Better-performing “25519” elliptic-curve cryptography(amazon.science) |
Better-performing “25519” elliptic-curve cryptography(amazon.science) |
https://github.com/firedancer-io/firedancer/pull/716
Ditto for sha256: https://github.com/firedancer-io/firedancer/pull/778
And sha512: https://github.com/firedancer-io/firedancer/pull/760
If you’re an optimization nerd, this codebase is wild.
Not that you are technically wrong, not at all, that's where Jump came from. It's just that this is all completely blockchain-driven optimization, but the b-word is so dirty now that we've gotta go back to using TradFi for the rep.
If the Amazon improvements are hacker news worthy (they are) this seems reasonable contextually.
Also, I worked for Jump for almost 12 years :)
What a wasteful and unproductive enterprise, considering the vast majority of the devised improvements never see the public eye.
Still, impressive work. Imagine if those brilliant minds behind this were focused somewhere else.
- make people click on ads
- make trading algos faster
- replace human artists
- build more efficient killing machines
- destroy any remaining concept of privacy
It might not seem like real work, but making money by reducing costs of market participants sounds like a good thing. I admit though, block trades might be harder now than before the rise of HFT.
If you could do warehousing/distributing/coordinating fresh foods in a way that reduced the difference in price between the farmer and the consumer and make money doing it, that would clearly be good work.
nb those abstractions do make sense when you can only afford to write a single implementation of the algorithm; then you're just talking about a high level programming language. But they frequently fail to achieve their goal when you're writing a second implementation for the sole purpose of being faster.
/* All the functions in this file are considered "secure", specifically:
- Constant time in the input, i.e. the input can be a secret[2]
- Small and auditable code base, incl. simple types
- Either, no local variables = no need to clear them before exit (most functions)
- Or, only static allocation + clear local variable before exit (fd_ed25519_scalar_mul_base_const_time)
- Clear registers via FD_FN_SENSITIVE[3]
- C safety
*/
[1] https://github.com/firedancer-io/firedancer/blob/main/src/ba...
[2] https://en.wikipedia.org/wiki/Elliptic_curve_point_multiplic...
[3] https://eprint.iacr.org/2023/1713
[4] https://libsodium.gitbook.io/doc/internals#security-first
[5] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/lin...
They follow a first principles approach (the lead has a few physics degrees) and opted to speed up the cryptography. The beauty of this, despite the bad views on blockchain, is that they freaking sped up the cryptography of commonly used algorithms more than anything open or closed source that I personally am aware of.
It’s a win in cryptography, much like this Amazon post is, except it’s slower than the firedancer implementation.
Really though? This mostly-untrue statement is the line that warrants adding hashtag #post-quantum-cryptography to the blogpost?
/?q X25519Kyber768Draft00: https://www.google.com/search?q=X25519Kyber768Draft00
If this further improvement becomes widely used, it would be interesting to see if it's enough to tip the scales towards ed25519 being more of the de facto "default" ssh key algorithm. My experience is that a decent number of people still use RSA keys most of the time, but I don't feel like I have nearly enough of a sample size to conclude anything significant from that.
That's pretty sweet. I'm currently using BoringSSL in a project as a supplement to OpenSSL (mostly because it is much easier to build for Windows users than requiring them to fiddle with msys2/vcpkg etc; the alternative is to rely on the Windows CNG API, but it lacks features like ed25519 support.) I wonder how much effort it would take to use aws-lc instead... Not that I'm that interested, BSSL is pretty good, but free performance and heavy automated verification is always nice :)
Related: one of the authors of this post, John Harrison, wrote a really good book about automated theorm proving about 15 years ago while working on floating point verification at Intel -- there's still no other book quite like this one, I think https://www.cl.cam.ac.uk/~jrh13/
Turns out someone else has already tried: https://github.com/aws/aws-lc/issues/1827
My immediate fear was that they optimized away the security features like absence of timing side channels, but they say they still have those.
They also claim to have formal proof of correctness, which is even more amazing, because they are not doing it on a symbolic level but on a machine instruction level. Apparently they tought their reasoning system the semantics of all the CPU instructions used in the assembler implementation.
I'll still wait what djb has to say about this, but it looks freaking amazing to me.
Is this implementation resistant to that?
If it isn't, it's kinda a footgun which shouldn't be published for general use.
That doesn't mean that this implementation doesn't have timing attacks, but the implementors claim they chose mechanisms which should be constant-time.
I mean, when implemented naively, yes, but the industry has been aware of timing attacks for decades such that this is table stakes for any crypto implementations.
From the article:
> We also do our best to execute the algorithms in constant time, to thwart side-channel attacks that infer secret information from the durations of computations.
https://github.com/awslabs/s2n-bignum (where most of the heavy lifting is done, per the article) further explicitly states that "Each function is moreover written in a constant-time style to avoid timing side-channels."
