I've got a compiled program mybinary that fails immediately with "image not found" when I try to run it. It isn't the path to the missing dynamic library that's the problem, it's that it needs to look at the library name that doesn't include the version number.
At execution of mybinary, I get this (edited for brevity):
dyld: Library not loaded: /usr/local/opt/icu4c/lib/libicuuc.68.dylib
Reason: image not found
Running otool -L mybinary shows this (also edited for brevity):
/usr/local/opt/icu4c/lib/libicuuc.68.dylib
I want it to look at libicuuc.dylib without the ".68". An ls -1 shows it exists:
/usr/local/opt/icu4c/lib/libicuuc.69.1.dylib
/usr/local/opt/icu4c/lib/libicuuc.69.dylib
/usr/local/opt/icu4c/lib/libicuuc.a
/usr/local/opt/icu4c/lib/libicuuc.dylib
Because the version referenced in mybinary is 68 (lower than 69, thus preventing Incompatible library version), I can fix mybinary by executing the following:
sudo install_name_tool -change /usr/local/opt/icu4c/lib/libicuuc.68.dylib /usr/local/opt/icu4c/lib/libicuuc.dylib mybinary
Now, mybinary launches just fine.
The goal, though, is to modify the Makefile for mybinary so that it will link to libicuuc.dylib.
How can I do this?
mybinary is deployed in a tar file to end users, and they are more likely to have a more current version of icu4c (in this example, 69) compared to mybinary in the tar file (in this example, 68).
Is there someway to tell the linker at compile time that I want to use libicuuc.dylib (no version# in the filename)? It would be fine if this behavior then applied to other linked libraries, i.e., I don't need to have it specific for libicuuc.dylib.
If that's not possible, I could live with modifying the Makefile to run install_name_tool but then my dilemma is that there doesn't seem to be an easy way to get libicuuc.68.dylib. It's not part of the compile console output, and parsing the output of otool -L mybinary is going to be messy. Is there any easy and elegant way to get libicuuc.68.dylib for passing to install_name_tool?
Related
I mean to compile example Advancing_front_surface_reconstruction from the CGAL package under Win10 + PortableApps Msys2.
For the library itself I use a precompiled package ($ pacman -Qs cgal gives local/mingw-w64-x86_64-cgal 4.14-1).
I used to be able to do that in my previous PC, see details below.
After transferring everything to a new PC (simply copying, it is PortableApps), I made a system update in Msys2 (required for other reasons).
I proceed as usual: mkdir build, cd build, ccmake ... Selecting Release mode, and verbosity ON, upon pressing g to generate files I got
CMake Error in CMakeLists.txt:
Imported target "CGAL::CGAL" includes non-existent path
"C:/building/msys64/mingw64/include"
in its INTERFACE_INCLUDE_DIRECTORIES. Possible reasons include:
* The path was deleted, renamed, or moved to another location.
* An install or uninstall procedure did not complete successfully.
* The installation package was faulty and references files it does not
provide.
Upon pressing e to exit the error message, the ccmake gui was exited and there were no files generated.
After googling a bit, this suggested creating the missing dirs, which seems merely a workaround (I did not try adapting this).
I created those dirs, and then I got
CMake Error in CMakeLists.txt:
Target "CGAL::CGAL" contains relative path in its
INTERFACE_INCLUDE_DIRECTORIES:
"C:/building/msys64/mingw64/include"
How can I fix this?
I found no suitable answers around.
Possibly related links, but which I found no way of relating to a suitable solution:
https://gitlab.kitware.com/cmake/cmake/commit/634bb33f3aa3b7f2181a896c025976b52171524a
https://cmake.org/cmake/help/v3.12/manual/cmake-buildsystem.7.html
NOTE:
The same happened with other examples.
EDIT:
I do not know what changed, but now upon pressing e to exit the error message (see above) all generated files are present, including the Makefile.
So I can proceed as described below in Previous state of affairs.
This happens either with or without the missing dirs (the error message changes, but I can generate my executables).
AFAICT, the only related change is that
I have just installed packages
, mingw-w64-x86_64-eigen3 (3.3.7-1)
, mingw-w64-x86_64-suitesparse (5.4.0-1)
, mingw-w64-x86_64-lapack (3.8.0-5)
, but I would not think this should matter.
Previous state of affairs
To compile and link CGAL examples, in my previous PC, some tweaking was needed.
Compilation went fine, but linking required a couple of tweaks.
