How to add linker option to CMake bootstrap? - cmake

I'm building CMake 3.12.4 from the release tarball on AIX. CMake is failing to link on the machine:
ld: 0711-781 ERROR: TOC overflow. TOC size: 65632 Maximum size: 65536
The error is detailed in the IBM technotes at ld: 0711-781 ERROR: TOC overflow. I want to add -bbigtoc linker option for the cmake recipe.
CMake's bootstrap does not appear to accept linker options for LDFLAGS as shown below.
How do I add a linker flag to the bootstrap process?
Here are the options CMake bootstrap accepts:
$ ./bootstrap --help
Usage: ./bootstrap [<options>...] [-- <cmake-options>...]
Options: [defaults in brackets after descriptions]
Configuration:
--help print this message
--version only print version information
--verbose display more information
--parallel=n bootstrap cmake in parallel, where n is
number of nodes [1]
--enable-ccache Enable ccache when building cmake
--init=FILE load FILE as script to populate cache
--system-libs use all system-installed third-party libraries
(for use only by package maintainers)
--no-system-libs use all cmake-provided third-party libraries
(default)
--system-curl use system-installed curl library
--no-system-curl use cmake-provided curl library (default)
--system-expat use system-installed expat library
--no-system-expat use cmake-provided expat library (default)
--system-jsoncpp use system-installed jsoncpp library
--no-system-jsoncpp use cmake-provided jsoncpp library (default)
--system-zlib use system-installed zlib library
--no-system-zlib use cmake-provided zlib library (default)
--system-bzip2 use system-installed bzip2 library
--no-system-bzip2 use cmake-provided bzip2 library (default)
--system-liblzma use system-installed liblzma library
--no-system-liblzma use cmake-provided liblzma library (default)
--system-libarchive use system-installed libarchive library
--no-system-libarchive use cmake-provided libarchive library (default)
--system-librhash use system-installed librhash library
--no-system-librhash use cmake-provided librhash library (default)
--system-libuv use system-installed libuv library
--no-system-libuv use cmake-provided libuv library (default)
--qt-gui build the Qt-based GUI (requires Qt >= 4.2)
--no-qt-gui do not build the Qt-based GUI (default)
--qt-qmake=<qmake> use <qmake> as the qmake executable to find Qt
--sphinx-info build Info manual with Sphinx
--sphinx-man build man pages with Sphinx
--sphinx-html build html help with Sphinx
--sphinx-qthelp build qch help with Sphinx
--sphinx-build=<sb> use <sb> as the sphinx-build executable
--sphinx-flags=<flags> pass <flags> to sphinx-build executable
Directory and file names:
--prefix=PREFIX install files in tree rooted at PREFIX
[/usr/local]
--bindir=DIR install binaries in PREFIX/DIR
[bin]
--datadir=DIR install data files in PREFIX/DIR
[share/cmake-3.12]
--docdir=DIR install documentation files in PREFIX/DIR
[doc/cmake-3.12]
--mandir=DIR install man pages files in PREFIX/DIR/manN
[man]
--xdgdatadir=DIR install XDG specific files in PREFIX/DIR
[share]
Here is the CMake link error:
cmake_build$ make VERBOSE=1
...
[ 91%] Linking CXX executable ../bin/cmake
cd /home/noloader/cmake_build/Source && /home/noloader/cmake_build/Boots
trap.cmk/cmake -E cmake_link_script CMakeFiles/cmake.dir/link.txt --verbose=1
/usr/bin/g++ -pthread -Wl,-bnoipath -Wl,-brtl -Wl,-bexpall CMakeFiles/cmake.dir
/cmakemain.cxx.o CMakeFiles/cmake.dir/cmcmd.cxx.o -o ../bin/cmake -Wl,-blibpath
:/opt/freeware/lib/gcc/powerpc-ibm-aix7.1.0.0/4.8.1/pthread:/opt/freeware/lib/pt
hread:/opt/freeware/lib/gcc/powerpc-ibm-aix7.1.0.0/4.8.1:/opt/freeware/lib:/usr/
lib:/lib libCMakeLib.a libCMakeServerLib.a libCMakeLib.a kwsys/libcmsys.a ../Uti
lities/cmexpat/libcmexpat.a ../Utilities/cmlibarchive/libarchive/libcmlibarchive
.a ../Utilities/cmliblzma/libcmliblzma.a ../Utilities/cmbzip2/libcmbzip2.a ../Ut
ilities/cmcompress/libcmcompress.a ../Utilities/cmcurl/lib/libcmcurl.a ../Utilit
ies/cmzlib/libcmzlib.a -lld ../Utilities/cmjsoncpp/libcmjsoncpp.a ../Utilities/c
mlibuv/libcmlibuv.a -lperfstat ../Utilities/cmlibrhash/libcmlibrhash.a
ld: 0711-781 ERROR: TOC overflow. TOC size: 65632 Maximum size: 65536
collect2: error: ld returned 12 exit status
make: 1254-004 The error code from the last command is 1.

