Background: I'm working on porting an Ada project from Vxworks to a Linux platform. The project heavily relies on usage of Ada's dynamic task priorities.

I've done a couple experiments which have left me confused. Ada task priorities seem to have no effect on the underlying Linux priority/niceness. Two identical tasks with different priorities take equally long to complete, even when pinned to one CPU core.

Question: What are the semantics of Ada tasks on Linux? Why am I seeing no effect from setting task priorities?