Re: CPU ordering with respect to krefs

[Greg KH <greg@kroah.com>]

On Mon, Apr 02, 2007 at 04:33:54PM +0200, Eric Dumazet wrote:
> On Mon, 2 Apr 2007 14:47:59 +0200
> Oliver Neukum <oneukum@suse.de> wrote:
> 
> > Hi,
> > 
> > some atomic operations are only atomic, not ordered. Thus a CPU is allowed
> > to reorder memory references to an object to before the reference is
> > obtained. This fixes it.
> > 
> > 	Regards
> > 		Oliver
> > Signed-off-by: Oliver Neukum <oneukum@suse.de>
> > ------
> > 
> > --- a/lib/kref.c	2007-04-02 14:40:40.000000000 +0200
> > +++ b/lib/kref.c	2007-04-02 14:40:50.000000000 +0200
> > @@ -21,6 +21,7 @@
> >  void kref_init(struct kref *kref)
> >  {
> >  	atomic_set(&kref->refcount,1);
> > +	smp_mb();
> >  }
> 
> I dont understand why smp_mb() is needed here, and not in
> spinlock_init() for example.

I think, after reading the Documentation/memory-barriers.txt and
Documentation/atomic_ops.txt documentation, that spin_lock_init() also
needs this kind of memory barrier.

>From what I can tell (Oliver, please correct me if I'm wrong, you know
this much better than I do), the issue is that atomic init has no memory
barrier, and you need to handle that.

thanks,

greg k-h