Use the "raw" helper to read the vAPIC register after an APIC-write trap
VM-Exit. Hardware is responsible for vetting the write, and the caller
is responsible for sanitizing the offset. This is a functional change,
as it means KVM will consume whatever happens to be in the vAPIC page if
the write was dropped by hardware. But, unless userspace deliberately
wrote garbage into the vAPIC page via KVM_SET_LAPIC, the value should be
zero since it's not writable by the guest.
This aligns common x86 with SVM's AVIC logic, i.e. paves the way for
using the nodecode path to handle APIC-write traps when AVIC is enabled.
Signed-off-by: Sean Christopherson <seanjc@xxxxxxxxxx>
---
arch/x86/kvm/lapic.c | 4 +---
1 file changed, 1 insertion(+), 3 deletions(-)
diff --git a/arch/x86/kvm/lapic.c b/arch/x86/kvm/lapic.c
index fbce455a9d17..2c88815657a9 100644
--- a/arch/x86/kvm/lapic.c
+++ b/arch/x86/kvm/lapic.c
@@ -2186,9 +2186,7 @@ EXPORT_SYMBOL_GPL(kvm_lapic_set_eoi);
/* emulate APIC access in a trap manner */
void kvm_apic_write_nodecode(struct kvm_vcpu *vcpu, u32 offset)
{
- u32 val = 0;
-
- kvm_lapic_reg_read(vcpu->arch.apic, offset, 4, &val);
+ u32 val = kvm_lapic_get_reg(vcpu->arch.apic, offset);
/* TODO: optimize to just emulate side effect w/o one more write */
kvm_lapic_reg_write(vcpu->arch.apic, offset, val);
--
2.35.0.263.gb82422642f-goog
