Re: [Lse-tech] Re: [RFC] Scalable statistics counters using kmalloc_percpu

[Dipankar Sarma <dipankar@in.ibm.com>]

On Mon, Jul 29, 2002 at 11:23:06AM -0700, Andrew Morton wrote:
> Ravikiran G Thirumalai wrote:
> > 
> > ...
> > 
> > > General comment:  we need to clean up the kernel_stat stuff.  We
> > > cannot just make it per-cpu because it is 32k in size already.  I
> > > would suggest that we should break out the disk accounting and
> > > make the rest of kernel_stat per CPU.
> > >
> > > That would be a great application of your interface, and a good
> > > way to get your interface merged ;)  Is that something which you
> > > have time to do?
> > 
> > Sure,... anything to get these interfaces merged :)
> 
> Well Rusty's point about just using the percpu API seemed
> reasonable - that's basically equivalent to statically defining
> the data.  In what situation is dynamic allocation needed?
> 
> Well, in modules for one.  Unless we work out a way to make the
> percpu API work with storage which is defined within modules.

Apart from modules, I can see two advantages -

1. static per_cpu data has to be allocated based on NR_CPUS because
of the current offset calculations. kmalloc_percpu() needs to allocate only
for cpu_possible() cpus.

2. If the statistics is a part of a dynamically allocated structure.

Thanks
-- 
Dipankar Sarma  <dipankar@in.ibm.com> http://lse.sourceforge.net
Linux Technology Center, IBM Software Lab, Bangalore, India.