On Wed, 2015-06-10 at 09:34 +0200, Christoph Hellwig wrote:

Hi Vishal,

I'm mostly worried about handling scalability to large CPU counts
properly.  If you think this is the best way to handle it that fine,
but please document the decisions on the changelog in a similar form
to what you did below.

Thanks Christoph.

I'll update the changelog for the next version. Scalability and
performance are of course at the top of my head - this implementation is
a first shot at what we think will work best, but if there are
experiments/changes/data that show otherwise, I'll be more than happy to
incorporate those.

	-Vishal