[PATCH net-next 3/8] net: enqueue_to_backlog() change vs not running device

[Eric Dumazet <edumazet@google.com>]

If the device attached to the packet given to enqueue_to_backlog()
is not running, we drop the packet.

But we accidentally increase sd->dropped, giving false signals
to admins: sd->dropped should be reserved to cpu backlog pressure,
not to temporary glitches at device dismantles.

While we are at it, perform the netif_running() test before
we get the rps lock, and use REASON_DEV_READY
drop reason instead of NOT_SPECIFIED.

Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 net/core/dev.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/net/core/dev.c b/net/core/dev.c
index 5d36a634f468ffdeaca598c3dd033fe06d240bd0..af7a34b0a7d6683c6ffb21dd3388ed678473d95e 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -4791,12 +4791,13 @@ static int enqueue_to_backlog(struct sk_buff *skb, int cpu,
 	unsigned long flags;
 	unsigned int qlen;
 
-	reason = SKB_DROP_REASON_NOT_SPECIFIED;
+	reason = SKB_DROP_REASON_DEV_READY;
+	if (!netif_running(skb->dev))
+		goto bad_dev;
+
 	sd = &per_cpu(softnet_data, cpu);
 
 	backlog_lock_irq_save(sd, &flags);
-	if (!netif_running(skb->dev))
-		goto drop;
 	qlen = skb_queue_len(&sd->input_pkt_queue);
 	if (qlen <= READ_ONCE(net_hotdata.max_backlog) &&
 	    !skb_flow_limit(skb, qlen)) {
@@ -4817,10 +4818,10 @@ static int enqueue_to_backlog(struct sk_buff *skb, int cpu,
 	}
 	reason = SKB_DROP_REASON_CPU_BACKLOG;
 
-drop:
 	sd->dropped++;
 	backlog_unlock_irq_restore(sd, &flags);
 
+bad_dev:
 	dev_core_stats_rx_dropped_inc(skb->dev);
 	kfree_skb_reason(skb, reason);
 	return NET_RX_DROP;
-- 
2.44.0.478.gd926399ef9-goog