bap

Binary Analysis Platform

2,061

273

2,061

View on GitHub

Top Related Projects

angr

7,537

A powerful and user-friendly binary analysis platform!

radare2

20,547

UNIX-like reverse engineering framework and command-line toolset

Capstone disassembly/disassembler framework for ARM, ARM64 (ARMv8), Alpha, BPF, Ethereum VM, HPPA, LoongArch, M68K, M680X, Mips, MOS65XX, PPC, RISC-V(rv32G/rv64G), SH, Sparc, SystemZ, TMS320C64X, TriCore, Webassembly, XCore and X86.

retdec

7,992

RetDec is a retargetable machine-code decompiler based on LLVM.

frida

15,995

Clone this repo to build Frida

Quick Overview

BAP (Binary Analysis Platform) is an open-source framework for binary analysis and reverse engineering. It provides a set of tools and libraries for analyzing machine code, disassembling binaries, and performing various types of program analysis. BAP supports multiple architectures and can be used for tasks such as vulnerability discovery, malware analysis, and program understanding.

Pros

Supports multiple architectures (x86, x86-64, ARM, MIPS, and more)
Extensible plugin system for customizing and adding new functionality
Provides a high-level intermediate representation (IR) for easier analysis
Integrates with popular reverse engineering tools like IDA Pro and Ghidra

Cons

Steep learning curve for beginners due to its complexity
Documentation can be sparse or outdated in some areas
Performance can be slower compared to some native disassemblers
Limited support for some less common architectures or file formats

Code Examples

Disassembling a binary:

open Bap.Std

let main () =
  let filename = "path/to/binary" in
  Project.from_file filename |> function
  | Ok project ->
      let disasm = Project.disasm project in
      Disasm.insns disasm |> Seq.iter ~f:(fun (mem, insn) ->
        printf "%a: %a\n" Memory.pp mem Insn.pp insn
      )
  | Error err -> printf "Error: %a\n" Error.pp err

let () = main ()

Lifting binary code to BAP IR:

open Bap.Std

let lift_to_ir project =
  let program = Project.program project in
  Term.enum sub_t program |> Seq.iter ~f:(fun sub ->
    printf "Sub: %s\n" (Sub.name sub);
    Term.enum blk_t sub |> Seq.iter ~f:(fun blk ->
      printf "  Block:\n";
      Term.enum def_t blk |> Seq.iter ~f:(fun def ->
        printf "    %a\n" Def.pp def
      )
    )
  )

let () =
  match Project.from_file "path/to/binary" with
  | Ok project -> lift_to_ir project
  | Error err -> printf "Error: %a\n" Error.pp err

Finding function calls:

open Bap.Std

let find_calls project =
  let program = Project.program project in
  Term.enum sub_t program |> Seq.iter ~f:(fun sub ->
    Term.enum blk_t sub |> Seq.iter ~f:(fun blk ->
      Term.enum jmp_t blk |> Seq.iter ~f:(fun jmp ->
        match Jmp.kind jmp with
        | Call c -> printf "Call at %a: %a\n" Addr.pp (Term.tid jmp |> Tid.name) Call.pp c
        | _ -> ()
      )
    )
  )

let () =
  match Project.from_file "path/to/binary" with
  | Ok project -> find_calls project
  | Error err -> printf "Error: %a\n" Error.pp err

Getting Started

To get started with BAP:

Install OPAM (OCaml Package Manager)
Install BAP using OPAM:
```
opam install bap
```
Create a new OCaml project and add BAP as a dependency in your dune file:
```
(executable
 (name your_project)
 (libraries bap))
```
Use the BAP library in your OCaml code as shown in the examples above.

Compile and run your project using dune:

dune build
dune exec ./your_project.exe

Competitor Comparisons

angr

7,537

A powerful and user-friendly binary analysis platform!

