15-410 Approved Readings
Advanced Configuration And Power Interface Specification
Practical File System Design with the Be File System
Daniel Bovet and Marco Cesati.
Understanding the Linux Kernel (2nd Edition or newer)
Note: this book has some tendency to function as
a play-by-play of what happens inside Linux (on an x86) as opposed to
a design book.
You may find the BSD book
or the Love book
Eppinger, Mummert, and Spector, ed.
Camelot and Avalon: a distributed transaction facility
This is the story of Mach the platform as opposed to Mach the
OS: Mach served as the basis for Camelot, a distributed transaction
system based on transactional virtual memory, and Avalon, a high-level
language built on Camelot. The system as developed didn't enter
popular use, but it's an interesting case study.
Maurice J. Bach.
Design of the Unix Operating System
This is a very clear overview of how Unix kernels
used to be constructed (some time ago).
Some people have found this book dry due to extensive
Joshua J. Bloch.
Though this isn't a book about an OS, there are
many related topics covered from a different perspective
(races and synchronization; access control and integrity).
Also, this book is full of things Java programmers need
to know (but often don't know they need to know).
If you are a working Java programmer, you probably want to
read the second edition rather than the first so you can
be up to date. If you are curious about Java, the first
edition is probably fine.
Joseph Boykin, David Kirschen, Alan Langerman, Susan LoVerso.
Programming under Mach,
Addison-Wesley Unix and Open Systems Series;
Addison-Wesley Publishing Company, Inc. 1993.
This is perhaps the most straightforward book on
what Mach provided to programmers. Past students have
found it somewhat dry.
Niels Ferguson and Bruce Schneier.
This is a good nuts-and-bolts book; if you
are interested in the human-factors, managerial,
political, or social implications of computer
security, you might prefer Secrets and Lies
Brian Goetz et al.
Java Concurrency in Practice
If you want to learn a lot more about concurrency,
this is a good chance to cover, in more detail than we
have time for, how to carefully define the problem and
some good tools to address it with.
Linux Kernel Development
We received some complaints about the 2003 edition
containing too much play-by-play material, but the
2005 edition has received favorable student reviews.
You may find that the BSD book
provides more historical and design perspective.
You should also know that people familiar with other
operating systems may dispute some historical descriptions
and claims of architectural novelty.
These both count as "the BSD book":
- Marshall Kirk McKusick, George V. Neville-Neil.
The Design and Implementation of the FreeBSD Operating System,
Pearson Education, 2004.
- Marshall Kirk McKusick, Keith Bostic, Michael J. Karels, John S.
The Design and Implementation of the 4.4 BSD Operating
Addison-Wesley Publishing Company, Inc., 1996.
You may read either edition of "the BSD book",
though the older one will describe a system which
seems oddly and perhaps frustratingly old.
Efficient Memory Programming
Note that some students reported a distracting
number of minor errors in this book.
Jim Mauro, Richard McDougall.
Solaris Internals: Core Kernel Architecture
Mark Russinovich, David Solomon, & Alex Ionescu.
Windows Internals, 5th edition
The fifth edition contains coverage of key topics
such as 64-bit mode and hypervisor support.
Secrets and Lies
Note: his Beyond Fear is probably
too layman-targeted for this class. Meanwhile,
Applied Cryptography is probably too dry.
If you're looking for nuts and bolts, try
Practical Cryptography (above); this is
a more conceptual/high-level book.
Using Speculative Execution to Automatically Hide I/O Latency (CMU-CS-01-172)
To predict which data your application will fetch, clone a copy and
run it in a world where I/O happens instantly but returns fake data.
Replace out-of-date/inaccurate clones with newer ones as necessary.
Performance Measurements of the First RAID Prototype
Note: this is a very dense piece of work,
but it is an excellent
example of what must be done to make real systems go fast. It is
also a good example of how to graduate even if your system doesn't
go as fast as you expected it to.
Richard P. Draves.
Control Transfer in Operating System Kernels
Note: this is the "continuations in the Mach kernel" dissertation.
