I'm using GCC 4.7.2 and LD 2.23 but when I add -flto to my compile options my compile time increases by over 20%! The manual seems to indicate that -fuse-linker-plugin is needed for the optimization to work. It also says that it's enabled by default with -flto but when I add it explicitly I see the following error in the link command:
g++: error: -fuse-linker-plugin is not supported in this configuration
According to manual, it should be supported by LD 2.21 or greater. Any idea why I'm getting this error? For reference here are examples of my full compile commands:
g++ -Wall -pipe -O3 -flto -fno-strict-aliasing -mtune=generic --no-exceptions -fPIC -c some.cc
g++ -o exec -Xlinker some1.o some2.o -static some1.a some2.a -Wl,--wrap,open -flto -fuse-linker-plugin
Running 'ld --help | grep plugin' shows "-plugin" option so I don't understand why GCC is complaining:
-plugin PLUGIN Load named plugin
-plugin-opt ARG Send arg to last-loaded plugin
Link time optimizations aren't supposed to reduce compilation time, but optimize runtime of your program.
#options, just add "-flto -fuse-linker-plugin" to your CFLAGS(or CXXFLAGS for c++) and LDFLAGS and it should work just fine.
#gold: ld --version is probably gonna return gnu LD, to switch to gold, make ld symlink which ld point to which ld.gold
e.g. ln -s /usr/bin/ld.gold /usr/bin/ld
Related
I am trying to cross compile my application for a arm based system.
I have 2 libraries compiled in the following way:
$ gcc -shared --sysroot=$DIR_PATH -o $LIBPATH/libfoo.so foo.o
$ gcc -shared --sysroot=$DIR_PATH -o $LIBPATH/libbar.so bar.o
A third library is compiled:
gcc -shared -o $LIBPATH/libfoobar.so --sysroot=$DIR_PATH -L$LIBPATH -Wl,rpath=$RUN_TIME_PATH foobar.o -lfoo -lbar
Then finally I compile a binary:
gcc -o app --sysroot=$DIR_PATH -L$LIBPATH -Wl,rpath=$RUN_TIME_PATH app.o -lfoobar
However when compiling app I get
warning: libfoo.so, needed by libfoobar.so, not found (try using -rpath or -rpath-link)
I believe you need to use -Wl,-rpath-link=$LIBPATH to tell the linker where to look to resolve runtime library references during the link operation.
More info can be found in the ld documentation: https://sourceware.org/binutils/docs-2.37/ld/Options.html
I noticed that in the default release configuration, qmake (qmake 3.1, qt 5.9.5 - whatever is installed on my Ubuntu build box) passes -Wl,O1 to g++ when linking. So the linking command line looks like
g++ -Wl,-O1 -flto -O2 -o program program.o lib1.a lib2.a ...
where -flto -O2 are the options that I'm passing via QMAKE_LFLAGS_RELEASE to enable LTO.
Now the question: why qmake has this -Wl,-O1 option and is it going to interfere with LTO?
QMake passes -Wl,O1, because it is meant to be a good default.
It will not harm LTO, because your -O2 option comes later and overrides the -Wl,O1.
From g++'s man page:
If you use multiple -O options, with or without level numbers, the
last such option is the one that is effective.
You can remove the -Wl,-O1 from your Makefile by specifying
QMAKE_LFLAGS_RELEASE -= -Wl,-O1
Preamble: this question is not about Oracle, instead I'd like to understand the fundamental difference between gcc-4 and gcc-6 in the handling of Position Independent Code.
So I have decided to try an Oracle 12c installation on a Debian stretch.
During the link stage with gcc-6, error messages like the following are issued:
/usr/bin/ld: /opt/oracle/product/12.2.0/lib/libpls12.a(pci.o):
relocation R_X86_64_32S against `.rodata.str1.4' can not be used when making a shared object;
recompile with -fPIC.
However, if I switch the compiler to use gcc-4.9, all the linking is done without any problems.
