Let me give you a bit of background before going further. It’s been almost a month since Cesar Cerrudo, security researcher and CEO of Argeniss publicly warned that the latest Microsoft operating system, Windows Server 2008, might not be as secure as it seems. On an advisory released on March 24, he stated:

“The problem discovered by Argeniss results from design issues that were not identified by Microsoft engineers during the Security Development Lifecycle (SDL), and allows accounts commonly used by Windows services (NETWORK SERVICE and LOCAL SERVICE) to bypass new Windows services protection mechanisms and elevate privileges to achieve complete control over the operating system.”

This security flaw was reported to affect Windows XP, Vista, Windows Server 2003 and the brand new Windows Server 2008. Although exploitation details were not given at that time, the advisory announced that further details would be given at the upcoming Hack-in-the-Box conference in Dubai (14-17 April 2008).

And it’s right at this point when this becomes really interesting. Microsoft quickly stressed that Cerrudo’s advisory described a “design issue” rather than a new vulnerability. Well, that was before Cerrudo’s paper, entitled “Token Kidnapping”, was presented in Dubai earlier this week. As to rectify is of wise people, Microsoft released on Thursday a pre-patch security advisory (951306) recognizing that they are “investigating new public reports of a vulnerability which could allow elevation of privilege from authenticated user to LocalSystem, affecting Windows XP Professional Service Pack 2 and all supported versions and editions of Windows Server 2003, Windows Vista, and Windows Server 2008”.

Some sources claim that the vulnerability addressed in that security advisory is the same discussed by Cesar Cerrudo in Hack-in-the-Box. This becomes quite evident after reading Thursday’s advisory and Cerrudo’s paper, although I haven’t found any official statement from Microsoft confirming so.

I must admit that this has been of benefit for both Microsoft and the researcher. On one side, Microsoft has had some time to research the problem before public details of exploitation are given, and on the other side, the researcher has created a great expectation on his talk. No doubt a win-win for everyone.

Now the question is, am I affected by this flaw? And in the absence of a security patch, what workarounds are available? Well, the good news is that no exploitation has been seen yet, however it’s worth to keep an eye on this, especially if you are a Web-hosting provider where customers could upload malicious code. Attending to Cerrudo’s paper, the main attack vectors are:

• Fully trusted ASP.Net code in IIS,
• SQL Server with users that have administrative privileges to load and run code,
• Any process with SeImpersonatePrivilege that loads and runs user-provided code, by acquiring a NetworkService token from the Microsoft Distributed Transaction Coordinator (MSDTC) service.

The following are some recommended workarounds (refer to Microsoft’s advisory for more details):

• On Windows XP and Server 2003: for IIS 6 don’t run ASP.Net in full trust and if classic ASP is enabled don’t allow users to execute binaries.
• On Windows Vista and 2008: for IIS 7, again, don’t run ASP.Net in full trust or don’t run websites (nor services) under NetworkServer or LocalService accounts. Use regular user accounts to run services instead.

Happy Sunday evening!

Further info on: http://www.argeniss.com and http://www.microsoft.com/technet/security/advisory/951306.mspx

PS. Earlier this week I got an email from my friend Adrian Pastor, also known as ‘pagvac‘ saying he was also giving a talk at HiTB 2008 in Dubai (Cracking into Embedded Devices and Beyond!). He is a great security professional and even a better chap. I promise I will post a link to his presentation on this blog when he makes it available.

Granted, I also gave him some tips on where to go in Dubai -views from the 24th floor bar at the Jumeirah Beach Hotel are awesome!-