> how do you tell where the fault happened, in an SRC or in a DEST access?
Easy, by the address of the faulting instruction. You can have two
different exception handlers, one pointed to by faults on src, one
pointed to by faults on dst. I think I lost track of why you would
want to do this, though. Is it just to detect "this shouldn't happen"
errors in the kernel?
Tom.