On Thu, Jul 14, 2011 at 04:26:39PM +0300, Avi Kivity wrote:
> On 07/14/2011 04:12 PM, Joerg Roedel wrote:
>> Makes sense. I'll probably remove the lazy allocation and initialize
>> both VMCBs at vcpu-creation time. The memory foodprint is the same as
>> before because the hsave area was also allocated at the beginning.
>
> Related, would we need a pool of n_vmcbs/vmcb02s?
Probably. This depends on how nested-svm will be used I think. It is not
very hard to add if really needed. Some kind of LRU is certainly needed
too then.
> I guess the condition for reusing an n_vmcb would be: same vmcb_gpa and
> at least one clean bit set?
Same vmcb_gpa is sufficient I think. I nothing is marked clean then it
is the same situation as if the vmcb_gpa is different.