[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CACT4Y+bvHWuyDZjoFhLAOiK1rfNQcfuSJCY0oS5xA=u8j04bXQ@mail.gmail.com>
Date: Tue, 15 Dec 2015 14:41:28 +0100
From: Dmitry Vyukov <dvyukov@...gle.com>
To: Peter Zijlstra <peterz@...radead.org>
Cc: Ingo Molnar <mingo@...hat.com>,
Arnaldo Carvalho de Melo <acme@...nel.org>,
LKML <linux-kernel@...r.kernel.org>,
syzkaller <syzkaller@...glegroups.com>,
Kostya Serebryany <kcc@...gle.com>,
Alexander Potapenko <glider@...gle.com>,
Sasha Levin <sasha.levin@...cle.com>,
Eric Dumazet <edumazet@...gle.com>
Subject: Re: perf: stalls in perf_install_in_context/perf_remove_from_context
On Tue, Dec 15, 2015 at 2:39 PM, Peter Zijlstra <peterz@...radead.org> wrote:
> On Tue, Dec 15, 2015 at 02:01:32PM +0100, Dmitry Vyukov wrote:
>
>
>> Call Trace:
>> <IRQ> [<ffffffff810e5e38>] sched_show_task+0x148/0x250 kernel/sched/core.c:4934
>
> <snip>
>
>> [<ffffffff823ff46c>] apic_timer_interrupt+0x8c/0xa0 arch/x86/entry/entry_64.S:520
>>
>> <EOI> [< inline >] ? arch_local_irq_enable ./arch/x86/include/asm/paravirt.h:816
>> <EOI> [< inline >] ? __raw_spin_unlock_irq include/linux/spinlock_api_smp.h:170
>> <EOI> [<ffffffff823fde9e>] ? _raw_spin_unlock_irq+0x2e/0x40 kernel/locking/spinlock.c:199
>> [< inline >] ? __raw_spin_unlock_irq include/linux/spinlock_api_smp.h:170
>> [<ffffffff823fde97>] ? _raw_spin_unlock_irq+0x27/0x40 kernel/locking/spinlock.c:199
>> [<ffffffff8119c8bb>] perf_install_in_context+0x6b/0xf0 kernel/events/core.c:2173
>> [<ffffffff811a492c>] SYSC_perf_event_open+0x63c/0xe40 kernel/events/core.c:8565
>> [<ffffffff811a7739>] SyS_perf_event_open+0x9/0x10 kernel/events/core.c:8261
>> [<ffffffff823fe676>] entry_SYSCALL_64_fastpath+0x16/0x7a arch/x86/entry/entry_64.S:185
>
>
> So this looks like its stuck in the retry loop in
> perf_install_in_context(). We do the raw_spin_unlock_irq(), which allows
> the above watchdog interrupt to happen, but make no progress.
>
> This means the task_function_call() thing keeps failing, which is
> extremely unlikely.
>
> Let me have a play with your test-case to see if it reproduces.
Yeah, both perf_install_in_context and perf_remove_from_context hang
in a very similar loops.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists