1

Environment Setup

In this chapter, we will discuss the basic steps of setting up the environment for future experiments with Clang . The setup is appropriate for Unix-based systems such as Linux and Mac OS (Darwin). In addition, you will get important information on how to download, configure, and build the LLVM source code. We will continue with a short session that explains how to build and use the LLVM debugger (LLDB ), which will be used as the primary tool for code investigation throughout the book. Finally, we will finish with a simple Clang tool that can check C/C++ files for compilation errors. We will use LLDB for a simple debug session for the created tool and clang internal. We will cover the following topics:

• Prerequisites

• Getting to know LLVM

• Source code compilation

• How to create a custom Clang tool

1.1 Technical requirements

Downloading and building LLVM code is very easy and does not require any paid tools. You will require the following:

• Unix-based OS (Linux, Darwin)

• Command line git

• Build tools: CMake and Ninja

We will use the debugger as the source investigation tool. LLVM has its own debugger, LLDB. We will build it as our first tool from LLVM monorepo: https://github.com/llvm/llvm-project.git.

Any build process consists of two steps. The first one is the project configuration and the last one is the build itself. LLVM uses CMake as a project configuration tool. It also can use a wide range of build tools, such as Unix Makefiles, and Ninja. It can also generate project files for popular IDEs such as Visual Studio and XCode. We are going to use Ninja as the build tool because it speeds up the build process, and most LLVM developers use it. You can find additional information about the tools here: https://llvm.org/docs/GettingStarted.html.

The source code for this chapter...

1.2 Getting to know LLVM

Let’s begin by covering some foundational information about LLVM, including the project history as well as its structure.

1.2.1 Short LLVM history

The Clang compiler is a part of the LLVM project. The project was started in 2000 by Chris Lattner and Vikram Adve as their project at the University of Illinois at Urbana–Champaign [26].

LLVM was originally designed to be a next-generation code generation infrastructure that could be used to build optimizing compilers for many programming languages. However, it has since evolved into a full-featured platform that can be used to build a wide variety of tools, including debuggers, profilers, and static analysis tools.

LLVM has been widely adopted in the software industry and is used by many companies and organizations to build a variety of tools and applications. It is also used in academic research and teaching and has inspired the development of similar projects in other fields.

The project received...

1.3 Source code compilation

We are compiling our source code in debug mode to make it suitable for future investigations with a debugger. We are using LLDB as the debugger. We will start with an overview of the build process and finish building the LLDB as a concrete example.

1.3.1 Configuration with CMake

Create a build folder where the compiler and related tools will be built:

$ mkdir build
$ cd build

The minimal configuration command looks like this:

$ cmake -DCMAKE_BUILD_TYPE=Debug ../llvm

The command requires the build type to be specified (e.g. Debug in our case) as well as the primary argument that points to a folder with the build configuration file. The configuration file is stored as CMakeLists.txt and is located in the llvm folder, which explains the ../llvm argument usage. The command generates Makefile located in the build folder, thus you can use the simple make command to start the build process.

We will use more advanced configuration...

1.4 Test project – syntax check with a Clang tool

For our first test project, we will create a simple Clang tool that runs the compiler and checks the syntax for the provided source file. We will create a so-called out-of-tree LLVM project, that is, a project that will use LLVM but will be located outside the main LLVM source tree.

Several actions are required to create the project:

• The required LLVM libraries and headers have to be built and installed

• We have to create a build configuration file for our test project

• The source code that uses LLVM has to be created

We will start with the first step and install the Clang support libraries and headers. We will use the following configuration command for CMake:

cmake -G Ninja -DCMAKE_BUILD_TYPE=Debug -DCMAKE_INSTALL_PREFIX=../install -DLLVM_TARGETS_TO_BUILD="X86" -DLLVM_ENABLE_PROJECTS="clang" -DLLVM_USE_LINKER=gold -DLLVM_USE_SPLIT_DWARF=ON -DBUILD_SHARED_LIBS=ON ../llvm

Figure 1.12: LLVM CMake configuration...

1.5 Summary

In this chapter, we covered the history of the LLVM project, obtained the source code for LLVM, and explored its internal structure. We learned about the tools used to build LLVM, such as CMake and Ninja. We studied the various configuration options for building LLVM and how they can be used to optimize resources, including disk space. We built Clang and LLDB in debug and release modes and used the resulting tools to compile a basic program and run it with the debugger. We also created a simple Clang tool and ran it with the LLDB debugger.

The next chapter will introduce you to the compiler design architecture and explain how it appears in the context of Clang . We will primarily focus on the Clang frontend, but we will also cover the important concept of the Clang driver – the backbone that manages all stages of the compilation process, from parsing to linking.

1.6 Further reading

• Getting Started with the LLVM System: https://llvm.org/docs/GettingStarted.html

• Building LLVM with CMake: https://llvm.org/docs/CMake.html

• Clang Compiler User’s Manual: https://clang.llvm.org/docs/UsersManual.html