Getting Started

Let’s fetch the llvm-libc sources and build them.

Install dependencies first:

$ sudo apt update
$ sudo apt install git cmake ninja-build clang gcc-multilib

$ git clone --depth=1 git@github.com:llvm/llvm-project.git /tmp/llvm-project
$ mkdir /tmp/llvm-project/build
$ cd /tmp/llvm-project/build
$ cmake ../runtimes -GNinja \
  -DLLVM_ENABLE_RUNTIMES="libc;compiler-rt" \
  -DCMAKE_BUILD_TYPE=Debug \
  -DCMAKE_CXX_COMPILER=clang++ \
  -DCMAKE_C_COMPILER=clang \
  -DLLVM_LIBC_FULL_BUILD=ON \
  -DLLVM_LIBC_INCLUDE_SCUDO=ON \
  -DCOMPILER_RT_BUILD_SCUDO_STANDALONE_WITH_LLVM_LIBC=ON \
  -DCOMPILER_RT_BUILD_GWP_ASAN=OFF \
  -DCOMPILER_RT_SCUDO_STANDALONE_BUILD_SHARED=OFF
$ ninja libc libm

This will produce the following artifacts:

llvm-project/build/libc/lib/libc.a
llvm-project/build/libc/lib/libm.a
llvm-project/build/libc/startup/linux/crt1.o
llvm-project/build/libc/include/**.h

We can then compile and run hello world via:

// hello.c
#include <stdio.h>
int main () { puts("hello world"); }

$ clang -nostdinc -nostdlib hello.c -I libc/include \
  -I $(clang -print-resource-dir)/include libc/startup/linux/crt1.o \
  libc/lib/libc.a
$ ./a.out
hello world

This was what we call a “full build” of llvm-libc. From here, you can visit Full Host Build for more info, Full Cross Build for cross compiling, Overlay Mode for mixing llvm-libc with another libc, or libc for GPUs for targeting GPUs.