Waiting for the Other Conflicker Shoe to Drop

OK, so by now you have probably read a 100 articles on Conflicker and the spread of the worm. I warned of impending trouble from the worm, which, thankfully did not emerge over the weekend. I really thought the traffic levels would be of importance, but indeed, there was little impact on global traffic levels. This is one of those cases where I am really glad I was wrong!

Now that the majority of the scanning and traffic spikes are over, we are waiting for the other shoe to drop on this attack. The initial worm spread and compromise was likely only the first phase of the attacker’s plans. They now have an immense network of bot-infected hosts at their command. What they will do with them and how they will focus these systems on compromise remains to be seen. Given the sophistication of Conflicker and the “intelligence” of its design and scanning code, the forthcoming use could be a pretty creative and powerful threat vector. We may well see some new form of attack or probe that we have not encountered before.

It is, of course, critical that organizations and individuals move to identify and mitigate any infected hosts. The less bot-infected hosts for the attackers to command, the better. The problem is that many of the compromised systems are in locales with limited IT knowledge resource levels. In many of the countries where infected systems are concentrated, IT admins and tech savvy users are difficult to locate and even harder to afford. This means that while some of the systems may get cleaned up, there is still likely to be a significant army of infected zombies for the bot-herder(s) to wield.

In general, in this case, other than mitigating compromised hosts, there is little you can do beyond standard security practices. You can deploy detective capabilities around logging and vision-enhancement tools like HoneyPoint, but other than the usual, there is little focused risk minimization you can do for this one.

My best advice is to remain vigilant, keep up to date and keep working to better the security across your organization. Eventually, the other shoe will drop, and when it does, we will have to do our best to turn aside whatever happens.

Danger: Conflicker Growing at Massive Rate **ALERT**

Just a quick word of caution, the MSI::HITME (HoneyPoint Internet Threat Monitoring Environment) is getting nailed by Conflicker worm scans. New hosts (not seen in the last 24 hours) are probing the HITME every 5 mins or so! Scanning for port 445/TCP is growing HUGELY, if not EXPONENTIALLY!

This is important to you for the following reasons if you are an IT person or Infosec person:

  • The rate of spread is quite high. Likely, we will see Internet wide traffic impacts over the weekend or by early Monday if it continues at present growth rate.
  • Even when it plateaus and tapers, this will mean a HUGE INCREASE in infected bot-net machines, the likes of which will likely compare to Kraken or Storm
  • On Monday, you should be prepared for worm war. People who took their machines home and got infected over the weekend will be returning it to your office on Monday or when they come back to work. Look for scanning on a large scale in many organizations.
  • You are likely to get “those calls” from a competitor or other company about “why is your network scanning mine” — always fun!

What can you do?

  • HoneyPoint users (Personal Edition and Security Server) should deploy Linux or virtual decoy hosts (no SAMBA/CIFS) with a HoneyPoint listening on 445/tcp. (Note that you can’t bind to 445 on Windows systems as Windows is using it to host the possibly vulnerable service) Investigate any host that probes that open port.
  • Make sure all servers and as many workstations as possible are patched! (do this NOW!!!!!)(Servers first!!!!)
  • Make sure all AV is up to date. Most AV will catch the overt worm, though evolution and mutation seem likely.
  • Prepare yourself and your team for the battle ahead.
  • If you are a NAC person, pray to the various “NAC Daemons” that your solution actually works and is configured to actually protect you in this event.
  • Obviously, make sure all of your Windows hosts are protected by a real firewall and that port 445 is NOT Internet exposed. (Goes without saying, but obviously not…)

Please, pay attention to this one. It looks “slammer/code red” nasty…..

