Re: [PATCH V5 12/14] trace: Protect tr->tracing_cpumask with get/put_online_cpus

[Daniel Bristot de Oliveira <bristot@redhat.com>]

On 6/24/21 11:39 PM, Steven Rostedt wrote:
> On Tue, 22 Jun 2021 16:42:30 +0200
> Daniel Bristot de Oliveira <bristot@redhat.com> wrote:
> 
>> diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
>> index 52fc9438b7b4..c14f33db147e 100644
>> --- a/kernel/trace/trace.c
>> +++ b/kernel/trace/trace.c
>> @@ -5053,7 +5053,13 @@ int tracing_set_cpumask(struct trace_array *tr,
>>  	arch_spin_unlock(&tr->max_lock);
>>  	local_irq_enable();
>>  
>> +	/*
>> +	 * tracing_cpumask is read by tracers that support CPU
>> +	 * hotplug.
>> +	 */
>> +	get_online_cpus();
>>  	cpumask_copy(tr->tracing_cpumask, tracing_cpumask_new);
>> +	put_online_cpus();
>>  
>>  	return 0;
> 
> Hmm, the tracing_cpumask is only touched in he work function, with the
> necessary locks. How is get_online_cpus() protecting it here?
> 
> That is, tracing_cpumask isn't touched in the path of bringing up or
> taking down a CPU, and shouldn't be an issue here.

You are right, it is not needed with the new hotplug method.

> Should I just drop this patch?

Yes! drop it.

Thanks Steven!
-- Daniel

> 
> -- Steve
>