Fuzzing, or fuzz testing, is an automated software testing technique that supplies invalid, unexpected, or random data (fuzz) to a computer program to detect vulnerabilities like crashes, code assertion failures, or memory leaks. It's particularly useful for testing programs that process structured inputs, such as file formats or network protocols, by generating semi-valid inputs that are accepted by the parser but can cause unexpected behavior in the program, revealing unhandled corner cases.

The concept of fuzzing originated from a 1988 class project led by Prof. Barton Miller at the University of Wisconsin. The project aimed to test the reliability of UNIX utilities by feeding them random inputs. This early form of fuzzing, which generated inputs without any knowledge about the program's structure, is now referred to as black box, generational, or dumb fuzzing.

Over the years, fuzzing has evolved and can be categorized based on its knowledge of the program's input structure (smart vs. dumb) and its awareness of the program's internal structure (white, grey, or black-box). Smart fuzzers generate inputs based on models of valid input formats, while dumb fuzzers generate inputs without such knowledge. White-box fuzzers analyze the program's source code to generate inputs, grey-box fuzzers use some form of instrumentation to observe program behavior, and black-box fuzzers operate without any knowledge of the program's internals.

Fuzzing has been instrumental in identifying vulnerabilities in critical software components, including widely used protocols and applications. Notable examples include the discovery of the Heartbleed vulnerability in OpenSSL and numerous bugs in web browsers through projects like Google's ClusterFuzz. It has also been used in high-profile security competitions, such as the DARPA Cyber Grand Challenge, to automatically find and patch vulnerabilities in software.

The development of fuzzing tools and services, such as Google's OSS-Fuzz, Microsoft's Project Springfield, and the open-source AFL (American Fuzzy Lop), has made fuzzing accessible to a wider audience of security researchers and developers. These tools automate the generation of inputs and the detection, triage, and minimization of failures, making it easier to identify and fix vulnerabilities in software.

Fuzzing is now a standard part of the security testing toolkit, used primarily to find vulnerabilities in software before they can be exploited by attackers. However, it's also used more broadly in quality assurance to identify bugs, demonstrating that while fuzzing can prove the presence of defects, it cannot prove their absence.

Fuzzed Out

principal vulnerability researcher
artist. coder. writer. musician.

Fuzzed Out

geeknik`s Lab'>">