More on Last Year's Cisco Source Code Theft

grazzy writes "The New York Times has a story about last year's theft of Cisco source code: The incident seemed alarming enough: a breach of a Cisco Systems network in which an intruder seized programming instructions for many of the computers that control the flow of the Internet. "

Comments

Stakkato

[natron 2.0]

so now the hunt is on for the elusive stakkato...

Question for an expert...

[wcitech]

I'm without a doubt no networking expert, so I'd like to ask one of you who is: if the source code for cisco's equipment is leaked, would that person have the ability to create some kind of virus/malware that could bring the internet to a screaching halt? What can they do, infect routers with viruses now? I guess I'm unclear on the real dangers in a situation like this.

Re:Question for an expert...

[dgatwood]

I'd be more worried about a timebomb. Somebody writes a worm that spreads to all the Cisco routers on the planet, then sits and waits until some specific time---say... the anniversary of the first movie theater opening---then thus down the IOS processors and locks the OS up in an endless reboot cycle.

If somebody did something like that, it could basically bring the majority of the internet to a grinding halt. By anybody's book, this is a bad thing. Indeed, that's why I've been saying for so many years that we need more diversity on the Internet, that we depend way too much on Cisco and their systems, and that the Internet isn't nearly as reliable as we think....

Doesn't make sense

[afidel]

Cisco uses two factor one time passwords for remote access. I don't see how planting a trojaned copy of SSH on the lab computers would give the hacker access to Cisco's systems.

Re:Doesn't make sense

[dr_dank]

From the tone of the article, it seems that a keylogger was used to grab passwords from a cisco user, which were then used from machines in that lab remotely exploited by an ssh rootkit.

Re:Doesn't make sense

[afidel]

Yes, but if things were working as they should have been a password logger doesn't do you ANY good. The password as entered by the user consists of two parts, the first part is their passphrase, which is like a traditional password, but the second part is a numeric sequence which is spat out by a numeric FOB which is synced to the master access server. These combined passwords must be entered within a narrow window, and once accepted are no longer valid. Someone must have f'd up big time to allow an internet accessible machine to bypass this very strong access mechanism.

Re:Doesn't make sense

[rcw-home]

Cisco uses two factor one time passwords for remote access. I don't see how planting a trojaned copy of SSH on the lab computers would give the hacker access to Cisco's systems.

I don't know how Cisco has their stuff set up, but it's easy to imagine such a breach playing out:

1. Black hat replaces ssh client at University lab computer.
   
2. Authorized but unwitting user uses University computer to VPN into Cisco's network and then uses the trojaned ssh client to connect to a computer on Cisco's network.
   
3. The trojaned ssh client is now able to execute arbitrary code as the unwitting user on an internal Cisco computer. It uploads an executable to the internal Cisco computer that regularly makes outgoing TCP connections (they could even look like web browser traffic) to a computer under the black hat's control. The black hat sends control commands through these connections which the executable gladly obeys.
   
4. The black hat is now free to scan the internal network to look for a host they can get root on, or hope that the user's account on the internal server they control will be used to connect to other internal systems, perhaps using more highly privileged accounts. (Any admins ever had to sit down at a users' computer and ssh into a server to fix something?) The longer the initial breakin is left unidentified, the better the chances of this occurring.
   
5. Eventually the black hat will strike paydirt and get root on a system. From then on, the rootkit that the black hat installs can use any credentials anyone uses to access any systems remotely. Ssh into something? It can run commands on the remote host. Connect to a file server? It can replace executables that you have write access to and wait for someone else to run them.
   

While an attacker would need a fairly deep understanding of the software infrastructure he is attacking and of the usage habits of the users there to pull this off, the same basic strategy is applicable to UNIX, Windows, anything. I remember reading several years ago that the breakins at Exodus and VA Linux happened this way.

We're only used to the stuff we hear about not doing any real damage, because it's all dumb worms running without anyone at the controls. Just because we can fend off that stuff doesn't mean that someone with determination, knowledge, and patience won't get in and stay in.

Re:Doesn't make sense

[dr_dank]

Its not easy to pull off, but if you can enter the sequence before it expires, you can gain entry. IIRC, for the RSA SecurIDs, its 90 seconds.

Timing..

[gmerideth]

Rather good timing that last night on "24" we see Cisco's name all over the screen's at the CTU command center and the actress works in the line "the Cisco network is defending itself" followed immediately by an Alienware laptop on the screen.

Just in time for major articles about how bad Cisco's security was that they had some source code stolen. /golfclap foxtv

And people wonder why I don't watch television. Sad..just sad.

Re:Alarming ?

[iztaru]

I think the problem with this could come from two corners: 1- The people at Cisco are not as confident with their source code as the people of Linux, Free/Net/OpenBSD 2- There are back doors in the Cisco systems for the government to use and they are afraid anyone else might find them useful!

Thef

[Doc Ruby]

What do Slashdot "authors" (editors) do all day? They publish about 35 stories in a 24 hour cycle, usually about 4 editors participating. That's about 1-2 stories an hour, with 1-2 authors overlapping shifts. The summaries take about 2 minutes max to read, and the stories take max 5-10 minutes. That seems ample time to catch dups, fix typos, spelling and punctuation errors. Why not? What else are they doing? Maybe they don't read Slashdot after they've published, so they don't see all the feedback on their poor editing performance.

