On Mon, Sep 27, 2010 at 12:22:13PM -0400, Avi Kivity wrote:
> On 09/27/2010 05:40 PM, Roedel, Joerg wrote:
> > On Mon, Sep 27, 2010 at 11:02:23AM -0400, Avi Kivity wrote:
> > > On 09/27/2010 04:58 PM, Avi Kivity wrote:
> > > > On 09/27/2010 03:16 PM, Joerg Roedel wrote:
> > > >> This patch adds the svm cpuid feature flags to the qemu
> > > >> intialization path. It also adds the svm features available
> > > >> on phenom to its cpu-definition and extends the host cpu
> > > >> type to support all svm features KVM can provide.
> > > >>
> > > >
> > > > This should really be "all svm features" since we'll later mask
them
> > > > with kvm capabilities. This way, if we later add more
capabilities,
> > > > we automatically gain support in userspace.
> >
> > Yes, that is what -cpu host does with these patches applied. The
> > svm_features variable is set to -1 in this case and masked out later
> > with the KVM capabilities.
> >
>
> Well, running current uq/master I get:
>
> has_svm: can't execute cpuid 0x8000000a
> kvm: no hardware support
>
> ?
Weird, it worked here as I tested it. I had it on qemu/master and with
all three patches. But patch 1 should not make the difference. I take a
look, have you pushed the failing uq/master?