URGENT! LL security exploit

Your argument seems to be that LL did wrong by resetting the passwords because the hash would have protected them. But that's wrong. It doesn't matter how they do it, and what arcane techniques they use for authentication: In the end, the attackers had everything required to start getting passwords within an hour or two. Not resetting the passwords would have meant them giving much more time to do a lot of damage.

Perhaps they did some already and it didn't go public yet. I wonder why it took two days, maybe because LL first didn't know the database was compromised, and has just noticed something strange?

It makes perfect sense, when you realize your system has been breached the first thing to do is to take it offline and examine it. And reset all passwords, of course. This will, no doubt, annoy lots of people, but it doesn't mean it's not the right action to take.



Now this is a much better subject for discussion. My wild guess is that they took the web servers offline because they found about a new vulnerability, or the servers were directly breached. They still didn't know they got the database until now.



What plan can there be? There was just one thing to do, reset the passwords. Then do the obvious thing, handle the tech support issues (which is fair to say they're not doing very well)

I apologize if I am restating the obvious then, I'm a bit distracted. But yes, this is all falling into place -- the sudden downtime, the immediate password reset.

Whee! That's all I know what to say.

oh Jeska that makes me feel SOOOOOOOO much better, i cant even get in to my $200 a MONTH Acct!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

I am so freking mad i could chew nails...how dare you do this...after 2 days my god..and i know my PETS name dammit, i sure the hell dont know why YOU dont.....

you guys hijacked my thread to waste technospeak on it.

I hate you all.

Okay, prove it.

string strPassword = XXXX;
integer iRounds = XXXX;
integer iNonce = 12345;
string strHash;

default {
    state_entry() {
        integer i;
        strHash = strPassword;
        llOwnerSay("Start!");
        for (i=0; i<iRounds; i++) 
            strHash = llMD5String(strHash,iNonce);
        llOwnerSay("Done.");
    }

    touch_start(integer total_number) {
        llSay(0, strHash);
    }
}

The output when clicked:



I'll give you L$10,000 if you recover the password in 48 hours or less. (Repost this script in the scripting forum showing the correct values for the two omitted variables, if this one goes away by then.)

The password is one lowercase English word. Lowercase a-z only, no numbers or special symbols. I even checked; it's in my dictionary.

Trivial, right? I mean, it's MD5. I gave you the source code. I even gave you the nonce value.

Come on... "any hash." Put up or shut up.

Again, you're not getting it. They got the database, which is a table with half a million hashes. It also has the salt, because it has to have it. They got into the webserver, which somewhere in it had the function that validates your password. Which means they also have the hashing count. Resuming, the attacker has:

1. Half a million hashes
   
2. Their associated salts
   
3. Account information
   
4. Whatever function is used to validate the password
   

So, for the attacker: iRounds is known, iNonce is known, strHash is known, plus account data is known to guess what the password might be. Somebody in that database, probably multiple people will have the password of "password". Determining who is only going to take minutes.

i think there was more than one security question, other than pets, i tried my fav hobby and it worked, try your moms maiden name etc...try anything but beat LL at is own game dammit, this is atrocious!!! and i bet it lands LL in the mud hole over this one, i got in and im still furious!

Now my partner cant get in great just great, what about all the ppl that owe us rent today that cant get in?????? This is costing us MONEY...and its just not rite, we did not create this problem, yet we are paying for it...yes LL i am NOT a happy, and satisfied customer about now...if that matters any at all to you...

Also, who cares about the stupid password.

Sounds like they have unencrypted Name, Address, Phone, and let's say 10% of the users used their mother's maiden name as the security question.

What they're really after are the roughly 5000 credit cards that they can now call up and transfer funds /open new accounts using the owners name, address, phone, and maiden name info.

So the encryption to worry about is NOT the password, but sixteen digit numbers. Brute force.

You said:



But I wrote a hash function in under five minutes using only tech that LL is known to have that demonstrates that you are completely full of it. Nothing for you to do now but backpeddle and start qualifying your ignorant claim.



I'd like to know how they got that from a SQL injection.



In the LL case, the iRounds is NOT known (if they even use a similar mechanism, about which you have no idea). You don't know if MD5 is used. You have no idea whether or how they apply the salt.

Actually, it seems sufficient to stop that sentence here:

You have no idea.

