From mboxrd@z Thu Jan 1 00:00:00 1970 From: Sebastian Andrzej Siewior Subject: Re: [BUGREPORT] Tasklet scheduled issue in Linux 3.4.x-rt Date: Fri, 28 Mar 2014 17:37:53 +0100 Message-ID: <20140328163753.GB14842@linutronix.de> References: <53144A57.4040204@huawei.com> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Ingo Molnar , Peter Zijlstra , LKML , linux-rt-users@vger.kernel.org, Li Zefan , Zhangwei To: Yijing Wang Return-path: Content-Disposition: inline In-Reply-To: <53144A57.4040204@huawei.com> Sender: linux-kernel-owner@vger.kernel.org List-Id: linux-rt-users.vger.kernel.org * Yijing Wang | 2014-03-03 17:24:39 [+0800]: >[2012-03-26 18:55:43][ 929.252312] WARNING: at kernel/softirq.c:773 _= _tasklet_action+0x51/0x1a0() >[2012-03-27 03:41:06][ 3647.886005] WARNING: at kernel/softirq.c:773 _= _tasklet_action+0x51/0x1a0() >[2012-03-27 03:42:04][ 3705.434418] WARNING: at kernel/softirq.c:799 _= _tasklet_action+0xae/0x1a0() >FC card hardware -------> FC driver interrupt handler --------->task= let_schedule(fc driver tasklet) ------->tasklet running, call function = process FC IO data. > here will disable FC card interrupt = here will enable FC card inte= rrupt again This looks okay. >We found the tasklet state is 0x1(mean state is TASKLET_STATE_SCHED),c= ount is 0, before we call tasklet_schedule(). >So the new tasklet can not add to CPU list. > >And I also add some dynamic debug in __tasklet_action(); after the iss= ue occur, I open the dynamic debug. >After we force the hardware reset to interrupt OS, we never found the = =46C driver tasklet running in dmesg(I identify the tasklet by its data= ). >I guess the FC tasklet is not in CPU global tasklet list. You guess correct. >I hope somebody can help to look at it. If I missing something, let me= know. The tasklet is always added to the local cpu, never cross. That list is always accessed with interrupts off. With TASKLET_STATE_SCHED set, the next step is to add the task let to the CPU's tasklet list. This isn't done if TASKLET_STATE_RUN is already set which means __tasklet_action() is already busy serving the tasklet. In that case it clears TASKLET_STATE_SCHED and invokes the tasklet again. After looking at it for a while I must say I have no idea how you managed to keep TASKLET_STATE_SCHED set. Further, each time TASKLET_STATE_RUN is cleared it is always with a cmpxchg() down to zero which means TASKLET_STATE_SCHED is removed earlier. That said, triggerring the warning at 773 is the first thing that went wrong. After it has been added to the list, the TASKLET_STATE_RUN is cleared again. I have no idea how it managed to remain still on except that __tasklet_common_schedule() is invoked which is protected by the SCHED bit=E2=80=A6 >Thanks! >Yijing. Sebastian