Sean Christopherson <seanjc@xxxxxxxxxx> writes:
> Explicitly zero the guest's CR3 and mark it available+dirty at RESET/INIT.
> Per Intel's SDM and AMD's APM, CR3 is zeroed at both RESET and INIT. For
> RESET, this is a nop as vcpu is zero-allocated. For INIT, the bug has
> likely escaped notice because no firmware/kernel puts its page tables root
> at PA=0, let alone relies on INIT to get the desired CR3 for such page
> tables.
>
> Cc: stable@xxxxxxxxxxxxxxx
> Signed-off-by: Sean Christopherson <seanjc@xxxxxxxxxx>
> ---
> arch/x86/kvm/x86.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index e77a5bf2d940..2cb38c67ed43 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -10899,6 +10899,9 @@ void kvm_vcpu_reset(struct kvm_vcpu *vcpu, bool init_event)
> kvm_set_rflags(vcpu, X86_EFLAGS_FIXED);
> kvm_rip_write(vcpu, 0xfff0);
>
> + vcpu->arch.cr3 = 0;
> + kvm_register_mark_dirty(vcpu, VCPU_EXREG_CR3);
kvm_register_mark_dirty() is redundant here as PATCH1 does
vcpu->arch.regs_avail = ~0;
vcpu->arch.regs_dirty = ~0;
just a few lines above. The dependency is, however, implicit and this
patch is marked for stable@ (well, PATCH1 has 8 Fixes: tags so I'd
expect it to get picked by everyone too, especially by robots) and
flipping two bits is cheap.
Reviewed-by: Vitaly Kuznetsov <vkuznets@xxxxxxxxxx>
> +
> /*
> * CR0.CD/NW are set on RESET, preserved on INIT. Note, some versions
> * of Intel's SDM list CD/NW as being set on INIT, but they contradict
--
Vitaly
