Skip to content

datatheorem/strongarm

Repository files navigation

strongarm

Build PyPI version PyVersion badge

strongarm is a full-featured, cross-platform ARM64 Mach-O analysis library.

strongarm is production-ready and is used throughout DataTheorem's iOS static analyzer stack.

REPL example

This repo contains multiple tools to explore strongarm and the API. In the scripts folder, several popular Mach-O analysis tools have been reimplemented in strongarm, to demonstrate real API usage. As strongarm is cross-platform, all of these tools are as well:

  • strongarm-cli: Static analysis REPL (try me!)
  • class-dump: Dump the Objective-C class information from a Mach-O with Objective-C declaration syntax
  • insert_dylib: Add a load command to a Mach-O
  • dsc_symbolicate: Given a dyld_shared_cache, generate a symbol map from the embedded system images
  • nm: List the symbol table of a Mach-O
  • lipo: Thin or fatten Mach-O files and slices
  • hexdump: Output the hex content of a byte range in a file
  • strings: Output the C-strings in a Mach-O
  • dump_entitlements: Print the code-signing information
  • bitcode_retriever: Extract the XAR archive containing LLVM bitcode from a Mach-O

Installation

strongarm is supported on macOS and Linux.

Via pip

pip install strongarm-ios

Via git (for local development)

To setup a local environment:

git clone ...
cd strongarm
python -m venv .venv
source .venv/bin/activate
pip install -U pip setuptools wheel 'pip-tools<7.0.0'
pip-sync requirements.txt requirements-dev.txt

If you modify requirements.in or requirements-dev.in:

pip-compile requirements.in
pip-compile requirements-dev.in
pip-sync requirements.txt requirements-dev.txt
git add requirements-dev.in requirements-dev.txt

Features

  • Access and cross-reference Mach-O info via an API
  • Dataflow analysis
  • Function-boundary detection

Mach-O parsing:

  • Metadata (architecture, endianness, etc)
  • Load commands
  • Symbol tables
  • String tables
  • Code signature
  • Dyld info
  • Objective-C info (classes, categories, protocols, methods, ivars, etc)

Mach-O analysis:

  • Cross-references (xrefs) of code and strings
  • Function boundary detection & disassembly
  • Track constant data movement in assembly
  • Dyld bound symbols & implementation stubs
  • Parse constant NSStrings and C strings
  • Basic block analysis

Mach-O editing:

  • Load command insertion
  • Write Mach-O structures
  • Byte-edit binaries

Quickstart

Pass an input file to MachoParser, which will read a Mach-O or FAT and provide access to individual MachoBinary slices.

import pathlib
from strongarm.macho import MachoParser, MachoBinary

# Load an input file
parser = MachoParser(pathlib.Path("~/Documents/MyApp.app/MyApp"))
# Read the ARM64 slice and perform some operations
binary: MachoBinary = parser.get_arm64_slice()
print(binary.get_entitlements().decode())
print(hex(binary.section_with_name("__text", "__TEXT").address))

Advanced analysis

Some APIs which require more memory or cross-referencing are available through MachoAnalyzer

from pathlib import Path
from strongarm.macho import MachoParser, MachoBinary, MachoAnalyzer

macho_parser = MachoParser(Path("~/Documents/MyApp.app/MyApp"))
binary: MachoBinary = macho_parser.get_arm64_slice()
# A MachoAnalyzer wraps a binary and allows deeper analysis
analyzer = MachoAnalyzer.get_analyzer(binary)

# Find all calls to -[UIAlertView init] in the binary
print(analyzer.objc_calls_to(["_OBJC_CLASS_$_UIAlertView"], ["init"], requires_class_and_sel_found=False))

# Print some interesting info
print(analyzer.imported_symbol_names_to_pointers)   # All the dynamically linked symbols which will be bound at runtime
print(analyzer.exported_symbol_names_to_pointers)   # All the symbols which this binary defines and exports
print(analyzer.get_functions())                     # Entry-point list of the binary. Each of these can be wrapped in an ObjcFunctionAnalyzer
print(analyzer.strings())                           # __cstring segment
print(analyzer.get_imps_for_sel("viewDidLoad"))     # Convenience accessor for an ObjcFunctionAnalyzer

