Click here to Skip to main content
Click here to Skip to main content
Go to top

How to Find a COM Object Connected to Internet Explorer

, 28 Jan 2008
Rate this:
Please Sign up or sign in to vote.
An article on how to retrieve information from a COM object to determine COM DLL filename.

Introduction

A programmer who works with Internet Explorer must pay attention to where the other extension DLLs are. This is because these other extension DLLs interact with Internet Explorer and generate new problems and features. The programmer has to deal with this situation. However, there are no documents about this. In this document, I will explain the detailed solution.

COM Component

COM is not a simple Win32 DLL. COM is created by the Interface and to call its function, use the Invoke function. For these reasons, the COM DLL's function is not in the EXPORT table and cannot retrieve its entry point by GetPRocAddress. This is the reason why we cannot access its real code and why we cannot know its base DLL.

We can find out about all of a DLL's base address and size with the ToolHelp API. However, in case of COM, the instance of the DLL is not in the DLL address area. So, we cannot compare the ToolHelp API's result. For example, if we use function ABCD, in the case of a simple DLL, we use it as ABCD(); and we can get the address of the function by ABCD;. However, in the case of COM, we have to use it via the Invoke function. So, there is no raw access point of the ABCD function (method).

Internet Explorer and Connection Point

In order to receive an event from Internet Explorer, connect a COM object using the Connection Point interface. Just like a Virtual Function in C++, a connected COM object's function will be called by Internet Explorer. If you want to know the details, please refer to books on COM. Connection objects have a Connection Point; this interface provides a method for enumerating connection points. To use this method, we can get the IUnknown interface pointer and the cookie value. The cookie value is used to connect to Internet Explorer.

To connect to Internet Explorer, an object uses the Advise and Unadvise functions. These functions' result value is a cookie. It is a connection identifier. One funny tip is that the Unadvise function disconnects an object from Internet Explorer even if the cookie value is not yours. The IUnknown object which you get by this method has to release by the Release function. This is because, in the process of Enumeration (on the Enumeration function), AddRef is called and it increases the reference count.

Go into the COM

Let's dig into the COM component. To know about COM, we start with implementation of a COM component. Open the Unknwn.h file and we can see the following code:

extern "C++"
{
    MIDL_INTERFACE("00000000-0000-0000-C000-000000000046")
    IUnknown
    {
    public:
        BEGIN_INTERFACE
        virtual HRESULT STDMETHODCALLTYPE QueryInterface(
            /* [in] */ REFIID riid,
            /* [iid_is][out] */ void __RPC_FAR *__RPC_FAR *ppvObject) = 0;

        virtual ULONG STDMETHODCALLTYPE AddRef( void) = 0;

        virtual ULONG STDMETHODCALLTYPE Release( void) = 0;

        template<class>
    HRESULT STDMETHODCALLTYPE QueryInterface(Q** pp)
    {
        return QueryInterface(__uuidof(Q), (void **)pp);
    }

        END_INTERFACE
    };
} // extern C++

The code says that the IUnknown interface consists of three Virtual functions. The Virtual function saves the implementation function's address. To understand this code, you have to know about the implementation of the Virtual function. Let's look at the following example to study the Virtual function:

class CVirtualTest {
public:
    virtual void VFunc1()
    {
    }

    virtual void VFunc2()
    {
    }
};

class CVirtualTest2 : public CVirtualTest{
public:
    void VFunc1()
    {
        DWORD dwEIP = 0;
        _asm {
            call GETEIP3
GETEIP3:
            pop eax
            mov dword ptr[dwEIP], eax
        }

        TRACE("CVirtualTest2::VFunc1 = EIP : %X\n", dwEIP);
    }

    void VFunc2()
    {
        DWORD dwEIP = 0;
        _asm {
            call GETEIP4
GETEIP4:
            pop eax
            mov dword ptr[dwEIP], eax
        }

        TRACE("CVirtualTest2::VFunc2 = EIP : %X\n", dwEIP);
    }
};

void main()
{
    CVirtualTest2 v;

    TRACE("v is %X\n", &v);

    v.VFunc1();
    v.VFunc2();
}

Let's try debugging. First, set the break point to the main function and start the application. This is the result of TRACE and the debugger after the creation of v.

We can see the addresses. The address of v is 0x0012F5F8 and that of __vfptr is 0x004154d8. The Virtual function's addresses are 0x00401014 and 0x0040100a. As you can see, the Virtual Function Table's real address is the real code's address and if you know this address, we can know where the code is. By IUnknown's definition, IUnknown is a struct (class) which has a Virtual function. So, if we know exactly about the Virtual Function Table, we can say "We know IUnknown." First, let's see the CVirtualTest's address (0x12F5F8).

