This could also be problematic.
Password Shucking https://www.youtube.com/watch?v=OQD3qDYMyYQ
First off, you'd have to assume the attacker knows the bcrypt hashes are bcrypt(md5(password)) – an attacker wouldn't always know this
Also it assumes there is password reuse, but that the password is strong enough that the md5 is uncracked.
> > I know I'm probably stupid but... how is this different from a dictionary attack? Instead of trying a list of known passwords, you try their md5s. If the md5 hasn't been cracked before, chances are that the password is strong enough to resist being cracked now. – nobody Jul 17 '20 at 14:55
> Because there are plenty of MD5s in the wild that A) just happen to not have been cracked yet because they weren't interesting enough to stand out, but B) once an attacker can figure out that that MD5 is inside a really interesting, high-value-target bcrypt, they might spend a lot more effort to crack that MD5. So it's not just a dictionary attack; it's a dictionary attack of passwords that are currently unknown but might be crackable with additional effort. And that effort is much less than trying to crack that password if it was only inside a pure bcrypt. – Royce Williams Jul 17 '20 at 15:00
https://security.stackexchange.com/questions/234794/is-bcryp...
So the assumption is: There is a breach A of an low-interest target with MD5 hashes and a breach B of a high-interest target with BCrypt(MD5) hashes. As A is not interesting enough, people don't invest the time to crack A's MD5s. But as B is super interesting they will use A as a dictionary source to then know on which MD5s they should invest a high amount of time, as it will help them crack the high-interest target B. Note that no specific user association takes place, like in the presentation about password shucking by Sam Croley (above Youtube link), where usernames/emails of A and B are correlated.
I think this is a bit more plausible than Croley's take on it. Because if I have identified a high interest individual, I would already invest a lot time to crack the MD5 password.
And yes, what you said bears repeating: All of this attack lives in the small space where the password is too strong to be cracked from a simple MD5 hash when you are mildly interested but not strong enough to prevent cracking when you are deeply interested – for varying degrees of mildly and deeply interested. Overall I would like to read about real world examples where this made the difference and how that password happened to fall into that region.
Still much better to have md5 directly in your db.
I fixed something like this just 4 years ago. :|
It's a very interesting attack, highly specific to the high-number of breaches, high password reuse environment we're in that enables at-scale password cracking.
I don't think it invalidates this advice completely. You should watch the talk and eventually add a global pepper (assuming it does not leak), and of course do the final bcrypt(md5(pass)) -> bcrypt(pass) migration upon user login.
https://security.stackexchange.com/questions/234794/is-bcryp...
bcrypt(md5(password) + salt) + salt
the problem with password shucking would be that they just do a bcrypt(md5) over the list of md5 hashes they have and check if they exist in your database.
but if each hash is salted they would need to run every their complete md5 hash list through bcrypt for each account instead of once per database.
I expect the parent's concern came from experience with PBKDF2, where the length is unbounded. It's good to consider possible denial of service attacks: if someone submits an enormous 1 MB password a PBKDF2 hasher can be knocked offline for 60s. Sha256 will quickly crunch that attack to a more manageable length.
https://en.m.wikipedia.org/wiki/Bcrypt
This sounds undesirable to me, so I'd support sha256 and an alternative.
I'd also recommend adding a max password length to any API. No point in allowing million character passwords.
Is that why websites sometimes have low maximum password length requirements ? Ex: must be less than 20 characters.
Websites (like some banks used to) that have less-than 20 char limits for passwords are purely bad security strategy.
There needs to be a limit, yes, but surely something like 100 characters, or even 50, would be more sensible.
Banks all have systems that will stop you from attempting to bruteforce them.
I've run across that at least once and it was a total pain to troubleshoot and figure out.
"A password one megabyte in size, for example, will require roughly one minute of computation to check when using the PBKDF2 hasher."
> If you care about the security of your account use a 20+ character random password and do not reuse that password at any other sites. There are several excellent password managers that can generate and remember such a password for you and make it easy to use. Here's a list: <link to list of password managers>.
> We allow all normal US printable characters in your password: upper and lower case A-Z, digits, and <list of punctuation>. Set your password generator to length 20+ and to use mixed case, digits, and symbols and you will be fine.
> If you think you might have to manually type this password at some point, you can use a reduced character set with a longer password. If you use just mixed case letters and digits make the password 22+ characters long. If you use just mixed case letters make it 23+ characters. Letters all of the same case? Make it 28+ characters. 32+ characters of hex is fine, too. Heck, you can make it all digits if you use at least 39 of them.
> If your password manager offers other options, such as patterns like groups of digits or pronounceable syllables separated by some symbol, that too is fine as long as you make it long enough. Make it long enough that your password manager gives it its highest strength rating.
> The exact upper limit on password length that our password entry fields allow might vary from time to time as we update the site, but will always be at least 64 characters.
NIST says there's no reason to limit password length any more[0]:
"Memorized secrets SHALL be at least 8 characters in length if chosen by the subscriber. Memorized secrets chosen randomly by the CSP or verifier SHALL be at least 6 characters in length and MAY be entirely numeric. If the CSP or verifier disallows a chosen memorized secret based on its appearance on a blacklist of compromised values, the subscriber SHALL be required to choose a different memorized secret. No other complexity requirements for memorized secrets SHOULD be imposed. A rationale for this is presented in Appendix A Strength of Memorized Secrets."