<![CDATA[A bit of a weird longshot, but does anyone have any resources on reducing the viable search space of inputs to a polynomial rolling hash where the multiplier is known, and the modulus is just over a 4-byte word?]]>A bit of a weird longshot, but does anyone have any resources on reducing the viable search space of inputs to a polynomial rolling hash where the multiplier is known, and the modulus is just over a 4-byte word?

I’ve given a couple of thoughts to cracking this but I keep hitting hard dead-ends to my knowledge.

Also I must state this isn’t for any actual security breaching, (you shouldn’t be using this kind of hash for anything remotely secure), this is mostly a pet project with a minor application in reverse engineering a few hashed strings in an older game with no reference in the code to the original strings

]]>https://board.circlewithadot.net/topic/3595ffce-3fd3-4d7b-a77e-0f4c2beb2cdc/a-bit-of-a-weird-longshot-but-does-anyone-have-any-resources-on-reducing-the-viable-search-space-of-inputs-to-a-polynomial-rolling-hash-where-the-multiplier-is-known-and-the-modulus-is-just-over-a-4-byte-wordRSS for NodeMon, 25 May 2026 14:07:30 GMTFri, 01 May 2026 18:54:29 GMT60<![CDATA[Reply to A bit of a weird longshot, but does anyone have any resources on reducing the viable search space of inputs to a polynomial rolling hash where the multiplier is known, and the modulus is just over a 4-byte word? on Fri, 01 May 2026 19:07:00 GMT]]>@MrL314 I think you're kinda stuck since this seems to just be a standard discrete logarithm problem

that said even with a pretty meagre GPU you could probably just brute-force the entire space in a few minutes

]]>https://board.circlewithadot.net/post/https://toot.cat/users/clarfonthey/statuses/116500884372096223https://board.circlewithadot.net/post/https://toot.cat/users/clarfonthey/statuses/116500884372096223Fri, 01 May 2026 19:07:00 GMT<![CDATA[Reply to A bit of a weird longshot, but does anyone have any resources on reducing the viable search space of inputs to a polynomial rolling hash where the multiplier is known, and the modulus is just over a 4-byte word? on Fri, 01 May 2026 18:54:29 GMT]]>For those more curious, the rolling part of the hash is:

h_i = s_i + h_i-1 * 31
over an int32

My current approach was to try to find a lattice basis for
KNOWN_HASH = sum(0<=k<n, s_i * 31^(n-1 - k)) mod 2^32 where n = length of string s
and then run some brutes over that search space since it’ll likely be significantly smaller than just regular bruting

However I’m both very interested in the math, and I don’t want to chug my computer on the problem more than I need to

]]>https://board.circlewithadot.net/post/https://peoplemaking.games/users/MrL314/statuses/116500835177688694https://board.circlewithadot.net/post/https://peoplemaking.games/users/MrL314/statuses/116500835177688694Fri, 01 May 2026 18:54:29 GMT