Braving the Cold (Boot) – A Sneak Peek of My Presentation at Black Hat

by Patrick McGregor

Cold boot attacks aren’t theoretical academic exercises. Cold boot attacks are real. And they’re serious.

In the past few years, companies have poured hundreds of millions of dollars into full disk encryption technologies. Companies expect full disk encryption to reduce the risk of exposure of sensitive information such as intellectual property or customer data. Reality often deviates from what is expected, however. Researchers from Princeton shocked the industry earlier in 2008 when they released a research paper that showed that low-cost “Cold Boot” attacks could be used to defeat the security of most full disk encryption systems. They recently even published all the tools needed to do this at home!

Some have argued that Cold Boot attacks are not serious security threats. I disagree! First, an unskilled person can capitalize on the exploit using simple, automated steps and publicly available tools. In fact, Cold Boot attacks require nothing more than plugging a USB drive into a laptop. Second, the physical target of a Cold Boot attack, such as a laptop, is very easily obtainable (see the recent Ponemon report on laptops lost/stolen in airports – scary!). Third, although many laptops and desktops are stolen via random acts of theft, it is well known that some criminals profit from organized, calculated data theft. It is only a matter of time before we hear of a high-profile data breach that results from a simple Cold Boot attack.

I am excited to present at Black Hat several innovations for preventing Cold Boot attacks. In addition to summarizing how a Cold Boot attack works, I’ll describe four new software techniques for hardening full disk encryption against the attacks. The software technology was developed by myself, Tim Hollebeek, Alexander Volynkin, and Matt White. All of us work for BitArmor, an exciting security startup based in Pittsburgh. Here’s a sneak peek:

· Wash up: Wipe keys immediately before certain OS state transitions, such as before the computer shuts down or goes into hibernation mode – accessing the memory will yield nothing.

· Take advantage of BIOS memory smashing: By strategically placing keys in certain regions of memory, we can rely on the BIOS boot process to overwrite keys before any operating system can dump the contents of memory.

· Is it chilly in here?: Using built-in temperature sensors, we can lock down the system in reaction to temperature drops that may indicate a Cold Boot attack is in progress.

· Create a virtual enclave for keys: We can implement special cryptographic, OS and processor architecture techniques to provide robust protection for keys against the most aggressive cold boot attacks. By creating a “virtual secure enclave” for encryption keys in software, an attacker cannot extract critical keys from memory – even if the RAM is super-cooled.

Hope you can join us at Black Hat as we take an in-depth look at the future of full disk encryption technology.

Format was fireside chat, with Craig Mundie, Microsoft's Chief Research and Strategy Officer sitting and talking with Chris Leach, Chief Information Security Officer at Affiliated Computer Services.  [fwiw, I personally don't love the fireside chat format.  Give me videos, fancying graphics and lots of acrobats on the stage ...]

I knew generally what Craig was going to talk about, but I was very interested to hear Craig's perspective and see how he thought about and talked about the end-to-end Trust topic.  In my opinion, this is one of the key topics that could help guide where Microsoft security efforts will go over the next 5 years, building on the past 5 years, and I am happy to see that leadership (Craig, Scott Charney) are approaching it as a dialog with industry and a recognition that it needs interoperability and industry support.

Two key topics stuck with me at the end of the keynote:

1. How security and privacy are very independent, supporting each other, while also having a tension between them.
2. Any technological efforts supporting End-to-end Trust will need to be very inclusive in order to work in heterogeneous environments.  Past infrastructure efforts (e.g. PKI) have demonstrated that the level of work and investment required means that it is more likely to hit roadblocks if existing business processes are excluded.

After the keynote, with the excellent assistance of Eric Green, I was able to pin down several Microsoft partners and get their thoughts on these two areas.  Listen to the attached mp3 to hear our discussions with these good folks:

Sandy Porter Director, Strategy Avoco Secure
Jeremiah Beckett President SecureVantage Technologies
Patrick McGregor, Ph.D. CEO BitArmor
Jon Callas CTO & CSO PGP Corporation
Conrad G. Bayer Senior Vice President IDA Strategy Avalaris, Inc.
Edward J. Gaudet Senior Vice President, Corporate Development and Marketing Liquid Machines

 

I did get a couple of these folks on video as well, so once I get that edited and uploaded, I'll update with links to those.

Additional information that is available on End to End Trust:

• Read Scott Charney’s Full Article about End to End Trust
• Read Microsoft's End to End Trust White Paper
• Join the dialogue. Go to Microsoft's End to End Trust forum, and let your voice be heard.

Best regards from RSA ~ Jeff

This is scary news and rightfully so. We have seen encryption vendors approach this differently.

• The don’t-worry, be happy approach: Some claim the attack is so esoteric, the customer need not worry – this is just research stuff.
• Leave it to us approach: Some claim to have solved the problem, but with no indication of what that means or how they do it.
• Increase your complexity approach: Some want you to increase the end-user complexity with process and unnatural actions to solve the problem. Not a good idea – every time we ask the end user to be responsible, we lose control and confidence that it was indeed secure. Transparency is the key to security..

We at BitArmor have taken another approach – the “solve the problem” approach. In fact, we had solved this problem, before it even became a known issue. Our CEO, Patrick McGregor is one of the researchers mentioned in the Princeton paper as having proposed architectural enhancements to prevent (the key word being prevent :))these attacks. From the paper:

“Others have proposed architectures that would routinely encrypt the contents of memory for security purposes [28, 27]. These would apparently prevent the attacks we describe..”

The “others” mentioned above, in case you were wondering, are McGregor et al… Check out his blog on his experience at Princeton...

Sorry if we seem to be bragging a bit – not often does a small startup from steeltown open up such a big can of whupass against a new broad new threat!

We have since applied (we had the technology already for a while) for multiple patents on technologies to solve these and similar attacks. Find out more on the BitArmor website (http://www.bitarmor.com/prevent-cold-boot-attacks/) for a high level look at how we deal with specific cold boot threats.

As soon as we can write up detailed information on exactly how we are dealing with the specific cold boot threats in our FDE (full disk encryption) as well as PFE (persistent file encryption) solutions, we will put it up here. Look for more information next week…