My Review of Samsung 840 EVO SSD Full Drive Encryption.


  1. Posts : 5
    Windows 7 Home Premium 64bit
       #1

    My Review of Samsung 840 EVO SSD Full Drive Encryption.


    A review of the Self Encrypting Drive feature of the Samsung 840 Evo.

    Industry leaders, including Samsung, are using deceptive and fraudulent language to sell “hardware based” “self encrypting” SSD's to people who want to protect their data with hardware based encryption. They are making claims that are untrue or misleading in order to peddle hardware to trusting customers.

    Self Encrypting SSD's based on TCG-Opal are a big fat lie.
    TCG-Opal is a scam.

    Encrypts on the fly!
    AES 256!
    Secure Erase!
    Self Encrypting Drive!
    Hardware based encryption!
    No impact on performance!

    That is what I read on many supposedly reputable, hardware review websites about the data encryption capabilities of the Samsung 840 Evo. So I bought one. But it isn't true. Don't believe the hype.

    The sites I visted had a bunch of buzz words and techie sounding slogans that mean zippo. All of the tech reviewers who reviewed the drive and stated in one way or another that the user's data on the drive would be safely encrypted using flashy sounding techno-babble should all turn in their self-awarded tech expert badges and stop writing. That includes the writers and editors from 'Tom's Hardware', 'AnandTech', YouTube tech “experts” and many, many others. Shame on you all.

    Samsung finally released their firmware update (9 December 2013)for the 840 Evo which makes the drive TCG-Opal compliant. TCG, 'The Trusted Computing Group' developed a standard for self encrypting drives like the 840 Evo. TCG appears to be an unholy trust of tech industry giants who can't seem to figure out how to actually deliver secure data encryption to the public.

    What is TCG-Opal compliant? Why did I believe I would be able to securely store my data on my now TCG-Opal compliant Samsung SSD? The Trusted Computing group said I would be able too if I have a TCG-Opal compliant drive.

    From the Trusted Computing Group website:

    https://www.trustedcomputinggroup.or...rypting_drives

    "Commonly Asked Questions and Answers on Self-encrypting Drives"

    How is the access to the drive secured to allow only the Authorized user to access it? Is there a boot- up password that is entered via a BIOS dialog?

    A: When the BIOS requests the Master Boot Record from the drive, the drive instead returns the pre-boot record to the user. This pre-boot record is a complete, though quite restricted OS, usually something simple like MS-DOS or LINUX. The pre-boot image requests the Authentication Credentials from the user, which are passed to and checked directly by the drive logic. If accepted, then the drive returns the MBR and the OS is loaded. Important point: This pre-boot authentication is the FIRST thing that happens and is controlled by the drive directly. This has the added advantages of not modifying the MBR, which many software encryption products do, and allowing the MBR to be encrypted like all other user accessible data.
    Controlled by the drive "directly". Not by the OS or a program installed on the OS. The password is independent of the BIOS. Great. That's how it should work. That's not how it really works though. But they say it does. Strange.

    “Many software encryption products” would include Truecypt which I have used for a couple years now. The idea then is that instead of the BIOS fetching a mini-OS from Truecrypt from the MBR on the HDD which allows me to enter a password which authenticates me and allows the drive to unlock the OS to load from my software encrypted drive....the mini-OS will load from a chip on the SSD and do the same process. Sounds good right? That's not what happens. But they say it does. Hmmm.

    Also from the TCG website:
    Trusted Computing Group - Solutions - Data Protection

    Stronger Security
    Encryption always on and transparent to the user - major compliance requirement
    Keys for encryption are generated in the drive and never leave the drive
    User authentication is performed by the drive before it will unlock, independent of the operating system
    So, controlled by the drive "directly". Great. Not by the OS or a program installed on the OS. Excellent. The password is independent of the BIOS. Just right. The way Truecrypt works. They state that vendors provide software for enterprise deployment, something not relevant to single users who buy a drive which they intend to use for their own data. Like I did.

    But it turns out TCG-Opal SED-SSD's do not really have "hardware based" encryption. The industry appears to be dancing about flailing weasel words that sound very profound but upon closer examination are revealed as cheap prevarications.

