Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751568AbbH2Cz0 (ORCPT ); Fri, 28 Aug 2015 22:55:26 -0400 Received: from e31.co.us.ibm.com ([32.97.110.149]:36883 "EHLO e31.co.us.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751525AbbH2CzY (ORCPT ); Fri, 28 Aug 2015 22:55:24 -0400 X-Helo: d03dlp02.boulder.ibm.com X-MailFrom: raghavendra.kt@linux.vnet.ibm.com X-RcptTo: netdev@vger.kernel.org Date: Sat, 29 Aug 2015 08:27:15 +0530 From: Raghavendra K T To: David Miller Cc: raghavendra.kt@linux.vnet.ibm.com, edumazet@google.com, kuznet@ms2.inr.ac.ru, jmorris@namei.org, yoshfuji@linux-ipv6.org, kaber@trash.net, jiri@resnulli.us, hannes@stressinduktion.org, tom@herbertland.com, azhou@nicira.com, ebiederm@xmission.com, ipm@chirality.org.uk, nicolas.dichtel@6wind.com, serge.hallyn@canonical.com, netdev@vger.kernel.org, linux-kernel@vger.kernel.org, anton@au1.ibm.com, nacc@linux.vnet.ibm.com, srikar@linux.vnet.ibm.com Subject: Re: [PATCH RFC V2 2/2] net: Optimize snmp stat aggregation by walking all the percpu data at once Message-ID: <20150829025715.GA5546@linux.vnet.ibm.com> Reply-To: Raghavendra K T References: <1440610653-14210-3-git-send-email-raghavendra.kt@linux.vnet.ibm.com> <20150827.113823.214019265460582055.davem@davemloft.net> <55E00238.10909@linux.vnet.ibm.com> <20150828.112413.424099339331017970.davem@davemloft.net> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline In-Reply-To: <20150828.112413.424099339331017970.davem@davemloft.net> User-Agent: Mutt/1.5.21 (2010-09-15) X-TM-AS-MML: disable X-Content-Scanned: Fidelis XPS MAILER x-cbid: 15082902-8236-0000-0000-00000EC00150 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 4810 Lines: 136 * David Miller [2015-08-28 11:24:13]: > From: Raghavendra K T > Date: Fri, 28 Aug 2015 12:09:52 +0530 > > > On 08/28/2015 12:08 AM, David Miller wrote: > >> From: Raghavendra K T > >> Date: Wed, 26 Aug 2015 23:07:33 +0530 > >> > >>> @@ -4641,10 +4647,12 @@ static inline void __snmp6_fill_stats64(u64 > >>> *stats, void __percpu *mib, > >>> static void snmp6_fill_stats(u64 *stats, struct inet6_dev *idev, int > >>> attrtype, > >>> int bytes) > >>> { > >>> + u64 buff[IPSTATS_MIB_MAX] = {0,}; > >>> + > ... > > hope you wanted to know the overhead than to change the current > > patch. please let me know.. > > I want you to change that variable initializer to an explicit memset(). > > The compiler is emitting a memset() or similar _anyways_. > > Not because it will have any impact at all upon performance, but because > of how it looks to people trying to read and understand the code. > > Hi David, resending the patch with memset. Please let me know if you want to resend all the patches. ----8<---- From: Raghavendra K T Subject: [PATCH RFC V2 2/2] net: Optimize snmp stat aggregation by walking all the percpu data at once Docker container creation linearly increased from around 1.6 sec to 7.5 sec (at 1000 containers) and perf data showed 50% ovehead in snmp_fold_field. reason: currently __snmp6_fill_stats64 calls snmp_fold_field that walks through per cpu data of an item (iteratively for around 90 items). idea: This patch tries to aggregate the statistics by going through all the items of each cpu sequentially which is reducing cache misses. Docker creation got faster by more than 2x after the patch. Result: Before After Docker creation time 6.836s 3.357s cache miss 2.7% 1.38% perf before: 50.73% docker [kernel.kallsyms] [k] snmp_fold_field 9.07% swapper [kernel.kallsyms] [k] snooze_loop 3.49% docker [kernel.kallsyms] [k] veth_stats_one 2.85% swapper [kernel.kallsyms] [k] _raw_spin_lock perf after: 10.56% swapper [kernel.kallsyms] [k] snooze_loop 8.72% docker [kernel.kallsyms] [k] snmp_get_cpu_field 7.59% docker [kernel.kallsyms] [k] veth_stats_one 3.65% swapper [kernel.kallsyms] [k] _raw_spin_lock Signed-off-by: Raghavendra K T --- net/ipv6/addrconf.c | 19 ++++++++++++++----- 1 file changed, 14 insertions(+), 5 deletions(-) Change in V2: - Allocate stat calculation buffer in stack (Eric) - Use memset to zero temp buffer (David) Thanks David and Eric for coments on V1 and as both of them pointed, unfortunately we cannot get rid of buffer for calculation without avoiding unaligned op. diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c index 21c2c81..9bdfba3 100644 --- a/net/ipv6/addrconf.c +++ b/net/ipv6/addrconf.c @@ -4624,16 +4624,22 @@ static inline void __snmp6_fill_statsdev(u64 *stats, atomic_long_t *mib, } static inline void __snmp6_fill_stats64(u64 *stats, void __percpu *mib, - int items, int bytes, size_t syncpoff) + int items, int bytes, size_t syncpoff, + u64 *buff) { - int i; + int i, c; int pad = bytes - sizeof(u64) * items; BUG_ON(pad < 0); /* Use put_unaligned() because stats may not be aligned for u64. */ put_unaligned(items, &stats[0]); + + for_each_possible_cpu(c) + for (i = 1; i < items; i++) + buff[i] += snmp_get_cpu_field64(mib, c, i, syncpoff); + for (i = 1; i < items; i++) - put_unaligned(snmp_fold_field64(mib, i, syncpoff), &stats[i]); + put_unaligned(buff[i], &stats[i]); memset(&stats[items], 0, pad); } @@ -4641,10 +4647,13 @@ static inline void __snmp6_fill_stats64(u64 *stats, void __percpu *mib, static void snmp6_fill_stats(u64 *stats, struct inet6_dev *idev, int attrtype, int bytes) { + u64 buff[IPSTATS_MIB_MAX]; + switch (attrtype) { case IFLA_INET6_STATS: - __snmp6_fill_stats64(stats, idev->stats.ipv6, - IPSTATS_MIB_MAX, bytes, offsetof(struct ipstats_mib, syncp)); + memset(buff, 0, sizeof(buff)); + __snmp6_fill_stats64(stats, idev->stats.ipv6, IPSTATS_MIB_MAX, bytes, + offsetof(struct ipstats_mib, syncp), buff); break; case IFLA_INET6_ICMP6STATS: __snmp6_fill_statsdev(stats, idev->stats.icmpv6dev->mibs, ICMP6_MIB_MAX, bytes); -- 1.7.11.7 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/