From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752800AbdKHQ2S (ORCPT ); Wed, 8 Nov 2017 11:28:18 -0500 Received: from mail-qt0-f175.google.com ([209.85.216.175]:50033 "EHLO mail-qt0-f175.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752577AbdKHQ2R (ORCPT ); Wed, 8 Nov 2017 11:28:17 -0500 X-Google-Smtp-Source: ABhQp+S3BVr+3tMdU41he12Z56jHpyhfF+cvBXyaVMD7SLXn7QC8AAZ8QAVJakLA0395Gki8OhsxJw== Date: Wed, 8 Nov 2017 08:28:13 -0800 From: Tejun Heo To: Steven Rostedt Cc: Sergey Senozhatsky , Sergey Senozhatsky , Petr Mladek , linux-kernel@vger.kernel.org, kernel-team@fb.com, Linus Torvalds , Andrew Morton Subject: Re: [PATCH 2/2] Subject: printk: Don't trap random context in infinite log_buf flush Message-ID: <20171108162813.GA983427@devbig577.frc2.facebook.com> References: <20171102135146.GN3252168@devbig577.frc2.facebook.com> <20171102135258.GO3252168@devbig577.frc2.facebook.com> <20171104042408.GB539@tigerII.localdomain> <20171107002254.GI3252168@devbig577.frc2.facebook.com> <20171107020434.GC1822@jagdpanzerIV> <20171107132331.GJ3252168@devbig577.frc2.facebook.com> <20171108112229.200858cb@gandalf.local.home> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20171108112229.200858cb@gandalf.local.home> User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hello, Steven. On Wed, Nov 08, 2017 at 11:22:29AM -0500, Steven Rostedt wrote: > > Oh, I was talking about tweaking the repro, but I'm not sure the above > > would change anything. > > No, it would change an awful lot. > > It would cause two printers to access the consoles at the same time, > which would lead to unpredictable behavior. Sure, I'll give it a shot later but in the repro there literally is no context which is safe, so I kinda have a difficult time imagining how that will make things materially better. Neither of the producers can get scheduled out no matter they switch back and forth between them or not, so rcu stalls seem rather unavoidable without introducing an independent context. Thanks. -- tejun