Managed C++ Wrapper for Unmanaged Static Library with static variables hangs - c++

The explanation of the problem is a little long-winded, please bear with me.
I have an unmanaged C++ static library that is used for financial application. It has business day conventions, swap conventions, bond conventions, etc. Most of the conventions rely on static global variables, which are initialized on first use. The library also initializes the holiday calendars on startup by running some queries against a SQL Server database using ODBC.
I have to interact with third-party software using web services. The only way to do this realistically is through C#. That isn't an issue, and I was making good progress. However, I hit a stumbling block when it became necessary to do some date calculations in C#. Since I didn't want to port all my C++ code to C#, I figured the most efficient way to achieve this would be by writing a Managed C++ Class Library DLL that is a wrapper around my unmanaged static library. Everything seems to work fine, I get no compile-time or link-time errors, and I can add the reference to the wrapper and see all the proper object definitions. However, when I try to run my application, it just hangs. I have tried playing around with a bunch of compiler setting for the wrapper DLL, to no avail. If I remove the project dependency on my unmanaged library, everything works fine. I have a strong suspicion that my liberal use of global static variables is causing issues. Is there are way to solve this problem, are at least figure out where the issue is? Example code is below.
Thanks,
Marc.
// FSAManaged.h
#pragma once
using namespace System;
//class XLDate;
namespace FSAManaged {
public ref class Futures
{
public:
static DateTime Expiry(String ^ code, DateTime date);
};
}
The implementation does not even rely on a dependency to the unmanaged static library:
// This is the main DLL file.
#include "stdafx.h"
#include "FSAManaged.h"
namespace FSAManaged
{
DateTime Futures::Expiry(String ^ code, DateTime date) {
return DateTime::Today;
}
}
For completeness' sake, here is AssemblyInfo.cpp:
#include "stdafx.h"
using namespace System;
using namespace System::Reflection;
using namespace System::Runtime::CompilerServices;
using namespace System::Runtime::InteropServices;
using namespace System::Security::Permissions;
//
// General Information about an assembly is controlled through the following
// set of attributes. Change these attribute values to modify the information
// associated with an assembly.
//
[assembly:AssemblyTitleAttribute("FSAManaged")];
[assembly:AssemblyDescriptionAttribute("")];
[assembly:AssemblyConfigurationAttribute("")];
[assembly:AssemblyCompanyAttribute("?????")];
[assembly:AssemblyProductAttribute("FSAManaged")];
[assembly:AssemblyCopyrightAttribute("??????")];
[assembly:AssemblyTrademarkAttribute("")];
[assembly:AssemblyCultureAttribute("")];
//
// Version information for an assembly consists of the following four values:
//
// Major Version
// Minor Version
// Build Number
// Revision
//
// You can specify all the value or you can default the Revision and Build Numbers
// by using the '*' as shown below:
[assembly:AssemblyVersionAttribute("1.0.*")];
[assembly:ComVisible(false)];
[assembly:CLSCompliantAttribute(true)];
[assembly:SecurityPermission(SecurityAction::RequestMinimum, UnmanagedCode = true)];

Use the debugger. If you test this from C# then Project + Properties, Debug, tick "Enabled unmanaged code debugging". Setting up the symbol server in Tools + Options, Debugging, Symbols is strongly recommended. Run.
When it hangs use Debug + Break All. Debug + Windows + Threads and double-click the thread that is supposed to be doing the job. Debug + Windows + Call stack to see what is going on. Post the stack trace in your question if you can't figure it out. Anything you see in the Output window and the Visual Studio status bar is relevant too.

Static C++ variables are initialized from DllMain. There are lot's of things you should not do in DllMain; triggering the load of yet another Dll being the most important one. This is easy to break if you call into other peoples libraries in from DllMain.
I suggest you make an Init function on your Dll, which you call after the dll is up and running.

