[PATCH net-next v3 4/5] veth: use persistent NAPI config

[Tianyi Gao <tianyi@cloudflare.com>]

veth deletes the NAPI instances on link down, when NAPI is no longer
needed for GRO or XDP, and when ethtool -L reduces the number of RX
queues. When veth adds them again, each queue gets a new NAPI ID and
loses its per-NAPI settings. An AF_XDP socket bound to the queue keeps
the old NAPI ID, so its busy polling stops.

Use netif_napi_add_config(), as Documentation/networking/napi.rst asks
drivers to do whenever possible. The queue then keeps its NAPI ID and
its settings. virtio_net made the same change in commit d5d715207e29
("virtio_net: Use persistent NAPI config").

Signed-off-by: Tianyi Gao <tianyi@cloudflare.com>
---
New in v3, as Sashiko suggested on v2 [1].

[1] https://lore.kernel.org/all/179470883411.2199510.13825316417997386083@kernel.org/
---
 drivers/net/veth.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/veth.c b/drivers/net/veth.c
index 470c4f95e4a1..1c7ee3bf65f0 100644
--- a/drivers/net/veth.c
+++ b/drivers/net/veth.c
@@ -1057,7 +1057,7 @@ static int veth_napi_enable_range(struct net_device *dev, int start, int end)
 	}
 
 	for (i = start; i < end; i++)
-		netif_napi_add(dev, &priv->rq[i].xdp_napi, veth_poll);
+		netif_napi_add_config(dev, &priv->rq[i].xdp_napi, veth_poll, i);
 
 	for (i = start; i < end; i++) {
 		struct veth_rq *rq = &priv->rq[i];
-- 
2.56.0