Pros of angr

More extensive documentation and tutorials
Larger and more active community support
Broader range of analysis capabilities, including symbolic execution

Cons of angr

Steeper learning curve for beginners
Higher resource consumption for complex analyses
Less focus on low-level binary details compared to BAP

Code Comparison

angr example:

import angr

proj = angr.Project('binary')
state = proj.factory.entry_state()
simgr = proj.factory.simulation_manager(state)
simgr.explore(find=0x400000)

BAP example:

open Bap.Std

let main proj =
  let prog = Project.program proj in
  let entry = Term.entry prog in
  let cfg = Program.cfg prog in
  Graphlib.depth_first_search (module Graphs.Cfg) ~start:entry cfg

Both examples demonstrate basic project setup and analysis initiation. angr uses Python with a more high-level approach, while BAP employs OCaml with a focus on low-level program representation. angr's code is more concise and intuitive for many users, but BAP offers finer control over binary analysis details.

radare2

20,547

UNIX-like reverse engineering framework and command-line toolset

Pros of radare2

More extensive and mature codebase with a larger community and ecosystem
Supports a wider range of architectures and file formats
Offers a more comprehensive set of features for reverse engineering and binary analysis

Cons of radare2

Steeper learning curve due to its complex command-line interface
Less focus on formal methods and academic research compared to BAP
Can be more challenging to integrate into automated workflows

Code Comparison

radare2:

RCore *core = r_core_new();
r_core_cmd_str(core, "aaa");
char *disasm = r_core_cmd_str(core, "pd 10");
printf("%s", disasm);
r_core_free(core);

BAP:

let proj = Project.create ~filename:"binary" () in
let disasm = Project.disasm proj in
Disasm.insns disasm |> Seq.iter ~f:(fun (mem, insn) ->
  printf "%a: %a\n" Memory.pp mem Insn.pp insn
);

Both examples demonstrate basic disassembly functionality, with radare2 using a C API and BAP using OCaml. radare2's approach is more imperative, while BAP's is more functional and type-safe.

capstone

7,542

Pros of Capstone

Lightweight and focused solely on disassembly
Supports a wide range of architectures (x86, ARM, MIPS, PowerPC, etc.)
Easy to integrate into existing projects due to its simplicity

Cons of Capstone

Limited to disassembly only, lacking advanced analysis features
Less comprehensive documentation compared to BAP
Smaller community and fewer extensions/plugins

Code Comparison

Capstone (C):

csh handle;
cs_insn *insn;
size_t count;

if (cs_open(CS_ARCH_X86, CS_MODE_64, &handle) != CS_ERR_OK)
    return -1;
count = cs_disasm(handle, code, code_size, address, 0, &insn);

BAP (OCaml):

let proj = Project.create ~rooter:"byteweight" ~loader:"llvm" filename in
let prog = Project.program proj in
let insns = Disasm.insns prog in
Seq.iter ~f:(fun (mem, insn) ->
  printf "%a: %a\n" Memory.pp mem Insn.pp insn
) insns

This comparison highlights the simplicity of Capstone for basic disassembly tasks, while BAP offers a more comprehensive approach with additional analysis capabilities. Capstone is ideal for projects requiring quick and efficient disassembly, whereas BAP is better suited for complex binary analysis tasks.

retdec

7,992

RetDec is a retargetable machine-code decompiler based on LLVM.

Pros of RetDec

More user-friendly with a graphical user interface (GUI) for easier interaction
Supports a wider range of architectures, including ARM, MIPS, and PIC32
Provides detailed documentation and examples for users

Cons of RetDec

Less extensible compared to BAP's plugin system
Slower decompilation process, especially for large binaries
Limited support for advanced analysis techniques like symbolic execution

Code Comparison

RetDec (C++):

retdec::config::Config config;
config.setInputFile(inputPath);
config.setOutputFile(outputPath);
auto decompiler = retdec::decompiler::createDecompiler(config);
decompiler->run();

BAP (OCaml):

let main () =
  Project.from_file input >>= fun project ->
  Project.with_program project ~f:(fun program ->
    Program.to_graph program |>
    Graph.to_dot |>
    print_endline)

