The Binary Analysis Platform is a reverse engineering and program analysis platform that targets binaries, i.e., compiled programs without the source code. BAP supports multiple architectures (more than 30), though the first tier architectures are x86, x86-64, and ARM. BAP operates by disassembling and lifting the binary code into the RISC-like BAP Instruction Language (BIL). Thus the analysis, implemented in BAP, is architecture independent in a sense that it will work equally well for all the supported architectures. The platform comes with a set of tools, libraries, and plugins. The main purpose of BAP is to provide a toolkit for automated program analysis. BAP is written in OCaml and it is the preferred language to write analysis, we have bindings to C, Python and Rust.
BAP is developed in CMU, Cylab and is sponsored by various grants from the United States Department of Defense, Siemens AG, and the Korea government, see sponsors for more information.
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/v1.3.0/{bap,libbap,libbap-dev}_1.3.0.deb
sudo dpkg -i {bap,libbap,libbap-dev}_1.3.0.deb
The binary release doesn't contain OCaml runtime, and is suitable only if you are not going to extend BAP using OCaml programming language (the recommended way). We recommend to use the OPAM package manager to install BAP and a development environment. After you've successfully installed OPAM, do the following:
opam init --comp=4.03.0 # install the compiler
eval `opam config env` # activate opam environment
opam depext --install bap # install bap
To use a specific version of llvm (we support 3.4, 3.8, and 4.0) specify the desired version explicitly via the conf-llvm
package, e.g.,
opam depext --install conf-llvm=3.8
opam depext --install conf-llvm=3.8 bap
Got any problems? Then visit our troubleshooting page or contact us directly via our Gitter chat.
The BAP main frontend is a command line utility called bap
. You can
use it to explore the binary, run existing analysis, plugin your own
behavior, load traces, and much more.
To dump a program in various formats use the --dump
option (or its short
equivalent, -d
), For example, let's run bap
on the
x86_64-linux-gnu-echo
file.
$ bap testsuite/bin/x86_64-linux-gnu-echo -d | grep 'sub print_endline' -A44
00000334: sub print_endline()
00000301:
00000302: v483 := RBP
00000303: RSP := RSP - 8
00000304: mem := mem with [RSP, el]:u64 <- v483
00000305: RBP := RSP
00000307: RSP := RSP - 0x20
0000030e: mem := mem with [RBP + 0xFFFFFFFFFFFFFFE8, el]:u64 <- RDI
0000030f: RAX := mem[RBP + 0xFFFFFFFFFFFFFFE8, el]:u64
00000310: mem := mem with [RBP + 0xFFFFFFFFFFFFFFF8, el]:u64 <- RAX
00000311: goto %00000312
00000312:
00000313: RAX := mem[RBP + 0xFFFFFFFFFFFFFFF8, el]:u64
00000314: RAX := pad:64[pad:32[mem[RAX]]]
00000315: v545 := low:8[low:32[RAX]]
0000031b: ZF := 0 = v545
0000031c: when ~ZF goto %0000032a
0000031d: goto %0000031e
0000031e:
0000031f: RDI := pad:64[0xA]
00000320: RSP := RSP - 8
00000321: mem := mem with [RSP, el]:u64 <- 0x400731
00000322: call @putchar with return %00000323
00000323:
00000324: RSP := RBP
00000325: RBP := mem[RSP, el]:u64
00000326: RSP := RSP + 8
00000327: v693 := mem[RSP, el]:u64
00000328: RSP := RSP + 8
00000329: return v693
0000032a:
0000032b: RAX := mem[RBP + 0xFFFFFFFFFFFFFFF8, el]:u64
0000032c: RDX := RAX + 1
0000032d: mem := mem with [RBP + 0xFFFFFFFFFFFFFFF8, el]:u64 <- RDX
0000032e: RAX := pad:64[pad:32[mem[RAX]]]
0000032f: RAX := pad:64[extend:32[low:8[low:32[RAX]]]]
00000330: RDI := pad:64[low:32[RAX]]
00000331: RSP := RSP - 8
00000332: mem := mem with [RSP, el]:u64 <- 0x40071C
00000333: call @putchar with return %00000312
By default, the --dump
options used the IR format, but you can
choose from various other formats. Use the --list-formats
option to
get the list. However, formats are provided by plugins, so just
because you don't see your preferred format listed doesn't mean you
can't generate it. Check OPAM for plugins which may provide the format
you want to read (the bap-piqi plugin provides protobuf, xml, and
json, which cover many use cases).
To discover what plugins are currently available, use the
--list-plugins
option. A short description will be printed for each
plugin. The --list-plugins
option also accepts a list of tags, that
will limit the output to plugins that match with the selected tags. For
the list of tags use the --list-tags
option. You can also use
the opam search bap
command, to get the information about bap packages,
available from OPAM.
