Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S934416Ab3CHQqD (ORCPT ); Fri, 8 Mar 2013 11:46:03 -0500 Received: from e7.ny.us.ibm.com ([32.97.182.137]:50921 "EHLO e7.ny.us.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S933517Ab3CHQqA (ORCPT ); Fri, 8 Mar 2013 11:46:00 -0500 Date: Fri, 8 Mar 2013 08:44:35 -0800 From: "Paul E. McKenney" To: Sasha Levin Cc: Steven Rostedt , Frederic Weisbecker , Thomas Gleixner , Ingo Molnar , Andrew Morton , paul.gortmaker@windriver.com, Dave Jones , "linux-kernel@vger.kernel.org" Subject: Re: irq_work: WARNING: at kernel/irq_work.c:98 irq_work_needs_cpu+0x8a/0xb0() Message-ID: <20130308164435.GI3268@linux.vnet.ibm.com> Reply-To: paulmck@linux.vnet.ibm.com References: <51397B96.7030008@oracle.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <51397B96.7030008@oracle.com> User-Agent: Mutt/1.5.21 (2010-09-15) X-TM-AS-MML: No X-Content-Scanned: Fidelis XPS MAILER x-cbid: 13030816-5806-0000-0000-00002044E1E1 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 7567 Lines: 119 On Fri, Mar 08, 2013 at 12:48:06AM -0500, Sasha Levin wrote: > Hi guys, > > While fuzzing with trinity inside a KVM tools guest it seemed to have hit the > new warning in kernel/irq_work.c: > > [ 486.527075] ------------[ cut here ]------------ > [ 486.527788] WARNING: at kernel/irq_work.c:98 irq_work_needs_cpu+0x8a/0xb0() > [ 486.528870] Modules linked in: > [ 486.529377] Pid: 0, comm: swapper/2 Tainted: G W 3.9.0-rc1-next-20130307-sasha-00047-g0a7d304-dirty #1037 > [ 486.530165] Call Trace: > [ 486.530165] [] warn_slowpath_common+0x8c/0xc0 > [ 486.530165] [] warn_slowpath_null+0x15/0x20 > [ 486.530165] [] irq_work_needs_cpu+0x8a/0xb0 > [ 486.530165] [] tick_nohz_stop_sched_tick+0x95/0x2a0 > [ 486.530165] [] __tick_nohz_idle_enter+0x189/0x1b0 > [ 486.530165] [] tick_nohz_idle_enter+0xa1/0xd0 > [ 486.530165] [] cpu_idle+0x77/0x180 > [ 486.530165] [] ? setup_APIC_timer+0xc9/0xce > [ 486.530165] [] start_secondary+0xe1/0xe8 > [ 486.530165] ---[ end trace dd075f5cfc2c4f26 ]--- > > Obviously this was happening when trinity tried to exercise the shutdown syscall. > > It was followed by RCU choking and causing a bunch of locked tasks, preventing > shutdown. I guess it's the result of whatever caused this warning above to > happen, but in-case it isn't, the relevant parts of the RCU hang are: > > [ 607.040283] INFO: task init:1 blocked for more than 120 seconds. > [ 607.042932] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. > [ 607.047046] init D ffff8800ba308000 2736 1 0 0x00000000 > [ 607.050498] ffff8800ba311b18 0000000000000002 0000000000000003 ffff8800ba308000 > [ 607.055110] ffff8800ba310000 ffff8800ba310010 ffff8800ba311fd8 ffff8800ba310000 > [ 607.058208] ffff8800ba310010 ffff8800ba311fd8 ffffffff8542c420 ffff8800ba308000 > [ 607.060611] Call Trace: > [ 607.060847] [] ? __mutex_lock_common+0x365/0x5d0 > [ 607.061462] [] schedule+0x55/0x60 > [ 607.061948] [] schedule_preempt_disabled+0x13/0x20 > [ 607.062590] [] __mutex_lock_common+0x3a5/0x5d0 > [ 607.063209] [] ? rcu_cleanup_dead_cpu+0x52/0x250 > [ 607.063840] [] ? free_cpumask_var+0x9/0x10 > [ 607.064453] [] ? rcu_cleanup_dead_cpu+0x52/0x250 > [ 607.065035] [] mutex_lock_nested+0x40/0x50 > [ 607.065606] [] rcu_cleanup_dead_cpu+0x52/0x250 > [ 607.066230] [] ? trace_hardirqs_on+0xd/0x10 > [ 607.066810] [] rcu_cpu_notify+0x1b4/0x1ef > [ 607.067375] [] notifier_call_chain+0xee/0x130 > [ 607.067975] [] __raw_notifier_call_chain+0x9/0x10 > [ 607.068631] [] __cpu_notify+0x1b/0x30 > [ 607.069165] [] cpu_notify_nofail+0x10/0x30 > [ 607.069749] [] _cpu_down+0x185/0x2e0 > [ 607.070319] [] disable_nonboot_cpus+0x88/0x1b0 > [ 607.070937] [] kernel_restart+0x16/0x60 > [ 607.071487] [] SYSC_reboot+0x18c/0x2a0 > [ 607.072020] [] ? rcu_cleanup_after_idle+0x23/0xf0 > [ 607.072635] [] ? rcu_eqs_exit_common+0x64/0x280 > [ 607.073251] [] ? user_exit+0xc5/0x100 > [ 607.073772] [] ? trace_hardirqs_on+0xd/0x10 > [ 607.074352] [] ? syscall_trace_enter+0x23/0x290 > [ 607.075054] [] SyS_reboot+0x9/0x10 > [ 607.075495] [] tracesys+0xdd/0xe2 > [ 607.075967] 4 locks held by init/1: > [ 607.076439] #0: (reboot_mutex){+.+.+.}, at: [] SYSC_reboot+0xe6/0x2a0 > [ 607.077276] #1: (cpu_add_remove_lock){+.+.+.}, at: [] cpu_maps_update_begin+0x12/0x20 > [ 607.078288] #2: (cpu_hotplug.lock){+.+.+.}, at: [] cpu_hotplug_begin+0x27/0x60 > [ 607.079260] #3: (rcu_preempt_state.onoff_mutex){+.+...}, at: [] rcu_cleanup_dead_cpu+0x52/0x250 > > [ 607.187177] rcu_preempt D ffff8800aa8884a8 5136 11 2 0x00000000 > [ 607.187890] ffff8800ba391c08 0000000000000002 ffff8800ba391bb8 000000078117e00a > [ 607.188674] ffff8800ba390000 ffff8800ba390010 ffff8800ba391fd8 ffff8800ba390000 > [ 607.189472] ffff8800ba390010 ffff8800ba391fd8 ffff8800ba308000 ffff8800ba388000 > [ 607.190581] Call Trace: > [ 607.190849] [] schedule+0x55/0x60 > [ 607.191336] [] schedule_timeout+0x276/0x2c0 > [ 607.191904] [] ? lock_timer_base+0x70/0x70 > [ 607.192460] [] schedule_timeout_uninterruptible+0x19/0x20 > [ 607.193132] [] rcu_gp_init+0x438/0x490 > [ 607.193646] [] ? trace_hardirqs_on+0xd/0x10 > [ 607.194216] [] rcu_gp_kthread+0xbc/0x2d0 > [ 607.194760] [] ? rcu_gp_init+0x490/0x490 > [ 607.195298] [] ? wake_up_bit+0x40/0x40 > [ 607.195823] [] ? rcu_gp_init+0x490/0x490 > [ 607.196364] [] kthread+0xe2/0xf0 > [ 607.196842] [] ? __lock_release+0x1da/0x1f0 > [ 607.197405] [] ? __init_kthread_worker+0x70/0x70 > [ 607.198022] [] ret_from_fork+0x7c/0xb0 > [ 607.198559] [] ? __init_kthread_worker+0x70/0x70 > > [ 609.414891] Showing all locks held in the system: > [ 609.415490] 4 locks held by init/1: > [ 609.415836] #0: (reboot_mutex){+.+.+.}, at: [] SYSC_reboot+0xe6/0x2a0 > [ 609.416708] #1: (cpu_add_remove_lock){+.+.+.}, at: [] cpu_maps_update_begin+0x12/0x20 > [ 609.417712] #2: (cpu_hotplug.lock){+.+.+.}, at: [] cpu_hotplug_begin+0x27/0x60 > [ 609.418668] #3: (rcu_preempt_state.onoff_mutex){+.+...}, at: [] rcu_cleanup_dead_cpu+0x52/0x250 > [ 609.419819] 1 lock held by rcu_preempt/11: > [ 609.420277] #0: (rcu_preempt_state.onoff_mutex){+.+...}, at: [] rcu_gp_init+0x169/0x490 > [ 609.421391] 2 locks held by khungtaskd/3087: > [ 609.421811] #0: (rcu_read_lock){.+.+..}, at: [] check_hung_uninterruptible_tasks+0x3c/0x390 > [ 609.422867] #1: (tasklist_lock){.+.+..}, at: [] debug_show_all_locks+0x5c/0x270 > [ 609.423841] 2 locks held by sh/7108: > [ 609.424199] #0: (&tty->ldisc_sem){.+.+.+}, at: [] tty_ldisc_ref_wait+0x1f/0x50 > [ 609.425134] #1: (&ldata->atomic_read_lock){+.+...}, at: [] n_tty_read+0x31e/0x980 > > It looks like rcu_gp_init() went sleeping with the onoff_mutex held and > never got it's wakeup, while the rcu_cleanup_dead_cpu() code is waiting > to grab on that mutex. If your workload was busy enough with enough non-SCHED_OTHER activity that SCHED_OTHER tasks (like RCU's grace-period kthread) don't get to run, this is expected behavior. If this is reproducible, could you please try using chrt to increase that kthread's priority? Thanx, Paul -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/