Kernel mode rootkits are more viable than has been realised and could be used to bypass more or less any security product in existence, researchers at Bromium have discovered after conducting a proof-of-concept attack using a modified variant of in the infamous TDL4 malware.
Due to be presented in more detail by the firm at this week's Security BSides event in London, the research involved 'tweaking' the TDL4 variant that had appeared to take advantage of the Windows kernel privilege zero day (CVE-2013-3660), discovered in June last year.
With a new payload, what this created was something lethal enough to overcome a variety of security layers the team tested against it such as antivirus, sandboxes and intrusion prevention, making it a sort of "Swiss Army knife" attack hiding behind ring zero.
"While many were aware of the discovery of the TDL4 rootkit rumoured to be using kernel exploit code at the end of last year, few paid it any serious attention. And that was a huge error of judgement," said Bromium's head of security, Rahul Kashyap.
"We've discovered it could prove lethal to many systems. By simply tweaking the exploit, we found we could bypass the typical security software you'd expect to encounter on a corporate user machine," he said.
The low-down is that kernel mode rootkits represent a greater threat than previous realised. The received wisdom is that rootkits execute in user mode with administrator privileges because this is more stable if easier to detect. Kernel mode is the Holy Grail but it is hard to pull off without causing a system crash that negates the invisibility factor - or perhaps not if Bromium's TDL4 experiment is representative of the state of the art.
But get a stable kernel mode rootkit into a PC using a zero day flaw (Bromium used an unpatched Windows system of course) and almost nothing could detect it.
"To work around these monitoring [security] tools does not require any sophistication," added Kashyap.
Exactly who might wield something this powerful is anyone's guess but one might start with national security agencies. Few other criminals would bother because, as Broimum admits, very few conventional attacks require the level of long-term persistence that a state-backed cyber-attack would crave.
Microsoft had released more than 80 kernel-level patches in 2013, the firm said. Despite its much-vaunted Security Development Lifecycle (SDL), this kind of flaw was still common.
"A layered approach is still the only way to protect organisations. But it's vital, when designing and architecting enterprise security, organisations are aware of fundamental technology limitations at each level, adding layers of protection that address these weaknesses, and in so doing make sure each layer counts," said Kashyap.
Bromium's motivation in pointing this out is, of course, to advertise its own Citrix Xen-based micro-virtualisation approach to the desktop security problem.
The full presentation will be made at the BSides conference on 29 April at Kensington and Chelsea Town Hall, London.