Towards Sustainable Systems with the Civil Infrastructure Platform by Jan Kiszka, Siemens AG

Overview Jan Kiszka from Siemens AG presents the Civil Infrastructure Platform (CIP), a Linux Foundation collaboration launched in April to address the unique requirements of long-lived industrial embedded systems that control critical infrastructure including transportation, energy distribution, healthcare, and building management. Unlike typical servers or mobile devices, these systems operate for decades with strict functional safety, real-time, security, and regulatory compliance requirements.

Key Challenges Industrial infrastructure systems differ fundamentally from consumer or cloud systems. They require industrial-grade reliability, functional safety (e.g., train brakes must work when commanded), real-time response guarantees, and ultra-long operational lifetimes spanning decades without replacement. However, they remain isolated until now, forcing each company to privately maintain its own kernel versions and components. The conservative update strategies and regulatory compliance costs make maintenance extremely expensive. Growing connectivity through IoT introduces new security requirements, yet systems still cannot tolerate failures that endanger human life or critical services.

Project Vision and Goals CIP aims to create a standardized, long-term maintained embedded Linux platform by pooling resources across member companies. Rather than each company independently maintaining custom kernel branches, CIP establishes shared long-term support extending beyond the 2–5 years typical of standard LTS kernels. The project commits to working upstream with open source communities rather than forking projects, ensuring quality and maintenance burden remain in mainline.

Core Activities **Super Long-Term Stable Kernel**: Ben Hutchings from Codethink has been announced as the super long-term kernel maintainer, bringing expertise from Debian and existing LTS work. The project will extend the lifetime of selected LTS kernel versions to 10+ years through a defined merge window for backporting hardware support and features, followed by strict security and functional fixes. Kernel version selection criteria include broad usage within member products, synchronization with existing LTSI initiatives, and focus on versions supporting upcoming products rather than older already-deployed versions. A specific version has not yet been announced pending steering committee approval.

**Testing Infrastructure**: Using the kernelCI framework, CIP establishes distributed testing on both reference hardware platforms and member-specific systems. This differs from generic CI in requiring real hardware validation, regulatory compliance documentation, and archiving of test results. The project aims for test cycle times of hours rather than days to respond quickly to security issues.

**User-Space Components**: Starting with a minimal bootable system, CIP focuses on commonly used, security-sensitive packages such as busybox, SSL libraries, and OpenSSH. The emphasis is identifying components that will remain stable and relevant over decades while ensuring reproducible builds after extended periods.

Project Structure and Membership Founding members include Siemens, Toshiba, Scott Fink, and Platt Home. The project is funded by member contributions and employs fixed developers supplemented by member company developers. Member benefits include influence over kernel version selection, component focus, and board/SOC support, plus demonstrated compatibility with the CIP base layer. Notably, results are released as open source, so non-members can use the output but cannot influence decisions.

Current Status and Next Steps The project established requirements and common components by mid-2024. Immediate tasks include formalizing the development process, operationalizing testing infrastructure, and releasing the first super long-term stable kernel version. Future phases will expand to additional user-space packages and new middleware components. The presentation emphasizes this domain moves slowly by nature but commits for the long term once established.

Key Differentiator Unlike other open source domains that have fragmented into proprietary forks and branches, CIP explicitly rejects forking and maintains a strict upstream-first policy, maintaining only what respective open source communities accept. This approach reduces replicated effort and enables knowledge sharing with adjacent domains like automotive, which face similar long-lifetime requirements.