An example command line produced for linking reads
/mingw64/bin/c++.exe -O3 -DNDEBUG -Wl,--enable-auto-import CMakeFiles/reconstruction_structured.dir/reconstruction_structured.cpp.o -o reconstruction_structured.exe -Wl,--out-implib,libreconstruction_structured.dll.a -Wl,--major-image-version,0,--minor-image-version,0 /mingw64/lib/lib/libmpfr.a /mingw64/lib/lib/libgmp.dll.a /mingw64/lib/lib/libCGAL.dll.a -lC:/building/msys64/mingw64/lib/libgmp.dll.a -lC:/building/msys64/mingw64/lib/libmpfr.a
The two types of required fixes are
Replacing /mingw64/lib/lib/ with /mingw64/lib/ in every (link.txt;build.make) file.
Replacing flag -lC:/building/msys64/mingw64/lib/libgmp.dll.a with -lgmp in every link.txt file. Similarly for -lmpfr.
Run this command to find all the places in the CGAL CMake files where the erroneous directory C:/building/ is mentioned:
grep -r '/building/' /mingw64/lib/cmake/CGAL
For each instance, either delete it or replace it with something more appropriate.
(Note: I haven't tried this myself.)
This erroneous directory is a general issue with MSYS2 which I attempted to fix at one point.
After How to solve "FATAL: kernel too old" when running gem5 in syscall emulation SE mode? I managed to run a statically linked hello world under certain conditions.
But if I try to run an ARM dynamically linked one against the stdlib with:
./out/common/gem5/build/ARM/gem5.opt ./gem5/gem5/configs/example/se.py -c ./a.out
it fails with:
fatal: Unable to open dynamic executable's interpreter.
How to make it find the interpreter? Hopefully without copying my cross' toolchain's interpreter on my host's root.
For x86_64 it works if I use my native compiler, and as expected strace says that it is using the native interpreter, but it does not work if I use a cross compiler.
The current FAQ says it is not possible to use dynamic executables: http://gem5.org/Frequently_Asked_Questions but I don't trust it, and then these presentations mention it:
http://www.gem5.org/wiki/images/0/0c/2015_ws_08_dynamic-linker.pdf
http://research.cs.wisc.edu/multifacet/papers/learning_gem5_tutorial.pdf
but not how to actually use it.
QEMU user mode has the -L option for that.
Tested in gem5 49f96e7b77925837aa5bc84d4c3453ab5f07408e
https://www.mail-archive.com/gem5-users#gem5.org/msg15582.html
Support for dynamic linking has been added in November 2019
At: https://gem5-review.googlesource.com/c/public/gem5/+/23066
It was working for sure at that point, but then it broke at some point and needs fixing.....
arm 32-bit https://gem5.atlassian.net/browse/GEM5-461
arm 64-bit https://gem5.atlassian.net/browse/GEM5-828
If you have a root filesystem to use, for example one generated by Buildroot you can do:
./build/ARM/gem5.opt configs/example/se.py \
--redirects /lib=/path/to/build/target/lib \
--redirects /lib64=/path/to/build/target/lib64 \
--redirects /usr/lib=/path/to/build/target/usr/lib \
--redirects /usr/lib64=/path/to/build/target/usr/lib64 \
--interp-dir /path/to/build/target \
--cmd /path/to/build/target/bin/hello
Or if you are using an Ubuntu cross compiler toolchain for example in Ubuntu 18.04:
sudo apt install gcc-aarch64-linux-gnu
aarch64-linux-gnu-gcc -o hello.out hello.c
./build/ARM/gem5.opt configs/example/se.py \
--interp-dir /usr/aarch64-linux-gnu \
--redirects /lib=/usr/aarch64-linux-gnu/lib \
--cmd hello.out
You have to add any paths that might contain dynamic libraries as a separate --redirect as well. Those are enough for C executables.
--interp-dir sets the root directory where the dynamic loader will be searched for, based on ELF metadata which says the path of the loader. For example, buildroot ELF files set that path to /lib/ld-linux-aarch64.so.1, and the loader is a file present at /path/to/build/target/lib/ld-linux-aarch64.so.1. As mentioned by Brandon, this path can be found with:
readelf -a $bin_name | grep interp
The main difficulty with syscall emulation dynamic linking, is that we want somehow:
linker file accesses to go to a magic directory to find libraries there
other file accesses from the main application to go to normal paths, e.g. to read an input file in the current working directory
and it is hard to detect if we are in the loader or not, especially because this can happen via dlopen in the middle of a program.
The --redirects option is a simple solution for that.
For example /lib=/path/to/build/target/lib makes it so that if the guest would access the C standard library /lib/libc.so.6, then gem5 sees that this is inside /lib and redirects the path to /path/to/build/target/lib/libc.so.6 instead.
The slight downside is that it becomes impossible to actually access files in the /lib directory of the host, but this is not common, so it works in most cases.
If you miss any --redirect, the dynamic linker will likely complain that the library was not found with a message of type:
hello.out: error while loading shared libraries: libstdc++.so.6: cannot open shared object file: No such file or directory
If that happens, you have to find the libstdc++.so.6 library in the target filesystem / toolchain and add the missing --redirect.
