Re: [PATCH] KVM: MMU: Do not unconditionally read PDPTE from guest memory

["Roedel, Joerg" <Joerg.Roedel@amd.com>]

On Tue, Aug 02, 2011 at 08:34:32AM -0400, Avi Kivity wrote:
> On 08/02/2011 12:31 PM, Roedel, Joerg wrote:
> > On Sun, Jul 31, 2011 at 04:08:44AM -0400, Avi Kivity wrote:
> > >  On 07/29/2011 02:31 PM, Roedel, Joerg wrote:
> > >  >  On Thu, Jul 28, 2011 at 04:36:17AM -0400, Avi Kivity wrote:
> > >  >  >   Architecturally, PDPTEs are cached in the PDPTRs when CR3 is reloaded.
> > >  >  >   On SVM, it is not possible to implement this, but on VMX this is possible
> > >  >  >   and was indeed implemented until nested SVM changed this to unconditionally
> > >  >  >   read PDPTEs dynamically.  This has noticable impact when running PAE guests.
> > >  >  >
> > >  >  >   Fix by changing the MMU to read PDPTRs from the cache, falling back to
> > >  >  >   reading from memory for the nested MMU.
> > >  >  >
> > >  >  >   Signed-off-by: Avi Kivity<avi@redhat.com>
> > >  >
> > >  >  Hmm, interesting. Sorry for breaking it. I tested the patch on nested
> > >  >  svm, it works fine.
> > >
> > >  Does pae-on-pae work for you?
> >
> > Only tested pae-on-longmode. I'll see if I can find my 32bit
> > installation again and test this too.
> 
> I wanted to test it since any mixup in where the PTPTRs were taken from 
> would be readily apparent.  But it crashes even without the patch.

Hmm, it works here on a Phenom II X6 box with and without the patch.
Host was your master-branch with pae enabled. The l1-guest was a vanilla
3.0-pae kernel and the l2-guest ran with 3.0-pae and 3.0-nonpae kernels.


	Joerg

-- 
AMD Operating System Research Center

Advanced Micro Devices GmbH Einsteinring 24 85609 Dornach
General Managers: Alberto Bozzo, Andrew Bowd
Registration: Dornach, Landkr. Muenchen; Registerger. Muenchen, HRB Nr. 43632