I was afraid reading this: This $299 tool is reportedly capable of decrypting BitLocker, PGP, and TrueCrypt disks in real-time. It seems to me that PGP and other sensible tools have to lock memory to ensure they where not be swapped, nor hibernated.

Seems to me that programmers know this, from as long time swap and hibernation files exists!

So I don't know the target operating system (as I like to work with efficient tools, I've never work with this. But I know his poor reputation about core system).

My question: Could that be true!? Or is this a hoax?

If true, this give me another argument to tell some to migrate, but as everybody know I'm paranoid, nobody would believe or understand the importance of this.

To be more explicit: I'm an old computer actor, so I know a lot of things but even, while I do some sensible installation, like crypted disks or messages, everything is locked by default (cannot be swapped and will be unmounted before hibernation). If I try to build sensible thing in a less secured manner, I'll be warned.

Working for people who don't know anything about computer must imply some responsibilities. So I understand, reading this, that a huge amount of people, working with confidence on his registered OS (they should pay dearly) are much less protected than they think!

Well known as a paranoid, I won't present this to my customers without your precious feed-back!

3 Answers 3


That sensational title is, unsurprisingly, missing a qualification: “decrypting BitLocker, PGP, and TrueCrypt disks in real-time” on misconfigured systems.

If you have encrypted data on a system, and you want that data to be protected against the theft of the device while it is hibernating, then you must encrypt the hibernation file (or disable hibernation altogether, which isn't very nice). If Bitlocker and the others on don't insist on encrypting the hibernation file, this is a bug in that tool.

(All the Linux distributions I've seen do this right, by the way: they always encrypt the swap space (which contains the hibernation image) when you select whole-disk or home directory encryption at installation time. It's possible to install without encrypting anything, and later set up encryption for some files while retaining cleartext swap, which as we've seen is insecure if the attacker may get hold of the hibernation image — so if you set up file encryption, also make sure to encrypt the swap space.)

Encryption software should also try to arrange not to be swapped out, though this is a less reliable vector of attack unless the attacker can both cause software to run on the machine (causing other programs to be swapped out) and get hold of the disk image, which is a rare combination.

  • The article seem promising that every host powered by this os is vulnerable. I know how to configure Linux, that's ok, but did you say that it is possible to configure this (poorly reputated) os to prevent hibernation or swaping for sensible datas? Jan 7, 2013 at 3:06
  • @F.Hauri AFAIK Truecrypt full-disk encryption is fine (though this article seems to say the contrary, but it is a bit vague and based partly on ad copy). I don't know about Bitlocker, it should be ok at least if the encryption key is in a TPM. Jan 7, 2013 at 10:20

As @Lucas says in his answer, and as it says in the article the OP linked to:

So, how does it work? Elcomsoft Forensic Disk Decryptor acquires the necessary decryption keys by analyzing memory dumps and/or hibernation files obtained from the target PC. You’ll thus need to get a memory dump from a running PC (locked or unlocked) with encrypted volumes mounted, via a standard forensic product or via a FireWire attack. Alternatively, decryption keys can also be derived from hibernation files if a target PC is turned off.

This is not new - this has always been possible, and in fact is often used by forensics teams where access to the running PC is available.

To be safe, you don't let others have access to your computer when it is on, (or off - see Evil Maid attack), ensure you don't leak information by leaving hibernation files, memory dumps or other unencrypted data around and generally follow the usual security guidance.


It's true and it's also very logical. If you want to encrypt something like a disk, then you need your encryption key to be stored in your memory. This is the same as what happens with for instance, single signon.

Now think about it, if you hibernate your computer, which has full disk encryption, where is the hibernation file stored? On your encrypted hard drive of course.

So you would need either a memory dump or access to the hibernation files. (you are at risk if they are contained in the hibernation file when you are not using full disk encryption)

  • On Linux, it's possible to store symmetric key outside of RAM and inside the CPU debug registers. Kernel patch is available at www1.informatik.uni-erlangen.de/tresor
    – Matrix
    Jan 5, 2013 at 8:42
  • Yes but it does not seem to be standard. Jan 5, 2013 at 10:16
  • On my linux, PGP allocated memory won't be swapped and pre-hiberation script force umount all crypted filesystems. Mar 16, 2013 at 6:00

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .