DCMAKE_INSTALL_PREFIX:PATH=<INSTALL_DIR>
-DCMAKE_CXX_FLAGS:STRING=${BENCHMARK_COMPILE_FLAGS}
-DBENCHMARK_USE_LIBCXX:BOOL=ON
- -DBENCHMARK_ENABLE_TESTING:BOOL=OFF)
-
-#==============================================================================
-# Benchmark tests configuration
-#==============================================================================
-add_custom_target(cxx-benchmarks)
-set(BENCHMARK_OUTPUT_DIR ${CMAKE_CURRENT_BINARY_DIR})
-set(BENCHMARK_INSTALL_DIR ${CMAKE_CURRENT_BINARY_DIR}/google-benchmark)
-
-add_library( cxx-benchmarks-flags INTERFACE)
-
-# TODO(cmake): remove. This is a workaround to prevent older versions of GCC
-# from failing the configure step because they don't support C++23.
-if (CMAKE_CXX_COMPILER_ID STREQUAL "GNU" AND CMAKE_CXX_COMPILER_VERSION VERSION_LESS "13.0")
- return()
-endif()
-#TODO(cmake): remove the `add_compile_options`. Currently we have to explicitly
-# pass the `std:c++latest` flag on Windows to work around an issue where
-# requesting `cxx_std_23` results in an error -- somehow CMake fails to
-# translate the `c++23` flag into `c++latest`, and the highest numbered C++
-# version that MSVC flags support is C++20.
-if (MSVC)
- add_compile_options(/std:c++latest)
-# ibm-clang does not recognize the cxx_std_23 flag, so use this as a temporary
-# workaround on AIX as well.
-elseif (${CMAKE_SYSTEM_NAME} MATCHES "AIX")
- add_compile_options(-std=c++23)
-else()
- target_compile_features( cxx-benchmarks-flags INTERFACE cxx_std_23)
-endif()
-
-target_compile_options(cxx-benchmarks-flags INTERFACE -fsized-deallocation -nostdinc++
- ${SANITIZER_FLAGS} -Wno-user-defined-literals -Wno-suggest-override)
-target_include_directories(cxx-benchmarks-flags INTERFACE "${LIBCXX_GENERATED_INCLUDE_DIR}"
- INTERFACE "${BENCHMARK_INSTALL_DIR}/include"
- INTERFACE "${LIBCXX_SOURCE_DIR}/test/support")
-target_link_options(cxx-benchmarks-flags INTERFACE -lm -nostdlib++
- "-L${BENCHMARK_INSTALL_DIR}/lib" "-L${BENCHMARK_INSTALL_DIR}/lib64"
- ${SANITIZER_FLAGS})
+ -DBENCHMARK_ENABLE_TESTING:BOOL=OFF
+ -DBENCHMARK_CXX_LIBRARIES:STRING="${BENCHMARK_CXX_LIBRARIES}")
-set(libcxx_benchmark_targets)
-
-function(add_benchmark_test name source_file)
- set(libcxx_target ${name}_libcxx)
- list(APPEND libcxx_benchmark_targets ${libcxx_target})
- add_executable(${libcxx_target} EXCLUDE_FROM_ALL ${source_file})
- target_link_libraries(${libcxx_target} PRIVATE cxx-benchmarks-flags)
- add_dependencies(${libcxx_target} cxx google-benchmark)
- add_dependencies(cxx-benchmarks ${libcxx_target})
- if (LIBCXX_ENABLE_SHARED)
- target_link_libraries(${libcxx_target} PRIVATE cxx_shared)
- else()
- target_link_libraries(${libcxx_target} PRIVATE cxx_static)
- endif()
- target_link_libraries(${libcxx_target} PRIVATE cxx_experimental benchmark)
- if (LLVM_USE_SANITIZER)
- target_link_libraries(${libcxx_target} PRIVATE -ldl)
- endif()
- set_target_properties(${libcxx_target}
