KVM: Move putting of vcpu->pid to kvm_vcpu_destroy()
authorSean Christopherson <sean.j.christopherson@intel.com>
Wed, 18 Dec 2019 21:55:29 +0000 (13:55 -0800)
committerPaolo Bonzini <pbonzini@redhat.com>
Mon, 27 Jan 2020 18:59:33 +0000 (19:59 +0100)
Move the putting of vcpu->pid to kvm_vcpu_destroy().  vcpu->pid is
guaranteed to be NULL when kvm_vcpu_uninit() is called in the error path
of kvm_vm_ioctl_create_vcpu(), e.g. it is explicitly nullified by
kvm_vcpu_init() and is only changed by KVM_RUN.

No functional change intended.

Acked-by: Christoffer Dall <christoffer.dall@arm.com>
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
virt/kvm/kvm_main.c

index 6b49603..483b6b8 100644 (file)
@@ -352,12 +352,6 @@ static int kvm_vcpu_init(struct kvm_vcpu *vcpu, struct kvm *kvm, unsigned id)
 
 static void kvm_vcpu_uninit(struct kvm_vcpu *vcpu)
 {
-       /*
-        * no need for rcu_read_lock as VCPU_RUN is the only place that
-        * will change the vcpu->pid pointer and on uninit all file
-        * descriptors are already gone.
-        */
-       put_pid(rcu_dereference_protected(vcpu->pid, 1));
        free_page((unsigned long)vcpu->run);
 }
 
@@ -365,6 +359,13 @@ void kvm_vcpu_destroy(struct kvm_vcpu *vcpu)
 {
        kvm_arch_vcpu_destroy(vcpu);
 
+       /*
+        * No need for rcu_read_lock as VCPU_RUN is the only place that changes
+        * the vcpu->pid pointer, and at destruction time all file descriptors
+        * are already gone.
+        */
+       put_pid(rcu_dereference_protected(vcpu->pid, 1));
+
        kvm_vcpu_uninit(vcpu);
        kmem_cache_free(kvm_vcpu_cache, vcpu);
 }