Potent malware link infects almost 300,000 webpages
Johnny Come Lately
A security researcher has identified a new attack that has infected almost 300,000 webpages with links that direct visitors to a potent cocktail of malicious exploits.
The SQL injection attacks started in late November and appear to be the work of a relatively new malware gang, said Mary Landesman, a researcher with ScanSafe, a web security firm recently acquired by Cisco Systems. Hacked sites contain an invisible iframe that silently redirects users to 318x .com (a space has been added to protect the clueless), which goes on to exploit known vulnerabilities in at least five applications.
At time of writing, this web search showed more than 294,000 webpages that contained the malicious script. Infected sites included yementimes .com, parisattitude .com and knowledgespeak .com.
People who visit infected pages receive an invisible link that pulls code from a series of sites tied to 318x .com. The code looks for insecure versions of Adobe Flash, Internet Explorer, and several other Microsoft applications, and when they are detected it exploits them to surreptitiously install malware known as Backdoor.Win3.Buzus.croo. The rootkit-enabled program logs banking credentials and may do other nefarious bidding, Landesman said.
At the moment, about two percent of the requests ScanSafe sees are for sites infected by the malicious link, an indication the threat is significant, Landesman said.
SQL injection attacks prey on web applications that fail to adequately inspect user supplied input before passing it off to a webserver's backend database. They are a favorite way of adding malicious links and content to third-party websites and were also the the chink that allowed Albert Gonzalez and other hackers the toehold they needed to steal more than 130 million credit card numbers from card processor Heartland Payment Systems and four other companies.
The fingerprints on this latest attack lead Landesman to believe the perpetrators are new to the SQL injection game. More sophisticated mass attacks using the method, such as the Gumblar infection inject unique, dynamically-generated links that prevent researchers from being able to locate them using web searches.
Gumblar also uploads exploits directly to infected sites, which greatly complicates white hat efforts to clean up the mess. Rather than shutting down a single site that's hosting the malware, thousands of mom and pop sites must be disinfected one at a time.
"I'm not convinced SQL injection is the method they're most accustomed to," Landesman said of the gang behind the most recent mass infection. "It's almost as if they're a seasoned attacker but this is their first foray into managing a wide-scale web attack." ®
It's not rocket science
True - and in the .NET world you can use urlscan.
But it becomes trickier when you're responsible for a large, complex web site with poor documentation (and, trust me, there are rather a lot of these). Retrofitting is non-trivial and you may also need to introduce security testing and change management processes to ensure that vulnerabilities are not inadvertently reintroduced.
SQL injection is easily prevented
if you parse your (PHP) input with:
$params['name'] = isset($_POST['name']) ? mysql_real_escape_string($_POST['name']) : '';
or if it's a numerical value you're expecting, even better:
$params['name'] = isset($_POST['name']) ? (int)($_POST['name']) : 0;
(use $_GET['name'] if you're passing parameters in the URL.)
And HTML injection is removed simply by:
$params['name'] = preg_replace("/\<.*\>/g", "", $params['name']);
Then you only work with the clean $params array. There's no reason any webmaster can't implement these most basic code checks. It's not rocket science.
Just gonna quickly blackhole that domain before anyone finds out