Skip to content
Abstraction layer for Xilinx FPGAs
Python Makefile
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
bal_xilinx Update configs Aug 9, 2019
docs/src Clean up RST source files Aug 8, 2019
examples Initial commit Aug 8, 2019
tests Initial commit Aug 8, 2019
.gitignore Initial commit Aug 8, 2019
MANIFEST.in Update configs Aug 9, 2019
Makefile Initial commit Aug 8, 2019
README.md Add BlackHat 2019 presentation link. Aug 9, 2019
setup.py Update configs Aug 9, 2019

README.md

BAL Xilinx Overview

The BAL Xilinx package is an implementation of the Binary Abstraction Layer (BAL) framework for Xilinx FPGA. It supports packing/unpacking of most of the bitstream, target device/encryption detection and pin modification (force the pin high/low). It is supposed to be a dependency of a project, not a boilerplate project to be customized. It was first introduced at Black Hat 2019 (presentation, research paper).

The full api documentation is available here.

Contributors:

  • Jatin Kataria (Researcher)
  • Rick Housley (Researcher)
  • Alex Massonneau (Developer)
  • Andrew O'Brien (Developer)

Installation

The BAL Xilinx package can be installed from PyPi with the following command:

pip install bal-xilinx

The BAL Xilinx package can also be installed from the repository. It requires the BAL package. See the installation instructions.

To install the BAL Xilinx package from source, run:

git clone https://github.com/RedBalloonShenanigans/bal-xilinx.git
cd bal-xilinx
pip install .

To generate the documentation, run:

pip install .[docs]
make html-docs

Tools

While this module is meant to be used as a dependency for another project, it currently offers a single tool called pin. It turns on/off a given pin for an existing Xilinx FPGA bitstream. It is accessible by running:

python -m bal_xilinx.tools.pin 

The documentation for the pin tool can be accessed by running it with the -h flag.

Methodology

The Xilinx converters rely heavily on format definitions contained in bal_xilinx/configs. The methodology used to create these JSON files will be published shortly.

Guide

This guide assumes familiarity with the BAL framework. As the name of the project implies, this is an implementation of it for Xilinx FPGA bitstreams. Complete examples are available under ./example.

Analyzers

There are 3 analyzers available:

  • bal_xilinx.analyzers.device_analyzer.XilinxDeviceAnalyzer Determines the type of device targeted by the bitstream.
  • bal_xilinx.analyzers.encryption_analyzer.XilinxEncryptionAnalyzer Determines if the FDRI packets are encrypted.
  • bal_xilinx.analyzers.visualizer_analyzer.XilinxVisualizerAnalyzer Generate the configuration data for the BAL visualizer.

Modifiers

There is currently a single analyzer:

  • bal_xilinx.modifiers.pin_modifier.XilinxPinModifer Force a pin to be low/high regardless of the logic executed by the FPGA.

Examples

Here is an example that puts together all the analyzers and modifiers available.

# Configure the default converters
xilinx_context_factory = default_xilinx_context(XilinxContextFactory(
    # Register the formats for the LX9 and LX45T
    default_xilinx_formats(XilinxFormatBuilder()).build()
))

lx9_bin = wget.download('https://redballoonsecurity.com/files/JwfEU4veQSNFao8h/lx9.bin')
with open(lx9_bin, "rb") as f:
    data = f.read()
    
bitstream_context = xilinx_context_factory.create(data)

# Get the device type targeted by the bitstream
device_type = bitstream_context.create_analyzer(XilinxDeviceAnalyzer)\
    .analyze()
print("Device info: {}".format(device_type))

# Determine if the FDRI packets are encrypted.
is_encrypted = bitstream_context.create_analyzer(XilinxEncryptionAnalyzer)\
    .analyze()
print("Is encrypted: {}".format(is_encrypted))

# Unpack the entire bitstream to make sure the visualization includes everything
bitstream_context.get_data().unpack_all()

print("Writing the visualizer config to {}".format(os.path.abspath("data.json")))
with open("data.json", "w") as f:
    # Generate the visualization config file
    visualizer_config = bitstream_context.create_analyzer(VisualizerAnalyzer) \
        .analyze()
    json.dump(visualizer_config,  f)

print("Pulling the pins low")
pin_modifier = bitstream_context.create_modifier(XilinxPinModifer)
for pin in ["P134", "P133", "P132", "P131", "P127", "P126", "P124", "P123"]:
    pin_modifier.modify(pin, False)


print("Analysis and modifications done!")

That is it for the guide. Have fun!

You can’t perform that action at this time.