    Hardware based encryption. Like a chip that encrypts the data and a chip that runs firmware that requires the user to enter a password to unlock the drive and access the data. A hardware/firmware combination would qualify as "hardware based". I would buy that. The lie is that the actual interface is based on installed SOFTWARE which makes the encryption chain...SOFTWARE based.

    TCG says, on their website that unlocking the drive on a TCG-Opal compliant drive is independent of both the BIOS and OS yet Samsung has produced a TCG-Opal compliant drive that does not offer a BIOS or OS or Software independent means of using a strong password for unlocking the drive. What they provide is a method for using the BIOS for a weak password that is dependent on the BIOS or a possibly stronger password method based on third party software that, naturally, runs on a software OS.

    The new Samsung 840 Evo firmware from 9 December 2013 updated my drive to be TCG-Compliant yet the Samsung Magician software states that:







    Samsung is lying
    Trusted Computing Group is lying.
    The industry is lying.

    The weasels who write the reviews (you know who you are) repeated the lies. Does that make them liars too?
    Well, the lazy tech writers didn't exactly lie. They merely read press releases, probably got some free hardware then regurgitated the language of the press releases to their readers without any critical examination of it. Disgraceful yes. Not deliberate lying though. They're just lazy tools. Cut and Pasters with a bucket full of benchmarking software.

    The Samsung Magician Software provides three possible levels of encryption:

    Level 0 breaks TCG-Opal compliance. It is dependent on BIOS. It is also no improvement because it offers no better protection than one would get from a drive without the SED feature. It is ATA-BIOS locking/unlocking. Weak. And that only if your BIOS supports it which it probably doesn't.

    Level 1 breaks TCG-Opal compliance. It is dependent on software therefore it is software based, not hardware based.

    Level 2 breaks TCG-Opal compliance. It is OS dependent and also software based since it is based on bundled software included with the OS.

    Levels 1 and 2 are software based meaning the installed software whether 3rd party or something bundled into the OS provides a weak link between user and hardware.

    All three levels also break TCG-Opal compliance because they all depend on software, hardware, firmware, BIOS cooperation that may or may not exist on most users systems.

    Is the Samsung 840 Evo really a SED SSD? Is it really 'hardware based' encryption? Encryption in chips is hardware based but a user must have a way to interact with the hardware to unlock the encryption to access their data. Using firmware (Free DOS or Linux on a chip) for this purpose should still qualify the encryption chain from hardware to user interface as 'hardware based'. I would call that 'hardware based' encryption. Requiring third party software to unlock the data nullifies the hardware based description though and it really then becomes software based encryption.

    Think of it this way. Truecrypt is software but it uses hardware (HDD, CPU, RAM, IO Channels) for encryption, decryption, locking, unlocking data and user interface. Truecypt then could just as accurately be called “hardware based” encryption as the “hardware encryption” of SED SSD's from manufacturers like Samsung based on the apparently malleable TCG-Opal “standard” which requires installed software to function even though they claim it doesn't.

    I wrote of “industry leaders”. Which ones? Go to TCG's website. They have a long list right there of TCG members. Promoters and contributors. Samsung is actually listed as an “advisor” to TCG. Really? Yes really:

    Trusted Computing Group - TCG Members

    The take away here is this. If you are person who intends to buy a self encrypting, solid state drive (SED SSD) because you believe your data will be safer than it would be on a non-encrypted drive...don't buy the self-encrypting drive. You will be wasting your money.

    If you are someone who already uses “software” based encryption like Truecrypt but you want better performance from 'hardware based' encryption, stick with what you already have if the encryption part is important to you. You will lose the data protection if you switch to “hardware based” encryption.

    If you want a really fast SSD but don't much care about protecting your data then buy an 840 Evo. The sucker is fast but the encryption is a lame joke.

    If you want to gift a SSD to someone or sell it and you want to be sure your data is erased then the secure erase function in the Magician software applied to the Evo will do what it says it does. Your data will not be retrievable by anyone. I'm pretty sure of that.

