[CI][C++] Add Windows ARM64 build

- Use GitHub hosted runner `windows-11-arm` to build with MSVC ARM64
- `simd-level` is set `NONE` for now as xsimd does not yet support MSVC ARM64
- `msys2` is installed and used for the timezone data, as it is not guaranteed to exist on windows-11-arm images
- A recent version of `cmake` is installed as support for finding Windows OpenSSL ARM64 builds was only added in 4.1.0 (see Kitware/CMake@bf52219)
- The Boost context implementation is set to `winfib` when building with MSVC ARM64 as it's the only supported implementation for that platform (see boostorg/context#296 (comment) and boostorg/context#315)
- Updated ccache installation script and version to enable a Windows ARM64 native version

Author: jgiannuzzi

.github/workflows/cpp.yml

@@ -279,6 +279,7 @@ jobs:
           ulimit -c unlimited  # must enable within the same shell
           ci/scripts/cpp_test.sh $(pwd) $(pwd)/build
 
+  # This job should be kept up-to-date with the equivalent job in cpp_extra.yml
   windows:
     name: ${{ matrix.title }}
     runs-on: ${{ matrix.os }}
@@ -287,12 +288,11 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        os:
-          - windows-2022
         include:
-          - os: windows-2022
+          - arch: x64
+            os: windows-2022
             simd-level: AVX2
-            title: AMD64 Windows 2022 AVX2
+            title: AMD64 Windows 2022 AVX2 C++
     env:
       ARROW_BOOST_USE_SHARED: OFF
       ARROW_BUILD_BENCHMARKS: ON
@@ -340,10 +340,17 @@ jobs:
       - name: Download Timezone Database
         shell: bash
         run: ci/scripts/download_tz_database.sh
+      - name: Install msys2 (for tzdata for ORC tests)
+        uses: msys2/setup-msys2@v2
+        id: setup-msys2
+      - name: Install cmake
+        shell: bash
+        run: |
+          ci/scripts/install_cmake.sh 4.1.2 /usr
       - name: Install ccache
         shell: bash
         run: |
-          ci/scripts/install_ccache.sh 4.6.3 /usr
+          ci/scripts/install_ccache.sh 4.12.1 /usr
       - name: Setup ccache
         shell: bash
         run: |
@@ -365,14 +372,14 @@ jobs:
       - name: Build
         shell: cmd
         run: |
-          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvarsall.bat" x64
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvarsall.bat" ${{ matrix.arch }}
           bash -c "ci/scripts/cpp_build.sh $(pwd) $(pwd)/build"
       - name: Test
         shell: cmd
         run: |
-          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvarsall.bat" x64
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvarsall.bat" ${{ matrix.arch }}
           # For ORC
-          set TZDIR=C:\msys64\usr\share\zoneinfo
+          set TZDIR=${{ steps.setup-msys2.outputs.msys2-location }}\usr\share\zoneinfo
           bash -c "ci/scripts/cpp_test.sh $(pwd) $(pwd)/build"
 
   windows-mingw:

.github/workflows/cpp_extra.yml

@@ -313,6 +313,113 @@ jobs:
           cd cpp/examples/minimal_build
           ../minimal_build.build/arrow-example
 
