I’m using GCC to compile a shared library for ARM Linux. Here’s my compiler options from CMakeLists.txt:
add_definitions( "-std=c++14 -fvisibility=hidden -fvisibility-inlines-hidden -Wall -Wno-psabi -march=native -mfpu=neon" )
I’ve just opened the resulting .so file in a disassembler. I was disappointed to see a lot of stuff there. It showed me names for everything, including all my internal classes, and functions that were never exported. Even the stuff from anonymous namespaces is still there. On the “Exports” tab on the disassembler, I only see a dozen of functions I actually export (plus just a few extra: .init_proc .term_proc _edata __bss_end__ and call_weak_fn).
On Windows, I only see these things if I have a PDB file for the module I’m disassembling. But I don’t ship my .PDB files.
I’d like the same behavior for GCC.
Is there a way to tell GCC to stop including that debug info (=the mangled name of each and every function) in the .so file, and actually compile these things into binary addresses?
You need to add -Wl,--strip-debug to your linker flags.
Related
I'm trying to use CUDA code inside MATLAB mex, under linux. With the "whole program compilation" mode, it works good for me. I take the following two steps inside Nsight:
(1) Add "-fPIC" as a compiler option to each .cpp or .cu file, then compile them separately, each producing a .o file.
(2) Set the linker command to be "mex" and add "-cxx" to indicate that the type of all the .o input files are cpp files, and add the library path for cuda. Also add a cpp file that contains the mexFunction entry as an additional input.
This works good and the resulted mex file runs well under MATLAB. After that when I need to use dynamical parallelism, I have to switch to the "separate compilation mode" in Nsight. I tried the same thing above but the linker produces a lot of errors of missing reference, which I wasn't able to resolve.
Then I checked the compilation and linking steps of the "separate compilation" mode. I got confused by what it is doing. It seems that Nsight does two compilation steps for each .cpp or .cu file and produces a .o file as well as a .d file. Like this:
/usr/local/cuda-5.5/bin/nvcc -O3 -gencode arch=compute_35,code=sm_35 -odir "src" -M -o "src/tn_matrix.d" "../src/tn_matrix.cu"
/usr/local/cuda-5.5/bin/nvcc --device-c -O3 -gencode arch=compute_35,code=compute_35 -gencode arch=compute_35,code=sm_35 -x cu -o "src/tn_matrix.o" "../src/tn_matrix.cu"
The linking command is like this:
/usr/local/cuda-5.5/bin/nvcc --cudart static --relocatable-device-code=true -gencode arch=compute_35,code=compute_35 -gencode arch=compute_35,code=sm_35 -link -o "test7" ./src/cu_base.o ./src/exp_bp_wsj_dev_mex.o ./src/tn_main.o ./src/tn_matlab_helper.o ./src/tn_matrix.o ./src/tn_matrix_lib_dev.o ./src/tn_matrix_lib_host.o ./src/tn_model_wsj_dev.o ./src/tn_model_wsj_host.o ./src/tn_utility.o -lcudadevrt -lmx -lcusparse -lcurand -lcublas
What's interesting is that the linker does not take the .d file as input. So I'm not sure how it dealt with these files and how I should process them with the "mex" command when linking?
Another problem is that the linking stage has a lot of options I don't understand (--cudart static --relocatable-device-code=true), which I guess is the reason why I cannot make it work like in the "whole program compilation" mode. So I tried the following:
(1) Compile in the same way as in the beginning of the post.
(2) Preserve the linking command as provided by Nsight but change to use "-shared" option, so that the linker produces a lib file.
(3) Invoke mex with input the lib file and another cpp file containing the mexFunction entry.
This way mex compilation works and it produces a mex executable as output. However, running the resulted mex executable under MATLAB produces a segmentation fault immediately and crashes MATLAB.
I'm not sure if this way of linking would cause any problem. More strangely, I found that the mex linking step seems to finish trivially without even checking the completeness of the executable, because even if I miss a .cpp file for some function that the mexFunction will use, it still compiles.
EDIT:
I figured out how to manually link into a mex executable which can run correctly under MATLAB, but I haven't figured out how to do that automatically under Nsight, which I can in the "whole program compilation" mode. Here is my approach:
(1) Exclude from build the cpp file which contains the mexFunction entry. Manually compile it with the command "mex -c".
(2) Add "-fPIC" as a compiler option to each of the rest .cpp or .cu file, then compile them separately, each producing a .o file.
(3) Linking will fail because it cannot find the main function. We don't have it since we use mexFunction and it is excluded. This doesn't matter and I just leave it there.
(4) Follow the method in the post below to manually dlink the .o files into a device object file
cuda shared library linking: undefined reference to cudaRegisterLinkedBinary
For example, if step (2) produces a.o and b.o, here we do
nvcc -gencode arch=compute_35,code=sm_35 -Xcompiler '-fPIC' -dlink a.o b.o -o mex_dev.o -lcudadevrt
Note that here the output file mex_dev.o should not exist, otherwise the above command will fail.
