On Wed, Dec 20, 2023 at 3:08 PM Eugenio Perez Martin
<epere...@redhat.com> wrote:
>
> On Wed, Dec 20, 2023 at 5:34 AM Jason Wang <jasow...@redhat.com> wrote:
> >
> > On Sat, Dec 16, 2023 at 1:28 AM Eugenio Pérez <epere...@redhat.com> wrote:
> > >
> > > The backend does not reset them until the vdpa file descriptor is closed
> > > so there is no harm in doing it only once.
> > >
> > > This allows the destination of a live migration to premap memory in
> > > batches, using VHOST_BACKEND_F_IOTLB_BATCH.
> > >
> > > Signed-off-by: Eugenio Pérez <epere...@redhat.com>
> > > ---
> > > hw/virtio/vhost-vdpa.c | 50 ++++++++++++++++--------------------------
> > > 1 file changed, 19 insertions(+), 31 deletions(-)
> > >
> > > diff --git a/hw/virtio/vhost-vdpa.c b/hw/virtio/vhost-vdpa.c
> > > index 449c3794b2..43f7c382b1 100644
> > > --- a/hw/virtio/vhost-vdpa.c
> > > +++ b/hw/virtio/vhost-vdpa.c
> > > @@ -587,11 +587,25 @@ static int vhost_vdpa_init(struct vhost_dev *dev,
> > > void *opaque, Error **errp)
> > > struct vhost_vdpa *v = opaque;
> > > assert(dev->vhost_ops->backend_type == VHOST_BACKEND_TYPE_VDPA);
> > > trace_vhost_vdpa_init(dev, v->shared, opaque);
> > > + uint64_t backend_features;
> > > + uint64_t qemu_backend_features = 0x1ULL <<
> > > VHOST_BACKEND_F_IOTLB_MSG_V2 |
> > > + 0x1ULL <<
> > > VHOST_BACKEND_F_IOTLB_BATCH |
> > > + 0x1ULL <<
> > > VHOST_BACKEND_F_IOTLB_ASID |
> > > + 0x1ULL << VHOST_BACKEND_F_SUSPEND;
> > > int ret;
> > >
> > > v->dev = dev;
> > > dev->opaque = opaque ;
> > > v->shared->listener = vhost_vdpa_memory_listener;
> > > +
> > > + if (vhost_vdpa_call(dev, VHOST_GET_BACKEND_FEATURES,
> > > &backend_features)) {
> > > + return -EFAULT;
> > > + }
> > > +
> > > + backend_features &= qemu_backend_features;
> > > +
> > > + dev->backend_cap = backend_features;
> > > + v->shared->backend_cap = backend_features;
> > > vhost_vdpa_init_svq(dev, v);
> > >
> > > error_propagate(&dev->migration_blocker, v->migration_blocker);
> > > @@ -599,6 +613,11 @@ static int vhost_vdpa_init(struct vhost_dev *dev,
> > > void *opaque, Error **errp)
> > > return 0;
> > > }
> > >
> > > + ret = vhost_vdpa_call(dev, VHOST_SET_BACKEND_FEATURES,
> > > &backend_features);
> > > + if (ret) {
> > > + return -EFAULT;
> > > + }
> > > +
> > > /*
> > > * If dev->shadow_vqs_enabled at initialization that means the
> > > device has
> > > * been started with x-svq=on, so don't block migration
> > > @@ -829,36 +848,6 @@ static int vhost_vdpa_set_features(struct vhost_dev
> > > *dev,
> > > return vhost_vdpa_add_status(dev, VIRTIO_CONFIG_S_FEATURES_OK);
> > > }
> > >
> > > -static int vhost_vdpa_set_backend_cap(struct vhost_dev *dev)
> >
> > How about keeping this function but just calling it in vhost_vdpa_init()?
> >
>
> Sure, that is possible. I need to remove the VhostOps
> vhost_set_backend_cap = vhost_vdpa_set_backend_cap, anyway, is that ok
> for you?
Fine with me.
Thanks
>
> Thanks!
>
