From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751785Ab1LKKrz (ORCPT ); Sun, 11 Dec 2011 05:47:55 -0500 Received: from mx2.mail.elte.hu ([157.181.151.9]:45890 "EHLO mx2.mail.elte.hu" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751519Ab1LKKrx (ORCPT ); Sun, 11 Dec 2011 05:47:53 -0500 Date: Sun, 11 Dec 2011 11:46:07 +0100 From: Ingo Molnar To: Pekka Enberg Cc: Thomas Gleixner , LKML , Sasha Levin Subject: Re: [patch 0/3] kvm tool: Serial emulation overhaul Message-ID: <20111211104607.GA10351@elte.hu> References: <20111210132220.083204833@linutronix.de> <20111211103047.GA19299@elte.hu> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20111211103047.GA19299@elte.hu> User-Agent: Mutt/1.5.21 (2010-09-15) X-ELTE-SpamScore: -2.0 X-ELTE-SpamLevel: X-ELTE-SpamCheck: no X-ELTE-SpamVersion: ELTE 2.0 X-ELTE-SpamCheck-Details: score=-2.0 required=5.9 tests=AWL,BAYES_00 autolearn=no SpamAssassin version=3.3.1 -2.0 BAYES_00 BODY: Bayes spam probability is 0 to 1% [score: 0.0000] 0.0 AWL AWL: From: address is in the auto white-list Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org btw., i just noticed another performance weirdness, running the latest version on -tip with such a 64-bit x86 kernel: make defconfig make kvmconfig # sidenote: sigh - this still needs a 'make oldconfig' and do a 'kvm run' into a shell, then I get a permanently busy kvm executable sucking up CPU time: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 28206 mingo 20 0 3586m 80m 1080 S 56.1 0.7 0:13.83 kvm 29462 mingo 20 0 15380 1392 900 R 0.3 0.0 0:00.02 top perf kvm top suggests simple idling around of 16 CPUs: -------------------------------------------------------------------------------------------------- PerfTop: 1635 irqs/sec kernel:91.1% us: 0.3% guest kernel: 8.5% guest us: 0.0% exact: 0.0% [1000Hz cycles], (all, 16 CPUs) -------------------------------------------------------------------------------------------------- samples pcnt function DSO _______ _____ _________________________ _______________________________ 155.00 17.9% apic_timer_interrupt /home/mingo/linux/linux/vmlinux 89.00 10.3% find_busiest_group /home/mingo/linux/linux/vmlinux 66.00 7.6% native_apic_mem_write /home/mingo/linux/linux/vmlinux 51.00 5.9% idle_cpu /home/mingo/linux/linux/vmlinux 29.00 3.3% __rcu_pending /home/mingo/linux/linux/vmlinux 25.00 2.9% run_posix_cpu_timers /home/mingo/linux/linux/vmlinux 24.00 2.8% _raw_spin_lock /home/mingo/linux/linux/vmlinux 19.00 2.2% cpumask_next_and /home/mingo/linux/linux/vmlinux 19.00 2.2% ktime_get /home/mingo/linux/linux/vmlinux 18.00 2.1% run_timer_softirq /home/mingo/linux/linux/vmlinux 16.00 1.8% run_rebalance_domains /home/mingo/linux/linux/vmlinux 16.00 1.8% native_sched_clock /home/mingo/linux/linux/vmlinux 15.00 1.7% rcu_exit_nohz /home/mingo/linux/linux/vmlinux 14.00 1.6% tick_nohz_stop_sched_tick /home/mingo/linux/linux/vmlinux 14.00 1.6% hrtimer_run_queues /home/mingo/linux/linux/vmlinux NO_HZ is set in the .config. Btw., 'perf kvm' is in need of some love: for example it is unable to pick up a vmlinux from the current directory and ./vmlinux does not work either, it needs /home/mingo/linux/linux/vmlinux specified explicitly - that sucks. Ideally perf could be taught to talk to the guest kvm process and get its vmlinux home position (even better: its guest kallsyms) from there - so if i typed 'perf kvm top' it would do the right thing all automagically. [ What a fantastc level of automation! ;-) ] Thanks, Ingo