From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753656AbaCaKZe (ORCPT ); Mon, 31 Mar 2014 06:25:34 -0400 Received: from www.linutronix.de ([62.245.132.108]:58853 "EHLO Galois.linutronix.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753256AbaCaKZc (ORCPT ); Mon, 31 Mar 2014 06:25:32 -0400 Message-ID: <53394296.5030906@linutronix.de> Date: Mon, 31 Mar 2014 12:25:26 +0200 From: Sebastian Andrzej Siewior User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:24.0) Gecko/20100101 Icedove/24.4.0 MIME-Version: 1.0 To: Yijing Wang CC: Ingo Molnar , Peter Zijlstra , LKML , linux-rt-users@vger.kernel.org, Li Zefan , Zhangwei Subject: Re: [BUGREPORT] Tasklet scheduled issue in Linux 3.4.x-rt References: <53144A57.4040204@huawei.com> <20140328163753.GB14842@linutronix.de> <533669C7.0@huawei.com> In-Reply-To: <533669C7.0@huawei.com> X-Enigmail-Version: 1.6 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit X-Linutronix-Spam-Score: -1.0 X-Linutronix-Spam-Level: - X-Linutronix-Spam-Status: No , -1.0 points, 5.0 required, ALL_TRUSTED=-1,SHORTCIRCUIT=-0.0001 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 03/29/2014 07:35 AM, Yijing Wang wrote: > Hi Sebastian, Hi Yijing, > Thanks for your reply and help to look at it, thanks! > > I also check the tasklet state machine changes, and didn't find > clue for this issue. So I Temporarily reverted Ingo's patch, without > this patch, my test is ok. > > Because this patch does not exist in the latest Linus kernel, so I > have not reported this issue to kernel bugzilla. This patch exists in all -RT releases up to 3.12. If there is an issue with it, it should be solved. If the sched bit set is and you can't get lock later then the tasklet has be to active. Finally, not getting the lock in the tasklet code itself means it is still occupied by the "add-to-the-list" part which actually can't happen according to the code. You said, that you have an eight-way. Is this also NUMA? If so, does this problem happen if you disable NUMA (i.e. run only one NUMA node and use only the memory that is directly attached to the node). > Finally, I would like to thank you again. > > Thanks! > Yijing. Sebastian