KVM: x86: Unify TSC logic

Move the TSC control logic from the vendor backends into x86.c
by adding adjust_tsc_offset to x86 ops.  Now all TSC decisions
can be done in one place.

Signed-off-by: Zachary Amsden <zamsden@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

arch/x86/include/asm/kvm_host.h

@@ -255,7 +255,6 @@ struct kvm_mmu {
 };
 
 struct kvm_vcpu_arch {
-	u64 host_tsc;
 	/*
 	 * rip and regs accesses must go through
 	 * kvm_{register,rip}_{read,write} functions.
@@ -336,9 +335,10 @@ struct kvm_vcpu_arch {
 
 	gpa_t time;
 	struct pvclock_vcpu_time_info hv_clock;
-	unsigned int hv_clock_tsc_khz;
+	unsigned int hw_tsc_khz;
 	unsigned int time_offset;
 	struct page *time_page;
+	u64 last_host_tsc;
 
 	bool nmi_pending;
 	bool nmi_injected;
@@ -520,6 +520,7 @@ struct kvm_x86_ops {
 	u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
 	int (*get_lpage_level)(void);
 	bool (*rdtscp_supported)(void);
+	void (*adjust_tsc_offset)(struct kvm_vcpu *vcpu, s64 adjustment);
 
 	void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
 

arch/x86/kvm/svm.c

@@ -715,6 +715,15 @@ static void svm_write_tsc_offset(struct kvm_vcpu *vcpu, u64 offset)
 	svm->vmcb->control.tsc_offset = offset + g_tsc_offset;
 }
 
+static void svm_adjust_tsc_offset(struct kvm_vcpu *vcpu, s64 adjustment)
+{
+	struct vcpu_svm *svm = to_svm(vcpu);
+
+	svm->vmcb->control.tsc_offset += adjustment;
+	if (is_nested(svm))
+		svm->nested.hsave->control.tsc_offset += adjustment;
+}
+
 static void init_vmcb(struct vcpu_svm *svm)
 {
 	struct vmcb_control_area *control = &svm->vmcb->control;
@@ -961,20 +970,6 @@ static void svm_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 	int i;
 
 	if (unlikely(cpu != vcpu->cpu)) {
-		u64 delta;
-
-		if (check_tsc_unstable()) {
-			/*
-			 * Make sure that the guest sees a monotonically
-			 * increasing TSC.
-			 */
-			delta = vcpu->arch.host_tsc - native_read_tsc();
-			svm->vmcb->control.tsc_offset += delta;
-			if (is_nested(svm))
-				svm->nested.hsave->control.tsc_offset += delta;
-		}
-		vcpu->cpu = cpu;
-		kvm_migrate_timers(vcpu);
 		svm->asid_generation = 0;
 	}
 
@@ -990,8 +985,6 @@ static void svm_vcpu_put(struct kvm_vcpu *vcpu)
 	++vcpu->stat.host_state_reload;
 	for (i = 0; i < NR_HOST_SAVE_USER_MSRS; i++)
 		wrmsrl(host_save_user_msrs[i], svm->host_user_msrs[i]);
-
-	vcpu->arch.host_tsc = native_read_tsc();
 }
 
 static unsigned long svm_get_rflags(struct kvm_vcpu *vcpu)
@@ -3553,6 +3546,7 @@ static struct kvm_x86_ops svm_x86_ops = {
 	.has_wbinvd_exit = svm_has_wbinvd_exit,
 
 	.write_tsc_offset = svm_write_tsc_offset,
+	.adjust_tsc_offset = svm_adjust_tsc_offset,
 };
 
 static int __init svm_init(void)

arch/x86/kvm/vmx.c

@@ -505,7 +505,6 @@ static void __vcpu_clear(void *arg)
 		vmcs_clear(vmx->vmcs);
 	if (per_cpu(current_vmcs, cpu) == vmx->vmcs)
 		per_cpu(current_vmcs, cpu) = NULL;
-	rdtscll(vmx->vcpu.arch.host_tsc);
 	list_del(&vmx->local_vcpus_link);
 	vmx->vcpu.cpu = -1;
 	vmx->launched = 0;
@@ -881,7 +880,6 @@ static void vmx_load_host_state(struct vcpu_vmx *vmx)
 static void vmx_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 {
 	struct vcpu_vmx *vmx = to_vmx(vcpu);
-	u64 tsc_this, delta, new_offset;
 	u64 phys_addr = __pa(per_cpu(vmxarea, cpu));
 
 	if (!vmm_exclusive)
@@ -898,14 +896,12 @@ static void vmx_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 		struct desc_ptr *gdt = &__get_cpu_var(host_gdt);
 		unsigned long sysenter_esp;
 
