So is this user specific salt word stored in a table somewhere, how does the company decrypt a salted password otherwise, and so if the salt is also stored somewhere alongside the encrypted password, couldn’t the hacker get his hands on both the salt and the password and use that to figure out the password?
Comment on Plex got hacked.
mic_check_one_two@lemmy.dbzer0.com 2 weeks agoYou can’t reverse the hash, but you can generate hashes until you find a match.
That’s called a rainbow table attack, and that’s why you should salt your hash. This salt basically appends a unique string of characters to every password before it goes into the hash. Let’s say your users are dumb and use “password” for their password. If a hacker has pre-generated a rainbow table, (which is extremely time and resource intensive), then they’ll instantly crack that as soon as they get a match on those common passwords. Even if they haven’t generated a rainbow table, they can just look for identical hashes and guess that those users are using common passwords.
But if you salt it, it’ll slow the hackers down drastically by invalidating their pre-generated table. Each user has their own salt stored alongside the username and hash, so User 1’s hash actually saw “password19,jJ03pa5/-@“ while user 2’s hash saw “passwords)205JrGp02?@-“. Because each of their salts are unique, their resulting hashes are unique too. Even though they used the same password. Now the hackers need to crack the hash for each user, by incorporating the existing salts for each user. They’ll start with the weak and common passwords first, which is why it’s still best practice to use strong passwords. But they can’t actually start the rainbow table process until after they have hacked the info, and they’ll need to create fresh tables for every single user.
BackgrndNoize@lemmy.world 2 weeks ago
WhyJiffie@sh.itjust.works 2 weeks ago
the salt does not need to be encrypted. the point of it is that it makes a generic rainbow table useless, because the crackers need to compute hashes themselves for all passwords.
as they said, the purpose of hashing is to slow down the crackers, because they need to find the string that produces that hash. a rainbow table cancels that, it makes password lookup for an account almost instantaneous. but a rainbow table is only really useful for unsalted hashes, because for salted hashes a different rainbow table is needed that takes the salt into account.
mic_check_one_two@lemmy.dbzer0.com 2 weeks ago
Yes, the salt is stored right alongside the username and hashed password. The point of the salt isn’t to be unknown; It’s to make every single password unique before it gets hashed, which invalidates the hackers’ pre-generated rainbow tables. It forces them to re-generate their table for each user. Even identical passwords will create different hashes, because the salt is different for each user. Essentially requiring the hacker to brute force every single password, even after they have the database downloaded.
Basically, the hash algorithms are known. There are a few common ones, but they’re all reliable. A rainbow table is generated by running potential passwords through each hash, and saving the results. For a simplified example: maybe for a certain hashing algorithm, “password” generates the hash “12345”. I have a pre-generated table with millions of potential passwords that tells me as much. And I have repeated this for all of the most popular hashes. This gigantic database (literally hundreds of GB in size) of millions of potential passwords and resulting hashes for the most popular algorithms is my rainbow table. This took hours of cooking my CPU to generate.
So I hack an unsalted password database, and find a bunch of hashes that are listed as “12345”. I can now guess that they’re probably using that specific hash algorithm, and can immediately crack a bunch of passwords purely because I have already brute-forced them before I hacked anything.
But now let’s say it’s a salted hash instead. When I hack the database, my pre-generated rainbow tables are useless. Because now “password” is not being hashed as “12345”. It’s being hashed as something entirely different, because the salt is added before it gets hashed. Even if multiple users use “password”, it still doesn’t help me because each of their salts is unique. So even if two different users use “password”, they’ll each return different hashes. So I need to recreate my rainbow table for every single user. Even if two users both used “password” I’ll still need to check each one individually, with their unique salts.
This doesn’t completely invalidate the breach, but it drastically slows down my ability to access individual accounts. The goal is simply to slow me down long enough for the company to be able to send out “hey, change your password” notifications, and for the users to do so. Without a salt, once I have the database, I instantly know which hash the company is using. And I can immediately access a bunch of accounts using my pre-generated rainbow table. But with the salt, I’m still forced to crack each user individually.
To be clear, weak passwords will still crack faster. A good password guessing attack doesn’t just brute force. It starts with known lists of common/popular/weak passwords, because that known list of weak passwords will often get you into an account extremely quickly.
Amir@lemmy.ml 2 weeks ago
If the hash is unique per person, hackers need to build a new table per person. It doesn’t matter if the hackers can get their hand on the salt; the point is that they cannot try the common passwords easily for all users; it takes N times as long where N is the number of users with a different salt (hopefully all of them)
ClanOfTheOcho@lemmy.world 2 weeks ago
Best explanation of salt I’ve ever seen. Thanks!
uninvitedguest@lemmy.ca 2 weeks ago
NaCl