blob: 2b5ef932de4a4bb29ee2f94f1f8e10b08977093c (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
|
From a9b204d1564702b704ad6fe74f10a102c7b87ba3 Mon Sep 17 00:00:00 2001
From: Eric Dumazet <edumazet@google.com>
Date: Sat, 3 Dec 2016 11:14:53 -0800
Subject: [PATCH 04/10] tcp: tsq: avoid one atomic in tcp_wfree()
Under high load, tcp_wfree() has an atomic operation trying
to schedule a tasklet over and over.
We can schedule it only if our per cpu list was empty.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
net/ipv4/tcp_output.c | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -886,6 +886,7 @@ void tcp_wfree(struct sk_buff *skb)
for (oval = READ_ONCE(tp->tsq_flags);; oval = nval) {
struct tsq_tasklet *tsq;
+ bool empty;
if (!(oval & TSQF_THROTTLED) || (oval & TSQF_QUEUED))
goto out;
@@ -898,8 +899,10 @@ void tcp_wfree(struct sk_buff *skb)
/* queue this socket to tasklet queue */
local_irq_save(flags);
tsq = this_cpu_ptr(&tsq_tasklet);
+ empty = list_empty(&tsq->head);
list_add(&tp->tsq_node, &tsq->head);
- tasklet_schedule(&tsq->tasklet);
+ if (empty)
+ tasklet_schedule(&tsq->tasklet);
local_irq_restore(flags);
return;
}
|