- PROPERTIES
- OUTPUT_NAME "${name}.bench.out"
- RUNTIME_OUTPUT_DIRECTORY "${BENCHMARK_OUTPUT_DIR}"
- CXX_EXTENSIONS NO)
- cxx_link_system_libraries(${libcxx_target})
-endfunction()
-
-
-#==============================================================================
-# Register Benchmark tests
-#==============================================================================
-set(BENCHMARK_TESTS
- algorithms.partition_point.bench.cpp
- algorithms/count.bench.cpp
- algorithms/equal.bench.cpp
- algorithms/find.bench.cpp
- algorithms/fill.bench.cpp
- algorithms/for_each.bench.cpp
- algorithms/lexicographical_compare.bench.cpp
- algorithms/lower_bound.bench.cpp
- algorithms/make_heap.bench.cpp
- algorithms/make_heap_then_sort_heap.bench.cpp
- algorithms/min.bench.cpp
- algorithms/minmax.bench.cpp
- algorithms/min_max_element.bench.cpp
- algorithms/mismatch.bench.cpp
- algorithms/pop_heap.bench.cpp
- algorithms/pstl.stable_sort.bench.cpp
- algorithms/push_heap.bench.cpp
- algorithms/ranges_contains.bench.cpp
- algorithms/ranges_ends_with.bench.cpp
- algorithms/ranges_make_heap.bench.cpp
- algorithms/ranges_make_heap_then_sort_heap.bench.cpp
- algorithms/ranges_pop_heap.bench.cpp
- algorithms/ranges_push_heap.bench.cpp
- algorithms/ranges_sort.bench.cpp
- algorithms/ranges_sort_heap.bench.cpp
- algorithms/ranges_stable_sort.bench.cpp
- algorithms/set_intersection.bench.cpp
- algorithms/sort.bench.cpp
- algorithms/sort_heap.bench.cpp
- algorithms/stable_sort.bench.cpp
- atomic_wait.bench.cpp
- atomic_wait_vs_mutex_lock.bench.cpp
- libcxxabi/dynamic_cast.bench.cpp
- libcxxabi/dynamic_cast_old_stress.bench.cpp
- allocation.bench.cpp
- deque.bench.cpp
- deque_iterator.bench.cpp
- exception_ptr.bench.cpp
- filesystem.bench.cpp
- format/write_double_comparison.bench.cpp
- format/write_int_comparison.bench.cpp
- format/write_string_comparison.bench.cpp
- format_to_n.bench.cpp
- format_to.bench.cpp
- format.bench.cpp
- formatted_size.bench.cpp
- formatter_float.bench.cpp
- formatter_int.bench.cpp
- function.bench.cpp
- hash.bench.cpp
- join_view.bench.cpp
- lexicographical_compare_three_way.bench.cpp
- map.bench.cpp
- monotonic_buffer.bench.cpp
- numeric/gcd.bench.cpp
- ordered_set.bench.cpp
- shared_mutex_vs_mutex.bench.cpp
- stop_token.bench.cpp
- std_format_spec_string_unicode.bench.cpp
- std_format_spec_string_unicode_escape.bench.cpp
- string.bench.cpp
- stringstream.bench.cpp
- system_error.bench.cpp
- to_chars.bench.cpp
- unordered_set_operations.bench.cpp
- util_smartptr.bench.cpp
- variant_visit_1.bench.cpp
- variant_visit_2.bench.cpp
- variant_visit_3.bench.cpp
- vector_operations.bench.cpp
- )
-
-foreach(test_path ${BENCHMARK_TESTS})
- get_filename_component(test_file "${test_path}" NAME)
- string(REPLACE ".bench.cpp" "" test_name "${test_file}")
- if (NOT DEFINED ${test_name}_REPORTED)
- message(STATUS "Adding Benchmark: ${test_file}")
- # Only report the adding of the benchmark once.
