Windows Vista's firewall can easily be subverted because of design decisions made by Microsoft, according to researcher at Symantec.
Orlando Padilla, a Symantec security response team member who authored a study released last week on how well Vista stands up to current malware, took the new operating system's firewall to task in a blog.
"[The firewall] poses a great limitation for malicious code looking to backdoor a host," said Padilla in the entry. "Unfortunately, the Unblock button may be accessed with the same privilege level as a standard user. This configuration of privileges creates a point of vulnerability that undermines the effectiveness of the firewall's policy in Windows Vista."
Padilla was unavailable for comment, but Javier Santoyo, manager of development in Symantec's research group, explained. "Subverting a firewall is nothing new, but Microsoft, with Vista and User Account Control [UAC], had the ability to strengthen the firewall." The company didn't take that opportunity, he said.
The problem, according to Santoyo, is that while Vista's firewall blocks all third-party and untrusted network traffic unless the user clicks the Unblock button, it's not hard for attackers to code their malware so that the software surreptitiously clicks the button. The SendMessage API call can be used to automate that function, said Padilla.
Microsoft could have guaranteed that only a click by a real live user would unblock the firewall for an application requesting internet access. "They could have coded it so only an interactive user could click the button," he said. It's possible that rival firewall vendors may take up the approach, Santoyo said.
The motivation for tricking Vista into allowing malware access to the internet is plain: "They could then download other malicious code" or hide the command-and-control traffic between an infected PC and the hacker using the machine as a spam zombie or denial-of-service attacker, said Santoyo.
"Assuming an attacker can perform the firewall unblock attack, most of the functionality commonly present in a bot is available," wrote Padilla in his research paper (download PDF).
"Yes, I think attackers will try this," said Santoyo. "It's not hard to do."
Microsoft officials could not be reached for comment.