On Tue, 13 Aug 2019 11:08:39 +0200, Paolo Bonzini <pbonzini@xxxxxxxxxx> wrote:
> On 09/08/19 18:00, Adalbert Lazăr wrote:
> > It should complete the commit fd34a9518173 ("kvm: x86: consult the page tracking from kvm_mmu_get_page() and __direct_map()")
> >
> > Signed-off-by: Adalbert Lazăr <alazar@xxxxxxxxxxxxxxx>
> > ---
> > arch/x86/kvm/mmu.c | 3 +++
> > 1 file changed, 3 insertions(+)
> >
> > diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
> > index 65b6acba82da..fd64cf1115da 100644
> > --- a/arch/x86/kvm/mmu.c
> > +++ b/arch/x86/kvm/mmu.c
> > @@ -2660,6 +2660,9 @@ static void clear_sp_write_flooding_count(u64 *spte)
> > static unsigned int kvm_mmu_page_track_acc(struct kvm_vcpu *vcpu, gfn_t gfn,
> > unsigned int acc)
> > {
> > + if (!kvmi_tracked_gfn(vcpu, gfn))
> > + return acc;
> > +
> > if (kvm_page_track_is_active(vcpu, gfn, KVM_PAGE_TRACK_PREREAD))
> > acc &= ~ACC_USER_MASK;
> > if (kvm_page_track_is_active(vcpu, gfn, KVM_PAGE_TRACK_PREWRITE) ||
> >
>
> If this patch is always needed, then the function should be named
> something like kvm_mmu_apply_introspection_access and kvmi_tracked_gfn
> should be tested from the moment it is introduced.
>
> But the commit message says nothing about _why_ it is needed, so I
> cannot guess. I would very much avoid it however. Is it just an
> optimization?
>
> Paolo
We'll retest to see if we still need kvm_mmu_page_track_acc().
The kvmi_tracked_gfn() check was used to keep the KVM code flow
"unchanged" as much as possible. Probably, we can get ride of it.
