New Denial-of-Service Attack Is a Killer

ancientribe writes "Hacker RSnake blogs about a newly discovered and deadly denial-of-service attack that could well be the next big threat to the Internet as a whole. It goes after a broadband Internet connection and KOs machines on the other end such that they stay offline even after the attack is over. It spans various systems, too: the pair of Swedish researchers who found it have already contacted firewall, operating system, and Web-enabled device vendors whose products are vulnerable to this attack." Listen to the interview (MP3) — English starts a few minutes in — and you might find yourself convinced that we have a problem. The researchers claim that they have been able to take down every system with a TCP/IP stack that they have attempted; and they know of no fix or workaround.

I cant believe this is the first comment,

Some DOS attack on Slashdot in progress?

Re:I cant believe this is the first comment,

Yeah, some stupid user deltree'd the whole site!

fearmongering

While it is pretty interesting, and disturbing, we are once again faced with a "The Internet Will Cease To Exist And Your Brain Will Explode" vulnerability. We dont know exactly how it works, we dont know exactly what to do to stop it, fixes are not available, and we are all doomed. The podcast goes into enough detail about how they discovered it to be replicated by skilled evildoers without too much trouble, but nobody knows how long, easy or invasive a fix is going to be.

Re:fearmongering

Sorry, but your entire argument is shot down by TFA. For those of you too lazy to read it, this gem "Robert and Jack are smart dudes. I've known them for years," clearly shows that your argument is moot. The author has known them for years from (presumably) T-Ball league. How can you argue with that?

(this having to wait 5 minutes between posts is a pain in the ass. Anyone else stuck with this restriction?)

Re:fearmongering

Sorry, but your entire argument is shot down by TFA. For those of you too lazy to read it, this gem "Robert and Jack are smart dudes. I've known them for years," clearly shows that your argument is moot.

Seriously....just saying "Yeah, these two dudes I know can break the whole Internet. Trust me. I've known them a long time." is just completely lame and useless.

The article is nothing more than fear mongering and fudfudfud (please tag appropriately). Unless there's something to the interview beyond "I know how to break the Interwebs!!!", I'm from Missouri on this one.

Re:fearmongering

(this having to wait 5 minutes between posts is a pain in the ass. Anyone else stuck with this restriction?)

My sig answers your question. :)

Re:fearmongering

(this having to wait 5 minutes between posts is a pain in the ass. Anyone else stuck with this restriction?)

Yes, limiting the possibilities to comment is clearly a bad idea. /. summaries have always been quite bad for as long I can remember it, but all the informational value is in the comments. Where else can you see a fearmongering article, people making some obvious remarks, getting insightful retorts to finally end on a +5 comment coming from a guy working in the lab TFA mentions ?

Slashdot, don't fear posters. Your moderation system filters spam (and as*holeness) with enough efficiency, don't add nagging features !

Re:fearmongering

Dude, did you just tell one of the guys who discovered this issue to *not* post links to his blog which may contain relevant information? I mean I'd understand if his blog had a ton of advertisements or something. But it's pretty much just blog entries and that's it.

Weird.

Pfffft

Doesn't affect me. I haven't used DOS in YEARS. Some folks need to move up to Windows 3.1. That is where it is at.

Re:Pfffft

What? WinME was simply repackaged Win98. Windows _NT_ was built by David Cutler, on a VMS foundation rather than a DOS foundation, because Cutler was one of the core authors of VMS and there were some fascinating lawsuits about his duplicating his old VMS work for Microsoft.

Re:Pfffft

Piffle yourself. The memory management, as an example, was definitely theft from DEC.

David was one of the three team leads on Starlet, which became VMS. And yes, according to DEC personnel of that era, he was very much a technical lead, if not the technical lead.

Re:Pfffft

Nope, NT 3.1 circa '93. We were an early adopter on a currently top of the line Pentium (1)--50 MHz, I believe. Thing would BSOD if you more than looked at it funny.

Transcript

Do people really have time to listen to podcasts unless they are commuting?

Is there a transcript???

