1
1
mirror of https://github.com/rui314/mold.git synced 2024-11-10 10:57:55 +03:00
mold: A Modern Linker 🦠
Go to file
2022-09-18 13:26:29 +08:00
.github [ELF][PPC64] Enable PPC64 tests if -DMOLD_ENABLE_QEMU_TESTS is specified 2022-09-10 15:59:24 +08:00
docs Update the man page 2022-09-13 15:53:03 +08:00
elf Refactor 2022-09-18 13:04:56 +08:00
macho Fix -Wunused-variable warnings 2022-09-14 15:00:49 +08:00
test [ELF] Do not promote hidden symbols to dynamic symbols 2022-09-18 12:32:56 +08:00
third-party Add libzstd to support --compress-debug-sections=zstd 2022-09-13 14:55:17 +08:00
.gitignore Ignore directories commonly used with CMake 2022-08-19 14:33:20 +08:00
archive-file.h Update comments 2022-05-02 09:03:28 +08:00
CMakeLists.txt [ELF] Add a stub for SPARC64 2022-09-16 14:00:21 +08:00
CMakeSettings.json Use clang-cl instead of MSVC's cl as a compiler on Windows 2022-08-12 21:21:01 -07:00
cmdline.h Suppress 'unused function' compiler warning 2022-07-04 08:05:12 +02:00
compress.cc Fix -Wunused-variable warnings 2022-09-14 15:00:49 +08:00
CONTRIBUTING.md Do not request a sign-off for small patches 2022-08-19 15:08:28 +08:00
create-symlink.cmake Eliminate all uses of Python3 2022-08-20 12:29:03 +08:00
demangle.cc Simplify 2022-08-12 19:37:57 +08:00
dist.sh Create an Docker multi-architecture build image 2022-09-15 15:30:21 +08:00
Dockerfile Create an Docker multi-architecture build image 2022-09-15 15:30:21 +08:00
filepath.cc Fixes for Windows string_view impl 2022-08-11 17:28:11 +02:00
filetype.h [ELF][SPARC64] Implement features 2022-09-16 18:06:19 +08:00
glob.cc Move glob pattern matchers out of elf directory 2022-06-05 17:31:32 +08:00
hyperloglog.cc Refactor 2022-01-16 12:55:23 +09:00
install-build-deps.sh Do not over-specify distro version 2022-09-14 11:24:23 +08:00
inttypes.h Refactor 2022-09-18 13:26:29 +08:00
LICENSE Add libzstd to support --compress-debug-sections=zstd 2022-09-13 14:55:17 +08:00
main.cc Remove dead code 2022-09-11 10:36:20 +08:00
Makefile Disable -Wsign-compare and -Wunused-function 2022-09-14 15:12:50 +08:00
mold.h Simplify 2022-09-14 11:17:07 +08:00
multi-glob.cc Refactor 2022-07-04 13:52:34 +08:00
output-file-unix.h Fix more compile errors for Windows 2022-08-13 00:55:47 -07:00
output-file-win32.h Fix build errors for Windows 2022-08-13 01:54:38 -07:00
output-file.h Fix more compile errors for Windows 2022-08-13 00:55:47 -07:00
perf.cc Simplify 2022-08-12 16:44:51 +08:00
README.md Update README 2022-09-09 18:08:02 +08:00
sha.h Fix compilation error with MSVC 2022-08-12 19:50:45 +08:00
strerror.cc Simplify 2022-08-12 19:37:57 +08:00
tar.cc [ELF] Reimplement --repro 2022-02-01 13:47:44 +09:00
update-git-hash.cmake Refactor 2022-08-21 16:48:39 +08:00
uuid.cc [Mach-O] Compute LC_UUID as a tree hash of ad-hoc code signatures 2022-05-27 11:57:27 +08:00

mold: A Modern Linker

mold is a faster drop-in replacement for existing Unix linkers. It is several times faster than the LLVM lld linker, the second-fastest open-source linker which I originally created a few years ago. mold is designed to increase developer productivity by reducing build time, especially in rapid debug-edit-rebuild cycles.

