I anxiously awaited reading and putting this book to use. Fuzzing is one of those "mystical" concepts that the people cranking out exploits were doing and I wanted to be able to use some of the publicly available fuzzers to fuzz for vulnerabilities and join the ranks.
From the back cover: "...Now, its your turn. In this book, renowned fuzzing experts show you how to use fuzzing to reveal weaknesses in your software before someone else does."
I thought the book excellently covered the theory portions of fuzzing. The format of theory/background of a fuzzing method (Environment Variable and Argument Fuzzing, Web Application and Server fuzzing, File Format Fuzzing, Network Protocol Fuzzing, Web Browser Fuzzing, and In-Memory Fuzzing) followed with that fuzzing method Automation or on Unix and then on Windows worked perfectly. It was a good structure and informative. The Automation or Unix and Windows sections fit in well with the theory sections before it.
I think the book falls a bit short on practical execution (case studies) of using the fuzzing tools. Granted I say this based on my own expectations of what I would like to see from a fuzzing book but also from what the authors say in the preface that we will get out of the book. They say, "We detail numerous vulnerabilities throughout the book and discuss how they might have been identifies through fuzzing." Some of the case studies are exactly what I expected like case studies in Chapter 10, the fuzzing with SPIKE section in Chapter 15, and the Complete Walkthru with Sulley in Chapter 21. Some of the others fall a bit short. I expected a lot more out of the ActiveX fuzzing sections (chapter 18), the Shockwave Flash example in Chapter 21 was useful for the discussion of creating a test case for a protocol but after 11 pages of mostly code in the last section we basically get told to load it into PaiMei and "go fuzz", and while the theory parts of chapter's 7 & 8 were great, telling me to find an AIX 5.3 box to see some example environment variables and argument vulnerabilities was less than useful. It would have been much more useful to use some of today's fuzzing tools to find some old vulnerabilities in something like *BSD or old RedHat distributions, something I might have in the lab or at least something I could install in VMWare.
Likes: Theory, background, discussion of how and why they built the "author built" fuzzers they cover in the book, some of the case studies gave me everything I needed to reproduce on my own in the lab. Providing the fuzzers on the companion website was great as well. The George Bush quotes were hilarious as well and made me look forward to each chapter so I could get another quote.
Dislikes: some of the case studies I don't think went into enough detail (no step by step instructions), I think the explanations of the blocks of code could have been better and numbering lines so we could refer to them in the text would have helped. The discussion of the existing frameworks was a little bit light (but we do get told to go the companion website for more info). Ideally we would have walked thru a couple of easy examples using multiple fuzzer frameworks to get us from advisory to EIP= 0x41414141. That would have been nice to see.
Overall a great book, it has a place on the bookshelf next to shellcoder's handbook and some other programming books and it will be used (many times) as a reference to play with the various fuzzers available out there.
Perhaps a more appropriate title would be: "Fuzzing for Dummies" or "Fuzzing 101"- but I mean this in a really good way. Why I say this is because of how the book is set up, starting with the background history of fuzzing, and many variations of what fuzzing really is. These are excellent so those who may not have this background don't jump in blindly to this area. For example, Chapter 3 goes into the Fuzzing Methods and Chapter 4 discusses Data Representation. While not lengthy discussions, they are good to set up for the actual doing part in the rest of the book
I liked that the book starts out with what fuzzing is good for, the steps that you have to take for it to be successful, and what fuzzing is not good at. It explains how vectors like access control issues, and design flaws fit into this category. Knowing this up front saves a lot of head banging later on down the road. It's also good that the authors point out that they are merely defining fuzzing in their specific realm: talk to others and you are going to find a whole different explanation. This is OK though- most of the security industry is like that.
Part II of the book starts to get into the heart of things, discussing the components required for fuzzing, more details into the tool they built called "WebFuzz" and then dive into the tests themselves. The author's openness in telling us what they did, then how it works, then tell you all the things to make it better makes this book even more valuable. Good efforts to share useful things and make them a community effort with proper guidance are never a bad thing. Plus, if you are interested in helping, this guidance gives you somewhere to start.
Essentially, this book gives you the blueprint of fuzzing and a bunch of ideas on how to get started down a more advanced path. Well written with good explanations of how the authors got where they got to as well a useful tool to get you started (located on their companion website), this book gives you the toolkit of building blocks for your future fuzzing endeavors.
I loved the layout of the book, with explanations, practical applications, and (mostly) working examples. There were two things I didn't like about the book. First, not all the examples worked. Specifically, the Protocol Informatics (PI) example will not run on any machine I have. When I searched for a solution, it led me to the second thing I don't like about the book: it appears the authors cribbed their section on PI from PI's own documentation. It's clear they didn't even try to run it on their own. It makes me question whether they really understand it; if not, why are they writing about it in their book? I also wonder what else they cribbed. I also wish they'd update the book's website more, as much as they refer to it in the text.
All that aside, I really did enjoy and appreciate the book as a whole, and it certainly gave me a great foundational knowledge of fuzzing.
In this book the authors do a number of things that are worth reading: o Document how and why SPIKE works (and implement their own block-based fuzzer sulley) o Go through the process of writing a .flv fuzzer o Go through the process of writing a Python ActiveX fuzzer, which was probably my favorite part. o Talk about the downsides of various kinds of fuzzing. For example, when is fuzzing with a genetic algorithm not the right thing to do?