Clang 10 has been released. I downloaded the source and build it myself as follows:
cmake -DCMAKE_BUILD_TYPE=Release -DPYTHON_EXECUTABLE=/usr/bin/python3 ../llvm
make -j16
make install
The resulting binary of clang is located at /usr/local/bin/clang-10, and its size is horriably 123MB!
If I install the clang-9 by dnf install clang, the binary size of clang-9 is just 201KB!
Any wrong steps?
Related
I am trying to build LLVM compilers so that I can enable OpenMP on the Apple M1.
I am using the LLVM development tree, (since I saw some OpenMP runtime go into that for this recently).
I have ended up with this script to invoke cmake:
# Xcode, Ninja
BUILD_SYSTEM=Ninja
BUILD_TAG=Ninja
cmake ../llvm \
-G$BUILD_SYSTEM -B ${BUILD_TAG}_build \
-DCMAKE_OSX_ARCHITECTURES='arm64' \
-DCMAKE_C_COMPILER=`which clang` \
-DCMAKE_CXX_COMPILER=`which clang++` \
-DCMAKE_BUILD_TYPE=Release \
-DCMAKE_BUILD_WITH_INSTALL_RPATH=1 \
-DCMAKE_INSTALL_PREFIX=$HOME/software/clang-12.0.0/arm64 \
-DLLVM_ENABLE_WERROR=FALSE \
-DLLVM_TARGETS_TO_BUILD='AArch64' \
-DLLVM_ENABLE_PROJECTS='clang;openmp,polly' \
-DLLVM_DEFAULT_TARGET_TRIPLE='aarch64-apple-darwin20.1.0'
The compilers used here are
$ /usr/bin/clang --version
Apple clang version 12.0.0 (clang-1200.0.32.27)
Target: arm64-apple-darwin20.1.0
Thread model: posix
InstalledDir: /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin
ninja can then successfully build clang, clang++ and the OpenMp runtime and install them. (As simple, Arm64 images targeting Arms64)
$ file ~/software/clang-12.0.0/arm64/bin/clang
/Users/jcownie/software/clang-12.0.0/arm64/bin/clang: Mach-O 64-bit executable arm64
$ ~/software/clang-12.0.0/arm64/bin/clang --version
clang version 12.0.0 (https://github.com/llvm/llvm-project.git 879c15e890b4d25d28ea904e92497f091f796019)
Target: aarch64-apple-darwin20.1.0
Thread model: posix
InstalledDir: /Users/jcownie/software/clang-12.0.0/arm64/bin
Which all looks sane, except that when I try to compile anything with them they are missing the include path to get system headers.
$ ~/software/clang-12.0.0/arm64/bin/clang hello.c
hello.c:1:10: fatal error: 'stdio.h' file not found
#include <stdio.h>
^~~~~~~~~
1 error generated.
So, after all that,
Does anyone know how to fix that include path problem?
Does anyone know how to configure and build a fat binary for the compilers (and libraries) so that the x86_64 embedded compiler targets x86_64 and the aarch64 binary aarch64? (This is what the Xcode clang and clang++ do...)
My attempt at this ended up with a compiler fat binary where both architectures targeted x86_64 :-(
Thanks
You can set -DDEFAULT_SYSROOT=/path/to/MacOSX11.1.sdk at build time or do export SDKROOT=/path/to/MacOSX11.1.sdk at runtime.
You need to compile with clang -arch arm64 -arch x86_64 to get a fat binary out of clang. You need to do this for Apple clang as well.
UPDATED 8 Feb 2021
Homebrew now supports the M1 based Arm machines, so using that is a better answer than the one below.
The info below is potentially still useful if you want to do this on your own, but using brew is likely to be much simpler.
Pre-brew answer
I haven't found a clean solution, but in case it helps anyone else, I do have a horrible hack.
The full recipe, then is configure with this script, then build and install.
