On Wed, Dec 29, 2010 at 2:52 PM, Michael S. Tsirkin <m...@redhat.com> wrote:
> I'll probably split this patch in two, and merge into the
> appropriate patches in the ioeventfd series.
>
> Compile-tested only so far, would appreciate feedback/test reports.
virtio-ioeventfd works as expected.
> diff --git a/hw/vhost.c b/hw/vhost.c
> index 6082da2..1d09ed0 100644
> --- a/hw/vhost.c
> +++ b/hw/vhost.c
> @@ -630,6 +630,17 @@ void vhost_dev_cleanup(struct vhost_dev *hdev)
> int vhost_dev_start(struct vhost_dev *hdev, VirtIODevice *vdev)
> {
> int i, r;
> + /* A binding must support vmstate notifiers (for migration to work) */
> + if (!vdev->binding->vmstate_change) {
> + fprintf(stderr, "binding does not support vmstate notifiers\n");
> + r = -ENOSYS;
> + goto fail;
> + }
> + if (!vdev->binding->set_guest_notifiers) {
> + fprintf(stderr, "binding does not support guest notifiers\n");
> + r = -ENOSYS;
> + goto fail;
> + }
Merge error? The set_guest_notifiers check is already present.
Stefan
