I have a problem with linking shared (dynamic) libraries .so or .dylib from my one project with another project.
I am using such Makefile:
#libraries
#custom
COMPARERS_STATIC_LIB_PATH=../comparers/output/debug/lib/libcomparers.a
COMPARERS_LIB_DIR=../comparers/output/debug/lib/
$(SHARED_LIBRARY): assertion.o
$(CC) $(CFLAGS) -L$(COMPARERS_LIB_DIR) -shared -o $(OUTPUTS_LIB_DIR)/$(SHARED_LIBRARY) $(OUTPUTS_DIR)/assertion.o -lcomparers
$(DYNAMIC_LIBRARY): assertion.o
$(CC) $(CFLAGS) -L$(COMPARERS_LIB_DIR) -dynamiclib -o $(OUTPUTS_LIB_DIR)/$(DYNAMIC_LIBRARY) $(OUTPUTS_DIR)/assertion.o -lcomparers
The directories and libraries are in place as on the below picture.
Linking with static library compiles and links correctly:
$(TARGET): $(LIBRARY)
$(CC) $(CFLAGS) -o $(OUTPUTS_BIN_DIR)/$(TARGET) src/main.c $(OUTPUTS_LIB_DIR)/$(LIBRARY) $(COMPARERS_STATIC_LIB_PATH)
The error I get:
gcc -g -Wall -D__USE_FIXED_PROTOTYPES__ -std=c99 -I./include -I../comparers/include -L../comparers/output/debug/lib/ -shared -o outputs/debug/lib/libunit_tests.so outputs/debug/assertion.o -lcomparers
ld: warning: directory not found for option '-L../comparers/output/debug/lib/'
ld: library not found for -lcomparers
clang: error: linker command failed with exit code 1 (use -v to see invocation)
make: *** [libunit_tests.so] Error 1
You have a typo, your $(COMPARERS_LIB_DIR) contains:
../comparers/output/debug/lib/
but must be:
../comparers/outputs/debug/lib/
^
concerning to your figure.
Related
I am working on a c project. https://github.com/eantoranz/gitmod It's broken up into separate c/h files. In my current Makefile, I take each one of the c files and compile them into their own .o file, then when I want to create the final program, I ask to compile a c file that has a main and provide all the other .o files. So far, so good (let me know if this approach is not correct).
As a next step, I want to separate the main part of the program from all the other pieces of the code and would like to pack all of the previous o files into a single so file (that had been compiled as just another .o file so far).
So, old approach was like this:
gitmod.o: src/gitmod.c include/gitmod.h lock.o root_tree.o thread.o object.o cache.o
$(CC) -c -o $# $< $(CFLAGS)
gitmod: src/main.c gitmod.o
$(CC) $< *.o -o $# $(CFLAGS)
These would be the last 2 steps of this process using the so replacing gitmod.o for libgitmod.so, after creating all the previous .o files:
libgitmod.so: src/gitmod.c include/gitmod.h lock.o root_tree.o thread.o object.o cache.o
$(CC) $< *.o -shared -o $# $(CFLAGS)
gitmod: src/main.c libgitmod.so
$(CC) $< -llibgitmod.so -o $# $(CFLAGS)
libgitmod.so would be the shared library that I would like to actually contain all the other sections of code (I really hope that is possible).
When make is going through libgitmod.so, I am getting this:
gcc src/gitmod.c *.o -shared -o libgitmod.so -Iinclude `pkg-config fuse3 libgit2 glib-2.0 --cflags --libs` -DGITMOD_DEBUG -Wall -g
/usr/bin/ld: object.o: warning: relocation against `stderr##GLIBC_2.2.5' in read-only section `.text'
/usr/bin/ld: /tmp/ccir4Ou9.o: relocation R_X86_64_PC32 against symbol `stderr##GLIBC_2.2.5' can not be used when making a shared object; recompile with -fPIC
/usr/bin/ld: final link failed: bad value
collect2: error: ld returned 1 exit status
make: *** [Makefile:32: libgitmod.so] Error 1
What am I doing wrong?
