Hacker News Reader

Comment by transpute | Hacker News Reader

Will the new CPUs implement DRTM (dynamic root of trust, similar to Intel TXT) for Windows 10 SystemGuard? This appears to be planned for Qualcomm Arm CPUs that support Windows 10 and x86 emulation.

https://cloudblogs.microsoft.com/microsoftsecure/2018/04/19/...

> "Windows Defender System Guard runtime attestation, which is built into the core Windows operating system, will soon be delivered in all editions of Windows. Windows Defender System Guard runtime attestation, like Credential Guard, takes advantage of the same hardware-rooted security technologies in virtualization-based security (VBS) to mitigate attacks in software."

View on HN

What's the benefit of hardware-based security tech? Is it actually doing anything special, or just doing what security software is doing, but in hardware?

Software based is very vulnerable to the evil maid attack.

BTW, Google uses hardware-based security on all of their servers:

https://www.theregister.co.uk/2017/01/16/google_reveals_its_...

Ok, so you move from the "evil maid" reinstalling the OS to having to replace the CPU/whole computer. Yeah that definitely looks like an advantage.

Yes but you can detect that. The enclave can sign a statement that you could verify with a public key for that enclave. So if you replace the hardware there is no way to do that anymore as you cannot extract the private key from the original enclave.

It can provide a signing chain from the bootloader upwards to verify that you're running the software you think you are.

The OS could do that as well, yet we see a lot of userspace-only exploits and hacks.

No, the OS cannot do that, because an attacker will replace your OS with a compromised one which lies to you that everything is peachy.

https://blog.invisiblethings.org/2011/09/07/anti-evil-maid.h...

Take for example the secure enclave on an iPhone. If the FaceID / TouchID was implemented in software then you could read it from memory if you compromise the A11 chip. Instead you have to now compromise both the secure enclave and the A11 since it is isolated from the A11.

Why would that be any more than 2x harder?

Because you need two compromises and hacking the secure enclave is a much harder proposition than finding a exploit that allows you to read the memory of an iPhone. Public information about how the secure enclave really works has been hard to come by.

Yes, it absolutely is more than 2x harder. The attack surface of the secure enclave is considerably smaller than the attack surface of the AXX chip as a whole, and you need a significant jailbreak/compromise before you could even target the SE.

The secure enclave would have a much smaller attack surface, due to only handling a much smaller set of features, I guess.

General-purpose processors have to be secure while executing untrusted code, providing a large number of features, and providing good performance.

The secure enclave isn't subject to these constraints, allowing for more conservative design decisions.

You've found a privilege-escalation attack that can let sandboxed apps escape their sandbox? Still secure if the chip can't run apps in the first place. You've found a bug in the USB disk mode emulation code? Still secure if the chip doesn't have any USB code on it. You've found a bug in branch prediction? Still secure if your chip didn't use it. You've found a way to abuse the third party developers' debugging interface? Still secure if your chip provides no such interface...

Reply on news.ycombinator.com