peripherals

This book is a must-have if you want to write your own device drivers for Linux, or if you just need a place to get started hacking the kernel.

All the basics are covered and you really don't need to know much about the kernel internals to read this book. If you're not a beginner it still contains valuable information and it's organized in a way to make it very easy to skip the areas you are already familiar with.

After a few chapters you'll be able to write your own driver from scratch. When you have read the whole book you'll also be familiar with some of the more complex concepts of kernel programming. You will not be an expert kernel hacker but you will have an excellent base for learning more yourself.

The only drawback is that it doesn't completely cover the newer kernels (2.2 and above), but although some details have changed, most of the concepts are still more or less the same, so the knowledge you gain can easily be adapted to the later versions.

Does contain lots of interesting info about Linux drivers and Linux in general. But the meat is more reference than tutorial. A really great tutorial begins chapter two, and so I thought here I'm really going to learn everything I need to know about creating Linux drivers. Didn't turn out that way. After you work the first exercise, that is, the tutorial example at the beginning of chapter two, you have seen the last of the complete examples. From here on it's code fragments and isolated functions. The author obviously could have written the book we need. But he didn't. It's a valuable book, but it's not a tutorial. What a beginner needs are whole, complete, real, listings of programs that work. Which reminds me, a real driver that drives a real device, presented in its entirety, with all details of how to compile it, and how to run it, would have been far more instructive than a "driver" that reads and writes only in memory so that it can be "portable" across many Linux platforms. A portable driver probably is a neat stunt that impresses existing gurus, but that's not the group that needs this book. To see what I'm driving at, look at Kernighan and Pike's "The UNIX Programming Environment." Their big programming project is indeed presented in fragments and isolated functions in their chapter eight, but the entire project just as it will appear on your disk is listed in the appendix. If Rubini had followed that model his book could have been really instructive. But he didn't. So there's an opportunity here. Some guru should set down and assemble these fragments into the book we need.

I can't understand the 5-star reviews of this book. I wonder if anyone who gave the book 5 stars ever tried to sit down and actually write a device driver. I doubt it.

The book suffers from two main problems:

1. Presentation is disorganized. The book reads as if the authors sat down and planned out what chapters to cover. That part is good. It does NOT seem like they planned further than that. The text within each chapter seems haphazard. Disorganized. Thrown together. The authors have no concept of when to start a new paragraph, so topics are strung together piecemeal. The whole book is confusing, making the reading of this book very frustrating.

2. The material is presented at a frenetic pace. As I was reading the book, it felt like there was a conveyor belt feeding me information, getting faster and faster without a break. The authors feel like they can throw everything but the kitchen sink at you in a very short time. Information, minute details, big ideas, analogies, and code get thrown at you fast and furious. It starts at chapter one and just gets worse from there, making the reading of this book downright difficult.

Difficult and confusing. A good characterization of the book.

In summary, this book is NOT a tutorial. It is NOT a guide. Don't make the mistake of buying this book expecting a gentle (or even a not-so-gentle) introduction to writing device drivers.

IMHO, the market is still open for a good book on the subject of writing device drivers for Linux.