** 1/25 11:00 AM Eastern Update: After talking with many other folks on twitter and with some wonderful visualization help from @pophop, it appears that the growth is linear, AND NOT EXPONENTIAL. Much of the growth is coming from consumer broadband, especially Asia and Europe. Given the oddity of the source host increases and data from other scans, I am wondering if the infection scans for a while and then goes into a sleep mode to await further instructions. More analysis and such on Monday. Thanks to all for the help, especially @pophop and SANS **

3 Links for Securing USB Drives

This project caught my eye. It is includes crypto and ease of use. It is called geek.menu and is based from the portableapps project. Installed and configured right, it makes an encrypted file system to protect your data if you lose the drive. It also allows you to easily configure some pretty powerful options around the apps you install. Check it out if you are a big thumb drive user.

This article is a great overview of risks from thumb drives. It should be a basic requirement for any user in the organization that gets provisioned one.

Lastly, for those of you want to make the most of security through obscurity to protect your precious USB thumb drive from discovery, check this article out about hiding your drive in the wall.

If you are both a thumb drive (USB drive) and a Windows user, you should probably read about the Conflicker malware. It is currently spreading wildly and can transit itself on USB drives. (Oooops, that was 4….)

Major Breach at Heartland Payment Systems

You’ve heard this story before. A major credit card company has experienced a massive breach. Tons and tons of data was stolen during the incident. They think they have it under control and are working with law enforcement. You should check your statements. Blah, blah, blah…

Once again, though, in this case, the company was certified as PCI compliant by their PCI auditors. If they were all compliant and filled to the brim with “fluffy, compliant goodness” then the attackers must have used some uber-hacking technique, right? Some bleeding edge tool or 0-day exploit that cut right through their defenses and rendered their compliant protections useless? Ummm…. NO…. The mighty technique that caused the damage? A sniffer!!!! (Some of the best technology that the late 80’s/early 90’s had to offer…)

How did I reach this conclusion? From their own press release:

“Last week, the investigation uncovered malicious software that compromised data that crossed Heartland’s network.” — sounds like a sniffer to me….(and a lot of other infosec folks…)

That’s right, the mighty sniffer strikes again. In the last couple of years, this same attack footprint has occurred over and over again. It has been largely successful. Why? Because companies don’t encrypt credit card data in transit across networks. Sure, many of them encrypt the database (not all, but many.) and some use various forms of endpoint protection, but many (way too many apparently) don’t encrypt the credit card data in transit across their networks.

Even worse, the PCI DSS DOES NOT REQUIRE THIS. That is how they can be compliant with PCI and still have this issue. What a cruel joke for consumers.

The DSS requires that organizations encrypt credit card data when it flows across “open, public” networks. Well, guess what, when your network gets compromised, even your “internal, private LAN”, it becomes “public” at least for the attackers. Misconfigure a firewall rule, get a workstation popped, allow a social engineer into the environment and that “private network” is not so private anymore, is it?

But, that never happens, right? Except when it does.

In my opinion, it is high time that organizations realize that compliance is not security. Compliance is a false goal set in sand. The real goal is risk management and data protection. In order to accomplish these goals, you have to make rational decisions and account for real threats, not just checklists compiled by some nebulous group of people in a “one size fits all fashion”. That is a fool’s errand.

As I have been saying for a while now, we have to start thinking differently about security. We have to forget the baselines and look at our risk from the view of a threat agent (a hacker, cyber-criminal, attacker, whatever!). We have to make rational choices that really do protect that which needs to be protected. We have to hope for the best and architect for abject failure. Anything less than that, and this is a story you we will just get to keep on telling….

Interested in learning more about “sniffing”? Click here for a great FAQ.

I also did an interview with Secure Computing Magazine about this. You can read that here.

Toata Moves On To Additional Targets

The Toata bot army has moved on to scanning for additional web-applications to target/catalog. Medium levels of scanning began last night and continue today. The new targets are:

/mantisbt/login_page.php

/tracker/login_page.php

/bugtracker/login_page.php

/bugtrack/login_page.php

/support/login_page.php

/bug/login_page.php

/bugs/login_page.php

/login_page.php

/statistics

/bin/statistics

/twiki/bin/statistics

/wiki/bin/statistics

/wikis/bin/statistics

/cgi-bin/twiki/bin/statistics

/cgi-bin/wiki/bin/statistics

/cgi-bin/wikis/bin/statistics

Check your systems to see if you have these files, if so, check with the responsible projects for updates. Consider additional monitoring and/or removal from service. Investigations should be performed, exploitation timelines and goals are unknown. It appears that Mantis Bugtracker and Twiki are the likely targets. Exploit vectors have not been researched at this time, though Mantis has had known XSS in the login page previously.

