A complete guide to Linux process scheduling

The subject of this thesis is process scheduling in wide purpose operating systems. For many years kernel hackers all over the world tried to accomplish the seemingly infeasible task of achieving good interaction on desktop systems and low latencies on heavily loaded server machines. Some progress has been made in this area since the rise of free software, but, in opinion of many, it is still far from perfect. Lots of beginner operating system enthusiasts find the existing solutions too complex to understand and, in light of almost complete lack of documentation along with common hostility of active kernel developers towards rookies, impossible to get hands on. Anyone who has the courage to wade into the dragon infested layer that is the scheduler, should be aware of the ideas behind current implementations before making any contributions. That is what this thesis is about showing how things work under the hood and how they developed to be like this. Every decision behind each concept in a kernel of an OS has its history and meaning. Here I will guide you through process scheduling mechanisms in currently stable Linux kernel as an object lesson on the matter. The work starts with an overview of the essentials of process abstraction in Linux, and continues with detailed code-level description of scheduling techniques involved in past and present kernels