We present an analytical framework for formulating partition
configuration problems in real-time systems with dependencies,
particularly applicable to modeling systems with
multiple criticality or security levels. Partition configuration
constraints for real-time tasks include affinity and conflict.
We also discuss the application of the framework to arbitrary
partition schedulers, harmonic partition execution,
and round robin partition execution (which is particularly
problematic). Our interest is in minimizing end-to-end latency,
though the computational complexity of the problem
prevents us from finding optimal results. We conclude with
some open problems.