Related

Can't find COM object from C++, although Guid it's registered

First of all happy new year to everyone, hope you're doing well!
I'm working on a C++ project in which I need to call a C# DLL I created following the first answer of this post. Once I have the DLL, I need to call it from Qt, so by using dumpcpp and the .tlb file generated by regasm, I managed to get the .cpp and .h files to use my classes. Just as a reference, the namespace of the classes is Wrapper, and the main class is Device with guid {DD4A4896-C105-4C60-839B-B18C99C8FE15}.
Once I have the generated files to use the DLL, if I try to create a Wrapper:: Device instance on Qt, I get the following error:
QAxBase::setControl: requested control {dd4a4896-c105-4c60-839b-b18c99c8fe15} could not be instantiated
QAxBase::qt_metacall: Object is not initialized, or initialization failed
It doesn't give any more information, so I tried to check if the guid was stored on the system registry (I used the regasm command explained on the previously quoted post, and It said that it was successful, but you never know). Opening Registry editor and searching for the Guid revealed that it's present at: Computer\HKEY_CLASSES_ROOT\WOW6432Node\CLSID\{DD4A4896-C105-4C60-839B-B18C99C8FE15}, which, as far as I know, is the right route for these guids, and it points to the right DLL.
I though It may be due to some kind ActiveQt problem, and as the previously quoted post explained how to use that DLL from VS C++, I decided to give it a try, using this as an another reference. I've finished with this code, which is supposed to create an instance of my Device object
#include <iostream>
#include <atlstr.h>
#import "C:\Users\javie\Documents\Wrapper\Wrapper\bin\x86\Release\netstandard2.0\Wrapper.tlb" named_guids raw_interfaces_only
inline void TESTHR(HRESULT x) { if FAILED(x) _com_issue_error(x); };
int main()
{
try
{
TESTHR(CoInitialize(0));
Wrapper::IDevicePtr devPtr = nullptr;
TESTHR(devPtr.CreateInstance("{DD4A4896-C105-4c60-839B-B18C99C8FE15}"));
}
catch (const _com_error& e)
{
CStringW out;
out.Format(L"Exception occurred. HR = %lx, error = %s", e.Error(), e.ErrorMessage());
MessageBoxW(NULL, out, L"Error", MB_OK);
}
CoUninitialize();// Uninitialize COM
std::cout << "Hello World!\n";
}
However, this doesn't work either, the createInstance method throws an exception of Class not registered and HR=80040154. Again, according to Registry editor, the class is registered, so I don't understand the error. I've also tried with devPtr.CreateInstance("Wrapper.Device"), devPtr.CreateInstance("Wrapper::Device") or `devPtr.CreateInstance("Wrapper::CLSID_Device") as the links I posted suggest, but in those cases I get another exception with HR=800401f3 and message Invalid class string.
It doesn't matter whether VS or Qt Creator are opened as administrator or not, I get the exact same error.
I have run out of ideas, and I really need to be able to use that DLL from Qt using the files generated by dumpcpp.
Does any one know what could be happening? It feels quite strange to me.
If your C++ application is 64-bit, that's the answer right there, because your C# component is 32-bit (or MSIL but registered to the 32-bit hive). In situations like these, a simple test using VBScript is always useful.
Write a simple VB Script (test.vbs)
Dim obj
Set obj = CreateObject("Wrapper.Device") ' or whatever your ProgID is
MsgBox TypeName(obj)
Now, run this macro 2 ways: with 32-bit and 64-bit versions of VBScript:
32-bit > c:\windows\SysWow64\cscript.exe test.vbs
64-bit > c:\windows\system32\cscript.exe test.vbs
This is assuming your C# component is dispatch compatible. If it's not, then it will still give you differing results that you can use to debug.
Assuming automation/IDispatch compatible, one will work and one won't if you have registered your component correctly.
Have you registered correctly? When I use regasm, I always use the the switches /tlb /codebase when registering the C# component for COM.
Ok, in case someone find the same error, I'll explain the solution I found.
The problem was that in my case, the C# class I developed depended on another 32 bits dll which was not registered on my PC. Once I registered the other dll, everything worked fine.
I don't know why VS kept telling me that the class was not registered when my class itselft was registered, it was one of its dependencies that wasn't registered.
Anyway, I discovered this thanks to Joseph's comments, so thanks a lot for your help.

