build: third_party を submodule から FetchContent に移行 (#97 + #98)

.github/workflows/build.yml

@@ -41,14 +41,11 @@ jobs:
 
     steps:
       - uses: actions/checkout@v6
-        with:
-          submodules: recursive
 
       # actions/checkout writes its own safe.directory, but later steps
       # invoke git from a fresh shell (`sh -e {0}`) where that scoped
       # config is not always picked up. Set it once globally so every
-      # `git ...` call in this job — checkout's submodule walk, the
-      # bootstrap-key step, etc. — sees the workspace as safe.
+      # `git ...` call in this job sees the workspace as safe.
       - name: Mark workspace as safe for git
         run: git config --global --add safe.directory "$GITHUB_WORKSPACE"
 
@@ -68,37 +65,6 @@ jobs:
           ccache --max-size=400M
           ccache --zero-stats
 
-      # Cache key invalidates when any submodule SHA or bootstrap script changes.
-      - name: Compute bootstrap cache key
-        id: bootstrap-key
-        run: |
-          git submodule status > .submodule-state
-          echo "hash=$(sha256sum .submodule-state build_tools/bootstrap*.sh | sha256sum | cut -c1-16)" >> "$GITHUB_OUTPUT"
-
-      - name: Restore third_party build cache
-        id: tp-cache
-        uses: actions/cache@v4
-        with:
-          path: |
-            third_party/masstree/lib*.a
-            third_party/masstree/config.h
-            third_party/masstree/config.status
-            third_party/mimalloc/out
-            third_party/googletest/build
-          key: tp-${{ runner.os }}-${{ steps.bootstrap-key.outputs.hash }}
-
-      - name: Build third-party (masstree)
-        if: steps.tp-cache.outputs.cache-hit != 'true'
-        run: ./build_tools/bootstrap.sh
-
-      - name: Build third-party (mimalloc)
-        if: steps.tp-cache.outputs.cache-hit != 'true'
-        run: ./build_tools/bootstrap_mimalloc.sh
-
-      - name: Build third-party (googletest)
-        if: steps.tp-cache.outputs.cache-hit != 'true'
-        run: ./build_tools/bootstrap_googletest.sh
-
       - name: Configure (top-level cmake)
         # Sanitizer adds significant link-time memory/cpu cost and CI does
         # not run binaries, so it adds no value here. Use Release for the

CMakeLists.txt

@@ -8,8 +8,6 @@ list(APPEND CMAKE_MODULE_PATH "${CMAKE_SOURCE_DIR}/cmake")
 # protocol binary via ccbench_add_protocol() (see cmake/CompileOptions.cmake).
 include(CompileOptions)
 
-set(THIRD_PARTY_DIR "${PROJECT_SOURCE_DIR}/third_party/")
-
 option(ENABLE_SANITIZER "enable sanitizer on debug build" ON)
 option(ENABLE_UB_SANITIZER "enable undefined behavior sanitizer on debug build" OFF)
 option(ENABLE_COVERAGE "enable coverage on debug build" OFF)
@@ -32,17 +30,12 @@ find_package(glog REQUIRED)
 find_package(Boost COMPONENTS filesystem)
 
 # ---------------------------------------------------------------------------
-# Imported targets for the bootstrap-built third-party static libraries.
-# Replaces the raw ${THIRD_PARTY_DIR}/.../*.a paths previously copy-pasted
-# into every protocol's target_link_libraries(...).
+# Third-party dependencies (masstree / mimalloc / googletest) are pulled in
+# via FetchContent — see cmake/ThirdParty.cmake. Exposes ccbench::masstree
+# / ccbench::mimalloc / GTest::gtest{,_main}, which replace the static
+# IMPORTED targets and raw third_party/*.a paths used previously.
 # ---------------------------------------------------------------------------
-add_library(ccbench::masstree STATIC IMPORTED GLOBAL)
-set_target_properties(ccbench::masstree PROPERTIES
-  IMPORTED_LOCATION "${THIRD_PARTY_DIR}/masstree/libkohler_masstree_json.a")
-
-add_library(ccbench::mimalloc STATIC IMPORTED GLOBAL)
-set_target_properties(ccbench::mimalloc PROPERTIES
-  IMPORTED_LOCATION "${THIRD_PARTY_DIR}/mimalloc/out/release/libmimalloc.a")
+include(ThirdParty)
 
 # ---------------------------------------------------------------------------
 # common/ as a real shared static library — replaces the per-binary

cc/ss2pl/test/CMakeLists.txt

@@ -28,19 +28,18 @@ foreach(src IN LISTS TEST_SOURCES)
             KEY_SORT=${CCBENCH_KEY_SORT}
     )
 
