Accelerating Low-End Edge Computing with Cross-Kernel Functionality Abstraction