Jens Axboe wrote:

> Looks good, I've been toying with something very similar for a long time
> myself.
> 

Here is another thing I just noticed that should further reduce the
locking by at least 1, sometimes 2 lock/unlock pairs per request.
At the cost of uglifying the code somewhat. Although it is pretty
nicely contained, so Jens you might consider it acceptable as is,
or we could investigate how to make it nicer if Kenneth reports some
improvement.

Note, this isn't runtime tested - it could easily have a bug.