I have a question about using CMake to link the shared library.
After building OpenBLAS from source and install it succesfully.
In /opt/OpenBLAS/include has header files and
in /opt/OpenBLAS/lib has shared and static libraries.
If I have a toy program call cpp_gemm.cpp, I can build it by typing
g++ -o cpp_gemm cpp_gemm.cpp -I/opt/OpenBLAS/include -L/opt/OpenBLAS/lib -lopenblas
And execute it without error message.
But I want to use CMake to build it, so I write following rules in CMakeLists.txt
cmake_minimum_required(VERSION 2.8.4)
project(CppGEMM)
include_directories(/opt/OpenBLAS/include)
link_directories(/opt/OpenBLAS/lib)
add_executable(cpp_gemm cpp_gemm.cpp)
target_link_libraries(cpp_gemm openblas)
After I create build folder and enter into it, I type
cmake ..
make
to build the program successfully.
But I will get
Illegal instruction (core dumped)
after execute it.
After searching the related problems by Google search, I found that
replace
link_directories(/opt/OpenBLAS/lib)
with
link_directories(/opt/OpenBLAS)
can solve my problem.
That is, I can build the program successfully and execute it without the
error message.
I'm confused, but I can not find any material to realize it.
Can anyone tell me why this occurred?
Many thanks!
Related
If you have an answer for this, or further information, I'd welcome it. I'm following advice from here, to offer some unsolicited help by posting this question then an answer I've already found for it.
I have a bare-metal ARM board for which I'm building a cross-toolchain, from sources for GNU binutils, gcc and gdb, and for SourceWare's Newlib. I got those four working and cross-built a DoNothing.c into an ELF file - but I couldn't disassemble it with this:
$ arm-none-eabi-objdump -S DoNothing.elf
The error was:
$ arm-none-eabi-objdump: error while loading shared libraries: libdebuginfod.so.1: cannot open shared object file: No such file or directory
I'll follow up with a solution.
The error was correct - my system didn't have libdebuginfod.so.1 installed - but I have another cross-binutils, installed from binary for a different target, and its objdump -S works fine on the same host. Why would one build of objdump complain about missing that shared library, when clearly not all builds of objdump need it?
First I tried rebuilding cross binutils, specifying --without-debuginfod as a configure option. No change, which seems odd: surely that should build tools that not only don't use debuginfod but which don't depend on it in any way. (If someone can answer that, or point out what I've misunderstood, it may help people.)
Next I figured debuginfod was inescapable (for my cross-tools built from source at least), so I'd install it to get rid of the error. It's a component of the elfutils package, but installing the latest elfutils available for my Ubuntu 20.04 system didn't bring libdebuginfod.so.1 with it.
I found a later one, for Arch Linux, whose package contents suggested it would - but its package format doesn't match Ubuntu's and installing it was going to involve a lot of work. Instead I opted to build it from the Arch Linux source package. However, running ./configure on that gave a couple of infuriatingly similar errors:
configure: checking libdebuginfod dependencies, --disable-libdebuginfod or --enable-libdebuginfo=dummy to skip
...
configure: error: dependencies not found, use --disable-libdebuginfod to disable or --enable-libdebuginfod=dummy to build a (bootstrap) dummy library.
No combination of those suggestions would allow configure for elfutils-0.182 to run to completion.
The problem of course was my own lack of understanding. The solution came from the Linux From Scratch project: what worked was to issue configure with both of the suggested options, like this:
$ ./configure --prefix=/usr \
--disable-debuginfod \
--enable-libdebuginfod=dummy \
--libdir=/lib
That gave a clean configure; make worked first time, as did make check and then sudo make install which of course installed libdebuginfod.so.1 as required. I then had an arm-none-eabi-objdump which disassembles cross-compiled ELF files without complaining.
i am new to stackflow and i previously i have no background in computer system and programming. However, now i need to run analysis under cygwin for my bioinformatics project. I encounter some error when i try to compile a file name 'zone_b.linux'using cygwin, to produce an executable program. The linux file is download from web https://github.com/haddocking/HADDOCK-binding-sites-tutorial/blob/master/ana_scripts/zone_b.linux. When i try to compile using the following command under cygwin it produce the following error:
$ gcc zone_b.linux
/usr/lib/gcc/i686-pc-cygwin/6.4.0/../../../libcygwin.a(libcmain.o): In
function `main':
/usr/src/debug/cygwin-2.9.0-3/winsup/cygwin/lib/libcmain.c:37: undefined
reference to `WinMain#16'
collect2: error: ld returned 1 exit status
Error description
I search the following error under stackoverflow, and i found two post with similar problem.
