I'm attempting to build TensorFlow's C++ library for Windows XP. While I've been able to build and use it on Windows 10, 32-bit XP isn't working. The background: I'm working on a COM module that calls fuctions from tensorflow.dll. My build environment:
Visual Studio 2017 15.7
CMake 3.11.1
TensorFlow 1.8
Windows 10
The sequence I use to build tensorflow.dll is:
Open "x64_x86 Cross Tools Command Prompt for VS 2017"
Try to force the use of functions availablbe in Win XP: set CXXFLAGS=/D_WINVER=0x0501 /D_WIN32_WINNT=0x0501
Add Git to path: set PATH=%PATH%;C:\Program Files (x86)\Git\bin
Fix CMake file for converting *.proto files to *.pb.h files as described here.
Configure CMake: cmake .. -A Win32 -T v141_xp,host=x64 -DCMAKE_SYSTEM_VERSION=7.0 -DCMAKE_BUILD_TYPE=Release -DPYTHON_EXECUTABLE=C:\Users\williams\AppData\Local\Continuum\Anaconda3\envs\tensorflow\python.exe -Dtensorflow_BUILD_SHARED_LIB=ON -Dtensorflow_BUILD_PYTHON_BINDINGS=OFF -Dtensorflow_WIN_CPU_SIMD_OPTIONS="/arch:IA32"
Build: cmake --build . --target tensorflow --config Release -- /fileLogger /m:1 /p:CL_MPCount=1
The last step also involves some manual labour as the build process doesn't copy .lib files from the 3rd part dependencies to where they are needed. For whatever reason, a bunch of INSTALL projects never get run so I had to do that manually each time the build would fail while looking for a missing lib file. Once that was done, the build completed successfully.
Next I copy my COM module (a DLL) and the TensorFlow DLL over to a Windows XP virtual machine for testing and try to register the COM module, but get an error LoadLibrary("MyDLL.dll") - The specified procedure could not be found. I don't know what procedure it is looking for, so the best I can offer is that Dependency Walker highlights WS2_32.DLL and tells me it can't find inet_ntop and inet_pton.
Any suggestions on how to build TensorFlow so that it doesn't use these two functions?
P.S. suggestions of "Stop using XP, its old and no longer supported" don't help here. Upgrading to Windows 10 is an absolute last resort because of the disruption it would cause at the facility where this software will be tested.
Edit 1:
These two functions inet_pton and inet_ntop were only used in one file that forms part of Google Cloud Storage support in TensorFlow. The build process generated a tensorflow_static.lib in addition to tensorflow.dll. Linking against the static version and adding a few dependencies that aren't included in tensorflow_static.lib got rid of the code using inet_* functions.
My COM module still isn't working on Windows XP though because the file tensorflow\core\platform\windows\env.cc uses functions like CloseThreadpoolWork, submit SubmitThreadpoolWork, etc. that were only introduced in Windows Vista. It looks like I'll have to replace them with something else, as I don't see an alternative implementation in TensorFlow.
Additionally, I found that tensorflow\contrib\cmake\CMakeLists.txt forces _WIN32_WINNT=0x0A00 and that CXXFLAGS is the wrong environment variable to use. Changing it to CMAKE_CXX_FLAGS at least gets my macro definitions included, FWIW.
It is almost impossible to port tensorflow to windows xp, because:
TF's platform depedent code requires some Windows APIs later than winxp such as Thread Pool API. This would possibly bypassed by using third party thread pool libs.
The nsync, protobuf and eigen, which are core parts of TF, use C++11 thread_local, which makes them unable to run-time load as dll, see https://learn.microsoft.com/en-us/cpp/parallel/thread-local-storage-tls?view=vs-2017 for details. This later feature can theoretically be replaced by old windows TLS API, which requires many modifications on TF's core framework.
Anyway, if you really need the xp support, good luck with that.
In the end I gave up on this as simply being impossible. Even replacing the thread pool functions with something from Boost didn't help. If someone else manages to get this working, I'll gladly accept that as the answer, but so far this looks impossible.
Related
Trying to build ArrayFire examples, everything goes well until I get to the CUDA ones. They are supposed to be skipped, since I have an AMD processor/GPU. However, during the build process, the CUDA section is built anyway, failing for obvious reasons, interrupting the rest of the process.
I could manually change the CMakeLists.txt files. However, is there a higher level way to let the build system (cmake) know that I do not have a CUDA compatible GPU?
It looks like the ArrayFire_CUDA_FOUND and CUDA_FOUND macros are erroneously defined on my system.
The ArrayFire CMake build provides a flag to disable the CUDA backend. Simply set AF_BUILD_CUDA to NO via -DAF_BUILD_CUDA=NO at the command line to disable CUDA.
i have a separate build-flow that can produce .out files for the MT3620/M4 core using the same gcc-arm compiler.... for reasons beyond the scope of this question, it's not practical for me to migrate this build-flow into cmake....
once i generate the .out file, my thinking was to simply envelop it within a project structure identical to the samples -- with an conforming app_manifest.json file.... from here, i could effectively run just the final step of the cmake build -- which generates the image package....
before i go down this path, is there anything obvious i'm missing here????