# Xcode, Ninja
BUILD_SYSTEM=Ninja
BUILD_TAG=ninja
INSTALLDIR=$HOME/software/clang-12.0.0/arm64
cmake ../llvm \
-G$BUILD_SYSTEM -B ${BUILD_TAG}_build \
-DCMAKE_OSX_ARCHITECTURES='arm64' \
-DCMAKE_C_COMPILER=`which clang` \
-DCMAKE_CXX_COMPILER=`which clang++` \
-DCMAKE_BUILD_TYPE=Release \
-DCMAKE_INSTALL_PREFIX=$INSTALLDIR \
-DLLVM_LOCAL_RPATH=$INSTALLDIR/lib \
-DLLVM_ENABLE_WERROR=FALSE \
-DLLVM_TARGETS_TO_BUILD='AArch64' \
-DLLVM_DEFAULT_TARGET_TRIPLE='aarch64-apple-darwin20.1.0' \
-DDEFAULT_SYSROOT="$(xcrun --show-sdk-path)" \
-DLLVM_ENABLE_PROJECTS='clang;openmp;polly;clang-tools-extra;libcxx;libcxxabi' \
# -DLLVM_ENABLE_PROJECTS='clang;openmp;polly'
That gives a compiler that finds the right headers, but won't link successfully if OpenMP is used because it doesn't pass on any useful -L path or add a necessary rpath.
To overcome that I created a small shell script that sits in my ~/bin, at the front of my $PATH, which adds those extra linker flags.
#
# A truly awful hack, but it seems necessary.
# Install this with execute permissions as clang and clang++ in
# a directory early in your path, so that it is executed when clang or
# clang++ is needed.
#
# For brew...
INSTALLDIR=/usr/local/opt/llvm
# For a local build.
INSTALLDIR=${HOME}/software/clang-12.0.0/arm64/
# Find out the name of this file, and then invoke the same file in the
# compiler installation, adding the necessary linker directives
CMD=`echo $0 | sed "s/\/.*\///"`
${INSTALLDIR}/bin/${CMD} -L${INSTALLDIR}/lib -Wl,-rpath,${INSTALLDIR}/lib $*
I am not recommending this particularly; there should clearly be a better way to make it work, but it'll do for now, and lets me get back to using the compiler rather than building it!
I was able to build with -DDEFAULT_SYSROOT="$(xcrun --show-sdk-path)" -DCMAKE_INSTALL_PREFIX=/Users/foo/lokal/ and install into the lokal/bin lokal/lib path. Once that is done you can use LD_LIBRARY_PATH=/Users/foo/lokal/lib and all the libraries should be found without mucking with anything else rpath related.
These are the step I follow to install doxygen 1.8.20:
apt-get install graphviz
apt-get install bison
apt-get install flex
wget http://ftp.gnu.org/pub/gnu/libiconv/libiconv-1.16.tar.gz
tar -xvzf libiconv-1.16.tar.gz
cd libiconv-1.16
./configure --prefix=/usr/local/libiconv
make
make install
wget http://doxygen.nl/files/doxygen-1.8.20.src.tar.gz
gunzip doxygen-1.8.20.src.tar.gz
tar -xf doxygen-1.8.20.src.tar
cd doxygen-1.8.20
mkdir build
cd build
cmake -G "Unix Makefiles" -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=/usr -Duse_libclang=ON ..
At this point I get the following output:
CMake Error at CMakeLists.txt:51 (find_package):
Could not find a package configuration file provided by "LLVM" with any of
the following names:
LLVMConfig.cmake
llvm-config.cmake
Add the installation prefix of "LLVM" to CMAKE_PREFIX_PATH or set
"LLVM_DIR" to a directory containing one of the above files. If "LLVM"
provides a separate development package or SDK, be sure it has been
installed.
-- Configuring incomplete, errors occurred!
See also "/home/doxygen-1.8.20/build/CMakeFiles/CMakeOutput.log".
Now I get LLVM package:
apt-get install llvm
cmake -G "Unix Makefiles" -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=/usr -Duse_libclang=ON ..
