This is a follow up to an earlier post about COMException error messages. http://blogs.msdn.com/jaredpar/archive/2005/03/25/402386.aspx

In my previous post we examined the process by which an HRESULT is converted to a COMException by the CLR.  Several people have asked me about the reverse, how do error messages make their way from the CLR to COM?  They propagate in much the same fasion.  When an exception crosses a managed to native barrier, the CLR will return a failing HRESULT to the COM object that made the call.  The CLR will also support ISupportErrorInfo on the RCW so the original calling COM object can store the error information appropriately.