OK, I made a pass through all the code relating to various aspects of addressing so I could get a better overall picture. I've got to say, the mods to map the interrupt handler CS.base value, and mapping in the monitor handler pages such that we can run code in the kernel module pages while still in the monitor/guest context rocks!!! Nice job folks. That gives us some excellent potential to squeeze more performance out of this puppy, for doing some monitor processing without a context switch to/from the host. Man, if we eventually can move certain components from the device emulation into that space, that would kick butt! Getting ahead of myself. :^) OK, now the private copies of memset() and memcpy() make a lot of sense! Off to go make some graphics... -Kevin