Thus my 2 questions:
Is there a change in the defaults for -fPIC and -fPIE between gcc version 4 and 6? Most probably yes, version 6 seems to use the 2 options by default.
More important for me: does gcc, version 6 have an option to use the version 4 behavior for the generation of position independent code? (Or will I sooner or later no more be able to link against old libraries because gcc-4 is no more available?)
Most likely the gcc-6 linker creates position independent executables by default. The problem can be reproduced as follows and solved by adding the linker flag -no-pie:
UNIX # gcc-6 -g -Wall -fno-pic -c helloworld.c -o helloworld.o
UNIX # gcc-6 -g -Wall helloworld.o -o helloworld
/usr/bin/ld: helloworld.o: relocation R_X86_64_32 against `.rodata' can not be used when making a shared object; recompile with -fPIC
/usr/bin/ld: final link failed: Nonrepresentable section on output
collect2: error: ld returned 1 exit status
UNIX # gcc-6 -g -Wall -no-pie helloworld.o -o helloworld
Indeed, after adding -no-pie to the gcc options used by Oracle, linking works without any errors.
Solution from broeni works fine. Some additional steps I made to make it work:
During installation, I modified the default linker tool from oracle, editing the file
/opt/oracle/product/12.2.0/db1/bin/orald
In the first lines, I forced to use GCC linker, and add the -no-pie option:
#if [ -z "$BASH_VERSION" -o -n "$ORALD_USE_GCC" ] ; then
exec gcc -no-pie "$#"
exit 1
#fi
tags: oracle 12c debian stretch
I am new to C++ compiling/linking.
I am trying to link all libraries statically with gcc, I tried using LDFLAGS=-static but did not work. Error message showed:
/bin/sh ./libtool --tag=CXX --mode=link /home/dizhang/lib/hdf5/bin/h5c++ -g -O2 -L/home/dizhang/lib/blitz/lib -L/home/dizhang/lib/libconfig/lib -o angora src/libangora.la -lblitz -lconfig++
libtool: link: /home/dizhang/lib/hdf5/bin/h5c++ -g -O2 -o angora -L/home/dizhang/lib/blitz/lib -L/home/dizhang/lib/libconfig/lib src/.libs/libangora.a -L/bgsys/drivers/V1R2M2/ppc64/comm/lib64 -L/bgsys/drivers/V1R2M2/ppc64/comm/lib -L/bgsys/drivers/V1R2M2/ppc64/spi/lib -lpthread -lm /home/dizhang/lib/blitz/lib/libblitz.a /bgsys/drivers/V1R2M2/ppc64/comm/lib/libmpichcxx-gcc.so /bgsys/drivers/toolchain/V1R2M2_base/gnu-linux/powerpc64-bgq-linux/lib/libstdc++.so /bgsys/drivers/V1R2M2/ppc64/comm/lib/libmpich-gcc.so /bgsys/drivers/V1R2M2/ppc64/comm/lib/libopa-gcc.so /bgsys/drivers/V1R2M2/ppc64/comm/lib/libmpl-gcc.so /home/dizhang/lib/libconfig/lib/libconfig++.a /bgsys/drivers/toolchain/V1R2M2_base-efix014/gnu-linux/powerpc64-bgq-linux/lib/libstdc++.so -Wl,-rpath -Wl,/bgsys/drivers/toolchain/V1R2M2_base-efix014/gnu-linux/powerpc64-bgq-linux/lib -Wl,-rpath -Wl,/bgsys/drivers/V1R2M2/ppc64/comm/lib -Wl,-rpath -Wl,/bgsys/drivers/toolchain/V1R2M2_base/gnu-linux/powerpc64-bgq-linux/lib -Wl,-rpath -Wl,/bgsys/drivers/toolchain/V1R2M2_base-efix014/gnu-linux/powerpc64-bgq-linux/lib -Wl,-rpath -Wl,/bgsys/drivers/V1R2M2/ppc64/comm/lib -Wl,-rpath -Wl,/bgsys/drivers/toolchain/V1R2M2_base/gnu-linux/powerpc64-bgq-linux/lib
/bgsys/drivers/toolchain/V1R2M2_base-efix014/gnu-linux/lib/gcc/powerpc64-bgq-linux/4.4.7/../../../../powerpc64-bgq-linux/bin/ld: attempted static link of dynamic object `/bgsys/drivers/V1R2M2/ppc64/comm/lib/libmpichcxx-gcc.so'
collect2: ld returned 1 exit status
I did some search and found that, telling Makefile -Wl -Bstatic may solve this problem, but how exactly I should change this in my Makefile?
