KVM: s390: Fix vcpu_load handling in interrupt code

Recent changes (KVM: make processes waiting on vcpu mutex killable)
now requires to check the return value of vcpu_load. This triggered
a warning in s390 specific kvm code. Turns out that we can actually
remove the put/load, since schedule will do the right thing via
the preempt notifiers.

Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>

arch/s390/kvm/interrupt.c

@@ -423,9 +423,7 @@ int kvm_s390_handle_wait(struct kvm_vcpu *vcpu)
 		set_current_state(TASK_INTERRUPTIBLE);
 		spin_unlock_bh(&vcpu->arch.local_int.lock);
 		spin_unlock(&vcpu->arch.local_int.float_int->lock);
-		vcpu_put(vcpu);
 		schedule();
-		vcpu_load(vcpu);
 		spin_lock(&vcpu->arch.local_int.float_int->lock);
 		spin_lock_bh(&vcpu->arch.local_int.lock);
 	}