Something like this:
export CFLAGS=...
export CXXFLAGS=...
export LDFLAGS='... -Wl,-bbigtoc'
./bootstrap ... --verbose
make VERBOSE=1 all
Also you might want to read this: http://lzsiga.users.sourceforge.net/aix-linking.html#Q0025

Related

Compiling mpfr 4.0.2 on AIX fails to configure TLS support

I'm trying to compile mpfr on AIX and get an error finding TLS support. My end goal is to have a working gcc (9.3) with libstdc++ support for threads and TLS.
I tried compiling with both an existing gcc on the machine (which is 8.3 and doesn't have TLS support, as it seems) and using the built-in xlC (V12.1) on the machine.
I compiled GMP-6.2.0:
(gcc attempt): ./configure --enable-shared --enable-static CFLAGS=-fPIC CXXFLAGS=-fPIC --prefix=$OUTDIR --host=powerpc-ibm-aix7.1.0.0
(xlc attempt): ./configure --enable-shared --enable-static CC=xlc CXX=xlc CFLAGS=-qPIC CXXFLAGS=-qPIC --prefix=$OUTDIR --host=powerpc-ibm-aix7.1.0.0
gmake all
gmake check
gmake install
In both cases gmp successfully tests and installs.
Trying to configure mpfr-4.0.2 results in
(gcc attempt): ./configure --enable-shared --enable-static CFLAGS=-fPIC CXXFLAGS=-fPIC --with-gmp=$OUTDIR --prefix=$OUTDIR --host=powerpc-ibm-aix7.1.0.0
(xlc attempt): ./configure --enable-shared --enable-static CC=xlc CXX=xlc CFLAGS="-qPIC -qtls" CXXFLAGS="-qPIC -qtls" --with-gmp=$OUTDIR --prefix=$OUTDIR --host=powerpc-ibm-aix7.1.0.0
results in the following error in config.log
(gcc attempt):
configure:16567: checking for TLS support
configure:16595: gcc -o conftest -fPIC -I{OUTDIR}/include -I./src -L{OUTDIR}/lib conftest.c >&5
ld: 0711-317 ERROR: Undefined symbol: __tls_get_addr
(xlc attempt):
configure:16567: checking for TLS support
configure:16595: xlc -o conftest -qPIC -qtls -I{OUTDIR}/include -I./src -L{OUTDIR}/lib conftest.c >&5
ld: 0711-317 ERROR: Undefined symbol: .__tls_get_addr
I've tried a dozen different flags and configurations but mostly I don't think I'm in the correct direction - every dependency seems circular. I believe compiling with xlc is the correct idea here, but I just can't figure how to get it properly support TLS.
I also thought that I can ignore this and continue, and later compile libstdc++ again with support for TLS - but that also brought me to a dead end.
Edit:
Taking more actions on using xlc: I tried compiling the generated conftest.c (https://gist.github.com/NitzanEgozy/265e54e3353aefc79558c860b543a98f) manually:
$ xlc -o conftest -qPIC -qtls -I$OUTDIR/include -I./src -L$OUTDIR/lib conftest.c -bnoquiet
(ld): halt 4
(ld): setfflag 4
(ld): savename conftest
(ld): filelist 6 1
(ld): i /lib/crt0.o
(ld): i conftest.o
(ld): lib /usr/vac/lib/libxlopt.a
(ld): lib /usr/vac/lib/libxlipa.a
(ld): lib /usr/vac/lib/libxl.a
(ld): lib /usr/lib/libc.a
LIBRARY: Shared object libc.a[shr.o]: 3301 symbols imported.
LIBRARY: Shared object libc.a[meth.o]: 2 symbols imported.
LIBRARY: Shared object libc.a[posix_aio.o]: 20 symbols imported.
LIBRARY: Shared object libc.a[aio.o]: 18 symbols imported.
LIBRARY: Shared object libc.a[pse.o]: 8 symbols imported.
LIBRARY: Shared object libc.a[dl.o]: 4 symbols imported.
LIBRARY: Shared object libc.a[pty.o]: 1 symbols imported.
LIBRARY: Shared object libc.a[cthread.o]: 25 symbols imported.
FILELIST: Number of previously inserted files processed: 6
(ld): resolve
RESOLVE: 33 of 7029 symbols were kept.
(ld): addgl /usr/lib/glink.o
ADDGL: Glink code added for 2 symbols.
(ld): er full
ld: 0711-318 ERROR: Undefined symbols were found.
The following symbols are in error:
Symbol Inpndx TY CL Source-File(Object-File) OR Import-File{Shared-object}
RLD: Address Section Rld-type Referencing Symbol
----------------------------------------------------------------------------------------------
.__tls_get_addr [30] ER PR conftest.c(conftest.o)
00000010 .text R_RBA [16] .main
Removing -qtls throws The '__thread' keyword is not supported on the target platform. The keyword is ignored.
AFAIK if native TLS is not available gcc emulates it using pthread keys, and i think gcc does not even support native TLS on aix. adding "-pthread" flag to the configuration will enable the appropriate threading support.
the current gcc from the ibm aixtoolbox has the appropriate default threading model enabled, hope this helps a little bit.
root#aixbuildhostng: /root # gcc -v
Using built-in specs.
COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=/opt/freeware/libexec/gcc/powerpc-ibm-aix7.1.0.0/8.3.0/lto-wrapper
Target: powerpc-ibm-aix7.1.0.0
Configured with: ../gcc-8.3.0/configure --prefix=/opt/freeware --mandir=/opt/freeware/man --infodir=/opt/freeware/info --with-local-prefix=/opt/freeware --with-as=/usr/bin/as --with-ld=/usr/bin/ld --enable-languages=c,c++,fortran,objc,obj-c++ --enable-version-specific-runtime-libs --disable-nls --enable-decimal-float=dpd --with-cloog=no --with-ppl=no --disable-libstdcxx-pch --enable-__cxa_atexit --host=powerpc-ibm-aix7.1.0.0
Thread model: aix
gcc version 8.3.0 (GCC)
mpfr 4.0.2 is also available as an rpm package:
root#aixbuildhostng: /root # yum search mpfr
Loaded plugins: aliases, allowdowngrade, basearchonly, changelog, downloadonly, fastestmirror, filter-data, keys, list-data, local, merge-conf, post-transaction-actions, priorities, protectbase, ps, remove-with-leaves, rpm-warm-cache,
: show-leaves, tmprepo, tsflags, verify, versionlock
Loading mirror speeds from cached hostfile
0 packages excluded due to repository protections
============================================================================================================= N/S Matched: mpfr =============================================================================================================
mpfr-devel.ppc : Development tools A C library for mpfr library
mpfr.ppc : A C library for multiple-precision floating-point computations
ad bootstrapping gcc via xlc...this quite a sophisticated task, i never managed to make this work.
https://gcc.gnu.org/install/specific.html#x-ibm-aix

