CSingleLock Instantiation gets a Debug Assert

[greve]

I'm getting a debug assert at the instantiation of a CSingleLock in the situation shown below

size_t CErrorNotifier::CImplementation::GetNoOfErrorsActualyStored(void) const
    {
    size_t NoOfErrorsActualyStored;
    CSingleLock LocalSync(&m_Sync);  //m_Sync is a CSemaphore-object inside the CErrorNotifier::CImplementation-object

    LocalSync.Lock();
    NoOfErrorsActualyStored= m_ActualNoOfErrorsStored;
    LocalSync.Unlock();

    return NoOfErrorsActualyStored;
    }

The problem shows up in IsDerivedFrom() during debugging:

BOOL CRuntimeClass::IsDerivedFrom(const CRuntimeClass* pBaseClass) const
{
	ASSERT(this != NULL);
	ASSERT(AfxIsValidAddress(this, sizeof(CRuntimeClass), FALSE));
	ASSERT(pBaseClass != NULL);
	ASSERT(AfxIsValidAddress(pBaseClass, sizeof(CRuntimeClass), FALSE));

	// simple SI case
	const CRuntimeClass* pClassThis = this;
	while (pClassThis != NULL)
	{
		if (pClassThis == pBaseClass)
			return TRUE;
#ifdef _AFXDLL
		pClassThis = (*pClassThis->m_pfnGetBaseClass)();
#else
		pClassThis = pClassThis->m_pBaseClass;
#endif
	}
	return FALSE;       // walked to the top, no match
}

The condition if (pClassThis == pBaseClass) is never met although the m_lpszClassName member of both pClassThis and pBaseClass are the same: "CSyncObject". Under what circumstances can there be different CRuntimeClass Instances?

[greve]

When I described the problem some days ago I forgot to mention that CErrorNotifier is part of a DLL. At that time I wasn't aware of the following: The problem occurs when the function shown (GetNoOfErrorsActualyStored()) is called from outside the DLL. When it is called from another function of the same DLL everything works as expected. Does anybody has a clue what the problem might be?

[MikeAThon]

Are the DLL and the "outside" both compiled with the same version of MFC and the same version of flags for VS?

[greve]

@MikeAThon:

Are the DLL and the "outside" both compiled with the same version of MFC

Yes, VC6

and the same version of flags for VS

Yes, as far as I can see:

"Outside":
- Compiler-Options
/nologo /MTd /W3 /Gm /GR /GX /ZI /Od /D "WIN32" /D "_DEBUG" /D "_WINDOWS" /D "_MBCS" /FR"C:\_JOBS\CurrentStuff\_Intermediates\ETACtrlComponents\\" /Fo"C:\_JOBS\CurrentStuff\_Intermediates\ETACtrlComponents\\" /Fd"C:\_JOBS\CurrentStuff\_Intermediates\ETACtrlComponents\\" /FD /GZ /c 
- Linker-Options
/nologo /subsystem:windows /incremental:yes /pdb:"C:\_JOBS\CurrentStuff\_Products\ETACtrlComponents\ETACtrl.pdb" /debug /machine:I386 /out:"C:\_JOBS\CurrentStuff\_Products\ETACtrlComponents\ETACtrl.exe" /pdbtype:sept 

DLL:
-Compiler-Options:
/nologo /MTd /W3 /Gm /GR /GX /ZI /Od /D "WIN32" /D "_DEBUG" /D "_WINDOWS" /D "_MBCS" /D "_USRDLL" /D "TYPELIBSPECTROSCOPICDEVICES_EXPORTS" /D "_WINDLL" /FR"C:\_JOBS\CurrentStuff\_Intermediates\ETACtrlComponents\\" /Fo"C:\_JOBS\CurrentStuff\_Intermediates\ETACtrlComponents\\" /Fd"C:\_JOBS\CurrentStuff\_Intermediates\ETACtrlComponents\\" /FD /GZ /c 
-Linker-Options:
/nologo /dll /incremental:yes /pdb:"C:\_JOBS\CurrentStuff\_Products\ETACtrlComponents\SpectroscopicDevices.pdb" /debug /machine:I386 /out:"C:\_JOBS\CurrentStuff\_Products\ETACtrlComponents\SpectroscopicDevices.dll" /implib:"C:\_JOBS\CurrentStuff\_Products\ETACtrlComponents\SpectroscopicDevices.lib" /pdbtype:sept

[VictorN]

I'm getting a debug assert at the instantiation of a CSingleLock in the situation shown below
...
The problem shows up in IsDerivedFrom() during debugging:

BOOL CRuntimeClass::IsDerivedFrom(const CRuntimeClass* pBaseClass) const
{
	ASSERT(this != NULL);
	ASSERT(AfxIsValidAddress(this, sizeof(CRuntimeClass), FALSE));
	ASSERT(pBaseClass != NULL);
	ASSERT(AfxIsValidAddress(pBaseClass, sizeof(CRuntimeClass), FALSE));

	// simple SI case
	const CRuntimeClass* pClassThis = this;
	while (pClassThis != NULL)
	{
		if (pClassThis == pBaseClass)
			return TRUE;
#ifdef _AFXDLL
		pClassThis = (*pClassThis->m_pfnGetBaseClass)();
#else
		pClassThis = pClassThis->m_pBaseClass;
#endif
	}
	return FALSE;       // walked to the top, no match
}

Where exactly (at what line) does the Debug assertion fail?

[greve]

Actually the 2nd assert in the constructor (IsKindOf()) is raised

CSingleLock::CSingleLock(CSyncObject* pObject, BOOL bInitialLock)
{
	ASSERT(pObject != NULL);
	ASSERT(pObject->IsKindOf(RUNTIME_CLASS(CSyncObject)));

	m_pObject = pObject;
	m_hObject = pObject->m_hObject;
	m_bAcquired = FALSE;

	if (bInitialLock)
		Lock();
}

IsKindOf() itself calls up IsDerivedFrom() shown in my initial post.

BOOL CObject::IsKindOf(const CRuntimeClass* pClass) const
{
	ASSERT(this != NULL);
	// it better be in valid memory, at least for CObject size
	ASSERT(AfxIsValidAddress(this, sizeof(CObject)));

	// simple SI case
	CRuntimeClass* pClassThis = GetRuntimeClass();
	return pClassThis->IsDerivedFrom(pClass);
}

IsDerivedFrom() returns FALSE due to the reason mentioned and thus IsKindOf() also returns FALSE which raises the assertion.

[GCDEF]

Use the debugger to look at pObject in this line.

ASSERT(pObject->IsKindOf(RUNTIME_CLASS(CSyncObject)));

What does the debugger think it is?

[greve]

pObject is of type CSemaphore and the member classCSyncObject is set to "CSyncObject".

[GCDEF]

pObject is of type CSemaphore and the member classCSyncObject is set to "CSyncObject".

You got that from the debugger and you're sure the ASSERT is coming from that line? If pObject is a CSyncObject, I don't see how ASSERT(pObject->IsKindOf(RUNTIME_CLASS(CSyncObject))); would fire.