To get information about a specific plugin named <PLUGIN>
use
the --<PLUGIN>-help
option, e.g., bap --llvm-help
.
The bap
utility works with whole binaries; if you have just few
bytes with which you would like to tackle, then bap-mc
is what you
are looking for.
An idiomatic way of using BAP is to extend it with a plugin. Suppose, you want to write some analysis. For example, let's estimate the ratio of jump instructions to the total amount of instructions (a value that probably correlates with a complexity of a program).
So, let's do it. Create an empty folder, then open your
favorite text editor
and write the following program in a jmp.ml
file:
open Core_kernel.Std
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 () = Project.register_pass' main
Before we run it, let's go through the code. The counter
object is a
visitor that has a 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. Finally, we register it with the
Project.register_pass'
function. Later the function can be invoked
from a command line, and it will get a project data structure, that
contains all the information that was recovered from a binary.
To compile the plugin simply run the following command:
bapbuild jmp.plugin
It is easier to run the pass, if it is installed, so let's do it:
bapbundle install jmp.plugin
Now we can test it:
$ bap /bin/true --pass=jmp
ratio = 974/7514 = 0.129625
$ bap /bin/ls --pass=jmp
ratio = 8917/64557 = 0.138126
OK, If the previous example doesn't make any sense to you, then you
can try our
Python bindings.
Install them with pip install bap
(you still need to install bap
beforehand). Here is the same example, but in Python:
import bap
from bap.adt import Visitor
class Counter(Visitor) :
def __init__(self):
self.jmps = 0
self.total = 0
def enter_Jmp(self,jmp):
self.jmps += 1
def enter_Term(self,t):
self.total += 1
proj = bap.run('/bin/true')
count = Counter()
count.run(proj.program)
print("ratio = {0}/{1} = {2}".format(count.jmps, count.total,
count.jmps/float(count.total)))
The same program in C will take too much space, and will not fit into the README format, but this is an example, of a simple diassembler in C:
#include <stdio.h>
#include <bap.h>
char data[] = "\x48\x8d\x00";
int main(int argc, const char **argv) {
bap_init(argc, argv);
if (bap_load_plugins() < 0) {
fprintf(stderr, "Failed to load BAP plugins\n");
return 1;
}
bap_disasm_basic_t *dis = bap_disasm_basic_create(BAP_ARCH_X86_64);
if (!dis) {
fprintf(stderr, "can't create a disassembler: %s\n", bap_error_get());
}
const int len = sizeof(data) - 1;
bap_code_t *code = bap_disasm_basic_next(dis, data, sizeof(data) - 1, 0x80000);
if (!code) {
fprintf(stderr, "can't disassemble instruction: %s\n", bap_error_get());
return 1;
}
bap_code_print(code);
bap_release(code);
bap_disasm_basic_close(dis);
return 0;
}
The example can be compiled with the following command (assuming that
the code is in the example.c
file):
make LDLIBS=-lbap example
BAP also ships an interactive toplevel, aka REPL. This is a shell-like program that will interactively evaluate OCaml instructions and print the results. Just run:
$ baptop
Now, you can play with BAP. The following example will open a file, build callgraph of a program, and a control flow graph with a dominance tree of a function.
open Core_kernel.Std;;
open Bap.Std;;
open Graphlib.Std;;
let rooter = Rooter.Factory.find "byteweight" |> Option.value_exn;;
let proj = Project.create ~rooter (Project.Input.file "/bin/true") |> ok_exn;;
let prog = Project.program proj;;
let cg = Program.to_graph prog;;
let sub = Term.first sub_t prog |> Option.value_exn;;
let cfg = Sub.to_cfg sub;;
module G = Graphs.Ir;;
let entry = Option.value_exn (Term.first blk_t sub);;
let dom_tree = Graphlib.dominators (module G) cfg (G.Node.create entry);;
Note: if you do not want to use baptop
or utop
, then you can
execute the following in any OCaml top-level:
#use "topfind";;
#require "bap.top";;
Other than API documentation, we have blog and wiki, where you can find some useful information. Also, we have a permanently manned chat in case of emergency. Look at the badge on top of the README file, and feel free to join.
BAP is a framework, so you don't need to change its code to extend it.We use the dependency injection principle with many injection points to allow the user to alter BAP behavior. However, bugs happen, so if you have any problems, questions or suggestions, please, don't hesitate to use our issue tracker. Submitting a pull request with a problem fix will make us really happy. However, we will only accepted pull requests that have MIT license.
If you wrote analysis with BAP, then don't hesitate to release it to OPAM, for the benefit of the community.
- DARPA VET Project
- Siemens AG
- Institute for Information & communications Technology Promotion(IITP) grant funded by the Korea government(MSIT) (No.2015-0-00565,Development of Vulnerability Discovery Technologies for IoT Software Security)
Please, contact us if you would like to become a sponsor.