Am I right or am I right?
-
I have been fretting over the question of hacked userids/passwords. As I understand it, the stolen password is not actually the password but a hashed version of it which the hacker then decodes by brute force and ignorance to arrive at the actual password. But what if that wasn't the actual password. What if the system I was using performed some sort of transformation before it hashed the password for storage. For example, when I log on to my account, I type in pass123 as my password, but unbeknown to me, the system translates that to 321ssap, hashes it and stores it. The hacker tries to log onto my userid and types in 321ssap as my password which is what he thinks it is. This gets translated to 123pass and hashed for checking. But that hash value is not the same as the one that is stored. Therefore an "Incorrect password" error message is generated. Am I missing something here?
xiecsuk wrote:
I type in pass123 as my password, but unbeknown to me, the system translates that to 321ssap, hashes it and stores it
This is very close. Actually, what you are thinking of is a salted hash. In other words, the user types in the password, the receiving application receives the password over some secure method (like HTTPS) then it 1. takes the original password 2. adds a secret salt value 3. hashes the entire thing (pass + salt) Since a hash is one-way, no one can really decrypt them -- that we know of. Same Is Same But look, every time you hash "ABCD" with the same algorithm, you get the same hash. That means a nefarious character can create tables of hashes (rainbow tables) of common words. Then, just compare the hash they have to the hashes stored in the datbase. Proper Salt Probably Prevents However, if the original value were properly salted, this would protect against that problem, unless the nefarious character should learn what the salt is. which is usually not the problem. Usually the problem is that the passwords are stored in the database in clear text or simply not salted.
-
Correct. To horse with passwords, you need a battery of scripts, to handle the myriad staple functions that protect them.
I wanna be a eunuchs developer! Pass me a bread knife!
Do we get extra points when understanding the reference here ? Or minus points for those who don't ?
-
xiecsuk wrote:
I type in pass123 as my password, but unbeknown to me, the system translates that to 321ssap, hashes it and stores it
This is very close. Actually, what you are thinking of is a salted hash. In other words, the user types in the password, the receiving application receives the password over some secure method (like HTTPS) then it 1. takes the original password 2. adds a secret salt value 3. hashes the entire thing (pass + salt) Since a hash is one-way, no one can really decrypt them -- that we know of. Same Is Same But look, every time you hash "ABCD" with the same algorithm, you get the same hash. That means a nefarious character can create tables of hashes (rainbow tables) of common words. Then, just compare the hash they have to the hashes stored in the datbase. Proper Salt Probably Prevents However, if the original value were properly salted, this would protect against that problem, unless the nefarious character should learn what the salt is. which is usually not the problem. Usually the problem is that the passwords are stored in the database in clear text or simply not salted.
What he suggests is that the password is transformed again before being hashed, using an algorithm proper to the website. I still fail to see why this is not clever (and it is probably not, otherwise they would all do it. Or maybe they all do it already and nobody knows ?).
-
xiecsuk wrote:
Am I missing something here?
Quite a lot, because it's pretty complicated. Firstly, you are assuming they've hashed the password, some (even large companies) don't they use a symmetric algorithm. If a company can send your password to you, say via e-mail, they are using a symmetric algorithm (at best).This is insecure. The idea is the hashing is one way, so the hash cannot be reversed, so password123 ---> 7FDEADBEEF or whatever. It the password table is stolen there are two immediate vulnerabilities. First off, if the system allows a bad password such as "password" this is going to be the most common value stored in the password field across the table, and you can work backwards through the most probable ones. The second vulnerability is something called a rainbow table, this reduces the amount of time it takes to reverse engineer a password that is going to lead to a particular hash. Worse, these tables are readily available, so you don't need the compuation time to get started. You can salt the password (adding a random bit of text which you store) in various ways, e.g. add then hash or has the password, salt then hash the result. This mitigates against the vunerabilities described above, but given enough time brute force strategies will always win - even if enough time means from now until after the end of the universe.
KeithBarrow.net[^] - It might not be very good, but at least it is free!
