Busted! Conficker's tell-tale heart uncovered Security experts have made a breakthrough in their five-month battle against the Conficker worm, with the discovery that the malware leaves a fingerprint on infected machines that is easy to detect using a variety of off-the-shelf network scanners. The finding means that, for the first time, administrators around the world have …
Heh, this is Dan Kaminsky, from the story. Just to make something very clear:
Tillmann Werner and Felix Leder are the Honeynet Project researchers who actually noticed the behavioral shift introduced by Conficker. I've been doing work in fingerprinting lately, so I saw the opportunity to make it quite a bit easier to track down infected nodes in large organizations, but again, it was Tillmann and Felix who actually designed the fingerprinting logic that ultimately all these other organizations are integrating into their vulnerability scanning systems.
This is one small part of what's actually some very fine research about Conficker. This is their baby, I've just been helping it fly.
Does this mean ISPs will now be able to scan for customer's infected PCs and block them from their networks?
The (L)users who did not pay attention to their security should face be made to realise that they have a responsibility to themselves and other internet users to secure their systems. Either that or swap to an OS that isn't riddled with security holes.
Well I'm a Python newb as well so this is what I did:
1. Download and install Python for Windows: http://www.python.org/download/windows/
2. Download and unpack Impacket to a folder: http://oss.coresecurity.com/repo/Impacket-stable.zip
3. Install Impacket by opening a dos prompt on the folder and doing "python setup.py install"
4. Open a dos prompt on the scanner directory and type "python scs.py <start-IP> <end-IP>" and watch it go.
My network was clean but it took a while to scan an entire class c
Can someone explain why your antivirus software wouldn't pick up which machine have Conflicker? Most entreprise a/v products already report back to a central server on your LAN anyway.
This reads like another of Dan Goodin's bi-weekly "My mate Dan Kaminsky told me he did this ..." stories.
Do we not get a quote from Graham from Sophos too?
Because all that would tell them is that it's going to contact a server and await instructions, which believe it or not the clever people already know.
It's the content of those instructions which isn't known, and won't be known until they're issued, which won't be until the deadline.
It's like saying "I wonder what the weather will be like in two months time... I'll just wind my PC clock forwards and look at the weather reports for 'today'" - it just doesn't work.
changing the system clock will only make the malware contact the controlling servers. If the malware writers have not issued any commands then nothing will happen. I would think that they will be issuing commands on 1st of April.
A secure OS..... no such thing...... Open source has and still has security issues......
Please, can we stop using this stupid argument now? Linux's superior security record is not down to obscurity. It is the majority webserver platform after all.
It comes down to massive, continuous source code peer review and good kernel design. Windows NT lacks both and unless they open source it (and wait a couple of years for the massive refactoring effort that would follow) the writing's on the wall for this decrepit VMS clone.
Caveat: I am not involved with this and have only just tried this myself.
Prerequisites: Python, py-pcapy and py-Impacket (that's *i*mpacket, it's capitalised and the font used here doesn't really make that clear). Runs fine here with python 2.5.4, py-pcapy 0.83.0_1 and py-Impacket 0.9.6.0 (FreeBSD 7-STABLE).
Run "python ./scs.py <start-ip> <end-ip>" or use the filename of a file that contains a list of IPs you require scanning as an argument in place of start and end IPs; the example supplied is Unix format, hence lack of CRs in notepad.
If you're using Windows, be aware that the Windows MSI Python package is compiled with VS 7.1 and the extensions you need to build (py-pcapy, py-Impacket) also need access to that compiler, so will require much buggering about with bits and pieces of visual studio if you don't have a copy. You'll also need WinPcap. You're well advised to use a *nix box to run this.
Example output when pointed at a lappy with the server service enabled:
----------------------------------
Simple Conficker Scanner
----------------------------------
scans selected network ranges for
conficker infections
----------------------------------
Felix Leder, Tillmann Werner 2009
{leder, werner}@cs.uni-bonn.de
----------------------------------
192.168.2.31 seems to be clean.
HUGE thanks to the authors. We've needed something like this since Conficker/Downadup reared its ugly head.