+  # This job should be kept up-to-date with the equivalent job in cpp.yml
+  windows:
+    needs: check-labels
+    name: ${{ matrix.title }}
+    runs-on: ${{ matrix.os }}
+    if: >-
+      needs.check-labels.outputs.force == 'true' ||
+      contains(fromJSON(needs.check-labels.outputs.ci-extra-labels || '[]'), 'CI: Extra') ||
+      contains(fromJSON(needs.check-labels.outputs.ci-extra-labels || '[]'), 'CI: Extra: C++')
+    timeout-minutes: 60
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - arch: arm64
+            os: windows-11-arm
+            simd-level: NONE # Currently set to NONE because xsimd does not support MSVC ARM64 yet
+            title: ARM64 Windows 11 C++
+    env:
+      ARROW_BOOST_USE_SHARED: OFF
+      ARROW_BUILD_BENCHMARKS: ON
+      ARROW_BUILD_SHARED: ON
+      ARROW_BUILD_STATIC: OFF
+      ARROW_BUILD_TESTS: ON
+      ARROW_DATASET: ON
+      ARROW_FLIGHT: OFF
+      ARROW_HDFS: ON
+      ARROW_HOME: /usr
+      ARROW_JEMALLOC: OFF
+      ARROW_MIMALLOC: ON
+      ARROW_ORC: ON
+      ARROW_PARQUET: ON
+      ARROW_SIMD_LEVEL: ${{ matrix.simd-level }}
+      ARROW_SUBSTRAIT: ON
+      ARROW_USE_GLOG: OFF
+      ARROW_VERBOSE_THIRDPARTY_BUILD: OFF
+      ARROW_WITH_BROTLI: OFF
+      ARROW_WITH_BZ2: OFF
+      ARROW_WITH_LZ4: OFF
+      ARROW_WITH_OPENTELEMETRY: OFF
+      ARROW_WITH_SNAPPY: ON
+      ARROW_WITH_ZLIB: ON
+      ARROW_WITH_ZSTD: ON
+      BOOST_SOURCE: BUNDLED
+      CMAKE_CXX_STANDARD: "17"
+      CMAKE_GENERATOR: Ninja
+      CMAKE_INSTALL_PREFIX: /usr
+      CMAKE_UNITY_BUILD: ON
+    steps:
+      - name: Disable Crash Dialogs
+        run: |
+          reg add `
+            "HKCU\SOFTWARE\Microsoft\Windows\Windows Error Reporting" `
+            /v DontShowUI `
+            /t REG_DWORD `
+            /d 1 `
+            /f
+      - name: Checkout Arrow
+        uses: actions/checkout@v5
+        with:
+          fetch-depth: 0
+          submodules: recursive
+      - name: Download Timezone Database
+        shell: bash
+        run: ci/scripts/download_tz_database.sh
+      - name: Install msys2 (for tzdata for ORC tests)
+        uses: msys2/setup-msys2@v2
+        id: setup-msys2
+      - name: Install cmake
+        shell: bash
+        run: |
+          ci/scripts/install_cmake.sh 4.1.2 /usr
+      - name: Install ccache
+        shell: bash
+        run: |
+          ci/scripts/install_ccache.sh 4.12.1 /usr
+      - name: Setup ccache
+        shell: bash
+        run: |
+          ci/scripts/ccache_setup.sh
+      - name: ccache info
+        id: ccache-info
+        shell: bash
+        run: |
+          echo "cache-dir=$(ccache --get-config cache_dir)" >> $GITHUB_OUTPUT
+      - name: Cache ccache
+        uses: actions/cache@v4
+        with:
+          path: ${{ steps.ccache-info.outputs.cache-dir }}
+          key: cpp-ccache-windows-${{ env.CACHE_VERSION }}-${{ hashFiles('cpp/**') }}
+          restore-keys: cpp-ccache-windows-${{ env.CACHE_VERSION }}-
+        env:
+          # We can invalidate the current cache by updating this.
+          CACHE_VERSION: "2022-09-13"
+      - name: Build
+        shell: cmd
+        run: |
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvarsall.bat" ${{ matrix.arch }}
+          bash -c "ci/scripts/cpp_build.sh $(pwd) $(pwd)/build"
+      - name: Test
+        shell: cmd
+        run: |
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvarsall.bat" ${{ matrix.arch }}
+          # For ORC
+          set TZDIR=${{ steps.setup-msys2.outputs.msys2-location }}\usr\share\zoneinfo
+          bash -c "ci/scripts/cpp_test.sh $(pwd) $(pwd)/build"
+
   report-extra-cpp:
     needs:
       - docker