Gregory S. Hartman,
Attentiveness: Reactivity at Scale
(abstract and 144M PDF)
This work attempts to explain why many applications fail to react in a timely fashion to your input
(the "cancel button problem"),
and lays out a roadmap for implementing attentive systems.
If you're in a hurry, you can probably get by with Chapters 1, 2, and 4.
Disconnected Operation in a Distributed File System
Disconnected operation is a very timely topic.
Synthesis: An Efficient Implementation of Fundamental
Operating System Services
The practical implications of this work are very unclear, but
it's definitely a good vehicle for thinking differently about
what a kernel might be.
Lily B. Mummert,
Exploiting weak connectivity
in a distributed file system
Adaptive mobility from the file system perspective.
Informed Prefetching and Caching (CMU-CS-97-204)
Instead of an application-blind cache eviction policy such as
LRU, use your cache space to hold data your particular application
is likely to want in the future.
Alfred Z. Spector,
Multiprocessing Architectures for Local Computer Networks,
This is old and not trivial to find (I think the E&S Library
has a copy), but it's fascinating...
Michael Wayne Young,
Exporting a User Interface to Memory Management
from a Communication-Oriented Operating System,
If multiple tracks are listed for an area, the suggestion
is to read one track's worth of papers on the expectation that
they have been chosen because they work well together.
- Autonet (N.B. The papers are interesting though the technology is dead)
(related topics: Xen, VMware)
The exokernel approach to extensibility (panel statement)
The operating system kernel as a secure programmable machine
Exterminate all operating system abstractions
Exokernel: an operating system architecture for
application-level resource management
Application performance and flexibility on exokernel systems
Fast and flexible Application-Level Networking on Exokernel Systems
- EROS - Yes, you can read about
but you need
to propose a particular list of papers, see below.
- Plan 9
- File system track
Plan 9 From Bell Labs
The Plan 9 File Server
Venti: A new approach to archival storage
- Network track
Plan 9 From Bell Labs
The Organization of Networks in Plan 9
The IL protocol
- Naming track
Plan 9 From Bell Labs
The Use of Name Spaces in Plan 9
Lexical File Names in Plan 9, or, Getting Dot-Dot Right
- Or suggest your own track of two to three other papers...
(related topics: Exokernel, VMware)
Xen and the Art of Virtualization
Safe Hardware Access with the Xen Virtual Machine Monitor
Live Migration of Virtual Machines
(related topics: Exokernel, Xen) - ok to read a subset of two or three
Fast Transparent Migration for Virtual Machines
VMI: An Interface for Paravirtualization
A Comparison of Software and Hardware Techniques for x86 Virtualization
The evolution of an x86 virtual machine monitor
(ACM--fetch while on campus)
- User-level isolation & execution of untrusted code (read two or three)
Native Client: A Sandbox for Portable, Untrusted
x86 Native Code (15 pages)
Vx32: Lightweight User-level Sandboxing on
the x86 (14 pages)
User-Level Sandboxing: a Safe and Efficient
Mechanism for Extensibility (13 pages)
- Stretching Linux
An Analysis of Linux Scalability to Many Cores (16 pages)
FlexSC: Flexible System Call Scheduling with Exception-Less
System Calls (14 pages)
- Scheduler activations (a different threading approach)
Scheduler Activations: Effective Kernel Support for
the User Level Management of Parallelism
An Implementation of Scheduler Activations on the NetBSD Operating System
- Filesystem reliability options (you can read one, two, or three)
Analysis and Evolution of Journaling File Systems
Soft Updates: A Technique for Eliminating Most Synchronous Writes in the Fast Filesystem
Journaling Versus Soft Updates: Asynchronous Meta-data Protection in File Systems
Remember, you are not required to read something on this list.
This is the list of pre-approved readings. If you want to
read something else,
make a suggestion on the book choice page.
If you are proposing a collection of papers
which has not already been approved,
please send me a list of titles, page counts, and URLs.
I would appreciate it if you would make submissions in
HTML according to the format used in this document so I
can easily add approved readings. For example:
<LI>Paper collection name
<LI><CITE>Paper Title 1</CITE> (NN pages),
<LI><CITE>Paper Title 2</CITE> (NN pages),
(Back to Book Report page)