I have an app that depends on a dynamic library that is not in a system location. If the library is located in the location from which the executable was linked and LD_LIBRARY_PATH is set to that directory, the application runs.
If the libraries are copied to another directory and LD_LIBRARY_PATH is reset, the application won't start and an undefined symbol error occurs, despite the fact that the symbol appears to be in the library.
Any ideas why this may happen?
Thanks,
Try ldd to show what path used:
ldd youprogram
Related
I have built some shared libraries on Ubuntu Linux 16.0.2 from source.
They are 64-bit libs.
I manually copied them to /usr/local/lib.
I verified that the /usr/local/lib path is indeed in one of the .conf files that ld.so.conf includes.
I then ran: sudo ldconfig to update the cache.
But then when I try to run my executable which tries to dynamically load one of the .so files that I previously copied into /usr/local/lib using dlopen, it fails.
In my code, I have:
dlopen ("foobar.so", RTLD_LAZY);
Can anybody tell me what I am doing wrong?
The dynamic linker normally doesn't access the paths recursively included from /etc/ld.so.conf directly, but it uses a cache.
You can update the cache with
sudo ldconfig
See ldconfig(8) for more details.
for dlopen to work, there's no directory list of places to find the shared object. So doing dlopen("somefile", ...); probably won't work.
You don't need to use any path or to put the shared object (or to comply with naming conventions) to use a shared object via dlopen(3). That's only a requirement of the dynamic linker that loads and links all the shared libraries at launch time: linux-vdso.so.1 (in 64bit)
To test, just put the shared in your local dir and try to open it with it's basename, like you post.
For a system library, there are more requirements, like defining a soname for the library, which is used by the loader to load the library and to construct the cache database index, so if you have no idea of what I'm talking about, you will not be able to use automatic loading procedure. If you want to see if an executable file has all the libraries it needs and where the loader finds them out, just run ldd(1) with the executable as argument, and you'll se the dependencies for automatic loading and how the dynamic linker resolves the paths.
I am trying to run an executable file - SaTScanBatch, executable of the SaTScan software - on a remote machine from the terminal.
Here is the error I get
/usr/lib64/libstdc++.so.6: version 'GLIBCXX_3.4.15' not found
The problem is similar to these:
/usr/lib/libstdc++.so.6: version `GLIBCXX_3.4.15' not found or How to fix: /usr/lib/libstdc++.so.6: version `GLIBCXX_3.4.15' not found or /usr/lib64/libstdc++.so.6: version `GLIBCXX_3.4.15' not found
What is different in my case is that I do not have admin rights, so I cannot add libraries in usr/lib file. I downloaded libstdc++.6.0.15 in my remote repository, and I want to use it to execute my file.
Here are the possibilities I tried
i) Modify environment variable LD_LIBRARY_PATH or LD_RUN_PATH or LD_PRELOAD to the path of libstdc++.so.6.0.15.
--> Did not change anything
ii) Include the library in a static way
gcc SaTScanBatch -static-libstdc++
--> -static option is not recognized, I guess the remote machine's GCC version is too old.
iii) Try to link the file to the additional library:
gcc SaTScanBatch -L /path/library -l stdc++
or similarly
gcc SaTScanBatch -Wl,-rpath,path/to/library
--> Error
/usr/bin/ld: warning: Cannot create .eh_frame_hdr section, --eh-frame-hdr ignored.
/usr/bin/ld: error in SaTScanBatch64(.eh_frame); no .eh_frame_hdr table will be created.
/usr/lib/gcc/x86_64-redhat-linux/4.4.6/../../../../lib64/crt1.o: In function _start:
(.text+0x20): undefined reference to "main"
collect2: ld returned 1 exit status
iv) Use PatchElf to link the file and the library --> I cannot install PatchElf because no admin rights
Thank you for any suggestion !
Modify environment variable LD_LIBRARY_PATH or LD_RUN_PATH or LD_PRELOAD to the path of libstdc++.so.6.0.15. --> Did not change anything
What exactly did you try?
