Re: [PATCH] mm: introduce sysctl file to flush per-cpu vmstat statistics

Re: [PATCH] mm: introduce sysctl file to flush per-cpu vmstat statistics

[Vlastimil Babka]

On 11/20/20 7:20 PM, Christopher Lameter wrote:
> On Tue, 17 Nov 2020, Marcelo Tosatti wrote:
> 
>> > So what we would need would be something like a sysctl that puts the
>> > system into a quiet state by completing all workqueue items. Idle all
>> > subsystems that need it and put the cpu into NOHZ mode.
>>
>> Are you suggesting that instead of a specific file to control vmstat
>> workqueue only, a more generic sysctl could be used?
> 
> Yes. Introduce a sysctl to quiet down the system. Clean caches that will
> trigger kernel threads and whatever else is pending on that processor.

Please CC linux-api on future postings that introduce stuff like this.

>> About NOHZ mode: the CPU should enter NOHZ automatically as soon as
>> there is a single thread running, so unclear why that would be needed.
> 
> There are typically pending actions that still trigger interruptions.
> 
> If you would immediately quiet down the system if there is only one thread
> runnable then you would compromise system performance through frequent
> counter folding and cache cleaning etc.

If someone goes through the trouble of setting up NOHZ, these 
disruptions should be only temporary and happen during the setup, no?

Re: [PATCH] mm: introduce sysctl file to flush per-cpu vmstat statistics

[Marcelo Tosatti]

On Tue, Nov 24, 2020 at 06:12:29PM +0100, Vlastimil Babka wrote:
> On 11/20/20 7:20 PM, Christopher Lameter wrote:
> > On Tue, 17 Nov 2020, Marcelo Tosatti wrote:
> > 
> > > > So what we would need would be something like a sysctl that puts the
> > > > system into a quiet state by completing all workqueue items. Idle all
> > > > subsystems that need it and put the cpu into NOHZ mode.
> > > 
> > > Are you suggesting that instead of a specific file to control vmstat
> > > workqueue only, a more generic sysctl could be used?
> > 
> > Yes. Introduce a sysctl to quiet down the system. Clean caches that will
> > trigger kernel threads and whatever else is pending on that processor.
> 
> Please CC linux-api on future postings that introduce stuff like this.

Thanks, will do.

> > > About NOHZ mode: the CPU should enter NOHZ automatically as soon as
> > > there is a single thread running, so unclear why that would be needed.
> > 
> > There are typically pending actions that still trigger interruptions.
> > 
> > If you would immediately quiet down the system if there is only one thread
> > runnable then you would compromise system performance through frequent
> > counter folding and cache cleaning etc.
> 
> If someone goes through the trouble of setting up NOHZ, these disruptions
> should be only temporary and happen during the setup, no?

There is no way for the userspace application to know when the
disruptions are gone (vmstat work function executed).

But yes, they are temporary and happen during setup.