Not much information

Hi,

Neither interview nor Link provides much information about the kind of attack. Between the lines they seem to be doing something with the ressource usage by manipulating tcp session parameters. But that's idle speculation for now.

CU, Martin

Re:Not much information

Here's a better story with more info: http://searchsecurity.techtarget.com/news/article/0,289142,sid14_gci1332898,00.html [techtarget.com] Looks like they're able to mess with the session parameters, as you said.

How it works

Many TCP servers use a technique known as a SYN cookie in order to prevent attackers using spoofed IP addresses from launching SYN flood denial-of-service attacks against them. The cookie is essentially a chosen TCP initial sequence number that is calculated using some specific hashed metadata that reflects the details of the specific TCP connection. Once the client returns a correct packet to the server, the server knows that the client isn't using a forged IP address.

Sockstress computes and stores so-called client-side SYN cookies and enables Lee and Louis to specify a destination port and IP address. The method allows them to complete the TCP handshake without having to store any values, which takes time and resources. "We can then say that we want to establish X number of TCP connections on that address and that we want to use this attack type, and it does it," Lee said.

In summary, it works by establishing tons and tons of connections using carefully-forged SYN cookies [wikipedia.org]. The irony? "SYN Cookies are the key element of a technique used to guard against SYN flood attacks". ROFLMAO.

And then it gets scarier:

From the wikipedia article:

The use of SYN Cookies does not break any protocol specifications, and therefore should be compatible with all TCP implementations.

Now, are you ready to scream?

the 2.6.26 Linux kernel added limited support of TCP options.

Scream.

Mod Parent Dow; Not a fix!

Now we see that a little bit of knowledge can be a dangerous thing.

The point that's in the grandparent's post is not that your own syn-cookies can be used against you. Syn cookies on your server are doing the right thing and are protecting you against normal syn floods.

What's happening in this attack is that the client side (the attacker) is using their own syn cookies to store information about connections on your server (instead of in their own memory). This allows them to handle more connections than otherwise. Unfortunately there is nothing you can do to stop this. They are using required behavior of the TCP stack for their information storage.

Some mitigation strategies that I can think of

The parents "fix" will make things slightly worse during this attack since turning off syn-cookies will mean that your server will have to track even more TCP connections. Not just those that are active, but also those that have just started. Of course, it will also make the new attack pointless since they can just do a normal syn-flood instead.

• Increase the TCP connection storage on your server to such a size that the DOS becomes impractical
• Ensure that TCP connections time out after some time if they have not been authorised to a particular user
• Impose a resource limit per authorised user on connections. Impose a separate resource limit on all non authorised users which will not interfere with authorised use.
• Use IPSEC to authorise all incoming connections / alternatively prioritise authorised sessions.

The best current full fix I can think of is to use IPSEC and ensure that all incoming connections are authorised. Your own users will still be able to DOS you, but at least you can hunt them down.

For those who can't listen to the interview

Here's a link to an article in English:

http://searchsecurity.techtarget.com/news/article/0,289142,sid14_gci1332898,00.html [techtarget.com]

From the article:

Many TCP servers use a technique known as a SYN cookie in order to prevent attackers using spoofed IP addresses from launching SYN flood denial-of-service attacks against them. The cookie is essentially a chosen TCP initial sequence number that is calculated using some specific hashed metadata that reflects the details of the specific TCP connection. Once the client returns a correct packet to the server, the server knows that the client isn't using a forged IP address.

Sockstress computes and stores so-called client-side SYN cookies and enables Lee and Louis to specify a destination port and IP address. The method allows them to complete the TCP handshake without having to store any values, which takes time and resources. "We can then say that we want to establish X number of TCP connections on that address and that we want to use this attack type, and it does it," Lee said.

Computing SYN cookies?

Sockstress computes and stores so-called client-side SYN cookies

This isn't supposed to be possible. SYN cookies [cr.yp.to] are supposed to contain at least 24 bits worth of entropy, produced by running a server-side secret through a one-way hashing function. You can easily obtain a SYN cookie by performing the initial SYN with the server. A SYN+ACK comes back which contains the SYN cookie (as the initial sequence number). The cookie so received is unique per TCP connection (IP address and port numbers at both ends), and valid only for a limited time. The server side does not maintain any state information until the cookie is returned in the client's ACK.

