1 ===================================================================
2 How To Cross-Compile Clang/LLVM using Clang/LLVM
3 ===================================================================
8 This document contains information about building LLVM and
9 Clang on host machine, targeting another platform.
11 For more information on how to use Clang as a cross-compiler,
12 please check http://clang.llvm.org/docs/CrossCompilation.html.
14 TODO: Add MIPS and other platforms to this document.
16 Cross-Compiling from x86_64 to ARM
17 ==================================
19 In this use case, we'll be using CMake and Ninja, on a Debian-based Linux
20 system, cross-compiling from an x86_64 host (most Intel and AMD chips
21 nowadays) to a hard-float ARM target (most ARM targets nowadays).
23 The packages you'll need are:
26 * ``ninja-build`` (from backports in Ubuntu)
27 * ``gcc-4.7-arm-linux-gnueabihf``
28 * ``gcc-4.7-multilib-arm-linux-gnueabihf``
29 * ``binutils-arm-linux-gnueabihf``
30 * ``libgcc1-armhf-cross``
31 * ``libsfgcc1-armhf-cross``
32 * ``libstdc++6-armhf-cross``
33 * ``libstdc++6-4.7-dev-armhf-cross``
38 For more information on how to configure CMake for LLVM/Clang,
41 The CMake options you need to add are:
43 * ``-DCMAKE_CROSSCOMPILING=True``
44 * ``-DCMAKE_INSTALL_PREFIX=<install-dir>``
45 * ``-DLLVM_TABLEGEN=<path-to-host-bin>/llvm-tblgen``
46 * ``-DCLANG_TABLEGEN=<path-to-host-bin>/clang-tblgen``
47 * ``-DLLVM_DEFAULT_TARGET_TRIPLE=arm-linux-gnueabihf``
48 * ``-DLLVM_TARGET_ARCH=ARM``
49 * ``-DLLVM_TARGETS_TO_BUILD=ARM``
51 If you're compiling with GCC, you can use architecture options for your target,
52 and the compiler driver will detect everything that it needs:
54 * ``-DCMAKE_CXX_FLAGS='-march=armv7-a -mcpu=cortex-a9 -mfloat-abi=hard'``
56 However, if you're using Clang, the driver might not be up-to-date with your
57 specific Linux distribution, version or GCC layout, so you'll need to fudge.
59 In addition to the ones above, you'll also need:
61 * ``'-target arm-linux-gnueabihf'`` or whatever is the triple of your cross GCC.
62 * ``'--sysroot=/usr/arm-linux-gnueabihf'``, ``'--sysroot=/opt/gcc/arm-linux-gnueabihf'``
63 or whatever is the location of your GCC's sysroot (where /lib, /bin etc are).
64 * Appropriate use of ``-I`` and ``-L``, depending on how the cross GCC is installed,
65 and where are the libraries and headers.
67 The TableGen options are required to compile it with the host compiler,
68 so you'll need to compile LLVM (or at least ``llvm-tblgen``) to your host
69 platform before you start. The CXX flags define the target, cpu (which in this case
70 defaults to ``fpu=VFP3`` with NEON), and forcing the hard-float ABI. If you're
71 using Clang as a cross-compiler, you will *also* have to set ``--sysroot``
72 to make sure it picks the correct linker.
74 When using Clang, it's important that you choose the triple to be *identical*
75 to the GCC triple and the sysroot. This will make it easier for Clang to
76 find the correct tools and include headers. But that won't mean all headers and
77 libraries will be found. You'll still need to use ``-I`` and ``-L`` to locate
78 those extra ones, depending on your distribution.
80 Most of the time, what you want is to have a native compiler to the
81 platform itself, but not others. So there's rarely a point in compiling
82 all back-ends. For that reason, you should also set the
83 ``TARGETS_TO_BUILD`` to only build the back-end you're targeting to.
85 You must set the ``CMAKE_INSTALL_PREFIX``, otherwise a ``ninja install``
86 will copy ARM binaries to your root filesystem, which is not what you
92 There are some bugs in current LLVM, which require some fiddling before
95 #. If you're using Clang as the cross-compiler, there is a problem in
96 the LLVM ARM back-end that is producing absolute relocations on
97 position-independent code (``R_ARM_THM_MOVW_ABS_NC``), so for now, you
102 -DLLVM_ENABLE_PIC=False
104 This is not a problem, since Clang/LLVM libraries are statically
105 linked anyway, it shouldn't affect much.
107 #. The ARM libraries won't be installed in your system.
108 But the CMake prepare step, which checks for
109 dependencies, will check the *host* libraries, not the *target*
110 ones. Below there's a list of some dependencies, but your project could
111 have more, or this document could be outdated. You'll see the errors
112 while linking as an indication of that.
114 Debian based distros have a way to add ``multiarch``, which adds
115 a new architecture and allows you to install packages for those
116 systems. See https://wiki.debian.org/Multiarch/HOWTO for more info.
118 But not all distros will have that, and possibly not an easy way to
119 install them in any anyway, so you'll have to build/download
122 A quick way of getting the libraries is to download them from
123 a distribution repository, like Debian (http://packages.debian.org/jessie/),
124 and download the missing libraries. Note that the ``libXXX``
125 will have the shared objects (``.so``) and the ``libXXX-dev`` will
126 give you the headers and the static (``.a``) library. Just in
129 The ones you need for ARM are: ``libtinfo``, ``zlib1g``,
130 ``libxml2`` and ``liblzma``. In the Debian repository you'll
131 find downloads for all architectures.
133 After you download and unpack all ``.deb`` packages, copy all
134 ``.so`` and ``.a`` to a directory, make the appropriate
135 symbolic links (if necessary), and add the relevant ``-L``
136 and ``-I`` paths to ``-DCMAKE_CXX_FLAGS`` above.
139 Running CMake and Building
140 --------------------------
142 Finally, if you're using your platform compiler, run:
146 $ cmake -G Ninja <source-dir> <options above>
148 If you're using Clang as the cross-compiler, run:
152 $ CC='clang' CXX='clang++' cmake -G Ninja <source-dir> <options above>
154 If you have ``clang``/``clang++`` on the path, it should just work, and special
155 Ninja files will be created in the build directory. I strongly suggest
156 you to run ``cmake`` on a separate build directory, *not* inside the
159 To build, simply type:
165 It should automatically find out how many cores you have, what are
166 the rules that needs building and will build the whole thing.
168 You can't run ``ninja check-all`` on this tree because the created
169 binaries are targeted to ARM, not x86_64.
174 After the LLVM/Clang has built successfully, you should install it
181 which will create a sysroot on the install-dir. You can then tar
182 that directory into a binary with the full triple name (for easy
183 identification), like:
187 $ ln -sf <install-dir> arm-linux-gnueabihf-clang
188 $ tar zchf arm-linux-gnueabihf-clang.tar.gz arm-linux-gnueabihf-clang
190 If you copy that tarball to your target board, you'll be able to use
191 it for running the test-suite, for example. Follow the guidelines at
192 http://llvm.org/docs/lnt/quickstart.html, unpack the tarball in the
193 test directory, and use options:
197 $ ./sandbox/bin/python sandbox/bin/lnt runtest nt \
199 --test-suite `pwd`/test-suite \
200 --cc `pwd`/arm-linux-gnueabihf-clang/bin/clang \
201 --cxx `pwd`/arm-linux-gnueabihf-clang/bin/clang++
203 Remember to add the ``-jN`` options to ``lnt`` to the number of CPUs
204 on your board. Also, the path to your clang has to be absolute, so
205 you'll need the `pwd` trick above.