(5) Use mex command to link all the .o files produced in step (2) and step (4), with all necessary libraries supplied.
This works and produces runnable mex executable. The reason I cannot automate step (1) inside Nsight is because if I change the compilation command to "mex", Nsight will also use this command to generate a dependency file (the .d file mentioned in the question text). And the reason I cannot automate step (4) and step (5) in Nsight is because it involves two commands, which I don't know how to put them in. Please let me know if you knows how to do these. Thanks!
OK, I figured out the solution. Here are the complete steps for compiling mex programs with "separate compilation mode" in Nsight:
Create a cuda project.
In the project level, change build option for the following:
Switch on -fPIC in the compiler option of "NVCC compiler" at the project level.
Add -dlink -Xcompiler '-fPIC' to "Expert Settings" "Command Line Pattern" of the linker "NVCC Linker"
Add letter o to "Build Artifact" -> "Artifact Extension", since by -dlink in the last step we are making the output a .o file.
Add mex -cxx -o path_to_mex_bin/mex_bin_filename ./*.o ./src/*.o -lcudadevrt to "Post Build Steps", (add other necessary libs)
UPDATE: In my actual project I moved the last step to a .m file in MATLAB, because otherwise if I do it while my mex program is running, it could cause MATLAB crash.
For files needs to be compiled with mex, change these build option for each of them:
Change the compiler to GCC C++ Compiler in Tool Chain Editor.
Go back to compiler setting of GCC C++ Compiler and change Command to mex
Change command line pattern to ${COMMAND} -c -outdir "src" ${INPUTS}
Several additional notes:
(1) Cuda specific details (such as kernel functions and calls to kernel functions) must be hidden from the mex compiler. So they should be put in the .cu files rather than the header files. Here is a trick to put templates involving cuda details into .cu files.
In the header file (e.g., f.h), you put only the declaration of the function like this:
template<typename ValueType>
void func(ValueType x);
Add a new file named f.inc, which holds the definition
template<>
void func(ValueType x) {
// possible kernel launches which should be hidden from mex
}
In the source code file (e.g., f.cu), you put this
#define ValueType float
#include "f.inc"
#undef ValueType
#define ValueType double
#include "f.inc"
#undef ValueType
// Add other types you want.
This trick can be easily generalized for templated classes to hide details.
(2) mex specific details should also be hidden from cuda source files, since the mex.h will alter the definitions of some system functions, such as printf. So including of "mex.h" should not appear in header files that can potentially be included in the cuda source files.
(3) In the mex source code file containing the entry mexFunction, one can use the compiler macro MATLAB_MEX_FILE to selectively compile code sections. This way th source code file can be compiled into both mex executable or ordinarily executable, allowing debugging under Nsight without matlab. Here is a trick for building multiple targets under Nsight: Building multiple binaries within one Eclipse project
First of all, it should be possible to set up Night to use a custom Makefile rather than generate it automatically. See Setting Nsight to run with existing Makefile project.
Once we have a custom Makefile, it may be possible to automate (1), (4), and (5). The advantage of a custom Makefile is that you know exactly what compilation commands will take place.
A bare-bones example:
all: mx.mexa64
mx.mexa64: mx.o
mex -o mx.mexa64 mx.o -L/usr/local/cuda/lib64 -lcudart -lcudadevrt
mx.o: mxfunc.o helper.o
nvcc -arch=sm_35 -Xcompiler -fPIC -o mx.o -dlink helper.o mxfunc.o -lcudadevrt
mxfunc.o: mxfunc.c
mex -c -o mxfunc.o mxfunc.c
helper.o: helper.c
nvcc -arch=sm_35 -Xcompiler -fPIC -c -o helper.o helper.c
clean:
rm -fv mx.mexa64 *.o
... where mxfunc.c contains the mxFunction but helper.c does not.
EDIT: You may be able achieve the same effect in the automatic compilation system. Right click on each source file and select Properties, and you'll get a window where you can add some compilation options for that individual file. For linking options, open Properties of the project. Do some experiments and pay attention to the actual compilation commands that show up in the console. In my experience, custom options sometimes interact with the automatic system in a weird way. If this method proves too troublesome for you, I suggest that you make a custom Makefile; this way, at least we are not caught by unexpected side-effects.
There is a set with - files with extension.с: avl_tree.c, buf_read.c, db_prep.c, file_process.c, global_header.c, traverser.c. Used include files are in folder/usr/gcc/4.4/bin/include except for jni.h, and libraries are in folder/usr/gcc/4.4/bin/lib. How from them to create.so the file (if it is possible specify all options in this command)? It me interests in communication by creation of native of methods by means of JNI.
You really should read the documentation of GCC. Notably invoking GCC. The program library howto is also relevant.
