(Ab)using GitHub Actions for building & testing kernels
summary
Shows how GitHub Actions runs kernel builds and tests at scale using unlimited free compute, with concrete recipes: workflow YAML, hosted Ubuntu runners (2 cores, 7 GB RAM, 14 GB disk), use of ccache and the Actions cache to cut rebuild time, and a matrix to run multiple defconfigs in parallel (example run: four configs finish in ~17 minutes; single build ~14m27s). Integrates problem matchers and smart sparse diff for actionable sparse annotations, boots kernels in QEMU, pushes results into Patchwork, and even triggers spurious email floods (250,000 emails).
AI-written summary · about
chapters (14)
- 0:00Intro
- 0:14(Ab)using GitHub Actions for building & testing kernels
- 1:35Automate your workflow from idea to production
- 2:41unlimited free compute
- 4:35let's build a kernel we need
- 5:30what's a workflow file?
- 8:55kernel build, pat2
- 11:28enter the matrix
- 15:31ccache
- 15:58problem matchers • Teach GitHub Actions what your output looks like "problematcher"
- 19:40booting in qemu
- 21:45Patchwork
- 24:38potential use cases
- 28:27that's all folks
related talks
45mA Call to (GitHub) Actions!
36mRunning the Runners: How CNCF Powers GitHub Actions at Scale
59mIntroduction to GitHub Actions
37mIntroduction to GitHub Actions: Understanding Key Terms and Building Your First...
32mReusing the BPF CI
45mLights, Camera, GitHub Actions!
26mRISCV patchwork CI
47mLights, Camera, GitHub Actions!