Differences

This shows you the differences between two versions of the page.

--- pgpfan:rsabad [2022/05/18 15:38] – Ambiguous b.walzer
+++ pgpfan:rsabad [2024/07/31 20:12] (current) – Provided a bit of context. b.walzer
@@ Line 35: / Line 35: @@
 >// Instead, developers are encouraged to choose a large d such that Chinese remainder theorem techniques can be used to speed up decryption. However, this approach’s complexity increases the probability of subtle implementation errors, which [[https://www.cs.tau.ac.il/~tromer/courses/infosec11/Boneh%20DeMillo%20Lipton%201997%20---%20On%20the%20importance%20of%20eliminating%20errors%20in%20cryptographic%20protocols.pdf|can lead to key recovery]].//
-The linked article doesn't describe any sort of implementation error. Instead it describes a completely theoretical hardware attack.
+The linked article doesn't describe any sort of implementation error. Instead it describes an attack based on hardware faults((This originally used the term "theoretical" to describe the attack. See the more recent [[https://eprint.iacr.org/2023/1711.pdf|Passive SSH Key Compromise via Lattices]], which shows that this sort of weakness existed at the rate of one per million historical SSH records. Almost all of these came from a single version of Zyxel SSH server. See section 4 of the paper.)).
+Almost all RSA implementations check generated signatures to prevent, say, a cosmic ray induced bit flip, from creating the possibility of a secret key leak. So the "implementation error" here would probably be a failure to do such a check. Very few people would consider a hardware failure to be any sort of a software implementation error. Fewer people would feel that the lack of a check for a hardware fault is an implementation error.
 >//Public Exponent//