Can't use MFC CObject class in VS 2017 application

I need to use MFC Serialization mechanism to serialize objects of class Product:
class Product : public CObject
{
protected:
string name;
int expiring;
double price;
public:
Product();
~Product();
virtual void input_data();
virtual void print_data();
};
This is simple Windows Console Application. I got an error on CObject: not a class or struct name.
I tried to make MFC Console Application following the instruction in this comment: https://stackoverflow.com/a/50320168/6543699. Now I got a lot of errors (identifier not found or identifier not declared). The text of errors is in Russian, so I don't copy them here. This is how it looks:
I don't know anything about MFC using and can't find guide where it described clearly. My questions are:
1) Is it possible to use CObject in console application (non-MFC) and how?
2) If not, what should I do to be able to use MFC serialazation? Maybe include some headers or some components were just missing while installation?
You can just adjust a console app in a couple of steps to use MFC. First is to include afx.h, like:
#include <iostream>
#include <afx.h>
Then you will want to link with the MFC dynamic libraries.
Project Properties > Configuration Properties > Advanced > Use MFC
Select: Use MFC in a Shared DLL
It should now compile with CObject.
My note, I would not use MFC serialization, at the least use Boost Serialization I gave up using any serialization a long time ago because of the constant need to maintain versioning. I found it a night mare. Unless you see that your object structure will remain fairly static, I would recommend using XML to database your objects. It is a little more work to get going but way more often than not, you don't need to worry about versioning as you make changes.

In Unity3d test scripts, how do I call a static function from another class?

I have two files in a Unity3d project. One is a test script that runs in edit mode. The other is a single class with static functions that I'd like to call from the test scripts.
here's my test script:
using UnityEngine;
using UnityEngine.TestTools;
using NUnit.Framework;
using System.Collections;
public class NewTestScript
{
[Test]
public void TestAnotherStaticFunction()
{
int a = NewBehaviourScript.FunctionUnderTest(1);
int b = 1;
// Use the Assert class to test conditions.
Assert.IsTrue(a == b);
}
}
here's my function under test:
using System.Collections;
using System.Collections.Generic;
using UnityEngine;
public class NewBehaviourScript : MonoBehaviour
{
/// <summary>
/// the stupidest function in the world,
/// used to verify tests.
/// </summary>
public static int FunctionUnderTest(int a)
{
return a;
}
}
This gives me the error from the Unity compiler (I'm not building outside of Unity):
Assets/TestS/NewTestScript.cs(12,17): error CS0103: The name `NewBehaviourScript' does not exist in the current context
These are running in edit-mode.
I've tried adding and removing the SuperTestNameSpace namespace from the function under test and the calling code.
I've attempted adding/removing files from the .asmdef file that was autogenerated by unity, although this usually leads to other compile errors.
My previous unit test experience is largely in Visual Studio or VSCode, and I'm trying to get my unity3d test experience to match my prior test environment experiences.
Is there a fundamentally limited functionality in the edit-mode tests, or am I missing something stupid?
Further elaboration on the assemblies involved. It looks like there are two assemblies at play here: Assembly-CSharp.dll contains my code under test and TestS.dll contains my testing code. I believe my questions boils down to: how do I add a reference from the TestS.dll assembly to the Assembly-CSharp.dll. I'd know how to do this in Visual Studio (either via the context menu in VS or directly editing the csproj file), however I don't see how to do it in Unity3d. Any edit I make to the csproj file is frequently overwritten by unity, and while there is a 'references' section in the inspector (see picture) I can't add Assembly-CSharp.dll as a reference.
These are the inspector settings for TestS.asmdef. While there's an option to add references, I can't add a reference to Assembly-CSharp.dll, which is where my code-under-test lives.
Ok, I figured this out. There were two things going on:
Editor tests need to be underneath a folder called editor. It's really annoying that the unity editor doesn't do this for you.
You need to have an assembly definition for the code under test and add a reference from the test code to the newly created assembly definition. This must be done in the Unity editor UI.
by default, unity adds your script code to an assembly called Assembly-CSharp.dll, and, for reasons unknown, this assembly isn't referenced by my edit mode test code. I'm not sure if this is a bug in Unity or if it's by design, but explicitly creating and referencing the assembly definition has cleared things up.
The main issue is currently you are trying to call the
NewBehaviourScript(1);
constructor which does not exist...
instead of the method
using SuperTestNameSpace;
//...
NewBehaviourScript.FunctionUnderTest(1);
or alternatively with the namespace in the call directly
SuperTestNameSpace.NewBehaviourScript.FunctionUnderTest(1);
Also make sure the filename matches exactly the class name. So in your case it should be
NewBehaviourScript.cs
Note that the .cs is not printed by Unity in the Project view so there it should say NewBehaviourScript.
Why does it not compile with the using SuperTestNameSpace;? What is the error?
If that exception
Assets/TestS/NewTestScript.cs(14,17): error CS0103: The name `NewBehaviourScript' does not exist in the current context
is only shown in VisualStudio but the script compiling fine in Unity especially after adding a new script it helps to simply close and restart VS.
In some cases it also helps to close Unity and VS, remove all files and folders except Assets and ProjectSettings (and if you are under version control anything that belongs to it like e.g. .git, .gitignore, .gitattributes etc) in particular delete the Library, .vs folder and all .csproj and .sln files.
Than open Unity again and let it recompile everything.
Make sure the file that contains your NewBehaviourScript class IS NOT inside an Editor folder.
Move both the scripts in the Assets (root) folder and try again.

