Re: [net-next,v2,1/4] net: use ndo_set_features to set offload features for bonding/bridge/team

[Simon Horman <horms@kernel.org>]

This is an AI-generated review of your patch. The human sending this
email has considered the AI review valid, or at least plausible.

For local reproduction steps see https://netdev-ai.bots.linux.dev/ai-local.html
---
net: use ndo_set_features to set offload features for bonding/bridge/team

This commit converts bonding, bridge, and team drivers to use the
ndo_set_features callback instead of manually calling
netdev_compute_master_upper_features() during port add/remove operations.

> diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
> index 90b71bd53040..2c84d3ce3b04 100644
> --- a/drivers/net/bonding/bond_main.c
> +++ b/drivers/net/bonding/bond_main.c

[ ... ]

> @@ -5953,6 +5957,7 @@ static const struct net_device_ops bond_netdev_ops = {
>  	.ndo_add_slave		= bond_enslave,
>  	.ndo_del_slave		= bond_release,
>  	.ndo_fix_features	= bond_fix_features,
> +	.ndo_set_features	= bond_set_features,
>  	.ndo_features_check	= passthru_features_check,
>  	.ndo_get_xmit_slave	= bond_xmit_get_slave,
>  	.ndo_sk_get_lower_dev	= bond_sk_get_lower_dev,

[ ... ]

> diff --git a/drivers/net/team/team_core.c b/drivers/net/team/team_core.c
> index b7282f5c9632..4906ea3717b1 100644
> --- a/drivers/net/team/team_core.c
> +++ b/drivers/net/team/team_core.c

[ ... ]

> @@ -2007,6 +2010,7 @@ static const struct net_device_ops team_netdev_ops = {
>  	.ndo_add_slave		= team_add_slave,
>  	.ndo_del_slave		= team_del_slave,
>  	.ndo_fix_features	= team_fix_features,
> +	.ndo_set_features	= team_set_features,
>  	.ndo_change_carrier     = team_change_carrier,
>  	.ndo_features_check	= passthru_features_check,
>  };

[ ... ]

> diff --git a/net/bridge/br_device.c b/net/bridge/br_device.c
> index f7502e62dd35..f82082050e36 100644
> --- a/net/bridge/br_device.c
> +++ b/net/bridge/br_device.c

[ ... ]

> @@ -456,6 +462,7 @@ static const struct net_device_ops br_netdev_ops = {
>  	.ndo_add_slave		 = br_add_slave,
>  	.ndo_del_slave		 = br_del_slave,
>  	.ndo_fix_features        = br_fix_features,
> +	.ndo_set_features        = br_set_features,
>  	.ndo_fdb_add		 = br_fdb_add,
>  	.ndo_fdb_del		 = br_fdb_delete,
>  	.ndo_fdb_del_bulk	 = br_fdb_delete_bulk,

[ ... ]

> diff --git a/net/core/dev.c b/net/core/dev.c
> index f48dc299e4b2..16b3fc620807 100644
> --- a/net/core/dev.c
> +++ b/net/core/dev.c
> @@ -8920,6 +8920,9 @@ static int __netdev_upper_dev_link(struct net_device *dev,
>  	__netdev_walk_all_upper_dev(upper_dev, __netdev_update_lower_level,
>  				    priv);
>
> +	/* re-compute all features after adding link */
> +	netdev_change_features(upper_dev);
> +
>  	return 0;
>
>  rollback:

Can the feature computation skip updating ancillary fields when features
don't change?

The new architecture routes computation through netdev_change_features() ->
__netdev_update_features() -> ndo_set_features(). However,
__netdev_update_features() has this check:

net/core/dev.c:__netdev_update_features() {
    ...
    if (dev->features == features)
        goto sync_lower;

This skips the ndo_set_features callback entirely when computed features
match current features. This means netdev_compute_master_upper_features() is
never called when dev->features doesn't change, even though it updates
several ancillary fields independent of dev->features:

- vlan_features
- hw_enc_features
- mpls_features
- gso_partial_features
- tso_max_size
- tso_max_segs
- hard_header_len
- needed_headroom
- needed_tailroom

For example, adding a second bond slave with identical dev->features but
different tso_max_size or vlan_features will not trigger ndo_set_features,
leaving the bond's ancillary fields stale.

In the old code, netdev_compute_master_upper_features() was called
unconditionally and always updated these fields regardless of whether
dev->features changed.

> @@ -9012,6 +9015,9 @@ static void __netdev_upper_dev_unlink(struct net_device *dev,
>  	__netdev_update_lower_level(upper_dev, priv);
>  	__netdev_walk_all_upper_dev(upper_dev, __netdev_update_lower_level,
>  				    priv);
> +
> +	/* re-compute all features after removing link */
> +	netdev_change_features(upper_dev);
>  }

The same issue applies here. When a port is removed but dev->features
doesn't change, the ancillary fields won't be recomputed.

> @@ -12874,8 +12880,6 @@ void netdev_compute_master_upper_features(struct net_device *dev, bool update_he
>
>  	netif_set_tso_max_segs(dev, tso_max_segs);
>  	netif_set_tso_max_size(dev, tso_max_size);
> -
> -	netdev_change_features(dev);
>  }
>  EXPORT_SYMBOL(netdev_compute_master_upper_features);