My L$10,000 is completely safe, from both hackers and you.

Are you serious? It took them two days to do anything about the database being comprimised. The hack took place on wednesday when all the website problems were happening. You think allowing account info to be in the wild for 2 full days before lifting a finger to take any action is getting on it before trouble hits, think again. Not to mention the vague information they're provided us with regards to what information was comprimised, and the lack of phone support till after the weekend doesn't seem like LL getting right on things to me. I guess I just have higher standards for personal information security.

Since you wrote the challenge post, I already coded a little program that tried 676000 passwords against your string using up to 100 MD5 iterations. Unless your iteration count is brutally high, I think I have quite good chances.

Best of luck. Please don't post anymore til you get it.

Edit: I will post the script (here or on the scripting forum) on Monday with nothing removed to prove that I'm on the level.

Eh, no.

Whatever happens to our little contest isn't very relevant. First, if they have the database, they will be able to guess the base algorithm quite easily. The two most common ones are MD5 and SHA1. MD5 and SHA1 generate outputs of different length, making it quite easy to guess which is it.

Next, you need the salt or nonce, which is going to be in the table.

Assuming they're using an algorithm like your, it's still quite trivial: They only need to try to decrypt their own password! Once they crack it, they'll know for sure the decryption method works, and will go on to work on the rest of the database.

I suppose you're not completely stupid though, and used some number in the millions, because that's really the only difficulty of the contest.

Unless I use MD5 on the penultimate SHA1 key to make it look like an MD5 key, making you spend the rest of your natural life using the wrong hash because it was so obvious which one I used...



They would be stupid to use an algorithm I coded in a weak scripting language in a few minutes.

Make the salt the key to a nonrepeating generator function and use the output of the generator function at each round.

Then use SHA1 on even numbered rounds and MD5 on odd numbered rounds.

Then swap the first character with the character in the (current round#) position.

Add a few more inefficient text operations to each round. They don't need to be well documented, as long as they're reproducible.

Keep going until it takes a fast CPU about 10ms to compute the hash. Add rounds once a year to keep the calculation time constant as CPU speeds increase.



They still need to know what algorithm they're using. They don't.



You will see the truth on Monday.

you guys hijacked my thread to waste technospeak on it.

I hate you all.

I had the same problem, finally after trying numerous misspellings I guess it. Hard to believe I misspelled it the first time??!!?? I am sceptical to say the least.

Try misspellings of the answer. Maybe you will get it right.

But yes this sucks, its as if LL tries to make things as miserable as they can for thier customers.

Actually quite the opposite, they all got off early, they are not answering the phone, I tried and got a recording, no telephone assistance until Monday.

Which reminds me, it's been about a year since I updated the password algorithm on the site I run. Excuse me, I'm needed elsewhere. Not to worry, I'll be back Monday.

Have a great weekend, and don't let your CPU overheat!

oh yeah so all they care about is us changing our passwords. Oh well to info that was taken

No need to wait so long.

string strPassword = "concentricity";
integer iRounds = 3981;
integer iNonce = 12345;
string strHash;

default {
    state_entry() {
        integer i;
        strHash = strPassword;
        llOwnerSay("Start!");
        for (i=0; i<iRounds; i++) 
            strHash = llMD5String(strHash,iNonce);
        llOwnerSay("Done.");
    }

    touch_start(integer total_number) {
        llSay(0, strHash);
    }
}

[18:47] Object: beacface99bd0dc785fc0347348d6c29

Cracked on a dual Athlon MP 2000+, using both CPUs at once, in 20 minutes, using a 234937 words list, trying up to 10000 iterations. Process worked at about 30 passwords/second with my sluggish perl script.

Now where are my L$10000?

If you don't remember your security question/don't have a valid email, they will probably ask you to mail or fax a copy of your driver's license. How else could they authenticate you? Asking the security question is already dumb because the attackers presumably have it.

Perhaps they could let people reset their password without email if they were coming from the same IP (or range maybe?) that they have used with SL in the last month.

Yeah - one of my alts got that one too, all the others made it back on just fine...

Well, let's see if you're indeed back this Monday. Wouldn't be very surprised if you weren't. Your departure's timing is quite suspicious.

Has anyone had difficulties changing passwords as no emails are arriving from SL ?