Including C++ headers in user mode programs built with NT DDK

So...I have a kernel mode component and a user mode component I'm putting together using the turnkey build environment of the NT DDK 7.1.0. The kernel component is all .c/.h/.rc files. The user mode component is .cpp/.c/.h/.rc files.
At first it seemed simplest to use build for both, as I saw you could modify the ./sources file of the user mode component to say something like:
TARGETNAME = MyUserModeComponent
TARGETTYPE = PROGRAM
UMTYPE = windows
UMENTRY = winmain
USE_MSVCRT = 1
That didn't seem to cause a problem and so I was pleased, until I tried to #include <string> (or <memory>, or whatever) Doesn't find that stuff:
error C1083: Cannot open include file: 'string': No such file or directory
Still, it's compiling the user mode piece with C++ language semantics. But how do I get the standard includes to work?
I don't technically need to use the DDK build tool for the user mode piece. I could make a visual studio solution. I'm a bit wary as I have bumped into other annoyances, like the fact that the DDK uses __stdcall instead of __cdecl by default... and there isn't any pragma or compiler switch to override this. You literally have to go into each declaration you care about and change it, assuming you have source to do so. :-/
I'm starting to wonder if this is just a fractal descent into "just because you CAN doesn't mean you SHOULD build user mode apps with the DDK. Here be dragons." So my question isn't just about this particular technical hurdle, but rather if I should abandon the idea of building a C++ user mode component with the DDK tools...just because the kernel component is pure C.
To build a user mode program with WINDDK you need to add some variables to your SOURCES file:
386_STDCALL=0 to use cdecl calling convention by default
USE_STL=1 to use STL
USE_NATIVE_EH=1 to add a support for exception handling
Everything else you already have.
I'll put my full SOURCES file for reference:
TARGETNAME = MyUserModeComponent
TARGETTYPE = PROGRAM
TARGETPATH = obj
UMTYPE = console
UMENTRY = main
USE_MSVCRT = 1
USE_NATIVE_EH=1
USE_STL=1
386_STDCALL=0
SOURCES= main.cpp
And main.cpp:
#include <iostream>
#include <string>
using namespace std;
int main()
{
string s = "bla bla bla!";
cout << s;
return 0;
}
Have fun!
Quick Answer
Abandon the idea of building user-mode components with DDK tools (although I find the concept fascinating :-P)
Your kernel mode component should be built separately from the user mode components as a matter of good practice.
Vague thoughts
Off the top of my head, and this really speaking from limited experience...there are a lot of subtle differences that can creep up if you try to mix the two together.
Using your own example of __cdecl vs __stdcall; You have two different calling conventions. _cdecl is all kernel stuff and all of the C++ methods are wrapped around in WINAPI (_stdcall) passing conventions and __stdcall will clean do auto stack clean up and expect frame pointers inserted all over the place. And if you by accident use compiler options to trigger a __fastcall, it would be a pain to debug.
You can definitely hack something together, but do you really want to keep track of that in your user-space code and build environment? UGH I say.
Unless you have very specific engineering reasons to mix the two environments, (and no a unified build experience is not a valid reason, because you can get that from a batch file called buildall.bat) I say use the separate toolchains.

