From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751240AbeEBNkf (ORCPT ); Wed, 2 May 2018 09:40:35 -0400 Received: from mail-vk0-f65.google.com ([209.85.213.65]:36701 "EHLO mail-vk0-f65.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750936AbeEBNkd (ORCPT ); Wed, 2 May 2018 09:40:33 -0400 X-Google-Smtp-Source: AB8JxZoToWWGgxP9ZX4AJUTd6l/2hCpulki4HE7Q9EsjksoioX0iKkyKRQPZ31g7GUK+qvR/sZNpGzT76I8SgaeFBsU= MIME-Version: 1.0 In-Reply-To: <20180426103133.GA6953@linaro.org> References: <20180412091822.GG12256@bigcity.dyn.berto.se> <20180412111519.GH12256@bigcity.dyn.berto.se> <20180412133031.GA551@linaro.org> <20180412223904.GJ12256@bigcity.dyn.berto.se> <20180420160013.GA13769@linaro.org> <20180422221827.GB27674@bigcity.dyn.berto.se> <20180423095420.GA23995@linaro.org> <20180425225603.GA26177@bigcity.dyn.berto.se> <20180426103133.GA6953@linaro.org> From: Geert Uytterhoeven Date: Wed, 2 May 2018 15:40:31 +0200 X-Google-Sender-Auth: TjGrEOftXi4he1oBMEmiDGs0370 Message-ID: Subject: Re: Potential problem with 31e77c93e432dec7 ("sched/fair: Update blocked load when newly idle") To: Vincent Guittot Cc: =?UTF-8?Q?Niklas_S=C3=B6derlund?= , Heiner Kallweit , Peter Zijlstra , "Paul E. McKenney" , Ingo Molnar , linux-kernel , Linux-Renesas Content-Type: text/plain; charset="UTF-8" Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Transfer-Encoding: 8bit X-MIME-Autoconverted: from quoted-printable to 8bit by mail.home.local id w42Degeo017972 Hi Vincent, On Thu, Apr 26, 2018 at 12:31 PM, Vincent Guittot wrote: > Le Thursday 26 Apr 2018 à 00:56:03 (+0200), Niklas Söderlund a écrit : >> Here are the result, sorry for the delay. >> >> On 2018-04-23 11:54:20 +0200, Vincent Guittot wrote: >> >> [snip] >> >> > >> > Thanks for the report. Can you re run with the following trace-cmd sequence ? My previous sequence disables ftrace events >> > >> > trace-cmd reset > /dev/null >> > trace-cmd start -b 40000 -p function -l dump_backtrace:traceoff -e sched -e cpu_idle -e cpu_frequency -e timer -e ipi -e irq -e printk >> > trace-cmd start -b 40000 -p function -l dump_backtrace -e sched -e cpu_idle -e cpu_frequency -e timer -e ipi -e irq -e printk >> > >> > I have updated the patch and added traces to check that scheduler returns from idle_balance function and doesn't stay stuck >> >> Once more I applied the change bellow on-top of c18bb396d3d261eb ("Merge >> git://git.kernel.org/pub/scm/linux/kernel/git/davem/net"). >> >> This time the result of 'trace-cmd report' is so large I do not include >> it here, but I attach the trace.dat file. Not sure why but the timing of >> sending the NMI to the backtrace print is different (but content the >> same AFIK) so in the odd change it can help figure this out: >> > > Thanks for the trace, I have been able to catch a problem with it. > Could you test the patch below to confirm that the problem is solved ? > The patch apply on-top of > c18bb396d3d261eb ("Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net") > > From: Vincent Guittot > Date: Thu, 26 Apr 2018 12:19:32 +0200 > Subject: [PATCH] sched/fair: fix the update of blocked load when newly idle > MIME-Version: 1.0 > Content-Type: text/plain; charset=UTF-8 > Content-Transfer-Encoding: 8bit > > With commit 31e77c93e432 ("sched/fair: Update blocked load when newly idle"), > we release the rq->lock when updating blocked load of idle CPUs. This open > a time window during which another CPU can add a task to this CPU's cfs_rq. > The check for newly added task of idle_balance() is not in the common path. > Move the out label to include this check. > > Fixes: 31e77c93e432 ("sched/fair: Update blocked load when newly idle") > Reported-by: Heiner Kallweit > Reported-by: Niklas Söderlund > Signed-off-by: Vincent Guittot Thanks! Tested-by: Geert Uytterhoeven Gr{oetje,eeting}s, Geert -- Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that. -- Linus Torvalds