From mboxrd@z Thu Jan 1 00:00:00 1970 From: Hangbin Liu Subject: Re: [PATCHv4 iproute2 1/2] lib/libnetlink: re malloc buff if size is not enough Date: Sat, 30 Sep 2017 21:54:56 +0800 Message-ID: <20170930135455.GW1754@leo.usersys.redhat.com> References: <1506605626-1744-1-git-send-email-haliu@redhat.com> <1506605626-1744-2-git-send-email-haliu@redhat.com> <20170929105440.7edaec1f@xeon-e3> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: netdev@vger.kernel.org, Michal Kubecek , Phil Sutter , Hangbin Liu To: Stephen Hemminger Return-path: Received: from mx1.redhat.com ([209.132.183.28]:44384 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751028AbdI3NzC (ORCPT ); Sat, 30 Sep 2017 09:55:02 -0400 Content-Disposition: inline In-Reply-To: <20170929105440.7edaec1f@xeon-e3> Sender: netdev-owner@vger.kernel.org List-ID: Hi Stephen, On Fri, Sep 29, 2017 at 10:54:40AM -0700, Stephen Hemminger wrote: > > Doubling the number of system calls per message is not going to make > users with 5,000,000 routes or 1000 vlans, or 10,000 tunnels happy. > Please rethink this. I tried to add 2500 vlans and 70,000 routes. Then show the result. The time looks reasonable. # ip link show | wc -l 5024 # time ip link show > /dev/null real 0m0.218s user 0m0.007s sys 0m0.210s # time iproute2/ip/ip link show > /dev/null real 0m0.221s user 0m0.008s sys 0m0.212s # time ip addr show > /dev/null real 0m0.299s user 0m0.094s sys 0m0.205s # time iproute2/ip/ip addr show > /dev/null real 0m0.302s user 0m0.099s sys 0m0.202s # ip -6 route show | wc -l 704458 # time ip -6 route show > /dev/null real 0m5.400s user 0m0.947s sys 0m4.453s # time iproute2/ip/ip -6 route show > /dev/null real 0m5.404s user 0m1.070s sys 0m4.333s Thanks Hangbin