When Radiation Meets Linux: Analyzing Soft Errors in Linux on COTS SoCs under Proton Irradiation
Authors: Saad Memon, Rafal Graczyk, Tomasz Rajkowski, Jan Swakon, Damian Wrobel, Sebastian Kusyk, Mike Papadakis
Abstract: The increasing use of Linux on commercial off-the-shelf (COTS) system-on-chip (SoC) in spaceborne computing inherits COTS susceptibility to radiation-induced failures like soft errors. Modern SoCs exacerbate this issue as aggressive transistor scaling reduces critical charge thresholds to induce soft errors and increases radiation effects within densely packed transistors, degrading overall reliability. Linux's monolithic architecture amplifies these risks, as tightly coupled kernel subsystems propagate errors to critical components (e.g., memory management), while limited error-correcting code (ECC) offers minimal mitigation. Furthermore, the lack of public soft error data from irradiation tests on COTS SoCs running Linux hinders reliability improvements. This study evaluates proton irradiation effects (20-50 MeV) on Linux across three COTS SoC architectures: Raspberry Pi Zero 2 W (40 nm CMOS, Cortex-A53), NXP i MX 8M Plus (14 nm FinFET, Cortex-A53), and OrangeCrab (40 nm FPGA, RISC-V). Irradiation results show the 14 nm FinFET NXP SoC achieved 2-3x longer Linux uptime without ECC memory versus both 40 nm CMOS counterparts, partially due to FinFET's reduced charge collection. Additionally, this work presents the first cross-architecture analysis of soft error-prone Linux kernel components in modern SoCs to develop targeted mitigations. The findings establish foundational data on Linux's soft error sensitivity in COTS SoCs, guiding mission readiness for space applications.
Explore the paper tree
Click on the tree nodes to be redirected to a given paper and access their summaries and virtual assistant
Look for similar papers (in beta version)
By clicking on the button above, our algorithm will scan all papers in our database to find the closest based on the contents of the full papers and not just on metadata. Please note that it only works for papers that we have generated summaries for and you can rerun it from time to time to get a more accurate result while our database grows.