Slashdot Mirror
Dual Boot Not Trusted, Rejected By Vista SP1
Alsee writes "Welcome to our first real taste of Trusted Computing: With Vista Enterprise and Vista Ultimate, Service Pack 1 refuses to install on dual boot systems. Trusted Computing is one of the many things that got cut from Vista, but traces of it remain in BitLocker, and that is the problem. The Service Pack patch to your system will invalidate your Trust chain if you are not running the Microsoft-approved Microsoft-trusted boot loader, or if you make other similar unapproved modifications to your system.
The Trust chip (the TPM) will then refuse to give you your key to unlock your own hard drive. If you are not running BitLocker then a workaround is available: Switch back to Microsoft's Vista-only boot mode, install the Service Pack, then reapply your dual boot loader. If you are running BitLocker, or if Microsoft resumes implementing Trusted Computing, then you are S.O.L."
It's possible to use the Vista bootloader to chainload GRUB rather than the other way around (which is the default for most Linux installs.)
Yes, it's a pain to set up, but so is any dual-boot setup.
I've upped my standards, so up yours.
I have Vista SP1 installed on a machine that uses GRUB to dual boot into Kubuntu, so it appears to work fine on systems without a TPM.
And no TPM in the laptop.
That's the whole point of the problem, TPM has begun causing issues. You don't have TPM, so you are not affected.
For large sets, this will be our guide even unto death, for the LORD will work for each type of data it is applied to...
no, you just have to have a version of Vista that supports BitLocker, whether it is on or off. Enterpise and ultimate are the only versions that support BL, so they are the ones that need the KB which is prerequisite to SP1 install (because SP1 upgrades some bitlocker features). Never Trust Trustworthy computing. it hasn't earned it.
This *may* be a corner case as most TPM's were shipped in the disabled state back when XP was still shipping.
Instead, how about testing the open source BIOS stack? Most of you have an unused box of recent vintage and I'm sure the projects can use the feedback.
FYI: An open sourced bios is an Achilles heel for Microsoft. Mobo OEM's will **jump** on a Free bios because it saves them money and elminating TPM saves them much more money.
Get involved!!
http://www.coreboot.org/Welcome_to_coreboot
http://openbios.info/Welcome_to_OpenBIOS
http://www.maxineudall.com/2010/02/should-economists-be-sued-for-malpractice.html
Terrorizing the poor multi billion $ business of M$
I thought it was: Shit Out of Luck
which is not in your list.
Obama's legacy: (N)othing (S)ecure (A)nywhere and (T)error (S)imulation (A)dministration
I have Vista Enterprise on a dual boot laptop with a TPM that I have never enabled. Installing SP1 did nothing adverse to the dual boot capability.
I won't use it. I just bought a laptop on Ebay, brand new, out of box, that came with the Home edition, great bargain at $421. First thing I did with it was actually start it up and say "No" on the AUP acceptance page. I immediately powered it off, put in my trust Ubuntu Hardy 64-bit install cd, wiped the disk, and installed a real operating system that will stay the fuck out of my way.
Sorry, Microsoft, but I'd call this Epic Fail. Trusted computing causes me to lose control of *my* computer. Problem is, Microsoft don't understand the definition of computer ownership.
Linux with ntfs-3g has been supporting full read/write on ntfs for some time, and works out of the box on my ubuntu hardy machine anyways.
"You know, Hobbes, some days even my lucky rocketship underpants don't help" -- Calvin
Actually most linux distros can read/write ntfs now
Two words: filesystem support.
Boot up Linux and all the stuff on your NTFS partition is read-only.
What? You know, Linux has had full NTFS Read/Write support for a while now, see :
http://www.linux-ntfs.org/
Also, ever heard about WUBI ?
jdb2
Not to mention it's fairly easy to get Windows to read ext2/3 partitions with the extfs driver.
Informative gives Karma but Funny doesn't. Therefore, people who appreciate the post and wish to give the user some karma will choose Informative.