First is the post from undefined reference to `WinMain#16'. It stated that the problem is due the Microsoft'linker uses a runtime library entry point(winMainCRTStartup) that calls Microsoft's non-standard WinMain instead of standard main. So, i try the post's suggestion of including the entry by following command
$ gcc zone_b.linux /entry:winMainCRTStartup
gcc: error: /entry:winMainCRTStartup: No such file or directory
However i get the error no such file or directory. I think maybe it is because i am running under cygwin not mingW.
Second post is the Undefined reference to WinMain in Cygwin. From the post, it said use -c compile flag to only produce object file. However, for my case, i am not using any -c. Therefore, i think it is not relevant to my issue.
I would appreciate if anyone could kindly explain to me since i am new to this computing area. Thank you.
zone_b.linux is the compiled and linked executable program to run on a linux machine. It is a 32-bit ELF binary file. It will not work on a Windows machine, even using cygwin or mingw32, without re-compulation.
You probably have to compile zone_b.f, a FORTRAN source file, using the gfortran compiler to create a zone_b.exe that is usable in cygwin. I saw no instructions for this, but try something like gcc zone_b.f and cross fingers. Be sure gfortran is installed using cygwin setup.
You will also need to (re-)build the other executables (cluster_struc and contact) by performing make in the ana_scripts directory. Any supplied executables (from the git clone ... or a downloaded .zip file) will not work under cygwin.
You will need to have perl and python installed. I think perl is installed by default. You can install python2 using cygwin setup. The python script looked like it will work with python2 or python3, whichever is the default. On cygwin, today, python2 is the default python. I don't do perl, so cross your fingers.
I am installing Open MPI v1.8.8 with CUDA v7.5 on my Linux Debian.
I have tested CUDA and it works, tested OpenMPI and it works too. But when i try to combine them into a program, i meet an error: cannot find cuda.h file . This is my scenario:
My program source code include these .h file
include "cuda.h"
include "mpi.h"
I run command:
mpicc <filePath> -o test
And error appear:cuda.h: No such file or directory
#include "cuda.h"`
omp_info give me : mca:mpi:base:param:mpi_built_with_cuda_support:value:false
I have googled , and i followed some methods i found:
./configure --with-cuda
./configure --with-cuda=/usr/local/cuda-7.5
( source link : http://mirror.its.dal.ca/openmpi/faq/?category=buildcuda)
After that, i remake all , remake install Open Mpi. I run: mpicc or mpirun, the compiler give me error : mpirun error mca: base: component find: unable to open /usr/local/lib/openmpi/mca_mpool_sm
I set up soft link : ln -s /usr/local/cuda/include /usr/include ( describe in link : Building CUDA-aware openMPI on Ubuntu 12.04 cannot find cuda.h).
But it cannot fix my issue.
Does anyone successfully install it? Please help me or share your experience.
Thanks.
I think you are confusing installation problems with incorrect compiler options. It will be necessary to explicity specify the include paths, library paths, and libraries for CUDA when compiling and linking host code with your mpi wrapped host compiler.
Something like:
mpicc -I/usr/local/cuda-7.5/include -L/usr/local/cuda-7.5/lib -o test <filePath> -lcuda
would be the normal way to build a simple MPI program which call the cuda driver APIs. You will need to add nvcc compilation for device code and host code which uses the runtime API.
The apparent lack of CUDA support in your MPI flavour is a separate question and one you should probably take up in another forum (like the user mailing list of the MPI flavour you use).
I am building GCC-4.4.2 ( open embedded recipe using bitbake version-1.8 ) for an x86 based system.
The configure options are as under:
"--enable-libssp --disable-bootstrap --disable-libgomp --disable-libmudflap"
It builds successfully after some minor changes and libstdc++.so.6.0.13 library is getting created.
My aim is to run Citrix receiver on an x86 based target. For this, I run storebrowse as follows:
./usr/lib/ICAClient/storebrowse
But I get the following errors.
./storebrowse: /usr/lib/libxml2.so.2: no version information available (required by ./storebrowse)
./storebrowse: relocation error: ./storebrowse: symbol _ZTISt15basic_streambufIwSt11char_traitsIwEE, version GLIBCXX_3.4 not defined in file libstdc++.so.6 with link time reference.
libxml2 error can be resolved by deleting and creating a new symbolic link to libxml. I am unable to resolve the symbol error (_ZTISt15basic_streambufIwSt11char_traitsIwEE). I am compiling using default gcc_4.4.2.bb recipe.
Do I need to make any changes in configure options..?
It looks like your executable is loading the system's (older) libstdc++. To fix this, either configure GCC with --enable-rpath or explicitly add RPATH to your call to g++ with -Wl,-rpath=/path/to/your/gcc/lib, or run your executable like this:
LD_LIBRARY_PATH=/path/to/your/gcc/lib ./usr/lib/ICAClient/storebrowse
I wrote those three possible solutions in order of my personal preference. :)
This page may be of some interest: https://gcc.gnu.org/onlinedocs/libstdc++/manual/abi.html
I'm in the process of attempting to port some code across from PC to Ubuntu, and am having some issues due to limited experience developing under linux.
