Conversation

Triss, dishonest and possibly illegal

i was disappointed because i couldnt find any good explanation out there about how post quantum cryptography works, so i wrote my own explanation.

xenia was kind enough to host it on its blog (because i have absolutely no intention of becoming a , so im not starting my own blogwebsite thingy), you can find the article here: https://blog.awoo.systems/posts/2024-10-04-kyber

1
0
0

@sys64738 to be fair, the trucha bug was way funnier than a simple rsa implementation mistake

1
0
0

@sys64738 (broadon's rsa implementation was perfectly fine, the problem is that they compared the hash with strncmp instead of memcmp)

2
0
0

@leo @sys64738 also they didn’t check the padding at all

0
0
0

@leo true. tho it still feels like if everything around it was standardized at once (instead of rsa, pkcs#1 1.5, sha1, asn1, etc all separately), there mightve been fewer footguns. similarly, if ecdsa told you what rng to use (as is the case for kyber/ml-kem), sony might've mplemented it correctly in the ps3 too

1
0
0

@sys64738 @leo the problem is more that all of these individually are deceptively simple on their own and it’s easy to have non-obvious issues with each one of them that completely destroys the security of the entire cryptosystem

1
0
1

Charlotte 🦝 therian

Edited 5 months ago

@sys64738 @leo like rsa requiring constant time bignum arithmetic, [existing code and algorithms for generic bignum math probably has variable time multiplication and exponentiation!], pkcs#1 v1.5 has multiple modes that are simply Invalid for signing (the 3ds gets this wrong), asn.1 is quite complex of a format that needs to be treated as untrusted by default, and several important languages used for writing cryptographic code in have no protection against out of bounds memory access

0
0
1