The flaw was found in OpenBSD's kernel and involves the way the OS handles certain kinds of IPv6 packets, according to the researchers at Core Security Technologies Inc. who discovered the problem. The vulnerability affects versions 3.1, 3.6, 3.8, 3.9, 4.0 and 4.1 of OpenBSD. Also, all other versions that support the IPv6 stack are thought to be vulnerable.
In order to exploit the flaw, an attacker need only be able to send fragmented IPv6 packets to a target system. This requires direct access to the target network, however the attacker's machine does not need to have its own IPv6 stack in order to make the exploit work, Core said. Users who don't need to route IPv6 traffic can block those packets using OpenBSD's native firewall.
Ivan Arce, chief technology officer of Boston-based Core, said that once an attacker knows what he is looking for, exploiting this vulnerability is trivial.
"It was difficult for us to spot and figure out what was going on because the kernel was crashing in a bunch of different places because the memory was corrupted," Arce said. "Once we figured it out, it's not difficult to exploit."
This is the first IPv6-related vulnerability that leads to a remote system compromise that Core has seen, Arce said, but it's unlikely to be the last.
"It's a complex protocol and some of the implementations are difficult," he said. "When things are that complex, it's likely you will run into security vulnerabilities as well."