Good link...What I am trying to do is slightly different though. In the post, the event sink is the .NET application it self. In my case, the event sink is another COM object and this sink COM object is instantiated from .NET application.

Thanks a lot! The specific section you gave a link for does look like what I am trying to do. Thanks :)

The sink callback happens inside the COM object. In other words, when sink establishes connection with connectionable object and then expects event from COM server, the execution control is still inside the COM sink. So the instance is alive while this happens. Could there be anything on .NET security side that I might be missing?

Hello, I have an Win32 COM application with GUI written in C++ (MFC). This GUI invokes a ATL COM client/DLL that acts as a sink to a ATL COM server. Everything works well. So we have 3 components here- MFC GUI app, COM client/sink and COM event server. Now, I need to replace this native MFC C++ GUI application with .NET application. So I imported the COM dll that acts as sink into my .NET application and all interfaces were available as expected. When I make a specific method call in this COM dll/object, this method establishes sink connection with COM server (using AtlAdvise) without any issues. However, when the COM server raises an event, the call reaches the sink, but it fails in InvokeFromFuncInfo (specifically it DispCallFunc fails) from ATLCOM.h giving Access Violation error (0xC0000005). So to represent this: 1. .NET GUI ---create instance--->>> COM DLL (SINK) ---establish connection --->>>> COM Event Server 2. COM Event Server ---raise event--->>>ATLCOM.h (in COM sink DLL) fails in DispCallFunc with Access Violation (0xC0000005). The above flow works perfectly well with my MFC GUI. Is there anything required for this connectionable object or sink that I may not be setting in my .NET appliaction? How can I at least delve deeper to see what is going on? thanks! Vikram