Our HoneyPoint Internet Threat Monitoring Environment (HITME) is tracking the scans, sources and payload evolutions. SANS and other groups have been notified.

More Toata Scans for a New RoundCube File

Last night, HITME began to pick up various sources scanning for a new file in the RoundCube Webmail product. The file “list.js” is being scanned for by the Toata bot and low levels of port 80 scans matching these probes are ongoing. SANS and the project owners have been informed.

No exploitation has been observed by us thus far in relationship to these scans, but cataloging is ongoing. Intent of the attacker is currently unknown, as is the vulnerability, if any, present in the file.

Following are the signatures captured from one host:

XXX received an alert from 88.191.50.206 at 2009-01-15 19:55:41 on port 80

Alert Data: GET /rc/program/js/list.js HTTP/1.1

Accept: */*

Accept-Language: en-us

Accept-Encoding: gzip, deflate

User-Agent: Toata dragostea mea pentru diavola

Host: xx.xx.xx.xx

Connection: Close

XXX received an alert from 88.191.50.206 at 2009-01-15 19:55:39 on port 80

Alert Data: GET /roundcubemail/program/js/list.js HTTP/1.1

Accept: */*

Accept-Language: en-us

Accept-Encoding: gzip, deflate

User-Agent: Toata dragostea mea pentru diavola

Host: xx.xx.xx.xx

Connection: Close

XXX received an alert from 88.191.50.206 at 2009-01-15 19:55:38 on port 80

Alert Data: GET /roundcube/program/js/list.js HTTP/1.1

Accept: */*

Accept-Language: en-us

Accept-Encoding: gzip, deflate

User-Agent: Toata dragostea mea pentru diavola

Host: xx.xx.xx.xx

Connection: Close

XXX received an alert from 88.191.50.206 at 2009-01-15 19:55:36 on port 80

Alert Data: GET /webmail/program/js/list.js HTTP/1.1

Accept: */*

Accept-Language: en-us

Accept-Encoding: gzip, deflate

User-Agent: Toata dragostea mea pentru diavola

Host: xx.xx.xx.xx

Connection: Close

XXX received an alert from 88.191.50.206 at 2009-01-15 19:55:35 on port 80

Alert Data: GET /email/program/js/list.js HTTP/1.1

Accept: */*

Accept-Language: en-us

Accept-Encoding: gzip, deflate

User-Agent: Toata dragostea mea pentru diavola

Host: xx.xx.xx.xx

Connection: Close

Once again, users of RoundCube Webmail are urged to ensure they are doing additional levels of monitoring, staying current on all patches/updates and taking other precautions. Consider removing RoundCube from Internet exposure until these and other ongoing issues are mitigated.

SANS Posts Info on Previous RoundCube Vulnerability

Looks like our work got more folks looking at RoundCube. SANS Storm Center has a posting that shows the exploit being used by attackers against the helpnetsecurity announced vulnerability in “html2text”.

The RoundCube folks have already released patches and done code cleanup to remove this and other known issues, including the msgimport.sh scripts from previous versions.

If you are a RoundCube user, please upgrade. Scans have slowed for this issue, but are still present and active at low levels.

Thanks to everyone who helped on this and to the RoundCube Webmail project team for their friendly, open approach to solving the problems and their rapid attention. It is refreshing to work with developers who are focused on solutions instead of wanting to fight about the source of the problems. Hats off to them!

PHP Threats Continue to Rise But More Work & Education Could Help

Threats against web applications developed in PHP continue to be an area of high activity and interest for attackers. PHP applications now represent a significant portion of the web-application attack footprints we see in our HoneyPoint Internet Threat Monitoring Environment (HITME). PHP scans and probes for new and emerging vulnerabilities are a common occurrence and one the driving forces behind our deployment of the HITME. Our unique insights into ongoing threat activities allows our vulnerability management and professional services clients to know that they are better protected, even against bleeding edge threats.

