KVM: x86: Rework request for immediate exit

The VMX implementation of enable_irq_window raised
KVM_REQ_IMMEDIATE_EXIT after we checked it in vcpu_enter_guest. This
caused infinite loops on vmentry. Fix it by letting enable_irq_window
signal the need for an immediate exit via its return value and drop
KVM_REQ_IMMEDIATE_EXIT.

This issue only affects nested VMX scenarios.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Gleb Natapov <gleb@redhat.com>

3 files changed, 13 insertions, 12 deletions

diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c
index 15c9cccd716b..7f896cbe717f 100644
--- a/arch/x86/kvm/svm.c
+++ b/arch/x86/kvm/svm.c
@@ -3632,7 +3632,7 @@ static int svm_interrupt_allowed(struct kvm_vcpu *vcpu)
         return ret;
 }
 
-static void enable_irq_window(struct kvm_vcpu *vcpu)
+static int enable_irq_window(struct kvm_vcpu *vcpu)
 {
         struct vcpu_svm *svm = to_svm(vcpu);
 
@@ -3646,6 +3646,7 @@ static void enable_irq_window(struct kvm_vcpu *vcpu)
                 svm_set_vintr(svm);
                 svm_inject_irq(svm, 0x0);
         }
+        return 0;
 }
 
 static void enable_nmi_window(struct kvm_vcpu *vcpu)
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 0f0cb3110626..74c525e2c608 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -4398,22 +4398,23 @@ static bool nested_exit_on_nmi(struct kvm_vcpu *vcpu)
                 PIN_BASED_NMI_EXITING;
 }
 
-static void enable_irq_window(struct kvm_vcpu *vcpu)
+static int enable_irq_window(struct kvm_vcpu *vcpu)
 {
         u32 cpu_based_vm_exec_control;
-        if (is_guest_mode(vcpu) && nested_exit_on_intr(vcpu)) {
+
+        if (is_guest_mode(vcpu) && nested_exit_on_intr(vcpu))
                 /*
                  * We get here if vmx_interrupt_allowed() said we can't
-                 * inject to L1 now because L2 must run. Ask L2 to exit
-                 * right after entry, so we can inject to L1 more promptly.
+                 * inject to L1 now because L2 must run. The caller will have
+                 * to make L2 exit right after entry, so we can inject to L1
+                 * more promptly.
                  */
-                kvm_make_request(KVM_REQ_IMMEDIATE_EXIT, vcpu);
-                return;
-        }
+                return -EBUSY;
 
         cpu_based_vm_exec_control = vmcs_read32(CPU_BASED_VM_EXEC_CONTROL);
         cpu_based_vm_exec_control |= CPU_BASED_VIRTUAL_INTR_PENDING;
         vmcs_write32(CPU_BASED_VM_EXEC_CONTROL, cpu_based_vm_exec_control);
+        return 0;
 }
 
 static void enable_nmi_window(struct kvm_vcpu *vcpu)
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 2a434bf3918d..c522260b5bbf 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -5692,7 +5692,7 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
         int r;
         bool req_int_win = !irqchip_in_kernel(vcpu->kvm) &&
                 vcpu->run->request_interrupt_window;
-        bool req_immediate_exit = 0;
+        bool req_immediate_exit = false;
 
         if (vcpu->requests) {
                 if (kvm_check_request(KVM_REQ_MMU_RELOAD, vcpu))
@@ -5734,8 +5734,6 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
                         record_steal_time(vcpu);
                 if (kvm_check_request(KVM_REQ_NMI, vcpu))
                         process_nmi(vcpu);
-                req_immediate_exit =
-                        kvm_check_request(KVM_REQ_IMMEDIATE_EXIT, vcpu);
                 if (kvm_check_request(KVM_REQ_PMU, vcpu))
                         kvm_handle_pmu_event(vcpu);
                 if (kvm_check_request(KVM_REQ_PMI, vcpu))
@@ -5757,7 +5755,8 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
                 if (vcpu->arch.nmi_pending)
                         kvm_x86_ops->enable_nmi_window(vcpu);
                 else if (kvm_cpu_has_injectable_intr(vcpu) || req_int_win)
-                        kvm_x86_ops->enable_irq_window(vcpu);
+                        req_immediate_exit =
+                                kvm_x86_ops->enable_irq_window(vcpu) != 0;
 
                 if (kvm_lapic_enabled(vcpu)) {
                         /*