From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-0.8 required=3.0 tests=HEADER_FROM_DIFFERENT_DOMAINS, MAILING_LIST_MULTI,SPF_PASS autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 02D6FC6778A for ; Tue, 3 Jul 2018 14:54:57 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 9EFDA21886 for ; Tue, 3 Jul 2018 14:54:56 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 9EFDA21886 Authentication-Results: mail.kernel.org; dmarc=none (p=none dis=none) header.from=goodmis.org Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=linux-kernel-owner@vger.kernel.org Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932517AbeGCOyx (ORCPT ); Tue, 3 Jul 2018 10:54:53 -0400 Received: from mail.kernel.org ([198.145.29.99]:57916 "EHLO mail.kernel.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932209AbeGCOyw (ORCPT ); Tue, 3 Jul 2018 10:54:52 -0400 Received: from gandalf.local.home (cpe-66-24-56-78.stny.res.rr.com [66.24.56.78]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.kernel.org (Postfix) with ESMTPSA id 15E372084D; Tue, 3 Jul 2018 14:54:50 +0000 (UTC) Date: Tue, 3 Jul 2018 10:54:49 -0400 From: Steven Rostedt To: LKML Cc: Ingo Molnar , Peter Zijlstra , Thomas Gleixner , Juri Lelli , Claudio Scordino , Viresh Kumar , Clark Williams Subject: [BUG] Caused by: sched/deadline: Move CPU frequency selection triggering points Message-ID: <20180703105449.2a211fb2@gandalf.local.home> X-Mailer: Claws Mail 3.16.0 (GTK+ 2.24.32; x86_64-pc-linux-gnu) MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org When looking to test SCHED_DEADLINE, I triggered a lockup. The lockup appears to be caused by WARN_ON() done inside the scheduling path, and I'm guessing it tried to grab the rq lock and caused a deadlock (all I would get would be the "--- cut here ---" line, and then nothing after that. But a bit of playing with the printks() I figured out it was happening at: assert_clock_updated() { SCHED_WARN_ON(rq->clock_update_flags < RQCF_ACT_SKIP); } I bisected it down to commit e0367b126 ("sched/deadline: Move CPU frequency selection triggering points"). Reverting it indeed makes the deadlock go away. I commented out the WARN_ON that was being triggered, and that let the system continue to run. I did the following change: --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -914,7 +914,12 @@ static inline void assert_clock_updated(struct rq *rq) * The only reason for not seeing a clock update since the * last rq_pin_lock() is if we're currently skipping updates. */ - SCHED_WARN_ON(rq->clock_update_flags < RQCF_ACT_SKIP); + if (rq->clock_update_flags < RQCF_ACT_SKIP) + trace_printk("WARN_ON: [%d] rq->clock_update_flags (%d) < %d)\n", + rq->cpu, rq->clock_update_flags, RQCF_ACT_SKIP); + else + trace_printk("GOOD: [%d] rq->clock_update_flags (%d) >= %d)\n", + rq->cpu, rq->clock_update_flags, RQCF_ACT_SKIP); } To see what was happening. I also, added trace_printk()s to all the updates to clock_update_flags, and ran my test again. Here's what I got: deadline_test-1393 [002] 162.127132: bprint: push_dl_task.part.40: WARN_ON: [1] rq->clock_update_flags (0) < 2) deadline_test-1393 [002] 162.127133: bprint: update_rq_clock: [1] clock was 0 now 4 deadline_test-1393 [002] 162.127134: bprint: rq_clock: GOOD: [1] rq->clock_update_flags (4) >= 2) It appears that we hit this in this path: push_dl_task { add_running_bw() { __add_running_bw() { cpufreq_update_util() { data->func(data, rq_clock(rq), flags); rq_clock() { assert_clock_updated() And here the clock isn't updated and we get the splat. Reverting the stated patch works because it added the call to cpufreq_update_util() that does the rq_clock() at an inappropriate time, which causes the splat. I'm not sure what the right answer to this is. Reverting obviously works, but I'm also guessing a proper placement of update_rq_clock() may also work. I just don't know where that placement is, as I don't understand the rq_clock() updates enough. Help? -- Steve