[PATCH net-next] inet: minor optimization for backlog setting in listen(2)

Set the backlog earlier in inet_dccp_listen() and inet_listen(),
then we can avoid the redundant setting.

Signed-off-by: Yafang Shao <redacted>
---
 net/dccp/proto.c                | 2 +-
 net/ipv4/af_inet.c              | 2 +-
 net/ipv4/inet_connection_sock.c | 1 -
 3 files changed, 2 insertions(+), 3 deletions(-)

diff --git a/net/dccp/proto.c b/net/dccp/proto.c
index 43733ac..658cd32b 100644
--- a/net/dccp/proto.c
+++ b/net/dccp/proto.c

@@ -948,6 +948,7 @@ int inet_dccp_listen(struct socket *sock, int backlog)
 	if (!((1 << old_state) & (DCCPF_CLOSED | DCCPF_LISTEN)))
 		goto out;
 
+	sk->sk_max_ack_backlog = backlog;
 	/* Really, if the socket is already in listen state
 	 * we can only allow the backlog to be adjusted.
 	 */

@@ -960,7 +961,6 @@ int inet_dccp_listen(struct socket *sock, int backlog)
 		if (err)
 			goto out;
 	}
-	sk->sk_max_ack_backlog = backlog;
 	err = 0;
 
 out:

diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
index 1fbe2f8..39066cd 100644
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c

@@ -208,6 +208,7 @@ int inet_listen(struct socket *sock, int backlog)
 	if (!((1 << old_state) & (TCPF_CLOSE | TCPF_LISTEN)))
 		goto out;
 
+	sk->sk_max_ack_backlog = backlog;
 	/* Really, if the socket is already in listen state
 	 * we can only allow the backlog to be adjusted.
 	 */

@@ -231,7 +232,6 @@ int inet_listen(struct socket *sock, int backlog)
 			goto out;
 		tcp_call_bpf(sk, BPF_SOCK_OPS_TCP_LISTEN_CB, 0, NULL);
 	}
-	sk->sk_max_ack_backlog = backlog;
 	err = 0;
 
 out:

diff --git a/net/ipv4/inet_connection_sock.c b/net/ipv4/inet_connection_sock.c
index 15e7f79..860e22a 100644
--- a/net/ipv4/inet_connection_sock.c
+++ b/net/ipv4/inet_connection_sock.c

@@ -874,7 +874,6 @@ int inet_csk_listen_start(struct sock *sk, int backlog)
 
 	reqsk_queue_alloc(&icsk->icsk_accept_queue);
 
-	sk->sk_max_ack_backlog = backlog;
 	sk->sk_ack_backlog = 0;
 	inet_csk_delack_init(sk);
 

-- 
1.8.3.1

Bitwise operation is a little faster.
So I replace after() with (flag & FLAG_SND_UNA_ADVANCED) as this flag is
already set before.

Cc: Joe Perches <joe@perches.com>
Signed-off-by: Yafang Shao <redacted>
---
 net/ipv4/tcp_input.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 2868ef2..0167015 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c

@@ -3610,7 +3610,7 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag)
 	if (flag & FLAG_UPDATE_TS_RECENT)
 		tcp_replace_ts_recent(tp, TCP_SKB_CB(skb)->seq);
 
-	if (!(flag & FLAG_SLOWPATH) && after(ack, prior_snd_una)) {
+	if (!(flag & FLAG_SLOWPATH) && (flag & FLAG_SND_UNA_ADVANCED)) {
 		/* Window is constant, pure forward advance.
 		 * No more checks are required.
 		 * Note, we use the fact that SND.UNA>=SND.WL2.

-- 
1.8.3.1

On 11/07/2018 03:20 AM, Yafang Shao wrote:

Bitwise operation is a little faster.

What about reducing this to a single conditional jump ?

if ((flag & (FLAG_SLOWPATH | FLAG_SND_UNA_ADVANCED)) == FLAG_SND_UNA_ADVANCED)  {

On 11/07/2018 03:20 AM, Yafang Shao wrote:

Set the backlog earlier in inet_dccp_listen() and inet_listen(),
then we can avoid the redundant setting.

Signed-off-by: Yafang Shao <redacted>
---

Reviewed-by: Eric Dumazet <edumazet@google.com>

On Wed, Nov 7, 2018 at 11:16 PM Eric Dumazet [off-list ref] wrote:

On 11/07/2018 03:20 AM, Yafang Shao wrote:

quoted

Bitwise operation is a little faster.

quoted

So I replace after() with (flag & FLAG_SND_UNA_ADVANCED) as this flag is
already set before.

Cc: Joe Perches <joe@perches.com>
Signed-off-by: Yafang Shao <redacted>
---
 net/ipv4/tcp_input.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 2868ef2..0167015 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c

@@ -3610,7 +3610,7 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag)
      if (flag & FLAG_UPDATE_TS_RECENT)
              tcp_replace_ts_recent(tp, TCP_SKB_CB(skb)->seq);

-     if (!(flag & FLAG_SLOWPATH) && after(ack, prior_snd_una)) {
+     if (!(flag & FLAG_SLOWPATH) && (flag & FLAG_SND_UNA_ADVANCED)) {
              /* Window is constant, pure forward advance.
               * No more checks are required.
               * Note, we use the fact that SND.UNA>=SND.WL2.

What about reducing this to a single conditional jump ?

if ((flag & (FLAG_SLOWPATH | FLAG_SND_UNA_ADVANCED)) == FLAG_SND_UNA_ADVANCED)  {

That's better.
Will change it.


Thanks
Yafang

From: Yafang Shao <redacted>
Date: Wed,  7 Nov 2018 19:20:16 +0800

Set the backlog earlier in inet_dccp_listen() and inet_listen(),
then we can avoid the redundant setting.

Signed-off-by: Yafang Shao <redacted>

Applied.