The Side Channel

« first day (2742 days earlier) ← previous day next day → last day (1934 days later) »

forest

12:38 AM

@kelalaka The argument I'm having occurred on IRC, not here on SE.

Also

-1

A: How does Twofish avoid weak keys if it uses key-dependent S-boxes?

The key to the answer lies within these two snippets:- These results help confirm our belief that, from a statistical standpoint, the Twofish S-box sets behave largely like a randomly chosen set of permutations. and There should be few or no pairs of keys that define the same S...

Paul answered by Twofish question, but I barely understand his answer.

He's claiming that $q_1$ and $q_0$ are S-boxes themselves...

Then he goes to describe one of the features of the permutations and claimed they are "like good randomness extractors" or akin to a CRC. I am so confused.

And what do preimages have to do with it?

I'm not sure that I really understand this answer. How are those permutations S-boxes, and how are they similar to CRC in any way? Why would you need to find preimages of a simple permutation just to create weak keys? Why do you quote the requirement that individual bit changes modify each S-box? What is the relevance of related key attacks here? The only thing I can take away from your answer is that it's hard to individually specify exact S-box bits, which is something I already mentioned in my question. — forest 1 min ago

Ella Rose

1:26 AM

Is this "permutation" meaning "bit transposition", or "permutation" as in "the output set is a permutation of the input set"?

Ella Rose

1:49 AM

I guess a better question is, how are q_1 and q_0 not s-boxes

forest

2:23 AM

@EllaRose I suppose they could be considered S-boxes being permutations.

They're just never referred to as such in literature discussing Twofish.

What gets me more is the comparison to a CRC than to an S-box.

Ella Rose

2:34 AM

I guess they're not the s-boxes

forest

3:07 AM

Is H(k || s) bad form in a KDF? Should H(H(k) || s) always be used instead?

forest

3:58 AM

nvm

10 hours later…

b degnan

1:51 PM

@kelalaka I hope to get around updating that LSFR answer this week. Terrible things here regarding my free time to do art with 1/2 of my collaborators on furlough.

@forest people still use IRC? :P

2 hours later…