Hacker News Reader

Comment by Sami_Lehtinen | Hacker News Reader

Since when has SHA1 been crackable in three days? AFAIK with proper passwords SHA1 is nearly impossible to crack. Did I get something wrong? Salting does make the process slower with bad passwords, but with good ones there's very little difference. Just my gut feeling.

View on HN

There's nothing to stop anyone from creating a reverse lookup table by systematically generating strings and their hash, and aside from dictionary words, this is an example of low hanging fruit. It's a repeated sequence of characters, which would be the second most obvious heuristic (after the dictionary) when trying to generate plausible passwords and their associated SHA1 hash.

Before the LinkedIn leak, I was guilty of sharing passwords across sites, and I got a wake up call by a "someone tried resetting your password" email. Started resetting them all, got to the point where I noticed there were a lot and I should probably use a password manager. Eventually I found out I had over 100 online accounts for various things, all of them now having unique passwords. I've deleted a handful of them (e.g. MySpace, MyCokeRewards). There are a couple more I'm trying to get deleted (e.g. Dominos) but haven't been successful yet.

I thought salting was to stop precisely this sort of attack, no?

From what I understand, salting prevents computing the tables ahead of time and the ability to cross-reference a single password hash against all other password hashes. It doesn't prevent people who have access to both the salt and the hash from performing the same attack against individual passwords.

Yes, that sounds correct. I thought the idea, though, was to make the salt non-obvious for this reason. I would hope people aren't storing passwords in a table with a "salt" column, but I don't really know. But I think you're right: if you know the salts and you want to use this approach to go after a small number of high-value targets it could work.

It seems like it's easy to find collisions from the hash.

https://www.schneier.com/blog/archives/2005/02/sha1_broken.h...

But in case of passwords, the input masking (limited character set, length etc) makes most of found collisions nonviable? Because can't just freely enter colliding bits for the algorithm.

Yes, here is more info on SHA1 being broken.

https://crypto.stackexchange.com/questions/3690/no-sha-1-col...

Reply on news.ycombinator.com