I tried searching -Wl in Makefile but it was not in the text.
Thanks,
Di
Looks like you are trying to build HDF5 with MPI support on BGQ.
As long as you end up passing a ".so" version of a lib to gcc on BGQ -- you will see this error. You may need to check how you are passing MPI info to HDF5's configure script.
In my case (building another project that uses CMake) passing BGQ's MPI compiler wrappers to CMake always created a problem where it would try to link MPI using the shared libs instead of the static ones. To resolve this, I had to make sure to explicitly specify the ".a" variants of the MPI libs.
Running the build script from ELLCC results in this error
gcc -DHAVE_CONFIG_H -I. -I../../../src/binutils/binutils -I. -I../../../src/binutils/binutils -I../bfd -I../../../src/binutils/binutils/../bfd -I../../../src/binutils/binutils/../include -I./../intl -DLOCALEDIR="\"/Library/Caches/Homebrew/ellcc--svn-HEAD/lib/share/locale\"" -Dbin_dummy_emulation=bin_vanilla_emulation -W -Wall -Wstrict-prototypes -Wmissing-prototypes -Wshadow -Werror -Wno-unused-value -Wno-shadow -MT nm.o -MD -MP -MF .deps/nm.Tpo -c -o nm.o ../../../src/binutils/binutils/nm.c
../../../src/binutils/binutils/nm.c:1690:28: error: 'sbrk' is deprecated
[-Werror,-Wdeprecated-declarations]
char *lim = (char *) sbrk (0);
^
/usr/include/unistd.h:582:7: note: 'sbrk' declared here
void *sbrk(int);
^
The following compilers have been used with the same result:
gcc 4.8
llvm-gcc 2.8
llvm 3.3
I had the same issue compiling binutils-2.24 on Mac OSX Mavericks 13.2.0 with clang. Thanks to Richard Pennington's suggestion, I was able to get binutils to compile by specifying a few other -Wno-error arguments to gcc by setting CFLAGS before running configure. Namely, these are the commands I ran to build and install binutils:
CFLAGS="-Wno-error=deprecated-declarations -Wno-error=unused-variable -Wno-error=unused-function" ./configure --prefix=/usr/local/toolchain-arm-linux-elf --target=arm-linux-elf
make
make install
EDIT: I just noticed that the binutils configure script accepts an --disable-werror argument, which disables gcc turning warnings into errors, and removes the need for the settings CFLAGS. With this argument, building could be done as follows:
./configure --prefix=/usr/local/toolchain-arm-linux-elf --target=arm-linux-elf --disable-werror
make
make install
This error is occurring because sbrk() is deprecated on OSX, -Werror is enabled for the binutils build, and the compiler (in this case "gcc" is an alias for clang) rightly complains about the use of sbrk(). I'll be looking into eliminating this error this weekend when I won't have to be at my day job. ;-)
I looked into it a bit more. This happened because the latest version of OS X (Mavericks) uses clang as its compiler and /usr/include/unistd.h has a deprecated declaration of sbrk().
The solution was to add a -Wno-error=deprecated-declarations option to the CFLAGS for binutils. I also had to make a few other changes to complete the Max OS build. You can find the latest stuff in the ELLCC subversion tree.