Re: [PATCH] perf: fix assertion failure in x86_pmu_start()

[Stephane Eranian]

On Tue, Feb 7, 2012 at 9:34 AM, Eric Dumazet <eric.dumazet@xxxxxxxxx> wrote:
> Le lundi 06 fÃvrier 2012 Ã 21:53 +0100, Stephane Eranian a Ãcrit :
>> The following patch fixes an issue introduced by the following
>> commit:
>> Â Â Â Â e050e3f0a71b ("perf: Fix broken interrupt rate throttling")
>>
>> The patch caused the following warning to pop up depending on
>> the sampling frequency adjustments:
>>
>> [89214.962603] ------------[ cut here ]------------
>> [89214.967441] WARNING: at arch/x86/kernel/cpu/perf_event.c:995 x86_pmu_start+0x79/0xd4()
>> [89214.975825] Hardware name: X8DTN
>> [89214.979268] Modules linked in:
>> [89214.982560] Pid: 0, comm: swapper/6 Not tainted 3.3.0-rc2-tip+ #1
>> [89214.988865] Call Trace:
>> [89214.991533] Â<IRQ> Â[<ffffffff81065cc7>] warn_slowpath_common+0x7e/0x97
>> [89214.998379] Â[<ffffffff81065cf5>] warn_slowpath_null+0x15/0x17
>> [89215.004428] Â[<ffffffff8103f626>] x86_pmu_start+0x79/0xd4
>> [89215.010042] Â[<ffffffff810e30d1>] perf_adjust_freq_unthr_context.part.63+0xef/0x123
>> [89215.018123] Â[<ffffffff810e318c>] perf_event_task_tick+0x87/0x1c1
>> [89215.024463] Â[<ffffffff810a2370>] ? tick_nohz_handler+0xda/0xda
>> [89215.030595] Â[<ffffffff8108b819>] scheduler_tick+0xd1/0xf3
>> [89215.036296] Â[<ffffffff810720b0>] update_process_times+0x5e/0x6f
>> [89215.042512] Â[<ffffffff810a23e0>] tick_sched_timer+0x70/0x99
>> [89215.048387] Â[<ffffffff810823f9>] __run_hrtimer+0x8c/0x148
>> [89215.054087] Â[<ffffffff81082add>] hrtimer_interrupt+0xc1/0x18c
>>
>> It was caused by the following call sequence:
>>
>> perf_adjust_freq_unthr_context.part() {
>> Â Â Âstop()
>> Â Â Âif (delta > 0) {
>> Â Â Â Â Â perf_adjust_period() {
>> Â Â Â Â Â Â Â if (period > 8*...) {
>> Â Â Â Â Â Â Â Â Â stop()
>> Â Â Â Â Â Â Â Â Â ...
>> Â Â Â Â Â Â Â Â Â start()
>> Â Â Â Â Â Â Â }
>> Â Â Â Â Â }
>> Â Â Â }
>> Â Â Â start()
>> }
>>
>> Which caused a double start and a double stop, thus triggering the assert
>> in x86_pmu_start().
>>
>> The patch fixes the problem by avoiding the double calls. We pass a new
>> argument to perf_adjust_period() to indicate whether or not the event
>> is already stopped. We can't just remove the start/stop from that function
>> because it's called from __perf_event_overflow where the event needs to
>> be reloaded via a stop/start back-toback call.
>>
>> The patch reintroduces the assertion in x86_pmu_start() which was removed
>> by commit:
>> Â Â Â 84f2b9b perf: Remove deprecated WARN_ON_ONCE()
>>
>> Signed-off-by: Stephane Eranian <eranian@xxxxxxxxxx>
>> ---
>
> This indeed fix the WARNING for me
>
> Acked-by: Eric Dumazet <eric.dumazet@xxxxxxxxx>
>
> But I still have these messages when doing a perf session.
>
> Machine seems to recover properly.
>
> Previous kernels were working without notice.
>
Doing what on what machine?

> [ Â300.553017] Uhhuh. NMI received for unknown reason 31 on CPU 2.
> [ Â300.553071] Do you have a strange power saving mode enabled?
> [ Â300.553115] Dazed and confused, but trying to continue
> [ Â300.775014] Uhhuh. NMI received for unknown reason 31 on CPU 1.
> [ Â300.775064] Do you have a strange power saving mode enabled?
> [ Â300.775107] Dazed and confused, but trying to continue
> [ Â303.250012] Uhhuh. NMI received for unknown reason 31 on CPU 0.
> [ Â303.250067] Do you have a strange power saving mode enabled?
> [ Â303.250110] Dazed and confused, but trying to continue
> [ Â303.278012] Uhhuh. NMI received for unknown reason 31 on CPU 7.
> [ Â303.278063] Do you have a strange power saving mode enabled?
> [ Â303.278106] Dazed and confused, but trying to continue
> [ Â305.839016] Uhhuh. NMI received for unknown reason 21 on CPU 5.
> [ Â305.839068] Do you have a strange power saving mode enabled?
> [ Â305.839112] Dazed and confused, but trying to continue
> [ Â305.907013] Uhhuh. NMI received for unknown reason 31 on CPU 4.
> [ Â305.907066] Do you have a strange power saving mode enabled?
> [ Â305.907109] Dazed and confused, but trying to continue
> [ Â306.953017] Uhhuh. NMI received for unknown reason 31 on CPU 1.
> [ Â306.953069] Do you have a strange power saving mode enabled?
> [ Â306.953111] Dazed and confused, but trying to continue
> [ Â308.585014] Uhhuh. NMI received for unknown reason 31 on CPU 6.
> [ Â308.585064] Do you have a strange power saving mode enabled?
> [ Â308.585108] Dazed and confused, but trying to continue
> [ Â309.239012] Uhhuh. NMI received for unknown reason 31 on CPU 0.
> [ Â309.239079] Do you have a strange power saving mode enabled?
> [ Â309.239137] Dazed and confused, but trying to continue
> [ Â309.426009] Uhhuh. NMI received for unknown reason 21 on CPU 7.
> [ Â309.426075] Do you have a strange power saving mode enabled?
> [ Â309.426132] Dazed and confused, but trying to continue
> [ Â309.592010] Uhhuh. NMI received for unknown reason 21 on CPU 2.
> [ Â309.592076] Do you have a strange power saving mode enabled?
> [ Â309.592133] Dazed and confused, but trying to continue
>
>
--
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/