So where's the one-click .EXE file for Windows users to at least tell them they have a problem or not, even if it doesn't remove the contamination itself ?
It's all well and good saying "It's not hard to download Python", but I'm sure it's equally, "Not hard to have effective security in place to stop such infections", but let's not forget it isn't just corporates having problems who (hopefully) employ competant sysadmins ... so let's put that nonsense to one side and get on with dealing with the problem before the clock runs out.
I llooked at the nmap site but couldn't see anything in the changelog which says what version I should be using, I don't care which paid for softwre will include detection, I want something I can download, run and breath a sigh of relief orknow which WAN cables to take the scissors to.
Full marks for the industry "responding", but so far it seems to be near zero marks in providing tools your average user can actually use ...
Unless someone knows differently ?
but i get an error and I'd quite like to get this running.
c:\python30\python.exe scs.py 10.226.40.40 10.226.40.254
gives invalid syntax error.
would one of the python genii on here mind telling a noob what they're doing wrong?
cheers
PH cos i have as much idea about Python as her.
Or, most probably, 2 days before the next Conficker C insignificant update. The other variants (e.g. the B variant, you know, the most widespread one) call home constantly already, and we're not dead yet.
""We have no idea what Conficker is going to do on April 1," Kaminsky said."
Well Mr Kaminsky might not know, but I think I have a pretty good idea. Nothing is gonna happen. Nothing noticeable at least.
El Reg is beginning to look a lot like the Daily Mail.
If you have the server service disabled on the machine you're scanning, the No resp. is what you'll get because port 445 is closed. Conficker uses SMB and it's this that the tool is querying. IPs with no active machine on them will also show No resp.
The code tells you what you'd get if you were infected:
if result[1]==0x5c450000 and result[3]==0x00000057:
print '[WARNING] %s seems to be infected by Conficker!' % ip
retval = 1
If you've turned off the server service on your boxen, congratulations, there's really no reason for client machines to use it. If you've turned off the server service, disabled autorun AND threatened your lusers with death if anything, anything at all connecting via USB or IEEE 1394 is found near the machines, you're *almost* immune from Conficker. If you've filled with Araldite the USB and IEEE 1394 ports on the client machines, removed all the DVD-RW drives and floppies, disabled the server service, riveted the cases shut and have a modified electric fence pulser on a stick to hand for lusers that never learn, GTFO, you're me ;o)
Ok, it's 1:12PM EST - Can anyone let us know where the NMAP signatures are going to be released? As someone else posted will they be -
" Does anyone know how the conflicker scanning functionality will be available on nmap? nse script or new nmap release? Any links to it?"
Please advise - I've run the Python script, although it did give a Not all options will be available - missing Crypto modules or something like that. It did it both on Windows and Linux. I'm new to Python, what Crypto modules are we talking about?
Folks,
Nmap creator Gordon Lyon, aka Fyodor, just emailed me to say he expects the Conficker update to be available within the next hour or so. For those who can't wait and don't mind mucking about with manual commands, the code is available at:
http://www.skullsecurity.org/blog/?p=209
Fyodor plans to announce availability of the patch at:
http://seclists.org/nmap-dev/2009/q1/index.html
Cheers,
Dan Goodin
Is anyone wondering whether the discovery of a fix on the eve of activation is a little suspicious? Can't quite articulate why I do but something's rattling in the back of my mind. (Perhaps some of my loose marbles.)
I'm still waiting for evidence the AV industry is funding some of the "bad guys"...
Paris because she'll get infected sooner or later. (Black helicopters are more appropriate but rather dull. Oh, wait...)
Yeah, I admire how well it's done and all that. But it's not that intelligent is it?
It stops AV packages from installing, running and updating.
Change the name of the package and watch it fly. ren norton.exe fluffypants.exe and it'll run just fine. Wont update, granted, but it's still a lapse in the intelligence there :)
Hello All,
I have the same errors seen by Anonymous Coward using the Simple conficker Scanner --
No resp.: xxxmachinexxx:445/tcp.
