net: rps: protect last_qtail with rps_input_queue_tail_save() helper

Removing one unnecessary reader protection and add another writer
protection to finish the locklessly proctection job.

Note: the removed READ_ONCE() is not needed because we only have to protect
the locklessly reader in the different context (rps_may_expire_flow()).

Signed-off-by: Jason Xing <kernelxing@tencent.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/core/dev.c

@@ -4507,7 +4507,7 @@ set_rps_cpu(struct net_device *dev, struct sk_buff *skb,
 		struct netdev_rx_queue *rxqueue;
 		struct rps_dev_flow_table *flow_table;
 		struct rps_dev_flow *old_rflow;
-		u32 flow_id;
+		u32 flow_id, head;
 		u16 rxq_index;
 		int rc;
 
@@ -4535,8 +4535,8 @@ set_rps_cpu(struct net_device *dev, struct sk_buff *skb,
 			old_rflow->filter = RPS_NO_FILTER;
 	out:
 #endif
-		rflow->last_qtail =
-			READ_ONCE(per_cpu(softnet_data, next_cpu).input_queue_head);
+		head = READ_ONCE(per_cpu(softnet_data, next_cpu).input_queue_head);
+		rps_input_queue_tail_save(&rflow->last_qtail, head);
 	}
 
 	rflow->cpu = next_cpu;
@@ -4619,7 +4619,7 @@ static int get_rps_cpu(struct net_device *dev, struct sk_buff *skb,
 		if (unlikely(tcpu != next_cpu) &&
 		    (tcpu >= nr_cpu_ids || !cpu_online(tcpu) ||
 		     ((int)(READ_ONCE(per_cpu(softnet_data, tcpu).input_queue_head) -
-		      READ_ONCE(rflow->last_qtail))) >= 0)) {
+		      rflow->last_qtail)) >= 0)) {
 			tcpu = next_cpu;
 			rflow = set_rps_cpu(dev, skb, rflow, next_cpu);
 		}