Here is a performance comparison of GNU gold, LLVM lld, and mold for linking final debuginfo-enabled executables of major large programs on a simulated 8-core 16-threads machine.

Link speed comparison

Program (linker output size) GNU gold LLVM lld mold
Chrome 96 (1.89 GiB) 53.86s 11.74s 2.21s
Clang 13 (3.18 GiB) 64.12s 5.82s 2.90s
Firefox 89 libxul (1.64 GiB) 32.95s 6.80s 1.42s

mold is so fast that it is only 2x slower than cp on the same machine. Feel free to file a bug if you find mold is not faster than other linkers.

mold officially supports x86-64, i386, ARM64, ARM32, 64-bit RISC-V and 32-bit RISC-V. Little-endian PowerPC64 ELFv2 is partially supported.

Why does the speed of linking matter?

If you are using a compiled language such as C, C++ or Rust, a build consists of two phases. In the first phase, a compiler compiles source files into object files (.o files). In the second phase, a linker takes all object files to combine them into a single executable or a shared library file.

The second phase takes a long time if your build output is large. mold can make it faster, saving your time and keeping you from being distracted while waiting for a long build to finish. The difference is most noticeable when you are in rapid debug-edit-rebuild cycles.

Install

Binary packages for the following systems are currently available.

Packaging status

How to build

mold is written in C++20, so if you build mold yourself, you need a recent version of a C++ compiler and a C++ standard library. GCC 10.2 or Clang 12.0.0 (or later) as well as libstdc++ 10 or libc++ 7 (or later) are recommended.

Install dependencies

To install build dependencies, run ./install-build-deps.sh in this directory. It recognizes your Linux distribution and tries to install necessary packages. You may want to run it as root.

Compile mold

git clone https://github.com/rui314/mold.git
cd mold
git checkout v1.4.2
make -j$(nproc) CXX=clang++
sudo make install

You may need to pass a C++20 compiler command name to make. In the above case, clang++ is passed. If it doesn't work for you, try a specific version of a compiler such as g++-10 or clang++-12.

By default, mold is installed to /usr/local/bin.

If you don't use a recent enough Linux distribution, or if for any reason make in the above commands doesn't work for you, you can use Docker to build it in a Docker environment. To do so, just run ./dist.sh in this directory instead of running make -j$(nproc). The shell script pulls a Docker image, builds mold and auxiliary files inside it, and packs them into a single tar file mold-$version-$arch-linux.tar.gz. You can extract the tar file anywhere and use mold executable in it.

How to use

A classic way to use mold

On Unix, the linker command (which is usually /usr/bin/ld) is invoked indirectly by the compiler driver (which is usually cc, gcc or clang), which is typically in turn indirectly invoked by make or some other build system command.

If you can specify an additional command line option to your compiler driver by modifying build system's config files, add one of the following flags to use mold instead of /usr/bin/ld:

  • Clang: pass -fuse-ld=mold

  • GCC 12.1.0 or later: pass -fuse-ld=mold

  • GCC before 12.1.0: -fuse-ld does not accept mold as a valid argument, so you need to use -B option instead. -B is an option to tell GCC where to look for external commands such as ld.

    If you have installed mold with make install, there should be a directory named /usr/libexec/mold (or /usr/local/libexec/mold, depending on your $PREFIX), and ld command should be there. The ld is actually a symlink to mold. So, all you need is to pass -B/usr/libexec/mold (or -B/usr/local/libexec/mold) to GCC.

If you haven't installed mold to any $PATH, you can still pass -fuse-ld=/absolute/path/to/mold to clang to use mold. GCC does not take an absolute path as an argument for -fuse-ld though.