Use LLVM in a cmake build

I'm trying to build my own project that use LLVM. I downloaded the source code and the precompiled package on the official web site (last version).
http://releases.llvm.org/download.html
I downloaded :
LLVM source code
Clang for Windows (64-bit)
FYI, I don't build LLVM... only want to use it !
I followed the instruction here :
http://llvm.org/docs/CMake.html#embedding-llvm-in-your-project
in the section : "Embedding LLVM in your project"
So, I added this code :
find_package(LLVM REQUIRED CONFIG)
message(STATUS "Found LLVM ${LLVM_PACKAGE_VERSION}")
message(STATUS "Using LLVMConfig.cmake in: ${LLVM_DIR}")
message("LLVM_INCLUDE_DIRS=${LLVM_INCLUDE_DIRS}")
message("LLVM_DEFINITIONS=${LLVM_DEFINITIONS}")
But I got several cmake error messages, here is my output :
-- Using LLVMConfig.cmake in: C:\\Luciad_src\\libs\\LLVM\\cmake\\modules
LLVM_INCLUDE_DIRS=
LLVM_DEFINITIONS=
CMake Error at C:/Luciad_src/libs/LLVM/cmake/modules/LLVM-Config.cmake:31 (list):
list sub-command REMOVE_ITEM requires two or more arguments.
Call Stack (most recent call first):
C:/Luciad_src/libs/LLVM/cmake/modules/LLVM-Config.cmake:256 (is_llvm_target_library)
components/query/CMakeLists.txt:15 (llvm_map_components_to_libnames)
Is there a problem with my script, or the packages I use ? Any idea ?
Thanks for your help
You can have the LLVM as binary or source package.
The binary package does not include CMake support. If you compare both installations (binary on the left, source after building and installing it on the right) you can see the difference:
LLVM-5.0.0-win64.exe
<=> llvm-5.0.1.src.tar.xz (build and installed)
So you need to build and install the source package first to get CMake support. On my Windows machine I needed a cmd shell with administrator rights, a Visual Studio installation, go to the downloaded and extracted sources and do:
> mkdir mybuilddir
> cd mybuilddir
> cmake ..
> cmake --build . --config Release --target INSTALL
If I now use your CMake example I get:
-- Found LLVM 5.0.1
-- Using LLVMConfig.cmake in: C:/Program Files (x86)/LLVM/lib/cmake/llvm
LLVM_INCLUDE_DIRS=C:/Program Files (x86)/LLVM/include
LLVM_DEFINITIONS=-DLLVM_BUILD_GLOBAL_ISEL -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64 -D_CRT_SECURE_NO_DEPRECATE -D_CRT_SECURE_NO_WARNINGS -D_CRT_NONSTDC_NO_DEPRECATE -D_CRT_NONSTDC_NO_WARNINGS -D_SCL_SECURE_NO_DEPRECATE -D_SCL_SECURE_NO_WARNINGS -DUNICODE -D_UNICODE -D__STDC_CONSTANT_MACROS -D__STDC_FORMAT_MACROS -D__STDC_LIMIT_MACROS
-- Configuring done

