How to say you’re vulnerable to code injection without saying you’re vulnerable to code injection.
Are they vulnerable though, if they already exclude it at the user input?
I yet have to learn SQL and is there a way to allow passwords with '); DROP TABLE… without being vulnerable to an injection?
nevermind i googled it, and there various ways to do so
This still smells though. Why is the raw, plain text password string getting anywhere near database queries in the first place?
I doubt it is. they probably have a WAF that blocks these strings though and didnt want to bother reconfiguring it
Prepared statements, mostly. You define the query using variables, turn that query into a language-dependent object, assign values to those variables, then execute the statement. The values will be passed verbatim, without any parsing.
Or, since we’re talking about a password, you could encode or encrypt it before inserting it into the query string. The fact that the website could be negatively affected by phrases in the cleartext password is very concerning.
At best, it means they’re storing your password instead of just a salted hash. And that’s horrible.
I noticed that upper case select, drop etc are not prohibited.
Poorly implemented user input filters are not a valid solution to being vulnerable to injection.
Good old Bobby Tables
No one in their right mind is storing plain text passwords, or letting them anywhere near the database.
You convert the password to a hash, and store that. And the hash will look nothing like the password the user typed.
You’re right. No one in their right mind would do that.
On the other hand, people not in their right mind often run things. Such as my old professional liability insurance. Which wrote the username and password in the yearly statements…
And also sent you the password through email if you forgot it…
Also you couldn’t change it…
There was a popular companion app to a game I play that’s stored passwords as MD5 hashes for years and when they got hacked they were able to decrypt everything.
Bonus point, the app was released multiple years after md5 was cracked.
Developers (including myself) cannot be trusted to implement the correct process 100% of the time. It’s happened too many times for it to be a single person issue and has transcended into a problem with software engineers
😬😬 that’s crazy but good to know
Lol. Yes, people do still build systems and store plain text passwords. I regularly get scammers sending me my throwaway passwords from crappy sites. Good thing I never reuse passwords, or email addresses.
Parameterized queries.
🥲 t Yep it’s that easy to do the right thing
Maybe they filtered those strings to be safe, and put the notice there to answer the invertible “why won’t it accept my password” queries.
It’s a shitty password engine. But not necessarily uncleansed
If they’re trying to protect themselves from code injection by rejecting certain user input like that, then they don’t actually know how to protect themselves from code injection correctly and there may be serious vulnerabilities that they’ve missed.
(I think it’s likely that, as others have said, they’re using off-the-shelf software that does properly sanitize user input, and that this is just the unnecessary result of management making ridiculous demands. Even then, it’s evidence of an organization that doesn’t have the right approach to security.)
I don’t know, maybe they saw that classic XKCD comic and now they’re thinking “hahah, I’m wise to your tricks, ya little shit”
This is the result of some doc writer or middle manager not fully understanding what they’ve been told.
So they’re not hashing or salting the passwords too. Cool…
They might be doing it in the DB query, but they’re definitely not sanitized beforehand.
Sanitization has nothing to do with salting and hashing.
If you do the salting and hashing in a database query you need to sanitize the input before you use it or you open yourself to SQL injection.
Databases have salting and hashing functions, after all
Which makes me want to try and insert a password of a few megabytes worth of text. Should be fine, since there is no max lenght defined, right?
If there is no overwrought prohibition of something I know that at least in America that means it’s
- Affirmatively legal and
- Legislatively encouraged by the FREEE Act
So give ’em hell!
That’s not how it works. The code always has access to the submitted plaintext password. It’s salted and hashed after it’s verified for complexity. The complexity verification can even be done in JavaScript.
Little Bobby drop tables
Obligatory Little Bobby Tables: https://xkcd.com/327/
And for those who feel like saying they’ve already seen it: https://xkcd.com/1053/
I don’t believe this is real. This isn’t real, right?
This is real - I took the screenshot myself.
