News for hackers. Stuff that matters.

A comment on A fast alternative to the modulo reduction (2016). Parent comment

Re:A fast alternative to the modulo reduction (2016) by rurban on Monday October 07, 2019 @04:09AM (#21178515)
Not fixed by a random seed and not fixed by siphash. Only fixable by not allowing a linear search in the collisions. (counting or promote to a tree)

The random seed can by extracted, exposed or calculated, and then siphash doesn't help you at all, it just makes everything 2x slower.

[ Parent | on HN ]
Re:A fast alternative to the modulo reduction (2016) by baby on Monday October 07, 2019 @12:53PM (#21183067)
How do you extract/calculate The random seed?

This is what all (non-vulnerable) programming languages do btw.

PS: saw your other post, interesting take.

https://news.ycombinator.com/item?id=12401920

I’ll venture a guess: a lot of practical attacks in the lab becomes completely impractical on a network.

[ Parent | on HN ]
Re:A fast alternative to the modulo reduction (2016) by minitech on Wednesday October 09, 2019 @06:03PM (#21208663)
I like how you’ve been saying for 3+ years that it’s possible to recover a SipHash key based on its output, something it’s specifically designed not to allow, but never given any evidence for it.
[ Parent | on HN ]