Very intriguing.

Well, technically [rɛindaːl], but “Rhine Dahl”, or even, “Rine Doll” is pretty close.

Although there’s a joke that it’s now pronounced “AES” since Rijndael was selected to be the standard’s algorithm. There’s even a wave file of it being pronounced as such, which serves both as the joke and the definitive guide to pronouncing Rijndael (he has to say “rhine dahl” in order to get the joke out). e

“Rhine Dahl” (Rijndael) is pronounced “A - E - S”.

Pronunciation Clip | rijndael.info

Most in the information security community have heard of the “cold-boot” attack against encryption products that was recently released by Princeton. They put out a video demonstration that showed how they could read encryption keys right out of RAM even after the machine had been rebooted.

The big story here really comes from two things:

1. Sensitive information is stored in memory while your computer is running.
2. RAM isn’t as volatile as people thought it was.

What Princeton showed was that you could take a machine down hard (which keeps computers from cleaning up their memory), immediately boot to a system that dumps the contents of RAM, and then extract sensitive information from the content you dumped.

I was asked at work to try and replicate this behavior at work, so I contacted the group at Princeton and asked for a copy of Bit-Unlocker. They informed me that the tool is not available to the public. So I put out some feelers here on the blog and via a couple of other resources and found another option.

Msramdump

What Princeton showed was that you could take a machine down hard (which keeps computers from cleaning up their memory), immediately boot to a system that dumps the contents of RAM, and then extract sensitive information from the content you dumped.

A reader by the name of Dwight pointed me to a tool by the name of msramdump, which was exactly what I was looking for. Msramdump uses SysLinux and some Windows utilities to allow one to boot a Windows system from a USB disk (in my case a 4GB thumb drive) and dump the RAM as it currently exists. What I’ve Done So Far

The tool works. I built myself an msramdump drive yesterday and was able to successfully dump the memory from a Windows XP laptop. I then pulled that captured data onto another Linux box and captured the output of running strings against the same data.

The results were very interesting - even without using any cooling techniques.

As part of my first test I performed a simple Google search for my last name (miessler) on the Windows XP box after logging in with some test domain Active Directory credentials. After cutting power to the laptop and capturing the image I was able to find two interesting pieces of information, and this is within a few seconds of grepping the strings output.

• I found a highly legible and mostly unmangled copy of my Google search — including my exact search term (miessler).

• Even more interestingly, I found all sorts of interesting Active Directory information, most notably my current AD password — albeit in a slightly mangled state due to the RAM decay.

That second one tripped me out, and I’m going to confirm it tomorrow, but if I can actually pull AD passwords out of memory using this technique — without even using any cooling — that’s going to blow my mind. I wasn’t even aware that Active Directory passwords were stored in memory in plain text.

Next Steps

1. Confirm that I can pull Active Directory passwords out of memory.
2. Start searching for the keys used by various encryption products.

I’ll be working on these tomorrow and will update the post tomorrow night. If you have any questions, thoughts, or ideas hit me over email or through the comments below:.

I’m looking to do a POC of the encryption attack recently put out by Purdue.

If you guys know of a way of getting a hold of a copy of the tool they used (bit-unlocker), or any other methods of easily dumping and searching the RAM from a system, do let me know.

Everyone loves Google. They want to be everything to everyone, and they’re getting pretty damn good at it. Once you start using their services it gets easier and easier to migrate more of your life to them. But there’s a slight problem.

Google, like most other similar services, encrypts login traffic but not your content. So the moment you’re signed in they switch to plain-text communications and send everything to you in the open.

This means your mail, the news sources you read, your calendar events — are all able to be read by someone with access to any part of the network between you and Google. This could be your employer at work, the wireless network at your local coffee shop, whatever. This isn’t good.

Here’s an email I just sent myself over the default (unencrypted) connection:

And here’s what I captured via tcpdump.

That’s the whole email there for anyone to see. Luckily it’s easy to fix…

1. Use Bookmarks for Your Google Services Create bookmarks (or modify them if you already have them) for Gmail, Google Calendar, Google Reader, and iGoogle (your Google homepage) using https instead of http, like so: https://mail.google.com/mail/. Do this for every service that you use at Google.
2. Don’t Click on Links Within Google to Take You to Your Services If you use their links Google will often take you to the unencrypted version because it’s easier on their servers. Use your links instead to ensure that your sessions are encrypted

The more we depend on Google (or any other monolithic service) the more we need to safeguard the information they have of ours. One way we can help is by demanding (via secure bookmarks and other methods) that they send us our mail, news feeds, calendars, and other information over a secure connection.:

[ Note: This is not a Google-specific problem. Most other services work in exactly the same way. The difference is that Google is so prolific and is becoming very successfully at getting people to use not only their email service but also their calendaring, news reader, instant messaging, their search (with history), etc. It's the all-in-one dynamic that makes it especially important to protect Google traffic. ]

I wonder if there’s any precedent for using encryption and a multi-location cache as an alternative copyright method. The encryption is like the sealed envelope, and the fact that various, unrelated sources have copies with the same basic date serves as the timestamp.

So you get to court and say,

This was my idea. I published it on March 13th, 2007 and it was progated to the following list of sites. The owners of those sites are here and are willing to testify that the cache times to their sites have not been altered.

Actually, now that I think about it, that won’t matter at all. All someone has to say is that they were developing the idea concurrently. Oh well, it’s still a good way of proving you had an idea at some point in the past via the court of public opinion…

For anyone interested, I just completed a short write-up on the Diffie-Hellman protocol.:

Understanding The Diffie-Hellman Protocol