Skip to content

Latest commit

 

History

History

reversing-python

README.md

Reversing Python Problem Creation Walkthrough

Pre-requisites

  1. You have cmgr installed and configured.

    • Refer to the setup page if this is not the case for you.

  2. You have done the Web CSS Problem Creation Walkthrough.

Overview

This problem uses a container as a service host and presents this service through a port. This problem is adapted from the live picoGym problem, Picker-I.

This problem can be used as a template for any challenge that needs to host a service that players connect to with netcat.

The following walkthrough will touch on new problem files and then look at how to debug a Dockerfile by running it with Docker instead of cmgr.

Walkthrough

File Listing

  1. picker-I.py this is the vulnerable script that we will be hosting as a service on this container.

  2. start.sh starts a listener that receives connections. This script is ran as the last step in the Dockerfile. For this problem, we use socat to connect the output of our vulnerable script to a port, allowing users to interact with our script through the network.

  3. setup-challenge.py This script generates the flag for the problem and saves it in the important file, /challenge/metadata.json, which is required for every cmgr problem.

  4. Dockerfile this is the main setup for our problem. We pull down a pinned Ubuntu image, update it and install the required packages. We create the /challenge directory with specific permissions so only root can access it. /challenge is an important directory and contains files that cmgr needs to deploy a problem. We add artifacts.tar.gz to this directory as well, which contains a copy of the source for the service being run on the container.

NOTE: If you were hosting a vulnerable service that allowed arbitrary remote code execution, you would want to consider separating the challenge building container from the container hosting the service. If players landed a shell on this machine, they would be able to read challenge management data, which usually is not a problem, but is not the greatest look for a CTF competition. See this problem for how to create multiple containers for one challenge.

Debugging your Dockerfile

For typical problem playtest deployment and testing strategy, see this section.

In this walkthrough we will demonstrate how to debug one's Dockerfile. Cmgr does a decent job on relaying the error that it received when it tried to build the Docker container, however, the entire build process is opaque to the cmgr user, and some more control and introspection of the docker build process goes a long way in being able to determine what is going wrong.

  1. Clone this repo

  2. Go to the reversing-python directory

    • cd start-problem-dev/example-problems/reversing-python/

  3. Switch Dockerfiles. Dockerfile.test has 1 discrepancy in it.

    • mv Dockerfile Dockerfile.good
    • mv Dockerfile.test Dockerfile

  4. Build the problem with cmgr

    • cmgr update

    • cmgr playtest picoctf/examples/reversing-python

    • Expected output:

      cmgr: [ERROR:  failed to build image: The command '/bin/sh -c python3 /challenge/setup-challenge.py' returned a non-zero code: 2]
error: failed to build image: The command '/bin/sh -c python3 /challenge/setup-challenge.py' returned a non-zero code: 2

    • We at least know which step failed, however, we can get much more info by building the container manually with Docker:

  5. Build the container manually.

    • docker build . --build-arg FLAG=picoCTF{deadbeef}

    • Expected output (tail end):

      > [7/9] RUN python3 /challenge/setup-challenge.py:
#0 0.333 python3: can't open file '/challenge/setup-challenge.py': [Errno 2] No such
file or directory

  6. Fix the issue.

    • RUN python3 /challenge/setup-challenge.py needs to be RUN python3 setup-challenge.py

Conclusion

With this walkthrough, we created an advanced problem that used a Dockerfile to create a container that presented a service to be interacted with in the shell.

This walkthrough also demonstrated how to gain more debugging insight into cmgr problems by building the container manually with docker which provides a lot more information about build failures.

Next problem

Return to the index