-
What he suggests is that the password is transformed again before being hashed, using an algorithm proper to the website. I still fail to see why this is not clever (and it is probably not, otherwise they would all do it. Or maybe they all do it already and nobody knows ?).
-
What he suggests is that the password is transformed again before being hashed, using an algorithm proper to the website. I still fail to see why this is not clever (and it is probably not, otherwise they would all do it. Or maybe they all do it already and nobody knows ?).
Because it wouldn't prevent the commonest way to bypass hashes: look for identical values. Even if you reverse every password before you hash it, every user that uses "password" gets the same hash value. Salting uses different values for each user - the username or row id value for example - and so every user gets a different hash value even if they share a password.
Bad command or file name. Bad, bad command! Sit! Stay! Staaaay...
-
What he suggests is that the password is transformed again before being hashed, using an algorithm proper to the website. I still fail to see why this is not clever (and it is probably not, otherwise they would all do it. Or maybe they all do it already and nobody knows ?).
"Clever" is never secure, because the security vanish the moment someone figure it out. And if attacker could get their hands on the DB then they also could get the application (possibly even source code). So as other said slow(!) salted hashes are secure because it doesn't matter that you know how to create the hash. The way back is expensive. If the "secret way" would be enough then everyone will just XOR the password with something "secret", right?
-- "My software never has bugs. It just develops random features."
-
Do we get extra points when understanding the reference here ? Or minus points for those who don't ?
Minus points for those that don't read XKCD regularly. ;)
-
What he suggests is that the password is transformed again before being hashed, using an algorithm proper to the website. I still fail to see why this is not clever (and it is probably not, otherwise they would all do it. Or maybe they all do it already and nobody knows ?).
Rage wrote:
I still fail to see why this is not clever
Depending on the function, it could be just as susceptible to rainbow attacks as not applying the function. Also, given the function, it might still be statistically breakable (e.g. "password" most common) which is even quicker if applicable. The reverse example he gives has both these flaws, especially given the apparent lack of salt. As Eddie Vluggen says, about as clever as prefixing with a constant- it'll increase the time taken to break, but not by much. [Edit] My understanding was the OP was trying to get at how a hashes work generally, rather than a specific implementation :~
KeithBarrow.net[^] - It might not be very good, but at least it is free!
-
Correct. To horse with passwords, you need a battery of scripts, to handle the myriad staple functions that protect them.
I wanna be a eunuchs developer! Pass me a bread knife!
-
I have been fretting over the question of hacked userids/passwords. As I understand it, the stolen password is not actually the password but a hashed version of it which the hacker then decodes by brute force and ignorance to arrive at the actual password. But what if that wasn't the actual password. What if the system I was using performed some sort of transformation before it hashed the password for storage. For example, when I log on to my account, I type in pass123 as my password, but unbeknown to me, the system translates that to 321ssap, hashes it and stores it. The hacker tries to log onto my userid and types in 321ssap as my password which is what he thinks it is. This gets translated to 123pass and hashed for checking. But that hash value is not the same as the one that is stored. Therefore an "Incorrect password" error message is generated. Am I missing something here?
Quote:
the hacker then decodes by brute force and ignorance to arrive at the actual password
No. What you propose is little different from another common technique that doesn't really improve security -- that of hashing multiple times. Attackers don't care what the original password was or how the stored value was calculated, they only care about what they can pass in to gain access.
-
No, no - they store your password and check it as a nice secure SHA-2 hash value. And in case you need to recover it (as you did) they store the plain text version in the same table! :laugh:
Bad command or file name. Bad, bad command! Sit! Stay! Staaaay...
I had a profile on a site that continually (daily) sent me "we want you back" emails. I ignored them, but last week I noticed that they included my username and password -- in case I had forgotten. :omg:
-
Do we get extra points when understanding the reference here ? Or minus points for those who don't ?
Both, and leave the zero free for those who know but don't respond.
I wanna be a eunuchs developer! Pass me a bread knife!
