You are here: Home / Projects / Safety Critical Linux / 
2024-03-29 - 15:20
OSADL Projects

OSADL Project: Safety Critical Linux

Safety Critical Linux - Working Group Proposal by Nicholas Mc Guire

next up previous
Next: Problem Statement Up: Safety Critical Linux Working Previous: Abstract

Introduction

The Distributed and Embedded System Lab, Lanzhou University, was founded in 2004, and has since then been focused on distributed and real-time systems based on open-source components. Not too surprising the topic of safety critical systems begin closely related to distributed real-time has become a main focus of our on-going research work.

In the context of the on-going research projects for Cooperate Technology of Simens AG Germany, basic research and evaluation of GNU/Linux in safety critical systems has been a focus of the group, which currently consists of 16 members, resulting in the identification of critical topics and technologies to enable the use of GNU/Linux for safety critical systems.

  • COTS development integration - Lifecycle issues
  • Selection of COTS components - criteria and procedures
  • Validation of COTS components - tools, procedures and data interpretation
  • Maintenance of COTS components - update/upgrade and patch issues, long term stability and migration issues

All these tasks have proven to be critical for acceptance in safety critical systems - not only due to validation and certification related limitations or shortcomings of OSS based COTS components, but especially because of the fact that safety critical systems in general have a very long product life-time and thus there are constraints especially in the maintenance area that need to be addressed.

A further issue - tightly coupled to the migration towards OSS based systems, is the management of legacy code, and with that standardization issues. Typically these questions are either answered by the vendor of the respective software - one of the common problems found with COTS components based on OSS is though that no such vendor exists as a formal entity - OSADL might consider taking this role in the context of the findings of this working group.

On the technological side, [] and [18] give a good indication of what technological requirements are critical for utilization of COTS in safety critical systems.

  • temporal stability and determinism
  • subsystem/process/thread level isolation
  • independence of components
  • drivers (especially non-mainstream)

Much of the criticism in [17], which was based on an assessment of 2.4.X Linux kernels, has been addressed in the 2.6.X development cycle - a detailed reassessment pending, it does seem that the shortcomings identified in 2.4.X have been fixed in general due to the developments related to real-time and high-availability.


next up previous
Next: Problem Statement Up: Safety Critical Linux Working Previous: Abstract
latex2html 2007-07-15

To top