It later broke at https://gem5.atlassian.net/browse/GEM5-430 but was fixed again.
Downsides of dynamic linking
Once I got dynamic linking to work, I noticed that it actually has the following downsides, which might or not be considerable depending on the application:
the dynamic linker has to run some instructions, and if you have a very minimal userland test executable, and are running on a low CPU like O3, then this startup can dominate runtime, so watch out for that
ExecAll does not show symbol names for stdlib functions, you just get offsets from some random nearest symbol e.g. #__end__+274873692728. Maybe something along these lines would work: Debugging shared libraries with gdbserver but not sure
dynamically jumping to a stdlib function for the first time requires going through the dynamic linking machinery, which can create problems if you are trying to control a microbench.
I actually already hit this once: the dynamic version of a program was doing something extra that and that compounded with a gem5 bug broke my experiment, and cost me a few hours of debugging.
Interpreters like Python and Java
Python and Java are just executables, and the script to execute an argument to the executable.
So in theory, you can run them in syscall emulation mode e.g. with:
build/ARM/gem5.opt configs/example/se.py --cmd /usr/bin/python --options='hello.py arg1 arg2'
In practice however hugely complex executable like interpreters are likely to have syscalls that not yet implemented given the current state of gem5 as of November 2019, see also: When to use full system FS vs syscall emulation SE with userland programs in gem5?
Generally it is not hard to implement / ignore uneeded calls though, so give it a shot. Related threads:
Java: Running Java programs in gem5(or any language which is not C)
Python: 3.6.8 aarch64 fails with "fatal: syscall unused#278 (#278) unimplemented.", test setup
Old answer
I have been told that as of 49f96e7b77925837aa5bc84d4c3453ab5f07408e (May 2018) there is no convenient / well tested way for running dynamically linked cross arch executables in syscall emulation: https://www.mail-archive.com/gem5-users#gem5.org/msg15585.html
I suspect however that it wouldn't be very hard to patch gem5 to support it. QEMU user mode already supports that, you just have to point to the root filesystem with -L.
The cross-compiled binary should have an .interp entry if it's a dynamic executable.
Verify it with readelf:
readelf -a $bin_name | grep interp
The simulator is setup to find this section in main executable when it loads the executable into the simulated address space. If this sections exists, a c-string is set to point to that text (normally something like /lib64/ld-linux-x86-64.so.2). The simulator then calls glibc's open function with that c-string as the parameter. Effectively, this opens the dynamic linker-loader for the simulator as a normal file. The simulator then maps the file into the simulated address space in phases with mmap and mmap_fixed.
For cross compilation, this code must fail. The error message follows it directly if the simulator cannot open the file. To make this work, you need to debug the opening process and possibly the subsequent pasting of the loader into the address space. There is mechanism to set the program's entry point into the loader rather than directly into the code section of the main binary. (It's done through the auxiliary vector on the stack.) You may need to play around with that as well.
The interesting code is (as of 05/29/19) in src/base/loader/elf_object.cc.
I encountered this problem after I just cross compiled the code. You can try to add "--static" after the command.
I have an interpreter.ml file that contains an interpreter and some type definitions.
I've developed some test batteries to check if the interpreter works well or not.
When I put the functions I use to test the behaviour of the interpreter in the same file of the interpreter all works well, but if I try to use a different file for the tests (let's say tests.ml) it did not load the interpreter functions and definitions.
the interpreter.ml and tests.ml are in the same folder
I tried both with open Interpreter and #use "./interpreter.ml" from inside tests.ml but it wont compile nor shut down the warnings in the IDE (kind of...I'm using Visual Studio Code on MacOs )
I've already tried to follow the official documentation but it won't compile with ocamlopt -c tests.ml
As a result of discussions, the executable is obtained by compiling the 2 files test.ml & interpreter.ml in the right order (test.ml relies on objects defined in interpreter.ml; as a consequence test.ml has to reference to interpreter objects either via the clause open Interpreter or by prefixing all relevant items with Interpreter ):
ocamlc -o exec interpreter.ml test.ml
ocamlbuild is easier as it resolves by itself the dependencies:
The following command:
ocamlbuild test.native
will produce the executable.
Ultimately, I'm trying to build Apache QPID to run in the HPE NonStop OSS environment (a Posix-like environment on the NonStop system). The latest version of QPID uses cmake to build so I first need to get cmake to work for that environment. My earlier attempts tried to build in OSS directly (I needed to build cmake first before trying to build QPID), but I ran into many problems there. So lately I'm trying to build in Windows using a set of cross-development tools (compilers etc.) for NonStop. I've downloaded a Windows version of cmake 2.8 (suggested by the QPID build instructions) and am trying to use that with the X-dev tools to build QPID for OSS.
