Bad Behaviour: Automating System-level Safety Testing Using Open Source Tools
summary
Claims that an open source toolchain reproducibly automates system-level safety tests: it uses subplot for scenario-driven tests derived from STPA, buildstream for deterministic, sandboxed builds with remote execution and artifact caching, and LAVA to deploy on QEMU or target boards and collect results. It supports fault injection via patched builds, parallel CI runners, and traceability from safety goal to requirement to test to evidence. Codethink demonstrates the approach on ISO 26262 workflows and reports successful deterministic construction and CI-produced certification artifacts.
AI-written summary · about
chapters (11)
- 0:00Intro
- 1:07Testing and software safety
- 2:20Why system-level testing?
- 3:30Challenges when testing for compliance
- 5:39Modeling bad behaviour
- 7:51Speaking a common language
- 10:43Driving automated tests with scenarios
- 12:14Subplot binding file example
- 14:15Ensure reproducibility with deterministic construction
- 16:58Deploying test instances
- 19:01Putting it all together
related talks
30mUsing FOSS as Part of a System Safety Mechanism
40mBOF: Open Source Projects in Safety Critical Applications
05mKeynote: Evolution of Open Source for Safe Usage in Critical Systems
32mDeveloping Open Source for Safety and Security- 05mKeynote: Evolution of Open Source for Safe Usage in Critical Systems
45mSafety Certifying an Open Source Project: The Example of Xen
40mAutomated Full System Testing on Hardware with OpenQA
43mEngineering Trust: Formulating Continuous Compliance for Open Sour...