[messaging] Encrypted Pulic Contact Discovery

[Justin King-Lacroix]

In fairness, you never did need special hardware to verify an RA; that was
half the point.

SGX also lets you attest application software, which is usually what people
care about on the open Internet. Frankly, I think that's its biggest
selling point.

(TXT is great if you're a datacenter infrastructure operator, where it's
basically a really nice pseudo-secure-boot measure that also gives you nice
guarantees for your encrypted hard disk, but it's mostly useless for
non-systems tasks.)

I'd be interested to know if the group sig scheme is the same, or
substantially similar to the, one as used in Direct Anonymous Attestation.


On 26 August 2015 at 13:14, Mike Hearn <mike at plan99.net> wrote:

> Yeah, TPM / TXT based RA has always been a half-baked joke. I am not
> surprised Intel ran out of patience with the wider ecosystem and just
> decided to do it all themselves. SGX looks a lot easier to program and much
> more likely to actually work than prior attempts. The fact that it solves
> memory bus attacks and is fully on-die is the icing on the cake.
>
> Interestingly, SGX chips identify themselves using a group signature
> scheme. Thus they can prove they are issued by Intel without providing any
> kind of unique identifier. It's called EPID:
>
> http://csrc.nist.gov/groups/ST/PEC2011/presentations2011/brickell.pdf
>
> So in theory you won't need any special hardware to verify an RA, just to
> generate one.
>
> I am guessing the earliest that ordinary Joe's like me will get to play
> with SGX is next year at best. Right now it seems to be in some sort of
> private testing period.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://moderncrypto.org/mail-archive/messaging/attachments/20150826/9c379dc3/attachment.html>