# Print the Objective-C class information
for objc_cls in analyzer.objc_classes():
    print(objc_cls.name)
    for objc_ivar in objc_cls.ivars:
        print(f"\tivar: {objc_ivar.name}")
    for objc_sel in objc_cls.selectors:
        print(f"\tmethod: {objc_sel.name} @ {hex(objc_sel.implementation)}")

Code analysis

Once you have a handle to a FunctionAnalyzer, representing a source code function, you can analyze the code:

from pathlib import Path
from strongarm.macho import MachoParser, MachoBinary, MachoAnalyzer
from strongarm.objc import ObjcFunctionAnalyzer

macho_parser = MachoParser(Path("~/Documents/MyApp.app/MyApp"))
binary: MachoBinary = macho_parser.get_arm64_slice()
analyzer = MachoAnalyzer.get_analyzer(binary)
function_analyzer = ObjcFunctionAnalyzer.get_function_analyzer_for_signature(binary, "ViewController", "viewDidLoad")
print(function_analyzer.basic_blocks)   # Find the basic block boundaries

# Print some interesting info about Objective-C method calls in the function
for instr in function_analyzer.instructions:
    if not instr.is_msgSend_call:
        continue
    
    # In an Objective-C message send, x0 stores the receiver and x1 stores the selector being messaged.
    classref = function_analyzer.get_register_contents_at_instruction("x0", instr)
    selref = function_analyzer.get_register_contents_at_instruction("x1", instr)
    
    class_name = analyzer.class_name_for_class_pointer(classref.value)
    selector = analyzer.selector_for_selref(selref.value).name
   
    # Prints "0x100000000: _objc_msgSend(_OBJC_CLASS_$_UIView, @selector(alloc));"
    print(f"{hex(instr.address)}: {instr.symbol}({class_name}, @selector({selector}));")

Modifying Mach-O's

You can also modify Mach-O's by overwriting structures or inserting load commands:

from pathlib import Path
from strongarm.macho import MachoParser, MachoBinary
from strongarm.macho.macho_definitions import MachoSymtabCommand

macho_parser = MachoParser(Path("~/Documents/MyApp.app/MyApp"))
# Overwrite a structure
binary: MachoBinary = macho_parser.get_arm64_slice()
new_symbol_table = MachoSymtabCommand()
new_symbol_table.nsyms = 0
modified_binary = binary.write_struct(new_symbol_table, binary.symtab.address, virtual=True)

# Add a load command
modified_binary = modified_binary.insert_load_dylib_cmd("/System/Frameworks/UIKit.framework/UIKit")

# Write the modified binary to a file
MachoBinary.write_binary(Path(__file__).parent / "modified_binary")

MachoBinary provides several functions to faciliate binary modifications.

As modifying a MachoBinary may invalidate its public attributes, these APIs return a new MachoBinary object, which is re-parsed with the edits.

# Write raw bytes or Mach-O structures to a binary
MachoBinary.write_bytes(self, data: bytes, address: int, virtual=False) -> MachoBinary
MachoBinary.write_struct(self, struct: Structure, address: int, virtual=False) -> MachoBinary

# Insert a load command
MachoBinary.insert_load_dylib_cmd(dylib_path: str) -> MachoBinary

# Flush a modified slice to a thin Mach-O file, or a list of slices to a FAT Mach-O file:
MachoBinary.write_binary(self, path: pathlib.Path) -> None
@staticmethod
MachoBinary.write_fat(slices: List[MachoBinary], path: pathlib.Path) -> None

To make several modifications to a MachoBinary while triggering only one extra parse, use a MachoBinaryWriter:

from pathlib import Path
from strongarm.macho import MachoParser, MachoBinary
from ctypes import c_uint64, sizeof
from strongarm.macho.macho_binary_writer import MachoBinaryWriter

macho_parser = MachoParser(Path("~/Documents/MyApp.app/MyApp"))
binary: MachoBinary = macho_parser.get_arm64_slice()
# Initialise a batch binary writer
writer = MachoBinaryWriter(binary)

# Make a series of changes to the binary
with writer:
    for i in range(5):
        writer.write_word(word=c_uint64(0xdeadbeef), address=0x1000 + (i * sizeof(c_uint64)), virtual=False)

# `writer.modified_binary` contains the re-parsed binary containing the provided changes
# Persist the modified binary to disk
writer.modified_binary.write_binary(Path(__file__) / "modified_binary")

License

AGPL license

About

Mach-O analysis library πŸ’ͺ

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Contributors 4

  •  
  •  
  •  
  •  

Languages