Ulrich Weigand wrote:
> Of course. That's why I've tried to encapsulate all access
> to guest memory into routines (read/write_guest_dword etc.).
> The page-present check should need to be added only there ...
>
> (While we're at it, we might also fix the code to actually
> respect guest segment registers ;-/ Then again, maybe that's
> not a priority for now ...)
OK, this was a good idea. Though, we need to enhance
this by passing return codes back from routines which may
access memory, and always checking the return code.
If the page isn't present etc, we need to stop emulating
the instruction and return to the kernel module to
deal with the page situation.
I just started looking at re-writing the main emulation()
function to have it use the same instruction decode function
I use for prescanning, as well.
-Kevin