- set(${test_name}_REPORTED ON CACHE INTERNAL "")
- endif()
- add_benchmark_test(${test_name} ${test_path})
-endforeach()
-
-if (LIBCXX_INCLUDE_TESTS)
- include(AddLLVM)
-
- configure_lit_site_cfg(
- ${CMAKE_CURRENT_SOURCE_DIR}/lit.cfg.py.in
- ${CMAKE_CURRENT_BINARY_DIR}/lit.cfg.py)
-
- configure_lit_site_cfg(
- ${CMAKE_CURRENT_SOURCE_DIR}/lit.site.cfg.py.in
- ${CMAKE_CURRENT_BINARY_DIR}/lit.site.cfg.py)
-
- set(BENCHMARK_LIT_ARGS "--show-all --show-xfail --show-unsupported ${LIT_ARGS_DEFAULT}")
-
- add_lit_target(check-cxx-benchmarks
- "Running libcxx benchmarks tests"
- ${CMAKE_CURRENT_BINARY_DIR}
- DEPENDS cxx-benchmarks cxx-test-depends
- ARGS ${BENCHMARK_LIT_ARGS})
-endif()
+add_dependencies(cxx-test-depends google-benchmark)
diff --git a/libcxx/test/benchmarks/algorithms/min.bench.cpp b/libcxx/test/benchmarks/algorithms/min.bench.cpp
index ee426c7e727aa4..9928afd5677378 100644
--- a/libcxx/test/benchmarks/algorithms/min.bench.cpp
+++ b/libcxx/test/benchmarks/algorithms/min.bench.cpp
@@ -78,6 +78,8 @@ BENCHMARK(BM_std_min<unsigned char>)->Apply(run_sizes);
BENCHMARK(BM_std_min<unsigned short>)->Apply(run_sizes);
BENCHMARK(BM_std_min<unsigned int>)->Apply(run_sizes);
BENCHMARK(BM_std_min<unsigned long long>)->Apply(run_sizes);
+#ifndef TEST_HAS_NO_INT128
BENCHMARK(BM_std_min<unsigned __int128>)->Apply(run_sizes);
+#endif
BENCHMARK_MAIN();
diff --git a/libcxx/test/benchmarks/atomic_wait_vs_mutex_lock.bench.cpp b/libcxx/test/benchmarks/atomic_wait_vs_mutex_lock.bench.cpp
index b6f7f40b5cd34b..221fc086d2a626 100644
--- a/libcxx/test/benchmarks/atomic_wait_vs_mutex_lock.bench.cpp
+++ b/libcxx/test/benchmarks/atomic_wait_vs_mutex_lock.bench.cpp
@@ -8,10 +8,6 @@
// UNSUPPORTED: c++03, c++11, c++14, c++17
-// To run this test, build libcxx and cxx-benchmarks targets
-// cd third-party/benchmark/tools
-// ./compare.py filters ../../../build/libcxx/benchmarks/atomic_wait_vs_mutex_lock.libcxx.out BM_atomic_wait BM_mutex
-
#include <atomic>
#include <mutex>
#include <numeric>
diff --git a/libcxx/test/benchmarks/formatter_int.bench.cpp b/libcxx/test/benchmarks/formatter_int.bench.cpp
index 68f0f18bde4d20..db0edab6ae0522 100644
--- a/libcxx/test/benchmarks/formatter_int.bench.cpp
+++ b/libcxx/test/benchmarks/formatter_int.bench.cpp
@@ -41,6 +41,7 @@ BENCHMARK(BM_Basic<uint64_t>);
BENCHMARK(BM_Basic<int64_t>);
// Ideally the low values of a 128-bit value are all dispatched to a 64-bit routine.
+#ifndef TEST_HAS_NO_INT128
template <class T>
static void BM_BasicLow(benchmark::State& state) {
using U = std::conditional_t<std::is_signed_v<T>, int64_t, uint64_t>;
@@ -52,7 +53,6 @@ static void BM_BasicLow(benchmark::State& state) {
for (auto value : data)
benchmark::DoNotOptimize(std::format_to(output.begin(), "{}", value));
}
-#ifndef TEST_HAS_NO_INT128
BENCHMARK(BM_BasicLow<__uint128_t>);
BENCHMARK(BM_BasicLow<__int128_t>);
diff --git a/libcxx/test/benchmarks/lit.cfg.py.in b/libcxx/test/benchmarks/lit.cfg.py.in
deleted file mode 100644
index a3ce2477b9a858..00000000000000
--- a/libcxx/test/benchmarks/lit.cfg.py.in
+++ /dev/null
@@ -1,23 +0,0 @@
-# -*- Python -*- vim: set ft=python ts=4 sw=4 expandtab tw=79:
-# Configuration file for the 'lit' test runner.
-import os
-import site
-
-site.addsitedir(os.path.join("@LIBCXX_SOURCE_DIR@", "utils"))
-from libcxx.test.googlebenchmark import GoogleBenchmark
-
-# Tell pylint that we know config and lit_config exist somewhere.
-if "PYLINT_IMPORT" in os.environ:
- config = object()
- lit_config = object()
-
-# name: The name of this test suite.
-config.name = "libc++ benchmarks"
-config.suffixes = []
-
-config.test_exec_root = "@CMAKE_CURRENT_BINARY_DIR@"
-config.test_source_root = "@CMAKE_CURRENT_BINARY_DIR@"
-
-config.test_format = GoogleBenchmark(
- test_sub_dirs=".", test_suffix=".bench.out", benchmark_args=config.benchmark_args
-)
diff --git a/libcxx/test/benchmarks/lit.site.cfg.py.in b/libcxx/test/benchmarks/lit.site.cfg.py.in
deleted file mode 100644
index 6d4b0ca3dae0ef..00000000000000
--- a/libcxx/test/benchmarks/lit.site.cfg.py.in
+++ /dev/null
@@ -1,10 +0,0 @@
- at LIT_SITE_CFG_IN_HEADER@
-
-import sys
-
-config.libcxx_src_root = "@LIBCXX_SOURCE_DIR@"
-config.libcxx_obj_root = "@LIBCXX_BINARY_DIR@"
-config.benchmark_args = "@LIBCXX_BENCHMARK_TEST_ARGS@".split(';')
-
-# Let the main config do the real work.
-lit_config.load_config(config, "@CMAKE_CURRENT_BINARY_DIR@/lit.cfg.py")
\ No newline at end of file
diff --git a/libcxx/test/configs/cmake-bridge.cfg.in b/libcxx/test/configs/cmake-bridge.cfg.in
index 139a6cafa2cfb3..b802af4984325a 100644
--- a/libcxx/test/configs/cmake-bridge.cfg.in
+++ b/libcxx/test/configs/cmake-bridge.cfg.in
@@ -30,3 +30,4 @@ config.substitutions.append(('%{target-include-dir}', '@LIBCXX_TESTING_INSTALL_P
config.substitutions.append(('%{lib-dir}', '@LIBCXX_TESTING_INSTALL_PREFIX@/@LIBCXX_INSTALL_LIBRARY_DIR@'))
config.substitutions.append(('%{module-dir}', '@LIBCXX_TESTING_INSTALL_PREFIX@/@LIBCXX_INSTALL_MODULES_DIR@'))
config.substitutions.append(('%{test-tools-dir}', '@LIBCXX_TEST_TOOLS_PATH@'))
+config.substitutions.append(('%{benchmark_flags}', '-I @LIBCXX_BINARY_DIR@/test/benchmarks/google-benchmark/include -L @LIBCXX_BINARY_DIR@/test/benchmarks/google-benchmark/lib -l benchmark'))
diff --git a/libcxx/utils/ci/run-buildbot b/libcxx/utils/ci/run-buildbot
index 0b72f8c4031031..3df7b00a8aa09d 100755
--- a/libcxx/utils/ci/run-buildbot
+++ b/libcxx/utils/ci/run-buildbot
@@ -187,11 +187,6 @@ function check-abi-list() {
)
}
-function check-cxx-benchmarks() {
- step "Running the benchmarks"
- ${NINJA} -vC "${BUILD_DIR}" check-cxx-benchmarks
-}
-
function test-armv7m-picolibc() {
clean
@@ -640,11 +635,6 @@ apple-system)
step "Running the libunwind tests"
${NINJA} -vC "${BUILD_DIR}/unwind" check-unwind
;;
-benchmarks)
- clean
- generate-cmake
- check-cxx-benchmarks
-;;
aarch64)
clean
generate-cmake -C "${MONOREPO_ROOT}/libcxx/cmake/caches/AArch64.cmake"
diff --git a/libcxx/utils/libcxx/test/config.py b/libcxx/utils/libcxx/test/config.py
index 9a71c494030d70..0840c46d7bfaea 100644
--- a/libcxx/utils/libcxx/test/config.py
+++ b/libcxx/utils/libcxx/test/config.py
@@ -52,7 +52,7 @@ def configure(parameters, features, config, lit_config):
)
# Print the basic substitutions
- for sub in ("%{cxx}", "%{flags}", "%{compile_flags}", "%{link_flags}", "%{exec}"):
+ for sub in ("%{cxx}", "%{flags}", "%{compile_flags}", "%{link_flags}", "%{benchmark_flags}", "%{exec}"):
note("Using {} substitution: '{}'".format(sub, _getSubstitution(sub, config)))
# Print all available features
diff --git a/libcxx/utils/libcxx/test/format.py b/libcxx/utils/libcxx/test/format.py
index 7e5281c0b74064..ccd27170e197be 100644
--- a/libcxx/utils/libcxx/test/format.py
+++ b/libcxx/utils/libcxx/test/format.py