ci/scripts/install_ccache.sh

@@ -30,10 +30,19 @@ prefix=$2
 mkdir -p /tmp/ccache
 case $(uname) in
   MINGW64*)
-    url="https://github.com/ccache/ccache/releases/download/v${version}/ccache-${version}-windows-x86_64.zip"
+    declare -A archs
+    archs=([64-bit]="x86_64"
+           [ARM 64-bit Processor]="aarch64")
+    arch=$(powershell -Command "(Get-CimInstance Win32_OperatingSystem).OSArchitecture")
+    if [ -z "${archs[$arch]}" ]; then
+      echo "Unsupported architecture on Windows: ${arch}"
+      exit 0
+    fi
+    arch=${archs[$arch]}
+    url="https://github.com/ccache/ccache/releases/download/v${version}/ccache-${version}-windows-${arch}.zip"
     pushd /tmp/ccache
     curl --fail --location --remote-name "${url}"
-    unzip -j "ccache-${version}-windows-x86_64.zip"
+    unzip -j "ccache-${version}-windows-${arch}.zip"
     chmod +x ccache.exe
     mv ccache.exe "${prefix}/bin/"
     popd

ci/scripts/install_cmake.sh

@@ -24,17 +24,13 @@ if [ "$#" -ne 2 ]; then
   exit 1
 fi
 
-declare -A archs
-archs=([x86_64]=x86_64
-       [arm64]=aarch64
-       [aarch64]=aarch64)
+declare -A linux_archs
+linux_archs=([x86_64]="x86_64"
+             [aarch64]="aarch64")
 
-arch=$(uname -m)
-if [ -z "${archs[$arch]}" ]; then
-  echo "Unsupported architecture: ${arch}"
-  exit 0
-fi
-arch=${archs[$arch]}
+declare -A windows_archs
+windows_archs=([64-bit]="x86_64"
+               [ARM 64-bit Processor]="arm64")
 
 version=$1
 prefix=$2
@@ -43,12 +39,25 @@ platform=$(uname)
 case ${platform} in
   Linux)
     platform=linux
+    arch=$(uname -m)
+    if [ -z "${linux_archs[$arch]}" ]; then
+      echo "Unsupported architecture on Linux: ${arch}"
+      exit 0
+    fi
+    arch=${linux_archs[$arch]}
     ;;
   Darwin)
     platform=macos
+    arch=universal
     ;;
   MSYS_NT*|MINGW64_NT*)
     platform=windows
+    arch=$(powershell -Command "(Get-CimInstance Win32_OperatingSystem).OSArchitecture")
+    if [ -z "${windows_archs[$arch]}" ]; then
+      echo "Unsupported architecture on Windows: ${arch}"
+      exit 0
+    fi
+    arch=${windows_archs[$arch]}
     ;;
   *)
     echo "Unsupported platform: ${platform}"
@@ -59,22 +68,20 @@ esac
 mkdir -p "${prefix}"
 url="https://github.com/Kitware/CMake/releases/download/v${version}/cmake-${version}-${platform}-"
 case ${platform} in
-  macos)
-    url+="universal.tar.gz"
-    curl -L "${url}" | tar -xzf - --directory "${prefix}" --strip-components=1
-    ln -s CMake.app/Contents/bin "${prefix}/bin"
-    ;;
   windows)
     url+="${arch}.zip"
     archive_name=$(basename "${url}")
     curl -L -o "${archive_name}" "${url}"
     unzip "${archive_name}"
     base_name=$(basename "${archive_name}" .zip)
-    mv "${base_name}"/* "${prefix}"
+    cp -a "${base_name}"/* "${prefix}"
     rm -rf "${base_name}" "${archive_name}"
     ;;
   *)
     url+="${arch}.tar.gz"
     curl -L "${url}" | tar -xzf - --directory "${prefix}" --strip-components=1
+    ;;&
+  macos)
+    ln -s CMake.app/Contents/bin "${prefix}/bin"
     ;;
 esac

cpp/cmake_modules/ThirdpartyToolchain.cmake

@@ -1118,6 +1118,11 @@ function(build_boost)
     # This is for https://github.com/boostorg/container/issues/305
     string(APPEND CMAKE_C_FLAGS " -Wno-strict-prototypes")
   endif()
+  if(MSVC AND "${CMAKE_SYSTEM_PROCESSOR}" STREQUAL "ARM64")
+    set(BOOST_CONTEXT_IMPLEMENTATION
+        winfib
+        CACHE STRING "" FORCE)
+  endif()
   set(CMAKE_UNITY_BUILD OFF)
 
   fetchcontent_makeavailable(boost)