[XPU] [CI] Xpu CI Refactor #5932

plusNew001 · 2026-01-07T12:00:00Z

Motivation

💡 If this PR is a Cherry Pick, the PR title needs to follow the format by adding the [Cherry-Pick] label at the very beginning and appending the original PR ID at the end. For example, [Cherry-Pick][CI] Add check trigger and logic(#5191)

💡 如若此PR是Cherry Pick，PR标题需遵循格式，在最开始加上[Cherry-Pick]标签，以及最后面加上原PR ID，例如[Cherry-Pick][CI] Add check trigger and logic(#5191)

Modifications

Usage or Command

Accuracy Tests

Checklist

Add at least a tag in the PR title.
- Tag list: [[FDConfig],[APIServer],[Engine], [Scheduler], [PD Disaggregation], [Executor], [Graph Optimization], [Speculative Decoding], [RL], [Models], [Quantization], [Loader], [OP], [KVCache], [DataProcessor], [BugFix], [Docs], [CI], [Optimization], [Feature], [Benchmark], [Others], [XPU], [HPU], [GCU], [DCU], [Iluvatar], [Metax]]
- You can add new tags based on the PR content, but the semantics must be clear.
Format your code, run pre-commit before commit.
Add unit tests. Please write the reason in this PR if no unit tests.
Provide accuracy results.
If the current PR is submitting to the release branch, make sure the PR has been submitted to the develop branch, then cherry-pick it to the release branch with the [Cherry-Pick] PR tag.

paddle-bot · 2026-01-07T12:00:08Z

Thanks for your contribution!

Copilot

Pull request overview

本PR对XPU CI测试框架进行了重大重构，主要变更包括：

目的: 重构XPU CI流程，将原有的单一工作流拆分为模块化的可复用工作流，并新增大量测试用例。

主要改动:

将CI工作流拆分为clone、build、4卡测试和8卡测试四个独立的可复用工作流
新增pytest测试框架，包含4卡和8卡两套完整的测试用例
添加conftest.py提供通用的测试辅助函数
新增多种场景的测试用例：PD分离、EP并行、MTP模式、VL模型、logprobs等

Reviewed changes

Copilot reviewed 5 out of 18 changed files in this pull request and generated 7 comments.

Show a summary per file

File	Description
.github/workflows/ci_xpu.yml	重构主工作流，改为编排调用子工作流，实现模块化CI流程
.github/workflows/_build_xpu.yml	新增XPU构建工作流，负责编译和上传wheel包
.github/workflows/_xpu_4cards_case_test.yml	新增4卡测试工作流，在XPU-P800-4Cards runner上执行测试
.github/workflows/_xpu_8cards_case_test.yml	新增8卡测试工作流，在XPU-P800-8Cards runner上执行测试
tests/xpu_ci/4cards_cases/conftest.py	新增4卡测试通用配置，提供进程管理、健康检查等辅助函数
tests/xpu_ci/8cards_cases/conftest.py	新增8卡测试通用配置，与4卡版本基本相同
tests/xpu_ci/4cards_cases/test_*.py	新增10个4卡测试用例，覆盖V1模式、W4A8量化、VL模型、MTP、EP并行、PD分离、logprobs等场景
tests/xpu_ci/8cards_cases/test_*.py	新增2个8卡测试用例，测试PD分离的TP4EP4和TP1EP4配置

Copilot · 2026-01-07T12:03:56Z

.github/workflows/_build_xpu.yml

+            '
+      - name: Package Upload
+        id: set_output
+        env:


语法错误：YAML文件中 env: 关键字后缺少值。第166行只有 env: 而没有定义任何环境变量，这会导致YAML解析错误。如果不需要环境变量，应该删除这一行；如果需要，应该添加相应的变量定义。

Suggested change

env:

Copilot · 2026-01-07T12:03:56Z

.github/workflows/_build_xpu.yml

+              bash custom_ops/xpu_ops/download_dependencies.sh develop
+            export CLANG_PATH=$(pwd)/custom_ops/xpu_ops/third_party/xtdk
+            export XVLLM_PATH=$(pwd)/custom_ops/xpu_ops/third_party/xvllm
+            bash build.sh
+            ls ./dist/*.whl
+            '


代码块结束位置缺少右引号。第159行之后的bash脚本应该有对应的闭合引号 '，但是直接跳到了第163行。这会导致shell脚本执行错误。

Copilot · 2026-01-07T12:03:57Z

.github/workflows/ci_xpu.yml

@@ -1,86 +1,43 @@
-name: CI_XPU
-
+name: CI-XPU


PR描述信息不完整。根据项目规范，PR描述至少应该解释为什么要进行这些修改以及解决了什么问题。建议补充以下内容：

重构的背景和动机

主要改进点（例如：将单个工作流拆分为多个可复用的工作流、增加了测试用例等）

对CI流程的影响

测试验证情况

Copilot · 2026-01-07T12:03:57Z

.github/workflows/_build_xpu.yml

+            python ${push_file} fastdeploy*.whl ${target_path}
+            target_path_stripped="${target_path#paddle-github-action/}"
+            WHEEL_PATH=https://paddle-github-action.bj.bcebos.com/${target_path_stripped}/${fd_wheel_name}
+            echo "wheel_path=${WHEEL_PATH}" >> $GITHUB_OUTPUT


