On 29 Sep 2026, at 10:34, David Marchand wrote:
> This is a preparation for separating ethernet and vHost DPDK netdevices.
>
> Rename netdev ops specific to ethernet ports, then add specific helpers
> for ethernet and vHost ports.
>
> With this split, the use of the device type becomes unnecessary in some
> places, but for the rest, rely on the netdev_class.
>
> Some notes on the additional cleanups:
> - a vHost port can not be offloaded, so the helpers related to
> the dpdk dpif-offload only need to care about is_eth_class(),
> - dpdk_set_rx_steer_config() is called only from
> netdev_dpdk_set_config(), so there is no point checking the type,
> - the construct() op never fails, so make it void,
Thanks David,
One small alignment nit. I tried to cut out most of the patch from the
replies in this series due to its size.
//Eelco
[...]
> +
> +static int
> +netdev_dpdk_eth_update_flags(struct netdev *netdev,
> + enum netdev_flags off, enum netdev_flags on,
> + enum netdev_flags *old_flagsp)
Alignment?
> +{
> + struct netdev_dpdk *dev = netdev_dpdk_cast(netdev);
> + int error;
> +
> + ovs_mutex_lock(&dev->mutex);
> + error = netdev_dpdk_eth_update_flags__(dev, off, on, old_flagsp);
> + ovs_mutex_unlock(&dev->mutex);
> +
> + return error;
> +}
[...]
_______________________________________________
dev mailing list
[email protected]
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
