Re: [rfc][patch] x86-64 new smp_call_function design

[Nick Piggin <npiggin@suse.de>]

On Wed, Feb 27, 2008 at 02:27:13PM +0100, Ingo Molnar wrote:
> 
> * Nick Piggin <npiggin@suse.de> wrote:
> 
> > This isn't finished yet, however I'd just like to ask for comments.
> 
> looks really interesting!
> 
> only one fundamental observation:
> 
> > +struct call_data {
> > +	spinlock_t lock;
> > +	struct list_head list;
> >  	void (*func) (void *info);
> >  	void *info;
> > +	unsigned int flags;
> > +	unsigned int refs;
> > +	cpumask_t cpumask;
> > +	struct rcu_head rcu_head;
> >  };
> 
> > +struct call_single_data {
> > +	struct list_head list;
> > +	void (*func) (void *info);
> > +	void *info;
> > +	unsigned int flags;
> > +};
> 
> the two structures are quite similar in size and role - why not have a 
> type field and handle them largely together? I think we should try to 
> preserve a single queue and a single vector - that would remove a number 
> of ugly special-cases from the patch.

A single queue will kill one of the big fundamental scalability
improvements of the call_single. That's the problem.