What zero string sanitation does to a mfr
We could still have some fun with ALTER TABLE
Some of the strongest and easy to remember passwords are just a few words strung together with a few numbers.
For example: Simpsons7-Purple4-Monkey1-Dishwasher8
Just remember “Simpsons Purple Monkey Dishwasher” and “7418”. You’re probably never going to forget that and I just tossed it into a password strength tester and it said it would take about 46 billion years to randomly guess it.
It would take me about 5 seconds because you just told me what it is genius
Now remember these types of passwords, all different for different services. It’s not a realistic expectation. Password managers are a must nowadays if you want to protect your accounts. But these types of passwords are also easier to type out.
My tactic is use a memorizeable passphrase as the unlock for the vault and assorted gibberish for anything in the vault
How about hunter2
Yeah, a bunch of asterisks works too.
Correct horse battery staple
Anything else and I can’t remember so I’m using this.
I’m told it’s very secure so I must be very secure. Right? Right?
I’ll never understand why spaces are commonly not allowed in passwords.
I’ve seen stupid developers do dumb stuff that makes keyboard remove and or add spaces to password fields. Making you type correctly but still fail.
Same with tabs.
Password strength checkers are taking an approach that’s naive for this case. The actual strength depends on the size of the dictionary and the number of words you randomly choose out of it.
Bcrypt has a length limit of 72 characters, so very long passwords generated this way can be silently truncated. Developers can avoid this problem by running sha256 on the input before giving it to bcrypt, but that isn’t common.
Or you could just use a password manager
For maximum security your password manager should have a password and you have no choice but to remember that password.
Memorising 1 password like that sure, but according to bitwarden I have 209 passwords, no way I can ever remember them all
I also tweak my base password based on the site. One site hack could lead to all your passwords being compromised no matter how long it is. Sure someone might be able to figure out the pattern if they analyzed it manually, but most hacks try to break into accounts en masse and they’re not going through passwords one by one.
Yeah, but pass123! is easier to type.
If the structure of it is known it becomes much faster. Word+single digit^4 isn’t all that hard.
For the vast majority of purposes, it’ll be fine. And certainly as long as that particular structure isn’t commonplace, it won’t be easy to guess anyway. But password strength testers don’t consider that - guessing “aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa” randomly also takes billions of years, so they can give a bit of a sense of false security.
Eh it’s still pretty hard.
If we check the numbers of English words from https://www.merriam-webster.com/help/faq-how-many-english-words and take a conservative estimate of 400 000 at the bottom of the page.
That means with the exact format of (word)(number)- 4 times has (without repeating words) 400000*9*399999*9*399998*9*399997*9 = 167957820891293697014400000 combinations. https://www.wolframalpha.com/input?i=400000*9*399998*9*399997*9*399996*9
The fastest super computer at the moment apparently sits at 1.1 quintillion Hz. Or 1.1 billion billion.
If that computer could make 1 guess every clock cycle it would still take it over 4 years (167957820891293697014400000 / 1.1quintillion = ~52 months ) to run through all possibilities.
Now that is a very fast computer, and we haven’t included the possibility of various numbers of words, different delimiter, or where and how often numbers appear. So unless you’ve really pissed off the US gov I don’t think you have to worry about it.
There’s a reason passphrases are the currently recommended way to generate secure passwords that are hard to guess but easy to memorize/type in.
Just use a password manager!
All of my password look like this, and even I don’t know what they are for any given site. And each site has a unique one.
.S"uB3U-_5X?e8XRa:2JEdit: I just saw your other comment, but you said plural “passwords”, so I’m leaving this up.
I consider this an invitation.
Looking at that I wouldn’t be surprised if those rules are just client-side validation.
surprised that they let you use
evalsubmits Drop Table as passphrase
Grabs popcorn
Ima just use my butthole with a biometric scanner.
Didn’t say anything about truncate!
Anyone taking bets on them sanitising user passwords?