At this point I get the following output:
CMake Error at CMakeLists.txt:52 (find_package):
Could not find a package configuration file provided by "Clang" with any of
the following names:
ClangConfig.cmake
clang-config.cmake
Add the installation prefix of "Clang" to CMAKE_PREFIX_PATH or set
"Clang_DIR" to a directory containing one of the above files. If "Clang"
provides a separate development package or SDK, be sure it has been
installed.
-- Configuring incomplete, errors occurred!
Get clang package:
apt-get install clang
Get the same error:
CMake Error at CMakeLists.txt:52 (find_package):
Could not find a package configuration file provided by "Clang" with any of
the following names:
ClangConfig.cmake
clang-config.cmake
Add the installation prefix of "Clang" to CMAKE_PREFIX_PATH or set
"Clang_DIR" to a directory containing one of the above files. If "Clang"
provides a separate development package or SDK, be sure it has been
installed.
-- Configuring incomplete, errors occurred!
See also "/home/doxygen-1.8.20/build/CMakeFiles/CMakeOutput.log".
Now What?
This was found to be a problem with the llvm and clang that get installed with apt-get install llvm and apt-get install clang. This installs version 6 of both.
Instead of these commands, do the following:
apt install llvm-10 clang-10 libclang-10-dev
Then running
cmake -G "Unix Makefiles" -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=/usr -Duse_libclang=ON ..
will produce successful output:
-- Found PythonInterp: /usr/bin/python (found version "2.7.17")
-- Found FLEX: /usr/bin/flex (found version "2.6.4")
-- Found BISON: /usr/bin/bison (found version "3.0.4")
-- Looking for pthread.h
-- Looking for pthread.h - found
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD - Failed
-- Looking for pthread_create in pthreads
-- Looking for pthread_create in pthreads - not found
-- Looking for pthread_create in pthread
-- Looking for pthread_create in pthread - found
-- Found Threads: TRUE
-- Looking for iconv_open
-- Looking for iconv_open - found
-- Performing Test ICONV_COMPILES
-- Performing Test ICONV_COMPILES - Success
-- Found Iconv: In glibc
-- One (and only one) of the ICONV_ACCEPTS_... tests must pass
-- Performing Test ICONV_ACCEPTS_NONCONST_INPUT
-- Performing Test ICONV_ACCEPTS_NONCONST_INPUT - Success
-- Performing Test ICONV_ACCEPTS_CONST_INPUT
-- Performing Test ICONV_ACCEPTS_CONST_INPUT - Failed
-- The javacc executable not found, using existing files
-- Configuring done
-- Generating done
-- Build files have been written to: /home/libiconv-1.16/doxygen-1.8.20/build
and proceed with make and make install
I have a tough need to compile Coreutils with llvm for other arch: arm/aarch64/mips/mips32/ppc/ppc32...
Since I install all the gcc-cross tools like mips-linux-gnu, powerpc64-linux-gnu and if I have a simple C program like that test.c
#include<stdio.h>
int main(){
printf("hello!");
return 0;
}
I can compile it to the arch, i.e.
clang --target=mips64-linux-gnuabi64 test.c -o test-mips64
➜ tests file test-mips64
test-mips64: ELF 64-bit MSB executable, MIPS, MIPS64 rel2 version 1 (SYSV), dynamically linked, interpreter /lib64/ld.so.1, BuildID[sha1]=7b33d55a0d08e6cd18d966341590dc351e346a78, for GNU/Linux 3.2.0, not stripped
I try to the same way for compile Coreutils that try to set
export CC=clang
export CXX=clang++
CFLAGS = "--target=mips64-linux-gnuabi64"
./configure --host=mips64-linux-gnuabi64
Howerver, every time got errors in configure or make...
How should I set the configure? Can I easily compile Coreuntils with llvm for other archs?
It's a bit tricky to get the command-line options right for cross-compiling. I got it to work with the commands below, assuming you're working on a Debian-based system (like Debian or Ubuntu). Here are the steps.
Install gcc-mips64-linux-gnuabi64 and gcc-powerpc64-linux-gnu.
Choose the correct arguments for CFLAGS
-B/usr/mips64-linux-gnuabi64/bin/ to indicate we want to use the linker ld within that directory. Do the same for powerpc.