Both repositories provide powerful binary analysis capabilities, but they cater to different user needs. RetDec offers a more accessible approach with its GUI and broader architecture support, making it suitable for users who prefer a streamlined experience. BAP, on the other hand, provides a more flexible and extensible platform for advanced users and researchers who require customizable analysis tools.

frida

15,995

Clone this repo to build Frida

Pros of Frida

Dynamic instrumentation: Allows real-time modification and analysis of running processes
Cross-platform support: Works on multiple operating systems and architectures
Scripting capabilities: Provides a powerful JavaScript API for customization

Cons of Frida

Limited static analysis capabilities compared to BAP
May require more manual effort for complex analysis tasks
Potential performance overhead due to runtime instrumentation

Code Comparison

Frida (JavaScript):

Interceptor.attach(ptr(0x1234), {
  onEnter: function(args) {
    console.log("Function called with args:", args[0], args[1]);
  }
});

BAP (OCaml):

let main proj =
  let open Bap.Std in
  Project.memory proj |> Memmap.to_sequence
  |> Seq.iter ~f:(fun (mem, _) ->
    printf "Memory region: %a\n" Memory.pp mem)

Key Differences

Frida focuses on dynamic analysis and runtime manipulation
BAP specializes in static analysis and program transformation
Frida uses JavaScript for scripting, while BAP uses OCaml
BAP provides more comprehensive binary analysis capabilities
Frida offers easier integration with existing tools and workflows

Convert designs to code with AI

Introducing Visual Copilot: A new AI model to turn Figma designs to high quality code using your components.

Try Visual Copilot

README

Binary Analysis Platform

Overview
Installation
Using
Learning
Contributing
Sponsors

Overview

The Carnegie Mellon University Binary Analysis Platform (CMU BAP) is a suite of utilities and libraries that enables analysis of binary programs. BAP supports x86, x86-64, ARM, MIPS, PowerPC and new architectures can be added using plugins. BAP includes various analyses, standard interpreter, microexecution interpreter, and a symbolic executor. BAP features its own domain-specific language, Primus Lisp, that is used for implementing analyses, specifying verification conditions, modeling functions (writing stubs), and even interfacing with the SMT solver. The toolkit repository includes various examples of program analysis tools that could be implemented with BAP and can be used as the starting point (in addition to the tutorial) for implementing custom analyses. BAP can be used as a framework with a single bap utility that is extended with plugins or it can be used as a library embedded in a user application, which could be written in OCaml or, in any other language, using C bindings. We also provide some minimal support for Python to make it easier to start learning BAP.

BAP was developed in CMU, Cylab and is sponsored by grants from the United States Department of Defense, Siemens, Boeing, ForAllSecure, and the Korea government, see sponsors for more information. BAP is used in various institutions and serves as a backbone for many interesting projects, some are highlighted below:

Installation

Using pre-build packages

We provide binary packages packed for Debian and Red Hat derivatives. For other distributions we provide tgz archives. To install bap on a Debian derivative:

wget https://github.com/BinaryAnalysisPlatform/bap/releases/download/v2.5.0/{bap,libbap,libbap-dev}_2.5.0.deb
sudo dpkg -i {bap,libbap,libbap-dev}_2.5.0.deb

From sources

Our binary packages do not include the OCaml development environment. If you are going to write an analysis in OCaml you need to install BAP from the source code using either opam or by cloning and building this repository directly. The opam method is the recommended one. Once it is installed the following three commands should install the platform in a newly created switch.

opam init --comp=4.14.1   # inits opam and install the OCaml compiler
opam install bap          # installs bap and its dependencies
eval $(opam env)`         # activates opam environment

Or, if you already have a switch where you would like install bap, then just do

opam install bap

The opam install bap command will also try to install the system dependencies of BAP using your operating system package manager. If it fails due to a missing system dependency, try installing it manually and then repeat the opam install bap command. If it still doesn't work, do not hesitate to drop by our chat and seek help there. It is manned with friendly people that will be happy to help.

