RE: [Xen-devel] [PATCH] This patch fixes several issues related tovmxassist

["Li, Xin B" <xin.b.li@xxxxxxxxx>]

>>Well, it's impossible to know for certain without looking at hidden 
>>descriptor state isn't it. Big real mode is only possible because the 
>>hidden state can be out of sync with the current execution mode.
>>
>>If that state can be made available to the address() function 
>>then it's 
>>implementation, and confidence in its correctness, becomes 
>trivial. Do 
>>we have that state available directly in the guest at that point? Or 
>>would we need to grab the state from Xen?
>>
>
>Keir, do you think this one seems better?
>When in real mode, we should always calculate address as ((seg 
>& 0xFFFF)
><< 4) + off, big real mode obey the same rule.
>When in real mode to protected mode transition, only the IP address
>calculation should use real mode addressing.
>-Xin

At this time, I'd like to push this patch in, since it really fixes
quite a few bugs, anyway, we still need revisit it later, to be
watertight is hard for now :-(
-Xin

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel