Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   kerneljasonxing@gmail.com
To:     davem@davemloft.net, kuba@kernel.org, ast@kernel.org,
        daniel@iogearbox.net, hawk@kernel.org, john.fastabend@gmail.com,
        andrii@kernel.org, kafai@fb.com, songliubraving@fb.com, yhs@fb.com,
        kpsingh@kernel.org, edumazet@google.com, atenart@kernel.org,
        alobakin@pm.me, weiwan@google.com, bjorn@kernel.org, arnd@arndb.de,
        memxor@gmail.com
Cc:     netdev@vger.kernel.org, linux-kernel@vger.kernel.org,
        bpf@vger.kernel.org, kerneljasonxing@gmail.com,
        Jason Xing <xingwanli@kuaishou.com>
Subject: [PATCH net] net: gro: flush the real oldest skb
Date:   Wed, 27 Oct 2021 16:49:44 +0800
Message-Id: <20211027084944.4508-1-kerneljasonxing@gmail.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

From: Jason Xing <xingwanli@kuaishou.com>

Prior to this patch, when the count of skbs of one flow is larger than
MAX_GRO_SKBS, gro_flush_oldest() flushes the tail of the list. However,
as we can see in the merge part of skb_gro_receive(), the tail of the
list is the newest, head oldest.

Here, we need to fetch the real oldest one and then process it to lower
the latency.

Fix: 07d78363dc ("net: Convert NAPI gro list into a small hash table.")
Signed-off-by: Jason Xing <xingwanli@kuaishou.com>
---
 net/core/dev.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/net/core/dev.c b/net/core/dev.c
index 7ee9fec..d52ebdb 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -6012,7 +6012,7 @@ static void gro_flush_oldest(struct napi_struct *napi, struct list_head *head)
 {
 	struct sk_buff *oldest;
 
-	oldest = list_last_entry(head, struct sk_buff, list);
+	oldest = list_first_entry(head, struct sk_buff, list);
 
 	/* We are called with head length >= MAX_GRO_SKBS, so this is
 	 * impossible.
-- 
1.8.3.1