ghidra/DevGuide.md

17 KiB
Raw Blame History

Developer's Guide

Environment

For specific information on required versions and download links please see the README.md file.

Quickstart

Follow the Advanced Development instructions in the README.md file to get your development environment setup quickly.

If possible please try to stick to the Apache License 2.0 license when developing for Ghidra. At times it may be necessary to incorporate other compatible licenses into Ghidra. Any GPL code must live in the top-level GPL/ directory as a totally standalone, independently buildable Ghidra module.

If you are contributing code to the Ghidra project, the preferred way to receive credit/recognition is Git commit authorship. Please ensure your Git credentials are properly linked to your GitHub account so you appear as a Ghidra contributor on GitHub. We do not have a standard for putting authors' names directly in the source code, so it is discouraged.

Common Gradle Tasks

Download non-Maven Central dependencies. This creates a dependencies directory in the repository root.

gradle -I gradle/support/fetchDependencies.gradle

Download Maven Central dependencies and setup the repository for development. By default, these will be stored at $HOME/.gradle/.

gradle prepdev

Generate nested Eclipse project files which can then be imported into Eclipse as "existing projects".

gradle cleanEclipse eclipse

Build native components for your current platform. Requires native tool chains to be present.

gradle buildNatives

Manually compile sleigh files. Ghidra will also do this at runtime when necessary.

gradle sleighCompile

Build Javadoc:

gradle createJavadocs

Build Python3 packages for the Debugger:

gradle buildPyPackage

Build Ghidra to build/dist in an uncompressed form. This will be a distribution intended only to run on the platform on which it was built.

gradle assembleAll

Build Ghidra to build/dist in a compressed form. This will be a distribution intended only to run on the platform on which it was built.

gradle buildGhidra

Tip: You may want to skip certain Gradle tasks to speed up your build, or to deal with a problem later. For example, perhaps you added some new source files and the build is failing because of unresolved IP header issues. You can use the Gradle -x <task> command line argument to prevent specific tasks from running:

gradle buildGhidra -x ip

Known Issues

  • There is a known issue in Gradle that can prevent it from discovering native toolchains on Linux if a non-English system locale is being used. As a workaround, set the following environment variable prior to running your Gradle task: LC_MESSAGES=en_US.UTF-8

Offline Development Environment

Sometimes you may want to move the Ghidra repository to an offline network and do development there. These are the recommended steps to ensure that you not only move the source repository, but all downloaded dependencies as well:

  1. gradle -I gradle/support/fetchDependencies.gradle
  2. gradle -g dependencies/gradle prepdev
  3. Move ghidra directory to different system
  4. gradle -g dependencies/gradle buildGhidra (on offline system)

NOTE: The -g flag specifies the Gradle user home directory. The default is the .gradle directory in the users home directory. Overriding it to be inside the Ghidra repository will ensure that all maven central dependencies that were fetched during the prepdev task will be moved along with the rest of the repo.

Developing GhidraDev Eclipse Plugin

Developing the GhidraDev Eclipse plugin requires the Eclipse PDE (Plug-in Development Environment), which can be installed via the Eclipse marketplace. It is also included in the Eclipse IDE for RCP and RAP Developers. To generate the GhidraDev Eclipse projects, execute:

gradle eclipse -PeclipsePDE

Import the newly generated GhidraDev projects into an Eclipse that supports this type of project.

Note: If you are getting compilation errors related to PyDev and CDT, go into Eclipse's preferences, and under Target Platform, activate /Eclipse GhidraDevPlugin/GhidraDev.target.

See Building GhidraDev for instructions on how to build the GhidraDev plugin.

Running tests

To run unit tests, do:

gradle unitTestReport

For more complex integration tests, do:

gradle integrationTest

For running both unit and integration tests and to generate a report do:

gradle combinedTestReport

Setup build in CI

For running tests in headless mode on Linux, in a CI environment, or in Docker, first do:

Xvfb :99 -nolisten tcp &
export DISPLAY=:99

This is required to make AWT happy.

Building Supporting Data

Some features of Ghidra require the curation of rather extensive databases. These include the Data Type Archives and Function ID Databases, both of which require collecting header files and libraries for the relevant SDKs and platforms. Much of this work is done by hand. The archives included in our official builds can be found in the ghidra-data repository.

Building Data Type Archives

This task is often done manually from the Ghidra GUI, and the archives included in our official build require a fair bit of fine tuning.

  1. From the CodeBrowser, select File -> Parse C Source
  2. From here you can create and configure parsing profiles, which lists headers and pre-processor options.
  3. Click Parse to File to create the Data Type Archive.
  4. The result can be added to an installation or source tree by copying it to Ghidra/Features/Base/data/typeinfo.