We use CMake to generate all our build stuff. Under windows I'm making VS2010 projects, and under Linux I'm making Eclipse projects. I've managed to get my OpenCV stuff ported across successfully, but am having major headaches trying to port my threaded boost apps.
Just so we're clear, the steps I have followed so-far on a clean Ubuntu 12 installation. (I've done 2 clean re-installs to try and fix potential library cock-ups, now I'm just giving up and asking):
Install Eclipse and Eclipse CDT using my package manager
Install CMake and CMake Gui using my package manager
Install libboost-all-dev using my package manager
So-far that's all I've done. I can create the eclipse project using CMake with no errors, so CMake is successfully finding my boost install. When I try and build through eclipse is when I get issues; The app I'm attempting to build uses boost::asio for some UDP I/O and boost::thread to create worker threads for the asio I/O services. I can successfully compile each module, but when I come to link I get spammed with errors such as:
/usr/bin/c++ CMakeFiles/RE05DevelopmentDemo.dir/main.cpp.o CMakeFiles/RE05DevelopmentDemo.dir/RE05FusionListener/RE05FusionListener.cpp.o CMakeFiles/RE05DevelopmentDemo.dir/NewEye/NewEye.cpp.o -o RE05DevelopmentDemo -rdynamic -Wl,-Bstatic -lboost_system-mt -lboost_date_time-mt -lboost_regex-mt -lboost_thread-mt -Wl,-Bdynamic
/usr/lib/gcc/x86_64-linux-gnu/4.6/../../../../lib/libboost_thread-mt.a(thread.o): In function `void boost::call_once<void (*)()>(boost::once_flag&, void (*)()) [clone .constprop.98]':
make[2]: Leaving directory `/home/david/Code/Build/Support/RE05DevDemo'
(.text+0xc8): undefined reference to `pthread_key_create'
/usr/lib/gcc/x86_64-linux-gnu/4.6/../../../../lib/libboost_thread-mt.a(thread.o): In function `boost::this_thread::interruption_enabled()':
(.text+0x540): undefined reference to `pthread_getspecific'
make[1]: Leaving directory `/home/david/Code/Build/Support/RE05DevDemo'
/usr/lib/gcc/x86_64-linux-gnu/4.6/../../../../lib/libboost_thread-mt.a(thread.o): In function `boost::this_thread::disable_interruption::disable_interruption()':
(.text+0x570): undefined reference to `pthread_getspecific'
/usr/lib/gcc/x86_64-linux-gnu/4.6/../../../../lib/libboost_thread-mt.a(thread.o): In function `boost::this_thread::disable_interruption::disable_interruption()':
(.text+0x59f): undefined reference to `pthread_getspecific'
Some Gotchas that I have collected from other StackOverflow posts and have already checked:
The boost libs are all present at /usr/lib
I am not getting any compile errors for inability to find the boost headers, so they must be getting found.
I am trying to link statically, but I believe eclipse should be passing the correct arguments to make that happen since my CMakeLists.txt includes SET(Boost_USE_STATIC_LIBS ON)
I'm officially out of ideas here, I have tried doing local builds of boost and a bunch of other stuff with no more success. I even re-installed Ubuntu to ensure I haven't completely fracked the libs directories and links with multiple weird versions or anything else. Any help would be muchly appreciated.
Correct mechanism is to use Threads package:
find_package(Threads)
#...
target_link_libraries(my_app ${CMAKE_THREAD_LIBS_INIT} ...)
See also cmake and libpthread
When you are building your targets, add -lpthread and it will compile.
See this other thread.
OK, so I found the solution.
It was to do with the absence of the -lpthread flag in the link command. In order to get CMake to link appropriately, then the TARGET_LINK_LIBRARIES line needs to be edited. My edit is:
Original:
TARGET_LINK_LIBRARIES( RE05DevelopmentDemo ${Boost_LIBRARIES} )
Modified and working:
IF(WIN32)
TARGET_LINK_LIBRARIES( RE05DevelopmentDemo ${Boost_LIBRARIES} )
ELSE(WIN32)
TARGET_LINK_LIBRARIES( RE05DevelopmentDemo ${Boost_LIBRARIES} pthread )
ENDIF(WIN32)
I'm guessing that I should probably change the ELSE(WIN32) to an elseif for or use the CMake command FindThreads to link in pthread if needed, but I'm not sure how to do that at the moment and have more imporatant things on my plate given the time I've lost. Interestingly enough, I noticed my link command now has two -lpthread flags appended at the end, one after another, but everything is still compiling quite happily.