Tracking Down Security Flaws
Besides ferreting out XSS flaws, Lemon is sniffing out other security problems, including response-splitting attacks (where an attacker sends a single HTTP request that forces the Web server to form an output stream, which is then interpreted by the target as two HTTP responses instead of one response). This type of flaw can be used in XSS attacks or similar exploits. Other security bugs Lemon is finding include cross-user defacement (temporary site defacement used in cases of information, ID or password theft), Web cache poisoning (a larger defacement where a cache used by multiple users is poisoned and causes visitors to think the site has been defaced or that a bogus site is in fact a genuine site), cookie poisoning (modification of the contents of a cookie in order to bypass security mechanisms), stack trace leaks (a stack trace shows where an error occurs in a program), encoding issues and character set bugs.Anantharaju said that since the tool is home-grown, its easy to integrate into Googles automated test environment and to extend based on the search giants specific needs. "We are constantly in the process of adding new attack vectors to improve the tool against known security problems," he said. Click here to read Larry Seltzers commentary on why fuzzing is the best available way to ferret out application security flaws. Still, testing experts find it curious that Google would brew such a tool at home. "As an outsider, Id say that it wouldnt be first thing Id spend time onto build my own Web application security toolsif I were an enterprise," said Matasano President Dave Goldsmith. When Google enters a market, that market changes. But, in spite of the search giant having launched a security blog recently and having sent tongues wagging on the subject of whether it would enter the security market, experts arent sweating Lemon. "Id be pretty surprised if they jumped into the Web application security market, into the scanning market," Goldsmith said. "I think they should take Web applications seriously. I dont know that building my own internal tools is one [option Id think of] when there are tools they could buy that are already available. But with enough Web applications to scan, Id look at purchasing over building. [Lemon has] got to be driven by the fact that theyve got a lot of applications to look at." Watchfires Allan said that he uses manual fuzzers himself "all the time." The reason is that they allow a user to apply human intelligence in looking at a Web application and determining how it should be fuzzed. "You get the ability to specify what and how it works," he said. "My guess is [Google] developed a customized environment. In order to control how input was being fuzzed, they [also] built [a fuzzer] on their own." At any rate, if Google were to be building a fuzzer for commercial releasesomething he strongly doubtsAllan said that he would be surprised, given its "do no evil" motto. "Make no mistake, what theyre building, if they release it to a malicious individual, it could be used for malicious attacks," he said. Google told eWEEK that it has no plans to market Lemon externally, at least not in the near future, given that its "highly customized for Google apps." The companys security team did in fact evaluate commercially available fuzzers but "felt that our specialized needs could be served best by developing our own tools," a spokesperson said. Editors Note: This story was updated to comments from a Google spokesperson.
Check out eWEEK.coms Security Center for the latest security news, reviews and analysis. And for insights on security coverage around the Web, take a look at eWEEKs Security Watch blog.
Fuzzing tools are powerful, but theyre also a bear to create. Why would Google create one from scratch, given that powerful scanners/fuzzers already exist in the commercial sector?