Hacktoberfest 2026：维护者为十月标记出来的 issue，仍然开放、适合新手。 浏览 Hacktoberfest issue

RISC-V Development Partners: local QEMU boot + kselftest pipeline seeking the upstream integration path

未关闭
  • 哪里不对？
#579 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

还没有人认领这个 Issue。

评估

难度
5/5
预计耗时
一周以上
新手友好度
30/100
Issue 类型
功能
描述清晰度
需要澄清
活跃度
活跃
技术栈
github-actions, linux
领域
ci-cd, operating-systems, testing

调研方向

首先查看链接的 kernelci-riscv 仓库和 docs/findings-pointer-masking.md，然后将所描述的 QEMU 启动、kselftest、config-drift 和 GitHub Actions workflow 与 KernelCI 可用的集成路径进行比较。当 upstream profile 或结果发布路径达成一致、重复方面的疑虑得到解决，并且任何所需的 upstream PR 或实验室注册工作范围都得到明确界定时，该 issue 即完成。

由索引模型根据 Issue 内容生成。

描述

Context

I am an intern at KUBUDS Tech (a RISC-V International member
organization), working on the RISC-V Development Partners SOW:
https://github.com/riscv-admin/dev-partners/issues/49

Repo: https://github.com/xjysiiau/kernelci-riscv

I am working on a localized KernelCI pipeline for RISC-V targets
(x86 host cross-build, QEMU virt -cpu max as the test target, seeded
openKylin image as the boot target, -snapshot so the image is never
modified). Current state:

  • boot test: cross-built kernel (v7.2.0-rc7, riscv defconfig) boots the
    test image to the login prompt; detection is login-prompt/serial based
    with timeout, producing structured result.json;
  • functional tests: cpuinfo (asserts Vector/Hypervisor ISA extensions)
    and an RVV vector_add self-test (vsetvl/vle/vadd/vse, VLEN reported);
  • kselftests: tools/testing/selftests/riscv (hwprobe/vector/sigreturn/mm/abi)
    built on the host and run inside the booted guest (10 binaries:
    9 pass, 1 known XFAIL - see below);
  • config drift detection: 17-option required contract (Vector/virtio/ext4/
    serial/KVM etc.) + full config diff with FAIL/WARN/INFO levels;
  • CI: GitHub Actions - cloud job (drift check + cross-build) plus a
    self-hosted runner (WSL) executing the full closed loop; per-run results
    are archived and a pass-rate trend table is auto-committed to the repo.

Known finding

abi/pointer_masking "constraint" assertions fail on ZPM-capable platforms
(QEMU -cpu max): the test expects PMLEN round-up without
PR_TAGGED_ADDR_ENABLE, while the kernel resets PMLEN to 0 (commit
3033b2b1e3). Tracked as XFAIL; details:
https://github.com/xjysiiau/kernelci-riscv/blob/main/docs/findings-pointer-masking.md
(reporting to linux-riscv as well)

Questions for the KernelCI community

  1. What is the currently recommended path for adding a RISC-V QEMU test
    profile: Maestro config in kci-dev, or first publishing results via
    KCIDB from our own runner?
  2. Is there existing RISC-V QEMU boot/kselftest coverage we should extend
    rather than duplicate?
  3. Is there interest in registering our self-hosted machine as a pull lab
    for these tests?

Status

Phase 1 (pipeline + boot test) done; Phase 2 (Vector functional tests,
kselftests, config drift detection) done except real-hardware
Hypervisor/KVM testing, for which we are looking for lab partners per the
SOW. Aiming for a Phase 3 upstream PR once the configuration format is
settled.

主要语言
Python
星标
14
派生
32
平均合并
15 小时 40 分钟
30 天内合并 PR
5

环境准备

这个项目没有提供开发容器、Dockerfile 或贡献指南，环境需要你自己搭建：先看它的 README，通用步骤见我们的新手贡献指南。

从这里开始

  1. 先读完整个 Issue，再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库，在一个分支上完成修改。
  4. 提交 Pull Request，并在描述里引用这个 Issue 编号。

kernelci/kernelci-project 的其他 Issue

查看 kernelci/kernelci-project 的全部 Issue

相似的 Issue

更多 Python Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。