-fPIC, which means Position Indepent Code, flags is required to create shared object.
You can add the flag in the compile flag, then you can create a shared object.
I searched all other post saying "clang: error: linker command failed with exit code 1 (use -v to see invocation)" but i cannot resolve my problem.
After I make I got this message:
ld: symbol(s) not found for architecture i386
clang: error: linker command failed with exit code 1 (use -v to see invocation)
make: *** [btest] Error 1
My Makefile is:
CC = gcc
CFLAGS = -O -Wall -m32
LIBS = -lm
all: btest fshow ishow
btest: btest.c bits.c decl.c tests.c btest.h bits.h
$(CC) $(CFLAGS) $(LIBS) -o btest bits.c btest.c decl.c tests.c
fshow: fshow.c
$(CC) $(CFLAGS) -o fshow fshow.c
ishow: ishow.c
$(CC) $(CFLAGS) -o ishow ishow.c
# Forces a recompile. Used by the driver program.
btestexplicit:
$(CC) $(CFLAGS) $(LIBS) -o btest bits.c btest.c decl.c tests.c
clean:
rm -f *.o btest fshow ishow *~
This is from CS:APP data lab
I have already checked xcode Architecture setting is Standard Architecture(64 bit intel)
I've written an entire program and its makefile, the only trouble is I have to idea how to implement the debugger. I've been looking at similar questions online but to no avail.
Here's my makefile:
# the variable CC is the compiler to use.
CC=gcc
# the variable CFLAGS is compiler options.
CFLAGS=-c -Wall
assign4: main1.o ObjectManager.o
$(CC) main1.o ObjectManager.o -o assign4
main1.o: main1.c ObjectManager.h
$(CC) $(CFLAGS) main1.c
ObjectManager.o: ObjectManager.c ObjectManager.h
$(CC) $(CFLAGS) ObjectManager.c
clean:
rm -rf *o assign4
I've tried to adjust the CFLAGS: to -g Wall, but all it says is:
Undefined symbols for architecture x86_64:
"_main", referenced from:
implicit entry/start for main executable
ld: symbol(s) not found for architecture x86_64
clang: error: linker command failed with exit code 1 (use -v to see invocation)
make: *** [ObjectManager.o] Error 1
Any suggestions to fix this?
Look at this rule:
ObjectManager.o: ...
$(CC) $(CFLAGS) ObjectManager.c
If CFLAGS is -c -Wall, that -c means that gcc will compile the source file, but not attempt to link it. This is exactly what was intended, to produce an object file and not an executable. One consequence of this is that the source dile (ObjectManager.c) need not contain a main function-- the compiler trusts you to provide one at link time.
But if CFLAGS is -g -Wall, the -g does not imply -c. So gcc will attempt to create an executable, and abort if main is not present.
Solution:
CFLAGS := -Wall -g # the -g can be added with a conditional if you like
ObjectManager.o: ...
$(CC) $(CFLAGS) -c ObjectManager.c
I'm trying to compile a few .c files that used msgpack-c functions into a shared library. I have the following Makefile:
MSGPACK_CS = msgpack.c
CFLAGS = -std=c99
MSGPACK_OBJECTS = $(subst .c,.o,$(MSGPACK_CS))
MSGPACK_LIBS = msgpack.so
all: $(MSGPACK_OBJECTS) $(MSGPACK_LIBS)
%.o: %.c
$(CC) -c -shared -fPIC $(CFLAGS) $<
$(MSGPACK_LIBS): $(MSGPACK_OBJECTS)
ld -Lmsgpack/.libs -share -o $# $(MSGPACK_OBJECTS) -lmsgpack
I can compile a program that uses msgpack without problem, but this gives me the following error:
msgpack.o: In function `msgpack_pack_int64':
/usr/local/include/msgpack/pack_template.h:373: undefined reference to `__stack_chk_fail_local'
ld: msgpack.so: hidden symbol `__stack_chk_fail_local' isn't defined
ld: final link failed: Bad value
Apparently the linkage process isn't going well, but I don't know what is wrong. What can I do?