缺少环境变量 GITHUB_OUTPUT 的存在性检查。如果在本地或非GitHub Actions环境中运行此脚本，$GITHUB_OUTPUT 可能未定义，导致脚本失败。建议添加环境变量检查或使用更健壮的输出方式。

Suggested change

echo "wheel_path=${WHEEL_PATH}" >> $GITHUB_OUTPUT

if [ -n "${GITHUB_OUTPUT:-}" ]; then

echo "wheel_path=${WHEEL_PATH}" >> "$GITHUB_OUTPUT"

else

# Fallback for non-GitHub Actions environments: print to stdout

echo "GITHUB_OUTPUT is not set. wheel_path=${WHEEL_PATH}"

fi

Copilot · 2026-01-07T12:03:57Z

.github/workflows/ci_xpu.yml

@@ -1,86 +1,43 @@
-name: CI_XPU
-
+name: CI-XPU


PR标题缺少必要的标签格式。根据项目规范，PR标题应该遵循 [标签]描述的格式。建议修改为：[XPU][CI] XPU CI Refactor，明确标注这是XPU和CI相关的改动。

Copilot · 2026-01-07T12:03:58Z

tests/xpu_ci/8cards_cases/conftest.py

+    """
+    停止所有相关进程（最小改动版，避免误杀 pytest）
+    """
+    xpu_id = get_xpu_id()  # noqa: F841


Variable xpu_id is not used.

Suggested change

xpu_id = get_xpu_id() # noqa: F841

Copilot · 2026-01-07T12:03:58Z

tests/xpu_ci/8cards_cases/conftest.py

+            # 真正 kill
+            for pid in safe_pids:
+                subprocess.run(f"kill -9 {pid}", shell=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
+        except Exception:


'except' clause does nothing but pass and there is no explanatory comment.

codecov-commenter · 2026-01-07T13:49:58Z

Codecov Report

✅ All modified and coverable lines are covered by tests.
⚠️ Please upload report for BASE (develop@dc170e3). Learn more about missing BASE report.

Additional details and impacted files

@@            Coverage Diff             @@
##             develop    #5932   +/-   ##
==========================================
  Coverage           ?   67.36%           
==========================================
  Files              ?      347           
  Lines              ?    44612           
  Branches           ?     6869           
==========================================
  Hits               ?    30055           
  Misses             ?    12340           
  Partials           ?     2217

Flag	Coverage Δ
GPU	`67.36% <ø> (?)`

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

Copilot

Pull request overview

Copilot reviewed 6 out of 18 changed files in this pull request and generated 4 comments.

Comments suppressed due to low confidence (1)

tests/xpu_ci/4cards_cases/conftest.py:367

在函数名称后有一个多余的空格。应该是 os.path.dirname(os.path.dirname(os.path.dirname(current_dir))) 而不是 os.path.dirname (os.path.dirname(os.path.dirname(current_dir)))。这会影响代码的一致性和可读性。

Copilot · 2026-01-09T06:51:48Z

tests/xpu_ci/8cards_cases/conftest.py

+    """获取scripts目录路径"""
+    # conftest.py在tests/xpu_ci_pytest/下,scripts在项目根目录下
+    current_dir = os.path.dirname(os.path.abspath(__file__))
+    project_root = os.path.dirname (os.path.dirname(os.path.dirname(current_dir)))


在函数名称后有一个多余的空格。应该是 os.path.dirname(os.path.dirname(os.path.dirname(current_dir))) 而不是 os.path.dirname (os.path.dirname(os.path.dirname(current_dir)))。这会影响代码的一致性和可读性。

Suggested change

project_root = os.path.dirname (os.path.dirname(os.path.dirname(current_dir)))

project_root = os.path.dirname(os.path.dirname(os.path.dirname(current_dir)))

Copilot · 2026-01-09T06:51:48Z

.github/workflows/ci_xpu.yml


 on:
  pull_request:
-    branches:
-      - develop
-      - 'release/*'
-  workflow_dispatch:
+    types: [opened, synchronize]
+    branches: [develop, release/**]
+permissions: read-all

 concurrency:
-  group: ${{ github.event.pull_request.number }}-xpu-ci
+  group: ${{ github.event.pull_request.number }}-${{ github.workflow }}
  cancel-in-progress: true

 jobs:
-  CI_XPU:
-    timeout-minutes: 60
-    runs-on: [self-hosted, XPU-P800-8Card]
-    steps:
-      - name: Print current runner name
-        run: |
-          echo "Current runner name: ${{ runner.name }}"
-      # Because the system version is lower than 2.23, the checkout cannot be used.
-      # - name: Checkout code
-      #   uses: actions/checkout@v4
+  clone:
+    name: FD-Clone-Linux-XPU
+    uses: ./.github/workflows/_clone_linux.yml