-		kvm_migrate_timers(vcpu);
 		kvm_make_request(KVM_REQ_TLB_FLUSH, vcpu);
 		local_irq_disable();
 		list_add(&vmx->local_vcpus_link,
 			 &per_cpu(vcpus_on_cpu, cpu));
 		local_irq_enable();
 
-		vcpu->cpu = cpu;
 		/*
 		 * Linux uses per-cpu TSS and GDT, so set these when switching
 		 * processors.
@@ -915,16 +911,6 @@ static void vmx_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 
 		rdmsrl(MSR_IA32_SYSENTER_ESP, sysenter_esp);
 		vmcs_writel(HOST_IA32_SYSENTER_ESP, sysenter_esp); /* 22.2.3 */
-
-		/*
-		 * Make sure the time stamp counter is monotonous.
-		 */
-		rdtscll(tsc_this);
-		if (tsc_this < vcpu->arch.host_tsc) {
-			delta = vcpu->arch.host_tsc - tsc_this;
-			new_offset = vmcs_read64(TSC_OFFSET) + delta;
-			vmcs_write64(TSC_OFFSET, new_offset);
-		}
 	}
 }
 
@@ -1153,6 +1139,12 @@ static void vmx_write_tsc_offset(struct kvm_vcpu *vcpu, u64 offset)
 	vmcs_write64(TSC_OFFSET, offset);
 }
 
+static void vmx_adjust_tsc_offset(struct kvm_vcpu *vcpu, s64 adjustment)
+{
+	u64 offset = vmcs_read64(TSC_OFFSET);
+	vmcs_write64(TSC_OFFSET, offset + adjustment);
+}
+
 /*
  * Reads an msr value (of 'msr_index') into 'pdata'.
  * Returns 0 on success, non-0 otherwise.
@@ -4108,6 +4100,7 @@ static struct kvm_vcpu *vmx_create_vcpu(struct kvm *kvm, unsigned int id)
 
 	cpu = get_cpu();
 	vmx_vcpu_load(&vmx->vcpu, cpu);
+	vmx->vcpu.cpu = cpu;
 	err = vmx_vcpu_setup(vmx);
 	vmx_vcpu_put(&vmx->vcpu);
 	put_cpu();
@@ -4347,6 +4340,7 @@ static struct kvm_x86_ops vmx_x86_ops = {
 	.has_wbinvd_exit = cpu_has_vmx_wbinvd_exit,
 
 	.write_tsc_offset = vmx_write_tsc_offset,
+	.adjust_tsc_offset = vmx_adjust_tsc_offset,
 };
 
 static int __init vmx_init(void)

arch/x86/kvm/x86.c

@@ -973,9 +973,9 @@ static int kvm_write_guest_time(struct kvm_vcpu *v)
 		return 1;
 	}
 
-	if (unlikely(vcpu->hv_clock_tsc_khz != this_tsc_khz)) {
+	if (unlikely(vcpu->hw_tsc_khz != this_tsc_khz)) {
 		kvm_set_time_scale(this_tsc_khz, &vcpu->hv_clock);
-		vcpu->hv_clock_tsc_khz = this_tsc_khz;
+		vcpu->hw_tsc_khz = this_tsc_khz;
 	}
 
 	/* With all the info we got, fill in the values */
@@ -1866,13 +1866,24 @@ void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 	}
 
 	kvm_x86_ops->vcpu_load(vcpu, cpu);
-	kvm_request_guest_time_update(vcpu);
+	if (unlikely(vcpu->cpu != cpu)) {
+		/* Make sure TSC doesn't go backwards */
+		s64 tsc_delta = !vcpu->arch.last_host_tsc ? 0 :
+				native_read_tsc() - vcpu->arch.last_host_tsc;
+		if (tsc_delta < 0)
+			mark_tsc_unstable("KVM discovered backwards TSC");
+		if (check_tsc_unstable())
+			kvm_x86_ops->adjust_tsc_offset(vcpu, -tsc_delta);
+		kvm_migrate_timers(vcpu);
+		vcpu->cpu = cpu;
+	}
 }
 
 void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu)
 {
 	kvm_x86_ops->vcpu_put(vcpu);
 	kvm_put_guest_fpu(vcpu);
+	vcpu->arch.last_host_tsc = native_read_tsc();
 }
 
 static int is_efer_nx(void)