TheRock/rocm-systems/.github/workflows/aqlprofile-continuous_integration.yml

name: AqlProfile Continuous Integration

on:
  schedule:
    - cron: '0 7 * * *'
  workflow_dispatch:
  push:
    branches:
      - develop
    paths:
      - 'projects/aqlprofile/**'
      - '!projects/aqlprofile/*.md'
      - '!projects/aqlprofile/CODEOWNERS'
      - '!projects/aqlprofile/source/docs/**'
      - '.github/workflows/aqlprofile-continuous_integration.yml'
      - '.github/actions/rocprofiler-sdk-util/resolve-ci-config/**'
  pull_request:
    paths:
      - 'projects/aqlprofile/**'
      - '!projects/aqlprofile/*.md'
      - '!projects/aqlprofile/CODEOWNERS'
      - '!projects/aqlprofile/source/docs/**'
      - '.github/workflows/aqlprofile-continuous_integration.yml'
      - '.github/actions/rocprofiler-sdk-util/resolve-ci-config/**'

concurrency:
  group: ${{ github.workflow }}-${{ github.ref }}
  cancel-in-progress: true

env:
  ROCM_PATH: "/opt/rocm"
  PYTHON_VENV_PATH: "aqlprofile"
  PYTHON_VENV_ACTIVATE: "aqlprofile/bin/activate"
  navi3_EXCLUDE_TESTS_REGEX: ""
  vega20_EXCLUDE_TESTS_REGEX: ""
  mi200_EXCLUDE_TESTS_REGEX: ""
  mi300_EXCLUDE_TESTS_REGEX: ""
  mi300a_EXCLUDE_TESTS_REGEX: ""
  mi325_EXCLUDE_TESTS_REGEX: ""
  navi4_EXCLUDE_TESTS_REGEX: ""
  navi3_EXCLUDE_LABEL_REGEX: ""
  vega20_EXCLUDE_LABEL_REGEX: ""
  mi200_EXCLUDE_LABEL_REGEX: ""
  mi300_EXCLUDE_LABEL_REGEX: ""
  mi300a_EXCLUDE_LABEL_REGEX: ""
  mi325_EXCLUDE_LABEL_REGEX: ""
  navi4_EXCLUDE_LABEL_REGEX: ""

jobs:
  ci_config:
    if: ${{ github.event_name != 'pull_request' || !github.event.pull_request.head.repo.fork }}
    name: Load CI config
    runs-on: ubuntu-latest
    permissions:
      contents: read
    outputs:
      gfx94x_runner: ${{ steps.runner.outputs.gfx94x_runner }}
      gfx94x_sandbox_runner: ${{ steps.runner.outputs.gfx94x_sandbox_runner }}
    steps:
      - name: Checkout workflow actions
        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
        with:
          sparse-checkout: .github/actions/rocprofiler-sdk-util/resolve-ci-config
          set-safe-directory: true

      - name: Resolve runner labels
        id: runner
        uses: ./.github/actions/rocprofiler-sdk-util/resolve-ci-config

  core-deb:
    if: ${{ github.event_name != 'pull_request' || !github.event.pull_request.head.repo.fork }}
    needs: ci_config
    name: Core • ${{ matrix.system.gpu }} • ${{ matrix.system.os }}
    # See: https://docs.github.com/en/free-pro-team@latest/actions/learn-github-actions/managing-complex-workflows#using-a-build-matrix
    strategy:
      fail-fast: false
      matrix:
        system:
          # - { gpu: 'navi4', runner: 'linux-gfx120X-gpu-rocm', os: 'ubuntu-22.04', build-type: 'RelWithDebInfo', gpu-target: 'gfx120X' }
          # - { gpu: 'navi3', runner: 'linux-gfx110X-gpu-rocm', os: 'ubuntu-22.04', build-type: 'RelWithDebInfo', gpu-target: 'gfx110X' }
          - { gpu: 'mi325', os: 'ubuntu-22.04', build-type: 'RelWithDebInfo', gpu-target: 'gfx94X' }

    runs-on: ${{ needs.ci_config.outputs.gfx94x_runner }}
    container:
      image: docker.io/rocm/rocprofiler-private:${{ matrix.system.os }}-${{ matrix.system.gpu-target }}-latest
      credentials:
        username: ${{ secrets.ROCPROFILER_AZURE_CI_USER }}
        password: ${{ secrets.ROCPROFILER_AZURE_CI_PASS }}
      env:
        DEBIAN_FRONTEND: noninteractive
      options: --privileged --cap-add=SYS_PTRACE --security-opt seccomp=unconfined

    permissions:
      contents: read
      pull-requests: write

    # define this for containers
    env:
      GIT_DISCOVERY_ACROSS_FILESYSTEM: 1
      GPU_RUNNER: ${{ matrix.system.gpu }}

    steps:
    - name: Git Symlink
      shell: bash
      run: |
        ln -s -f /usr/bin/git /usr/local/bin/git

    - uses: actions/checkout@v5
      with:
        sparse-checkout: projects/aqlprofile
        set-safe-directory: true

    - name: Install requirements
      timeout-minutes: 10
      shell: bash
      run: |
        git config --global --add safe.directory '*'
        apt-get update
        apt-get install -y build-essential cmake g++-11 g++-12 python3-pip libgtest-dev libgmock-dev
        update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-11 10 --slave /usr/bin/g++ g++ /usr/bin/g++-11 --slave /usr/bin/gcov gcov /usr/bin/gcov-11
        update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-12 20 --slave /usr/bin/g++ g++ /usr/bin/g++-12 --slave /usr/bin/gcov gcov /usr/bin/gcov-12