Problem debugging C++ with an Eclipse based IDE

This is a weird question in that I'm not sure where to start looking.
First of all, I haven't done any C++ programming for the last 10 years so it could be me thats forgotten a few things. Secondly, the IDE I'm using is Eclipse based (which I've never used) and customized for Samsung bada based mobile development (it kicks off an emulator for debugging purposes)
I'm posting my code samples as images because the StackOverflow WYSIWYG editor seems to have a problem parsing C++.
[EDIT] Due to complaints I've edited my question to remove the images. Hope that helps :)
I have the following header file...
#include <FApp.h>
#include <FBase.h>
#include <FGraphics.h>
#include <FSystem.h>
#include <FMedia.h>
using namespace Osp::Media;
using namespace Osp::Graphics;
class NineAcross :
public Osp::App::Application,
public Osp::System::IScreenEventListener
{
public:
static Osp::App::Application* CreateInstance(void);
public:
NineAcross();
~NineAcross();
public:
bool OnAppInitializing(Osp::App::AppRegistry& appRegistry);
private:
Image *_problematicDecoder;
};
...and the following cpp file...
#include "NineAcross.h"
using namespace Osp::App;
using namespace Osp::Base;
using namespace Osp::System;
using namespace Osp::Graphics;
using namespace Osp::Media;
NineAcross::NineAcross()
{
}
NineAcross::~NineAcross()
{
}
Application* NineAcross::CreateInstance(void)
{
// Create the instance through the constructor.
return new NineAcross();
}
bool NineAcross::OnAppInitializing(AppRegistry& appRegistry)
{
Image *workingDecoder;
workingDecoder->Construct();
_problematicDecoder->Construct();
return true;
}
Now, in my cpp file, if I comment out the line that reads _problematicDecoder->Construct();...I'm able to set a breakpoint and happily step over the call to Constuct() on workingDecoder. However, as soon as I uncomment the line that reads _problematicDecoder->Construct();... I end up with the IDE telling me...
"No source available for "Osp::Media::Image::Construct()"
In other words, why can I NOT debug this code when I reference Image *image from a header file?
Any ideas?
Thanks :-)
This usually means you're stepping through some code which you do not posses its source.
I assume here that Osp::Media::Image is a class supplied by Samsung or similar for which you do not have the cpp file. So this means the debugger can't show you the current code line while you're at a function of Osp::Media::Image.
Alternatively, there's a good chance you do have all of the source code for this class, but Eclipse doesn't know where it is. In this case you can add the correct directories under the Debug Configurations window.
Ok, problem solved.
The idea is to first new up an instance of Image like so...
_decoder = new Osp::Media::Image();
And then do _decoder->Construct().
Funny enough, this seems blatantly obvious to me now coming from the C# world, though why the code I posted for workingDecoder works is still somewhat mysterious to me. The fact the sample projects pre-loaded with the bada IDE don't seem to make a call to new() leads me to believe that perhaps those samples are outdated our out of synch.
Either that or I really AM wildly out of the C++ loop.
Anyway thanks so much for the effort guys.
Appreciated :)