Re: [RFC] Light weight event counters (V2) instead of page state

[Christoph Lameter <clameter@sgi.com>]

On Tue, 13 Jun 2006, Martin Peschke wrote:

> Did you consider using the statistics infrastructure available in -mm?
> (lib/statistic.c, include/linux/statistic.h, Documentation/statistics.txt)

No its not in Linus tree and has even more overhead than the existing 
counters since it always checks if a certain counters is on or off and it 
also does indirect calls to an update function.

It does not use local_t at all and takes an smp_processor_id() parameter 
to various functions which makes the future use of cpu_local_* functions 
not possible (should that implementation get fixed).

> To me that looks like a good application for the statistics infrastructure,
> which shows statistics through debugfs instead of procfs. Btw., it completely
> unburdens exploiting kernel code from the delivery of statistics to users.

The light weight counter implementation here provides the output of 
/proc/vmstat. That is a fixed interface.