Decoding Safe(ty) Linux Architectural Approaches for Critical Systems
Talkat Open Source Summit Japan + AI_dev: Open Source GenAI & ML Summit + Automotive Linux Summit 2025 · 2025-12-08..12-10·
summary
Clear distinction between safety Linux vs safe Linux: either put the safety argument into the kernel or keep Linux quality‑managed and handle safety at system level. Three concrete architectures are compared: three architectures: kernel, supervised, hypervisor, with examples like instrument‑cluster display content checkers, watchdogs, hardware safety IP and ISO 26262 part 8/12 routes for pre‑existing software. Proven‑in‑use gets called out as misleading — proven‑in‑use requires identical config — and full kernel certification is noted as technically possible but economically impractical.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 3:34Eliza project overview
- 8:53Safety vs safe Linux
- 12:25Architectural approaches
- 18:50Hypervisor tradeoffs
- 25:13Trustworthiness in safety
- 30:53Safety tradeoffs and certification
- 34:25Q&A
related talks
40mBuilding Safe Systems with Linux
42mEnabling Linux in Safety-Critical Applications
58mSafety-critical systems from the inside
44mPanel Discussion: Enabling Linux in Safety Applications
52mCollaborate on Linux for Use in Safety-Critical Systems?
50mWhen Human Life Depends on Software - Introduction to Safety-Critical Systems
09mAspects of Dependable Linux Systems
40mNVIDIA's Approach for Achieving ASIL B Qualified Linux