On Tue, 2016-02-23 at 13:53 +0100, Paolo Abeni wrote:
> ndo_set_rx_headroom controls the align value used by tun devices to
> allocate skbs on frame reception.
> When the xmit device adds a large encapsulation, this avoids an skb
> head reallocation on forwarding.
>
> The measured improvement when forwarding towards a vxlan dev with
> frame size below the egress device MTU is around 6% when tunneling over
> ipv6.
>
> In case of ipv4 tunnels there is no improvement, since the tun
> device default alignment provides enough headroom to avoid the skb
> head reallocation, at least on hosts with 64 bytes cacheline.
>
> Signed-off-by: Paolo Abeni <pab...@redhat.com>
> ---
> drivers/net/tun.c | 16 +++++++++++++++-
> 1 file changed, 15 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/net/tun.c b/drivers/net/tun.c
> index 88bb8cc..5812693 100644
> --- a/drivers/net/tun.c
> +++ b/drivers/net/tun.c
> @@ -187,6 +187,7 @@ struct tun_struct {
> #define TUN_USER_FEATURES (NETIF_F_HW_CSUM|NETIF_F_TSO_ECN|NETIF_F_TSO| \
> NETIF_F_TSO6|NETIF_F_UFO)
>
> + int align;
This needs to be initialized to NET_SKB_PAD, to preserved the current
behavior.
I'll fix it in v2.
Paolo
