[SystemSafety] State of the art for "safe Linux"

[Paul Sherwood]

On 2024-08-05 13:01, Robert P Schaefer wrote:
> your "#where to next" sounds like “turtles all the way down”
> 
> why not break the cycle of attempting to square the circle and
> 
> model deterministic (non-software or at least non-complex software) 
> designs for monitoring non-deterministic systems?

Yes, that makes sense, and we are doing that also. But complexity is 
rising significantly even for microcontroller designs, and they're only 
deterministic until the hardware fails.

A colleague pointed out that on a recent project, the so-called "safety 
island" processor featured so many transistors that they had to 
constrain how much of the cpu (and how much memory) they were using, in 
order to remain within the target failure rates.

In any case, it seems to me that we can and should be measuring software 
failure rates.

Thanks for your feedback Robert!

br
Paul