Re: uvm_object::vmobjlock

[Eduardo Horvath <eeh%NetBSD.org@localhost>]

On Thu, 28 Jan 2010, Mindaugas Rasiukevicius wrote:

> Antti Kantee <pooka%cs.hut.fi@localhost> wrote:
> > > Unless anyone objects, I would like to change struct
> > > uvm_object::vmobjlock to be dynamically allocated with mutex_obj_alloc
> > > ().  It allows us to:  1) share the lock among objects by holding a
> > > reference  2) avoid false-sharing on locks. Note that struct
> > > vnode::v_interlock becomes a pointer, which means a chunk of mechanical
> > > changes.
> > 
> > 1) can you give an example of where this is useful?  (i'm not saying it
> >    isn't, i just don't like thinking ;)
> 
> Further locking improvements:
> 
> http://mail-index.netbsd.org/tech-kern/2009/01/11/msg003986.html

Since this is supposed to be an optimizaton, do you have performance 
measurements?  And since it's in MI code, on multiple platforms?

Eduardo