-      - name: Code Checkout
-        env:
-          docker_image: ccr-2vdh3abv-pub.cnc.bj.baidubce.com/paddlepaddle/fastdeploy-xpu:ci
-        run: |
-          REPO="https://github.com/${{ github.repository }}.git"
-          FULL_REPO="${{ github.repository }}"
-          REPO_NAME="${FULL_REPO##*/}"
-          BASE_BRANCH="${{ github.base_ref }}"
-          # Clean the repository directory before starting
-          docker run --rm --net=host -v $(pwd):/workspace -w /workspace \
-          -e "REPO_NAME=${REPO_NAME}" \
-          -e "BASE_BRANCH=${BASE_BRANCH}" \
-          ${docker_image} /bin/bash -c '
-            if [ -d ${REPO_NAME} ]; then
-              echo "Directory ${REPO_NAME} exists, removing it..."
-              rm -rf ${REPO_NAME}
-            fi
-          '
-          git config --global user.name "FastDeployCI"
-          git config --global user.email "[email protected]"
-          git clone ${REPO} ${REPO_NAME} -b ${BASE_BRANCH}
-          cd FastDeploy
-          if [ "${{ github.event_name }}" = "pull_request" ]; then
-            git fetch origin pull/${{ github.event.pull_request.number }}/head:pr/${{ github.event.pull_request.number }}
-            git merge pr/${{ github.event.pull_request.number }}
-            git log -n 3 --oneline
-          else
-            git checkout ${{ github.sha }}
-            git log -n 3 --oneline
-          fi
+  xpu_build_test:
+    name: xpu_build_test
+    needs: [clone]
+    uses: ./.github/workflows/_build_xpu.yml
+    with:
+      FASTDEPLOY_ARCHIVE_URL: ${{ needs.clone.outputs.repo_archive_url }}
+      DOCKER_IMAGE: ccr-2vdh3abv-pub.cnc.bj.baidubce.com/paddlepaddle/fastdeploy-xpu:ci

-      - name: Run CI unittest
-        env:
-          docker_image: ccr-2vdh3abv-pub.cnc.bj.baidubce.com/paddlepaddle/fastdeploy-xpu:ci
-        run: |
-          runner_name="${{ runner.name }}"
-          last_char="${runner_name: -1}"
+  xpu_4cards_case_test:
+    name: xpu_4cards_case_test
+    needs: [clone, xpu_build_test]
+    uses: ./.github/workflows/_xpu_4cards_case_test.yml
+    with:
+      FASTDEPLOY_ARCHIVE_URL: ${{ needs.clone.outputs.repo_archive_url }}
+      DOCKER_IMAGE: ccr-2vdh3abv-pub.cnc.bj.baidubce.com/paddlepaddle/fastdeploy-xpu:ci
+      FASTDEPLOY_WHEEL_URL: ${{ needs.xpu_build_test.outputs.wheel_path }}
+      MODEL_PATH: /ssd3/model

-          if [[ "$last_char" == "1" ]]; then
-            xpu_id="4"
-          else
-            xpu_id="0"
-          fi
-          PARENT_DIR=$(dirname "$WORKSPACE")
-          echo "PARENT_DIR:$PARENT_DIR"
-          docker run --rm --net=host --cap-add=SYS_PTRACE --privileged --shm-size=64G  \
-          -v $(pwd):/workspace -w /workspace \
-          -v "/ssd3:/ssd3" \
-          -e "MODEL_PATH=/ssd3/model" \
-          -e "http_proxy=$(git config --global --get http.proxy)" \
-          -e "https_proxy=$(git config --global --get https.proxy)" \
-          -e "no_proxy=bcebos.com,mirrors.tuna.tsinghua.edu.cn,127.0.0.1,localhost" \
-          -e "XPU_ID=${xpu_id}" \
-           ${docker_image} /bin/bash -c "
-          git config --global --add safe.directory /workspace/FastDeploy
-          cd FastDeploy
-          bash scripts/run_xpu_ci_pytest.sh
-          "
+  xpu_8cards_case_test:
+    name: xpu_8cards_case_test
+    needs: [clone, xpu_build_test]
+    uses: ./.github/workflows/_xpu_8cards_case_test.yml
+    with:
+      FASTDEPLOY_ARCHIVE_URL: ${{ needs.clone.outputs.repo_archive_url }}
+      DOCKER_IMAGE: ccr-2vdh3abv-pub.cnc.bj.baidubce.com/paddlepaddle/fastdeploy-xpu:ci
+      FASTDEPLOY_WHEEL_URL: ${{ needs.xpu_build_test.outputs.wheel_path }}
+      MODEL_PATH: /ssd3/model