> Our implementations of x/Ed25519 are designed with constant time in mind. They perform exactly the same sequence of basic CPU instructions regardless of the input values, and they avoid any CPU instructions that might have data-dependent timing.
When I see CVE-fests like — https://people.redhat.com/~hkario/marvin/ — … I just do not come away with that impression.
[Widely used] Cryptographic Rust crates offering "constant time" operations in "pure Rust" — but Rust has no primitives for doing constant time operations, so it's only through hopes and prayers that it might actually work, and with no guarantee anywhere that it actually should.
(Other, less timing attack related stuff, but e.g., major companies still not supporting anything beyond RSA.)
I wouldn't be surprised if a lot of people still use RSA for SSH keys for one or more of the following reasons:
1. A lot of tutorials about generating SSH Keys were written before ed25519, so if they follow an old tutorial they'll probably be generating an RSA key.
2. Older versions of OpenSSH, that you'd find on CentOS 7 and below, would default to RSA if you didn't specify a key type when running ssh-keygen.
3. There are some systems out there that don't support ed25519, though they are becoming rarer. If you have to deal with those systems then you're forced to use RSA (at least for that system).
4. Some of us have been using SSH keys from way before OpenSSH add support for ed25519 keys in 2014, so any long lived SSH keys won't be ed25519 keys (wow, ed25519 has now been about in OpenSSH for over 10 years).
I'm in my twenties and still have that reaction. I know elliptic curves exist, I even sort-of-kind-of have an awareness of how they work, but if I was asked to name one cryptosystem that used public and private keys, I'd definitely say RSA first and not elliptic curves.
Azure Devops is a big one.
I just remember the flag as being vaguely similar the name of the monster robot from RoboCop.
That’s not really anecdotal. Generating an ed25519 key is barely more than generating a random 256-bit value. Generating an RSA key is significantly more work.
My also naive (an possibly out of date) understanding is key generation is much faster in with ecc, and that signing is faster too, but verifying is faster for rsa. So switching from a RSA to an ECC server certificate saves bytes on the wire, because keys are smaller, and saves server cpu because signing is faster, but may increase client cpu because verification is slower. The byte savings may make up for the increase in cpu though.
Interesting! I wonder if this new algorithm is intended to help with that. I'm super curious if the smaller payload does indeed make a difference (with the current algorithm) like you mention; I know that with databases and filesystems, compression is commonly used to shift the balance from I/O to CPU due to disk writes being slow (with reduced storage size being a side benefit but not usually the main motivation), but I also know that cryptographic verification being too slow can be an anti-feature if it makes brute forcing feasible, so the amount of CPU work needed might be pretty high still.
* https://www.amazon.science/blog/formal-verification-makes-rs...
RSA signature verification is already very fast and TLS doesn't use RSA for encryption anymore so the problem reduces to optimizing signing operations.
$ openssl speed rsa ecdsa
sign verify sign/s verify/s
rsa 1024 bits 0.000117s 0.000008s 8518.7 132449.2
rsa 2048 bits 0.000884s 0.000025s 1130.6 39499.3
sign verify sign/s verify/s
256 bits ecdsa (nistp256) 0.0000s 0.0001s 33210.9 11483.0
384 bits ecdsa (nistp384) 0.0009s 0.0008s 1070.6 1268.9
Not even when FIPS 140-3 was (finally) finalized in 2019, and testing began in 2020?
https://csrc.nist.gov/projects/cryptographic-module-validati... includes mentions of EdDSA, and Curve25519 is listed among the "Recommended Curves for U.S. Federal Government Use" on page 15 of https://nvlpubs.nist.gov/nistpubs/SpecialPublications/NIST.S....
(I guess the problem is that various crypto implementations need to get recertified under the new standard...)
edit: it looks like AWS-LC [0] and boringcrypto [1] have both been validated under FIPS 140-3. Azure's OpenSSL crypto [2] has only been validated under FIPS 140-2 as far as I can tell.
[0] https://csrc.nist.gov/projects/cryptographic-module-validati...
[1] https://csrc.nist.gov/projects/cryptographic-module-validati...
[2] https://csrc.nist.gov/projects/cryptographic-module-validati...
Not to mention it also suggests there is a way to "compare" minds. I would not choose myself to do somethings, but that does not mean I despise automatically people choosing to.
cryptopals.com also doesn't cover any elliptive curve crypto until you get into the last set.
As of OpenSSH 9.5 the default has changed, so you don't have to specify anything:
* ssh-keygen(1): generate Ed25519 keys by default. Ed25519 public keys
are very convenient due to their small size. Ed25519 keys are
specified in RFC 8709 and OpenSSH has supported them since version 6.5
(January 2014).
$ ssh-keygen -t <TAB>
dsa ecdsa ecdsa-sk ed25519 ed25519-sk rsa
People have to forever be on guard that you might at any point pivot to all taxation is theft or how you have formed your own micro nation that consists entirely of yourself and thus have diplomatic immunity from all prosecution. Because it happens. Or maybe you have a once in a lifetime deal to buy this receipt like object for some hideous art that is guaranteed to appreciate in value millions of percent. It's just the crowd that has aggregated around crypto currencies includes a lot of untrustworthy people.
