Re: [PATCH v2 1/2] net: dsa: ensure linearized SKBs in case of tail taggers

[Florian Fainelli]

On 7/21/2021 4:35 PM, Vladimir Oltean wrote:
> On Wed, Jul 21, 2021 at 11:56:41PM +0200, Lino Sanfilippo wrote:
> > The function skb_put() that is used by tail taggers to make room for the
> > DSA tag must only be called for linearized SKBS. However in case that the
> > slave device inherited features like NETIF_F_HW_SG or NETIF_F_FRAGLIST the
> > SKB passed to the slaves transmit function may not be linearized.
> > Avoid those SKBs by clearing the NETIF_F_HW_SG and NETIF_F_FRAGLIST flags
> > for tail taggers.
> > Furthermore since the tagging protocol can be changed at runtime move the
> > code for setting up the slaves features into dsa_slave_setup_tagger().
> >
> > Suggested-by: Vladimir Oltean <olteanv@xxxxxxxxx>
> > Signed-off-by: Lino Sanfilippo <LinoSanfilippo@xxxxxx>
> > ---
> >   net/dsa/slave.c | 14 +++++++++-----
> >   1 file changed, 9 insertions(+), 5 deletions(-)
> >
> > diff --git a/net/dsa/slave.c b/net/dsa/slave.c
> > index 22ce11cd770e..ae2a648ed9be 100644
> > --- a/net/dsa/slave.c
> > +++ b/net/dsa/slave.c
> > @@ -1808,6 +1808,7 @@ void dsa_slave_setup_tagger(struct net_device *slave)
> >   	struct dsa_slave_priv *p = netdev_priv(slave);
> >   	const struct dsa_port *cpu_dp = dp->cpu_dp;
> >   	struct net_device *master = cpu_dp->master;
> > +	const struct dsa_switch *ds = dp->ds;
> >   
> >   	slave->needed_headroom = cpu_dp->tag_ops->needed_headroom;
> >   	slave->needed_tailroom = cpu_dp->tag_ops->needed_tailroom;
> > @@ -1819,6 +1820,14 @@ void dsa_slave_setup_tagger(struct net_device *slave)
> >   	slave->needed_tailroom += master->needed_tailroom;
> >   
> >   	p->xmit = cpu_dp->tag_ops->xmit;
> > +
> > +	slave->features = master->vlan_features | NETIF_F_HW_TC;
> > +	if (ds->ops->port_vlan_add && ds->ops->port_vlan_del)
> > +		slave->features |= NETIF_F_HW_VLAN_CTAG_FILTER;
> > +	slave->hw_features |= NETIF_F_HW_TC;
> > +	slave->features |= NETIF_F_LLTX;
> > +	if (slave->needed_tailroom)
> > +		slave->features &= ~(NETIF_F_SG | NETIF_F_FRAGLIST);
> >   }
> >   
> >   static struct lock_class_key dsa_slave_netdev_xmit_lock_key;
> > @@ -1881,11 +1890,6 @@ int dsa_slave_create(struct dsa_port *port)
> >   	if (slave_dev == NULL)
> >   		return -ENOMEM;
> >   
> > -	slave_dev->features = master->vlan_features | NETIF_F_HW_TC;
> > -	if (ds->ops->port_vlan_add && ds->ops->port_vlan_del)
> > -		slave_dev->features |= NETIF_F_HW_VLAN_CTAG_FILTER;
> > -	slave_dev->hw_features |= NETIF_F_HW_TC;
> > -	slave_dev->features |= NETIF_F_LLTX;
> >   	slave_dev->ethtool_ops = &dsa_slave_ethtool_ops;
> >   	if (!is_zero_ether_addr(port->mac))
> >   		ether_addr_copy(slave_dev->dev_addr, port->mac);
> > --
> > 2.32.0
>
> I would have probably changed the code in dsa_slave_create just like
> this:
>
> -	slave->features = master->vlan_features | NETIF_F_HW_TC;
> +	slave->features = NETIF_F_HW_TC;
> ...
> -	slave_dev->vlan_features = master->vlan_features;
>
> and in dsa_slave_setup_tagger:
>
> +	vlan_features = master->vlan_features;
> +	slave->features &= ~vlan_features;
> +	if (slave->needed_tailroom)
> +		vlan_features &= ~(NETIF_F_SG | NETIF_F_FRAGLIST);
> +	slave->features |= vlan_features;
> +	slave->vlan_features = vlan_features;
>
> no need to move around NETIF_F_HW_TC and NETIF_F_LLTX. Makes sense?
>
> And I would probably add:
>
> Fixes: 91da11f870f0 ("net: Distributed Switch Architecture protocol support")

Agreed, with those fixed:

Reviewed-by: Florian Fainelli <f.fainelli@xxxxxxxxx>
--
Florian