UML 2.5.1 Combined Fragments: Parallel and Critical Region Nesting

The UML 2.5.1 specification provides the par operator to model concurrent interactions and the critical region operator to ensure atomicity. While these fragments are documented for managing execution flow, ambiguity arises when these operators are nested within one another.

The goal is to define a consistent semantic behavior for a critical region placed inside a par fragment. Because UML is a modeling language rather than an execution engine, different CASE tool implementations may interpret the priority of atomicity versus concurrency differently, potentially leading to inconsistent architectural interpretations.

• Does a critical region nested within a par fragment block all other concurrent threads in the parallel block, or only those targeting the same resource?
• What is the standard behavior for resolving a deadlock scenario where two critical regions within the same par fragment depend on each other?