Granted, there are all kinds of whackos in crypto, but we should only be concerned about the immoral ones trying to scam us out of our money: SBF, Do-Kwon, and the like.
If you're math-avoidant, stick with high-level abstractions like NaCL and TLS. There's nothing wrong with that!
A professor talking about and demonstrating cryptography at the level of individual algorithms is doing their class a disservice if they say "none of the math will be on the test". The algorithms are enough to put something together that seems like it works; the math is what you need to find out if your resulting system actually does work. It's where many of the fun bug classes live.
As another example, the Amazon implementation also doesn't refer to gcc's "strub" attribute which zeroises the function's stack upon return or exception of functions working on crypto secrets.[4][5] OpenSSL doesn't either.[3] firedancer _does_ use the "strub" attribute to allow gcc to zeroise the function's stack.[9]
Is there a performance impact? [6] has the overhead at 0% for X25519 for implementing CPU register and stack zeroisation. Compiling the Linux kernel with "CONFIG_ZERO_CALL_USED_REGS=1" for x64_64 (impacting all kernel functions) was found to result in a 1-1.5% performance penalty.[7][8]
[1] https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attribute...
[2] https://clang.llvm.org/docs/AttributeReference.html#zero-cal...
[3] https://github.com/openssl/openssl/discussions/24321
[4] https://gcc.gnu.org/onlinedocs/gcc-14.2.0/gcc/Common-Type-At...
[5] https://gcc.gnu.org/onlinedocs/gcc/Stack-Scrubbing.html
[6] https://eprint.iacr.org/2023/1713.pdf
[7] https://www.phoronix.com/review/zero-used-regs/5
[8] https://lore.kernel.org/lkml/20210505191804.4015873-1-keesco...
[9] FD_FN_UNSANITIZED: https://github.com/firedancer-io/firedancer/blob/master/src/...
> Someday there will probably be a TLS1.4/2.0 with PQ, and also FIPS-140 -4?
> Are there additional ways to implement NIST PQ finalist algos with openssl?
- open-quantum-safe/oqs-provider [implements mlkem512 through mlkem1024 and x25519_mlkem768]
Kanav left, they liquidated a huge staked ETH position a few months ago (+ a bunch of other coins), and the SEC/CFTC is all over them for the Terra Luna fiasco.
Folks have said jump is gonna die for 20+ years. They’ve been around 30ish…
For users that have AVX-512, which isn't widely available (AMD Zen 4 / Zen 5, Sapphire Rapids)...
We should celebrate Amazon's improvements and we should celebrate these improvements. Both are great for the future of technology, regardless of why they were initially developed. Improving tech and keeping it open source is good for all.
You can make the determination on your own, but there will be an obvious large showing of firedancer folks and some exciting updates for the project.
That said, I'd argue that for the vast majority of software engineers the type of stuff they're dealing with can be dealt with without needing to know the math. For example, you don't need to understand the math to behind the algorithms to know that bcrypt is a reasonable password hashing algorithm and that sha1 and md5 are not, or that salts are used to mitigate issues when users reuse passwords. These are principles that you can understand at a high level without fully understanding the underlying details. If anything, I think that overemphasis on requiring people to learn and understand the math has the effect of over-focusing on simpler algorithms that aren't actually what people want to be using in practice due to the fact that they're easier to teach and often foundational in conveying concepts that would need to be learned to understand the more complicated algorithms.
If using cryptographic algorithms directly requires knowing the math, then I'd agree that most people shouldn't be using them directly, but I'd go further and say that a lack of libraries that are safe for people to use for software engineering without understanding the implementation is a failing of the ecosystem; as much as "regular" software engineering people (like myself!) can struggle with the math behind cryptography, I think that a lot of people developing cryptographic libraries struggle with building reasonable abstractions and making user-friendly APIs (which is a skill I think in general is not emphasized enough for most software engineers, to the detriment of everyone).
My thing here is just: learn the math! Or do something else. I did! There is so much to do in our industry.
I'm not sure I understand what you mean here by "something else in our industry". Are you arguing that I'm not qualified to be a software engineer due to not understanding the math behind elliptic curves, or did you miss my repeated use of phrases like "the vast majority of software engineers" rather than some specialty where cryptography implementation details details are more important? If the latter, I can reassure you that I don't work in cryptography, work on any cryptographic libraries, or have any specific responsibilities related to security beyond the general idea that all software being written should be secure. If the former, I'll have to respectfully disagree, and suggest that maybe even if you aren't willing to consider that you're wrong about the math being a hard requirement for someone being qualified as a software engineer, it's worth considering that you almost certainly don't have enough information to conclude whether a stranger on the internet is qualified based on reading some of their comments.