Building FID Databases

This task is often done manually from the Ghidra GUI, and the archives included in our official build require a fair bit of fine tuning. You will first need to import the relevant libraries from which you'd like to produce a FID database. This is often a set of libraries from an SDK. We include a variety of Visual Studio platforms in the official build. The official .fidb files can be found in the ghidra-data repository.

  1. From the CodeBrowser, select File -> Configure
  2. Enable the "Function ID" plugins, and close the dialog.
  3. From the CodeBrowser, select Tools -> Function ID -> Create new empty FidDb.
  4. Choose a destination file.
  5. Select Tools -> Function ID -> Populate FidDb from programs.
  6. Fill out the options appropriately and click OK.

If you'd like some details of our fine tuning, take a look at building_fid.txt.

Debugger Development

We have recently changed the Debugger's back-end architecture. We no longer user JNA to access native Debugger APIs. We only use it for pseudo-terminal access. Instead, we use Python3 and a protobuf-based TCP connection for back-end integration.

Additional Dependencies

In addition to Ghidra's normal dependencies, you may want the following:

  • WinDbg for Windows x64
  • GDB 13 or later for Linux
  • LLDB 10 or later for macOS

The others (e.g., JNA) are handled by Gradle via Maven Central.

Architecture Overview

There are several Eclipse projects each fitting into a larger architectural picture. These all currently reside in the Ghidra/Debug directory, but will likely be re-factored into the Framework and Feature directories later. Each project is listed "bottom up" with a brief description and status.

  • ProposedUtils - a collection of utilities proposed to be moved to other respective projects.
  • AnnotationValidator - an experimental annotation processor for database access objects.
  • Framework-TraceModeling - a database schema and set of interfaces for storing machine state over time.
  • Framework-AsyncComm - a collection of utilities for asynchronous communication (packet formats and completable-future conveniences).
  • Framework-Debugging - specifies interfaces for debugger models and provides implementation conveniences. This is mostly deprecated.
  • Debugger - the collection of Ghidra plugins and services comprising the Debugger UI.
  • Debugger-rmi-trace - the wire protocol, client, services, and UI components for Trace RMI, the new back-end architecture.
  • Debugger-agent-dbgeng - the connector for WinDbg (via dbgeng.dll) on Windows x64.
  • Debugger-agent-dbgmodel - an experimental connector for WinDbg Preview (with TTD, via dbgmodel.dll) on Windows x64. This is deprecated, as most of these features are implemented in Debugger-agent-dbgeng for the new architecture.
  • Debugger-agent-dbgmodel-traceloader - an experimental "importer" for WinDbg trace files. This is deprecated.
  • Debugger-agent-gdb - the connector for GDB (13 or later recommended) on UNIX.
  • Debugger-swig-lldb - the Java language bindings for LLDB's SBDebugger, also proposed upstream. This is deprecated. We now use the Python3 language bindings for LLDB.
  • Debugger-agent-lldb - the connector for LLDB (10 or later recommended) on macOS, UNIX, and Windows.
  • Debugger-gadp - the connector for our custom wire protocol the Ghidra Asynchronous Debugging Protocol. This is deprecated. It's replaced by Debugger-rmi-trace.
  • Debugger-jpda - an in-development connector for Java and Dalvik debugging via JDI (i.e., JDWP). This is deprecated and not yet replaced.

The Trace Modeling schema records machine state and markup over time. It rests on the same database framework as Programs, allowing trace recordings to be stored in a Ghidra project and shared via a server, if desired. Trace "recording" is a de facto requirement for displaying information in Ghidra's UI. The back-end connector has full discretion over what is recorded by using Trace RMI. Typically, only the machine state actually observed by the user (or perhaps a script) is recorded. For most use cases, the Trace is small and ephemeral, serving only to mediate between the UI components and the target's model. It supports many of the same markup (e.g., disassembly, data types) as Programs, in addition to tracking active threads, loaded modues, breakpoints, etc.

