[PATCH net-next 01/21] ipv6: remove unnecessary dst_hold() in ip6_fragment()

[Wei Wang <tracywwnj@gmail.com>]

From: Wei Wang <weiwan@google.com>

In ipv6 tx path, rcu_read_lock() is taken so that dst won't get freed
during the execution of ip6_fragment(). Hence, no need to hold dst in
it.

Signed-off-by: Wei Wang <weiwan@google.com>
Acked-by: Martin KaFai Lau <kafai@fb.com>
---
 net/ipv6/ip6_output.c | 4 ----
 1 file changed, 4 deletions(-)

diff --git a/net/ipv6/ip6_output.c b/net/ipv6/ip6_output.c
index 8b8efb0e55bf..5baa6fab4b97 100644
--- a/net/ipv6/ip6_output.c
+++ b/net/ipv6/ip6_output.c
@@ -698,8 +698,6 @@ int ip6_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
 		ipv6_hdr(skb)->payload_len = htons(first_len -
 						   sizeof(struct ipv6hdr));
 
-		dst_hold(&rt->dst);
-
 		for (;;) {
 			/* Prepare header of the next frame,
 			 * before previous one went down. */
@@ -742,7 +740,6 @@ int ip6_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
 		if (err == 0) {
 			IP6_INC_STATS(net, ip6_dst_idev(&rt->dst),
 				      IPSTATS_MIB_FRAGOKS);
-			ip6_rt_put(rt);
 			return 0;
 		}
 
@@ -750,7 +747,6 @@ int ip6_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
 
 		IP6_INC_STATS(net, ip6_dst_idev(&rt->dst),
 			      IPSTATS_MIB_FRAGFAILS);
-		ip6_rt_put(rt);
 		return err;
 
 slow_path_clean:
-- 
2.13.1.518.g3df882009-goog