PR描述不够完整。根据PR模板要求，描述信息至少应该包含：1) 为什么进行这些修改的动机；2) 具体做了哪些修改；3) 如何使用或运行相关命令。当前PR只填写了模板框架，没有实际内容。建议补充：重构XPU CI的原因、新的测试组织结构说明、以及如何运行这些测试用例的具体指令。

Copilot · 2026-01-09T06:51:48Z

.github/workflows/_build_xpu.yml

+            python3 ${push_file} fastdeploy*.whl ${target_path}
+            target_path_stripped="${target_path#paddle-github-action/}"
+            WHEEL_PATH=https://paddle-github-action.bj.bcebos.com/${target_path_stripped}/${fd_wheel_name}
+            echo "wheel_path=${WHEEL_PATH}" >> $GITHUB_OUTPUT


在 shell 命令中使用了未引用的变量，这可能导致命令注入风险。例如在构建命令时直接拼接了变量值。虽然这些变量来自环境变量，但最好使用参数化的方式或确保对这些值进行适当的验证和转义，以避免潜在的安全问题。

Suggested change

echo "wheel_path=${WHEEL_PATH}" >> $GITHUB_OUTPUT

echo "wheel_path=${WHEEL_PATH}" >> "$GITHUB_OUTPUT"

Copilot · 2026-01-09T06:51:49Z

tests/xpu_ci/8cards_cases/conftest.py

+        except Exception:
+            pass


'except' clause does nothing but pass and there is no explanatory comment.

Suggested change

except Exception:

pass

except Exception as exc:

# Ignore errors during process cleanup to avoid breaking test teardown

print(f"[WARN] safe_kill_cmd failed for '{cmd}': {exc}")

Copilot

Pull request overview

Copilot reviewed 6 out of 18 changed files in this pull request and generated 5 comments.

Comments suppressed due to low confidence (1)

tests/xpu_ci/4cards_cases/conftest.py:367

代码中存在多余的空格。在函数名os.path.dirname和左括号之间有一个不必要的空格，应该移除。这会影响代码的规范性。

Copilot · 2026-01-09T07:50:53Z

.github/workflows/_xpu_8cards_case_test.yml

+          ${docker_image} /bin/bash -c '
+          echo "安装lsof工具..."
+          apt install -y lsof
+
+          # 设置XPU_VISIBLE_DEVICES
+          export XPU_VISIBLE_DEVICES="0,1,2,3,4,5,6,7"
+          echo "XPU_VISIBLE_DEVICES=$XPU_VISIBLE_DEVICES"
+
+          # 下载和安装xre
+          echo "下载和安装xre..."
+          mkdir -p /workspace/deps
+          cd /workspace/deps
+          if [ ! -d "xre" ]; then
+              wget -q https://klx-sdk-release-public.su.bcebos.com/xre/kl3-release/5.0.21.21/xre-Linux-x86_64-5.0.21.21.tar.gz
+              tar -zxf xre-Linux-x86_64-5.0.21.21.tar.gz && mv xre-Linux-x86_64-5.0.21.21 xre
+          fi
+          cd -
+          export PATH=/workspace/deps/xre/bin:$PATH
+
+          # 重启XPU卡
+          echo "重启XPU卡..."
+          xpu-smi -r -i $XPU_VISIBLE_DEVICES
+          xpu-smi
+          set -e
+          git config --global --add safe.directory /workspace/FastDeploy
+          cd FastDeploy
+          python -m pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple
+          python -m pip install -r requirements.txt
+          echo "安装PaddlePaddle..."
+          # 针对不同分支和tag使用不同的PaddlePaddle安装包
+          if [[ "${PADDLE_WHL_URL}" != "" ]];then
+            python -m pip install ${PADDLE_WHL_URL}
+          elif [[ "${PADDLEVERSION}" != "" ]];then
+            python -m pip uninstall paddlepaddle-xpu fastdeploy-xpu -y
+            python -m pip install paddlepaddle-xpu==${PADDLEVERSION} -i https://www.paddlepaddle.org.cn/packages/stable/xpu-p800/
+          else
+            python -m pip uninstall paddlepaddle-xpu fastdeploy-xpu -y
+            # python -m pip install --pre paddlepaddle-xpu -i https://www.paddlepaddle.org.cn/packages/nightly/xpu-p800/
+            python -m pip install https://paddle-whl.bj.bcebos.com/nightly/xpu-p800/paddlepaddle-xpu/paddlepaddle_xpu-3.4.0.dev20260107-cp310-cp310-linux_x86_64.whl
+          fi
+          echo "安装上游任务编译的fastdeploy-xpu..."
+          python -m pip install ${FASTDEPLOY_WHEEL_URL}
+          rm -rf fastdeploy
+          python -m pip install ${FASTDEPLOY_WHEEL_URL} --no-deps --target=/workspace/FastDeploy
+          echo "============================安装测试依赖============================"
+          python -m pip install openai -U
+          python -m pip uninstall -y triton
+          python -m pip install triton==3.3.0
+          python -m pip install pytest
+          python -m pip install pytest-timeout
+          echo "============================开始运行pytest测试============================"
+          export PYTHONPATH=/workspace/FastDeploy/
+          python -m pytest -v -s --tb=short tests/xpu_ci/8cards_cases/
+          exit_code=$?
+
+          if [ $exit_code -eq 0 ]; then
+              echo "============================8卡cases测试通过!============================"
+          else
+              echo "============================8卡cases测试失败,请检查日志!============================"
+              exit $exit_code
+          fi
+          '


