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/dccp/ipv6.c | |
| parent | inet: reqsk_alloc() needs to take care of dead listeners (diff) | |
| download | linux-dev-3b24d854cb35383c30642116e5992fd619bdc9bc.tar.xz linux-dev-3b24d854cb35383c30642116e5992fd619bdc9bc.zip | |
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 '')
| -rw-r--r-- | net/dccp/ipv6.c | 7 |
1 files changed, 5 insertions, 2 deletions
diff --git a/net/dccp/ipv6.c b/net/dccp/ipv6.c index 4663a01d5039..71bf1deba4c5 100644 --- a/net/dccp/ipv6.c +++ b/net/dccp/ipv6.c @@ -642,6 +642,7 @@ discard: static int dccp_v6_rcv(struct sk_buff *skb) { const struct dccp_hdr *dh; + bool refcounted; struct sock *sk; int min_cov; @@ -670,7 +671,7 @@ static int dccp_v6_rcv(struct sk_buff *skb) lookup: sk = __inet6_lookup_skb(&dccp_hashinfo, skb, __dccp_hdr_len(dh), dh->dccph_sport, dh->dccph_dport, - inet6_iif(skb)); + inet6_iif(skb), &refcounted); if (!sk) { dccp_pr_debug("failed to look up flow ID in table and " "get corresponding socket\n"); @@ -699,6 +700,7 @@ lookup: goto lookup; } sock_hold(sk); + refcounted = true; nsk = dccp_check_req(sk, skb, req); if (!nsk) { reqsk_put(req); @@ -752,7 +754,8 @@ discard_it: return 0; discard_and_relse: - sock_put(sk); + if (refcounted) + sock_put(sk); goto discard_it; } |