Skip to content

Latest commit

 

History

History
115 lines (82 loc) · 4.54 KB

README.md

File metadata and controls

115 lines (82 loc) · 4.54 KB
Raw

nRF5x Firmware disassembly tools - Proof of Concept

This repository contains a collection of tools to reverse NRF5x firmwares.

How to install dependencies

pip3 install sqlite3 SQLAlchemy tqdm intelhex

NRF parsing and database population

The nrfparse.py python script must be run once to generate and populate the nRF.db SQLite3 database. It extracts, from each Development Kit archive in the developer.nordicsemi/nRF5_SDK directory to disk in the SDKs directory, the different SoftDevices (s110, s210, ...etc) and their associated headers and linkers files and the firmware in its IntelHex format if contained in the archive.

The parsing then extracts for each SoftDevice version:

  • Associated SDK version (14.0.0, ...)
  • Associated NRF (nrf51422, nrf51822, ...)
  • Associated RAM and FLASH addresses and sections length for binary mapping
  • Associated SVC_BASE and SVC_LAST ranges
  • Associated SVCALLS' structures
  • Associtade SVCALLS prototypes that will help nrfreverse using IDA python to define types of SVCALLS and rename functions and parameters

The data is then commited to the nRF.db.

NRF identification given a .hex or .bin version of the firmware

The nrfidentify.py python script is run with two arguments: bin or hex and NRF firmware in its .bin or .hex format.

usage: nrfident.py [-h] {bin,hex} FILE

positional arguments:
  {bin,hex}   the object format bfdname
  FILE        input file to identify

optional arguments:
  -h, --help  show this help message and exit

The script computes the SD firmware's signature, then looks for it in the nRF.db database.

If not found, the strings in the binary can be used to try an identification, an approximate signature is then generated.

The following information are then extracted from the nRF.db:

  • Associated SDK version
  • Associated NRF
  • Associated RAM and FLASH addresses and sections length for binary mapping in IDA pro

A file named nRF_version is then generated containing the firmware's signature.

This file is used with the nRF.db by nrfreverse.py that mainly uses IDApython to redefine types and rename SVCALL functions. The nRF.db and nRF_version can be be copied to the %PROGRAMFILES%\IDA\python folder.


python3 nrfident.py bin firmwares/s132.bin
############################ nRF5-tool ############################ 
#####                                                         ##### 
#####                Identifying nRF5x firmwares              ##### 
#####                                                         ##### 
################################################################### 


Binary file provided firmwares/s132.bin

Computing signature from binary
100%|██████████████████████████████████████████████████████████████████████| 
Signature:  d082a85351ee18ecfdc9dcb01352f5df3d938a2270bcadec2ec083e9ceeb3b1e
Searching for signature in nRF.db
100%|██████████████████████████████████████████████████████████████████████| 
=========================
SDK version:  14.0.0
SoftDevice version: s132
NRF: nrf52832
=========================
SDK version:  14.1.0
SoftDevice version: s132
NRF: nrf52832
                               ==================
nRF5x signature written to file nRF_ver in current directory
nRF_ver path must be provided when running nrfreverse.py from IDA

                                     *****
                                  Binary mapping
                                     *****

SoftDevice  :  s132
Card version :  xxaa
           *****
RAM address  :  0x20001368
RAM length   :  0xec98
ROM address  :  0x23000
ROM length   :  0x5d000

NRF firmware reversing in IDA pro

The NRF firmware is mapped in IDA pro using associated FLASH and RAM addresses and lengths.

The nrfreverse.py python script must be in the same directory as the nRF.db database and the nRF_ver file that was generated by nrfident.py.

It parses the assembly code of the given SoftDevice firmware to find SVC opcodes.

It selects the appropriate functions' prototypes from the nRF.db and outputs them in the python output window of IDA pro.

Further improvements

  1. Automatically mapping the binary according to the RAM and FLASH addresses and length in IDA pro.

  2. Recover classic libc functions based on their corresponding signatures

  3. Moar testing !