ci.yml

# This CI workflow is triggered by:
#   - push to main branch
#   - pull request
#   - workflow dispatch
#
# For pull requests, we run default builds and tests for:
#   - Linux: gfx94X gfx110X
#   - Windows: gfx110X
# If you want to trigger jobs for additional targets, please add a defined label (ex: gfx120X-linux) to the pull request
#
# For push to main branch, all AMD families will built and tested from `amdgpu_family_matrix.py`.
#
# Note: If a test machine is not available for a specific AMD GPU family in `amdgpu_family_matrix.py`, tests will be skipped.

name: CI

on:
  push:
    branches:
      - main
      - release/therock-*
  workflow_dispatch:
    inputs:
      linux_amdgpu_families:
        type: string
        description: "Insert comma-separated list of Linux GPU families to build and test. ex: gfx94X, gfx1201X"
        default: ""
      linux_test_labels:
        type: string
        description: "If enabled, reduce test set on Linux to the list of labels prefixed with 'test:'. ex: test:rocprim, test:hipcub"
        default: ""
      linux_use_prebuilt_artifacts:
        type: boolean
        description: "If enabled, the CI will pull Linux artifacts using artifact_run_id and only run tests"
      windows_amdgpu_families:
        type: string
        description: "Insert comma-separated list of Windows GPU families to build and test. ex: gfx94X, gfx1201X"
        default: ""
      windows_test_labels:
        type: string
        description: "If enabled, reduce test set on Windows to the list of labels prefixed with 'test:' ex: test:rocprim, test:hipcub"
        default: ""
      windows_use_prebuilt_artifacts:
        type: boolean
        description: "If enabled, the CI will pull Windows artifacts using artifact_run_id and only run tests"
      additional_label_options:
        type: string
        description: "If enabled, these additional labels options will be applied. ex: test_runner:oem"
        default: ""
      artifact_run_id:
        type: string
        description: "If provided, the tests will run on this artifact ID"
        default: ""
  pull_request:
    types:
      - labeled
      - opened
      - synchronize

permissions:
  contents: read

concurrency:
  # A PR number if a pull request and otherwise the commit hash. This cancels
  # queued and in-progress runs for the same PR (presubmit) or commit
  # (postsubmit). The workflow name is prepended to avoid conflicts between
  # different workflows.
  group: ${{ github.workflow }}-${{ github.event.number || github.sha }}
  cancel-in-progress: true

jobs:
  setup:
    uses: ./.github/workflows/setup.yml
    with:
      build_variant: "release"

  linux_build_and_test:
    name: Linux::${{ matrix.variant.family }}::${{ matrix.variant.build_variant_label }}
    needs: setup
    if: >-
      ${{
        needs.setup.outputs.linux_variants != '[]' &&
        needs.setup.outputs.enable_build_jobs == 'true'
      }}
    strategy:
      fail-fast: false
      matrix:
        variant: ${{ fromJSON(needs.setup.outputs.linux_variants) }}
    uses: ./.github/workflows/ci_linux.yml
    secrets: inherit
    with:
      amdgpu_families: ${{ matrix.variant.family }}
      artifact_group: ${{ matrix.variant.artifact_group }}
      test_runs_on: ${{ matrix.variant.test-runs-on }}
      build_variant_label: ${{ matrix.variant.build_variant_label }}
      build_variant_suffix: ${{ matrix.variant.build_variant_suffix }}
      build_variant_cmake_preset: ${{ matrix.variant.build_variant_cmake_preset }}
      test_labels: ${{ needs.setup.outputs.linux_test_labels }}
      artifact_run_id: ${{ inputs.artifact_run_id }}
      expect_failure: ${{ matrix.variant.expect_failure == true }}
      use_prebuilt_artifacts: ${{ inputs.linux_use_prebuilt_artifacts == true && 'true' || 'false' }}
      rocm_package_version: ${{ needs.setup.outputs.rocm_package_version }}
      test_type: ${{ needs.setup.outputs.test_type }}
      sanity_check_only_for_family: ${{ matrix.variant.sanity_check_only_for_family == true }}
    permissions:
      contents: read
      id-token: write

  windows_build_and_test:
    name: Windows::${{ matrix.variant.family }}::${{ matrix.variant.build_variant_label }}
    needs: setup
    if: >-
      ${{
        needs.setup.outputs.windows_variants != '[]' &&
        needs.setup.outputs.enable_build_jobs == 'true'
      }}
    strategy:
      fail-fast: false
      matrix:
        variant: ${{ fromJSON(needs.setup.outputs.windows_variants) }}
    uses: ./.github/workflows/ci_windows.yml
    with:
      amdgpu_families: ${{ matrix.variant.family }}
      artifact_group: ${{ matrix.variant.artifact_group }}
      test_runs_on: ${{ matrix.variant.test-runs-on }}
      build_variant_label: ${{ matrix.variant.build_variant_label }}
      build_variant_suffix: ${{ matrix.variant.build_variant_suffix }}
      build_variant_cmake_preset: ${{ matrix.variant.build_variant_cmake_preset }}
      test_labels: ${{ needs.setup.outputs.windows_test_labels }}
      artifact_run_id: ${{ inputs.artifact_run_id }}
      expect_failure: ${{ matrix.variant.expect_failure == true }}
      use_prebuilt_artifacts: ${{ inputs.windows_use_prebuilt_artifacts == true && 'true' || 'false' }}
      rocm_package_version: ${{ needs.setup.outputs.rocm_package_version }}
      test_type: ${{ needs.setup.outputs.test_type }}
      sanity_check_only_for_family: ${{ matrix.variant.sanity_check_only_for_family == true }}
    permissions:
      contents: read
      id-token: write

  # build_python_packages:
  #   name: Build Python Packages
  #   uses: ./.github/workflows/build_python_packages.yml

  ci_summary:
    name: CI Summary
    if: always()
    needs:
      - setup
      - linux_build_and_test
      - windows_build_and_test
    runs-on: ubuntu-24.04
    steps:
      - name: Output failed jobs
        run: |
          echo '${{ toJson(needs) }}'

          # Build a list of failed jobs, but ignore those marked continue-on-error
          FAILED_JOBS="$(echo '${{ toJson(needs) }}' \
            | jq --raw-output '
              to_entries
              | map(select(
                  (.value.result != "success" and .value.result != "skipped")
                  and (.value.outputs.continue_on_error | not)
                ))
              | map(.key)
              | join(",")
            ' \
          )"

          if [[ -n "${FAILED_JOBS}" ]]; then
            echo "The following jobs failed: ${FAILED_JOBS}"
            exit 1
          else
            echo "All required jobs succeeded (continue-on-error jobs ignored)."
          fi