Re: WARN_ON_ONCE(in_irq() || in_serving_softirq()

From: Paul E. McKenney
Date: Fri Nov 15 2013 - 14:21:57 EST

Next message: Paul E. McKenney: "Re: performance regressions by "seqcount: Add lockdep functionalityto seqcount/seqlock structures""
Previous message: H. Peter Anvin: "Re: [PATCH 0/3] Early use of boot service memory"
In reply to: Paul E. McKenney: "WARN_ON_ONCE(in_irq() || in_serving_softirq()"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

On Mon, Nov 11, 2013 at 10:36:02AM -0800, Paul E. McKenney wrote:
> Hello, Lai,
>
> I am hitting the new warning in rcu_read_unlock_special() that checks for
> (in_irq() || in_serving_softirq()). Please see below for the splat.
> I actually managed to get two CPUs hitting this simultaneously, so got
> two splats.
>
> My first thought is to revert the WARN_ON_ONCE(), going back to:
>
> if (in_irq() || in_serving_softirq()) {
> local_irq_restore(flags);
> return;
> }
>
> >From what I can see, the scheduling-clock tick is setting
> RCU_READ_UNLOCK_NEED_QS, which is causing the softirq handler's
> RCU read-side critical section to enter rcu_read_unlock_special().
>
> Another fix would be to check for t->rcu_read_unlock_special == 0
> in the previous "if (special & RCU_READ_UNLOCK_NEED_QS) {" check.

This check seems to do the trick.

