use for page_state accounting fields

use for page_state accounting fields

[Marcelo Tosatti]

Hi Andrew,

I suppose you wrote the page_state per-CPU statistics structure.

There are some fields, for instance pgactivate/pgdeactivate, that
do not seem to be used anywhere. Sure, they are useful for statistics, 
but no place in the kernel exports them to userspace AFAICS.

        unsigned long pgactivate;       /* pages moved inactive->active */
        unsigned long pgdeactivate;     /* pages moved active->inactive */

Counting them is somewhat expensive I believe (need to disable IRQ), based
on the assumption that these days any cycle is a loss.

So, from my POV we should 

a) export them to userspace
b) surround them by CONFIG_DEBUG_MMSTATS or something similar

Tell me I'm wrong.


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>

Re: use for page_state accounting fields

[Marcelo Tosatti]

On Mon, Aug 16, 2004 at 02:31:49PM -0700, Andrew Morton wrote:
> Marcelo Tosatti <marcelo.tosatti@cyclades.com> wrote:
> >
> > Hi Andrew,
> > 
> > I suppose you wrote the page_state per-CPU statistics structure.
> > 
> > There are some fields, for instance pgactivate/pgdeactivate, that
> > do not seem to be used anywhere. Sure, they are useful for statistics, 
> > but no place in the kernel exports them to userspace AFAICS.
> > 
> >         unsigned long pgactivate;       /* pages moved inactive->active */
> >         unsigned long pgdeactivate;     /* pages moved active->inactive */
> > 
> > Counting them is somewhat expensive I believe (need to disable IRQ), based
> > on the assumption that these days any cycle is a loss.
> > 
> > So, from my POV we should 
> > 
> > a) export them to userspace
> > b) surround them by CONFIG_DEBUG_MMSTATS or something similar
> > 
> > Tell me I'm wrong.
> 
> Take a peek in /proc/vmstat ;)

Doh. 

Is there any tool which reads these statistics and makes use of them? Number of 
inactivations/activations per timeframe, etc?

Thanks
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>

Re: use for page_state accounting fields

[Andrew Morton]

Marcelo Tosatti <marcelo.tosatti@cyclades.com> wrote:
>
> Hi Andrew,
> 
> I suppose you wrote the page_state per-CPU statistics structure.
> 
> There are some fields, for instance pgactivate/pgdeactivate, that
> do not seem to be used anywhere. Sure, they are useful for statistics, 
> but no place in the kernel exports them to userspace AFAICS.
> 
>         unsigned long pgactivate;       /* pages moved inactive->active */
>         unsigned long pgdeactivate;     /* pages moved active->inactive */
> 
> Counting them is somewhat expensive I believe (need to disable IRQ), based
> on the assumption that these days any cycle is a loss.
> 
> So, from my POV we should 
> 
> a) export them to userspace
> b) surround them by CONFIG_DEBUG_MMSTATS or something similar
> 
> Tell me I'm wrong.

Take a peek in /proc/vmstat ;)
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>

Re: use for page_state accounting fields

[Andrew Morton]

Marcelo Tosatti <marcelo.tosatti@cyclades.com> wrote:
>
> > Take a peek in /proc/vmstat ;)
> 
> Doh. 
> 
> Is there any tool which reads these statistics and makes use of them? Number of 
> inactivations/activations per timeframe, etc?

Not that I am aware of.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>