PHP security issues are so common that the folks at BreakingPoint Labs call it “one of the most commonly attacked pieces of software on the Internet today”. Even when deployed in so called, “safe mode”, PHP applications can still present a high level of risk. Until, at least, the release and wide scale adoption of PHP 6, issues are likely to continue to abound, maybe even beyond that if the attacker underground has anything to say about it.

PHP security problems also represent a major portion of known web vulnerabilities, especially over the course of 2008. Syhunt, the makers of Sandcat Pro, a web application vulnerability scanner and partner to MSI, has even created Sandcat4PHP, a special source code scanner to help organizations proactively secure their PHP applications during development. Recently, Syhunt created these images that show the impact that PHP vulnerabilities are having on their work. PHP security issues represent an overwhelming margin of their work for the year.

All of this is not to say that PHP development is a bad thing. In fact, PHP developed applications have empowered many new cutting edge applications, fueled the growth of web 2.0 and been a powerhouse for bringing average users the web maturity that they have come to expect. Combining the ease of PHP with the power of MySQL, Apache and other open source tools has become a virtual standard for the online world. PHP applications CAN BE DONE SECURELY, they just require additional work and effort to create secure code, just like any other language. The ease of PHP makes it a great language for learning development, but we, as a community, need to help even those budding developers among us learn the basics of creating secure code. Techniques like input validation, proper sanitization, strong authentication and role-based access controls need to be a core part of our outreach teaching to developers.

In the meantime, while education is being worked on, it might be a wise idea to take a check around your environment and audit any PHP applications in production or planned for use in the near future. Additional work, tools or monitoring may be required to better handle the risk you find. Let us know if we can be of any help or if you desire additional insight into PHP security problems. Keep your eyes on PHP, though, its powerful, flexible capabilities make it a big player in the future of the web!

** Have feedback on this post? Please feel free to leave a comment, drop me a line via email or send me a tweet to @lbhuston on twitter. Thanks for reading! **

Round Cube Webmail Probes Spreading Rapidly

Our HoneyPoint Security Server deployment has identified a set of 0-day scans and probes against the Round Cube Webmail system.

The probes are originating from infected Linux systems world wide and appear to be spreading rapidly. Infection of systems via a bot-net client or other form of malware is likely. The extent of compromise is currently unknown, but complete compromise or escalation to complete compromise may be possible.

Research and work with the developers is ongoing. Users of Round Cube Webmail systems should take steps to remove their systems from Internet access and/or implement additional controls for monitoring and protection. Removal of the msgimport.sh script file is highly encouraged, though additional entry points may emerge in the future.

New versions of the application may not have the msgimport.sh file present.

The current version of the attack is probing for the following files:

/nonexistenshit

/mail/bin/msgimport

/bin/msgimport

/rc/bin/msgimport

/roundcube/bin/msgimport

/webmail/bin/msgimport

Our HoneyPoint deployment has been reconfigured to trap additional data about this threat and additional information may be available soon. The MSI technical team is working with our clients to ensure they are protected against this and other emerging threats. Our threat detection capability, provided to us by our HoneyPoint line of products gives us uniquely deep insight and visibility into bleeding edge threats. As always, we strive to use that knowledge to protect our clients and the Internet at large.

More information can be found on this issue by following @lbhuston and/or @honeypoint on Twitter. You can also check back on our blog or schedule a call with one of our team members if you have additional needs.

** Update: @around 2:30pm Eastern, the “Toata” bot-net added the signature to its scans as well. In less than 24 hours there are now at least 2 known bot-nets scanning for the issue. Any bets on how long it will take before “morfeus” scans for it too??? Also, note that the URL request from “Toata” has a double // typo in it….

** Another Update: Syhunt has added tests to Sandcat for the issue. They are now available via update mechanism for clients.