When I use the NMAP, I get this --
Host script results:
| smb-check-vulns:
| MS08-067: NOT RUN
| Connficker: ERROR: NT_STATUS_OBJECT_NAME_NOT_FOUND
|_ regsvc DoS: NOT RUN (add --script-args=unsafe=1 to run)
Final times for host: srtt: 0 rttvar: 5000 to: 100000
Per Chronos's explanation, that may mean my server service is not running. However, the server service is on, and the "netstat -na" show the machine listening on port 445.
Any suggestion on what another explanation of what the errors may mean?
Thanks,
local_vi
I misspoke, sorry for the confusion. The port is still open, but not accepting inbound connections was what I meant. The port is still used by the workstation service and you're quite right that it is listening on all interfaces both TCP and UDP.
With Server enabled:
192.168.2.31 seems to be clean.
Done
With Server disabled:
No resp.: 192.168.2.31:445/tcp.
Done
I can repeat this ad-nauseum. Could it possibly be something simple such as the Windows Firewall (on the client) getting into the mix?
Instead of asking all windows users/admins to install ported linux apps and libs like python/perl/py-randomscriptnamethatdoeswhatkaminskydoesntknowhowtodohimself.py Kaminsky should learn a real programming language like C/C++/Delphi/VB/etc. I know it's hard but he won't regret it.
Sorry ! you used the phrase real programming language and VB in the same posting, you really should stop being so silly !
Python (& other UNIX originated) scripting languages get the job done with the minimum of
fuss, if you have the language (shell) installed it works. Unlike the joys of recompiling for every platform, processor (there are other archetectures other than Intel) and even OS revisions as the average C programmer has to suffer.
Personally I don't think it's too much to ask a windows admin to be able install Python and run a couple of scripts. Though to be honest I would expect most admins to access to a UN*X or Linux machine these days.
I'm not a linux or un*x snob / fanboi BTW but I don't believe in putting all my eggs in one basket.
The windows mindset is why several companys I worked for used UNIX systems in the 1st place. It prevented somebody with a machine at home thinking that they had the capability ot fix a server issues.
PS.
Yeah I know theres shed loads of really piss poor UNIX admins (you know who you are).
PPS
Yeah I have been known to even lower myself to use WSH on windows when I want to automate something (eg when iTunes destroys my library yet again)
Not me, Mark, I can get all of my Win clients to give either result depending on the state of the Server service, which was disabled by default here long, long before MS08-067 was discovered - any file-sharing needed goes over the central shares, not from individual machines. I was replying to the AC above who can get nothing but No resp. results. I thought it may be a firewall myself, but I'm not assuming anything because a) (s)he's posting on the Reg so (s)he's not likely to overlook such obvious issues and b) I don't fully understand the conversation the scanner is having with SMB and I haven't had time this morning to fully analyse it as an ET update buggered up the emerging.rules (line 116 if anyone's interested) and my snort inline fell over, cutting everyone off the IPv4 web. Cron jobs... :o(
Interestingly, if you point this scanner at a Samba server running with everything enabled, you get the same No resp. result.
> Please get rid of the penguin and "good Jobs" logos. Then perhaps we would get less posters thinking obscurity = security.
*sigh* There is world of difference between running an obscure system and hoping an obscure security problem on a popular system isn't found.
An obscure problem on a popular OS is a large target for malware writers. A large problem on an obscure OS is a very small target because exploits are *not* random events, worms and viruses are written for windows and not OSX/Linux for the same reason that games are written for windows and not *nix - the effort doesn't justify the ROI.
Regardless of the security situation, running *nix hosts reduces your *risk* of being affected by malware, which is a much larger problem than being targeted individually by a skilled hacker.
The security problem is not solved by everyone running *nix - that just moves the malware-writers' focus, but individuals *can* reduce their risk (relative to the average Windows user) by switching to a *nix based OS which doesn't have a huge target painted on it.
I'll be interested to see if much happens after 1st April. If its bad, Apple will have a large number of smug users doing PR.
That a guy from IOActive comes out two days prior to the "worm's attack" to claim that with a "last ditch effort" we avoided catastrophe.