Error when linking GSL with -static

I have written a programm in c++. Linking and runiing is working, as long as I don't use the "-static" option for g++. But I have to run it from an Antergos USB-Live Stick with default settings and there is no GSL included. In the manual of GSL they recommend
$ g++ -c main.cpp
$ g++ -static main.o -lgsl -lgslcblas -lm -lnlopt
But for this code I receive an error message:
/usr/bin/ld: cannot find -lgsl
/usr/bin/ld: cannot find -lgslcblas
collect2: Fehler: ld gab 1 als End-Status zurück
I tried it as this question, but it didn't work for me. When I run
$ g++ -O2 -o test main.cpp -lgsl -lgslcblas -lnlopt -lm
$ lld test
it prints
linux-vdso.so.1 (0x00007fffa5b95000)
libgsl.so.19 => /usr/lib/libgsl.so.19 (0x00007f8748c9a000)
libgslcblas.so.0 => /usr/lib/libgslcblas.so.0 (0x00007f8748a5d000)
libstdc++.so.6 => /usr/lib/libstdc++.so.6 (0x00007f87486d5000)
libm.so.6 => /usr/lib/libm.so.6 (0x00007f87483d1000)
libgcc_s.so.1 => /usr/lib/libgcc_s.so.1 (0x00007f87481ba000)
libc.so.6 => /usr/lib/libc.so.6 (0x00007f8747e1c000)
/lib64/ld-linux-x86-64.so.2 (0x00007f87490fe000)
So I tried to create a symlink, but I do have also "libgsl.so"
$ ls /usr/lib/libgsl
libgslcblas.so libgslcblas.so.0.0.0 libgsl.so.19
libgslcblas.so.0 libgsl.so libgsl.so.19.3.0
Am I doing something stupid? Thank your for your help.
When you pass -lgsl, by default you request the linker to
find and link either the shared library libgsl.so or the static
library libgsl.a and to prefer the shared library, if both are found
in the same search directory. The linker will search, first, in any
directories you have specified with the -L/path/to/search options,
in the order you specified, and then in its default search directories
(/usr/lib, etc.). Likewise for -lgslcblas.
But when you pass the linkage option -static to gcc/g++, it prevents
linking with any shared libraries. Shared libraries, libgsl.so, libgslcblas.so
will be ignored. Static libraries libgsl.a, libgslblas.a, must be
found, in some or other of the search directories, for the linkage to
succeed.
The linker is saying:
/usr/bin/ld: cannot find -lgsl
/usr/bin/ld: cannot find -lgslcblas
because it can't find those static libraries - presumably because you
haven't installed them.
You do not say what linux distro you are working on, but if the package
that provides libgsl and libgslcblas is called, say, libgsl[suffix]
then there will be a corresponding package called libgsl-dev, libgsl-devel,
or similar. This will be the development version of the package,
for the use of people who want to develop software that links with libgsl
or libgslcblas. The development package will require the libgsl package as a dependency
- so it will install the same stuff - and will in addition contain the
library's header files and the static version of the library.
So you need to install the libgsl development package for your distro.
For Ubuntu, for example, that is libgsl-dev:
Later
I gather that your distro, Arch Linux, does not do separate dev packages. You
need to build the static libraries from source. To do that you will need
at least to have installed:
GNU Make
GNU autotools (autoconf, automake, libtool)
GCC (C compiler)
texinfo
Then to make a default build:
Get the gsl source package from https://savannah.gnu.org/git/?group=gsl
either by cloning the git repo or downloading a current tar.gz tarball
and extracting it.
cd into the package directory.
run ./autogen.sh. This will succeed provided the GNU autotools prerequisites
are fulfilled.
run ./configure --enable-maintainer-mode (as ./autogen.sh will have prompted you).
This will succeed provided that the package dependencies are satisfied
and environment sanity checks pass.
run make
If make completes without errors - which will take a matter of minutes -
then, as root, run make install.
If all is well, this will install your missing static libraries:
/usr/local/lib/libgsl.a
/usr/local/lib/libgslcblas.a
You should not need to modify your linkage command for the linker to find
them: /usr/local/lib is a default linker search path.

