Fuzz testing or Fuzzing is a Black Box software testing technique, which basically consists in finding implementation bugs using malformed/semi-malformed data injection in an automated fashion.
- 1 A trivial example
- 2 History
- 3 Fuzzer implementations
- 4 Comparison with cryptanalysis
- 5 Attack types
- 6 Application fuzzing
- 7 Protocol fuzzing
- 8 File format fuzzing
- 9 Fuzzers advantages
- 10 Fuzzers limitations
- 11 Why Fuzz?
- 12 Fuzzers from OWASP
- 13 Technical resources on OWASP
- 14 References
- 15 Fuzzing tools
A trivial example
Lets's consider an integer in a program, which stores the result of a user's choice between 3 questions. When the user picks one, the choice will be 0, 1 or 2. Which makes three practical cases. But what if we transmit 3, or 255 ? We can, because integers are stored a static size variable. If the default switch case hasn't been implemented securely, the program may crash and lead to "classical" security issues: (un)exploitable buffer overflows, DoS, ...
Fuzzing is the art of automatic bug finding, and it's role is to find software implementation faults, and identify them if possible.
Fuzz testing was developed at the University of Wisconsin Madison in 1989 by Professor Barton Miller and his students. Their (continued) work can be found at http://www.cs.wisc.edu/~bart/fuzz/ ; it's mainly oriented towards command-line and UI fuzzing, and shows that modern operating systems are vulnerable to even simple fuzzing.
A fuzzer is a program which injects automatically semi-random data into a program/stack and detect bugs.
The data-generation part is made of generators, and vulnerability identification relies on debugging tools. Generators usually use combinations of static fuzzing vectors (known-to-be-dangerous values), or totally random data. New generation fuzzers use genetic algorithms to link injected data and observed impact. Such tools are not public yet.
Comparison with cryptanalysis
The number of possible tryable solutions is the explorable solutions space. The aim of cryptanalysis is to reduce this space, which means finding a way of having less keys to try than pure bruteforce to decrypt something.
Most of the fuzzers are:
- protocol/file-format dependant
- data-type dependant
- First, because the fuzzer has to connect to the input channel, which is bound to the target.
- Second, because a program only understands structured-enough data. If you connect to a web server in a raw way, it will only respond to listed commands such as GET (or eventually crash). It will take less time to start the string with "GET ", and fuzz the rest, but the drawback is that you'll skip all the tests on the first verb.
In this regard, Fuzzers try to reduce the number of unuseful tests, i.e. the values we already know that there's little chance they'll work: you reduce unpredictability, in favor of speed.
A fuzzer would try combinations of attacks on:
- numbers (signed/unsigned integers/float...)
- chars (urls, command-line inputs)
- metadata : user-input text (id3 tag)
- pure binary sequences
A common approach to fuzzing is to define lists of "known-to-be-dangerous values" (fuzz vectors) for each type, and to inject them or recombinations.
- for integers: zero, possibly negative or very big numbers
- for chars: escaped, interpretable characters / instructions (ex: For SQL Requests, quotes / commands...)
- for binary: random ones
Please refer to OWASP's Fuzz Vector's resource for real-life fuzzing vectors examples and methodology.
Protocols and file formats imply norms, which are sometimes blurry, very complicated or badly implemented : that's why developers sometimes mess up in the implementation process (because of time/cost constraints). That's why it can be interesting to take the opposite approach: take a norm, look at all mandatory features and constraints, and try all of them; forbidden/reserved values, linked parameters, field sizes. That would be conformance testing oriented fuzzing.
Whatever the fuzzed system is, the attack vectors are within it's I/O. For a desktop app:
- the UI (testing all the buttons sequences / text inputs)
- the command-line options
- the import/export capabilities (see file format fuzzing below)
For a web app: urls, forms, user-generated content, RPC requests, ...
A protocol fuzzer sends forged packets to the tested application, or eventually acts as a proxy, modifying requests on the fly and replaying them.
File format fuzzing
A file format fuzzer generates multiple malformed samples, and opens them sequentially. When the program crashes, debug information is kept for further investigation.
One can attack:
- the parser layer (container layer): file format constraints, structure, conventions, field sizes, flags, ...
- the codec/application layer: lower-level attacks, aiming at the program's deeper internals
One example of file format related vulnerabilities: MS04-028 (KB833987) Microsoft's JPEG GDI+ vulnerability was a zero sized comment field, without content.
Surprisingly, file format fuzzers are not that common, but tend to appear these days; some examples:
- A generic file format fuzzer : Ilja van Sprundel's mangle.c; "it's usage is very simple, it takes a filename and headersize as input. it will then change approximatly between 0 and 10% of the header with random bytes." (from the author)
- Zzuf can act as a fuzzed file generator, http://sam.zoy.org/zzuf/
- One may use tools like Hachoir as a generic parser for file format fuzzer development.
The great advantage of fuzz testing is that the test design is extremely simple, and free of preconceptions about system behavior (from Wikipedia http://en.wikipedia.org/wiki/Fuzz_testing).
The systematical/random approach allows this method to find bugs that would have often been missed by human eyes. Plus, when the tested system is totally closed (say, a SIP phone), fuzzing is one of the only means of reviewing it's quality.
Fuzzers usually tend to find simple bugs; plus, the more a fuzzer is protocol-aware, the less weird errors it will find. This is why the exhaustive / random approach is still popular among the fuzzing community.
Another problem is that when you do some black-box-testing, you usually attack a closed system, which increases difficulty to evaluate the dangerosity/impact of the found vulnerability (no debugging possibilities).
The purpose of fuzzing relies on the assumption that there are bugs within every program, which are waiting to be discovered. Therefore, a systematical approach should find them sooner or later.
Fuzzing can add another point of view to classical software testing techniques (hand code review, debugging) because of it's non-human approach. It doesn't replace them, but is a reasonable complement, thanks to the limited work needed to put the procedure in place.
Recent fuzzing initiatives:
- The Month of Kernel Bugs, which revealed an Apple Wireless flaw; see http://kernelfun.blogspot.com/; mainly used file system fuzzing tools
- The Month of Browser Bugs (review here); number of bugs found: MSIE: 25 Apple Safari: 2 Mozilla: 2 Opera: 1 Konqueror: 1; used DHTML, Css, DOM, ActiveX fuzzing tools
Fuzzers from OWASP
Fuzzing with WebScarab: a framework for analysing applications that communicate using the HTTP and HTTPS protocols
JBroFuzz: a web application fuzzer
WSFuzzer: real-world manual SOAP pen testing tool
Technical resources on OWASP
Wikipedia article - http://en.wikipedia.org/wiki/Fuzz_testing
Fuzzing-related papers - http://www.threatmind.net/secwiki/FuzzingPapers
Untidy - XML Fuzzer (Now integrated into Peach)