@@ -29,7 +29,7 @@ def _getTempPaths(test):
def _checkBaseSubstitutions(substitutions):
substitutions = [s for (s, _) in substitutions]
- for s in ["%{cxx}", "%{compile_flags}", "%{link_flags}", "%{flags}", "%{exec}"]:
+ for s in ["%{cxx}", "%{compile_flags}", "%{link_flags}", "%{benchmark_flags}", "%{flags}", "%{exec}"]:
assert s in substitutions, "Required substitution {} was not provided".format(s)
def _executeScriptInternal(test, litConfig, commands):
@@ -220,11 +220,15 @@ class CxxStandardLibraryTest(lit.formats.FileBasedTest):
The test format operates by assuming that each test's configuration provides
the following substitutions, which it will reuse in the shell scripts it
constructs:
- %{cxx} - A command that can be used to invoke the compiler
- %{compile_flags} - Flags to use when compiling a test case
- %{link_flags} - Flags to use when linking a test case
- %{flags} - Flags to use either when compiling or linking a test case
- %{exec} - A command to prefix the execution of executables
+ %{cxx} - A command that can be used to invoke the compiler
+ %{compile_flags} - Flags to use when compiling a test case
+ %{link_flags} - Flags to use when linking a test case
+ %{flags} - Flags to use either when compiling or linking a test case
+ %{benchmark_flags} - Flags to use when compiling benchmarks. These flags should provide access to
+ GoogleBenchmark but shouldn't hardcode any optimization level or other settings,
+ since the benchmarks should be run under the same configuration as the rest of
+ the test suite.
+ %{exec} - A command to prefix the execution of executables
Note that when building an executable (as opposed to only compiling a source
file), all three of %{flags}, %{compile_flags} and %{link_flags} will be used
@@ -254,6 +258,7 @@ class CxxStandardLibraryTest(lit.formats.FileBasedTest):
def getTestsForPath(self, testSuite, pathInSuite, litConfig, localConfig):
SUPPORTED_SUFFIXES = [
+ "[.]bench[.]cpp$",
"[.]pass[.]cpp$",
"[.]pass[.]mm$",
"[.]compile[.]pass[.]cpp$",
@@ -331,6 +336,20 @@ def execute(self, test, litConfig):
"%dbg(EXECUTED AS) %{exec} %t.exe",
]
return self._executeShTest(test, litConfig, steps)
+ elif filename.endswith(".bench.cpp"):
+ if "enable-benchmarks=no" in test.config.available_features:
+ return lit.Test.Result(
+ lit.Test.UNSUPPORTED,
+ "Test {} requires support for benchmarks, which isn't supported by this configuration".format(
+ test.getFullName()
+ ),
+ )
+ steps = [
+ "%dbg(COMPILED WITH) %{cxx} %s %{flags} %{compile_flags} %{link_flags} %{benchmark_flags} -o %t.exe",
+ ]
+ if "enable-benchmarks=run" in test.config.available_features:
+ steps += ["%dbg(EXECUTED AS) %{exec} %t.exe"]
+ return self._executeShTest(test, litConfig, steps)
else:
return lit.Test.Result(
lit.Test.UNRESOLVED, "Unknown test suffix for '{}'".format(filename)
diff --git a/libcxx/utils/libcxx/test/googlebenchmark.py b/libcxx/utils/libcxx/test/googlebenchmark.py
deleted file mode 100644
index d260484c837702..00000000000000
--- a/libcxx/utils/libcxx/test/googlebenchmark.py
+++ /dev/null
@@ -1,125 +0,0 @@
-from __future__ import absolute_import
-import os
-import subprocess
-import sys
-
-import lit.Test
-import lit.TestRunner
-import lit.util
-from lit.formats.base import TestFormat
-
-kIsWindows = sys.platform in ["win32", "cygwin"]
-
-
-class GoogleBenchmark(TestFormat):
- def __init__(self, test_sub_dirs, test_suffix, benchmark_args=[]):
- self.benchmark_args = list(benchmark_args)
- self.test_sub_dirs = os.path.normcase(str(test_sub_dirs)).split(";")
-
- # On Windows, assume tests will also end in '.exe'.