linking with particular version of a library

I am linking boost libraries with my .cpp files. The build machine has boost 1.55 in /usr/lib64 and I have boost 1.57 in my local directory. The cmake generates the following link command.
/home/ramki/mpich-install/bin/mpicxx -fopenmp -fexceptions -fno-use-linker-plugin CMakeFiles/factor.dir/factor.cpp.o CMakeFiles/factor.dir/factor_jobs.cpp.o -o factor -rdynamic -lboost_serialization -lboost_iostreams -lboost_program_options -lboost_mpi -llapack -llapacke -lblas -lpthread -lm -lz factorization/libfactorization.a
The above link command does not specify the version of the boost libraries. Because of this I get the following error.
/usr/bin/ld: warning: libboost_serialization.so.1.57.0, needed by /home/ramki/libraries/boost_1_57_0//lib/libboost_mpi.so, may conflict with libboost_serialization.so.1.55.0
Because of this nature of linking, when I use ldd to dump the linked libraries of the executable, I see it linked with couple of libboost 1.55 libraries. If the machine in which I run this executable does not have boost 1.55, it does not start at all.
In the CMakeLists.txt and CMakeCache.txt, I see that the find_package is discovering the 1.57 libraries.
find_package(Boost 1.57.0 COMPONENTS serialization iostreams program_options mpi REQUIRED).
However during linking it is not introducing the version of the library. How do I instruct the cmake to do the following.
linking libraries to use a particular version. For eg., -l:libboost_mpi.so.1.57.0
specify the library path for this version -L library path explicitly. It should NOT link with the library under /usr/lib64.
Ramki
Not sure if this will work for other libraries. Atleast for boost before find_package on boost, set (Boost_REALPATH ON). This will set Boost_LIBRARIES with full path as
/export5/home/ramki/libraries/boost_1_57_0/lib/libboost_serialization.so.1.57.0;/export5/home/ramki/libraries/boost_1_57_0/lib/libboost_iostreams.so.1.57.0;/export5/home/ramki/libraries/boost_1_57_0/lib/libboost_program_options.so.1.57.0;/export5/home/ramki/libraries/boost_1_57_0/lib/libboost_mpi.so.1.57.0.
Use this Boost_LIBRARIES with target_link_libraries(theTarget ${Boost_LIBRARIES}). Thus instead of linking with libboost_mpi.so that could be link to other versions, we are linking with the library of the correct version.

Using libdl.so in MinGW

I want to generate a dll file in MinGW, I have several object dependencies in order to do that, one of my object dependencies is libdl.so, I add this object in unix simply as :
g++ xx.o yy.o /usr/lib/libdl.so -o module.so
but in MinGW, I don't have any idea how to add this object. any ideas?
There is a MinGW port of libdl that you can use just like under Unix. Quote from the website:
This library implements a wrapper for dlfcn, as specified in POSIX and SUS, around the dynamic link library functions found in the Windows API.
It requires MinGW to build.
You may get pre-built binaries (with MinGW gcc 3.4.5) and a bundled source code from the Downloads section.
The following commands build and install it in a standard MinGW installation (to be run from your MinGW shell):
./configure --prefix=/ --libdir=/lib --incdir=/include && make && make install
To compile your library as a DLL, use the following command:
g++ -shared xx.o yy.o -ldl -o module.dll
I encountered the same problem (msys2, 32bit version of compiler etc.).
For me I found out that the libdl.a was available in /usr/lib but not in /mingw32/lib. I was able to solve the problem by linking it to the /mingw32/lib folder:
ln -s /usr/lib/libdl.a /mingw32/lib