在工作流文件中，多次重复相同的代码块来处理环境准备、依赖安装等。例如，下载xre、安装PaddlePaddle、安装FastDeploy等步骤在_xpu_4cards_case_test.yml和_xpu_8cards_case_test.yml中几乎完全相同。建议将这些共同的步骤提取为一个单独的可复用的工作流或使用复合操作（composite action），以减少重复并提高可维护性。

Copilot · 2026-01-09T07:50:53Z

.github/workflows/_xpu_8cards_case_test.yml

+          else
+            python -m pip uninstall paddlepaddle-xpu fastdeploy-xpu -y
+            # python -m pip install --pre paddlepaddle-xpu -i https://www.paddlepaddle.org.cn/packages/nightly/xpu-p800/
+            python -m pip install https://paddle-whl.bj.bcebos.com/nightly/xpu-p800/paddlepaddle-xpu/paddlepaddle_xpu-3.4.0.dev20260107-cp310-cp310-linux_x86_64.whl


工作流文件中硬编码了特定日期的PaddlePaddle安装包URL（paddlepaddle_xpu-3.4.0.dev20260107）。这个硬编码的日期版本会使CI依赖于一个特定的nightly build，可能导致以下问题：1) 如果该版本被删除，CI将失败；2) 无法自动获取最新的修复和改进。建议使用更灵活的方式，比如使用官方的nightly channel或将版本号作为可配置参数。

Copilot · 2026-01-09T07:50:53Z

.github/workflows/_build_xpu.yml

+              # python -m pip install --pre paddlepaddle-xpu -i https://www.paddlepaddle.org.cn/packages/nightly/xpu-p800/
+              python -m pip install https://paddle-whl.bj.bcebos.com/nightly/xpu-p800/paddlepaddle-xpu/paddlepaddle_xpu-3.4.0.dev20260107-cp310-cp310-linux_x86_64.whl


工作流文件中硬编码了特定日期的PaddlePaddle安装包URL（paddlepaddle_xpu-3.4.0.dev20260107）。这个硬编码的日期版本会使CI依赖于一个特定的nightly build，可能导致以下问题：1) 如果该版本被删除，CI将失败；2) 无法自动获取最新的修复和改进。建议使用更灵活的方式，比如使用官方的nightly channel或将版本号作为可配置参数。

Suggested change

# python -m pip install --pre paddlepaddle-xpu -i https://www.paddlepaddle.org.cn/packages/nightly/xpu-p800/

python -m pip install https://paddle-whl.bj.bcebos.com/nightly/xpu-p800/paddlepaddle-xpu/paddlepaddle_xpu-3.4.0.dev20260107-cp310-cp310-linux_x86_64.whl

python -m pip install --pre paddlepaddle-xpu -i https://www.paddlepaddle.org.cn/packages/nightly/xpu-p800/

# If a specific nightly wheel is required, please pass it via PADDLE_WHL_URL instead of hardcoding it here

Copilot · 2026-01-09T07:50:54Z

tests/xpu_ci/8cards_cases/conftest.py

+    """获取scripts目录路径"""
+    # conftest.py在tests/xpu_ci_pytest/下,scripts在项目根目录下
+    current_dir = os.path.dirname(os.path.abspath(__file__))
+    project_root = os.path.dirname (os.path.dirname(os.path.dirname(current_dir)))


代码中存在多余的空格。在函数名os.path.dirname和左括号之间有一个不必要的空格，应该移除。这会影响代码的规范性。

Suggested change

project_root = os.path.dirname (os.path.dirname(os.path.dirname(current_dir)))

project_root = os.path.dirname(os.path.dirname(os.path.dirname(current_dir)))

Copilot · 2026-01-09T07:50:54Z

.github/workflows/_xpu_4cards_case_test.yml

+            # python -m pip install --pre paddlepaddle-xpu -i https://www.paddlepaddle.org.cn/packages/nightly/xpu-p800/
+            python -m pip install https://paddle-whl.bj.bcebos.com/nightly/xpu-p800/paddlepaddle-xpu/paddlepaddle_xpu-3.4.0.dev20260107-cp310-cp310-linux_x86_64.whl


