μ-Kernels must and can be small

Abstract

For a general acceptance, μ-kernels must be fast and not burden applications. For fulfilling these conditions, cache architectures require μ-kernels to be small. The L4 μ-kernel shows that smallness can be achieved.