From mboxrd@z Thu Jan 1 00:00:00 1970 Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755035AbeAOJNr (ORCPT + 1 other); Mon, 15 Jan 2018 04:13:47 -0500 Received: from Galois.linutronix.de ([146.0.238.70]:39328 "EHLO Galois.linutronix.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755007AbeAOJNo (ORCPT ); Mon, 15 Jan 2018 04:13:44 -0500 Date: Mon, 15 Jan 2018 10:13:29 +0100 (CET) From: Thomas Gleixner To: Keith Busch cc: LKML Subject: Re: [BUG 4.15-rc7] IRQ matrix management errors In-Reply-To: <20180115030255.GA13921@localhost.localdomain> Message-ID: References: <20180115025759.GG13580@localhost.localdomain> <20180115030255.GA13921@localhost.localdomain> User-Agent: Alpine 2.20 (DEB 67 2015-01-07) MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Return-Path: On Sun, 14 Jan 2018, Keith Busch wrote: > I hoped to have a better report before the weekend, but I've run out of > time and without my machine till next week, so sending what I have and > praying someone more in the know will have a better clue. > > I've a few NVMe drives and occasionally the IRQ teardown and bring-up > is failing. Resetting the controllers to re-run interrupt allocation > results in device interrupts from not occuring at all. > > It appears I need at least 2 NVMe drives for the problem. The NVMe > driver initializes controllers in parallel, and multiple threads calling > pci_alloc_irq_vectors and/or pci_free_vectors at about the same time > seems required to trigger the issue. > > Here are the relavent warnings. I also have trace events from irq_matrix > and irq_vectors, but from a different run that won't match up to the > timestamps below, but can send if that's helpful. The dmesg is not that interesting. The traces definitely are if you can identify the point where it goes into lala land. Thanks, tglx