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 Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id BBF9DC25B0E for ; Mon, 15 Aug 2022 09:40:10 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S242060AbiHOJkJ (ORCPT ); Mon, 15 Aug 2022 05:40:09 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:60322 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S242063AbiHOJkE (ORCPT ); Mon, 15 Aug 2022 05:40:04 -0400 Received: from frasgout.his.huawei.com (frasgout.his.huawei.com [185.176.79.56]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 44D501F2D4; Mon, 15 Aug 2022 02:40:03 -0700 (PDT) Received: from fraeml714-chm.china.huawei.com (unknown [172.18.147.201]) by frasgout.his.huawei.com (SkyGuard) with ESMTP id 4M5q1V1fVHz67yfJ; Mon, 15 Aug 2022 17:37:02 +0800 (CST) Received: from lhrpeml500005.china.huawei.com (7.191.163.240) by fraeml714-chm.china.huawei.com (10.206.15.33) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2375.24; Mon, 15 Aug 2022 11:40:00 +0200 Received: from localhost (10.202.226.42) by lhrpeml500005.china.huawei.com (7.191.163.240) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2375.24; Mon, 15 Aug 2022 10:40:00 +0100 Date: Mon, 15 Aug 2022 10:39:58 +0100 From: Jonathan Cameron To: Yicong Yang CC: , , , , , , , , , , , , , , , , , , Subject: Re: [PATCH] drivers/perf: Change WARN_ON() to dev_err() on irq_set_affinity() failure Message-ID: <20220815103958.000016c9@huawei.com> In-Reply-To: <20220815092815.11597-1-yangyicong@huawei.com> References: <20220815092815.11597-1-yangyicong@huawei.com> X-Mailer: Claws Mail 4.0.0 (GTK+ 3.24.29; i686-w64-mingw32) MIME-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit X-Originating-IP: [10.202.226.42] X-ClientProxiedBy: lhrpeml500001.china.huawei.com (7.191.163.213) To lhrpeml500005.china.huawei.com (7.191.163.240) X-CFilter-Loop: Reflected Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Mon, 15 Aug 2022 17:28:15 +0800 Yicong Yang wrote: > From: Yicong Yang > > The WARN_ON() on irq_set_affinity() failure is misused according to the [1] > and may crash people's box unintentionally. This may also be redundant since > in the failure case we may also trigger the WARN and dump the stack in the > perf core[2] for a second time. > > So change the WARN_ON() to dev_err() to just print the failure message. > > [1] https://github.com/torvalds/linux/blob/master/include/asm-generic/bug.h#L74 > [2] https://github.com/torvalds/linux/blob/master/kernel/events/core.c#L313 > > Suggested-by: Greg KH > [https://lore.kernel.org/lkml/YuOi3i0XHV++z1YI@kroah.com/] > Signed-off-by: Yicong Yang Looks like progress in a sensible direction to me. Reviewed-by: Jonathan Cameron Kind of unrelated question inline. > > diff --git a/drivers/perf/arm_smmuv3_pmu.c b/drivers/perf/arm_smmuv3_pmu.c > index 00d4c45a8017..05e1b3e274d7 100644 > --- a/drivers/perf/arm_smmuv3_pmu.c > +++ b/drivers/perf/arm_smmuv3_pmu.c > @@ -646,7 +646,8 @@ static int smmu_pmu_offline_cpu(unsigned int cpu, struct hlist_node *node) > > perf_pmu_migrate_context(&smmu_pmu->pmu, cpu, target); > smmu_pmu->on_cpu = target; > - WARN_ON(irq_set_affinity(smmu_pmu->irq, cpumask_of(target))); > + if (irq_set_affinity(smmu_pmu->irq, cpumask_of(target))) > + dev_err(smmu_pmu->dev, "Failed to set interrupt affinity\n"); > > return 0; > } > @@ -892,7 +893,8 @@ static int smmu_pmu_probe(struct platform_device *pdev) > > /* Pick one CPU to be the preferred one to use */ > smmu_pmu->on_cpu = raw_smp_processor_id(); > - WARN_ON(irq_set_affinity(smmu_pmu->irq, cpumask_of(smmu_pmu->on_cpu))); > + if (irq_set_affinity(smmu_pmu->irq, cpumask_of(smmu_pmu->on_cpu))) > + dev_err(dev, "Failed to set interrupt affinity\n"); In this case we have the option to fail probe. Failing to set affinity means we are broken anyway, so perhaps that is cleaner than carrying on. As a side note, I wonder if other drivers could benefit from what I think is a micro optimization to short cut calling the hp handlers when the decision of which CPU is easy... > > err = cpuhp_state_add_instance_nocalls(cpuhp_state_num, > &smmu_pmu->node);