工作流文件中硬编码了特定日期的PaddlePaddle安装包URL（paddlepaddle_xpu-3.4.0.dev20260107）。这个硬编码的日期版本会使CI依赖于一个特定的nightly build，可能导致以下问题：1) 如果该版本被删除，CI将失败；2) 无法自动获取最新的修复和改进。建议使用更灵活的方式，比如使用官方的nightly channel或将版本号作为可配置参数。

Suggested change

# python -m pip install --pre paddlepaddle-xpu -i https://www.paddlepaddle.org.cn/packages/nightly/xpu-p800/

python -m pip install https://paddle-whl.bj.bcebos.com/nightly/xpu-p800/paddlepaddle-xpu/paddlepaddle_xpu-3.4.0.dev20260107-cp310-cp310-linux_x86_64.whl

python -m pip install --pre paddlepaddle-xpu -i https://www.paddlepaddle.org.cn/packages/nightly/xpu-p800/

Copilot

Pull request overview

Copilot reviewed 8 out of 18 changed files in this pull request and generated 5 comments.

Comments suppressed due to low confidence (4)

tests/xpu_ci/4cards_cases/conftest.py:367

在conftest.py的第367行，括号前有一个多余的空格：os.path.dirname (os.path.dirname(...)。应该修改为os.path.dirname(os.path.dirname(...)以符合Python代码规范。
tests/xpu_ci/8cards_cases/test_pd_21b_tp1ep4.py:140
缺少空行。在第138行的cleanup_resources()之后、第139行的if not download_and_build_xdeepep():之前缺少空行，这影响了代码的可读性。根据PEP 8规范，在逻辑代码块之间应该有空行分隔。
tests/xpu_ci/8cards_cases/test_pd_21b_tp4ep4.py:140
在第138行和第139行之间缺少空行。第138行末尾有一个空行（第138行），但在第139行的if语句之前应该有空行分隔。这影响了代码的可读性。根据PEP 8规范，在不同的逻辑代码块之间应该有空行。
tests/xpu_ci/4cards_cases/conftest.py:365
注释内容包含错误的路径信息。在第365行的注释中写道"conftest.py在tests/xpu_ci_pytest/下"，但实际上从文件路径来看，conftest.py应该在tests/xpu_ci/4cards_cases/目录下，而不是tests/xpu_ci_pytest/。这个注释会误导代码维护者。建议修改为准确的路径描述。

Copilot · 2026-01-09T11:45:59Z

.github/workflows/ci_xpu.yml


 on:
  pull_request:
-    branches:
-      - develop
-      - 'release/*'
-  workflow_dispatch:
+    types: [opened, synchronize]
+    branches: [develop, release/**]
+permissions: read-all


PR描述不完整。根据项目要求，PR描述至少需要说明为什么进行这些修改以及解决了什么问题。当前PR描述中的Motivation和Modifications部分都是空的。建议补充说明：1) 为什么要重构XPU CI；2) 具体做了哪些改动；3) 这些改动解决了什么问题或带来了什么改进。

Copilot · 2026-01-09T11:45:59Z

tests/xpu_ci/8cards_cases/conftest.py

+                    model_id = data["data"][0].get("id", "unknown")
+                    print(f"\n模型就绪!模型ID: {model_id}, 总耗时 {elapsed} 秒")
+                    return True
+        except (json.JSONDecodeError, Exception) as e:  # noqa: F841


代码中存在未使用的异常变量。在第190行，异常变量e被捕获但未使用，并带有# noqa: F841注释。建议移除未使用的异常变量，将代码改为except (json.JSONDecodeError, Exception):以符合Python最佳实践。

Suggested change

except (json.JSONDecodeError, Exception) as e: # noqa: F841

except (json.JSONDecodeError, Exception):

Copilot · 2026-01-09T11:46:00Z

tests/xpu_ci/8cards_cases/conftest.py

+    os.system("sysctl -w kernel.msgmax=131072")
+    os.system("sysctl -w kernel.msgmnb=33554432")


使用os.system()执行系统命令存在安全风险。在第475-476行，直接使用os.system()执行sysctl命令。建议使用subprocess.run()替代os.system()，因为它提供更好的错误处理和安全性。应该修改为：

subprocess.run(["sysctl", "-w", "kernel.msgmax=131072"], check=False) subprocess.run(["sysctl", "-w", "kernel.msgmnb=33554432"], check=False)

Suggested change

os.system("sysctl -w kernel.msgmax=131072")

os.system("sysctl -w kernel.msgmnb=33554432")

subprocess.run(["sysctl", "-w", "kernel.msgmax=131072"], check=False)

subprocess.run(["sysctl", "-w", "kernel.msgmnb=33554432"], check=False)

