On Mon, Sep 07, 2020 at 06:53:23AM -0400, Jason Wang wrote:
>
>
> ----- Original Message -----
> > If the memory map changes before the driver status is
> > VIRTIO_CONFIG_S_DRIVER_OK, don't attempt to create resources because it
> > may fail. For example, if the VQ is not ready there is no point in
> > creating resources.
> >
> > Fixes: 1a86b377aa21 ("vdpa/mlx5: Add VDPA driver for supported mlx5
> > devices")
> > Signed-off-by: Eli Cohen <e...@nvidia.com>
> > ---
> > drivers/vdpa/mlx5/net/mlx5_vnet.c | 3 +++
> > 1 file changed, 3 insertions(+)
> >
> > diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.c
> > b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> > index 9df69d5efe8c..c89cd48a0aab 100644
> > --- a/drivers/vdpa/mlx5/net/mlx5_vnet.c
> > +++ b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> > @@ -1645,6 +1645,9 @@ static int mlx5_vdpa_change_map(struct mlx5_vdpa_net
> > *ndev, struct vhost_iotlb *
> > if (err)
> > goto err_mr;
> >
> > + if (!(ndev->mvdev.status & VIRTIO_CONFIG_S_DRIVER_OK))
> > + return 0;
> > +
>
> I'm not sure I get this.
>
> It looks to me if set_map() is called before DRIVER_OK, we won't build
> any mapping?
>
What would prevent that? Is it some qemu logic you're relying upon?
With current qemu 5.1 with lack of batching support, I get plenty calls
to set_map which result in calls to mlx5_vdpa_change_map().
If that happens before VIRTIO_CONFIG_S_DRIVER_OK then Imay fail (in case
I was not called to set VQs ready).
>
> > restore_channels_info(ndev);
> > err = setup_driver(ndev);
> > if (err)
> > --
> > 2.26.0
> >
> >
>
