Linux Is Not Suitable as the Kernel Foundation for an ARINC 653 Platform

Integrated Modular Avionics (IMA) has become the foundational architecture of modern civil aircraft.To support mixed‑criticality software running concurrently on shared computing resources, the IMA model relies on strong guarantees of time partitioning, space partitioning, and fault containment. ARINC 653 is the de facto standard that specifies these guarantees through a partitioned operating environment, a two‑level scheduling model, and the well‑defined APplication EXecutive (APEX) interface. In recent years, as the low‑altitude economy, general aviation, and eVTOL industries expand, many domestic and international teams have shown growing interest in developing avionics platforms. However, commercial ARINC 653‑compliant operating systems—such as VxWorks 653, PikeOS, and INTEGRITY‑178B—tend to be expensive, export‑restricted, and bound by stringent licensing conditions. This encourages some new entrants to consider open‑source Linux as a low‑cost alternative for building ARINC 653 platforms. From an airworthiness and high‑assurance software engineering standpoint, however, the Linux kernel is architecturally inconsistent with the requirements of ARINC 653, DO‑178C, and DO‑297. The Linux design philosophy—time‑sharing scheduling, dynamic virtual memory, monolithic in‑kernel drivers, extensive subsystem interactions, and rapid evolutionary codebase—directly contradicts the determinism, analyzability, static configuration, and fault‑isolation principles mandated in civil aviation. This paper synthesizes engineering experience from avionics development and certification activities and presents eight core architecture‑based arguments that demonstrate why Linux cannot serve as the kernel foundation of an ARINC 653 platform. The analysis also clarifies the limitations of “Linux‑based ARINC 653 emulation” approaches and proposes viable, fully certifiable alternatives aligned with aviation industry standards.