I have a code by name mexCodeCuda.cu which uses dynamic parallelism. In the matlab interface, when I try to execute the following commands, I get the following error :
system('nvcc --compile mexCodeCuda.cu -o mexCodeCuda.o --compiler-options -fPIC -m64 -rdc=true -gencode arch=compute_35,code=sm_35 -O3 -lineinfo -use_fast_math -lcudadevrt -I/cm/shared/apps/MATLAB/extern/include')
system('nvcc -arch=sm_35 -dlink mexCodeCuda.o -o dlink.o')
eval('mex mexCodeCuda.o dlink.o -L'CUDA_LIB_PATH '-lcudart')
It gives the following error :
Error using mex
/usr/bin/ld: dlink.o: relocation R_X86_64_32S against `__nv_module_id' can
not be used when making a shared object; recompile with -fPIC
dlink.o: could not read symbols: Bad value
collect2: error: ld returned 1 exit status
Can anyone help me ?
Thanks in advance...
A few changes:
Add -fPIC to the second nvcc command (using --compiler-options, just as you did with the first nvcc).
Specify the cudadevrt runtime option on the mex command (-lcudadevrt), just as you have with the first nvcc call.
And you may need to add -fPIC to the mex linking via LDCXXFLAGS or LDFLAGS. Not sure.
Related
I write this command:
gcc drr.c -o drr -pthread -lcdsl -L./../synch_implementations -I./../synch_implementations
And I can't understand why I get this response:
usr/bin/ld: ./../synch_implementations/libcdsl.a(cdsl_queue_list.o): relocation R_X86_64_32 against `.rodata' can not be used when making a PIE object; recompile with -fPIC
usr/bin/ld: ./../synch_implementations/libcdsl.a(cdsl_deque_list.o): relocation R_X86_64_32S against symbol `cdsl_deque_list_lock_based_pthread_push_tail' can not be used when making a PIE object; recompile with -fPIC
usr/bin/ld: ./../synch_implementations/libcdsl.a(cdsl_lock.o): relocation R_X86_64_32S against symbol `cdsl_pthread_lock_request' can not be used when making a PIE object; recompile with -fPIC
usr/bin/ld: final link failed: Nonrepresentable section on output
ollect2: error: ld returned 1 exit status
Can anyone help me understand what is wrong?
Notes:
1)I use ubuntu (via VirtualBox).
2)I am very inexperienced, so probably this is not a very bright question.
Solution:
When you compile ./../synch_implementations/libcdsl.a you need to add an extra "-fPIC" flag to produce position-independent-code (see comments above).
Note:
The PIC flag is in general needed when producing library object code since it creates objects relative to a global offset table (GOT) which subsequently allows to link library objects with your code.
I'm compiling some C++ functionality into Matlab by using Mex compilation to produce a .mexa64 binary.
Honestly I don't know much about the guts of the symbols created in the .mexa64 binary and how it allows Matlab to call C++, but I'm starting with a simple question related to the error I eventually get when the command mex calls ld.
The ld error is:
/usr/bin/ld.real: duplicate version tag `MEX'
The question is, what does this ld error "duplicate version tag" mean? Other appreciated perspectives are: what does it mean in relation to the Matlab compilation I'm trying to do specifically; and where can I find comprehensive ld error message documentation?
This latter point, the comprehensive ld documentation, I've found particularly hard to find. The manual page for ld didn't seem to be helpful for this.
