From mboxrd@z Thu Jan 1 00:00:00 1970 From: Sebastian Andrzej Siewior Subject: Re: Large latencies with 4.4.1-rt6 Date: Thu, 25 Feb 2016 16:29:30 +0100 Message-ID: <20160225152930.GC16161@linutronix.de> References: <1A2AF5F768264E479963C4D8C015EB691B692C5C@dlrexmbx03.intra.dlr.de> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Cc: linux-rt-users@vger.kernel.org To: Martin.Wirth@dlr.de Return-path: Received: from www.linutronix.de ([62.245.132.108]:54023 "EHLO Galois.linutronix.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1760202AbcBYP3e (ORCPT ); Thu, 25 Feb 2016 10:29:34 -0500 Content-Disposition: inline In-Reply-To: <1A2AF5F768264E479963C4D8C015EB691B692C5C@dlrexmbx03.intra.dlr.de> Sender: linux-rt-users-owner@vger.kernel.org List-ID: * Martin.Wirth@dlr.de | 2016-02-19 17:02:20 [+0000]: >Hi, Hi, >I today switched from 4.1.15-rt17 to 4.4.1-rt6. >Now I'm seeing large latencies of more than 1 ms. >They occur sporadic ( every few minutes) and show up for example running > >cyclictest -l10000000 -m -Sp99 -i200 > >I tried to catch a trace and got the following: > > > <...>-3709 2d...2.. 2us : set_next_entity <-pick_next_task_fair > <...>-3709 2d...2.. 2us : update_stats_wait_end <-set_next_entity > firefox-3736 2d...2.. 3us : finish_task_switch <-__schedule > firefox-3736 2d...2.. 3us#: _raw_spin_unlock_irq <-finish_task_switch > firefox-3736 2d...... 1091us : smp_apic_timer_interrupt <-apic_timer_interrupt > firefox-3736 2d...... 1092us : irq_enter <-smp_apic_timer_interrupt >Does the above trace ring a bell or is it useless? This does not ring a bell. According to the trace your CPU took a break after the unlock until the next interrupt. If this a fucntion trace I would expect some functions after that. You could try the v4.4 line as of: v4.4-rt2 v4.4-rt3 v4.4.1-rt5 to see if I introduced this somewhere by accident. However I haven't seen this on any of my boxes. >Thanks, > >Martin Sebastian