We can see 0x004154D8 in 0x0012F5F8. Aren't you familiar with this? Right! It is the same as __vfptr in the Debugger Window. In other words, __vfptr is the first member of struct when class only has the Virtual function. Let's go to the __vfptr's address.

We can see 0x00401014, 0x0040100A at 0x004154D8. These are the addresses of VFunc1 and VFunc2. At last, we get the Virtual function's address. Collect this, go to the class object address and we can get the Virtual Function Table's address. Go to the Virtual Function Table's address and we can get each function's address.

Integrate Codes

Back to the subject: find the connected Connection Points of Internet Explorer. This work can be done with IConnectionPoint::EnumConnections. We can get the IEnumConnections object from the EnumConnections method and get each connection information by the IEnumConnections::Next method. This returns CONNECTDATA. CONNECTDATA contains the IUnknown interface and we can get IDispatch through it. The IDispatch interface has a similar form as IUnknown and inherits from IUnknown.

IEnumConnections* pConn = NULL;
hr = pConnectionPoint->EnumConnections(&pConn);
if(!SUCCEEDED(hr)) {
    return ;
}

CONNECTDATA sConnData;
ULONG uRet = 0;

while(true) {
    uRet = 0;
    memset(&sConnData, 0, sizeof(CONNECTDATA));
    hr = pConn->Next(1, &sConnData, &uRet);

    if(hr != S_OK || uRet != 1)
        break;

    LPVOID* lpVFT = (LPVOID*)(sConnData.pUnk);

    if(IsBadReadPtr(lpVFT, sizeof(LPVOID)) == FALSE) {
        CString szOutput;
        szOutput.Format("Filename : %s, 0x%x", 
            GetProcessFileName( (DWORD)(*lpVFT) ), *lpVFT );
        OutputDebugString(szOutput):
    }

    sConnData.pUnk->Release();
}

In this code, we cast the IUnknown pointer to LPVOID* and confirm the address by using IsBadReadPtr. lpVFT's content is the Virtual Function Table's address. So, *lpVFT is lpVFT[0]. In conclusion, it is the Virtual Function Table's first element, the first function's address. To compare this address with the ToolHelp API's result, we can see what DLL has the function. The GetProcessFileName function does this work in the recently used code.

CString GetProcessFileName(DWORD dwAddress)
{
    // reload....
    BOOL        bRet     = FALSE;
    BOOL        bFound     = FALSE;
    HANDLE    hModuleSnap = NULL;
    MODULEENTRY32 me32 = {0};
    DWORD dwBase, dwSize;

    hModuleSnap = CreateToolhelp32Snapshot(TH32CS_SNAPMODULE, GetCurrentProcessId());
    if (hModuleSnap == INVALID_HANDLE_VALUE)
        return "";

    me32.dwSize = sizeof(MODULEENTRY32);

    if (Module32First(hModuleSnap, &me32))  {
        do  {
            dwBase = (DWORD)me32.modBaseAddr;
            dwSize = (DWORD)me32.modBaseSize;

            if((dwBase < dwAddress) && ((dwBase + dwSize) > dwAddress)) {
                CloseHandle(hModuleSnap);
                return me32.szExePath;
            }
        } while (Module32Next(hModuleSnap, &me32));
    }

    CloseHandle (hModuleSnap);

    return "";
}

Points of Interest

The COM component's implementation is interesting. These days, Windows is almost wrapped by the COM component and we cannot do programming without COM. So, it's important to understand COM component's implementation.

History

  • 20th January, 2008: First release

License

This article, along with any associated source code and files, is licensed under The Code Project Open License (CPOL)

Share

About the Author

Yonghwi Kwon
Software Developer
United States United States
I started to write software since 1999 and have developed various products including security solutions and system utilities.
 
Microsoft Visual C++ MVP (from 2008 to present)
Website: http://rodream.net

Comments and Discussions

 
Generalgood job~! Pinmembernomark28-Jan-08 16:33 
GeneralRe: good job~! PinmemberKwon Yong Hwi29-Jan-08 0:25 

General General    News News    Suggestion Suggestion    Question Question    Bug Bug    Answer Answer    Joke Joke    Rant Rant    Admin Admin   

Use Ctrl+Left/Right to switch messages, Ctrl+Up/Down to switch threads, Ctrl+Shift+Left/Right to switch pages.

| Advertise | Privacy | Mobile
Web02 | 2.8.140916.1 | Last Updated 28 Jan 2008
Article Copyright 2008 by Yonghwi Kwon
Everything else Copyright © CodeProject, 1999-2014
Terms of Service
Layout: fixed | fluid