Re: [PATCH 0/3] Generalise ARM perf-events backend for oprofile

[Will Deacon <will.deacon@arm.com>]

Ben,

On Wed, 2010-08-25 at 02:41 +0100, Benjamin Herrenschmidt wrote:
> On Mon, 2010-08-23 at 06:57 -0400, Christoph Hellwig wrote:
> > On Mon, Aug 23, 2010 at 11:46:18AM +0100, Matt Fleming wrote:
> > > The perf-events backend for OProfile that Will Deacon wrote in
> > > 8c1fc96f6fd1f361428ba805103af0d0eee65179 ("ARM: 6072/1: oprofile: use
> > > perf-events framework as backend") is of use to more architectures
> > > than just ARM. Move the code into drivers/oprofile/ so that SH can use
> > > it instead of the nearly identical copy of its OProfile code.
> >
> > Nice, I didn't know such a backend already existed.  Now that you
> > have made it generic we should aim towards making it the only oprofile
> > backend and getting rid of all the duplication.
> 
> Even better would be to do the surgery at a higher level and provide the
> oprofile API without the oprofile buffer management. My experience is
> that it doesn't scale, and on heavily threaded large SMP setup, there is
> an enormous amount of time wasted contending on the global oprofile
> buffer mutex.
> 
Ideally, the OProfile userspace tools would talk native perf and we
could do away with oprofilefs and the oprofile buffer altogether.
Compatibility with the old [read current] OProfile tools can be
maintained using these patches until the API is deprecated.

Or is that a bit too optimistic? :)

Will