[PATCH v3 08/18] KVM: arm64: Introduce hyp context

[Andrew Scull]

On Mon, Sep 07, 2020 at 02:29:11PM +0100, Marc Zyngier wrote:
> On Thu, 03 Sep 2020 14:52:57 +0100,
> Andrew Scull <ascull at google.com> wrote:
> > 
> > During __guest_enter, save and restore from a new hyp context rather
> > than the host context. This is preparation for separation of the hyp and
> > host context in nVHE.
> > 
> > Signed-off-by: Andrew Scull <ascull at google.com>
> > ---
> >  arch/arm64/include/asm/kvm_hyp.h        |  3 ++-
> >  arch/arm64/kernel/image-vars.h          |  1 +
> >  arch/arm64/kvm/arm.c                    | 10 ++++++++++
> >  arch/arm64/kvm/hyp/entry.S              | 10 +++++-----
> >  arch/arm64/kvm/hyp/include/hyp/switch.h |  2 +-
> >  arch/arm64/kvm/hyp/nvhe/switch.c        |  2 +-
> >  arch/arm64/kvm/hyp/vhe/switch.c         |  2 +-
> >  7 files changed, 21 insertions(+), 9 deletions(-)
> > 
> > diff --git a/arch/arm64/include/asm/kvm_hyp.h b/arch/arm64/include/asm/kvm_hyp.h
> > index 1e2491da324e..0b525e05e5bf 100644
> > --- a/arch/arm64/include/asm/kvm_hyp.h
> > +++ b/arch/arm64/include/asm/kvm_hyp.h
> > @@ -12,6 +12,7 @@
> >  #include <asm/alternative.h>
> >  #include <asm/sysreg.h>
> >  
> > +DECLARE_PER_CPU(struct kvm_cpu_context, kvm_hyp_ctxt);
> >  DECLARE_PER_CPU(unsigned long, kvm_hyp_vector);
> >  
> >  #define read_sysreg_elx(r,nvh,vh)					\
> > @@ -89,7 +90,7 @@ void activate_traps_vhe_load(struct kvm_vcpu *vcpu);
> >  void deactivate_traps_vhe_put(void);
> >  #endif
> >  
> > -u64 __guest_enter(struct kvm_vcpu *vcpu, struct kvm_cpu_context *host_ctxt);
> > +u64 __guest_enter(struct kvm_vcpu *vcpu);
> >  
> >  void __noreturn hyp_panic(void);
> >  #ifdef __KVM_NVHE_HYPERVISOR__
> > diff --git a/arch/arm64/kernel/image-vars.h b/arch/arm64/kernel/image-vars.h
> > index 54bb0eb34b0f..9f419e4fc66b 100644
> > --- a/arch/arm64/kernel/image-vars.h
> > +++ b/arch/arm64/kernel/image-vars.h
> > @@ -71,6 +71,7 @@ KVM_NVHE_ALIAS(kvm_update_va_mask);
> >  /* Global kernel state accessed by nVHE hyp code. */
> >  KVM_NVHE_ALIAS(arm64_ssbd_callback_required);
> >  KVM_NVHE_ALIAS(kvm_host_data);
> > +KVM_NVHE_ALIAS(kvm_hyp_ctxt);
> >  KVM_NVHE_ALIAS(kvm_hyp_vector);
> >  KVM_NVHE_ALIAS(kvm_vgic_global_state);
> >  
> > diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c
> > index b6442c6be5ad..ae4b34f91e94 100644
> > --- a/arch/arm64/kvm/arm.c
> > +++ b/arch/arm64/kvm/arm.c
> > @@ -47,6 +47,7 @@ __asm__(".arch_extension	virt");
> >  #endif
> >  
> >  DEFINE_PER_CPU(struct kvm_host_data, kvm_host_data);
> > +DEFINE_PER_CPU(struct kvm_cpu_context, kvm_hyp_ctxt);
> 
> [back to this patch after having reviewed a few of the subsequent
> ones]
> 
> Given the variety of contexts you are introducing, I wonder if the
> best course of action for most of this isn't simply to use the EL2
> stack rather than defining ad-hoc structures.
> 
> The host save/restore you are introducing in a later patch certainly
> could do without a separate data structure (hello, struct pt_regs) and
> the hyp/host churn.
> 
> What do you think?

We could define the start of the stack to be the host context (IIRC,
TF-A does something along those lines). Maybe there is some locality
benefit?

The percpu definitions become less onerous in code with David's percpu
series as the mapping to EL2 is done in bulk rather than per item.

Ptrauth switching is something that doesn't fall under pt_regs (it's no
longer in this series, but will need to be switched later on). I had
chosen to reuse the existing structs but a host-specilized context might
be preferred?