requested to post the mex config file:
<config Name="g++" ShortName="g++" Manufacturer="GNU" Version="" Language="C++" Priority="A" Location="$GCC_DIR">
<Details CompilerExecutable="$CXX" CompilerDefines="$DEFINES" CompilerFlags="$CXXFLAGS" OptimizationFlags="$CXXOPTIMFLAGS" DebugFlags="$CXXDEBUGFLAGS" IncludeFlags="$INCLUDE" LinkerLibraries="$LINKLIBS" LinkerOptimizationFlags="$LDOPTIMFLAGS" LinkerDebugFlags="$LDDEBUGFLAGS" CompilerDefineFormatter="-D%s" LinkerLibrarySwitchFormatter="-l%s" LinkerPathFormatter="-L%s"/>
<vars CMDLINE1="$CXX -c $DEFINES $INCLUDE $CXXFLAGS $OPTIM $SRC -o $OBJ" CMDLINE2="$LDXX $LDFLAGS $LDTYPE $LINKOPTIM $LINKEXPORT $OBJS $CXXLIBS $LINKLIBS -o $EXE" CMDLINE3="rm -f $OBJ" CXX="$GCC" DEFINES="-D_GNU_SOURCE $MATLABMEX" MATLABMEX="-DMATLAB_MEX_FILE " CXXFLAGS="-ansi -fexceptions -fPIC -fno-omit-frame-pointer -pthread -std=c++11" INCLUDE="-I"$MATLABROOT/extern/include" -I"$MATLABROOT/simulink/include"" CXXOPTIMFLAGS="-O -DNDEBUG" CXXDEBUGFLAGS="-g" LDXX="$CXX" LDFLAGS="-pthread -Wl,--no-undefined " LDTYPE="-shared" LINKEXPORT="-Wl,--version-script,"$MATLABROOT/extern/lib/$ARCH/mexFunction.map"" LINKLIBS="-Wl,-rpath-link,$MATLABROOT/bin/$ARCH -L"$MATLABROOT/bin/$ARCH" -lmx -lmex -lmat -lm -lstdc++" LDOPTIMFLAGS="-O" LDDEBUGFLAGS="-g" OBJEXT=".o" LDEXT=".mexa64"> </vars>
<client>
<engine LINKLIBS="$LINKLIBS -leng" LINKEXPORT="" LDEXT="" LDTYPE=""/>
<mbuild DEFINES="-D_GNU_SOURCE $MATLABMEX -DUNIX -DX11 -DGLNXA64 -DGCC" CXXFLAGS="-ansi -pthread -std=c++11" INCLUDE="$INCLUDE -I"$MATLABROOT/extern/include/cpp"" LDFLAGS="-pthread -Wl,-rpath-link,$MATLABROOT/bin/$ARCH" LINKEXPORT="" LINKLIBS="-L"$MATLABROOT/runtime/$ARCH" -lm -lmwmclmcrrt" LDEXT="" MATLABMEX="" LDTYPE=""/>
</client>
<locationFinder>
<GCC>
<cmdReturns name="which g++"/>
</GCC>
<CPPLIB_DIR>
<cmdReturns name="g++ -print-file-name=libstdc++.so" diagnostic="GCC appears to be installed, but supporting C++ libraries were not found. You must first install the gcc-c++ packge, or similiar, to complete this action."/>
</CPPLIB_DIR>
</locationFinder>
</config>
I'm trying to install mutt on CentOS 6.7 and ./configure --prefix=$HOME/local runs fine but at make install step, I run into this error below that I don't know how to solve. I tried with mutt 1.6.1 and 1.6.0 and this same error came up so it's probably some system setting that wasn't correct. How could I fix it?
gcc -std=gnu99 -Wall -pedantic -Wno-long-long -g -O2 -o mutt addrbook.o alias.o attach.o base64.o browser.o buffy.o color.o crypt.o cryptglue.o commands.o complete.o compose.o copy.o curs_lib.o curs_main.o date.o edit.o enter.o flags.o init.o filter.o from.o getdomain.o group.o handler.o hash.o hdrline.o headers.o help.o hook.o keymap.o main.o mbox.o menu.o mh.o mx.o pager.o parse.o pattern.o postpone.o query.o recvattach.o recvcmd.o rfc822.o rfc1524.o rfc2047.o rfc2231.o rfc3676.o score.o send.o sendlib.o signal.o sort.o status.o system.o thread.o charset.o history.o lib.o muttlib.o editmsg.o mbyte.o mutt_idna.o url.o ascii.o crypt-mod.o safe_asprintf.o patchlist.o conststrings.o pgp.o pgpinvoke.o pgpkey.o pgplib.o gnupgparse.o pgpmicalg.o pgppacket.o crypt-mod-pgp-classic.o smime.o crypt-mod-smime-classic.o resize.o -lncursesw -lidn
/opt/gnu/bin/ld: curs_lib.o: undefined reference to symbol 'wtimeout'
/lib64/libtinfo.so.5: error adding symbols: DSO missing from command line
collect2: error: ld returned 1 exit status
make[2]: *** [mutt] Error 1
OK, I somehow fixed it.. not really understand why this fixed it.