Very often, some builder is used to drive the build. GNU make is often used and has a good tutorial documentation. If your Makefile-s are complex, you may also want to use GNU remake to debug them (remake is a debugging variant for make).
You usually want to compile each individual C source file into position independent code because shared objects have PIC code. You can use
gcc -Wall -fPIC -o foo.pic.o foo.c
to compile a C source foo.c into a position independent object file foo.pic.o and you may need some other compiler options (e.g. -I to add include directories, or -D to define some preprocessor symbols, -g for debugging, and -O for optimizing).
I strongly suggest to enable almost all warnings with -Wall (and to improve your code till no warnings are given; this will improve a little bit your code's quality).
Then you have to link all these *.pic.o files together into a shared object with
gcc -shared *.pic.o -o foo.so
You can link some shared libraries into a shared object.
You may want to read Levine's book on linkers and loaders
Of course if you use GNU make you'll have rules in your Makefile for all this.
You could use GNU libtool also.
Maybe dlopen(3) could interest you.
The question should probably give more information.
Most sets of sources have a Makefile, configure script or some other item to set up to make the output (the .so library you want).
gcc -dynamic -o file.so file.c
will create an so file from one of the source files, but you probably want a single so from all of them.
I have a shared library which is supposed to export only one function which is marked with __attribute__ ((visibility ("default"))). It also links with another static library (fftw), and
#include<fftw3.h>
is preceded with:
#pragma GCC visibility push(hidden)
The linker command used:
g++.exe -fvisibility=hidden -shared -o mylib.dll -Wl,--out-implib,mylib.dll.a -Wl,--no-whole-archive libfftw3.a libfftw3_omp.a -lgomp
Now the resulting library is huge and if I check the exported functions it includes ALL fftw functions, and ALL function from my files. It looks like mingw ignores visibility options. I read that previously it gave warning about -fvisibility, but now it compiles with no warnings whatsoever.
Does mingw and gcc 4.6.1 support visibility flags? If yes, how do I get rid of all unnecessary stuff in my shared library?
Mingw is a Windows port of GCC toolchain but Windows dll are not Linux so. Especially the link part is different. To specify the visibility with MingGW you have to go the Windows way and annotate your classes and functions with :
__declspec(dllexport) while compiling the library
__declspec(dllimport) while linking
If you want multiplatform support for the GCC toolchain you can add a header in your project doing that for you. For a step by step example and lots of details have a look at GCC's visibility guide.
Windows PE object files do not have visibility attributes. The closest is dllexport/dllimport, but that's only for shared libraries (DLL's). So either you don't mark all FFTW functions with __declspec(dllexport), and hope linking the static library does The Right Thing (tm), or you take care not to link to FFTW if linking with your library.
It should warn about bad visibility attributes, perhaps you need to turn up the warning level -Wall -Wextra -pedantic).
I'm building a library that needs to be dynamically linked to my project. The output is a .so file, so I think I'm on the right track. I'm concerned by the way it's being linked at compile time - by specifying the location of its makefile and depending on a bunch of macros, which I've never encountered before.
Can I assume that since I'm building a .so library (rather than a .a) that I'm in fact dynamically linking? Or is it possible for .so libs to be statically linked, in which case I need to rip apart the make/config files to better understand what's going on?
Thanks,
Andrew
I'm not familiar with internal structure of executables and shared objects, so I could only give some practical hints.
Assuming you use gcc, it should have -shared option when linking object files into library - this way ld (called by gcc) makes shared object instead of executable binary.
gcc -shared -o libabc.so *.o ...
When you link some application with this libabc.so it should link without errors and after that with ldd command you should be able to see libabc.so among its dependencies.
$ ldd app
...
libabc.so => ...............
I want to use CUDPP library in my project. I've downloaded sources from project page. Unfortunatly, when I ran "make", there was only static library build. I've looked into Makefile files and haven't found any dynamic lib configuration. I don't want to keep static library with the project - it's totally non-portable way.
My question is: how can I build .so dynamic library of CUDPP, without writing my own Makefile/compiling it manually? Maybe someone already did it?
EDIT: I've replaced "g++" with "g++ -fPIC", "gcc" with "gcc -fPIC" and "nvcc" with "nvcc -Xcompiler -fpic". When I unpack obj files from archive, and link them to shared lib, I've got no error. However, my application crashes at start, when linked with this library.
when you compile pass the flag -Xcompiler -fpic to nvcc. If you link against any cuda libraries make sure you've linked to the shared libs, otherwise you can't link it. Hopefully that's all you need.
Are you also using -shared to create the library? You shouldn't need to extract anything from an archive if it is working correctly.
If you run ldd on your executable it will show you what dynamic linking is required by the app and you can check that the -fPIC etc. worked correctly. Also make sure that the .so library is on your LD_LIBRARY_PATH (sorry if that's obvious, no harm in checking).