cicd: add cicd for cuda device.

.github/workflows/build_x86_64_cuda.yaml

@@ -0,0 +1,122 @@
+name: xLLM Build x86_64 CUDA
+
+on: 
+  workflow_dispatch:
+  push:
+    branches: [main]
+    paths-ignore:
+      - '.github/**'
+      - 'cibuild/**'
+      - 'cmake/**'
+      - 'docs/**'
+      - 'third_party/**'
+      - 'tools/**'
+      - '*.md'
+      - '*.txt'
+      - '*.yml'
+  pull_request:
+    branches: [main]
+    types: [opened, synchronize, reopened]
+    paths-ignore:
+      - 'cmake/**'
+      - 'docs/**'
+      - 'third_party/**'
+      - 'tools/**'
+      - '*.md'
+      - '*.txt'
+      - '*.yml'
+  pull_request_review:
+    types: [submitted]
+    paths:
+      - '.github/**.yaml'
+      - 'cibuild/**.sh'
+      - 'setup.py'
+      - 'examples/generate.py'
+
+env:
+  JOBNAME: xllm-x86_64-cuda-cibuild-${{ github.run_id }}
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ startsWith(github.ref, 'refs/pull/') }}
+
+jobs:
+  # need to review code first when sensitive files are modified.
+  check-sensitive:
+    runs-on: [self-hosted]
+    outputs:
+      requires_approval: ${{ steps.check_sensitive.outputs.requires_approval }}
+      do_build: ${{ steps.decide.outputs.do_build }}
+    steps:
+      - name: Checkout Code
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0  # Ensure we can compare commits
+
+      - name: Install jq
+        run: yum install -y jq
+
+      - name: Check if sensitive files were changed
+        id: check_sensitive
+        run: |
+          sensitive_files=(
+            ".github/**.yaml"
+            "cibuild/**.sh"
+            "setup.py"
+          )
+          changed_files=$(git diff --name-only ${{ github.event.pull_request.base.sha }} ${{ github.sha }})
+          requires_approval="false"
+          while IFS= read -r changed_file; do
+            [[ -z "$changed_file" ]] && continue
+            for pattern in "${sensitive_files[@]}"; do
+              if [[ "$changed_file" == $pattern ]]; then
+                requires_approval="true"
+                break 2
+              fi
+          done
+          done < <(git diff --name-only "${{ github.event.pull_request.base.sha }}" "${{ github.sha }}")
+          echo "requires_approval=$requires_approval" >> $GITHUB_OUTPUT
+
+      - name: Decide whether to check build
+        id: decide
+        run: |
+          event="${{ github.event_name }}"
+          if [[ "$event" == "workflow_dispatch" || "$event" == "push" ]]; then
+            echo "do_build=true" >> $GITHUB_OUTPUT
+          elif [[ "$event" == "pull_request" ]]; then
+            if  [ "${{ steps.check_sensitive.outputs.requires_approval }}" == "true" ]; then
+              echo "do_build=false" >> $GITHUB_OUTPUT
+            else
+              echo "do_build=true" >> $GITHUB_OUTPUT
+            fi
+          elif [[ "$event" == "pull_request_review" ]]; then
+            # Since pull_request_review now only triggers when sensitive files are modified,
+            # we only need to check if the review is approved
+            if [[ "${{ github.event.review.state }}" == "approved" ]]; then
+              echo "do_build=true" >> $GITHUB_OUTPUT
+            else
+              echo "do_build=false" >> $GITHUB_OUTPUT
+            fi
+          else
+            echo "do_build=false" >> $GITHUB_OUTPUT
+          fi
+
+  build:
+    needs: check-sensitive
+    if: >
+      (github.event_name == 'workflow_dispatch' || github.event_name == 'push') ||
+      needs.check-sensitive.outputs.do_build == 'true'
+    runs-on: [self-hosted]
+    steps:
+    - name: Checkout Code
+      timeout-minutes: 5
+      uses: actions/checkout@v4
+      with:
+        submodules: true
+
+    - name: Build
+      if: ${{ success() }}
+      timeout-minutes: 60
+      run: |
+        chmod +x ./cibuild/build_cuda.sh
+        bash cibuild/build_cuda.sh 'pip install pre-commit -i https://pypi.tuna.tsinghua.edu.cn/simple; python setup.py build --device cuda'

.github/workflows/build_x86_64_mlu.yaml

@@ -119,4 +119,4 @@ jobs:
       timeout-minutes: 60
       run: |
         chmod +x ./cibuild/build_mlu.sh
-        bash cibuild/build_mlu.sh 'pip install pre-commit -i https://pypi.tuna.tsinghua.edu.cn/simple; VCPKG_ROOT=/export/home/mlu_vcpkg_src python setup.py build --device mlu'
+        bash cibuild/build_mlu.sh 'pip install pre-commit -i https://pypi.tuna.tsinghua.edu.cn/simple; python setup.py build --device mlu'

cibuild/build_cuda.sh

@@ -0,0 +1,29 @@
+#!/bin/bash
+set -e
+
+function error() {
+  echo "Require build command, e.g. python setup.py build --device cuda"
+  exit 1
+}
+
+IMAGE="quay.io/jd_xllm/xllm-ai:xllm-dev-cuda-x86"
+
+RUN_OPTS=(
+  --rm
+  -t
+  --privileged
+  --ipc=host
+  --network=host
+  --pid=host
+  --shm-size '128gb'
+  -v /export/home:/export/home
+  -v /export/home/mlu_vcpkg_cache:/root/.cache/vcpkg # cuda and mlu vcpkg cache is same
+  -w /export/home
+)
+
+CMD="$*"
+[[ -z "${CMD}" ]] && error
+
+[[ ! -x $(command -v docker) ]] && echo "ERROR: 'docker' command is missing." && exit 1
+
+docker run "${RUN_OPTS[@]}" "${IMAGE}" bash -c "set -euo pipefail; cd $(pwd); ${CMD}"

cibuild/build_mlu.sh

@@ -2,7 +2,7 @@
 set -e
 
 function error() {
-  echo "Require build command, e.g. python setup.py build"
+  echo "Require build command, e.g. python setup.py build --device mlu"
   exit 1
 }
 

cibuild/build_npu.sh

@@ -2,7 +2,7 @@
 set -e
 
 function error() {
-  echo "Require build command, e.g. python setup.py build"
+  echo "Require build command, e.g. python setup.py build --device a2"
   exit 1
 }