After configure, in the Makefile, change the LDFLAGS = line to LDFLAGS = -lncurses -ltinfo, then make install and it solved the issue.
Credit to this post: http://www.pclinuxos.com/forum/index.php?topic=129566.0, Tara-Rain-9.5ß's answer
I can not get Perl/Tk to install via CPAN on OS X Mountain Lion. It errors with an error in a file with which google does not help:
/Users/villadelfia/perl5/perlbrew/perls/perl-5.17.8/bin/perl5.17.8 /Users/villadelfia/perl5/perlbrew/perls/perl-5.17.8/lib/5.17.8/ExtUtils/xsubpp -typemap /Users/villadelfia/perl5/perlbrew/perls/perl-5.17.8/lib/5.17.8/ExtUtils/typemap -typemap /Users/villadelfia/Downloads/Tk-804.030/Tk/typemap IO.xs > IO.xsc && mv IO.xsc IO.c
Warning: Found a 'CODE' section which seems to be using 'RETVAL' but no 'OUTPUT' section. in IO.xs, line 235
cc -c -I.. -I/usr/X11R6/include -I/usr/local/include/freetype2 -fno-common -DPERL_DARWIN -fno-strict-aliasing -pipe -fstack-protector -I/usr/local/include -O3 -DVERSION=\"804.03\" -DXS_VERSION=\"804.03\" "-I/Users/villadelfia/perl5/perlbrew/perls/perl-5.17.8/lib/5.17.8/darwin-2level/CORE" -Wall -Wno-implicit-int -Wno-comment -Wno-unused -D__USE_FIXED_PROTOTYPES__ IO.c
IO.xs:210:10: error: invalid argument type 'void' to unary expression
if (!SvUPGRADE(buf, SVt_PV))
^~~~~~~~~~~~~~~~~~~~~~~
1 error generated.
make[1]: *** [IO.o] Error 1
make: *** [subdirs] Error 2
As you can see, I run perlbrew as well.
Any ideas what could cause this?
From the 5.17.7 release notes:
SvUPGRADE() is no longer an expression. Originally this macro (and its
underlying function, sv_upgrade()) were documented as boolean, although
in reality they always croaked on error and never returned false. In 2005
the documentation was updated to specify a void return value, but
SvUPGRADE() was left always returning 1 for backwards compatibility. This
has now been removed, and SvUPGRADE() is now a statement with no return
value.
So this is now a syntax error:
if (!SvUPGRADE(sv)) { croak(...); }
If you have code like that, simply replace it with
SvUPGRADE(sv);
or to to avoid compiler warnings with older perls, possibly
(void)SvUPGRADE(sv);
It has already been reported.
I am trying to configure Eclipse/Helios on a Win7/64 Professional with CDT and MinGW. When I try and compile:
#include <iostream>
using namespace std ;
int main() {
cout << "What a lovely day\n" ;
return(0) ;
}
I get the output/error:
make all
Building file: ../SeeProgram.cpp
Invoking: Cross G++ Compiler
C:\Tools\MinGW\bin\g++ -O0 -g3 -Wall -c -fmessage-length=0 -MMD -MP -MF"SeeProgram.d" -MT"SeeProgram.d" -o"SeeProgram.o" "../SeeProgram.cpp"
/bin/sh: C:ToolsMinGWbing++: command not found
make: *** [SeeProgram.o] Error 127 .
In the next-to-last line, it should have read C:\Tools\MinGW\bin\g++, but it left out the back-slashes. I have not seen any other discussion of this, and have tried numerous approaches, but I am stumped at the moment. Any thoughts?
Many thanks.
Have you tried escaping the backslashes when you enter the location, e.g.:
C:\\Tools\\MingGW\\bin\\g++
This setup seems similar to yours, maybe it could provide a hint:
http://max.berger.name/howto/cdt/ar01s03.jsp#wincompiler