- exe_suffix = str(test_suffix)
- if kIsWindows:
- exe_suffix += ".exe"
-
- # Also check for .py files for testing purposes.
- self.test_suffixes = {exe_suffix, test_suffix + ".py"}
-
- def getBenchmarkTests(self, path, litConfig, localConfig):
- """getBenchmarkTests(path) - [name]
-
- Return the tests available in gtest executable.
-
- Args:
- path: String path to a gtest executable
- litConfig: LitConfig instance
- localConfig: TestingConfig instance"""
-
- # TODO: allow splitting tests according to the "benchmark family" so
- # the output for a single family of tests all belongs to the same test
- # target.
- list_test_cmd = [path, "--benchmark_list_tests"]
- try:
- output = subprocess.check_output(list_test_cmd, env=localConfig.environment)
- except subprocess.CalledProcessError as exc:
- litConfig.warning(
- "unable to discover google-benchmarks in %r: %s. Process output: %s"
- % (path, sys.exc_info()[1], exc.output)
- )
- raise StopIteration
-
- nested_tests = []
- for ln in output.splitlines(False): # Don't keep newlines.
- ln = lit.util.to_string(ln)
- if not ln.strip():
- continue
-
- index = 0
- while ln[index * 2 : index * 2 + 2] == " ":
- index += 1
- while len(nested_tests) > index:
- nested_tests.pop()
-
- ln = ln[index * 2 :]
- if ln.endswith("."):
- nested_tests.append(ln)
- elif any([name.startswith("DISABLED_") for name in nested_tests + [ln]]):
- # Gtest will internally skip these tests. No need to launch a
- # child process for it.
- continue
- else:
- yield "".join(nested_tests) + ln
-
- def getTestsInDirectory(self, testSuite, path_in_suite, litConfig, localConfig):
- source_path = testSuite.getSourcePath(path_in_suite)
- for subdir in self.test_sub_dirs:
- dir_path = os.path.join(source_path, subdir)
- if not os.path.isdir(dir_path):
- continue
- for fn in lit.util.listdir_files(dir_path, suffixes=self.test_suffixes):
- # Discover the tests in this executable.
- execpath = os.path.join(source_path, subdir, fn)
- testnames = self.getBenchmarkTests(execpath, litConfig, localConfig)
- for testname in testnames:
- testPath = path_in_suite + (subdir, fn, testname)
- yield lit.Test.Test(
- testSuite, testPath, localConfig, file_path=execpath
- )
-
- def execute(self, test, litConfig):
- testPath, testName = os.path.split(test.getSourcePath())
- while not os.path.exists(testPath):
- # Handle GTest parametrized and typed tests, whose name includes
- # some '/'s.
