One problem with debugging exceptions from a JIT attached debugger is that the call stack usually shows the flow through the OS exception processing code, rather than the call stack exactly where the exception occurred.

I've long known that I could pull an EIP out of the exception record structure to get the real address of interest. However, Mike Stall's blog takes the whole process even further. 

If I were actively writing books or articles, this is exactly the kind of cool stuff I'd be covering.