Over at rbb.antitree.com, you’ll see the details of a new project of mine: To build a Raspberry Pi environment to make it easy for anyone to run a Tor Bridge node. The goal here has been to release an RBP image that is minimalist (in terms of storage consumption as well as resource consumption) and provides the necessary tools to run and maintain a Tor Bridge Node on a Raspberry Pi.

Bridges

A reminder, a Bridge Node is a type of Tor node (like relay, exit, entry) that is a way of evading censorship to join the Tor Network. This is done by secretly hosting bridges that are not shared with the public so there’s no way for a censoring tool to merely block all Tor nodes. On top of that, an Obfuscated Bridge is one that further defends against various fingerprinting attacks of the Tor protocol. With an obfuscated bridge, communications from the client to the bridge appear to be benign traffic rather than Tor traffic.

Challenge Installing Tor

It’s odd how less-than-simple the process of running a relay on a Pi is. If you want to run a relay on a RBP, some sites will merely say install Rasbpian and run apt-get install Tor. The problem with this is that the Debian repos are very far behind from the latest version of Tor (like at least one major revision behind). The logical conclusion would be to use the Tor Project’s debian repo’s then. The problem here is that there are no repos for Rasbperry Pi’s ARM architecture. One solution was to use something similar to the Launchpad PPA hosting that lets you run a simple repo to deliver a .deb package. But launchpad does not support ARM architecture (and doesn’t seem to plan to do so in the near future).

So the result is I’ve built a github repo that hosts the Tor .deb packages for the latest stable release. It’s not pretty, but it does the job and I know that it will work well. That was the first piece of the puzzle.

Host Hardening

The Raspberry Pi images out there are designed for people that want to learn programming in Scratch and play with GPIO pins for some kind of maker project. They’re not ideal for providing a secure operating environment. So I built a Debian-based image from the ground up, with the latest packages and only the required packages. I’ve customized the image to not log anything across reboots (mounting /var/log as a tmpfs). You can read most of the design of the OS here.

I’ve also secured SSH (which many of the Raspberry Pi images don’t do) by autogenerating SSH keys the first time it’s boot. The alternative is to ship an image that has the same SSH keys allowing MITM attacks. Again, these images are designed for makers.

Torpi-config

The part I spent the most time on, and is hopefully the most useful, is I took the structure of the raspi-config tool that is shipped with Raspbian, and I convirted it into a Tor configuration tool. This will give you a text-based wizard to guide users through configuring Tor, keeping obfsproxy up-to-date and perform basic systems administration on the device.

Roadmap

It’s fully functional but there are a lot of things I’d like to improve upon. I’ve released it to solicit feedback and see how much more effort is necessary to get it where I want. Here are some of the other items on the roadmap:

• Add the ability to update Tor to the latest stable release over github (securely)
• Improve torpi-config to cover other use cases like configuring WiFi or a hidden service
• Print out the specific ports that need to be forwarded through the router for the obfuscated bridge
• Clean up some of the OS configuration stuff

 

 

For a few years now, I’ve been stating my plans for the rest of the year, and reviewing how the previous year went. Here’s the review:

Last Year

Major Con Presentation

One of my goals was to be accepted into a “Major Con” for some definition of that word. This year I presented at Derbycon, GRRCon, Defcon Skytalks, BSides Detroit, and the Rochester Security Summit. I will say that my moon shot was to be accepted into 30C3 but I was aptly turned down.

Intel/OSINT/OPSEC Project

This intel/OSINT/OPSEC topic has bugged for a few years now and thanks to Edward Snowden, I think I chose a good year to work on it. I put a ton of research time into formalizing what an intelligence gathering campaign would look like, and even implementing it. I even wrote some tools to help me get the job done. I presented my research at GrrCon, RSS, and BSides Detroit. I’ll admit, it’s a bit of a fluffy, opaque subject to talk about, which is why I really wanted to do the research and be done with it. The output from the research is just some new tools in my aresenal.

Hardware Hack into a PCB

Last year, although I was messing around with my hardware hacking project (that I was asked to take down), I never moved it from a proof-of-concept breadboard, to my own custom circuit. This year, I was able to build a couple circuits and get the fab’d. But I didn’t design them so I still think that’s cheating. Going through the process of loading an Arduinio with AVR software was a big enough step.

iButton Door System

This was a failure. Some of us still talk about it but I didn’t build an iButton door system. I have all t he hardware and Raspberry Pis to do it, I just haven’t put the time in.

Mannequin

My poor mannequin has been around for years. I’ve chopped her head open and loaded her up with an xbee controlled arduino, I’ve made her my T-Shirt model for BSides Rochester, but this was to be the year of her demise. I accomplished this in a fantastic fashion though using Tannerite – an explosive that we packed inside of her.

3D Printing A Model

This was just a fail. I didn’t print anything really. We used a 3D printer to make the badges for BSidesROC this year, but I never actually went through the process myself.

Unplanned Accomplishments

There have been some interesting unplanned accomplishments this year:

• Becoming a minister and performing in someone’s wedding ceremony
• Going to Korea, twice
• Building a silicone brain
• Having a thermite party to destroy all of my old media
• Operating a back hoe
• Receiving my first DMCA request

Next Year

Grown-up Things

This is the year I know I’m going to have to and want to do some what I would call Grown-Up Things. Things that aren’t necessarily about completely full blow chaos and fun. One being learning about how businesses work, forming an LLC, and paying attention to financials. There’s some other things but where’s the fun in discussing that. I just know that this year will be filled with a lot of “Adult” opportunities.

 Crypto

Although I have a decent understanding of crypto, I’d like to put some time in and develop this into a skill. A friend of mine is taking the Stanford Cryptography class and I’m hoping we can learn that together. But beyond that, I’d like to apply it to some actual research. Maybe doing some basic crypto audits of something like BitMessage. I’ll never be a cryptographer, but I’d like to be able to identify and exploit poor cryptographic implementations.

Development

I’m have a decent ability to make something in Python, but it’s all scripting. I’ve never taken a class or anything that would give me any kind of structured development style. My goal for this coming year is to further build my development skills beyond just scripting and hacking things together. Ideally I’d like to join a development team on a project of some kind.

Bitcoin

Gah – Bitcoin… when I say it out loud it sounds so stupid. But this year I’ll be putting time into learning how the bitcoin protocol works, the community that supports it, and slightly riding the roller coaster as it goes up and down. Last year I was doing intelligence when Edward Snowden released all his intel, this year I may be doing Bitcoin when we watch the first crypto currency become regulated.

Hardware RE

This year hasn’t taken me into much hardware reverse engineering lately. I’ll be looking for an interesting project to spend some time on.

Halloween time again. Last year I tried to do a simple little hardware project to make my emoticon pumpkins glow. That’s cute and all but not very difficult.

This year, I decided to work on this idea I’ve had for more than a year and a half. The Brain – a silicone based brain with controllable LEDs inside. I have some ideas of what to do next with it, but this first iteration is just to be a fun decoration for Halloween.

 

Why The Brain

Back when Interlock was moving into their current space, there was this cool area in the center of it, that was surrounded by windows. That turned into the network room and I really wanted to use that window for something. Show something cool in the window or whatever. I came up with this idea that I would have a brain to represent network activity. When a host goes down, the brain reflects that. If there was no Internet connectivity, the brain would show that too. The first version of the brain is not to that level yet, but it’s in the right direction.

Brain 1.0

Brain 1.0 is a Platsil GEL-10, silicone brain with a hollowed center. In the center is a plastic project box housing an Arduino with Neo Pixels attached to it. A Neo Pixel is an Adafruit project that is meant to be a low cost, multi-color LED that you can daisy-chain, or string together in-line. There’s really no reason to use Neo Pixels for this project besides the fact that I had some already.

Parts:

• Halloween Brain Jello mold from Amazon
• Platsil GEL-10 from BITY Mold Supply
• Tupperware container donated to the cause
• XL Breaking Bad meth making gloves
• Mixing containers

Making the Brain:

This was the most interesting part to me. I picked up the type of PlatSil that is a 1 to 1 compound either by volume or by mass so I didn’t need to worry about mixing too much. I took 500ML of A and mixed it with 500ML of B. This stuff has a 6 minute lifetime from the time you start mixing to the time it starts to harden. There are ways to slow this down, but again, I didn’t need to do that. I spent 2 minutes mixing because some guy on YouTube said this is important, and my recent adventures in Thermite taught me the lesson that they’re serious.

Before I poured it in, I used a can of Pol-Ease 2300 release which is used to keep the brain separated from the Jello mold. I was reminded the hard way what happens when you forget this. Pouring it into the mold was pretty simple but I made a small clay holder for it so I could make sure it stayed level. After the contents were dumped in, I sunk the plastic project container that was going to be my hollowed inside.

The whole things hardens within 30 minutes but because mine was in the garage in October, it was more like an hour.

Because this stuff isn’t very cheap, I did a demo mold just to make sure I was on the right track.

PlatSil Gel-10:

My goal was to create a mold of a brain that was rubbery and brain like. This Platsil line of chemicals are designed to create molds for other things. There wasn’t a lot of people making actual things from the material itself but I really like the texture and toughness of using it as the model. I will say that it is 100% overkill for what I wanted. There’s probably someone that can recommend a better, cheaper, alternative but for me this worked in the time frame I needed it to. They have a bunch of different types and I really wanted light to diffuse through it so I got that translucent version. It still comes out pretty white depending on how thick of a mold you’re making.

Neo Pixel:

Neo Pixels are really slick. They have 4 leads on them. Power, Ground, signal in, and a signal out. The biggest benefit is that each pixel is individually addressable without the need for multiple connections. Pixel 0 connects to pixel 1 that connects to pixel N through a single  wire connected to your microcontroller or whatever you’re using.

 

Power takes +5v, and there is a warning about memory consumption especially with smaller Arduinos and extremely long chains of Neo Pixels (up to 500 at 30 FPS). My 4 didn’t mind.

Adafruit has a Neo Pixel library that you can use pretty easily, even if you just want to hack one of their demos.

Arduino:

This is my hacked code to make the brain throb between red and pink. Again, a Neo Pixel is overkill for doing this but it’s fun none-the-less and I’ll be upgrading it next iteration.

https://gist.github.com/antitree/7188144

This is a presentation I gave about embedded security at the last 2600 meeting. This mostly just referencing other people’s work like Joe Grand and Travis Goodspeed who are embedded security gods.