[PATCH net-next] veth: use peer attributes to set peer real number of queues

[PATCH net-next] veth: use peer attributes to set peer real number of queues

[Eric Dumazet]

veth_newlink() creates the peer from its own attributes (tbp), which
come from the nested VETH_INFO_PEER ifinfomsg when present, so the
peer num_tx_queues/num_rx_queues follow the peer IFLA_NUM_TX_QUEUES
and IFLA_NUM_RX_QUEUES.

veth_init_queues(peer, ...) is then called with the attributes of
the first device (tb) instead, so the decision to reduce the peer to
a single real queue is based on the wrong device:

  ip link add va type veth peer vb numtxqueues 4
	-> vb has 4 tx queues, but only 1 real tx queue.

  ip link add va numtxqueues 4 type veth peer vb
	-> vb keeps nr_possible_cpus real tx queues.

Pass tbp, which is tb when no peer attributes were given.

Fixes: 9d3684c24a52 ("veth: create by default nr_possible_cpus queues")
Signed-off-by: Eric Dumazet <edumazet@kernel.org>
Cc: Qiliang Yuan <odys.yuan@gmail.com>
---
 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 38f263d4f06b406eaf0a801334272cd438386c14..ba8953fef979aac3868d39c8fb9b19becb47d8e9 100644
--- a/drivers/net/veth.c
+++ b/drivers/net/veth.c
@@ -1924,7 +1924,7 @@ static int veth_newlink(struct net_device *dev,
 	priv = netdev_priv(peer);
 	rcu_assign_pointer(priv->peer, dev);
 	netdev_hold(dev, &priv->peer_tracker, GFP_KERNEL);
-	err = veth_init_queues(peer, tb);
+	err = veth_init_queues(peer, tbp);
 	if (err)
 		goto err_peer_queues;
 
-- 
2.53.0