Virtualization - Feb 2023 - [PATCH v2] virtio-net: Fix probe of virtio-net on kvmtool

On Thu, Feb 23, 2023 at 07:38:25PM +0000, Rob Bradford via B4 Relay
wrote:
> From: Rob Bradford <rbradford at rivosinc.com>
> 
> kvmtool does not support the VIRTIO_NET_F_CTRL_GUEST_OFFLOADS feature
> but does advertise the VIRTIO_NET_F_GUEST_TSO{4,6} features. Check that
> the VIRTIO_NET_F_CTRL_GUEST_OFFLOADS feature is present before setting
> the NETIF_F_GRO_HW feature bit as otherwise an attempt will be made to
> program the virtio-net device using the ctrl queue which will fail.
> 
> This resolves the following error when running on kvmtool:
> 
> [    1.865992] net eth0: Fail to set guest offload.
> [    1.872491] virtio_net virtio2 eth0: set_features() failed (-22); wanted
0x0000000000134829, left 0x0080000000134829
> 
> Signed-off-by: Rob Bradford <rbradford at rivosinc.com>
> ---
> Changes in v2:
> - Use parentheses to group logical OR of features 
> - Link to v1:
>   https://lore.kernel.org/r/20230223-virtio-net-kvmtool-v1-1-fc23d29b9d7a
at rivosinc.com
> ---
>  drivers/net/virtio_net.c | 7 +++----
>  1 file changed, 3 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 61e33e4dd0cd..f8341d1a4ccd 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -3780,10 +3780,9 @@ static int virtnet_probe(struct virtio_device *vdev)
>  	}
>  	if (virtio_has_feature(vdev, VIRTIO_NET_F_GUEST_CSUM))
>  		dev->features |= NETIF_F_RXCSUM;
> -	if (virtio_has_feature(vdev, VIRTIO_NET_F_GUEST_TSO4) ||
> -	    virtio_has_feature(vdev, VIRTIO_NET_F_GUEST_TSO6))
> -		dev->features |= NETIF_F_GRO_HW;
> -	if (virtio_has_feature(vdev, VIRTIO_NET_F_CTRL_GUEST_OFFLOADS))
> +	if ((virtio_has_feature(vdev, VIRTIO_NET_F_GUEST_TSO4) ||
> +	    virtio_has_feature(vdev, VIRTIO_NET_F_GUEST_TSO6)) &&
> +	    virtio_has_feature(vdev, VIRTIO_NET_F_CTRL_GUEST_OFFLOADS))
>  		dev->hw_features |= NETIF_F_GRO_HW;
This will disable GRO/LRO on kvmtool completely causing a significant
performance regression.

Jason, isn't this what
	commit dbcf24d153884439dad30484a0e3f02350692e4c
	Author: Jason Wang <jasowang at redhat.com>
	Date:   Tue Aug 17 16:06:59 2021 +0800

	    virtio-net: use NETIF_F_GRO_HW instead of NETIF_F_LRO

was supposed to address?


And apropos this:
    
    Fix this by using NETIF_F_GRO_HW instead. Though the spec does not
    guarantee packets to be re-segmented as the original ones,
    we can add that to the spec, possibly with a flag for devices to
    differentiate between GRO and LRO.

this never happened. What's the plan exactly?



>  	dev->vlan_features = dev->features;
> 
> ---
> base-commit: c39cea6f38eefe356d64d0bc1e1f2267e282cdd3
> change-id: 20230223-virtio-net-kvmtool-87f37515be22
> 
> Best regards,
> -- 
> Rob Bradford <rbradford at rivosinc.com>

On Fri, Feb 24, 2023 at 4:25?PM Michael S. Tsirkin <mst at redhat.com>
wrote:
>
> On Thu, Feb 23, 2023 at 07:38:25PM +0000, Rob Bradford via B4 Relay wrote:
> > From: Rob Bradford <rbradford at rivosinc.com>
> >
> > kvmtool does not support the VIRTIO_NET_F_CTRL_GUEST_OFFLOADS feature
> > but does advertise the VIRTIO_NET_F_GUEST_TSO{4,6} features. Check
that
> > the VIRTIO_NET_F_CTRL_GUEST_OFFLOADS feature is present before setting
> > the NETIF_F_GRO_HW feature bit as otherwise an attempt will be made to
> > program the virtio-net device using the ctrl queue which will fail.
> >
> > This resolves the following error when running on kvmtool:
> >
> > [    1.865992] net eth0: Fail to set guest offload.
> > [    1.872491] virtio_net virtio2 eth0: set_features() failed (-22);
wanted 0x0000000000134829, left 0x0080000000134829
> >
> > Signed-off-by: Rob Bradford <rbradford at rivosinc.com>
> > ---
> > Changes in v2:
> > - Use parentheses to group logical OR of features
> > - Link to v1:
> >  
https://lore.kernel.org/r/20230223-virtio-net-kvmtool-v1-1-fc23d29b9d7a at
rivosinc.com
> > ---
> >  drivers/net/virtio_net.c | 7 +++----
> >  1 file changed, 3 insertions(+), 4 deletions(-)
> >
> > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > index 61e33e4dd0cd..f8341d1a4ccd 100644
> > --- a/drivers/net/virtio_net.c
> > +++ b/drivers/net/virtio_net.c
> > @@ -3780,10 +3780,9 @@ static int virtnet_probe(struct virtio_device
*vdev)
> >       }
> >       if (virtio_has_feature(vdev, VIRTIO_NET_F_GUEST_CSUM))
> >               dev->features |= NETIF_F_RXCSUM;
> > -     if (virtio_has_feature(vdev, VIRTIO_NET_F_GUEST_TSO4) ||
> > -         virtio_has_feature(vdev, VIRTIO_NET_F_GUEST_TSO6))
> > -             dev->features |= NETIF_F_GRO_HW;
> > -     if (virtio_has_feature(vdev, VIRTIO_NET_F_CTRL_GUEST_OFFLOADS))
> > +     if ((virtio_has_feature(vdev, VIRTIO_NET_F_GUEST_TSO4) ||
> > +         virtio_has_feature(vdev, VIRTIO_NET_F_GUEST_TSO6))
&&
> > +         virtio_has_feature(vdev, VIRTIO_NET_F_CTRL_GUEST_OFFLOADS))
> >               dev->hw_features |= NETIF_F_GRO_HW;
>
> This will disable GRO/LRO on kvmtool completely causing a significant
> performance regression.
>
> Jason, isn't this what
>         commit dbcf24d153884439dad30484a0e3f02350692e4c
>         Author: Jason Wang <jasowang at redhat.com>
>         Date:   Tue Aug 17 16:06:59 2021 +0800
>
>             virtio-net: use NETIF_F_GRO_HW instead of NETIF_F_LRO
>
> was supposed to address?
>
Yes, I've asked a similar question in another thread.
>
> And apropos this:
>
>     Fix this by using NETIF_F_GRO_HW instead. Though the spec does not
>     guarantee packets to be re-segmented as the original ones,
>     we can add that to the spec, possibly with a flag for devices to
>     differentiate between GRO and LRO.
>
> this never happened. What's the plan exactly?
It's in the backlog, but I'm out of bandwidth for doing that now.

Thanks
>
>
>
>
> >       dev->vlan_features = dev->features;
> >
> > ---
> > base-commit: c39cea6f38eefe356d64d0bc1e1f2267e282cdd3
> > change-id: 20230223-virtio-net-kvmtool-87f37515be22
> >
> > Best regards,
> > --
> > Rob Bradford <rbradford at rivosinc.com>
>