--target=mips64-linux-gnuabi64 to indicate what our target for compilation is. Do the same for powerpc.
-I/usr/mips64-linux-gnuabi64/include to include header files. Do the same for powerpc.
Use ./configure --host=mips64-linux-gnuabi to configure for mips64 and ./configure --host=powerpc64-linux-gnueabi to configure for powerpc64.
Here are the commands to compile for mips64:
make clean
CFLAGS="-B/usr/mips64-linux-gnuabi64/bin/ --target=mips64-linux-gnuabi64 -I/usr/mips64-linux-gnuabi64/include" \
./configure --host=mips64-linux-gnuabi
make
And the commands to compile for powerpc64:
make clean
CFLAGS="-B/usr/powerpc64-linux-gnu/bin/ --target=powerpc64-linux-gnueabi -I/usr/powerpc64-linux-gnu/include" \
./configure --host=powerpc64-linux-gnueabi
make
Here is the output of file ./src/ls to demonstrate that it is a powerpc64 executable:
$ file ./src/ls
./src/ls: ELF 64-bit MSB executable, 64-bit PowerPC or cisco 7500, version 1 (SYSV), dynamically linked, interpreter /lib64/ld64.so.1, for GNU/Linux 3.2.0, BuildID[sha1]=97fe33981ca0112160f44a6fb678d6dc1b462114, not stripped
Below is a Dockerfile that can be used to reproducibly cross-compile coreutils for mips64 and powerpc64.
# Cross-compile GNU coreutils for mips64 and powerpc64 using clang.
# With help from https://medium.com/#wolfv/cross-compiling-arm-on-travis-using-clang-and-qemu-2b9702d7c6f3
FROM debian:buster
# Install compile-time dependencies.
RUN apt-get update \
&& apt-get install --yes \
clang \
curl \
gcc-mips64-linux-gnuabi64 \
gcc-powerpc64-linux-gnu \
make \
perl \
&& rm -rf /var/lib/apt/lists/*
# Download source code for release.
WORKDIR /tmp/coreutils
RUN curl -fsSL https://ftp.gnu.org/gnu/coreutils/coreutils-8.32.tar.xz \
| tar xJ --strip-components 1
# Compile and install for mips64.
RUN CFLAGS="-B/usr/mips64-linux-gnuabi64/bin/ --target=mips64-linux-gnuabi64 -I/usr/mips64-linux-gnuabi64/include" \
./configure --host=mips64-linux-gnuabi --prefix=/opt/coreutils-mips \
&& make \
&& make install
# Compile and install for powerpc64.
RUN make clean \
&& CFLAGS="-B/usr/powerpc64-linux-gnu/bin/ --target=powerpc64-linux-gnueabi -I/usr/powerpc64-linux-gnu/include" \
./configure --host=powerpc64-linux-gnueabi --prefix=/opt/coreutils-powerpc64 \
&& make \
&& make install
# Keep only the compiled programs from the previous stage.
FROM debian:buster
COPY --from=0 /opt /opt
I am current working on a simple build tool in Python that maybe help you.
Unfortunately, still at moment, lacks clang implementation, but works fine with GCC and MSVC.
Basically the thing mix Json parameters files to generate command line building.
CppMagic
Environment: Ubuntu 18.04 bionic.
after git clone llvm-project from https://github.com/llvm/llvm-project.
I generated build credentials using Cmake commandline like this:
cmake -DLLVM_TARGETS_TO_BUILD=X86 \
-DLLVM_TARGET_ARCH=X86 \
-DCMAKE_BUILD_TYPE="Release" \
-DLLVM_BUILD_EXAMPLES=1 \
-DCLANG_BUILD_EXAMPLES=1 \
-G "Unix Makefiles" \
../llvm/
after makefile was generated, I then use make to build the project make -j8. somehow after build completion, clang and clang++ etc. are nowhere to be found in /build/bin/.
To my knowledge, if I didn't specify LLVM_ENABLE_PROJECTs, it will build LLVM and clang both by default. why are clang executables missing?
