Re: [PATCH] Do not change worker's running cpu in cmci_rediscover().

[Tang Chen <tangchen@cn.fujitsu.com>]

Hi all,

Would you please help to review this patch ?

Thanks. :)


On 09/28/2012 03:54 PM, Lai Jiangshan wrote:
> Add CC: Tejun Heo, Peter Zijlstra.
>
> Hi, Tejun
>
> This is a bug whose root cause is the same as
> https://bugzilla.kernel.org/show_bug.cgi?id=47301.
>
> Acked-by: Lai Jiangshan<laijs@cn.fujitsu.com>
>
> thanks,
> Lai
>
>
> On 09/27/2012 05:19 PM, Tang Chen wrote:
>> 1. cmci_rediscover() is only called by the CPU_POST_DEAD event handler, which
>> means the corresponding cpu has already dead. As a result, it won't be accessed
>> in the for_each_online_cpu loop.
>> So, we could change the if(cpu == dying) statement into a BUG_ON().
>>
>> 2. cmci_rediscover() used set_cpus_allowed_ptr() to change the current process's
>> running cpu, and migrate itself to the dest cpu. But worker processes are not
>> allowed to be migrated. If current is a worker, the worker will be migrated to
>> another cpu, but the corresponding  worker_pool is still on the original cpu.
>>
>> In this case, the following BUG_ON in try_to_wake_up_local() will be triggered:
>> BUG_ON(rq != this_rq());
>>
>> This will cause the kernel panic.
>>
>> This patch removes the set_cpus_allowed_ptr() call, and put the cmci rediscover
>> jobs onto all the other cpus using system_wq. This could bring some delay for
>> the jobs.
>>
>> The following is call trace.
>>
>> [ 6155.451107] ------------[ cut here ]------------
>> [ 6155.452019] kernel BUG at kernel/sched/core.c:1654!
>> ......
>> [ 6155.452019] RIP: 0010:[<ffffffff810add15>]  [<ffffffff810add15>] try_to_wake_up_local+0x115/0x130
>> ......
>> [ 6155.452019] Call Trace:
>> [ 6155.452019]  [<ffffffff8166fc14>] __schedule+0x764/0x880
>> [ 6155.452019]  [<ffffffff81670059>] schedule+0x29/0x70
>> [ 6155.452019]  [<ffffffff8166de65>] schedule_timeout+0x235/0x2d0
>> [ 6155.452019]  [<ffffffff810db57d>] ? mark_held_locks+0x8d/0x140
>> [ 6155.452019]  [<ffffffff810dd463>] ? __lock_release+0x133/0x1a0
>> [ 6155.452019]  [<ffffffff81671c50>] ? _raw_spin_unlock_irq+0x30/0x50
>> [ 6155.452019]  [<ffffffff810db8f5>] ? trace_hardirqs_on_caller+0x105/0x190
>> [ 6155.452019]  [<ffffffff8166fefb>] wait_for_common+0x12b/0x180
>> [ 6155.452019]  [<ffffffff810b0b30>] ? try_to_wake_up+0x2f0/0x2f0
>> [ 6155.452019]  [<ffffffff8167002d>] wait_for_completion+0x1d/0x20
>> [ 6155.452019]  [<ffffffff8110008a>] stop_one_cpu+0x8a/0xc0
>> [ 6155.452019]  [<ffffffff810abd40>] ? __migrate_task+0x1a0/0x1a0
>> [ 6155.452019]  [<ffffffff810a6ab8>] ? complete+0x28/0x60
>> [ 6155.452019]  [<ffffffff810b0fd8>] set_cpus_allowed_ptr+0x128/0x130
>> [ 6155.452019]  [<ffffffff81036785>] cmci_rediscover+0xf5/0x140
>> [ 6155.452019]  [<ffffffff816643c0>] mce_cpu_callback+0x18d/0x19d
>> [ 6155.452019]  [<ffffffff81676187>] notifier_call_chain+0x67/0x150
>> [ 6155.452019]  [<ffffffff810a03de>] __raw_notifier_call_chain+0xe/0x10
>> [ 6155.452019]  [<ffffffff81070470>] __cpu_notify+0x20/0x40
>> [ 6155.452019]  [<ffffffff810704a5>] cpu_notify_nofail+0x15/0x30
>> [ 6155.452019]  [<ffffffff81655182>] _cpu_down+0x262/0x2e0