- testPath, namePrefix = os.path.split(testPath)
- testName = namePrefix + "/" + testName
-
- cmd = [testPath, "--benchmark_filter=%s$" % testName] + self.benchmark_args
-
- if litConfig.noExecute:
- return lit.Test.PASS, ""
-
- try:
- out, err, exitCode = lit.util.executeCommand(
- cmd,
- env=test.config.environment,
- timeout=litConfig.maxIndividualTestTime,
- )
- except lit.util.ExecuteCommandTimeoutException:
- return (
- lit.Test.TIMEOUT,
- "Reached timeout of {} seconds".format(litConfig.maxIndividualTestTime),
- )
-
- if exitCode:
- return lit.Test.FAIL, ("exit code: %d\n" % exitCode) + out + err
-
- passing_test_line = testName
- if passing_test_line not in out:
- msg = "Unable to find %r in google benchmark output:\n\n%s%s" % (
- passing_test_line,
- out,
- err,
- )
- return lit.Test.UNRESOLVED, msg
-
- return lit.Test.PASS, err + out
diff --git a/libcxx/utils/libcxx/test/params.py b/libcxx/utils/libcxx/test/params.py
index 6c2713ac92fdb8..2c5cb169c0a9a3 100644
--- a/libcxx/utils/libcxx/test/params.py
+++ b/libcxx/utils/libcxx/test/params.py
@@ -364,6 +364,16 @@ def getSuitableClangTidy(cfg):
AddFeature("libcpp-has-no-experimental-syncstream"),
],
),
+ # TODO: This can be improved once we use a version of GoogleBenchmark that supports the dry-run mode.
+ # See https://github.com/google/benchmark/issues/1827.
+ Parameter(
+ name="enable_benchmarks",
+ choices=["no", "run", "dry-run"],
+ type=str,
+ default="run",
+ help="Whether to run the benchmarks in the test suite, to only dry-run them or to disable them entirely.",
+ actions=lambda mode: [AddFeature(f"enable-benchmarks={mode}")],
+ ),
Parameter(
name="long_tests",
choices=[True, False],
diff --git a/libcxxabi/test/configs/cmake-bridge.cfg.in b/libcxxabi/test/configs/cmake-bridge.cfg.in
index 1dd6b3367e4387..edf80c8f2a7321 100644
--- a/libcxxabi/test/configs/cmake-bridge.cfg.in
+++ b/libcxxabi/test/configs/cmake-bridge.cfg.in
@@ -32,6 +32,7 @@ config.substitutions.append(('%{include}', '@LIBCXXABI_TESTING_INSTALL_PREFIX@/i
config.substitutions.append(('%{cxx-include}', '@LIBCXXABI_TESTING_INSTALL_PREFIX@/@LIBCXXABI_INSTALL_INCLUDE_DIR@'))
config.substitutions.append(('%{cxx-target-include}', '@LIBCXXABI_TESTING_INSTALL_PREFIX@/include/%{triple}/c++/v1'))
config.substitutions.append(('%{lib}', '@LIBCXXABI_TESTING_INSTALL_PREFIX@/@LIBCXXABI_INSTALL_LIBRARY_DIR@'))
+config.substitutions.append(('%{benchmark_flags}', ''))
if @LIBCXXABI_USE_LLVM_UNWINDER@:
config.substitutions.append(('%{maybe-include-libunwind}', '-I "@LIBCXXABI_LIBUNWIND_INCLUDES_INTERNAL@"'))
diff --git a/libunwind/test/configs/cmake-bridge.cfg.in b/libunwind/test/configs/cmake-bridge.cfg.in
index 20b61e788ab180..b804c210f0bbc0 100644
--- a/libunwind/test/configs/cmake-bridge.cfg.in
+++ b/libunwind/test/configs/cmake-bridge.cfg.in
@@ -32,3 +32,4 @@ if not @LIBUNWIND_ENABLE_THREADS@:
config.substitutions.append(('%{install-prefix}', '@LIBUNWIND_TESTING_INSTALL_PREFIX@'))
config.substitutions.append(('%{include}', '@LIBUNWIND_TESTING_INSTALL_PREFIX@/include'))
config.substitutions.append(('%{lib}', '@LIBUNWIND_TESTING_INSTALL_PREFIX@/@LIBUNWIND_INSTALL_LIBRARY_DIR@'))
+config.substitutions.append(('%{benchmark_flags}', ''))
More information about the libcxx-commits
mailing list