LD_LIBRARY_PATH should be set to the directory containing the new libstdc++.so.6 not the the file itself, and you need to export the environment variable so it's available to child processes, not just to your shell. And you need a symlink from libstdc++.so.6 to libstdc++.so.6.0.15 because the dynamic loader is going to look for that name, not libstdc++.so.6.0.15
LD_RUN_PATH is used during linking to bake a path into the executable. It does nothing at run-time when trying to run the executable. If you use it you need to set it to the directory that will contain the libstdc++.so.6 file on the remote machine. Again, it needs to be set to a directory, not the path of the file.
LD_PRELOAD is something different again, and almost certainly not what you want. You would use it to force a particular shared library to be loaded before anything else when the executable runs. That can be used to pre-load a newer libstdc++ but it's usually better to use LD_LIBRARY_PATH to set a path at run-time or LD_RUN_PATH to set a path at link-time.
Try to link the file to the additional library:
The commands you showed are nonsense, they don't include any objects in the link, so you're trying to create an executable out of nothing. That's why you get the error undefined reference to "main"
For your scenario I would recommend using LD_LIBRARY_PATH.
My project includes a library and example projects for how to use it. I place the library in the "bin" folder along with all executable examples. I can run the example projects on the machine where they were compiled but when I try to run them on another machine I get:
./example: error while loading shared libraries: libMyLib.so: cannot
open shared object file: No such file or directory
This makes no sense since the library is in the same folder. What is causes it to ignore the library on other machines?
Just because the library is in the same directory as the executable doesn't mean it will look there for it. By default on linux, executables will only look in a limited set of directories, set by ldconfig and the LD_LIBRARY_PATH environment variable.
One trick that is very useful is to link your program with the extra linker option
-Wl,-rpath,'$ORIGIN'
which will cause the executable to also look in the directory the executable is in for shared objects.
You can usually set this by adding to your Makefile:
LDFLAGS := -Wl,-rpath,'$$ORIGIN'
Note the double-$ here -- make will interpret this as a make variable which expands to just $
The current directory is not necessarily a place where the dynamic linker will look for dynamic libraries. The directory where the executable is much less.
You might want to check ldconfig to see where it looks for them.
I'm using eclipse cdt to compile and run C++ application.
My_main_program needs specifically libjpeg.so.62.
My Ubuntu system previously have libjpeg.so.9 at /usr/local/lib/. I happened to compiled and run using libjpeg.so.9 before run-time compatibility errors was raised.
Then I deleted all libjpeg.* and installed libjpeg.la, libjpeg.so, libjpeg.so.62 and libjpeg.so.62.0.0 from source. Then I run ldconfig.
I can build the project. The problem is the dynamic linker keeps searching for libjpeg.so.9 and throwing
'error while loading shared libraries: libjpeg.so.9: cannot open shared object file: No such file or directory'
at run-time.
This problem is killing me.
I have checked that the symlink of libjpeg.so is correct.
Please help!
I can build the project. The problem is the dynamic linker keeps searching for libjpeg.so.9 and throwing
'error while loading shared libraries: libjpeg.so.9: ... No such file ...
You need to understand a couple of things:
A shared library may have SONAME dynamic tag (visible with readelf -d foo.so | grep SONAME).
If an executable is linked against such a library, the SONAME is recorded as a NEEDED dynamic tag (in the executable), regardless of what the library itself is called. That is, you can name the library foo.so, foo.so.1234, or anything else. IF the library has SONAME of libbar.so.7, then the executable will require libbar.so.7, no matter what [1].
On to your problem. Your executable fails to load libjpeg.so.9, therefor we conclude that it is being linked (at build time) with a shared library which has SONAME: libjpeg.so.9.
I deleted all libjpeg.* and installed libjpeg.so.62
You must not have deleted the libjpeg.so that is used at executable build time (which is somewhere other than /usr/local/lib). That library still has SONAME: libjpeg.so.9, and is causing you grief.
You can find out which libraries are being used at link time by passing -Wl,-t flag on the link line.
[1] Not strictly true: if the executable doesn't need any symbols from foo.so, and if --as-needed linker option is in effect, then NEEDED: libbar.so.7 will not be recorded after all.
