KVM: Fix unneeded instruction skipping during task switching.

There is no need to skip instruction if the reason for a task switch
is a task gate in IDT and access to it is caused by an external even.
The problem  is currently solved only for VMX since there is no reliable
way to skip an instruction in SVM. We should emulate it instead.
Signed-off-by: NGleb Natapov <gleb@redhat.com>
Signed-off-by: NAvi Kivity <avi@redhat.com>

arch/x86/include/asm/svm.h

@@ -225,6 +225,7 @@ struct __attribute__ ((__packed__)) vmcb {
 #define SVM_EVTINJ_VALID_ERR (1 << 11)
 
 #define SVM_EXITINTINFO_VEC_MASK SVM_EVTINJ_VEC_MASK
+#define SVM_EXITINTINFO_TYPE_MASK SVM_EVTINJ_TYPE_MASK
 
 #define	SVM_EXITINTINFO_TYPE_INTR SVM_EVTINJ_TYPE_INTR
 #define	SVM_EXITINTINFO_TYPE_NMI SVM_EVTINJ_TYPE_NMI

arch/x86/kvm/svm.c

@@ -1825,17 +1825,28 @@ static int task_switch_interception(struct vcpu_svm *svm,
 				    struct kvm_run *kvm_run)
 {
 	u16 tss_selector;
+	int reason;
+	int int_type = svm->vmcb->control.exit_int_info &
+		SVM_EXITINTINFO_TYPE_MASK;
 
 	tss_selector = (u16)svm->vmcb->control.exit_info_1;
+
 	if (svm->vmcb->control.exit_info_2 &
 	    (1ULL << SVM_EXITINFOSHIFT_TS_REASON_IRET))
-		return kvm_task_switch(&svm->vcpu, tss_selector,
-				       TASK_SWITCH_IRET);
-	if (svm->vmcb->control.exit_info_2 &
-	    (1ULL << SVM_EXITINFOSHIFT_TS_REASON_JMP))
-		return kvm_task_switch(&svm->vcpu, tss_selector,
-				       TASK_SWITCH_JMP);
-	return kvm_task_switch(&svm->vcpu, tss_selector, TASK_SWITCH_CALL);
+		reason = TASK_SWITCH_IRET;
+	else if (svm->vmcb->control.exit_info_2 &
+		 (1ULL << SVM_EXITINFOSHIFT_TS_REASON_JMP))
+		reason = TASK_SWITCH_JMP;
+	else if (svm->vmcb->control.exit_int_info & SVM_EXITINTINFO_VALID)
+		reason = TASK_SWITCH_GATE;
+	else
+		reason = TASK_SWITCH_CALL;
+
+
+	if (reason != TASK_SWITCH_GATE || int_type == SVM_EXITINTINFO_TYPE_SOFT)
+		skip_emulated_instruction(&svm->vcpu);
+
+	return kvm_task_switch(&svm->vcpu, tss_selector, reason);
 }
 
 static int cpuid_interception(struct vcpu_svm *svm, struct kvm_run *kvm_run)

arch/x86/kvm/vmx.c

@@ -3038,22 +3038,40 @@ static int handle_task_switch(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
 	struct vcpu_vmx *vmx = to_vmx(vcpu);
 	unsigned long exit_qualification;
 	u16 tss_selector;
-	int reason;
+	int reason, type, idt_v;
+
+	idt_v = (vmx->idt_vectoring_info & VECTORING_INFO_VALID_MASK);
+	type = (vmx->idt_vectoring_info & VECTORING_INFO_TYPE_MASK);
 
 	exit_qualification = vmcs_readl(EXIT_QUALIFICATION);
 
 	reason = (u32)exit_qualification >> 30;
-	if (reason == TASK_SWITCH_GATE && vmx->vcpu.arch.nmi_injected &&
-	    (vmx->idt_vectoring_info & VECTORING_INFO_VALID_MASK) &&
-	    (vmx->idt_vectoring_info & VECTORING_INFO_TYPE_MASK)
-	    == INTR_TYPE_NMI_INTR) {
-		vcpu->arch.nmi_injected = false;
-		if (cpu_has_virtual_nmis())
-			vmcs_set_bits(GUEST_INTERRUPTIBILITY_INFO,
-				      GUEST_INTR_STATE_NMI);
+	if (reason == TASK_SWITCH_GATE && idt_v) {
+		switch (type) {
+		case INTR_TYPE_NMI_INTR:
+			vcpu->arch.nmi_injected = false;
+			if (cpu_has_virtual_nmis())
+				vmcs_set_bits(GUEST_INTERRUPTIBILITY_INFO,
+					      GUEST_INTR_STATE_NMI);
+			break;
+		case INTR_TYPE_EXT_INTR:
+			kvm_clear_interrupt_queue(vcpu);
+			break;
+		case INTR_TYPE_HARD_EXCEPTION:
+		case INTR_TYPE_SOFT_EXCEPTION:
+			kvm_clear_exception_queue(vcpu);
+			break;
+		default:
+			break;
+		}
 	}
 	tss_selector = exit_qualification;
 
+	if (!idt_v || (type != INTR_TYPE_HARD_EXCEPTION &&
+		       type != INTR_TYPE_EXT_INTR &&
+		       type != INTR_TYPE_NMI_INTR))
+		skip_emulated_instruction(vcpu);
+
 	if (!kvm_task_switch(vcpu, tss_selector, reason))
 		return 0;
 
@@ -3306,7 +3324,7 @@ static void vmx_complete_interrupts(struct vcpu_vmx *vmx)
 	vector = idt_vectoring_info & VECTORING_INFO_VECTOR_MASK;
 	type = idt_vectoring_info & VECTORING_INFO_TYPE_MASK;
 
-	switch(type) {
+	switch (type) {
 	case INTR_TYPE_NMI_INTR:
 		vmx->vcpu.arch.nmi_injected = true;
 		/*

arch/x86/kvm/x86.c

@@ -3935,7 +3935,10 @@ int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int reason)
 		kvm_x86_ops->set_rflags(vcpu, eflags & ~X86_EFLAGS_NT);
 	}
 
-	kvm_x86_ops->skip_emulated_instruction(vcpu);
+	/* set back link to prev task only if NT bit is set in eflags
+	   note that old_tss_sel is not used afetr this point */
+	if (reason != TASK_SWITCH_CALL && reason != TASK_SWITCH_GATE)
+		old_tss_sel = 0xffff;
 
 	/* set back link to prev task only if NT bit is set in eflags
 	   note that old_tss_sel is not used afetr this point */