News for hackers. Stuff that matters.
Re:How to defeat Ed25519 and EdDSA using faults by Lery on Friday October 06, 2017 @04:22AM (#15415210)
You don't really want to verify the signature you've just generated, if you care about perf, because verification is twice as costly as signing.
Re:How to defeat Ed25519 and EdDSA using faults by loup-vaillant on Friday October 06, 2017 @04:38AM (#15415295)
I know, it would slow down the whole thing by a factor of 3.

Monocypher can "only" produce 7000 signatures per seconds on a single thread on my laptop (i5 Skylake Intel). With the verification, I would waste a whooping 0.3ms on each signature, reducing the throughput to 2000 signatures per seconds.

Unless you're running a signing intensive application on the scale of Let's Encrypt, you won't really care about the slow-down. If you don't fully trust your hardware however, faults might be a real problem.

Re:How to defeat Ed25519 and EdDSA using faults by rightos on Friday October 06, 2017 @07:57AM (#15416052)
Even let's encrypt is issuing only a handful of certificates per second. Perhaps it'd matter more if used on a per connection basis.
Re:How to defeat Ed25519 and EdDSA using faults by tptacek on Friday October 06, 2017 @10:19AM (#15416960)
1. Nobody is attacking Lets Encrypt with fault attacks.

2. If you can induce faults on Lets Encrypt's signing infrastructure you have better attacks to deploy.

3. Lets Encrypt doesn't use Elliptic Curve in the first place.

Re:How to defeat Ed25519 and EdDSA using faults by loup-vaillant on Friday October 06, 2017 @10:24AM (#15416998)
Of course, I was only speculating about the scale required to make several thousands of signature per second. Your second point stands, though. It's probably a waste of CPU to mitigate fault attacks in a Lets Encrypt like use case.
Re:How to defeat Ed25519 and EdDSA using faults by rightos on Saturday October 07, 2017 @02:39AM (#15422583)
I wasn't suggesting any of those things weren't true, merely a back-of-the-napkin estimate of the sort of load a function like this would be sufficient for.
Re:How to defeat Ed25519 and EdDSA using faults by wolf550e on Friday October 06, 2017 @01:21PM (#15418489)
Let's Encrypt use HSMs to sign, and more than 90% of signatures produced are for OCSP, not for issuance. Though I have failed to find the link for this, I saw them publish these numbers.
Re:How to defeat Ed25519 and EdDSA using faults by marcosdumay on Friday October 06, 2017 @10:03AM (#15416847)
Yet, if you then proceed to send your signed document over a network, for it to be rejected 10 minutes later because of a bad signature and then you have to start from the beginning again, that verification cost is irrelevant.
Re:How to defeat Ed25519 and EdDSA using faults by loup-vaillant on Friday October 06, 2017 @10:26AM (#15417004)
How frequent do you expect failures to be? (That would attack induced failures). There's a threshold below which the "10 minutes later" cost less than wasting 3ms every single time.
Re:How to defeat Ed25519 and EdDSA using faults by AstralStorm on Friday October 06, 2017 @12:15PM (#15417946)
It is much more likely that the fault is due to corruption during data transfer anyway and you get to handle this case.