    If you are a compliance manager in a huge IT department then you're probably not reading this because it isn't relevant to what you do. But if you are reading this and you are looking at the enterprise level functionality of a third party software system for encrypting employee laptops for instance using Evo's or something similar you need to triple check everything you think you know about it and grill the vendor again for real because if your job depends on the data being secure after you sign off on it you are in deep trouble. The whole system is flaky as hell.

    Lastly, If you are concerned there may be a “back door” built into SSD, hardware based encryption you can stop worrying. Near as I can tell there isn't even a front door.

    Mongo
    Last edited by MongoFongo; 12 Dec 2013 at 20:54.
      My Computer


  2. Posts : 1
    Windows 8.1
       #2

    Cheers for taking the time and effort to actually investigate this. One thing that leaves me confused is when you say..

    "If you want to gift a SSD to someone or sell it and you want to be sure your data is erased then the secure erase function in the Magician software applied to the Evo will do what it says it does. Your data will not be retrievable by anyone. I'm pretty sure of that."

    My understanding (which may be flawed) is that the way the erasure tool works is the same way the drive works by setting a BIOS/drive password.. which is that when you use the samsung erase tool it wipes the encryption key off the drive, therefor making the data useless. Setting the BIOS password is the sort of the same thing, it encrypts the key with the BIOS / HDD ATA password, and if that password is secure enough then its the same as wiping the encryption key. Otherwise, how can you be so sure that the odds are better of you not being able to recover that encryption key over you being able to decrypt the encrypted key (with a decent password)?
      My Computer


  3. Posts : 1
    Windows 7 Professional 64bit
       #3

    Hi,
    How can anybody goes for security in non open source yet? After all those Snowden's news.
      My Computer


  4. Posts : 1
    Windows 7/8/8.1/2008R2/2012
       #4

    MongoFongo's review is FUD


    While I appreciate MongoFongo taking the time to write a review on the encryption capabilities of this drive, there is so much FUD associated with it that it is a disservice to the security community. I felt I needed to comment on this. I don't work for Samsung and have no vested interest in this other than doing my part to stop the FUD.

    With regard to the following statements made by MongoFongo (in particular):

    "The take away here is this. If you are person who intends to buy a self encrypting, solid state drive (SED SSD) because you believe your data will be safer than it would be on a non-encrypted drive...don't buy the self-encrypting drive. You will be wasting your money."

    This Samsung drive is ALWAYS AES-256 encrypted. The only bit of importance is whether or not you have enabled a password that controls access to the decryption key in secure storage on the drive. By default there is no password assigned, and the device automatically obtains the key and decrypts.

    The ATA Security (BIOS) feature, what is referenced here as Class 0, is what allows the password to be submitted to the drive by the user. I suppose you can argue semantics over whether or not it is "software", but regardless it's pretty hard to get data from the keyboard to the hard drive without it.

    If you want to prove to yourself that this drive indeed provides AES-256 self-encryption capabilities, then all you have to do is inspect the hex of the raw disk. Pretty easy for any security professional to do.

    Your data would be safer by using this drive and enabling the password. I challenge you to find any security expert that would disagree with this statement.
      My Computer


  5. Posts : 6
    Windows 7 Home Premium 32bit
       #5

    Searching for "interface" description, but got misinformation


    I was just looking for a description what kind of interface was provided to enter the password, and I thank the MongoFongo for that description.

    But in fact I agree with securityguy that the rest, and complaints by MongoFongo really miss the point of "hardware encryption".

    OK it would certainly be nice if this was a completely OS independent operation, but that really has nothing to do with hardware encryption.

    Hardware encryption implies that the encrypting and decrypting is being done by the hardware.
    The fact that you have to run a separate program to provide the password to the hardware is not relevant to that question. It does mean that you have to have that program written for each operating system you use, so that you can use the drive, but that is about it.

    If I have a display card that says it can draw a circle in hardware they mean that you send the hardware a command with the needed parameters and the hardware draws the circle.

    If on the other hand you are drawing the circle in "software" that implies that the program calculates out where each dot in the circle should go, and then tells the display hardware to put in a point at each of these spots.