Every back end (or "adapter" or "connector" or "agent") employs the Trace RMI client to populate a trace database. As a general rule in Ghidra, no component is allowed to access a native API and reside in the same JVM as the Ghidra UI. This allows us to contain crashes, preventing data loss. To accommodate this requirement — given that debugging native applications is almost certainly going to require access to native APIs — we've developed the Trace RMI protocol. This also allows us to better bridge the language gap between Java and Python, which is supported by most native debuggers. This protocol is loosely coupled to Framework-TraceModeling, essentially exposing its methods via RMI, as well as some methods for controlling the UI. The protocol is built using Google's Protobuf library, providing a potential path for back-end implementations in alternative languages. We provide the Trace RMI server as a Ghidra component implemented in Java and the Trace RMI client as a Python3 package. A back-end implementation may be a stand-alone executable or script that accesses the native debugger's API, or a script or plugin for the native debugger. It then connects to Ghidra via Trace RMI to populate the trace database with information gleaned from that API. It should provide a set of diagnostic commands to control and monitor that connection. It should also use the native API to detect session and target changes so that Ghidra's UI consistently reflects the debugging session.

The old system relied on a "recorder" to discover targets and map them to traces in the proper Ghidra language. That responsibility is now delegated to the back end. Typically, it examines the target's architecture and immediately creates a trace upon connection.

Developing a new connector

So Ghidra does not yet support your favorite debugger? We believe the new system is much less daunting than the previous. Still, please finish reading this guide, and look carefully at the ones we have so far, and perhaps ask to see if we are already developing one. Of course, in time you might also search the internet to see if others are developing one. There are quite a few caveats and gotchas, the most notable being that this interface is still in some flux. When things go wrong, it could be because of, without limitation:

  1. A bug on your part
  2. A bug on our part
  3. A design flaw in the interfaces
  4. A bug in the debugger/API you're adapting

We are still (yes, still) in the process of writing up this documentation. In the meantime, we recommend using the GDB and dbgeng agents as examples. Be sure to look at the Python code src/main/py! The deprecated Java code src/main/java is still included as we transition.

You'll also need to provide launcher(s) so that Ghidra knows how to configure and start your connector. These are just shell scripts. We use bash scripts on Linux and macOS, and we use batch files on Windows. Try to include as many common use cases as makes sense for the debugger. This provides the most flexibility to users and examples to power users who might create derivative launchers. Look at the existing launchers for examples.

For testing, please follow the examples for GDB. We no longer provide abstract classes that prescribe requirements. Instead, we just provide GDB as an example. Usually, we split our tests into three categories:

  • Commands
  • Methods
  • Hooks

The Commands tests check that the user CLI commands, conventionally implemented in commands.py, work correctly. In general, do the minimum connection setup, execute the command, and check that it produces the expected output and causes the expected effects.

The Methods tests check that the remote methods, conventionally implemented in methods.py, work correctly. Many methods are just wrappers around CLI commands, some provided by the native debugger and some provided by commands.py. These work similarly to the commands test, except that they invoke methods instead of executing commands. Again, check the return value (rarely applicable) and that it causes the expected effects.

The Hooks tests check that the back end is able to listen for session and target changes, e.g., knowing when the target stops. The test should not "cheat" by executing commands or invoking methods that should instead be triggered by the listener. It should execute the minimal commands to setup the test, then trigger an event. It should then check that the event in turn triggered the expected effects, e.g., updating PC upon the target stopping.

Whenever you make a change to the Python code, you'll need to re-assemble the package's source.

gradle assemblePyPackage

This is required in case your package includes generated source, as is the case for Debugger-rmi-trace. If you want to create a new Ghidra module for your connector (recommended) use an existing one's build.gradle as a template. A key part is applying the hasPythonPackage.gradle script.

Adding a new platform

If a connector already exists for a suitable debugger on the desired platform, then adding it may be very simple. For example, many platforms are supported by GDB, so even though we're currently focused on x86-64 (and to some extent arm64) support, we've provided the mappings for many. These mappings are conventionally kept in each connector's arch.py file.

In general, to update arch.py, you need to know:

  1. What the platform is called (including variant names) by the debugger
  2. What the processor language is called by Ghidra
  3. If applicable, the mapping of target address spaces into Ghidra's address spaces
  4. If applicable, the mapping of target register names to those in Ghidra's processor language

In most cases (3) and (4) are already implemented by the included mappers. Naturally, you'll want to test the special cases, preferably in automated tests.

Emulation

The most obvious integration path for 3rd-party emulators is to write a "connector." However, p-code emulation is an integral feature of the Ghidra UI, and it has a fairly accessible API. Namely, for interpolation between machines states recorded in a trace, and extrapolation into future machine states. Integration of such emulators may still be useful to you, but we recommend trying the p-code emulator to see if it suits your needs for emulation in Ghidra before pursuing integration of another emulator. We also provide out-of-the-box QEMU integration via GDB.

Contributing

When submitting help tickets and pull requests, please tag those related to the debugger with "Debugger" so that we can triage them more quickly.