>> [ 6155.452019]  [<ffffffff81655236>] cpu_down+0x36/0x50
>> [ 6155.452019]  [<ffffffff813d3eaa>] acpi_processor_remove+0x50/0x11e
>> [ 6155.452019]  [<ffffffff813a6978>] acpi_device_remove+0x90/0xb2
>> [ 6155.452019]  [<ffffffff8143cbec>] __device_release_driver+0x7c/0xf0
>> [ 6155.452019]  [<ffffffff8143cd6f>] device_release_driver+0x2f/0x50
>> [ 6155.452019]  [<ffffffff813a7870>] acpi_bus_remove+0x32/0x6d
>> [ 6155.452019]  [<ffffffff813a7932>] acpi_bus_trim+0x87/0xee
>> [ 6155.452019]  [<ffffffff813a7a21>] acpi_bus_hot_remove_device+0x88/0x16b
>> [ 6155.452019]  [<ffffffff813a33ee>] acpi_os_execute_deferred+0x27/0x34
>> [ 6155.452019]  [<ffffffff81090589>] process_one_work+0x219/0x680
>> [ 6155.452019]  [<ffffffff81090528>] ? process_one_work+0x1b8/0x680
>> [ 6155.452019]  [<ffffffff813a33c7>] ? acpi_os_wait_events_complete+0x23/0x23
>> [ 6155.452019]  [<ffffffff810923be>] worker_thread+0x12e/0x320
>> [ 6155.452019]  [<ffffffff81092290>] ? manage_workers+0x110/0x110
>> [ 6155.452019]  [<ffffffff81098396>] kthread+0xc6/0xd0
>> [ 6155.452019]  [<ffffffff8167c4c4>] kernel_thread_helper+0x4/0x10
>> [ 6155.452019]  [<ffffffff81671f30>] ? retint_restore_args+0x13/0x13
>> [ 6155.452019]  [<ffffffff810982d0>] ? __init_kthread_worker+0x70/0x70
>> [ 6155.452019]  [<ffffffff8167c4c0>] ? gs_change+0x13/0x13
>>
>> Signed-off-by: Tang Chen<tangchen@cn.fujitsu.com>
>> Signed-off-by: Miao Xie<miaox@cn.fujitsu.com>
>> ---
>>   arch/x86/kernel/cpu/mcheck/mce_intel.c |   34 +++++++++++++++++--------------
>>   1 files changed, 19 insertions(+), 15 deletions(-)
>>
>> diff --git a/arch/x86/kernel/cpu/mcheck/mce_intel.c b/arch/x86/kernel/cpu/mcheck/mce_intel.c
>> index 38e49bc..f7d9795 100644
>> --- a/arch/x86/kernel/cpu/mcheck/mce_intel.c
>> +++ b/arch/x86/kernel/cpu/mcheck/mce_intel.c
>> @@ -163,34 +163,38 @@ void cmci_clear(void)
>>   	raw_spin_unlock_irqrestore(&cmci_discover_lock, flags);
>>   }
>>
>> +static long cmci_rediscover_work_func(void *arg)
>> +{
>> +	int banks;
>> +
>> +	/* Recheck banks in case CPUs don't all have the same */
>> +	if (cmci_supported(&banks))
>> +		cmci_discover(banks, 0);
>> +
>> +	return 0;
>> +}
>> +
>>   /*
>>    * After a CPU went down cycle through all the others and rediscover
>>    * Must run in process context.
>>    */
>>   void cmci_rediscover(int dying)
>>   {
>> -	int banks;
>> -	int cpu;
>> -	cpumask_var_t old;
>> +	int cpu, banks;
>>
>>   	if (!cmci_supported(&banks))
>>   		return;
>> -	if (!alloc_cpumask_var(&old, GFP_KERNEL))
>> -		return;
>> -	cpumask_copy(old,&current->cpus_allowed);
>>
>>   	for_each_online_cpu(cpu) {
>> -		if (cpu == dying)
>> -			continue;
>> -		if (set_cpus_allowed_ptr(current, cpumask_of(cpu)))
>> +		BUG_ON(cpu == dying);
>> +
>> +		if (cpu == smp_processor_id()) {
>> +			cmci_rediscover_work_func(NULL);
>>   			continue;
>> -		/* Recheck banks in case CPUs don't all have the same */
>> -		if (cmci_supported(&banks))
>> -			cmci_discover(banks, 0);
>> -	}
>> +		}
>>
>> -	set_cpus_allowed_ptr(current, old);
>> -	free_cpumask_var(old);
>> +		work_on_cpu(cpu, cmci_rediscover_work_func, NULL);
>> +	}
>>   }
>>
>>   /*
>
>