The instruction above will get you the latest stable release of BAP. If you're interested in our rolling releases, which are automatically updated every time a commit to the master branch happens, then you can create a new switch that uses our testing repository

opam switch create bap-testing --repos \
    default,bap=git+https://github.com/BinaryAnalysisPlatform/opam-repository#testing 4.14.1
opam install bap

After it is added, the bap repository will take precedence over the stable repository and you will get the freshly picked BAP packages straight from the farm.

If you want to build BAP manually or just want to tackle with BAP internals, then you can clone this repository and build it manually. We suggest your starting with a fresh environment without BAP being installed, to prevent clashes, or even better to use a local switch, e.g.,

git clone git@github.com:BinaryAnalysisPlatform/bap.git && cd bap
opam switch create . --deps-only
dune build && dune install

The snippet above will clone bap, create a fresh local switch, install the necessary dependencies, including the system one, and, finally, build and install bap with dune. Alternatively, if you already have a switch where you want to build and install bap, you can use

git clone git@github.com:BinaryAnalysisPlatform/bap.git && cd bap
opam install . --deps-only
dune build && dune install

to install bap and its dependencies into the currently selected switch.

Using

BAP, like Docker or Git, is driven by a single command-line utility called bap. Just type bap in your shell and it will print a message which shows BAP capabilities. The disassemble command will take a binary program, disassemble it, lift it into the intermediate architecture agnostic representation, build a control flow graph, and finally apply staged user-defined analysis in a form of disassembling passes. Finally, the --dump option (-d in short) will output the resulting program in the specified format. This is the default command, so you don't even need to specify it, e.g., the following will disassembled and dump the /bin/echo binary on your machine:

bap /bin/echo -d

Note, that unlike objdump this command will build the control flow graph of a program. If you just want to dump each instruction of a binary one after another (the so-called linear sweep disassembler mode), then you can use the objdump command, e.g.,

bap objdump /bin/echo --show-{insn=asm,bil}

If your input is a blob of machine code, not an executable, then you can use the raw loader, e.g.,

bap objdump /bin/echo --loader=raw --raw-base=0x400000 --show-{insn=asm,bil}

The raw loader takes a few parameters, like offsets, lengths, and base addresses, which makes it a swiss-knife that you can use as a can opener for formats that are not known to BAP. The raw loader works for all commands that open files, e.g., if the raw loader is used together with the disassemble command, BAP will still automatically identify function starts and build a suitable CFG without even knowing where the code is in the binary,

bap /bin/echo --loader=raw --raw-base=0x400000 -d

If you would like to play manually with bytes, e.g., type the instruction encoding manually and see how BAP disassembles it and what semantics it has, then mc is the command you're looking for. It is named for the corresponding utility in LLVM and stands for machine code and has the same interface as the objdump command except that it takes an ASCII encoding of instruction instead of a binary file, e.g.,

bap mc --show-{insn=asm,bil} -- 48 83 ec 08

bap mc --show-{insn=asm,bil} "\x48\x83\xec\x08"

It recognizes a few input formats (including llvm-mc is using for its -show-encoding option). Consult the documentation for more detailed information.

Extending

Writing your own analysis

BAP is a plugin-based framework and if you want to develop a new analysis you can write a plugin, build it, install, and it will work with the rest of the BAP without any recompilation. There are many extension points that you could use to add new analysis, change existing, or even build your own applications. We will start with a simple example, that registers a disassembling pass to the disassemble command. Suppose that we want to write an analysis that estimates the ratio of jump instructions to the total number of instructions in the binary. We will start by creating an empty file named jmp.ml in an empty folder (the folder name doesn't matter). Next, using our favorite text editor we will put the following code into it:

open Core_kernel
open Bap_main
open Bap.Std

let counter = object
  inherit [int * int] Term.visitor
  method! enter_term _ _ (jmps,total) = jmps,total+1
  method! enter_jmp _ (jmps,total) = jmps+1,total
end

let main proj =
  let jmps,total = counter#run (Project.program proj) (0,0) in
  printf "ratio = %d/%d = %g\n" jmps total (float jmps /. float total)

let () = Extension.declare @@ fun _ctxt ->
   Project.register_pass' main;
   Ok ()

Now we can build, install, and run our analysis using the following commands:

bapbuild jmp.plugin
bapbundle install jmp.plugin
bap /bin/echo --pass=jmp

Let's briefly go through the code. The counter object is a visitor that has the state consisting of a pair of counters. The first counter keeps track of the number of jmp terms, and the second counter is incremented every time we enter any term. The main function just runs the counter and prints the output. We declare our extension use the Extension.declare function from the Bap_main library. An extension is just a function that receives the context (which could be used to obtain configuration parameters). In this function, we register our main function as a pass using the Project.register_pass function.