    The program/CPU has to do all that extra work and it has to send many "draw dots" commands instead of one "draw circle" command.

    In the case of this drive and encryption, the software part is doing only one thing, and that is getting a password from the user and sending it to the hardware as a command. Once done, it is not part of the process.

    And there are a few more commands like "delete encryption key", which just tells the drive to remove that key, which makes all the data encrypted by it irretrievable.
      My Computer


  6. Posts : 13
    Windows 7 Pro
       #6

    I am trying to understand in detail how SSDs that claim to do FDE (Full Disk Encryption) / SED (Self-Encrypting Drive) actually work, since I am very concerned about security. I am particularly concerned with the Samsung 840 EVO, since that is a drive I would like to buy.

    I made this long posting on Arstechnica several days ago; read it if you want to see where I am coming from. I cited this post in my Arstechnica one, and am now commenting here, since I have some specific questions that people on this post seem to be informed about.

    securityguy said:
    This Samsung drive is ALWAYS AES-256 encrypted. The only bit of importance is whether or not you have enabled a password that controls access to the decryption key in secure storage on the drive. By default there is no password assigned, and the device automatically obtains the key and decrypts.
    Right, that is my understanding too.


    securityguy said:
    The ATA Security (BIOS) feature, what is referenced here as Class 0, is what allows the password to be submitted to the drive by the user.
    This is one of my most pressing questions: how do you guard access to the crypto keys on the FDE/SED drive?

    Here is my current understanding:

    1) One way to guard the crypto keys appears to be the old school ATA password that every ATA drive (conventional or SSD, independent of encryption) is supposed to support. On non-encrypted drives, the ATA password merely does drive locking, which is known to be almost useless (maybe only good for protecting your drive against the crackhead who stole your laptop, but not against a semi-intelligent adversary like law enforcement). On encrypted drives, apparently, the drive's firmware instead (or additionally) uses the ATA password as a key to encrypt/decrypt the drive's actual crypto keys. In MongoFongo's original post, his screen shot of Samsung Magician Class 0 is this security mode.

    2) another way to guard the crypto keys appears to be that the drive has a special operating system that gets booted and prompts the user for a password that is used to access the drive's actual crypto keys. In MongoFongo's original post, he cited this TCG FAQ webpage, and his screen shot of Samsung Magician TCG Opal is this security mode.

    3) In MongoFongo's original post, his final screen shot of Samsung Magician TCG Opal is some "Encrypted Drive" option. I assume that that is the Encrypted Hard Drive feature in Windows 8?

    Is my above understanding correct?

    Assuming it is, the next point that I want to make is that technique 1) (ATA password) appears to be highly defective. This is because ATA drives usually ship from the factory in a wrongly named "high security mode" in which a second (besides the User) password also exists. This other password is called the Master password, is factory preset, and it backdoors the device. Read this account for proof. Yes, in theory you can put the drive into "maximum security mode" which disables the Master password. (For Samsung drives, Samsung Magician presumably has an option for this? In the previous link, the guy used Linux's hdparm command on an Intel drive.) But in that account link, look at the comment by Anonymous in which he claims that the BIOS (for Dell at least) seemingly stores ATA passwords too, which may be very insecure. Anyone have comments about this?

    So, I conclude that technique 2) (boot a mini OS that prompts for password) is probably the only safe approach. It should mostly bypass the BIOS (other than the BIOS triggering the mini OS load). It should be totally operating system independent (very important to me; I hate Microsoft on many levels, and will probably use Linux in the future). How do you set this mode, however? Presumably with Samsung Magician for their drives, if you happen to be using Windows. Otherwise, you are SOL?

    I would love to get informed feedback on all this. securityguy, if you are listening...
      My Computer


 

  Related Discussions
Our Sites
Site Links
About Us
Windows 7 Forums is an independent web site and has not been authorized, sponsored, or otherwise approved by Microsoft Corporation. "Windows 7" and related materials are trademarks of Microsoft Corp.

© Designer Media Ltd
All times are GMT -5. The time now is 19:52.
Find Us