Re: [PATCH tip/locking/core v9 4/6] locking/pvqspinlock: Collect slowpath lock statistics

[Waiman Long <waiman.long@hpe.com>]

On 11/05/2015 11:43 AM, Peter Zijlstra wrote:
> On Thu, Nov 05, 2015 at 11:29:29AM -0500, Waiman Long wrote:
>> On 11/02/2015 11:40 AM, Peter Zijlstra wrote:
>>> On Fri, Oct 30, 2015 at 07:26:35PM -0400, Waiman Long wrote:
>>>> This patch enables the accumulation of kicking and waiting related
>>>> PV qspinlock statistics when the new QUEUED_LOCK_STAT configuration
>>>> option is selected. It also enables the collection of data which
>>>> enable us to calculate the kicking and wakeup latencies which have
>>>> a heavy dependency on the CPUs being used.
>>>>
>>>> The statistical counters are per-cpu variables to minimize the
>>>> performance overhead in their updates. These counters are exported
>>>> via the sysfs filesystem under the /sys/kernel/qlockstat directory.
>>>> When the corresponding sysfs files are read, summation and computing
>>>> of the required data are then performed.
>>> Why did you switch to sysfs? You can create custom debugfs files too.
>> I was not aware of that capability. So you mean using debugfs_create_file()
>> using custom file_operations. Right?
> Yep.
>
>> That doesn't seem to be easier than
>> using sysfs. However, I can use that if you think it is better to use
>> debugfs.
> Mostly I just wanted to point out that it was possible; you need not
> change to sysfs because debugfs lacks the capability.
>
> But now that you ask, I think debugfs might be the better place, such
> statistics (and the proposed CONFIG symbol) are purely for debug
> purposes, right?

Davidlohr had asked me to use per-cpu counters to reduce performance 
overhead so that they can be usable in production system. That is 
another reason why I move to sysfs.

BTW, do you have comments on the other patches in the series? I would 
like to collect all the comments before I renew the series.

Cheers,
Longman