Beware : the new Intel ICH10R has an integrated TPM.
Intel Macs use EFI instead of a BIOS, and EFI uses GUID Partition Tables (GPT) instead of MBR.
The space that the MBR used to sit in is reserved in GPT, so when a legacy system reads, uses, or modifies the partition table, it only changes the old MBR partition table, which is not actually used to boot. In contrast, Boot Camp's dual-boot features only use the GPT, which means that as far as Vista knows, it IS the only boot loader involved.
Not at all....
Booting is handled by the EFI, and any operating system booted under the legacy BIOS emulation wouldn't be able to do a thing about it!
Quicken's cock-up was that it was writing to parts of the MBR that DOS/Windows didn't use - but GRUB/LILO did. In this case, it would do the same thing, since it's unlikely that Vista has changed how such things work.
Microsoft's choice to 'intentionally not read Mac floppy disks' likely involves not having support for MFS/HFS, and not seeing any real need to reverse-engineer them to implement them.
The bug would be in the enforcement of the check when it does not apply, not in the very existence of it.
Do you agree that a full disk encryption product needs to protect the data from unauthorized access in every way possible?
If you agree to the above, do you assert that despite that, it should allow access to the data when the environment is verifiably NOT what it expects?
I'm not suggesting that the Windows boot loader is infallible (far from it), but it seems like you are suggesting that the FDE solution should continue on its merry way when it has detected an obvious deviation from the environment that it was designed to work in? We make sacrifices in usability and performance when we want to ensure that our data is safe. This disabling would obviously be purposeful. However, what I am saying is that if it is triggered when it does not apply (when FDE isn't enabled, for example), *THAT* is a bug.
Its only in Vista Enterprise or Vista Ultimate, which support disk encryption.
Intron: the portion of DNA which expresses nothing useful.
I'm running Vista Ultimate 64bit with GRUB for Ubuntu, but BitLocker is turned off. No problems here thankfully.
> Would not TrueCrypt be the better option?
It depends on what you need. This is an old and true as hell slogan - security as strong as the system's weakest element.
So for example it does not matter if you use the bestests the strongest the most sexy cryptographic algorithms for your Truecrypt installation if it is easy to get your keys from memory using other ways.
Such way would be for example *booting* the system into tiny supervisor.
This is fairly new concept of attack but it is possible as hell. All new VT technologies introduced sometime ago are now finding their way into consumer systems. Security researched warned about this since ca. 2003.
Now that MS is trying to think ahead and protect from such attacks it is Bad. But if they wouldn't it would also be Bad.
Does TrueCrypt enforce a chain of trust down to the hardware?
I believe it does. You can load any OS you want or put the disk in another machine and still not be able to decrypt the "hidden" partition, even if you know of its existence.
You misunderstood the question. TPM and full disk encryption, used in this way, ensures that every piece of software from the bootloader on up is either considered trusted or not. It starts this chain of trust in the hardware, which is considered much harder to trojan than software (like the bootloader or OS.)
Put another way, TPM conceivably protects you from software keyloggers by verifying the signature of the bootloader, the OS loader, and the OS itself before allowing you to decrypt your data. If anything in the chain has been modified, it won't release the keys, thus protecting your data. Unless Truecrypt interfaces with TPM, merely knowing the key is enough to decrypt the data, regardless of the computer that you put the disk in. Truecrypt adds a layer of deniability, but that's not the same thing.
No, they do. I think a lot of people here misunderstand what TPM is meant to actually do and what it's supposed to be good for; and what it is useless for. (Frankly, I'm not sure Microsoft fully understood.)
It's because the MBR has *changed* that means the chain isn't signed with something that will allow the system state register to authenticate with the TPM key storage; the register contents will have changed because the SHA-1 fingerprints changed, so you're not going to be able to get a coherent response from the TPM regarding any keys you've stored in it if you've taken ownership already. Without resetting the token and destroying the keys, that is.
