Browse Books

Reviewer: Jeffrey B. Putnam

The growth of open-source software has surely been good for the software development industry. It has also been good for those who like to tinker, understand, and even just play around. Surely, of all the open-source projects, Linux is at the forefront; it has grown from a toy to a major player. Linux consists of two major pieces: the kernel and the supporting code (which includes most of what users see, including compilers, shells, graphical user interface (GUI) systems, applications, and Web servers). This book covers the Linux kernel (although it also discusses supporting systems in passing). The kernel is a fascinating beast in itself: it has thousands of contributors and millions of lines of code; it supports a huge variety of devices on a number of central processing units (CPUs); and it is always moving (which way it is currently going-sometimes forward, sometimes backward-often depends on who is measuring what). Furthermore, it contains both code that is at the edge of operating system (OS) development and code that has been around for years. The source code is available to anyone, and anyone can read it and change it. (Getting changes incorporated into a distributed kernel is rather more difficult, however.) The complication, naturally, is that kernel code doesn't look like most user-level code, since the runtime environment for the kernel has very different constraints. Because of the different environment and the different requirements for a kernel, kernel code uses a specific set of coding practices, data structures, and locking mechanisms. The book covers the basic organization of the Linux kernel, and it discusses many things that are specific to kernels in general (for example, scheduling, process management, memory management, device drivers, and synchronization) and Linux in particular. This book is a good starting place for beginners because the level of detail is just about right: not so high as to ignore the interesting topics, nor so low as to overwhelm readers with long stretches of code. (In fact, most of the actual code snippets are less than a page long.) As a result, it is ideal supplementary reading for an introduction to an OSs course. For people who are interested in coding in the kernel, or for an OSs course that focuses more on implementation details, this book could be used nicely as a primary text, with a secondary text (such as Understanding the Linux kernel [1]), the Linux kernel source itself, and the Linux kernel mailing list providing more detail. The writing is clear and well organized, providing pointers to Linux 2.6.34. (The current release is 2.6.38, and version 2.6.39 is in the release candidate stage.) Since the kernel is a moving target (although much more stable than earlier in its lifetime), some changes are inevitable. In summary, though this is not a replacement for a general OSs text, it is an excellent introduction to the Linux kernel and to OS implementation in general. It is also quite a good read: it's sometimes casual, sometimes precise, and leavened with humor. Online Computing Reviews Service