LLVM - Revision history

Klinger: Category:Applications Category:Languages removed, Category:Development added

2025-03-25T21:26:08Z

Category:Applications Category:Languages removed, Category:Development added

← Older revision		Revision as of 21:26, 25 March 2025
Line 71:		Line 71:
	}		}
	</syntaxhighlight>		</syntaxhighlight>
	[[Category:~~Applications]]~~		[[Category:Development]]
	~~[[Category:Languages~~]]

TheComputerGuy: Add a section about using "pkgsLLVM"

2024-07-26T18:40:33Z

Add a section about using "pkgsLLVM"

← Older revision		Revision as of 18:40, 26 July 2024
Line 62:		Line 62:
	</syntaxHighlight>		</syntaxHighlight>

			== Building Nixpkgs/NixOS with LLVM ==
			It is technically possible to build Nixpkgs and NixOS with LLVM, however many packages are broken due to the differences between GCC and Clang. Tristan Ross, one of the maintainers of LLVM, started a project back in June of 2024 to improve the state of LLVM compiled packages. Currently, the project is located on GitHub at [https://github.com/RossComputerGuy/nixpkgs-llvm-ws RossComputerGuy/nixpkgs-llvm-ws]. Many packages have been fixed upstream but it is best to try the Flake as not all fixes are merged.

			=== Using LLVM in Nixpkgs without the Nixpkgs LLVM Workspace ===
			To use LLVM to build packages in Nix, it can be done by either using the <code>pkgsLLVM</code> attribute or by overriding a derivation and changing out the <code>stdenv</code>. Changing out the <code>stdenv</code> is as simple as this:<syntaxhighlight lang="nix">
			pkgs.hello.override {
			stdenv = pkgs.llvmPackages.stdenv;
			}
			</syntaxhighlight>
	[[Category:Applications]]		[[Category:Applications]]
	[[Category:Languages]]		[[Category:Languages]]

imported>NiklasGollenstede: add python3 depenedency, which is required for newer versions of llvm

2023-06-17T17:52:22Z

add python3 depenedency, which is required for newer versions of llvm

← Older revision		Revision as of 17:52, 17 June 2023
Line 14:		Line 14:
	buildInputs = [		buildInputs = [
	bashInteractive		bashInteractive
			python3
	ninja		ninja
	cmake		cmake

imported>Igodlab: libcxx & libcxxabi should be enabled in -DLLVM_ENABLE_RUNTIMES rather than -DLLVM_ENABLE_PROJECTS

2023-01-05T19:57:11Z

libcxx & libcxxabi should be enabled in -DLLVM_ENABLE_RUNTIMES rather than -DLLVM_ENABLE_PROJECTS

← Older revision		Revision as of 19:57, 5 January 2023
Line 34:		Line 34:
	"-DLLVM_INSTALL_TOOLCHAIN_ONLY=ON"		"-DLLVM_INSTALL_TOOLCHAIN_ONLY=ON"
	# change this to enable the projects you need		# change this to enable the projects you need
	"-DLLVM_ENABLE_PROJECTS=clang;libcxx;libcxxabi"		"-DLLVM_ENABLE_PROJECTS=clang"
			# enable libcxx* to come into play at runtimes
			"-DLLVM_ENABLE_RUNTIMES=libcxx;libcxxabi"
	# this makes llvm only to produce code for the current platform, this saves CPU time, change it to what you need		# this makes llvm only to produce code for the current platform, this saves CPU time, change it to what you need
	"-DLLVM_TARGETS_TO_BUILD=host"		"-DLLVM_TARGETS_TO_BUILD=host"

imported>Nix: add applications and languages categories

2021-09-20T06:02:36Z

add applications and languages categories

← Older revision		Revision as of 06:02, 20 September 2021
Line 58:		Line 58:
	$ ./inst/bin/clang $CFLAGS -o main main.c		$ ./inst/bin/clang $CFLAGS -o main main.c
	</syntaxHighlight>		</syntaxHighlight>

			[[Category:Applications]]
			[[Category:Languages]]

imported>Mic92: add llvm tutorial

2020-10-28T18:02:08Z

add llvm tutorial

New page

From [[Wikipedia:LLVM|Wikipedia]]:
:The LLVM compiler infrastructure project is a "collection of modular and reusable compiler and toolchain technologies" used to develop compiler front ends and back ends.

== Build LLVM/clang from source ==

LLVM/clang when compiled from source won't find headers/libraries/startup files without compiler wrapper. However it is possible to use the following <code>shell.nix</code> to get started and provide the missing flags to the compiler manually

<syntaxHighlight lang=nix>
with import <nixpkgs> {};
let
gccForLibs = stdenv.cc.cc;
in stdenv.mkDerivation {
name = "llvm-env";
buildInputs = [
bashInteractive
ninja
cmake
llvmPackages_latest.llvm
];

# where to find libgcc
NIX_LDFLAGS="-L${gccForLibs}/lib/gcc/${targetPlatform.config}/${gccForLibs.version}";
# teach clang about C startup file locations
CFLAGS="-B${gccForLibs}/lib/gcc/${targetPlatform.config}/${gccForLibs.version} -B ${stdenv.cc.libc}/lib";

cmakeFlags = [
"-DGCC_INSTALL_PREFIX=${gcc}"
"-DC_INCLUDE_DIRS=${stdenv.cc.libc.dev}/include"
"-GNinja"
# Debug for debug builds
"-DCMAKE_BUILD_TYPE=Release"
# inst will be our installation prefix
"-DCMAKE_INSTALL_PREFIX=../inst"
"-DLLVM_INSTALL_TOOLCHAIN_ONLY=ON"
# change this to enable the projects you need
"-DLLVM_ENABLE_PROJECTS=clang;libcxx;libcxxabi"
# this makes llvm only to produce code for the current platform, this saves CPU time, change it to what you need
"-DLLVM_TARGETS_TO_BUILD=host"
];
}
</syntaxHighlight>

After that you can execute the following commands to get a working clang:

<syntaxHighlight lang=console>
$ git clone https://github.com/llvm/llvm-project/
$ mkdir build && cd build
$ cmake $cmakeFlags ../llvm-project/llvm
$ ninja
# installs everything to ../inst
$ ninja install
$ cd ..
</syntaxHighlight>

Then assuming you have a test program called <code>main.c</code>:

<syntaxHighlight lang=console>
$ ./inst/bin/clang $CFLAGS -o main main.c
</syntaxHighlight>