Blaze Platform

Monorepo for Blaze, Flint, and binary lifter backends

Running

Using Docker

DOCKER_REGISTRY is an optional environment variable that is used in docker-compose.yaml to refer to a Docker registry that stores the images built by Dockerfile and .gitlab-ci.yml. If you use a Docker registry, set this environment variable (e.g., export DOCKER_REGISTRY=gitlab.example.com:1234) to use the images stored in the registry. Otherwise, leave it unset.

docker-compose run { SERVICE_NAME } [ ARGS... ]

From source

For the most part, stack run { TARGET } [ ARGS... ], but see each package's README.md for more information.

Building

Using Docker

As simple as docker build --platform=linux/amd64 . or docker-compose build. See the comments about DOCKER_REGISTRY in the Running > Using Docker section. You can increase the GHC optimization level with --build-arg OPTIM=-O2 for example (the default is -O0).

For now, Blaze has a hard dependency on both ghidra-haskell as well as binaryninja-haskell, even if you only plan on using one or the other at run-time. We have plans to make these soft dependencies, but what this means is, for now, you need to have both Binary Ninja and Ghidra available. Luckily, Ghidra is a FOSS project, and we can just grab it while building the Docker image. Unfortunately, there isn't any way to download Binary Ninja headlessly¹, so you'll need to provide your own. Simply copy BinaryNinja.zip and license.dat (commercial or headless license) into this directory before running any docker build or docker-compose build/docker-compose up commands. The version of BinaryNinja.zip doesn't matter, only that it's a linux build.

Note

Because of our hard dependency on Binary Ninja, which only provides AMD64 builds for Linux, our whole Docker image is forced to be AMD64. If you run a different architecture (e.g., on Apple Silicon devices), then Docker has to emulate AMD64 (e.g., through QEMU, Rosetta, etc). This impacts build times significantly. For instance, on a 2023 M2 MacBook Pro, building from scratch takes about 40 minutes. You may want to consider using a Docker remote context with an AMD64 machine.

From source

For the most part, stack build { TARGET }, but see each package's README.md for more information.

Footnotes

1. There is the official version switcher script, but that needs an existing installation to use, so we can't bootstrap with it. There is also the official headless download script, but that only works for "headless" licenses, and not for "commercial" licenses. ↩