diff options
| author | Paolo Abeni <pabeni@redhat.com> | 2017-06-21 10:24:40 +0200 |
|---|---|---|
| committer | David S. Miller <davem@davemloft.net> | 2017-06-21 11:38:11 -0400 |
| commit | dd99e425be23294a9a91b365bd04f9b255fb72e8 (patch) | |
| tree | 0bd383473708ab4a183c3a138b6cee363ae4d569 /net/ipv4/udp.c | |
| parent | da2e9cf03b8fccbb69dd1e215bb1e554ce8e8cbe (diff) | |
| download | lwn-dd99e425be23294a9a91b365bd04f9b255fb72e8.tar.gz lwn-dd99e425be23294a9a91b365bd04f9b255fb72e8.zip | |
udp: prefetch rmem_alloc in udp_queue_rcv_skb()
On UDP packets processing, if the BH is the bottle-neck, it
always sees a cache miss while updating rmem_alloc; try to
avoid it prefetching the value as soon as we have the socket
available.
Performances under flood with multiple NIC rx queues used are
unaffected, but when a single NIC rx queue is in use, this
gives ~10% performance improvement.
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'net/ipv4/udp.c')
| -rw-r--r-- | net/ipv4/udp.c | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c index f3450f092d71..067a607917f9 100644 --- a/net/ipv4/udp.c +++ b/net/ipv4/udp.c @@ -1949,6 +1949,7 @@ static int udp_queue_rcv_skb(struct sock *sk, struct sk_buff *skb) } } + prefetch(&sk->sk_rmem_alloc); if (rcu_access_pointer(sk->sk_filter) && udp_lib_checksum_complete(skb)) goto csum_error; |