If you are using Rust

Create .cargo/config.toml in your project directory with the following:

[target.x86_64-unknown-linux-gnu]
linker = "clang"
rustflags = ["-C", "link-arg=-fuse-ld=/path/to/mold"]

where /path/to/mold is an absolute path to mold exectuable. Please make sure you have installed clang.

If you want to use mold for all projects, put the above snippet to ~/.cargo/config.toml.

If you are using macOS, you can modify config.toml in a similar manner. Here is an example with mold installed via Homebrew.

[target.x86_64-apple-darwin]
linker = "clang"
rustflags = ["-C", "link-arg=-fuse-ld=mold"]

[target.aarch64-apple-darwin]
linker = "clang"
rustflags = ["-C", "link-arg=-fuse-ld=mold"]
mold -run

It is sometimes very hard to pass an appropriate command line option to cc to specify an alternative linker. To deal with the situation, mold has a feature to intercept all invocations of ld, ld.lld or ld.gold and redirect it to itself. To use the feature, run make (or another build command) as a subcommand of mold as follows:

mold -run make <make-options-if-any>

Internally, mold invokes a given command with LD_PRELOAD environment variable set to its companion shared object file. The shared object file intercepts all function calls to exec(3)-family functions to replace argv[0] with mold if it is ld, ld.gold or ld.lld.

On macOS

mold/macOS is available as an alpha version. It can be used to build not only macOS apps but also iOS apps because their binary formats are the same.

The command name of mold/macOS is ld64.mold. If you build mold on macOS, it still produces mold and ld.mold, but these executables are useful only for cross compilation (i.e. building Linux apps on macOS.)

If you find any issue with mold/macOS, please file it to our GitHub Issues.

GitHub Actions

You can use our setup-mold GitHub Action to speed up GitHub-hosted continuous build. GitHub Actions runs on a two-core machine, but mold is still significantly faster than the default GNU linker there especially when a program being linked is large.

Verify that you are using mold

mold leaves its identification string in .comment section in an output file. You can print it out to verify that you are actually using mold.

$ readelf -p .comment <executable-file>

String dump of section '.comment':
  [     0]  GCC: (Ubuntu 10.2.0-5ubuntu1~20.04) 10.2.0
  [    2b]  mold 9a1679b47d9b22012ec7dfbda97c8983956716f7

If mold is in .comment, the file is created by mold.

Online manual

Since mold is a drop-in replacement, you should be able to use it without reading its manual. But just in case you need it, it's available online at here. You can also read the same manual by man mold.

Why is mold so fast?

One reason is because it simply uses faster algorithms and efficient data structures than other linkers do. The other reason is that the new linker is highly parallelized.

Here is a side-by-side comparison of per-core CPU usage of lld (left) and mold (right). They are linking the same program, Chromium executable.

CPU usage comparison in htop animation

As you can see, mold uses all available cores throughout its execution and finishes quickly. On the other hand, lld failed to use available cores most of the time. In this demo, the maximum parallelism is artificially capped to 16 so that the bars fit in the GIF.

For details, please read design notes.

License

mold is available under AGPL. Note that that does not mean that you have to license your program under AGPL if you use mold to link your program. An output of the mold linker is a derived work of the object files and libraries you pass to the linker but not a derived work of the mold linker itself.

Besides that, you can also buy a commercial, non-AGPL license with technical support from our company, Blue Whale Systems PTE LTD. If you are a big company, please consider obtaining it before making hundreds or thousands of developers of your company to depend on mold. mold is mostly a single-person open-source project, and just like other open-source projects, we are not legally obligated to keep maintaining it. A legally-binding commercial license contract addresses the concern. By purchasing a license, you are guaranteed that mold will be maintained for you. Please contact us for a commercial license inquiry.

Acknowledgement

We accept donations via GitHub Sponsors and OpenCollective. We thank you to everybody who sponsors our project. In particular, we'd like to acknowledge the following people and organizations who have sponsored $128/month or more: