On Tue, Jul 18, 2017 at 11:59:02AM -0500, Jintack Lim wrote:
> Forward the EL1 virtual memory register traps to the virtual EL2 if they
> are not coming from the virtual EL2 and the virtual HCR_EL2.TVM or TRVM
> bit is set.
I noticed that all these recursive patches don't change how we program
the physical HCR_EL2. Is that because we always respect the guest
hypervisor's configuration of the virtual HCR_EL2 into the physical one
when running the VM?
If so, perhaps we should add a single sentence in the commit messages
about that.
>
> This is for recursive nested virtualization.
>
> Signed-off-by: Jintack Lim <jintack.lim@xxxxxxxxxx>
> ---
> arch/arm64/kvm/sys_regs.c | 24 ++++++++++++++++++++++++
> 1 file changed, 24 insertions(+)
>
> diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
> index 3559cf7..3e4ec5e 100644
> --- a/arch/arm64/kvm/sys_regs.c
> +++ b/arch/arm64/kvm/sys_regs.c
> @@ -135,6 +135,27 @@ static inline bool el12_reg(struct sys_reg_params *p)
> return (p->Op1 == 5);
> }
>
> +/* This function is to support the recursive nested virtualization */
it's just 'recursive nested virtualization', not 'the recursive nested
virtualization', and I also think 'recursive virtualization' is
sufficient.
> +static bool forward_vm_traps(struct kvm_vcpu *vcpu, struct sys_reg_params *p)
> +{
> + u64 hcr_el2 = vcpu_sys_reg(vcpu, HCR_EL2);
> +
> + /* If a trap comes from the virtual EL2, the host hypervisor handles. */
> + if (vcpu_mode_el2(vcpu))
> + return false;
> +
> + /*
> + * If the virtual HCR_EL2.TVM or TRVM bit is set, we need to foward
> + * this trap to the virtual EL2.
> + */
> + if ((hcr_el2 & HCR_TVM) && p->is_write)
> + return true;
> + else if ((hcr_el2 & HCR_TRVM) && !p->is_write)
> + return true;
> +
> + return false;
> +}
> +
> /*
> * Generic accessor for VM registers. Only called as long as HCR_TVM
> * is set. If the guest enables the MMU, we stop trapping the VM
> @@ -152,6 +173,9 @@ static bool access_vm_reg(struct kvm_vcpu *vcpu,
> if (el12_reg(p) && forward_nv_traps(vcpu))
> return kvm_inject_nested_sync(vcpu, kvm_vcpu_get_hsr(vcpu));
>
> + if (!el12_reg(p) && forward_vm_traps(vcpu, p))
> + return kvm_inject_nested_sync(vcpu, kvm_vcpu_get_hsr(vcpu));
why do you need the !el12_reg(p) check here?
> +
> /*
> * Redirect EL1 register accesses to the corresponding EL2 registers if
> * they are meant to access EL2 registers.
> --
> 1.9.1
>
Thanks,
-Christoffer