If they are actually computing SYN cookies on the client side, it's evidence of a weak SYN cookie implementation. Computation of the cookie should be infeasible without access to the server-side secret. Of course, this may be a case of sloppy reporting. As usual, we aren't given all the details of this earth-shattering vulnerability. We are simply left to guess whether these folks (and those that report on them) know what they're talking about or not.

They could be guessing cookies, and that would explain the "it will hurt intermediate systems" excuse they used for not demonstrating it, since they'd need to flood the peer TCP with millions of randomly-guessed initial sequence numbers. Incidentally, if this is a TCP SYN-flood attack of this sort, the "after effects" they mention have to do with the fact that all the TCP connections must time out naturally -- a process which might take several minutes per connection, depending on the configuration of the listening server application. The process is naturally limited by bandwidth and the size of the TCP state table: you have to be able to send successful fake ACKs fast enough to fill the TCP state table. All the usual mitigations for TCP SYN floods apply, such as increasing the state table size and reducing the timeout for open but idle connections.

It's not at all clear that this is any worse than the kind of DDoS attack that a typical botnet can unleash. In that case, you get thousands of perfectly real TCP connections from multiple addresses almost simultaneously. So maybe this attack doesn't require a botnet, but I don't see that it's a big new threat (as I've described it).

pff

Typical /. reaction to potential danger:

"Hah. Until I don't taste nuclear winter snow I don't believe that's gonna happen'"

Give the man his nuke. He earned it.

I'm safe

This doesn't me since use I UDP all communications communications for.

How this really SEEMS to work...

The observation: You can use a SYN-cookie like trick on the client side as well for an attacker:

You send SYNs where the initial seq # = H(sip, dip, sport, dport).

Now when you get a SYN/ACK back, you can send the ACK to complete the handshake. You can use the ACK field back from the server to know where you are in what data to send (just subtract the value from the initial sequence # to know what the next piece of data to send is), and you can know where you are in the received data (if necessary) by storing just the server's initial sequence #.

As a result, you can now interact with the server without having to maintain ANY TCP session state, or just a single word (the server's initial seq #), allowing the attacker to use vastly fewer resources to tie up server resources.

On one hand, this is a cool trick, and potentially useful for an attacker: if you have only a couple of machines and really want to tie up server resources, you can use this quite quickly.

But OTOH, attackers already have so many zombie resources that this really doesn't necessarily buy the attacker all that much: If you have 10K machines banging on a server, the 10K machines have a good 2000x more state than the servers. So who cares about stateholding requirements on the zombie side? Thus I think its only really relevant if you wanted to DOS google, akamai, or some similar very-high-resource infrastructure.

And as the attacker can't SPOOF packets with this (it needs to see the SYN/ACK), the zombies can be filtered if the DOS is detected and the attacker's identified as well.

Re:Go for it, take on my machine!

Unless it's a generic vulnerability in the TCP spec, in which case almost every implementation of it would be vulnerable - including all those Linux machines. Linux is not some magical shield, it takes responsible use to keep it secure.

Re:Go for it, take on my machine!

Of course Linux is not a magical shield. But having a diverse eco-system is known to protect against many attacks.

One of the reasons stories about how the banana is going extinct come up every few years is because the "modern" banana that most people in the over developed world can buy, are all clones! One disease can attack all the plants in the same manner.

In the same way, computers that have the same OS tend to be vulnerable to the same attack. Because there are a lot more OSs based around Linux (and BSD), people running these OSs are less vulnerable, because they are in a diverse eco-system. Especially when these kernels and the user-land tools are FLOSS.

As such, yes, it maybe a generic vulnerability in the TCP spec. (though how likely is that?), however, it is not specified, which is why I asked if it did affect *nix.

If nothing else, due to the nature of FLOSS, the attack could quickly be coded around as soon as it is known, and then pushed out to many many people running auto-update systems (such as Debian, Ubuntu and similar). (Even if that breaks the spec.)

Re:Go for it, take on my machine!

Thief! That's MY address!