Minutes 11 Jan 2024

Host: Paul Albertella

Participants: Igor Stoppa, Sebastian Hetze, Pete Brink, Luigi Pellecchia

Agenda:

• Ongoing topics
• OSEP progress summary for Working Group Updates [1]
  • Approx. 20 minutes talk on 16th Jan
• Current topics and plans for this year

Discussion:

Ongoing topics

• Igor’s document(s)
  • Ready to publish
  • NVIDIA version in PDF form - RTF form for collaboration on future refinements
  • Description of Linux kernel architecture for ARM64
  • Plan to create collaborative checklist of sources of interference known issues
  • Paul will share links to where to upload docs
• Proven in Use document
  • Goal was to explain what PiU means and why we do not consider it a viable approach for making safety arguments for Linux on the basis of its use in other contexts
  • Sebastian particularly interested in this from the perspective of whether PiU could be used to argue that, if Linux had been established for safety-related use in e.g. automotive, then this could be used to justify its use in e.g. rail
  • Pete: There are establish PiU approaches, but collecting the data to support the argument for a specific version of the software in question is difficult to apply to Linux
  • May be more reasonable to apply PiU for e.g. a Library, but would need to be applied in a component qualification context.
  • Igor: Real problem are in the BSP (board support package) part for integration between kernel and hardware - different drivers, etc
  • Luigi: Also the safety scope within which Linux is used may be larger or just different in the new context, which might make the argument invalid.

Basil

• Requirements management tool
• Plan a future session to discuss this

• Update / future plans

Checklist as an enabler for future discussions

• Identify problems that we need to work on and in what contexts they are relevant
• Opportunity to dig into the ‘gory details’ of why or why not Linux could be viable for a given use
• How can we define a ‘core part’ of the kernel?
• Use the existing definition(s) of this?

ASPICE as a topic

• Are there similar frameworks in other domains?
• Could we discuss this in the WG, and understand how it may be relevant and whether there any particular challenges in applying it in the context of Linux and Open Source

Roles of Linux in a system with respect to safety

• Define model(s) of reference systems
• Might be only as a source of interference
• Possible link to Systems WG

Next week:

• Plan to start discussion about Basil