You want another way of doing this? Don't take ownership of the TPM to store the keys, but put 'em on a thumbdrive and use a secure passphrase (10 word Diceware, for example) to unlock them; this is also a supported mode of operation under BitLocker (assuming you trust the Elephant diffuser as being part of a reasonable cipher mode; frankly, I'm not that happy with it and prefer OCB or XTS modes, or failing that Linux's aes-cbc-essiv:sha256)... doing it the "thumbdrive way" is highly recommended when a TPM isn't available or wanted. Putting the hard disk encryption keys in the TPM isn't necessarily a good idea; they are recoverable given some effort, and that's not really what the TPM tech is for.
This is all entirely by design; it's closing an actual security hole whereby a trojaned MBR could capture your encryption keys. Obviously this is unsuitable for any dual-booting setup. TPM just isn't designed to work with that kind of scenario; it's really more of a system for verifying extremely stable system images such as you might find on a server or tightly-controlled corporate workstation that you want to be able to have a reasonable degree of confidence hasn't had the MBR tampered with because it's a trusted client that handles classified data (and any tampering with the software whatsoever would decertify it).
You control the chain of trust when you take ownership of the TPM; they do work just fine with Linux, and Linux does have support for them - if you want to know and prove to another system that the bootloader, BIOS, and kernel haven't changed since the state you knew was good, you can do that (although the proof is only as good as the integrity of the TPM).
They're just hardware tokens coupled with a signed BIOS/bootloader/kernel, really. Handling the actual key management that results from that, or what you do with it, is entirely up to you.
Vista using the TPM for BitLocker is hardly plug-and-play, and quite unsuitable for many scenarios (many TPMs out there don't even support TCG1.2); there's always TrueCrypt or PGP Whole Disk Encryption or one of the many other solutions available if you want a little more flexibility and control.
In particular, it's not really about DRM. None of the DRM systems proposed or deployed have ever used it, or are likely to ever use any part of it, as a key storage blackbox, because an entirely homogeneous image just isn't something you can guarantee on any consumer box (that's one reason it's not even on or in the vast majority of OEM and consumer motherboards/chips). It's perhaps a bit more practical for laptops...
Also, TPM implementations are quite breakable where the attacker has physical access and ownership of the machine and plenty of time. PCs aren't even consoles, and look what we've done to those...
It's meant to be one interlocking part of a whole enterprise security solution. It sure as heck isn't a "magic crypto chip" that will lock up your PC, and it shares none of the common criteria with DRM scenarios (which are, of course, just as doomed if they use a hardware blackbox as if they use a software blackbox, because the plaintext is always available...). In fact, having a TPM around if you're running Linux, will at least make sure you always have a secure entropy source for /dev/random...
Informative has the benefit of generating a "Why is this informative!" post, which leads to people replying "Informative gives Karma but Funny doesn't. Therefore, people who appreciate the post and wish to give the user some karma will choose Informative." and getting rated Informative, which generates Karma itself.
It's kind of a huge karma circleje-..dependency.
Too right, I just modded it informative too, and your post as well, so your ka... oh wait. whoops.
What if Tetris was invented by Nazis?
Informative gives Karma but Funny doesn't. Therefore, people who appreciate the post and wish to give the user some karma will choose Informative.
What I don't understand is why anyone would care... Slashdot Karma is competing with Kool-Aid Fun Points for score that has the least impact on my life.
The enemies of Democracy are
This should definitely be modded Informative.
I'm so excited I just made water in my pantaloons!
I think the point is just to be nice :)
" Thou shalt devote thyself wholeheartedly to evil or to the good".
It's because what people are really saying is +1 satire./P
Oh, well heh, I think modding someone funny for being funny is nice enough for a little o' that real life karma. :)
The enemies of Democracy are
So "informative" is the new "funny"?
Damn!
Because you don't have a TPM chip, I'd guess.
Oh no! You guys started an infinite Karma loop!
Modern copyright is theft of culture from everyone and it retards the progress of the useful arts and sciences.
