diff --git a/sys/amd64/include/vmm.h b/sys/amd64/include/vmm.h
--- a/sys/amd64/include/vmm.h
+++ b/sys/amd64/include/vmm.h
@@ -329,7 +329,9 @@
 cpuset_t vm_active_cpus(struct vm *vm);
 cpuset_t vm_debug_cpus(struct vm *vm);
 cpuset_t vm_suspended_cpus(struct vm *vm);
+bool vm_start_single_cpu(struct vm *vm, struct vcpu *vcpu);
 cpuset_t vm_start_cpus(struct vm *vm, const cpuset_t *tostart);
+void vm_await_start_single_cpu(struct vm *vm, struct vcpu *vcpu);
 void vm_await_start(struct vm *vm, const cpuset_t *waiting);
 #endif	/* _SYS__CPUSET_H_ */
 
diff --git a/sys/amd64/vmm/io/vlapic.c b/sys/amd64/vmm/io/vlapic.c
--- a/sys/amd64/vmm/io/vlapic.c
+++ b/sys/amd64/vmm/io/vlapic.c
@@ -1128,8 +1128,9 @@
 				break;
 
 			/* vCPU i is waiting for SIPI. */
-			CPU_SETOF(i, &dmask);
-			vm_await_start(vlapic->vm, &dmask);
+			vm_await_start_single_cpu(vlapic->vm,
+			    vm_vcpu(vlapic->vm, i));
+
 			break;
 		}
 
@@ -1152,10 +1153,10 @@
 			/*
 			 * Ignore SIPIs in any state other than wait-for-SIPI
 			 */
-			CPU_SETOF(i, &dmask);
-			dmask = vm_start_cpus(vlapic->vm, &dmask);
-			if (CPU_EMPTY(&dmask))
+			if (vm_start_single_cpu(vlapic->vm,
+				vm_vcpu(vlapic->vm, i))) {
 				break;
+			}
 
 			vmexit = vm_exitinfo(vlapic->vcpu);
 			vmexit->exitcode = VM_EXITCODE_SPINUP_AP;
@@ -1166,10 +1167,7 @@
 			break;
 		}
 
-		/*
-		 * Ignore SIPIs in any state other than wait-for-SIPI
-		 */
-		ipimask = vm_start_cpus(vlapic->vm, &dmask);
+		ipimask = dmask;
 		break;
 	default:
 		return (1);
@@ -1207,6 +1205,13 @@
 		vm_await_start(vcpu_vm(vcpu), &vme->u.ipi.dmask);
 		break;
 	case APIC_DELMODE_STARTUP:
+		/*
+		 * Ignore SIPIs in any state other than wait-for-SIPI
+		 */
+		vme->u.ipi.dmask = vm_start_cpus(vcpu_vm(vcpu), &vme->u.ipi.dmask);
+		if (CPU_EMPTY(&vme->u.ipi.dmask)) {
+			*retu = false;
+		}
 		break;
 	default:
 		return (1);
diff --git a/sys/amd64/vmm/vmm.c b/sys/amd64/vmm/vmm.c
--- a/sys/amd64/vmm/vmm.c
+++ b/sys/amd64/vmm/vmm.c
@@ -2496,6 +2496,12 @@
 	return (vm->suspended_cpus);
 }
 
+bool
+vm_start_single_cpu(struct vm *vm, struct vcpu *vcpu)
+{
+	return (CPU_TEST_CLR_ATOMIC(vcpu->vcpuid, &vm->startup_cpus));
+}
+
 /*
  * Returns the subset of vCPUs in tostart that are awaiting startup.
  * These vCPUs are also marked as no longer awaiting startup.
@@ -2512,6 +2518,12 @@
 	return (set);
 }
 
+void
+vm_await_start_single_cpu(struct vm *vm, struct vcpu *vcpu)
+{
+	CPU_SET_ATOMIC(vcpu->vcpuid, &vm->startup_cpus);
+}
+
 void
 vm_await_start(struct vm *vm, const cpuset_t *waiting)
 {
diff --git a/sys/sys/cpuset.h b/sys/sys/cpuset.h
--- a/sys/sys/cpuset.h
+++ b/sys/sys/cpuset.h
@@ -64,6 +64,8 @@
 #define	CPU_CLR_ATOMIC(n, p)		__BIT_CLR_ATOMIC(CPU_SETSIZE, n, p)
 #define	CPU_SET_ATOMIC(n, p)		__BIT_SET_ATOMIC(CPU_SETSIZE, n, p)
 #define	CPU_SET_ATOMIC_ACQ(n, p)	__BIT_SET_ATOMIC_ACQ(CPU_SETSIZE, n, p)
+#define CPU_TEST_CLR_ATOMIC(n, p)	__BIT_TEST_CLR_ATOMIC(CPU_SETSIZE, n, p)
+#define CPU_TEST_SET_ATOMIC(n, p)	__BIT_TEST_SET_ATOMIC(CPU_SETSIZE, n, p)
 #define	CPU_AND_ATOMIC(n, p)		__BIT_AND_ATOMIC(CPU_SETSIZE, n, p)
 #define	CPU_OR_ATOMIC(d, s)		__BIT_OR_ATOMIC(CPU_SETSIZE, d, s)
 #define	CPU_COPY_STORE_REL(f, t)	__BIT_COPY_STORE_REL(CPU_SETSIZE, f, t)