Re: [PATCH] virtio_net: Do not send RSS key if it is not supported

[Breno Leitao]

Hello Xuan,

On Mon, Mar 25, 2024 at 01:57:53PM +0800, Xuan Zhuo wrote:
> On Fri, 22 Mar 2024 03:21:21 -0700, Breno Leitao <leitao@debian.org> wrote:
> > Hello Xuan,
> >
> > On Fri, Mar 22, 2024 at 10:00:22AM +0800, Xuan Zhuo wrote:
> > > On Thu, 21 Mar 2024 09:54:30 -0700, Breno Leitao <leitao@debian.org> 
> > > wrote:
> >
> > > > 4) Since the command above does not have a key, then the last
> > > >    scatter-gatter entry will be zeroed, since rss_key_size == 0.
> > > >     sg_buf_size = vi->rss_key_size;
> > >
> > >
> > >
> > >   if (vi->has_rss || vi->has_rss_hash_report) {
> > >           vi->rss_indir_table_size =
> > >                   virtio_cread16(vdev, offsetof(struct virtio_net_config,
> > >                           rss_max_indirection_table_length));
> > >           vi->rss_key_size =
> > >                   virtio_cread8(vdev, offsetof(struct virtio_net_config, 
> > > rss_max_key_size));
> > >
> > >           vi->rss_hash_types_supported =
> > >               virtio_cread32(vdev, offsetof(struct virtio_net_config, 
> > > supported_hash_types));
> > >           vi->rss_hash_types_supported &=
> > >                           ~(VIRTIO_NET_RSS_HASH_TYPE_IP_EX |
> > >                             VIRTIO_NET_RSS_HASH_TYPE_TCP_EX |
> > >                             VIRTIO_NET_RSS_HASH_TYPE_UDP_EX);
> > >
> > >           dev->hw_features |= NETIF_F_RXHASH;
> > >   }
> > >
> > >
> > > vi->rss_key_size is initiated here, I wonder if there is something wrong?
> >
> > Not really, the code above is never executed (in my machines). This is
> > because `vi->has_rss` and `vi->has_rss_hash_report` are both unset.
> >
> > Looking further, vdev does not have the VIRTIO_NET_F_RSS and
> > VIRTIO_NET_F_HASH_REPORT features.
> >
> > Also, when I run `ethtool -x`, I got:
> >
> >     # ethtool  -x eth0
> >     RX flow hash indirection table for eth0 with 1 RX ring(s):
> >     Operation not supported
> >     RSS hash key:
> >     Operation not supported
> >     RSS hash function:
> >         toeplitz: on
> >         xor: off
> >         crc32: off
> 
> 
> The spec saies:
>       Note that if the device offers VIRTIO_NET_F_HASH_REPORT, even if it
>       supports only one pair of virtqueues, it MUST support at least one of
>       commands of VIRTIO_NET_CTRL_MQ class to configure reported hash
>       parameters:
> 
>       If the device offers VIRTIO_NET_F_RSS, it MUST support
>       VIRTIO_NET_CTRL_MQ_RSS_CONFIG command per 5.1.6.5.7.1.
> 
>       Otherwise the device MUST support VIRTIO_NET_CTRL_MQ_HASH_CONFIG command
>       per 5.1.6.5.6.4.
> 
> 
> So if we have not anyone of `vi->has_rss` and `vi->has_rss_hash_report`,
> we should return from virtnet_set_rxfh directly.

Makes sense. Although it is not clear to me how vi->has_rss_hash_report
is related here, but, I am convinced that we shouldn't do any RSS
operation if the device doesn't have the RSS feature, i.e, vi->has_rss
is false.

That said, I am thinking about something like this. How does it sound?

        diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
        index 5a7700b103f8..8c1ad7361cf2 100644
        --- a/drivers/net/virtio_net.c
        +++ b/drivers/net/virtio_net.c
        @@ -3780,6 +3780,9 @@ static int virtnet_set_rxfh(struct net_device 
*dev,
                struct virtnet_info *vi = netdev_priv(dev);
                int i;
         
        +       if (!vi->has_rss)
        +               return -EOPNOTSUPP;
        +
                if (rxfh->hfunc != ETH_RSS_HASH_NO_CHANGE &&
                    rxfh->hfunc != ETH_RSS_HASH_TOP)
                        return -EOPNOTSUPP;

Thanks!