You need to explicitly specify LLVM_ENABLE_PROJECT and put the source code folder of clang in the same level of llvm source folder. The folder name must match the value of LLVM_ENABLE_PROJECT
I'm trying real hard to install vowpal wobbit and it fails when i run the make file, throwing:
cd library; make; cd ..
g++ -g -o ezexample temp2.cc -L ../vowpalwabbit -l vw -l allreduce -l boost_program_options -l z -l pthread
ld: library not found for -lboost_program_options collect2: ld returned 1 exit status make[1]: *** [ezexample] Error 1'
I then added the links to the boost library here by specifying -L/usr/local/lib
Now I get the following error:
g++ -g -o ezexample temp2.cc -L/usr/local/lib ../vowpalwabbit -l vw -l allreduce -l boost_program_options -l z -l pthread
ld: library not found for -lvw
collect2: ld returned 1 exit status
make: *** [ezexample] Error 1
I happened to get everything working on OS X 10.7 as follows:
Make sure you have a working Boost installation. As indicated on the Getting started page, usually we only need header files, but some Boost libraries must be built separately, including the program_options library which is used to process options from command line or config file. Go into your boost folder, and then at your shell prompt:
$ ./bootstrap.sh
$ ./bjam
This will compile and build everything. You should now have a bin.v2/ directory in your boost directory, with all built libraries for your system (static and threaded libs).
$ ls bin.v2/libs/
date_time iostreams python serialization test
filesystem math random signals thread
graph program_options regex system wave
More importantly, extra Boost libraries are made available in the stage/lib/ directory. For me, these are Mach-O 64-bit dynamically linked shared library x86_64.
The include path should be your_install_dir/boost_x_xx_x, where boost_x_xx_x is the basename of your working Boost. (I personally have boost_1_46_1 in /usr/local/share/ and I symlinked it to /usr/local/share/boost to avoid having to remember version number.) The library path (for linking) should read your_install_dir/boost_x_xx_x/stage/lib. However, it might be best to symlink or copy (which is what I did) everything in usual place, i.e. /usr/local/include/boost for header files, and /usr/local/lib for libraries.
Edit the Makefile from the vowpal_wabbit directory, and change the include/library paths to reflect your current installation. The Makefile should look like this (first 12 lines):
COMPILER = g++
UNAME := $(shell uname)
ifeq ($(UNAME), FreeBSD)
LIBS = -l boost_program_options -l pthread -l z -l compat
BOOST_INCLUDE = /usr/local/include
BOOST_LIBRARY = /usr/local/lib
else
LIBS = -l boost_program_options -l pthread -l z
BOOST_INCLUDE = /usr/local/share/boost # change path to reflect yours
BOOST_LIBRARY = /usr/local/share/boost/stage/lib # idem
endif
Then, you are ready to compile vowpal_wabbit (make clean in case you already compiled it):
$ make
$ ./vw --version
6.1
$ make test
You can also install vowpal wabbit on OS X using brew:
brew install vowpal-wabbit
Or you can just install boost, and then install vw from the github repo.
brew install boost
For installation on CentOS 7 (6.5 perl version is too old for latest vw source code), I've found the instructions at http://wkoplitz.blogspot.be/2012/12/vowpal-wabbit-on-centos.html to work fine:
yum install zlib-devel boost-devel
yum groupinstall "Development Tools"
git clone git://github.com/JohnLangford/vowpal_wabbit.git
cd vowpal_wabbit
./autogen.sh
make
make test
Good news:
As of the latest release VowpalWabbit version 9.1.0, vw no longer relies on Boost program_options
From the release highlights:
Removal of Boost Program Options dependency
For a long time we have depended on Boost Program Options
for command line options parsing. In this release, we have > replaced this dependency with our own implementation of
command line parsing. Apart from one place where we depend > on Boost Math in standalone mode, this means that VW core
and the command line tool are free of Boost dependencies
hopefully making the code a bit easier to build and package.
Vowpal Wabbit 9.1.0 release notes