Skip to content

Latest commit

 

History

History
299 lines (240 loc) · 11.3 KB

getting_started.md

File metadata and controls

299 lines (240 loc) · 11.3 KB
Raw

2021 MITRE Collegiate eCTF Getting Started

Version control on your local computer and MITRE server

On your local computer, you can fetch, pull and push as you normally would using git fetch origin and git push origin. If MITRE pushes out updated code, you can fetch this new code using git fetch mitre.

On the MITRE development server, the repo is cloned at ~/2021-ectf-UB-Cacti-design. A deployment key from the server is added to this repo, so you can pull and fetch on the MITRE server but not push.

Requirements

You may skip this step if running on MITRE-provided development servers.

The following packages are required:

  • make
  • docker
  • python3

Setting up Docker

You may skip this step if running on MITRE-provided development servers.

To use Docker without sudo, you need to be in the docker group (from https://docs.docker.com/engine/install/linux-postinstall/).

  1. Create the docker group (if not already done)
sudo groupadd docker
  1. Add your user to the docker group
sudo usermod -aG docker $USER
  1. Re-evaluate user group
newgrp docker
  1. Verify you can run docker commands without sudo
docker run hello-world

Creating and Launching a Deployment

To launch the example echo server and client (and earn the Boot Reference flag), run the following:

NOTE: the values of any arguments to make may be modified to suit your deployment. You should just be aware that if multiple team members are working at the same time on the same system, different SCEWL ID ranges and/or deployment names should be used to avoid naming collisions

1) Create deployment

make create_deployment DEPLOYMENT=echo

This will build a Docker container using dockerfiles/1a_create_sss.Dockerfile and dockerfiles/1b_create_controller_base.Dockerfile, which creates a new deployment with an SSS and a base image for SCEWL Bus Controllers to be built off of. The SSS container will be tagged echo/sss:latest and the base SCEWL Bus Controller container will be tagged echo/controller:base.

Arguments:

  • DEPLOYMENT - the name of the deployment that will prefix all Docker tags

Containers in deployment:

  • echo/sss:latest - Contiains the SSS and any deployment-wide secrets
  • echo/controller:base - Contains a base controller with any common packages or libraries pre-built

2) Add SEDs

make add_sed DEPLOYMENT=echo SED=echo_server SCEWL_ID=10 NAME=echo_server
make add_sed DEPLOYMENT=echo SED=echo_client SCEWL_ID=11 NAME=echo_client CUSTOM='TGT_ID=10'
make add_sed DEPLOYMENT=echo SED=echo_client SCEWL_ID=12 NAME=echo_client CUSTOM='TGT_ID=11'

These three invocations create first the echo server and then two echo clients. Using the below arguments, three Dockerfiles are used to build the echo server and then the echo client. First, 2a_build_cpu.Dockerfile is used to build the user code that will run on each CPUs, which are tagged as the containers echo/cpu:echo_server_10, echo/cpu:echo_client_11, and echo/cpu:echo_client_12. Next, 2b_create_sed_secrets.Dockerfile is used to create device-specific secrets for each SED, modifying and retagging echo/sss:latest. Finally, 2c_build_controller.Dockerfile is used to build the SCEWL Bus Controller, using the secrets stored in echo/sss:latest, and tagging the containers as echo/controller:echo_server_10, echo/controller:echo_client_11, and echo/controller:echo_client_12.

Arguments:

  • DEPLOYMENT - the same as before (here, echo)
  • SED - the directory in cpu/seds/ that contains the SED (here, echo_server and echo_client);
  • SCEWL_ID - the SCEWL ID of the SED (here, 10, 11, and 12)
  • NAME - the name of the SED (here, echo_server and echo_client), which makes the tags of the Docker containers easier to decypher in the future
  • CUSTOM - (Optional) Passes any string to the SED's makefile, in this case defining the SCEWL ID of the echo servers.

Containers in deployment:

  • echo/sss:latest - Contains the SSS, any deployment-wide secrets, and the SED-specific secrets for the echo server and clients
  • echo/controller:base - Contains a base controller with any common packages or libraries pre-built
  • echo/cpu:echo_server_10 - Contains the CPU for the echo server
  • echo/cpu:echo_client_11 - Contains the CPU for the first echo client
  • echo/cpu:echo_client_12 - Contains the CPU for the second echo client
  • echo/controller:echo_server_10 - Contains the SCEWL Bus Controller for the echo server
  • echo/controller:echo_client_11 - Contains the SCEWL Bus Controller for the first echo client
  • echo/controller:echo_client_12 - Contains the SCEWL Bus Controller for the second echo client

3) Remove SEDs

make remove_sed DEPLOYMENT=echo SCEWL_ID=12 NAME=echo_client

This would remove the second echo_client device from the deployment, removing the CPU and SCEWL Bus Controller images and using 3_remove_sed.Dockerfile to remove any SED secrets from the SSS container and retagging it as echo/sss:latest.

Arguments:

  • The arguments DEPLOYMENT, DEVICE_ID, and NAME must be the same as what were given in add_sed for the desired SED.

Containers in deployment:

  • echo/sss:latest - Contiains the SSS, any deployment-wide secrets, and the SED-specific secrets for the echo server and client
  • echo/controller:base - Contains a base controller with any common packages or libraries pre-built
  • echo/cpu:echo_server_10 - Contains the CPU for the echo server
  • echo/cpu:echo_client_11 - Contains the CPU for the first echo client
  • echo/controller:echo_server_10 - Contains the SCEWL Bus Controller for the second echo server
  • echo/controller:echo_client_11 - Contains the SCEWL Bus Controller for the first echo client

4) Launch deployment

make deploy DEPLOYMENT=echo START_ID=10 END_ID=12 SOCK_ROOT=$PWD/socks FAA_SOCK=faa.sock \
    MITM_SOCK=mitm.sock SC_PROBE_SOCK=sc_probe.sock SC_RECVR_SOCK=sc_recvr.sock

This will launch ectf/ectf-radio:latest (from Docker Hub) and echo/sss:latest in the background, deploying the radio waves emulator and SSS, respectively. Once these are both live, they should stay up thoughout the lifetime of the system.

