feat: add non-container job

.github/workflows/_build_linux_self_host.yml

@@ -33,9 +33,15 @@ on:
         required: false
         type: boolean
         default: true
+      use-container:
+        description: "Whether to use container"
+        required: false
+        type: boolean
+        default: true
 
 jobs:
   build_wheels_self_hosted:
+    if: ${{ inputs.use-container }}
     name: Build wheels and Upload (Linux x86_64, self-hosted runner)
     runs-on: ${{ inputs.runner }}
     container:
@@ -141,3 +147,122 @@ jobs:
         if: always()
         run: |
           rm -rf /opt/hostedtoolcache/Python
+
+  build_wheels_self_hosted_no_container:
+    if: ${{ !inputs.use-container }}
+    name: Build wheels and Upload (Linux x86_64, self-hosted runner)
+    runs-on: ${{ inputs.runner }}
+    defaults:
+      run:
+        shell: bash
+    env:
+      DEBIAN_FRONTEND: noninteractive
+      TERM: xterm-256color
+    timeout-minutes: 2160
+    steps:
+      - name: Check environment
+        shell: bash
+        run: |
+          cat /etc/os-release || true
+          echo "-----------------------------"
+          cat /etc/lsb-release || true
+          echo "-----------------------------"
+          lscpu || true
+          echo "-----------------------------"
+          df -h || true
+          echo "-----------------------------"
+          free -h || true
+
+      - name: Install tools
+        shell: bash
+        run: |
+          apt-get update && apt-get install -y --no-install-recommends \
+            curl \
+            ca-certificates \
+            sudo \
+            software-properties-common \
+            wget \
+            unzip \
+            zip \
+            git \
+            build-essential \
+            gcc \
+            g++ \
+            clang \
+            ninja-build \
+            keyboard-configuration \
+            time
+
+      - name: Install gh
+        shell: bash
+        run: |
+          sudo mkdir -p -m 755 /etc/apt/keyrings
+          out=$(mktemp)
+          wget -nv -O$out https://cli.github.com/packages/githubcli-archive-keyring.gpg
+          cat $out | sudo tee /etc/apt/keyrings/githubcli-archive-keyring.gpg > /dev/null
+          sudo chmod go+r /etc/apt/keyrings/githubcli-archive-keyring.gpg
+          echo "deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/githubcli-archive-keyring.gpg] https://cli.github.com/packages stable main" | sudo tee /etc/apt/sources.list.d/github-cli.list > /dev/null
+          sudo apt update
+          sudo apt install gh -y
+
+      - uses: actions/checkout@v4
+
+      - name: Configure Git safe directory
+        shell: bash
+        run: |
+          git config --global --add safe.directory $(pwd)
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: ${{ inputs.python-version }}
+
+      - uses: mjun0812/setup-cuda@v1
+        with:
+          version: ${{ inputs.cuda-version }}
+
+      - name: Install build dependencies
+        shell: bash
+        run: |
+          sudo apt install -y ninja-build clang
+          pip install -U pip setuptools==75.8.0 wheel packaging psutil
+
+      - name: Build wheels
+        timeout-minutes: 2160
+        id: build_wheels
+        shell: bash
+        run: |
+          chmod +x build_linux.sh
+          ./build_linux.sh ${{ inputs.flash-attn-version }} ${{ inputs.python-version }} ${{ inputs.torch-version }} ${{ inputs.cuda-version }}
+          wheel_name=$(basename $(ls flash-attention/dist/*.whl | head -n 1))
+          echo "WHEEL_NAME=$wheel_name" >> $GITHUB_OUTPUT
+
+      - name: Install Test
+        shell: bash
+        run: |
+          pip uninstall -y flash-attn > /dev/null 2>&1
+          pip install --no-cache-dir flash-attention/dist/${{ steps.build_wheels.outputs.WHEEL_NAME }}
+          python -c "import flash_attn; print(flash_attn.__version__)"
+
+      - name: Upload Release Asset
+        if: ${{ inputs.is-upload }}
+        shell: bash
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        run: |
+          tag_name=${{ github.ref_name }}
+          wheel_path="flash-attention/dist/${{ steps.build_wheels.outputs.WHEEL_NAME }}"
+
+          # Check if the file exists
+          if [ ! -f "$wheel_path" ]; then
+            echo "Error: Wheel file not found at $wheel_path"
+            exit 1
+          fi
+
+          # Upload the release asset using GitHub CLI
+          gh release upload "$tag_name" "$wheel_path" --clobber
+
+      - name: Clean up
+        shell: bash
+        if: always()
+        run: |
+          rm -rf /opt/hostedtoolcache/Python

.github/workflows/test-self-hosted-linux.yml

@@ -18,7 +18,8 @@ jobs:
         torch-version: ["2.9.1"]
         # https://developer.nvidia.com/cuda-toolkit-archive
         cuda-version: ["13.0.1"]
-        runner: ["self-hosted", "openci-runner-beta"]
+        runner: ["openci-runner-beta"]
+        use-container: [false]
     with:
       flash-attn-version: ${{ matrix.flash-attn-version }}
       python-version: ${{ matrix.python-version }}
@@ -26,3 +27,4 @@ jobs:
       cuda-version: ${{ matrix.cuda-version }}
       is-upload: false
       runner: ${{ matrix.runner}}
+      use-container: ${{ matrix.use-container}}