KVM: lapic: adjust preemption timer correctly when goes TSC backward

TSC_OFFSET will be adjusted if discovers TSC backward during vCPU load.
The preemption timer, which relies on the guest tsc to reprogram its
preemption timer value, is also reprogrammed if vCPU is scheded in to
a different pCPU. However, the current implementation reprogram preemption
timer before TSC_OFFSET is adjusted to the right value, resulting in the
preemption timer firing prematurely.

This patch fix it by adjusting TSC_OFFSET before reprogramming preemption
timer if TSC backward.

Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krċmář <rkrcmar@redhat.com>
Cc: Yunhong Jiang <yunhong.jiang@intel.com>
Signed-off-by: Wanpeng Li <wanpeng.li@hotmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 19f9f9e05c2a812fd07d07ce0e69223ec2accc25..699f8726539a2044616b82bd9cf792a8617f8001 100644 (file)
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -2743,16 +2743,16 @@ void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
                if (tsc_delta < 0)
                        mark_tsc_unstable("KVM discovered backwards TSC");
 
-               if (kvm_lapic_hv_timer_in_use(vcpu) &&
-                               kvm_x86_ops->set_hv_timer(vcpu,
-                                       kvm_get_lapic_tscdeadline_msr(vcpu)))
-                       kvm_lapic_switch_to_sw_timer(vcpu);
                if (check_tsc_unstable()) {
                        u64 offset = kvm_compute_tsc_offset(vcpu,
                                                vcpu->arch.last_guest_tsc);
                        kvm_x86_ops->write_tsc_offset(vcpu, offset);
                        vcpu->arch.tsc_catchup = 1;
                }
+               if (kvm_lapic_hv_timer_in_use(vcpu) &&
+                               kvm_x86_ops->set_hv_timer(vcpu,
+                                       kvm_get_lapic_tscdeadline_msr(vcpu)))
+                       kvm_lapic_switch_to_sw_timer(vcpu);
                /*
                 * On a host with synchronized TSC, there is no need to update
                 * kvmclock on vcpu->cpu migration