Thanx, Paul

> Other thoughts?
>
> Thanx, Paul
>
> ------------------------------------------------------------------------
>
> [ 192.542052] ------------[ cut here ]------------
> [ 192.542054] ------------[ cut here ]------------
> [ 192.542072] WARNING: CPU: 1 PID: 674 at /home/paulmck/public_git/linux-rcu/kernel/rcu/tree_plugin.h:367 rcu_read_unlock_special+0x260/0x270()
> [ 192.542074] Modules linked in:
> [ 192.542080] CPU: 1 PID: 674 Comm: rcu_torture_rea Not tainted 3.12.0-rc1+ #1
> [ 192.542081] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2007
> [ 192.542085] 000000000000016f ffff88001fc43cc8 ffffffff817eaf38 0000000000000102
> [ 192.542087] 0000000000000000 ffff88001fc43d08 ffffffff81045907 0000000000000001
> [ 192.542089] 0000000000000002 0000000000004bcf ffffffff83ccbd40 ffff88001e1cc0c0
> [ 192.542090] Call Trace:
> [ 192.542100] <IRQ> [<ffffffff817eaf38>] dump_stack+0x4f/0x84
> [ 192.542106] [<ffffffff81045907>] warn_slowpath_common+0x87/0xb0
> [ 192.542109] [<ffffffff81045945>] warn_slowpath_null+0x15/0x20
> [ 192.542111] [<ffffffff81094ea0>] rcu_read_unlock_special+0x260/0x270
> [ 192.542114] [<ffffffff8108ddce>] __rcu_read_unlock+0x5e/0x60
> [ 192.542117] [<ffffffff8108f1c1>] rcu_torture_read_unlock+0x21/0x30
> [ 192.542133] [<ffffffff810914f5>] rcu_torture_timer+0x135/0x150
> [ 192.542137] [<ffffffff810913c0>] ? rcu_torture_reader+0x310/0x310
> [ 192.542144] [<ffffffff8105149a>] call_timer_fn+0x7a/0x200
> [ 192.542146] [<ffffffff81051420>] ? del_timer+0x70/0x70
> [ 192.542148] [<ffffffff81052165>] run_timer_softirq+0x215/0x2f0
> [ 192.542151] [<ffffffff8109671f>] ? ktime_get+0x4f/0xe0
> [ 192.542153] [<ffffffff810913c0>] ? rcu_torture_reader+0x310/0x310
> [ 192.542156] [<ffffffff8104a1a9>] __do_softirq+0xd9/0x2d0
> [ 192.542158] [<ffffffff8104a4ce>] irq_exit+0x7e/0xa0
> [ 192.542163] [<ffffffff8102e1d5>] smp_apic_timer_interrupt+0x45/0x60
> [ 192.542170] [<ffffffff817fe6ca>] apic_timer_interrupt+0x6a/0x70
> [ 192.542177] <EOI> [<ffffffff81071d86>] ? finish_task_switch+0x46/0xf0
> [ 192.542179] [<ffffffff81071d86>] ? finish_task_switch+0x46/0xf0
> [ 192.542186] [<ffffffff817f64cc>] ? _raw_spin_unlock_irq+0x2c/0x60
> [ 192.542188] [<ffffffff817f64c6>] ? _raw_spin_unlock_irq+0x26/0x60
> [ 192.542190] [<ffffffff81071dc3>] finish_task_switch+0x83/0xf0
> [ 192.542192] [<ffffffff81071d86>] ? finish_task_switch+0x46/0xf0
> [ 192.542194] [<ffffffff817f46ca>] __schedule+0x3ba/0x860
> [ 192.542197] [<ffffffff817f4c84>] schedule+0x24/0x70
> [ 192.542199] [<ffffffff81091190>] rcu_torture_reader+0xe0/0x310
> [ 192.542201] [<ffffffff810913c0>] ? rcu_torture_reader+0x310/0x310
> [ 192.542204] [<ffffffff810910b0>] ? rcutorture_trace_dump+0x30/0x30
> [ 192.542209] [<ffffffff81068096>] kthread+0xd6/0xe0
> [ 192.542211] [<ffffffff81071d86>] ? finish_task_switch+0x46/0xf0
> [ 192.542215] [<ffffffff81067fc0>] ? flush_kthread_work+0x190/0x190
> [ 192.542217] [<ffffffff817fda2c>] ret_from_fork+0x7c/0xb0
> [ 192.542220] [<ffffffff81067fc0>] ? flush_kthread_work+0x190/0x190
> [ 192.542222] ---[ end trace 8519fcb7dea5ceee ]---
> [ 192.543027] WARNING: CPU: 2 PID: 670 at /home/paulmck/public_git/linux-rcu/kernel/rcu/tree_plugin.h:367 rcu_read_unlock_special+0x260/0x270()
> [ 192.543027] Modules linked in:
> [ 192.543027] CPU: 2 PID: 670 Comm: rcu_torture_rea Tainted: G W 3.12.0-rc1+ #1
> [ 192.543027] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2007
> [ 192.543027] 000000000000016f ffff88001fc83cc8 ffffffff817eaf38 0000000000000102
> [ 192.543027] 0000000000000000 ffff88001fc83d08 ffffffff81045907 ffff88001e36ffd8
> [ 192.543027] 0000000000000002 0000000000004bcf ffffffff83ccbd40 ffff88001e1c8000
> [ 192.543027] Call Trace:
> [ 192.543027] <IRQ> [<ffffffff817eaf38>] dump_stack+0x4f/0x84
> [ 192.543027] [<ffffffff81045907>] warn_slowpath_common+0x87/0xb0
> [ 192.543027] [<ffffffff81045945>] warn_slowpath_null+0x15/0x20
> [ 192.543027] [<ffffffff81094ea0>] rcu_read_unlock_special+0x260/0x270
> [ 192.543027] [<ffffffff8108ddce>] __rcu_read_unlock+0x5e/0x60
> [ 192.543027] [<ffffffff8108f1c1>] rcu_torture_read_unlock+0x21/0x30
> [ 192.543027] [<ffffffff810914f5>] rcu_torture_timer+0x135/0x150
> [ 192.543027] [<ffffffff810913c0>] ? rcu_torture_reader+0x310/0x310
> [ 192.543027] [<ffffffff8105149a>] call_timer_fn+0x7a/0x200
> [ 192.543027] [<ffffffff81051420>] ? del_timer+0x70/0x70
> [ 192.543027] [<ffffffff81052165>] run_timer_softirq+0x215/0x2f0
> [ 192.543027] [<ffffffff8109671f>] ? ktime_get+0x4f/0xe0
> [ 192.543027] [<ffffffff810913c0>] ? rcu_torture_reader+0x310/0x310
> [ 192.543027] [<ffffffff8104a1a9>] __do_softirq+0xd9/0x2d0
> [ 192.543027] [<ffffffff8104a4ce>] irq_exit+0x7e/0xa0
> [ 192.543027] [<ffffffff8102e1d5>] smp_apic_timer_interrupt+0x45/0x60
> [ 192.543027] [<ffffffff817fe6ca>] apic_timer_interrupt+0x6a/0x70
> [ 192.543027] <EOI> [<ffffffff817f64cc>] ? _raw_spin_unlock_irq+0x2c/0x60
> [ 192.543027] [<ffffffff817f64c6>] ? _raw_spin_unlock_irq+0x26/0x60
> [ 192.543027] [<ffffffff817f4a1d>] __schedule+0x70d/0x860
> [ 192.543027] [<ffffffff817f4c84>] schedule+0x24/0x70
> [ 192.543027] [<ffffffff81091190>] rcu_torture_reader+0xe0/0x310
> [ 192.543027] [<ffffffff81071d86>] ? finish_task_switch+0x46/0xf0
> [ 192.543027] [<ffffffff810913c0>] ? rcu_torture_reader+0x310/0x310
> [ 192.543027] [<ffffffff810910b0>] ? rcutorture_trace_dump+0x30/0x30
> [ 192.543027] [<ffffffff81068096>] kthread+0xd6/0xe0
> [ 192.543027] [<ffffffff81071d86>] ? finish_task_switch+0x46/0xf0
> [ 192.543027] [<ffffffff81067fc0>] ? flush_kthread_work+0x190/0x190
> [ 192.543027] [<ffffffff817fda2c>] ret_from_fork+0x7c/0xb0
> [ 192.543027] [<ffffffff81067fc0>] ? flush_kthread_work+0x190/0x190
> [ 192.543027] ---[ end trace 8519fcb7dea5ceef ]---
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Next message: Paul E. McKenney: "Re: performance regressions by "seqcount: Add lockdep functionalityto seqcount/seqlock structures""
Previous message: H. Peter Anvin: "Re: [PATCH 0/3] Early use of boot service memory"
In reply to: Paul E. McKenney: "WARN_ON_ONCE(in_irq() || in_serving_softirq()"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]