Try linking with the gcc driver instead of calling ld directly. ld doesn't know about the gcc support libs that are needed for the C runtime:
gcc -Lmsgpack/.libs -shared -o $# $(MSGPACK_OBJECTS) -lmsgpack
If this still doesnt't work, you might need to add -fno-stack-protector to your CFLAGS to supress runtime stack checking.
When I try and #include "json/json.h" in a .cu file, then run make, I get the following error:
nvcc -o sound main.o process.o -L /usr/lib -lopencv_core -lopencv_imgproc -lopencv_highgui -O3 -arch=sm_20 -Xcompiler -Wall -Xcompiler -Wextra -m64
/usr/local/cuda/bin/crt/link.stub:90:13: warning: ‘void __cudaRegisterLinkedBinary(const __fatBinC_Wrapper_t*, void (*)(void**), void*)’ defined but not used [-Wunused-function]
process.o: In function `count_tracks()':
tmpxft_00006061_00000000-3_process.cudafe1.cpp:(.text+0x75): undefined reference to `Json::Value::Value(Json::ValueType)'
tmpxft_00006061_00000000-3_process.cudafe1.cpp:(.text+0x7d): undefined reference to `Json::Value::~Value()'
collect2: ld returned 1 exit status
make: *** [student] Error 1
in reference to when I try and create a Json::Value. I've tried moving around where I link JsonCpp library, and I wasn't having this problem #including jsoncpp and creating a Json::Value in the main.cpp of the project. It just doesn't seem to be working correctly when in a .cu file.
Here is the pertinent stuff from my Makefile, which I got from Udacity's CUDA course and modified to fit my needs:
NVCC=nvcc
CXX = g++
LDFLAGS = -L ~/parallelcomputing/soundcloud/jsoncpp/build/debug/lib -ljsoncpp
INC = -I ~/parallelcomputing/soundcloud/jsoncpp/include
OPENCV_LIBPATH=/usr/lib
OPENCV_INCLUDEPATH=/usr/include
OPENCV_LIBS=-lopencv_core -lopencv_imgproc -lopencv_highgui
CUDA_INCLUDEPATH=/usr/local/cuda/include
NVCC_OPTS=-O3 -arch=sm_20 -Xcompiler -Wall -Xcompiler -Wextra -m64
GCC_OPTS=-O3 -Wall -Wextra -m64
student: main.o process.o Makefile
$(NVCC) -o sound main.o process.o -L $(OPENCV_LIBPATH) $(OPENCV_LIBS) $(NVCC_OPTS)
main.o: main.cpp
g++ -c main.cpp $(GCC_OPTS) $(LDFLAGS) $(INC) -I $(CUDA_INCLUDEPATH) -I $(OPENCV_INCLUDEPATH)
process.o: process.cu
nvcc -c process.cu $(NVCC_OPTS) $(LDFLAGS) $(INC)
clean:
rm -f *.o *.png hw
You've got LDFLAGS defined in your makefile, but you're not using it in the link phase that I can see.
As a result, -ljsoncpp doesn't show up in the link command you posted, that is showing the error.
Add LDFLAGS to your link phase:
student: main.o process.o Makefile
$(NVCC) -o sound main.o process.o $(LDFLAGS) -L $(OPENCV_LIBPATH) $(OPENCV_LIBS) $(NVCC_OPTS)
(And while we're cleaning up your makefile, LDFLAGS contains link specification, and is not relevant in, and can be safely deleted from, the subsequent compile targets.)
EDIT:
Since that is not working, but you say the link is successful with a .cpp file, try linking the executable with g++ instead of nvcc:
LDFLAGS2=-L/usr/local/cuda/lib64 -lcudart
student: main.o process.o Makefile
$(CXX) -o sound main.o process.o $(LDFLAGS) $(LDFLAGS2) -L $(OPENCV_LIBPATH) $(OPENCV_LIBS)