Fast-key erasure RNG using AES-NI

[vks]

I ported jedisct1/aes-stream to Rust: vks/aesrng. Would this be of interest to this repository?

[newpavlov]

Not in this repo, but it can be a good addition to CSRNGs repo. BTW it will be nice if you'll implement traits from rand_core crate. Can this algorithm be written generically over block cipher, so you could use aesni crate, without losing performance?

[newpavlov]

Also can you provide any specifications of the implemented algorithm? We would like to avoid inclusion of home brew algorithms.

@tarcieri
What do you think?

[vks]

it will be nice if you'll implement traits from rand_core crate.

Already done on my local branch, I just need to update it to use the recently published crate.

Also can you provide any specifications of the implemented algorithm?

The algorithm was suggested on djb's blog. I think it is implemented for various ciphers in SUPERCOP (see fastrandombytes). This gives it some legitimacy, but IMHO it still has to be considered homebrewn.

Can this algorithm be written generically over block cipher, so you could use aesni crate, without losing performance?

Maybe, but it is currently written using very low-level SIMD, so it is probably a lot of work to do this.

[tarcieri]

@newpavlov it's an interesting algorithm that provides a nicely fast CSPRNG wherever hardware AES is available. I'd be in favor of including it

[newpavlov]

@vks
Can you please create a PR to CSRNGs repository?

[vks]

I will look into it, but not before August 13.

…

On Mon, Aug 6, 2018, 19:28 Artyom Pavlov ***@***.***> wrote: @vks <https://github.com/vks> Can you please create a PR to CSRNGs <https://github.com/RustCrypto/CSRNGs> repository? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#1 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AACCtDN0fMsj4DsVi3LWXI3V2iPE1gzDks5uOHy9gaJpZM4TAJ0d> .

[vks]

Closed in favor of RustCrypto/CSRNGs#2.