iv/linux
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

net: add __sock_wfree() helper

Browse Source 
Hosts sending lot of ACK packets exhibit high sock_wfree() cost
because of cache line miss to test SOCK_USE_WRITE_QUEUE

We could move this flag close to sk_wmem_alloc but it is better
to perform the atomic_sub_and_test() on a clean cache line,
as it avoid one extra bus transaction.

skb_orphan_partial() can also have a fast track for packets that either
are TCP acks, or already went through another skb_orphan_partial()

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit is contained in:
Eric Dumazet
2016-05-02 10:56:27 -07:00
committed by David S. Miller
parent e34b1638d0
commit 1d2077ac01
3 changed files with 26 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
net/core/sock.c
View File

@ -1655,6 +1655,17 @@ void sock_wfree(struct sk_buff *skb)
}
EXPORT_SYMBOL(sock_wfree);
/* This variant of sock_wfree() is used by TCP,
* since it sets SOCK_USE_WRITE_QUEUE.
*/
void __sock_wfree(struct sk_buff *skb)
{
struct sock *sk = skb->sk;
if (atomic_sub_and_test(skb->truesize, &sk->sk_wmem_alloc))
__sk_free(sk);
}
void skb_set_owner_w(struct sk_buff *skb, struct sock *sk)
{
skb_orphan(skb);
@ -1677,8 +1688,21 @@ void skb_set_owner_w(struct sk_buff *skb, struct sock *sk)
}
EXPORT_SYMBOL(skb_set_owner_w);
/* This helper is used by netem, as it can hold packets in its
* delay queue. We want to allow the owner socket to send more
* packets, as if they were already TX completed by a typical driver.
* But we also want to keep skb->sk set because some packet schedulers
* rely on it (sch_fq for example). So we set skb->truesize to a small
* amount (1) and decrease sk_wmem_alloc accordingly.
*/
void skb_orphan_partial(struct sk_buff *skb)
{
/* If this skb is a TCP pure ACK or already went here,
* we have nothing to do. 2 is already a very small truesize.
*/
if (skb->truesize <= 2)
return;
/* TCP stack sets skb->ooo_okay based on sk_wmem_alloc,
* so we do not completely orphan skb, but transfert all
* accounted bytes but one, to avoid unexpected reorders.