Playing with Plugins for HoneyPoint

I have been playing with various plugins lately for HoneyPoint. In this case, I wanted to show the output of two plugins I am playing with currently.

The first one is the TweetCLI plugin that I have written about before. In this example, I am going to show an event that has come in and what the plugins did for me.

The TweetCLI plugin posted the following to the @HoneyPoint feed on Twitter:

Suspicious Activity Captured From: 41.205.122.150 on port 23

Then, the console also executed a plugin I lovingly call AutoPoke. It basically does a whois look up of the address and performs a basic nmap TCP port scan of a few common ports. This produced the following output:

OrgName: African Network Information Center

OrgID: AFRINIC

Address: 03B3 – 3rd Floor – Ebene Cyber Tower

Address: Cyber City

Address: Ebene

Address: Mauritius

City: Ebene

StateProv:

PostalCode: 0001

Country: MU

ReferralServer: whois://whois.afrinic.net

NetRange: 41.0.0.0 – 41.255.255.255

CIDR: 41.0.0.0/8

NetName: NET41

NetHandle: NET-41-0-0-0-1

Parent:

NetType: Allocated to AfriNIC

NameServer: NS1.AFRINIC.NET

NameServer: NS-SEC.RIPE.NET

NameServer: NS.LACNIC.NET

NameServer: TINNIE.ARIN.NET

Comment:

RegDate: 2005-04-12

Updated: 2005-07-12

OrgAbuseHandle: GENER11-ARIN

OrgAbuseName: Generic POC

OrgAbusePhone: +230 4666616

OrgAbuseEmail: abusepoc@afrinic.net

OrgTechHandle: GENER11-ARIN

OrgTechName: Generic POC

OrgTechPhone: +230 4666616

OrgTechEmail: abusepoc@afrinic.net

# ARIN WHOIS database, last updated 2008-12-29 19:10

# Enter ? for additional hints on searching ARIN’s WHOIS database.

Starting Nmap 4.68 ( http://nmap.org ) at 2008-12-30 xxx AST

Interesting ports on 41.205.122.150:

PORT STATE SERVICE

21/tcp open ftp

22/tcp open ssh

23/tcp filtered telnet

25/tcp closed smtp

79/tcp closed finger

80/tcp filtered http

110/tcp closed pop3

135/tcp filtered msrpc

136/tcp closed profile

137/tcp closed netbios-ns

138/tcp closed netbios-dgm

139/tcp filtered netbios-ssn

443/tcp closed https

445/tcp filtered microsoft-ds

1433/tcp closed ms-sql-s

3389/tcp closed ms-term-serv

5800/tcp closed vnc-http

5801/tcp closed vnc-http-1

5900/tcp closed vnc

5901/tcp closed vnc-1

6666/tcp closed irc

6667/tcp closed irc

6668/tcp closed irc

6669/tcp closed irc

Nmap done: 1 IP address (1 host up) scanned in 2.330 seconds

This output is kind of fun (at least to me) to watch. I get real time info about where scans and probes are coming from. I also get real time port info from the scanning hosts. Over time, this gives me some pretty interesting insight into common postures of hosts that appear to be compromised or infected.

In this case, this particular host was interesting because of the source. Our global HoneyPoint deployments don’t see too many offending hosts from this particular region. Over time, if I see more activity originating from there or the like, then I can decide if the threat levels in that area are increasing, but none the less, even this first one is interesting. A quick review of the host shows a likely vulnerable ssh deployment, which may indicate that the host is compromised and/or bot-net infected. Of course, this is all supposition, but interesting (to me) anyway.

Now you know how I spend my time. I love to watch the ebb and flow of attacks, probes and scans. I like to know the sources and virtual “look and feel” of the victim systems. I suppose that is where many of the capabilities in HoneyPoint come from. I think they are just toys that I would like to play with, thus they end up in the product. Do you have some plugins you would like to see or some new HoneyPoint toys or functions you would enjoy? If so, drop me a line. We are working on the plans for HPSS 3.xx as we speak, so now would be a great time to hear a want list from the public!

Thanks for reading!