clock menu more-arrow no yes

Filed under:

Google uses 'fuzzing' to automate Chrome security tests

New, 8 comments

Google automates hundreds of virtual machines and over 6,000 chrome instances to "fuzz test" Chrome's security.

Google Chrome Security
Google Chrome Security

With projects like Pwn2Own, Google has often made it apparent — to both users and developers — that it sees security as one of its highest priorities in regard to the Chrome browser. An update to The Chromium Blog introduces a security testing environment that it affectionately calls "ClusterFuzz." The humorously named environment uses "fuzz testing," or the distribution of millions of test cases across "several hundred" virtual machines to identify bugs en masse.

Google is stating that ClusterFuzz runs approximately 6,000 simultaneous Chrome in order to analyze crashes, identify patterns in bugs and exploits, and quickly seek out possible fixes. Exploring the specific tests and methodologies that ClusterFuzz employs can get technical in a hurry, but the "Core Principles" page of Chromium Projects is a good jumping off point for those who are curious.

ClusterFuzz was brought online at the end of last year, and since that time it has identified 95 vulnerabilities and was able to find fixes for 44 of them before they were rolled into a stable release. This rapid security testing benefits not only Google and the Chromium Project, but also open source projects WebKit and FFmpeg, where Google makes upstream contributions as bugs are discovered.