From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Dumazet Subject: [PATCH] net: reorder fields of struct socket Date: Sat, 14 Mar 2009 10:03:54 +0100 Message-ID: <49BB72FA.1080401@cosmosbay.com> References: <49B4B909.7050002@cosmosbay.com> <20090313.145152.121603300.davem@davemloft.net> <49BADE87.40407@cosmosbay.com> <20090313.153851.11725991.davem@davemloft.net> <49BAE1F3.5020105@cosmosbay.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: kchang@athenacr.com, netdev@vger.kernel.org, bmb@athenacr.com To: David Miller Return-path: Received: from gw1.cosmosbay.com ([212.99.114.194]:49758 "EHLO gw1.cosmosbay.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751198AbZCNJEG (ORCPT ); Sat, 14 Mar 2009 05:04:06 -0400 In-Reply-To: <49BAE1F3.5020105@cosmosbay.com> Sender: netdev-owner@vger.kernel.org List-ID: On x86_64, its rather unfortunate that "wait_queue_head_t wait" field of "struct socket" spans two cache lines (assuming a 64 bytes cache line in current cpus) offsetof(struct socket, wait)=0x30 sizeof(wait_queue_head_t)=0x18 This might explain why Kenny Chang noticed that his multicast workload was performing bad with 64 bit kernels, since more cache lines ping pongs were involved. This litle patch moves "wait" field next "fasync_list" so that both fields share a single cache line, to speedup sock_def_readable() Signed-off-by: Eric Dumazet diff --git a/include/linux/net.h b/include/linux/net.h index 4515efa..4fc2ffd 100644 --- a/include/linux/net.h +++ b/include/linux/net.h @@ -129,11 +129,15 @@ struct socket { socket_state state; short type; unsigned long flags; - const struct proto_ops *ops; + /* + * Please keep fasync_list & wait fields in the same cache line + */ struct fasync_struct *fasync_list; + wait_queue_head_t wait; + struct file *file; struct sock *sk; - wait_queue_head_t wait; + const struct proto_ops *ops; }; struct vm_area_struct;