Could it possibly be that there was no catastrophe in the future and that the AV companies simply used Conficker to scare up support and sales? On Wednesday, those who were claiming Conficker was going to be a disaster are going to need an explanation as to why nothing happend, and I promise you that this lame ass excuse is thrown out there. And of course, the press will eat it up rather than ask the question "Was there any real danger, ever?".
thanks much to the guys for banging this together! The one thing it needs: a ping test first and then discard further testing of that address if there is no response. I've seen that if you hit an IP that isn't returning port-unr or simply isn't on the network...the scanner seems to sit forever trying the SMB port.
Alright, I'm not the systems or network admin but instead just a help desk guy. I got all of the tools installed and, thanks to this thread, got the scanner working. In order to actually have a functional scan, you need firewalls dropped and port 445 opened?
Ultimately our systems admin didn't want to run it because he didn't want to drop the firewall temporarily.
*puts hands up to rid himself of responsibility* I'm just a cog in the machine.
obviously dropping the firewalls on all the machines wouldn't be a great idea... if one machine actually was infected it would spread to all the others. This is only for catching the most obvious gross offenders.
Still it's 0:15 GMT 1st of April and the machines haven't taken over yet...
wait! what was than noise???
Something strange is coming out of my adsl modem!
AAAAAAAAArrrrrrrrrrrrrrrrrrrg
"Does this mean ISPs will now be able to scan for customer's infected PCs and block them from their networks?"
Not really - the ISPs will get as far as your router with their scan and no further. Remember the majority of folks are on broadband these days, complete with broadband router with a little thing called NAT at the front.
Unless there's an outbreak of routers infected with Conflicker, I don't think the ISPs can do much about this. Not to mention that even though it would be done with good intentions, it's still against the Computer Misuse Act.
Mind you, why leave it up to the ISPs? Feel free to run python scs.py 1.1.1.1 255.255.255.255 and let us know how you get on saving the world
Hi local vi, I didn't think you'd overlook that. I'm not getting any new ideas right now, although a bit more information might help someone a bit brighter come up with a suggestion. OS version? Members of a AD domain? Group policies? File and printer sharing state on the clients (I *think* this is independent of the Server service state but I haven't checked)? Any of these may affect the results.
One thing you may want to try is remote management via the computer management tool; if that fails, either remote registry or file and printer sharing is restricted somehow. Remote registry shouldn't make a difference to the scanner, but remote management requires both services IIRC. Also check any firewall on the machine that you're running the scanner on. *nix firewalls tend to egress filter as well as ingress by default.
I don't think you need telling but [I'm going to go into annoying arsehole mode and say it anyway -Ed] these settings would be better returned to the restricted state once you've found out why the scanner is having a hard time talking over SMB. Where there's one piece of malware with this distribution method, there's bound to be copycats. All of the discussed restrictions contribute to stopping such things. Only run what you need for day-to-day operations and disable the chaff, thereby reducing your potential entry vector surface. That's how OpenBSD got its reputation rather than by any super-secret methods that other OSen couldn't copy if pushed. There's nowt special about it other than the default settings overriding the quality, or lack thereof, of the admin ;o)
It is now 10:30-ish AM and cyberspace as we know it hasn't come crashing down around our ears, I haven't seen a significant rise in detected spam (if anything, e-mail volume is lower than yesterday's figures), I'm not being DDoS'd and my web server doesn't seem to be getting hit by Dfind, Morfeus F-ing scanner and such crap any more than usual. Some will see that as proof that the foregoing security efforts and discussions were hype. I see it as the security experts, the Conficker Cabal, reporters making people aware and the people who worked on the various detection mechanisms having succeeded with the prevention efforts. Well done, if it's not too early to be handing out the kudos.
Have I missed something, or can you only get infected by Conflicker if you haven't installed the Microsoft patch which mitigates it? ie, the one which came out last October - http://www.microsoft.com/technet/security/Bulletin/MS08-067.mspx
If I've not missed something, Snort has had signatures to detect which machines don't have this installed for months now.
What extra value does this new tool ad?
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