Update:
I have also check ldd executable and it returns libjpeg.so.62
This means that the executable that you run ldd on is correct, but the executable that actually run is not, and they must be different executables.
Update 2:
You're right. ldd executable shows both libjpeg.so.62 and libjpeg.so.9 are included
Actually, no, I wasn't. But I will be right this time.
What's happening is that your executable correctly records NEEDED: libjpeg.so.62 (you can verify this with the following command: readelf -d /path/to/exe | grep 'NEEDED.*libjpeg').
But you also have some other shared library (one of the ones listed in ldd output), which has not been rebuilt, and still has a dependency on libjpeg.so.9.
You can find that library by running readelf -d /path/to/libXXX.so | grep 'NEEDED.*libjpeg\.so\.9' on all libraries listed in ldd output.
Once you find it, you'll have to rebuild it so it also depends on libjpeg.so.62.
I have a core file generated on a remote system that I don't have direct access to. I also have local copies of the library files from the remote system, and the executable file for the crashing program.
I'd like to analyse this core dump in gdb.
For example:
gdb path/to/executable path/to/corefile
My libraries are in the current directory.
In the past I've seen debuggers implement this by supplying the option "-p ." or "-p /=."; so my question is:
How can I specify that libraries be loaded first from paths relative to my current directory when analysing a corefile in gdb?
Start gdb without specifying the executable or core file, then type the following commands:
set solib-absolute-prefix ./usr
file path/to/executable
core-file path/to/corefile
You will need to make sure to mirror your library path exactly from the target system. The above is meant for debugging targets that don't match your host, that is why it's important to replicate your root filesystem structure containing your libraries.
If you are remote debugging a server that is the same architecture and Linux/glibc version as your host, then you can do as fd suggested:
set solib-search-path <path>
If you are trying to override some of the libraries, but not all then you can copy the target library directory structure into a temporary place and use the solib-absolute-prefix solution described above.
I'm not sure this is possible at all within gdb but then I'm no expert.
However I can comment on the Linux dynamic linker. The following should print the path of all resolved shared libraries and the unresolved ones.
ldd path/to/executable
We need to know how your shared libraries were linked with your executable. To do this, use the following command:
readelf -d path/to/executable | grep RPATH
Should the command print nothing, the dynamic linker will use standard locations plus the LD_LIBRARY_PATH environment variable to find the shared libraries.
If the command prints some lines, the dynamic linker will ignore LD_LIBRARY_PATH and use the hardcoded rpaths instead.
If the listed rpaths are absolute, the only solution I know is to copy (or symlink) your libraries to the listed locations.
If the listed rpaths are relative, they will contain a $ORIGIN which will be replaced at run time by the path of the executable. Move either the executable or the libraries to match.
For further informations, you could start with:
man ld.so
I found this excerpt on developer.apple.com
set solib-search-path path
If this variable is set, path is a
colon-separated list of directories to
search for shared libraries.
solib-search-path' is used after
solib-absolute-prefix' fails to
locate the library, or if the path to
the library is relative instead of
absolute. If you want to use
solib-search-path' instead of
solib-absolute-prefix', be sure to
set `solib-absolute-prefix' to a
nonexistant directory to prevent GDB
from finding your host's libraries.
EDIT:
I don't think using the above setting prepends the directories I added, but it does seem to append them, so files missing from my current system are picked up in the paths I added. I guess setting the solib-absolute-prefix to something bogus and adding directories in the solib-search-path in the order I need might be a full solution.
You can also just set LD_PRELOAD to each of the libraries or LD_LIBRARY_PATH to the current directory when invoking gdb. This will only cause problems if gdb itself tries to use any of the libraries you're preloading.
One important note:
if you're doing a cross compiling and trying to debug with gdb, then
after you've done file ECECUTABLE_NAME if you see smth. like :
Using host libthread_db library "/lib/libthread_db.so.1"
then check whether you have libthread_db for your target system. I found a lot of similar problems on the web. Such problem cannot be solved just using "set solib-", you has to build libthread_db using your cross-compiler as well.