ip_md_tunnel_xmit unnecessarily references the dst_entry from the
dst_cache when interacting with the cache.

Reduce this overhead by avoiding the redundant refcount increments.

This is only possible in flows where the cache is used. Otherwise, we
fall-back to a referenced dst.

This change is safe since ipv4 supports noref xmit under RCU which is
already the case for ip_md_tunnel_xmit.

Signed-off-by: Marek Mietus <mmietus97@yahoo.com>
---
 net/ipv4/ip_tunnel.c | 25 ++++++++++++++++---------
 1 file changed, 16 insertions(+), 9 deletions(-)

diff --git a/net/ipv4/ip_tunnel.c b/net/ipv4/ip_tunnel.c
index 6aa045793048..323d085cc377 100644
--- a/net/ipv4/ip_tunnel.c
+++ b/net/ipv4/ip_tunnel.c
@@ -578,6 +578,7 @@ void ip_md_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
 	const struct iphdr *inner_iph;
 	struct rtable *rt = NULL;
 	struct flowi4 fl4;
+	dstref_t dstref;
 	__be16 df = 0;
 	u8 tos, ttl;
 	bool use_cache;
@@ -608,20 +609,26 @@ void ip_md_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
 		goto tx_error;
 
 	use_cache = ip_tunnel_dst_cache_usable(skb, tun_info);
-	if (use_cache)
-		rt = dst_cache_get_ip4(&tun_info->dst_cache, &fl4.saddr);
+	if (use_cache) {
+		rt = dst_cache_get_ip4_rcu(&tun_info->dst_cache, &fl4.saddr);
+		dstref = dst_to_dstref_noref(&rt->dst);
+	}
 	if (!rt) {
 		rt = ip_route_output_key(tunnel->net, &fl4);
 		if (IS_ERR(rt)) {
 			DEV_STATS_INC(dev, tx_carrier_errors);
 			goto tx_error;
 		}
-		if (use_cache)
-			dst_cache_set_ip4(&tun_info->dst_cache, &rt->dst,
-					  fl4.saddr);
+		if (use_cache) {
+			dst_cache_steal_ip4(&tun_info->dst_cache, &rt->dst,
+					    fl4.saddr);
+			dstref = dst_to_dstref_noref(&rt->dst);
+		} else {
+			dstref = dst_to_dstref(&rt->dst);
+		}
 	}
 	if (rt->dst.dev == dev) {
-		ip_rt_put(rt);
+		dstref_drop(dstref);
 		DEV_STATS_INC(dev, collisions);
 		goto tx_error;
 	}
@@ -630,7 +637,7 @@ void ip_md_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
 		df = htons(IP_DF);
 	if (tnl_update_pmtu(dev, skb, rt, df, inner_iph, tunnel_hlen,
 			    key->u.ipv4.dst, true)) {
-		ip_rt_put(rt);
+		dstref_drop(dstref);
 		goto tx_error;
 	}
 
@@ -647,13 +654,13 @@ void ip_md_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
 
 	headroom += LL_RESERVED_SPACE(rt->dst.dev) + rt->dst.header_len;
 	if (skb_cow_head(skb, headroom)) {
-		ip_rt_put(rt);
+		dstref_drop(dstref);
 		goto tx_dropped;
 	}
 
 	ip_tunnel_adj_headroom(dev, headroom);
 
-	iptunnel_xmit(NULL, dst_to_dstref(&rt->dst), skb, fl4.saddr, fl4.daddr, proto, tos, ttl,
+	iptunnel_xmit(NULL, dstref, skb, fl4.saddr, fl4.daddr, proto, tos, ttl,
 		      df, !net_eq(tunnel->net, dev_net(dev)), 0);
 	return;
 tx_error:
-- 
2.51.0