    - name: List Files
      shell: bash
      run: |
        echo "PATH: ${PATH}"
        echo "LD_LIBRARY_PATH: ${LD_LIBRARY_PATH}"
        which-realpath() { echo -e "\n$1 resolves to $(realpath $(which $1))"; echo "$($(which $1) --version &> /dev/stdout | head -n 1)"; }
        for i in python3 git cmake ctest gcc g++ gcov; do which-realpath $i; done
        cat ${{ env.ROCM_PATH }}/.info/version
        ls -la
        pwd

    - name: Configure, Build, and Test
      timeout-minutes: 30
      shell: bash
      run:
        PATH=~/.local/bin:${{ env.ROCM_PATH }}/bin:${PATH}
        LD_LIBRARY_PATH=$(pwd)/build:${{ env.ROCM_PATH }}/lib:$LD_LIBRARY_PATH
        ctest --output-on-failure --verbose -DCTEST_SOURCE_DIRECTORY="$(pwd)"
          -DCTEST_BINARY_DIRECTORY="$(pwd)/build" -DAQLPROFILE_BUILD_NUM_JOBS="16" -DCTEST_SITE="${{ needs.ci_config.outputs.gfx94x_runner }}"
          -DCTEST_BUILD_NAME=PR_${{ github.ref_name }}_${{ github.repository }}-${{ matrix.system.os }}-${{ matrix.system.gpu }}-core
          -DCMAKE_CTEST_ARGUMENTS=""
          -DAQLPROFILE_BUILD_TESTS=ON
          -DAQLPROFILE_EXTRA_CONFIGURE_ARGS=""
          -S $(pwd)/projects/aqlprofile/dashboard.cmake

  core-rpm:
    if: ${{ github.event_name != 'pull_request' || !github.event.pull_request.head.repo.fork }}
    needs: ci_config
    name: Core • ${{ matrix.system.gpu }} • ${{ matrix.system.os }}
    # See: https://docs.github.com/en/free-pro-team@latest/actions/learn-github-actions/managing-complex-workflows#using-a-build-matrix
    strategy:
      fail-fast: false
      matrix:
        system:
          - { gpu: 'mi325', os: 'rhel-8.8', build-type: 'RelWithDebInfo', gpu-target: 'gfx94X' }
          - { gpu: 'mi325', os: 'rhel-9.5', build-type: 'RelWithDebInfo', gpu-target: 'gfx94X' }
          - { gpu: 'mi325', os: 'sles-15.6', build-type: 'RelWithDebInfo', gpu-target: 'gfx94X' }

    runs-on: ${{ needs.ci_config.outputs.gfx94x_runner }}
    container:
      image: docker.io/rocm/rocprofiler-private:${{ matrix.system.os }}-${{ matrix.system.gpu-target }}-latest
      credentials:
        username: ${{ secrets.ROCPROFILER_AZURE_CI_USER }}
        password: ${{ secrets.ROCPROFILER_AZURE_CI_PASS }}
      options: --privileged --cap-add=SYS_PTRACE --security-opt seccomp=unconfined

    permissions:
      contents: read
      pull-requests: write

    # define this for containers
    env:
      GIT_DISCOVERY_ACROSS_FILESYSTEM: 1
      GPU_RUNNER: ${{ matrix.system.gpu }}

    steps:
    - uses: actions/checkout@v5
      with:
        sparse-checkout: projects/aqlprofile
        set-safe-directory: true

    - name: Install requirements
      timeout-minutes: 10
      shell: bash
      run: |
        git config --global --add safe.directory '*'
        python3 -m venv ${{ env.PYTHON_VENV_PATH }}
        source ${{ env.PYTHON_VENV_ACTIVATE }}
        export PATH=/opt/rh/gcc-toolset-11/root/usr/bin:$PATH
        python3 -m pip install --upgrade pip
        python3 -m pip install cmake

    - name: List Files
      shell: bash
      run: |
        source ${{ env.PYTHON_VENV_ACTIVATE }}
        echo "PATH: ${PATH}"
        echo "LD_LIBRARY_PATH: ${LD_LIBRARY_PATH}"
        which-realpath() { echo -e "\n$1 resolves to $(realpath $(which $1))"; echo "$($(which $1) --version &> /dev/stdout | head -n 1)"; }
        for i in python3 git cmake ctest gcc g++ gcov; do which-realpath $i; done
        cat ${{ env.ROCM_PATH }}/.info/version
        ls -la
        pwd

    - name: Configure, Build, and Test
      timeout-minutes: 30
      shell: bash
      run:
        PATH=~/.local/bin:${{ env.ROCM_PATH }}/bin:${PATH}
        LD_LIBRARY_PATH=$(pwd)/build:${{ env.ROCM_PATH }}/lib:$LD_LIBRARY_PATH
        ctest --output-on-failure --verbose -DCTEST_SOURCE_DIRECTORY="$(pwd)"
          -DCTEST_BINARY_DIRECTORY="$(pwd)/build" -DAQLPROFILE_BUILD_NUM_JOBS="16"
          -DCTEST_SITE=${{ needs.ci_config.outputs.gfx94x_runner }}
          -DCTEST_BUILD_NAME=PR_${{ github.ref_name }}_${{ github.repository }}-${{ matrix.system.os }}-${{ matrix.system.gpu }}-core
          -DCMAKE_CTEST_ARGUMENTS=""
          -DAQLPROFILE_BUILD_TESTS=ON
          -DAQLPROFILE_EXTRA_CONFIGURE_ARGS=""
          -S $(pwd)/projects/aqlprofile/dashboard.cmake