I have Vista Enterprise in a dual-boot laptop with TPM and grub as the primary boot loader, and SP1 installed without any problems at all, and never altered the boot loader. It's 64-bit Vista, which is typically even more stringent with the code checks than 32-bit.
Were Microsoft not attaching it to a KB article, I'd have called it FUD, but I will say that I have not experienced it at all.
You can never go home again... but I guess you can shop there.
http://port25.technet.com/archive/2006/10/13/Using-Vista_2700_s-Boot-Manager-to-Boot-Linux-and-Dual-Booting-with-BitLocker-Protection-with-TPM-Support.aspx
Why is this rated informative?
The scenario in question is a stolen laptop. Adequate physical security? Are you kidding me?
Starbucks, Harbuckle of Breath.
Uh. Mods are now definitely literally on crack. Not behaving in an incomprehensible and unpredictable manner, they are putting the pipe to their lips and inhaling the smoke from burning crack cocaine.
Assume I was drunk when I posted this.
Respectively... No. Yes. No. Maybe.
Game! - Where the stick is mightier than the sword!
How many Vista Enterprise or Ultimate users really dual boot? Since this article is dated four months ago and this is the first we're hearing about it, I'm guessing not many.
We need a new category called Infunmative....
Dear sir, if you find my posts funny, please mark them funny so that I know you got the joke and don't think you got confused and took me serious.
I don't give a fuck about karma. Anyone willing to make the effort can have theirs pegged at the cap if they wanted, anyway. (Karma whores don't deserve it, and those that don't care about karma and just post things that are interesting and informative are always at the cap anyway.)
The cake is a pie
I have a dual-boot setup with Ubuntu 8.04 and Vista Ultimate. Linux was loaded first then Vista with the bootloader replaced with EasyBCD v1.72 from NeoSmart. Service Pack 1 installed w/o any problems at all.
I multi-boot with several 64-bit Linux distros and 64-bit Ultimate Vista on a Dell Vostro 400 I bought back in February (does this have the TPM stuff?). Grub is installed on the MBR and I don't have BitLocker enabled in Vista (why would I - can't read the disks in Linux if I did!). I installed Vista SP1 when it came out and had absolutely no problems (I may have had to re-install GRUB on the MBR, but I do that so often that I consider it no big deal). So am I the odd one out?
First, note that Iam the story submitter.
Second, and more important, note that I am a programmer and have I read the Trusted Platform Module technical specification from cover to cover. The 332 page technical spec.
The goal is to allow you to trust that your computer has not been compromised by a third party
Demonstrably incorrect. That is NOT the fundamental design criteria of the Trust chip.
You could get all of that functionality from a virtually identical design that did not secure the computer AGAINST the owner. If you are up for the technical details, you could for example have an identical chip with identical capabilities, except that you permit the owner to get a printed copy of his PrivEK when he buys the system. That alone would be minimally sufficient to grant the owner ultimate control of his system, but for technical reasons the chip should also have the capability to export the RootStorageKey encrypted to the PrivEK, as this makes things massively simpler benefiting security.
I forget the page number, but at one point somewhere in the latter half, the technical spec EXPLICITLY refers to the the owner as an "attacker". The specification explicitly details the measures that must be taken to secure the system AGAINST THE OWNER.
AGAINST
THE
OWNER.
Q.E.D. The fact that the technical specification for the chip repeatedly places the HIGHEST PRIORITY of forbidding the owner to ever obtain his own key (which would provide him ultimate control of his own computer) demonstrates that in fact the purpose of the design is to secure the computer against the owner. As the grandparent put it:
Trusted computing is all about allowing vendors like microsoft to trust the computer to work in thier partners interests rather than the users.
Of course, if you pour concrete over my house and take other insane measures to lock me out of my own home, yeah.... that does also incidentally have the effect of keeping other people out of my home too. The point here is that the owner is denied the key to his own house. Trying to advertise that as a security system securing the home FOR the owner is obviously a comically bogus argument.
-
- - You can't take something off the Internet! That's like trying to take pee out of a swimming pool.
