Security research on Operating system security measures, where to start?

Question

I am researching on operating system security, And I don't know where to look for more information on this.

I could find the following sources, but I need more guidance.

Sources:

1. Microkernel security evaluation[2012 Kevin C. Kurtz]
2. Operating System Security and Secure Operating Systems
3. Linux kernel vulnerabilities: State-of-the-art defenses and open problems[2011]
4. Evolution of the Windows Kernel Architecture [Dave Probert, Ph.D 2009]
5. Can we make operating systems reliable and secure? [Tanenbaum, A.S 2006.]
6. Monolithic kernel vs. Microkernel
7. Hybrid vs. monolithic OS kernels: a benchmark comparison [2006]
8. An Evaluation Framework for Kernel-Mode Security Modules
9. Protecting Commodity Operating System Kernels from Vulnerable Device Drivers
10. The Pros and Cons of Unix and Windows Security Policies
11. OPERATING SYSTEM CONCEPTS ABRAHAM SILBERSCHATZ 9ed
12. A Multiple-Criteria Approach to Ranking Computer Operating Systems
13. Choosing Secure Platforms in the Enterprise
14. Linux vs windows total cost of ownership comparison
15. XNU: a security evaluation

website links:
http://pastebin.ca/3160959

I am looking for questions like,

1. Is there any theoretical Operating system architecture that is inherently secure?
2. What have been done in making operating systems more secure?
3. If we are to assess an operating systems security, is there any guideline or standard measures to do so?
4. Is there any theoretical parameter to make an operating system secure?

Is there any parameter for benchmarking OS kernels and rank them?

For example :multics OS has a different design and implementation of stack that makes very hard to do a buffer overflow, or code injection, etc.

Is there any parameter like OWASP Top 10 to rank and benchmark a kernel, or how do I have to gather them?

Answer 1

The issue is that "security" as a standalone notion doesn't exists. You have different threats, and you can have specialized OS designed to address these threats.

A possible threat is that the OS code contains a bug opening the possibility for this code to act in an unexpected way (no matter if this "unexpected" behaviour is exploitable or simply a bug).

This threat is mostly a concern for specific embedded software. It is addressed by micro-kernels based OS. Micro-kernel having less code, this leave less room for bugs but also opens the possibility to formally prove the OS kernel correctness.

Another threat is information leakage from end-user general purpose environments (think about trojan, exploitable software, etc.).

This threat is address by OS compartimentation: it will divide the desktop environment into several containers, restricting the communication between the container themselves and the communication between the containers and the external environment (network, USB, drives, etc.). Example of commercial solutions implementing this are PolyXene and Bromium, a free alternative is Qubes OS.

Otherwise, more general solution can be achieved using an hardened version of an already existing OS. Here again, the question will be "Hardened against what threat?". This will provide an OS modified in a certain way to address a certain threat (Is it a server environment protecting against remote exploits? A multi-user environment protecting against one user accessing other user's data? Etc.).