Copilot · 2026-01-09T11:46:00Z

tests/xpu_ci/8cards_cases/conftest.py

+    result = subprocess.run("wget -q https://paddle-qa.bj.bcebos.com/xpu_third_party/xDeepEP.tar.gz", shell=True)
+    if result.returncode != 0:
+        print("下载xDeepEP失败")
+        return False
+
+    print("解压xDeepEP...")
+    result = subprocess.run("tar -xzf xDeepEP.tar.gz", shell=True)
+    if result.returncode != 0:
+        print("解压xDeepEP失败")
+        return False
+
+    print("编译xDeepEP...")
+    result = subprocess.run("cd xDeepEP && bash build.sh && cd -", shell=True)


shell命令注入风险。在第340、346、352行使用subprocess.run()执行shell命令时，shell=True参数可能导致命令注入风险。虽然当前代码使用的是硬编码的URL，但建议使用列表形式的参数而不是字符串+shell=True的方式，以提高安全性。例如，第340行可以改为：

result = subprocess.run(["wget", "-q", "https://paddle-qa.bj.bcebos.com/xpu_third_party/xDeepEP.tar.gz"])

Suggested change

result = subprocess.run("wget -q https://paddle-qa.bj.bcebos.com/xpu_third_party/xDeepEP.tar.gz", shell=True)

if result.returncode != 0:

print("下载xDeepEP失败")

return False

print("解压xDeepEP...")

result = subprocess.run("tar -xzf xDeepEP.tar.gz", shell=True)

if result.returncode != 0:

print("解压xDeepEP失败")

return False

print("编译xDeepEP...")

result = subprocess.run("cd xDeepEP && bash build.sh && cd -", shell=True)

result = subprocess.run(

["wget", "-q", "https://paddle-qa.bj.bcebos.com/xpu_third_party/xDeepEP.tar.gz"]

)

if result.returncode != 0:

print("下载xDeepEP失败")

return False

print("解压xDeepEP...")

result = subprocess.run(["tar", "-xzf", "xDeepEP.tar.gz"])

if result.returncode != 0:

print("解压xDeepEP失败")

return False

print("编译xDeepEP...")

result = subprocess.run(["bash", "build.sh"], cwd="xDeepEP")

Copilot · 2026-01-09T11:46:00Z

tests/xpu_ci/8cards_cases/conftest.py

+    """
+    停止所有相关进程（最小改动版，避免误杀 pytest）
+    """
+    xpu_id = get_xpu_id()  # noqa: F841


代码中存在未使用的变量。在第50行，变量xpu_id被获取但带有# noqa: F841注释来抑制"未使用变量"的警告，这表明该变量确实未被使用。如果不需要这个变量，应该删除它；如果需要用于将来的功能，应该添加注释说明其用途。

Suggested change

xpu_id = get_xpu_id() # noqa: F841

plusNew001 added 3 commits January 7, 2026 11:54

refactor xpu ci

5d01fdc

refactor xpu ci

f0ab26a

refactor xpu ci

6e49a77

Copilot AI review requested due to automatic review settings January 7, 2026 12:00

plusNew001 temporarily deployed to Metax_ci January 7, 2026 12:00 — with GitHub Actions Inactive

paddle-bot bot added the contributor External developers label Jan 7, 2026

Copilot started reviewing on behalf of plusNew001 January 7, 2026 12:00 View session

Copilot AI reviewed Jan 7, 2026

View reviewed changes

refactor xpu ci

a74a4fd

plusNew001 temporarily deployed to Metax_ci January 7, 2026 12:09 — with GitHub Actions Inactive

refactor xpu ci

2803b6d

plusNew001 temporarily deployed to Metax_ci January 8, 2026 02:31 — with GitHub Actions Inactive

refactor xpu ci

252c51b

plusNew001 temporarily deployed to Metax_ci January 8, 2026 02:44 — with GitHub Actions Inactive

refactor xpu ci

2e1819c

plusNew001 temporarily deployed to Metax_ci January 8, 2026 02:59 — with GitHub Actions Inactive

refactor xpu ci

2e296cf

plusNew001 temporarily deployed to Metax_ci January 8, 2026 03:18 — with GitHub Actions Inactive

refactor xpu ci

ad4ad49

plusNew001 had a problem deploying to Metax_ci January 8, 2026 06:09 — with GitHub Actions Error

Merge branch 'develop' into xpu-ci-refactor3

7b854f0

plusNew001 had a problem deploying to Metax_ci January 8, 2026 06:10 — with GitHub Actions Failure

