Re: [PATCH 2/4] arm/arm64: KVM: Use set/way op trapping to track the state of the caches

[Christoffer Dall <christoffer.dall@linaro.org>]

On Fri, Jan 09, 2015 at 11:38:19AM +0000, Marc Zyngier wrote:

[...]

> >> @@ -258,12 +272,24 @@ bool access_sctlr(struct kvm_vcpu *vcpu,
> >>                 const struct coproc_params *p,
> >>                 const struct coproc_reg *r)
> >>  {
> >> +     bool was_enabled = vcpu_has_cache_enabled(vcpu);
> >> +     bool now_enabled;
> >> +
> >>       access_vm_reg(vcpu, p, r);
> >>
> >> -     if (vcpu_has_cache_enabled(vcpu)) {     /* MMU+Caches enabled? */
> >> -             vcpu->arch.hcr &= ~HCR_TVM;
> >> +     now_enabled = vcpu_has_cache_enabled(vcpu);
> >> +
> >> +     /*
> >> +      * If switching the MMU+caches on, need to invalidate the caches.
> >> +      * If switching it off, need to clean the caches.
> >> +      * Clean + invalidate does the trick always.
> > 
> > couldn't a clean here then be writing bogus to RAM on an initial boot
> > of the guest?
> 
> It shouldn't. We're supposed to start a VM with a fully invalid cache
> (what's why we call coherent_cache_guest_page). And this is no different
> from what we had before, as an invalidate by set/way was upgraded to
> clean+invalidate.
> 

right, that makes sense.

Thanks,
-Christoffer