diff options
| author | Eric Dumazet <edumazet@google.com> | 2016-04-01 08:52:17 -0700 |
|---|---|---|
| committer | David S. Miller <davem@davemloft.net> | 2016-04-04 22:11:20 -0400 |
| commit | 3b24d854cb35383c30642116e5992fd619bdc9bc (patch) | |
| tree | a2a0c8c8ad3f7f974d6a9b11b17865edfb62d6d1 /net/ipv4/inet_diag.c | |
| parent | 3a5d1c0e7cb5ba91aabbd7e28626e3cc925f8093 (diff) | |
| download | op-kernel-dev-3b24d854cb35383c30642116e5992fd619bdc9bc.zip op-kernel-dev-3b24d854cb35383c30642116e5992fd619bdc9bc.tar.gz | |
tcp/dccp: do not touch listener sk_refcnt under synflood
When a SYNFLOOD targets a non SO_REUSEPORT listener, multiple
cpus contend on sk->sk_refcnt and sk->sk_wmem_alloc changes.
By letting listeners use SOCK_RCU_FREE infrastructure,
we can relax TCP_LISTEN lookup rules and avoid touching sk_refcnt
Note that we still use SLAB_DESTROY_BY_RCU rules for other sockets,
only listeners are impacted by this change.
Peak performance under SYNFLOOD is increased by ~33% :
On my test machine, I could process 3.2 Mpps instead of 2.4 Mpps
Most consuming functions are now skb_set_owner_w() and sock_wfree()
contending on sk->sk_wmem_alloc when cooking SYNACK and freeing them.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'net/ipv4/inet_diag.c')
| -rw-r--r-- | net/ipv4/inet_diag.c | 3 |
1 files changed, 1 insertions, 2 deletions
diff --git a/net/ipv4/inet_diag.c b/net/ipv4/inet_diag.c index ea8df52..bd591eb 100644 --- a/net/ipv4/inet_diag.c +++ b/net/ipv4/inet_diag.c @@ -775,13 +775,12 @@ void inet_diag_dump_icsk(struct inet_hashinfo *hashinfo, struct sk_buff *skb, for (i = s_i; i < INET_LHTABLE_SIZE; i++) { struct inet_listen_hashbucket *ilb; - struct hlist_nulls_node *node; struct sock *sk; num = 0; ilb = &hashinfo->listening_hash[i]; spin_lock_bh(&ilb->lock); - sk_nulls_for_each(sk, node, &ilb->head) { + sk_for_each(sk, &ilb->head) { struct inet_sock *inet = inet_sk(sk); if (!net_eq(sock_net(sk), net)) |