As of today you will need to decide to install Azure Sphere SDK Preview for Visual Studio or Azure Sphere SDK Preview for Windows (You cannot install both SDKs on the same PC.)
To develop applications, however, you need to set up your development environment and possibly install additional tools. The Azure Sphere SDK for Windows supports development with the CLI only. If you've installed this version of the SDK, you'll need to install CMake and Ninja on your PC.
If you plan to develop real-time capable applications (RTApps), you might need to install a compiler for the embedded processor and set up hardware to display output from the app.
Both Quickstarts "Build an RTapp on cmd" and "Build an RTApp with Visual Studio" require that you use CMake at some stage and those are the supported methods.
Update: New documentation improvements will be done soon to address the option of using a different build system to cmake and "yes" it should work as you suggest.
I use Cygwin to build source code to DLL used by windows app.
When I use GCC core / GCC g++, the app crash if it calls function (which includes printf or malloc) in DLL.
When I use Mingw64-x86_64-gcc-core / Mingw64-x86_64-gcc-g++ it reports error like sys/socket.h：No such file or directory.
Can anyone explain how to do it? Thanks.
The first problem is due to the tentative to build a stand alone DLL (not depending on cygwin1.dll) using cygwin only specific tools.
You have collision between multiple malloc and other C library call present in cygwin1.dll.
The second is due to the fact that sys/socket.h does not exist on Windows
see for possible solution:
Using sys/socket.h functions on windows
So you need to define what is your target : Cygwin/Posix or Windows and choose programming style and tools accordingly, you can not mix.
I want to use KDL (Kinematics and Dynamics Library) in robot control box. But robot control box uses SCons as their build system while KDL uses CMake.
It turned out that the control box doesn't have CMake installed. Should I install CMake in the control box? Or write SCons file for compiling KDL?
====================================================
My question is ambiguous. Sorry for that. And unfortunately, I cannot show the link of Control Box, it's not public. Here is link of KDL installation manual.
http://www.orocos.org/kdl/installation-manual
Let me make it more clear.
Forget all of previous question above and all about Control box, KDL. Let's say that you want to use one library. But the library can be built using CMake according to installation manual. Your PC doesn't have CMake installed but it has SCons, and unfortunately you should not install CMake on your PC.
If you can only use SCons, what can you do?
I know this situation is not usual, I want to know your opinion.
To answer your initial question: Yes, you should always try to install CMake, if that is a build requirement for you library and if you need to build that library from the sources.
To answer your later question: Replacing or rewriting the build system scripts is a major effort and not advisable. In general there is no script to convert build-systems. Such script might help to make the manual transformation. If you have a look at LLVM's effort to replace Autotools by CMake or Boost replacing it's own build system by CMake, you find out it takes several people several years and still not everybody is satisfied.
Often you don't need to build the library yourself. Either there are already built packages from the project directly of from your distribution (Debian etc. packages) or third party packagers like Mac Ports or NuGet.
In your case KDL provides Debian/Ubuntu packages.
Additional KDL is part of ROS, which is experimental in Homebrew for OS X.
there!
I'd like to execute example program, Surface_reconstruction_points_3.
I think the program needs additionally 3rd library, MPFI.
So I downloaded the library from http://mpfi.gforge.inria.fr/.
And I unzipped it at proper folder and linked it to the system path.
After configurating the example program using cmake-gui,
I found some error message, 'Could NOT find MPFI (missing: MPFI_LIBRARIES)'.
How can I install MPFI library in Window using visual studio 2008?
Since MPFI is a GNU library, it is completely supported on Windows.
CGAL provides this third-party libraries inside its installation, but Windows has its own replacement.
I installed this correctly with Visual Studio, but you need to uncheck those third-party libraries so CGAL won't mess with the own windows replacements.
Maybe doing this kind of installation isn't working well for all the sample codes inside the installation, but it works fine for development. I used CGAL without those third-libraries for Voronoi and 2D and 3D triangulation and it works fine.
For a clean installation, delete all you installed previously, make sure the new installation isn't messing with the bad previous installation.
EDIT:
About not using those optional libraries for Windows, I just received a similar mail through the CGAL mailing list and these could be the steps for newer versions of CGAL (>4.2):
You have to comment 3 lines in the CMake machinery:
In CGAL-4.1/CMakeList.txt you have to comment:
#list (INSERT CGAL_ESSENTIAL_3RD_PARTY_LIBRARIES 0 GMP MPFR)
#hide_variable(CGAL_ESSENTIAL_3RD_PARTY_LIBRARIES)
In CGAL-4.1/cmake/modules/UseCGAL.cmake you have to comment:
#use_essential_libs()
On the other hand, you can search on internet for precompiled versions of MPFI. For instance, you can find precompiled GMP, MPFR and MPFI for Windows on http://vegas.loria.fr/rs/ .