A little bit more complex example, as well as an example that uses Python, can be found in our tutorial.

Building a plugin with dune

You can also build and install bap plugins using dune. For that, you need to define a library and use the plugin stanza that uses this library. Below is the template dune file,

(library
 (name FOO)
 (public_name OUR-FOO.plugin)
 (libraries bap bap-main))

(plugin
 (name FOO)
 (package OUR-FOO)
 (libraries OUR-FOO.plugin)
 (site (bap-common plugins)))

Eveything that is capitalized in the above snippet is a placeholder that you shall substitute with appropriate private and public names for your plugin. Notice, that the .plugin extension is not necessary, but is percieved as a good convention.

Interactive REPL

BAP also ships an interactive toplevel utility baptop. This is a shell-like utility that interactively evaluates OCaml expressions and prints their values. It will load BAP libraries and initialize all plugins for you, so you can interactively explore the vast world of BAP. The baptop utility can also serve as a non-interactive interpreter, so that you can run your OCaml scripts, e.g., baptop myscript.ml or you can even specify it using sha-bang at the top of your file, e.g., #!/usr/bin/env baptop. We built baptop using UTop, but you can easily use any other OCaml toplevel, including ocaml itself, just load the bap.top library, e.g., for vanilla ocaml toplevel use the following directives

#use "topfind";;
#require "bap.top";;

Learning

We understand that BAP is huge and it is easy to get lost. We're working constantly on improving documentation ensuring that every single function in BAP API is thoroughly documented. But writing higher-level guidelines in the form of manuals or tutorials is much harder and very time consuming, especially given how different the goals of our fellow researchers and users. Therefore we employ a backward-chaining approach and prefer to answer real questions rather than prematurely trying to address all possible questions. We will be happy to see you in your chat that features searchable, indexed by Google, archive.

We are writing, occasionally, to our blog and wiki and are encouraging everyone to contribute to both of them. You can also post your questions on stackoverflow or discuss BAP on the OCaml board. We also have a cute discord channel, which has much less traffic than our gitter.

Contributing

BAP is built by the community and we're welcome all contributions from authors that are willing to share them under the MIT license. If you don't think that your analysis or tool suits this repository (e.g., it has a limited use, not fully ready, doesn't meet our standards, etc), then you can consider contributing to our bap-plugins repository that is a collection of useful BAP plugins that are not mature enough to be included in the main distribution. Alternatively, you can consider extending our toolkit with your tool.

Of course, there is no need to submit your work to one of our repositories. BAP is a plugin-based framework and your code could be hosted anywhere and have any license (including proprietary). If you want to make your work available to the community it would be a good idea to release it via opam.

Top Related Projects

Convert designs to code with AI

Introducing Visual Copilot: A new AI model to turn Figma designs to high quality code using your components.

Try Visual Copilot

Top Related Projects

Quick Overview

Pros

Cons

Code Examples

Getting Started

Competitor Comparisons

Pros of angr

Cons of angr

Code Comparison

Pros of radare2

Cons of radare2

Code Comparison

Pros of Capstone

Cons of Capstone

Code Comparison

Pros of RetDec

Cons of RetDec

Code Comparison

Pros of Frida

Cons of Frida

Code Comparison

Key Differences

Convert designs to code with AI

README

Binary Analysis Platform

Table of contents

Overview

Installation

Using pre-build packages

From sources

Using

Extending

Writing your own analysis

Building a plugin with dune

Interactive REPL

Learning

Contributing

Sponsors

Top Related Projects

Convert designs to code with AI