Crate disassemble

Disassemble

This crate provides basic functionality for working with disassembled code. It provides (or will provide) functionality for performing:

• Working with code that has already been compiled to machine code for any CPU, bytecode, compiler IR or JIT compiler output like that of V8 or the JVM. You provide an adapter to teach Instruction how to report the correct information for your generated code.
• Reconstructing the control flow graph from a body of compiled code.
• (Future) Reconstructing loops and higher level control flow constructs.
• (Future) Performing data flow analysis.
• (Future) Generating HTML and other rich output formats to assist in visualizing structure and higher level presentations of the data derived from the generated code.
• (Future) Writing decompilers that generate C or other languages from lower level generated code.
• (Future) Writing tools for reverse engineering. Many tools for reverse engineering require extracting and inferring higher level structure from the low level generated code. This crate provides those capabilities for re-use by anyone.
• (Future) Writing views that display disassembled code for debuggers, profilers and other systems level tools.

The actual disassembly with the implementation of Instruction and other elements of the system will be provided by other crates that integrate with other systems, such as the Capstone Engine. This crate is written such that it should work with nearly any machine code, VM bytecode or JIT compiler output, given an appropriate implementation of Instruction.

It is possible (likely?) that some functionality from within this crate may move in the future to a separate crate for broader re-use. This might impact Symbol among other things.

Installation

This crate works with Cargo and is on crates.io. Add it to your Cargo.toml like so:

[dependencies]
disassemble = "0.0.1"

Then, let rustc know that you're going to use this crate at the top of your own crate:

extern crate disassemble;

Future Directions

In the future, we want to extend this library to support a number of additional features:

• Implement DOT output for the ControlFlowGraph. This can probably be done with the help of [petgraph].
• HTML output modes?
• Implement loop finding. (Havlak)
• Implement the Capstone Engine backend as a separate crate.
• Make Instruction aware of operands, registers
• Data flow support. Memory SSA?
• Should we deal with mangled symbols at this level?
• So much more!

Contributions

Contributions are welcome.

Structs

Address	The location of something in an address space.
BasicBlock	A basic block is a sequence of instructions with no inward-bound branches except to the entry point and no outward-bound branches except at the exit.
BasicBlockEdge	Information about an edge between 2 basic blocks.
ControlFlowGraph	A control flow graph.
Function	A function within a program.
Symbol	A symbol within an executable or library. This is a named address.

Enums

EdgeType

When is this edge taken? Conditionally or unconditionally?

Traits

Instruction

An assembly instruction, bytecode operation, VM operation, etc.