Arguments:

  • DEPLOYMENT - is the name of the deployment from before
  • START_ID and END_ID - the start and end ranges of possible SCEWL IDs of the deployment, respectively (start is inclusive, end is exclusive, so in this case only SCEWL IDs 10 and 11 are expected)
  • SOCK_ROOT - the ABSOLUTE -- NOT RELATIVE path to a directory to bind the backend Unix sockets to
  • FAA_SOCK - The name of the Unix socket for the FAA transceiver to attach to
  • MITM_SOCK - The name of the Unix socket for the MitM transceiver to attach to
  • SC_PROBE_SOCK - The name of the Unix socket for the side-channel probe to attach to
  • SC_RECVR_SOCK - The name of the Unix socket for the side-channel receiver to attach to

Running containers:

  • ectf/ectf-radio:latest - the radio waves emulator
  • echo/sss/:latest - the SSS

5) Launch FAA transceiver

make launch_faa FAA_SOCK=socks/faa.sock

This launches the FAA transceiver, which will open a command prompt to control the transceiver that you may use to interract with the FAA channel of the deployed SEDs. Press <enter> to print any received messages. Type help or ? for detailed commands.

6) Launch SEDs

In another terminal:

make launch_sed_d DEPLOYMENT=echo SCEWL_ID=10 NAME=echo_server SOCK_ROOT=$PWD/socks
make launch_sed DEPLOYMENT=echo SCEWL_ID=11 NAME=echo_client SOCK_ROOT=$PWD/socks

This will launch two SEDs, first the echo server and then the echo client. Arguments are as described above. NOTE: launch_sed is for launching non-interactive SEDs of which you would like to see the output (though you won't be able to input anything through STDIN), launch_sed_d is for launching non-interactive SEDs that will be detached and backgrounded, and launch_sed_i is for launching interactive SEDs, whose STDIN will be attached to the terminal input.

After they spin up, if you press <enter> on the FAA transceiver, you should see a new message with your Boot Reference flag.

If you see Could not connect to /socks/scewl_bus.sock! after the first step, try to 1) remove all the sockets rm -rf ./socks/*; 2) make sure no container was running when you start.

Faster method

Alternatively, you can define each of the static Makefile arguments as environmental variables, reducing repetition and simplifying the invocations. The following is from tools/deploy_echo.sh:

To launch tools/deploy_echo.sh, make sure to do it from the root directory:

export DEPLOYMENT=echo
export SOCK_ROOT=$PWD/socks
export SSS_SOCK=sss.sock
export FAA_SOCK=faa.sock
export MITM_SOCK=mitm.sock
export START_ID=10
export END_ID=12

# create deployment
make create_deployment
make add_sed SED=echo_server SCEWL_ID=10 NAME=echo_server
make add_sed SED=echo_client SCEWL_ID=11 NAME=echo_client CUSTOM='TGT_ID=10'

# launch deployment
make deploy

# launch transceiver in background
python3 tools/faa.py $SOCK_ROOT/$FAA_SOCK &

# launch seds detatched
make launch_sed_d NAME=echo_server SCEWL_ID=10
sleep 1
make launch_sed_d NAME=echo_client SCEWL_ID=11

# bring transceiver back into foreground
fg

Man-in-the-Middle (MitM) Interface

make launch_mitm MITM_SOCK=socks/mitm.sock

The MitM interface (tools/mitm.py) allows you to intercept, modify, inject, or drop packets being sent over the wireless interface, although the reference implementation only records and forwards the messages unmodified. The MitM interface should be launched after the deployment is live, but may be started, closed, and restarted at will (the radio waves emulator should automatically detect if the connection opens or closes, and rout packets accordingly). The MITM_SOCK argument should be the same as what was used in make deploy.

Debugging an SED

To facilitate debugging, we have provided a tool to easily launch an SED and attach GDB to it. To run it, first deploy the deployment (with relevant arguments):

make deploy DEPLOYMENT=<deployment> FAA_SOCK=<faa.sock> MITM_SOCK=<mitm.sock> \
    SOCK_ROOT=<socks> START_ID=<start_id> END_ID=<end_id>

And then launch the sed, automatically attatching GDB to the SCEWL Bus Controller:

make launch_sed_gdb DEPLOYMENT=<deployment> SCEWL_ID=<id> NAME=<name> SOCK_ROOT=<socks>

NOTE: For this to work, your 2c_build_controller.Dockerfile must place the controller's ELF (potentially compiled with the extension .axf) at /controller.elf

Docker tricks

To view all running Docker containers:

docker ps

To kill the Docker container with process ID 12345:

docker kill 12345

To kill all Docker containers (be aware not to kill the containers of others on the server):

docker kill $(docker ps -q)

You can streamline this by adding alias dockerka='docker kill $(docker ps -q)' to your .bashrc.

To run a command in the Docker container test:deployment:

docker run test:deployment echo "this echo command will be run in the container"

Docker can chew up disk space, so if you need to reclaim some, first prune all unused images:

docker image prune -a

If that isn't enough, you can prune all of the containers:

docker container prune

NOTE: these will remove all of the cached containers, so the next builds may take a longer time