From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932516Ab1LPSsD (ORCPT ); Fri, 16 Dec 2011 13:48:03 -0500 Received: from mx1.cpanel.net ([208.74.121.68]:53836 "EHLO mx1.cpanel.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1760395Ab1LPSr5 convert rfc822-to-8bit (ORCPT ); Fri, 16 Dec 2011 13:47:57 -0500 From: Nate Custer Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 8BIT Subject: Re: [RFT PATCH] blkio: alloc per cpu data from worker thread context( Re: kvm deadlock) Date: Fri, 16 Dec 2011 12:47:55 -0600 Message-Id: <7E70F89E-F763-4642-8F2F-5FA1992C7778@cpanel.net> Cc: Jens Axboe , Avi Kivity , Marcelo Tosatti , kvm@vger.kernel.org, linux-kernel To: Vivek Goyal Mime-Version: 1.0 (Apple Message framework v1084) X-Mailer: Apple Mail (2.1084) X-AntiAbuse: This header was added to track abuse, please include it with any abuse report X-AntiAbuse: Primary Hostname - mx1.cpanel.net X-AntiAbuse: Original Domain - vger.kernel.org X-AntiAbuse: Originator/Caller UID/GID - [47 12] / [47 12] X-AntiAbuse: Sender Address Domain - cpanel.net Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Dec 15, 2011, at 1:47 PM, Vivek Goyal wrote: > Ok, I continued to develop on the patch which tries to allocate per cpu > stats from worker thread context. Here is the patch. > > Can the reporter please try out the patch and see if it helps. I am not > sure if deadlock was because of mutex issue or not, but it should help > get rid of lockdep warning. > > This patch is on top of for-3.3/core branch of jens's linux-block tree. > If it does not apply on your kernel version, do let me know the version > you are testing with and I will generate another version of patch. > > If testing results are good, I will break down the patch in small pieces > and post as a series separately. > > Thanks > Vivek Running on a fedora-16 box with the patch applied to the linux-block tree I still had deadlocks. In fact it seemed to happen much faster and with ligher workloads. I was able to get netconsole setup and a full stacktrace is posted here: http://pastebin.com/9Rq68exU Nate