Re: [PATCH 0/3 v2] Use lockless lists for smp_call_function_single()

[Christoph Hellwig <hch@infradead.org>]

On Thu, Dec 19, 2013 at 08:31:54AM -0700, Jens Axboe wrote:
> On Wed, Dec 18 2013, Jan Kara wrote:
> >   Hello,
> > 
> >   these three patches make smp_call_function_single() use lockless lists and
> > cleanup abuse of csd.list in the block layer. I have some use for lockless
> > smp_call_function_single() in printk code and that's why I've picked up
> > Christoph's patch. Jens, can you have a look at the series please?
> > 
> > Changes since v1:
> > * Reordered patches to a more sensible order.
> 
> Thanks Jan, I'll throw this into the testing mix for 3.14. Will be a
> little delayed over christmas, with holidays and family visiting.

Did you make any progress?  Btw, I'd love if you could queue up the
patch to use __smp_call_function_single directly in the block layer
for 3.14.  Independent of wether a llist using
__smp_call_function_single is faster than one using the lock we get
down from a list plus llist to just one of them, so I can't see how
that patch will cause an issue.  With all the prep patches already in
it can be queue up independently of Jan's series as well.