Update ci_xpu.yml

9ad38cd

plusNew001 had a problem deploying to Metax_ci January 8, 2026 06:12 — with GitHub Actions Failure

refactor xpu ci

8864436

plusNew001 temporarily deployed to Metax_ci January 8, 2026 06:50 — with GitHub Actions Inactive

refactor xpu ci

dcffe64

plusNew001 temporarily deployed to Metax_ci January 8, 2026 08:25 — with GitHub Actions Inactive

refactor xpu ci

48b4db1

plusNew001 temporarily deployed to Metax_ci January 8, 2026 09:03 — with GitHub Actions Inactive

refactor xpu ci

a4623d5

plusNew001 temporarily deployed to Metax_ci January 8, 2026 11:15 — with GitHub Actions Inactive

refactor xpu ci

33a8609

plusNew001 temporarily deployed to Metax_ci January 8, 2026 13:02 — with GitHub Actions Inactive

refactor xpu ci

91e3c8b

Copilot AI review requested due to automatic review settings January 9, 2026 06:47

plusNew001 temporarily deployed to Metax_ci January 9, 2026 06:47 — with GitHub Actions Inactive

Copilot started reviewing on behalf of plusNew001 January 9, 2026 06:47 View session

Copilot AI reviewed Jan 9, 2026

View reviewed changes

refactor xpu ci

3bc4585

plusNew001 temporarily deployed to Metax_ci January 9, 2026 06:56 — with GitHub Actions Inactive

refactor xpu ci

803f9ca

Copilot AI review requested due to automatic review settings January 9, 2026 07:46

plusNew001 temporarily deployed to Metax_ci January 9, 2026 07:46 — with GitHub Actions Inactive

Copilot started reviewing on behalf of plusNew001 January 9, 2026 07:46 View session

Copilot AI reviewed Jan 9, 2026

View reviewed changes

refactor xpu ci

bd5e605

plusNew001 temporarily deployed to Metax_ci January 9, 2026 08:45 — with GitHub Actions Inactive

refactor xpu ci

f47dea1

Copilot AI review requested due to automatic review settings January 9, 2026 11:41

plusNew001 temporarily deployed to Metax_ci January 9, 2026 11:41 — with GitHub Actions Inactive

Copilot started reviewing on behalf of plusNew001 January 9, 2026 11:41 View session

Copilot AI reviewed Jan 9, 2026

View reviewed changes

-            echo "wheel_path=${WHEEL_PATH}" >> $GITHUB_OUTPUT
+            if [ -n "${GITHUB_OUTPUT:-}" ]; then
+              echo "wheel_path=${WHEEL_PATH}" >> "$GITHUB_OUTPUT"
+            else
+              # Fallback for non-GitHub Actions environments: print to stdout
+              echo "GITHUB_OUTPUT is not set. wheel_path=${WHEEL_PATH}"
+            fi

	project_root = os.path.dirname (os.path.dirname(os.path.dirname(current_dir)))
	project_root = os.path.dirname(os.path.dirname(os.path.dirname(current_dir)))

	echo "wheel_path=${WHEEL_PATH}" >> $GITHUB_OUTPUT
	echo "wheel_path=${WHEEL_PATH}" >> "$GITHUB_OUTPUT"

-        except Exception:
-            pass
+        except Exception as exc:
+            # Ignore errors during process cleanup to avoid breaking test teardown
+            print(f"[WARN] safe_kill_cmd failed for '{cmd}': {exc}")

		# python -m pip install --pre paddlepaddle-xpu -i https://www.paddlepaddle.org.cn/packages/nightly/xpu-p800/
		python -m pip install https://paddle-whl.bj.bcebos.com/nightly/xpu-p800/paddlepaddle-xpu/paddlepaddle_xpu-3.4.0.dev20260107-cp310-cp310-linux_x86_64.whl

	# python -m pip install --pre paddlepaddle-xpu -i https://www.paddlepaddle.org.cn/packages/nightly/xpu-p800/
	python -m pip install https://paddle-whl.bj.bcebos.com/nightly/xpu-p800/paddlepaddle-xpu/paddlepaddle_xpu-3.4.0.dev20260107-cp310-cp310-linux_x86_64.whl
	python -m pip install --pre paddlepaddle-xpu -i https://www.paddlepaddle.org.cn/packages/nightly/xpu-p800/

	except (json.JSONDecodeError, Exception) as e: # noqa: F841
	except (json.JSONDecodeError, Exception):

		os.system("sysctl -w kernel.msgmax=131072")
		os.system("sysctl -w kernel.msgmnb=33554432")

[XPU] [CI] Xpu CI Refactor #5932

Are you sure you want to change the base?

[XPU] [CI] Xpu CI Refactor #5932

Uh oh!

Conversation

plusNew001 commented Jan 7, 2026

Motivation

Modifications

Usage or Command

Accuracy Tests

Checklist

Uh oh!

paddle-bot bot commented Jan 7, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Jan 7, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 7, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 7, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 7, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 7, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 7, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 7, 2026

Choose a reason for hiding this comment

Uh oh!

codecov-commenter commented Jan 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 9, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

codecov-commenter commented Jan 7, 2026 •

edited

Loading