-    target_include_directories(${test_name}
-            PRIVATE ${PROJECT_SOURCE_DIR}/../third_party/googletest/googletest/include
-    )
-
+    # All three deps now come from FetchContent (cmake/ThirdParty.cmake) —
+    # GTest::gtest{,_main} carry their own INTERFACE_INCLUDE_DIRECTORIES so
+    # we no longer need the manual include path override.
     target_link_libraries(${test_name}
             Boost::filesystem
             gflags::gflags
             glog::glog
-            ${PROJECT_SOURCE_DIR}/../third_party/mimalloc/out/release/libmimalloc.a
-            ${PROJECT_SOURCE_DIR}/../third_party/masstree/libkohler_masstree_json.a
+            ccbench::mimalloc
+            ccbench::masstree
             Threads::Threads
-            ${PROJECT_SOURCE_DIR}/../third_party/googletest/build/lib/libgtest.a
-            ${PROJECT_SOURCE_DIR}/../third_party/googletest/build/lib/libgtest_main.a
+            GTest::gtest
+            GTest::gtest_main
             )
     set_compile_options(${test_name})
     add_test(

cmake/ThirdParty.cmake

@@ -0,0 +1,136 @@
+# cmake/ThirdParty.cmake
+#
+# Centralised FetchContent declarations for ccbench's third-party
+# dependencies. Replaces tracked submodules under third_party/ +
+# the per-library build_tools/bootstrap_*.sh scripts.
+#
+# Targets exposed to the rest of the build:
+#   ccbench::masstree   — static archive built from upstream sources
+#                         (libkohler_masstree_json.a; same 11 objects the
+#                         old build_tools/bootstrap.sh produced)
+#   ccbench::mimalloc   — alias for the upstream mimalloc-static target
+#   GTest::gtest /
+#   GTest::gtest_main   — provided directly by upstream googletest
+#
+# All revisions are pinned to commit SHAs / tags so the build is
+# reproducible and FetchContent never silently follows a moving branch.
+
+include(FetchContent)
+include(ExternalProject)
+
+# --- masstree --------------------------------------------------------------
+#
+# Upstream masstree-beta is not a CMake project (autotools + hand-rolled
+# Makefile). We mirror the old build_tools/bootstrap.sh exactly:
+#   ./bootstrap.sh && ./configure --disable-assertions
+#   make CXXFLAGS='-g -W -Wall -O3 -fPIC'
+#   ar cr libkohler_masstree_json.a <11 objects> && ranlib …
+# done at configure time via ExternalProject_Add, then exposed as a
+# STATIC IMPORTED target so the rest of the CMake graph treats it the
+# same way as before.
+#
+# The fork URL + commit SHA come from the previous .gitmodules /
+# `git submodule status` (see issue #97).
+
+set(CCBENCH_MASSTREE_REPO  "https://github.com/thawk105/masstree-beta.git")
+set(CCBENCH_MASSTREE_TAG   "b3c5d054b66b08374d7a6ff5a0faeaf28b041a38")
+set(CCBENCH_MIMALLOC_REPO  "https://github.com/microsoft/mimalloc.git")
+set(CCBENCH_MIMALLOC_TAG   "v2.3.2")  # commit 02a2f5d, matches old submodule
+set(CCBENCH_GOOGLETEST_REPO "https://github.com/google/googletest.git")
+set(CCBENCH_GOOGLETEST_TAG  "f8d7d77c06936315286eb55f8de22cd23c188571")
+
+FetchContent_Declare(
+  masstree
+  GIT_REPOSITORY "${CCBENCH_MASSTREE_REPO}"
+  GIT_TAG        "${CCBENCH_MASSTREE_TAG}"
+  GIT_SHALLOW    FALSE   # SHA pin needs the full object, not a tag-only fetch
+)
+
+# Populate() leaves us with masstree_SOURCE_DIR / masstree_BINARY_DIR but
+# does NOT add it to the build graph (the upstream tree has no
+# CMakeLists.txt). We then drive its autotools build out-of-tree.
+FetchContent_GetProperties(masstree)
+if(NOT masstree_POPULATED)
+  FetchContent_Populate(masstree)
+endif()
+
+set(_masstree_archive "${masstree_SOURCE_DIR}/libkohler_masstree_json.a")
+set(_masstree_config_h "${masstree_SOURCE_DIR}/config.h")
+
+# Mirror the historical bootstrap.sh object set verbatim — these are the
+# exact 11 translation units the old script packed into the archive.
+set(_masstree_objs
+  json.o string.o straccum.o str.o msgpack.o clp.o
+  kvrandom.o compiler.o memdebug.o kvthread.o misc.o)
+
+add_custom_command(
+  OUTPUT  "${_masstree_archive}" "${_masstree_config_h}"
+  COMMAND ${CMAKE_COMMAND} -E echo "Building masstree (libkohler_masstree_json.a)"
+  COMMAND ./bootstrap.sh
+  COMMAND ./configure --disable-assertions
+  COMMAND make -j CXXFLAGS=-g\ -W\ -Wall\ -O3\ -fPIC
+  COMMAND ar cr libkohler_masstree_json.a ${_masstree_objs}
+  COMMAND ranlib libkohler_masstree_json.a
+  WORKING_DIRECTORY "${masstree_SOURCE_DIR}"
+  COMMENT "masstree: bootstrap + configure + make + ar"
+  VERBATIM
+)
+add_custom_target(masstree_build DEPENDS "${_masstree_archive}" "${_masstree_config_h}")
+
+# Expose masstree as an INTERFACE library (rather than STATIC IMPORTED) so
+# we can use add_dependencies() to wire the upstream-build custom target
+# into every consumer — IMPORTED targets don't propagate build-order deps
+# reliably across all CMake versions we care about.
+add_library(ccbench_masstree INTERFACE)
+target_include_directories(ccbench_masstree INTERFACE "${masstree_SOURCE_DIR}")
+target_link_libraries(ccbench_masstree INTERFACE "${_masstree_archive}")
+add_dependencies(ccbench_masstree masstree_build)
+add_library(ccbench::masstree ALIAS ccbench_masstree)
+
+# --- mimalloc --------------------------------------------------------------
+#
+# mimalloc is a regular CMake project; FetchContent_MakeAvailable drops it
+# straight into the build graph and gives us the upstream
+# `mimalloc-static` target. We alias it so the rest of the tree keeps
+# linking against ccbench::mimalloc unchanged.
+#
+# Note: we keep the same commit (v2.3.2) the old submodule was pinned to —
+# bumping the allocator version is out of scope for issue #98.
+
+set(MI_BUILD_STATIC ON  CACHE BOOL "" FORCE)
+set(MI_BUILD_SHARED OFF CACHE BOOL "" FORCE)
+set(MI_BUILD_OBJECT OFF CACHE BOOL "" FORCE)
+set(MI_BUILD_TESTS  OFF CACHE BOOL "" FORCE)
+set(MI_OVERRIDE     OFF CACHE BOOL "" FORCE)
+
+FetchContent_Declare(
+  mimalloc
+  GIT_REPOSITORY "${CCBENCH_MIMALLOC_REPO}"
+  GIT_TAG        "${CCBENCH_MIMALLOC_TAG}"
+  GIT_SHALLOW    TRUE
+)
+FetchContent_MakeAvailable(mimalloc)
+
+# Upstream exports `mimalloc-static`; expose it under ccbench:: so callers
+# don't have to know which upstream name to use.
+add_library(ccbench::mimalloc ALIAS mimalloc-static)
+
+# --- googletest ------------------------------------------------------------
+#
+# Standard FetchContent + MakeAvailable. The googletest CMake project
+# exports `GTest::gtest` / `GTest::gtest_main` targets that the test
+# subdirectories (currently only cc/ss2pl/test) link against directly.
+
+set(BUILD_GMOCK    OFF CACHE BOOL "" FORCE)
+set(INSTALL_GTEST  OFF CACHE BOOL "" FORCE)
+# Match MSVC's runtime to the rest of the build — no-op on Linux but
+# keeps the project portable if anyone ever builds tests on Windows.
+set(gtest_force_shared_crt ON CACHE BOOL "" FORCE)
+
+FetchContent_Declare(
+  googletest
+  GIT_REPOSITORY "${CCBENCH_GOOGLETEST_REPO}"
+  GIT_TAG        "${CCBENCH_GOOGLETEST_TAG}"
+  GIT_SHALLOW    FALSE  # SHA pin
+)
+FetchContent_MakeAvailable(googletest)

