[PATCH] KVM: arm64: Don't access PMCR_EL0 when no PMU is available

[Alexandru Elisei]

Hi Marc,

On 12/10/20 8:30 AM, Marc Zyngier wrote:
> We reset the guest's view of PMCR_EL0 unconditionally, based on
> the host's view of this register. It is however legal for an
> imnplementation not to provide any PMU, resulting in an UNDEF.
>
> The obvious fix is to skip the reset of this shadow register
> when no PMU is available, sidestepping the issue entirely.
> If no PMU is available, the guest is not able to request
> a virtual PMU anyway, so not doing nothing is the right thing
> to do!
>
> It is unlikely that this bug can hit any HW implementation
> though, as they all provide a PMU. It has been found using nested
> virt with the host KVM not implementing the PMU itself.
>
> Fixes: ab9468340d2bc ("arm64: KVM: Add access handler for PMCR register")
> Signed-off-by: Marc Zyngier <maz at kernel.org>
> ---
>  arch/arm64/kvm/sys_regs.c | 4 ++++
>  1 file changed, 4 insertions(+)
>
> diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
> index bc15246775d0..6c64d010102b 100644
> --- a/arch/arm64/kvm/sys_regs.c
> +++ b/arch/arm64/kvm/sys_regs.c
> @@ -923,6 +923,10 @@ static void reset_pmcr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *r)
>  {
>  	u64 pmcr, val;
>  
> +	/* No PMU available, PMCR_EL0 may UNDEF... */
> +	if (!kvm_arm_support_pmu_v3())
> +		return;
> +

reset_pmcr() is called from kvm_reset_vcpu()->kvm_reset_sys_regs(). Before calling
kvm_reset_sys_regs(), kvm_reset_vcpu() returns -EINVAL if the VCPU has the PMUv3
feature but the host doesn't have a PMU.

It looks to me like the undef can happen only when the VCPU feature isn't set and
the hardware doesn't have a PMU. How about we change the test to check for
kvm_vcpu_has_pmu() to avoid executing the extra instructions, which are not needed
because the VM won't have a PMU?

On the other hand, kvm_pmu_vcpu_reset() is happy to do the reset even if the VCPU
feature isn't set, the host doesn't support a PMU, and before PMCR_EL0 is
initialized. It's up to you, the kvm_arm_support_pmu_v3() check looks consistent
with how PMU reset is handled.

Thanks,
Alex

>  	pmcr = read_sysreg(pmcr_el0);
>  	/*
>  	 * Writable bits of PMCR_EL0 (ARMV8_PMU_PMCR_MASK) are reset to UNKNOWN