From: Tarik Nesh-Nash (tarikn_at_microsoft.com)
Date: Mon Jan 12 2004 - 13:47:40 PST
I would recommend reading "The nucleus of a multiprogramming system" By
Brinch-Hansen as an introduction to this paper. Brinch makes a clear
distinction between the operating system and the Kernel. The kernel in
a way is considered part of the hardware. Also, he focused on the
difference between policy and mechanism for a successful Kernel. The
HYDRA document seems an extension and an application of Brinch idea of a
Kernel. Unlike the "Supervisor" described in Dennis paper, the HYDRA
kernel is built with the goal to be as a framework for different types
of operating systems. The HYDRA is supposed to be "flexible, efficient
and reliable" to support any any number of systems (the only limitation
will be the hardware).
The design principles described were mainly a collection of previous
ideas and research, it is interesting to see how the Dijkstra's layering
idea is inflexible for this project. I should note here that the papers
seem to have similar jargon, it may be hard to assimilate first but
things get better with reading related topics.
The paper seems to focus on protection. Using the same separation
method between policy and mechanism, the author aims to build a system
that provides support and enable different security measures implemented
at the OS level. An execution level description is provided to
distinguish between procedure, LNS and process. The notions of
procedure and process were not new, they were discussed in earlier
papers. However, the notion of capability seems innovative as a
security measure at the Kernel level.
A detailed example is provided at the end of the paper, it uses most of
the technical details in a simple understandable way. It is worth
reading the document a second time after reading the example, so the
reader can gets a deeper understanding of these details.
This archive was generated by hypermail 2.1.6 : Mon Jan 12 2004 - 13:47:50 PST