include/masstree_wrapper.hh

@@ -13,19 +13,23 @@
 // フォーマッターを利用すると，辞書順のために下記2行が入れ替わる．
 // しかし，依存関係があるため，config -> compiler の順にしなければ
 // 大量のエラーが出てしまう． そのため，改行を防ぐため空行を空けている
-#include "../third_party/masstree/config.h"
-
-#include "../third_party/masstree/compiler.hh"
-
-#include "../third_party/masstree/kvthread.hh"
-#include "../third_party/masstree/masstree.hh"
-#include "../third_party/masstree/masstree_insert.hh"
-#include "../third_party/masstree/masstree_print.hh"
-#include "../third_party/masstree/masstree_remove.hh"
-#include "../third_party/masstree/masstree_scan.hh"
-#include "../third_party/masstree/masstree_stats.hh"
-#include "../third_party/masstree/masstree_tcursor.hh"
-#include "../third_party/masstree/string.hh"
+//
+// masstree のヘッダは ccbench::masstree (cmake/ThirdParty.cmake) の
+// INTERFACE_INCLUDE_DIRECTORIES から angle-bracket 形式で引き当てる。
+// FetchContent によりソースは build/_deps/masstree-src/ に置かれる。
+#include <config.h>
+
+#include <compiler.hh>
+
+#include <kvthread.hh>
+#include <masstree.hh>
+#include <masstree_insert.hh>
+#include <masstree_print.hh>
+#include <masstree_remove.hh>
+#include <masstree_scan.hh>
+#include <masstree_stats.hh>
+#include <masstree_tcursor.hh>
+#include <string.hh>
 
 #include "atomic_wrapper.hh"
 #include "debug.hh"