From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753356Ab2JICUt (ORCPT ); Mon, 8 Oct 2012 22:20:49 -0400 Received: from e28smtp01.in.ibm.com ([122.248.162.1]:52480 "EHLO e28smtp01.in.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752572Ab2JICUq (ORCPT ); Mon, 8 Oct 2012 22:20:46 -0400 Message-ID: <507389F4.9060906@linux.vnet.ibm.com> Date: Tue, 09 Oct 2012 10:20:36 +0800 From: Michael Wang User-Agent: Mozilla/5.0 (X11; Linux i686; rv:15.0) Gecko/20120827 Thunderbird/15.0 MIME-Version: 1.0 To: Peter Zijlstra CC: linux-kernel@vger.kernel.org, mingo@redhat.com, wfg@linux.intel.com Subject: Re: [PATCH] sched: fix should_resched() to avoid do schedule in atomic References: <1347937984-5817-1-git-send-email-wangyun@linux.vnet.ibm.com> In-Reply-To: <1347937984-5817-1-git-send-email-wangyun@linux.vnet.ibm.com> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit x-cbid: 12100902-4790-0000-0000-000004FADCAF Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 09/18/2012 11:13 AM, Michael Wang wrote: > This patch try to fix the BUG: > > [ 0.043953] BUG: scheduling while atomic: swapper/0/1/0x10000002 > [ 0.044017] no locks held by swapper/0/1. > [ 0.044692] Pid: 1, comm: swapper/0 Not tainted 3.6.0-rc1-00420-gb7aebb9 #34 > [ 0.045861] Call Trace: > [ 0.048071] [] __schedule_bug+0x5e/0x70 > [ 0.048890] [] __schedule+0x91/0xb10 > [ 0.049660] [] ? vsnprintf+0x33a/0x450 > [ 0.050444] [] ? lg_local_lock+0x6/0x70 > [ 0.051256] [] ? wait_for_xmitr+0x31/0x90 > [ 0.052019] [] ? do_raw_spin_unlock+0xa5/0xf0 > [ 0.052903] [] ? _raw_spin_unlock+0x22/0x30 > [ 0.053759] [] ? up+0x1b/0x70 > [ 0.054421] [] __cond_resched+0x1b/0x30 > [ 0.055228] [] _cond_resched+0x45/0x50 > [ 0.056020] [] mutex_lock_nested+0x28/0x370 > [ 0.056884] [] ? console_unlock+0x3a2/0x4e0 > [ 0.057741] [] __irq_alloc_descs+0x39/0x1c0 > [ 0.058589] [] io_apic_setup_irq_pin+0x2c/0x310 > [ 0.060042] [] setup_IO_APIC+0x101/0x744 > [ 0.060878] [] ? clear_IO_APIC+0x31/0x50 > [ 0.061695] [] native_smp_prepare_cpus+0x538/0x680 > [ 0.062644] [] ? do_one_initcall+0x12c/0x12c > [ 0.063517] [] ? do_one_initcall+0x12c/0x12c > [ 0.064016] [] kernel_init+0x4b/0x17f > [ 0.064790] [] ? do_one_initcall+0x12c/0x12c > [ 0.065660] [] kernel_thread_helper+0x6/0x10 > > The process to trigger the BUG is: > > native_smp_prepare_cpus() > preempt_disable() //preempt_count++ > __irq_alloc_descs() > mutex_lock() > might_sleep() //should_resched() return true > __schedule() > preempt_disable() //preempt_count++ > schedule_bug() //preempt_count > 1, report bug > > So the issue is that should_resched() should not return true while the preempt > already disabled. > > This patch will fix the issue, then might_sleep() won't do schedule in atomic > any more. Could I get some comments on the patch? Regards, Michael Wang > > Reported-by: Fengguang Wu > Signed-off-by: Michael Wang > --- > kernel/sched/core.c | 2 +- > 1 files changed, 1 insertions(+), 1 deletions(-) > > diff --git a/kernel/sched/core.c b/kernel/sched/core.c > index b38f00e..2b7cd15 100644 > --- a/kernel/sched/core.c > +++ b/kernel/sched/core.c > @@ -4171,7 +4171,7 @@ SYSCALL_DEFINE0(sched_yield) > > static inline int should_resched(void) > { > - return need_resched() && !(preempt_count() & PREEMPT_ACTIVE); > + return need_resched() && !preempt_count(); > } > > static void __cond_resched(void) >