Re:Thef

[Doc Ruby]

Moderation 0
50% Interesting
20% Troll
20% Redundant

Where's another post running a time analysis of Slashdot editing? Even given Slashdot's absence of features to prevent comment redundancy, isn't a chorus of "not again!" appropriate? And how is my coherent, accurate comment, which I haven't seen before, a "Troll"?

Perhaps this comment is just the criticism uberpost, destined to point out all the serious flaws in Slashdot's publishing system model. If so, here's some constructive suggestions for fixing it:

1. Submitted story queue filter: editors see a story, with links listed separately (already a Slashcode function). Links already published in a previous story are indicated, linked to the previously published story. Publishing such links includes an "ongoing coverage" indication in the new published story.
   
   
2. Submission spell/grammar checker
   
   
3. Submission link checker: links in stories in submission queue are interlinked through a Slashdot redirection script which sets a flag. Until each link's flag has been set, by following the link (through the script) to the linked object, the story cannot be published.
   
   
4. Mod comments: Negative moderation must be accompanied by an explanatory comment, which can be viewed by metamoderators. Metamoderation gets more "teeth", with 3 "unfair" metamods cutting modpoints for a month, and 3 such suspensions cutting modpoints forever.
   

Cisco VPN Client

[Malc]

Tell me again why our IT department insists on using this buggy Cisco VPN Client POS that causes me grief on a daily basis...

This is actually kinda funny

[Jetifi]

I mean, 'cybersecurity' bigheads are all worried about Terrorists disabling our Internet Infostructure etc., but in real life it turns out that any vulnerabilities that could be used to break into (e.g.) the JPL, White Sands, the DoD etc. have already been exploited by petulant teenagers.

So in this sense, the script kiddies of the Internet are kinda like an early warning system: it's almost certain that before someone with serious intentions finds a nasty flaw and uses it, it'll be discovered by some kid who will promptly boast about it on IRC.

How lucky we are that terrorists find themselves vastly outnumbered by people with too much free time on their hands!

Re:Contradiction? Sorta.

[RealProgrammer]

As odd as it sounds, both are correct. A sophisticated intruder could compromise security with the stolen code. Or not.

But for the sake of argument, suppose they do find flaws in Cisco's code. An exploit shows up on rootkit.org or someplace. It should be apparent from the exploit which flaws they're using, and so Cisco cleans up the flaw. In the long run, customers are actually safer.

It's sort of a backasswards way to open source your code.

I wouldn't believe everything in this story.

[Anonymous Coward]

Note that this article is by John Markoff, who has established a history of trying to milk arrests of people for his own personal profit, by not only sensationalizing security breeches, but assisting authorities to the point of being on site during the arrests, and doing book and movie deals afterwards.

Don't confuse this story with independent journalism; Markoff is out to make a mint here, however he can.

Markoff reportedly was pissed of at Kevin Mitnick for spurning a movie deal, and later set himself up to write "the Kevin Mitnick story", earning over a million dollars in the process.

Here's a link: http://www.labmistress.com/kevins_story.php

So one really has to wonder what the Truth is here, and whether Markoff is just trying to screw over some teenage kid in Europe in order to make another million off of it.

So I'd take anything that John Markoff has to say with a LARGE grain of salt. The same goes for the New York Times, which has officially encouraged this practice.

The real truth is probably out there; but I wouldn't expect to hear it from either John Markoff or the NY Times.

We got hit.

[glockenspieler]

My laboratory was hit. We're all linux machines. Turns out that I still had an account on a system at Stanford where I was faculty and I transferred some files via scp to my machine at my current university. 4-5 days later, i see some logins from Stanford to my machine but I because I had been using the Stanford account recently, it just didn't register.

One day later, I'm on another lab machine using my lab /home directory (different from my main machine) and i notice a program (it was either brk.c or dobrk.c I think) that was on an unpatched system, allowed a priviledge escalation. I switch to root and look at the history and see a command to stop recording the command history but he (and the article indicates the person is male) misstypied it so i could see that he logged into this machine from mine, grabbed the source code for the exploit from a warez site, compiled, ran, got root, and just tooled around a little.

Because our machines are pretty isolated and don't have any hint of financial stuff, he seemed to just drop it. I called the sysadmin at Stanford, turned out that on a machine with over 500 accounts (i won't say which department), the machine had been rooted about 2 months prior and every password was being captured during that time. The breakin was tracked back through a couple of departments, then back to University of Michigan, then to Uppsala.

Three valuable and perhaps obvious lessons here. Local priviledge escalation exploits are important even if your system has very few users. Keep your system patched (duh...), and remember, if you log onto your machine from another, ask yourself "What do I know about the integrity of this machine?". I really assumed that my stanford account was pretty secure and so I didn't even think about logging from that machine to my current one. No more.

The other interesting thing was that the local exploit used on my machines was announced well after the Stanford machine was hit. I don't think I ever heard of how that machine was comprimised.