I believe the whole point of the TPM chip is that it performs the checking before we gain control.
If the checks pass, the TPM key is then "available" for that boot.
If the checks fail, the TPM key is locked away.
I wholeheartedly agree with the "why would you want to run Vista" comment though!
For me, the only reason is PC gaming, but manufacturer support is currently still good for XP, and the DX9 vs DX10 difference is small.
Come DX11, things may change, but that's ages away.
People who appreciate the post and don't really understand the rating system, that is. The correct way to deal with this is to rate the post "Underrated". This gives the poster karma without hanging any new (and inappropriate) tag on the post.
Uh. Mods are now definitely literally on crack. Not behaving in an incomprehensible and unpredictable manner, they are putting the pipe to their lips and inhaling the smoke from burning crack cocaine.
Name a better way to spend a Thursday morning with mod points in your account!
"intentional way of censoring what somebody considers a really sensitive topic"
I've also suspected this is possible a number of times. Companies like Sony, for example, have been shown up for using such tactics as Gorilla Marketing, to get their message across and employing bloggers to appear to be independent reviewers, when in fact they are working as part of an organized PR campaign So its well within the concepts of Gorilla Marketing style behavior to work to manipulate popular forum discussions. I wouldn't be at all surprised if many big companies and even some governments could be playing these same disinformation style games. Its interesting how manipulations to the Wikipedia have been detected and proven to be occurring. Forum style discussions need some way to detect organized disinformation/manipulation campaigns, but that's not going to be so easy to detect, but over time, at least more people are becoming aware of these disinformation games.
There are 10 kinds of people in the world... those who understand binary and those who don't.
The nasty thing isn't that Funny doesn't give karma, it's that Overrated and other downmods still take away karma on a post marked as Funny. So, if a post gets moderated up to +5 Funny, then gets two Overrated downmods, the poster loses karma overall. Over time, this can eat away at someone's karma, especially if they're writing a lot of humorous posts that don't go down well with everyone.
Is EFI planned to replace BIOS in the non-mac world?
If you ask Intel, yes. If you ask the rest of the world, meh. I don't think anyone would argue that BIOS should stay, it's a crusty old POS that's been hacked on top of hacks over the years to keep supporting new things, but what should replace it is very debatable.
Can Linux bootloaders and whatnot play nicely with EFI?
http://sourceforge.net/projects/elilo/
http://refit.sourceforge.net/
The former is a Linux-focused bootloader for all EFI platforms, rEFIt is a generic loader built with Intel Macs in mind. I have no idea if it can run on other EFI platforms.
Heck, can Windows?
Yes and no. Windows has had an EFI loader for a few years now, as it's required for Itanium. That was finally brought to normal processors with Server 2008 and Vista SP1, x64 only. So if you're 32 bit or running anything but the latest versions of Windows, you're stuck with the BIOS.
If so, can one even BUY a motherboard that uses EFI? As I'm planning to build a system on which I can (hopefully) run both windows and linux, I'd like to try to avoid the whole MBR shenanigans.
It seems MSI is shipping a MB they call "EFINITY" and a few OEMs supposedly have started using EFI on their custom boards, but in the non-Mac x86 world it's still pretty rare.
I used to get high on life, but I developed a tolerance. Now I need something stronger.
I'm failing to see why this is a big deal. Software is in place to check for a piece of third party code intercepting your encryption key... It successfully detects GRUB as such software, and stops. So what?
This is a flaw of the trusted computing architecture. If the partition of the trusted OS (Vista) is encrypted, Multiboot does not break trust, because the other OS cannot decrypt the partition. But in trusted computing, if an untrusted bootloader loads a trusted OS the chain of trust is broken.
If trusted computing were designed with the user's interest in mind, the user would be able to decide that the bootloader he is using (grub) is trusted, sign it with a key which enables that bootloader only on his computer, and get on with his life. But now we have to wait for Microsoft to implement and sign a real bootloader... good luck with that.