Re: [PATCH 09/12] mm: compaction: avoid possible NULL pointer dereference in kcompactd_cpu_online

[Miaohe Lin <linmiaohe@huawei.com>]

On 2022/4/19 11:55, Andrew Morton wrote:
> Please cc David H on memhotplug stuff.

I'm relying on the scripts/get_maintainer.pl to take me the right people now.
Will take care of it. Many thanks for your remind!

> 
> On Mon, 18 Apr 2022 22:12:50 +0800 Miaohe Lin <linmiaohe@huawei.com> wrote:
> 
>> It's possible that kcompactd_run could fail to run kcompactd for a hot
>> added node and leave pgdat->kcompactd as NULL. So pgdat->kcompactd should
>> be checked here to avoid possible NULL pointer dereference.
>>
>> ..
>>
>> --- a/mm/compaction.c
>> +++ b/mm/compaction.c
>> @@ -3052,7 +3052,8 @@ static int kcompactd_cpu_online(unsigned int cpu)
>>  
>>  		if (cpumask_any_and(cpu_online_mask, mask) < nr_cpu_ids)
>>  			/* One of our CPUs online: restore mask */
>> -			set_cpus_allowed_ptr(pgdat->kcompactd, mask);
>> +			if (pgdat->kcompactd)
>> +				set_cpus_allowed_ptr(pgdat->kcompactd, mask);
>>  	}
>>  	return 0;
>>  }
> 
> Why not fail to bring the node online if kcompactd_run() failed?

kcompactd_run() is allowed to fail since it's introduced via commit 698b1b30642 ("mm, compaction: introduce kcompactd").

> 
> Also, should we panic the system if kcompactd_run() failed in
> kcompactd_init()?

So I think this might not be a critical issue. We could live with it anyway.
And in fact, CONFIG_COMPACTION is even not enabled in some systems.

Thanks!

> 
> .
>