Linux Group Files Complaint With EU Over SecureBoot

[mjg59]

Are you asserting that an unhackable system is possible?

Other than brute force attacks, sure, why wouldn't it be?

[duby229]

So when you said, you were wrong?

No. If it isnt hacked now, it will be. That goes for anything and everything. Most hacks are not brute force. Most of them use exploits from undocumented behavior.

[mjg59]

No. If it isnt hacked now, it will be. That goes for anything and everything. Most hacks are not brute force. Most of them use exploits from undocumented behavior.

You didn't say that they all would be hacked. You said they already had been. There are many implementations in the wild that have never been hacked even though hackers have been attempting to, so you're just wrong. History does not guarantee that Secure Boot will be fundamentally compromised.

[duby229]

Which brings us full circle again. Like I said we'll see.

[mjg59]

Which brings us full circle again. Like I said we'll see.

No, it doesn't. There's the bit where you said history proved that all of these things would be hacked. History doesn't say that. They haven't all been hacked. You're just wrong.

[Serge]

Other than brute force attacks, sure, why wouldn't it be?

Well, I'm not a security expert, and I realize that you are, so I'm not going to make a fool of myself by contradicting you. I'm just surprised, as this is the first time I've seen an actual expert say that unbreakable security is possible.

[mjg59]

Well, I'm not a security expert, and I realize that you are, so I'm not going to make a fool of myself by contradicting you. I'm just surprised, as this is the first time I've seen an actual expert say that unbreakable security is possible.

Well, in practice any given implementation generally turns out to have flaws (I've certainly found flaws in specific Secure Boot implementations, and I have no doubt that there are others), and it's difficult to prove it in advance, so pointing at a specific machine and saying "This is unbreakable" would be a pretty astonishing thing to say. Having said that, the design principles behind Secure Boot itself have been incredibly heavily reviewed - it's effectively the same Authenticode system that Microsoft use for signing drivers and executables, and nobody's demonstrated an exploit against those yet despite it being one of the most attractive targets. It was easier for the Stuxnet developers to use keys that were physically stolen from hardware companies than it was to break the underlying cryptography...

[duby229]

No, it doesn't. There's the bit where you said history proved that all of these things would be hacked. History doesn't say that. They haven't all been hacked. You're just wrong.

But they will be. And you'll see.History does show that everything unhackable in fact gets hacked. The security industry may be one step ahead, but thats only because something has to be made before it gets hacked. You think that the cryptography needs to be broken for a hack to be successful, but the truth is that is rarely the case. More often then not the cryptography doesnt get broken. An exploit is found for some behavior that wasnt expected.

Anything that hasnt yet been hacked will be.

[duby229]

Well, in practice any given implementation generally turns out to have flaws (I've certainly found flaws in specific Secure Boot implementations, and I have no doubt that there are others), and it's difficult to prove it in advance, so pointing at a specific machine and saying "This is unbreakable" would be a pretty astonishing thing to say. Having said that, the design principles behind Secure Boot itself have been incredibly heavily reviewed - it's effectively the same Authenticode system that Microsoft use for signing drivers and executables, and nobody's demonstrated an exploit against those yet despite it being one of the most attractive targets. It was easier for the Stuxnet developers to use keys that were physically stolen from hardware companies than it was to break the underlying cryptography...

Who's to say that something similar wont happen here? If enough keys become available then then what is stop it?

EDIT: You don't have to break the cryptography. In at least one case that I know of from an ARM device that I have it was the devices own bootloader that was used to break the boot lock. The cryptography was never broken, but it still was effectively unlocked.

[mjg59]

Who's to say that something similar wont happen here? If enough keys become available then then what is stop it?

An effective blacklisting mechanism?

EDIT: You don't have to break the cryptography. In at least one case that I know of from an ARM device that I have it was the devices own bootloader that was used to break the boot lock. The cryptography was never broken, but it still was effectively unlocked.

Sure, and as I keep saying it's likely that specific implementations will fall prey to this. But a generic flaw that affects all machines with Secure Boot? I doubt it.