One big issue I've run into has to do with how cmake does things to test compilers and so forth early on. It will invoke the compiler to create an intermediate object file from C (and/or C++) source file and after that it will invoke the compiler to link an object file from the intermediate file. It seems that cmake prefers to add .obj to file names to create the intermediate object file name. This will work OK with my cross-compiler when creating the file (the name passed with -o to the compiler) but it will not work when passing this name for link purposes. Here is a short bit of the output per the CMakeError.log file (from trying build an OSS version of cmake 2.8 itself):
Determining if the C compiler works failed with the following output:
Change Dir: C:/Source/cmake-2.8.0/bld/CMakeFiles/CMakeTmp
Run Build Command:C:/cygwin/bin/make.exe "cmTryCompileExec/fast"
/usr/bin/make -f CMakeFiles/cmTryCompileExec.dir/build.make CMakeFiles/cmTryCompileExec.dir/build
make[1]: Entering directory '/cygdrive/c/Source/cmake-2.8.0/bld/CMakeFiles/CMakeTmp'
"C:/Program Files (x86)/CMake 2.8/bin/cmake.exe" -E cmake_progress_report C:/Source/cmake-2.8.0/bld/CMakeFiles/CMakeTmp/CMakeFiles 1
Building C object CMakeFiles/cmTryCompileExec.dir/testCCompiler.c.obj
/cygdrive/c/NonStop/tndm_cmplrs-j20/usr/bin/c89.exe -o CMakeFiles/cmTryCompileExec.dir/testCCompiler.c.obj -c C:/Source/cmake-2.8.0/bld/CMakeFiles/CMakeTmp/testCCompiler.c
Linking C executable cmTryCompileExec
/cygdrive/c/NonStop/tndm_cmplrs-j20/usr/bin/c89.exe "CMakeFiles/cmTryCompileExec.dir/testCCompiler.c.obj" -o cmTryCompileExec
c89.exe: error: Invalid input file extension"CMakeFiles/cmTryCompileExec.dir/testCCompiler.c.obj".
The cross-compiler fails because it requires intermediate object files to use .o for the extension, in order to determine they are intermediate object files. There is no way to get the c89 compiler to recognize testCCompiler.c.obj as a file type it knows what to do with.
So I've been searching (trying to find a local expert, but no one in my organization knows cmake; also numerous Google searches but could not find an answer) to see if there is any way to get cmake to change the name of the output file it uses for these type of compiles and tests. I've found info and then set CMAKE_C_OUTPUT_EXTENSION in a toolchain file:
SET(CMAKE_C_OUTPUT_EXTENSION ".o")
but that has made no difference.
If I can find a way to get cmake to create object files with names like testCCompiler.c.o instead of testCCompiler.c.obj, then the c89 cross-compiler would work.
Is it possible to do this?
UPDATE: I've managed to figure out that setting CMAKE_C_OUTPUT_EXTENSION in the toolchain file doesn't help. This gets overwritten in the CMakeCInformation.cmake (depending on whether UNIX is set or not). I also tracked down that UNIX gets set to true in Platform/UnixPaths.cmake, which gets INCLUDEd by various Platform files. So I've created a Modules/Platform/OSS.cmake file which includes it to takes care of that. I'll probably need/want to add other settings there later as I determine more flags for compilers etc that should be set to specific values for the OSS environment.
I've tried official howto but failed. I got error:
"The procedure entry point InterlockedCompareExchange#12 could not be located in the dynamic link library libstdc++-6.dll"
The problem was due the old gcc compiler, bundled with DevKit from rubyinstaller.org (4.5 vs 4.8 on my PC). Use MSYS instead. Assume we have zeromq source inside D:\libs\zeromq, then the procedure is:
Download GUI MinGW installer.
Install base and MSYS (if you already have working gcc compiler you probably only need MSYS).
Launch MSYS environment by executing C:\MinGW\msys\1.0\msys.bat.
Follow Using MSYS with MinGW section:
mount c:/mingw /mingw
cd /d/libs/zeromq
./configure --prefix=/mingw
make
Copy /d/libs/zeromq/src/.libs/libzmq.dll to your desired place.
In fact I needed to use ZeroMQ with C++, so I downloaded zmq.hpp, moved it to include directory, and compiled hwserver.cpp to test it:
C:\MinGW\bin\g++.exe -o hwserver hwserver.cpp -L. -lzmq -ID:\libs\zeromq\include
It worked, but when I launch it I got:
Assertion failed!
Program: D:\tmp\zmq\hwserver.exe
File: D:\libs\zeromq\include/zmq.hpp, Line 280
Expression: rc == 0
This application has requested the Runtime to terminate it in an unusual way.
Please contact the application's support team for more information.
I've managed to get rid of this failure by commenting lines 279, 280. Similar issue