On Feb 12, 2007, at 1:56 PM, Perry E. Metzger wrote:

> I believe the newlock stuff was started by Jason Thorpe, actually, but
> Andrew certainly brought it to fruition.

That's correct.  I started the effort back in 2002 -- I realized that  
we were never going to get certain of our MP issues fixed until we had  
better synchronization APIs.  I used Solaris as a model, for a couple  
of reasons:

1- Solaris is widely regarded as a very high-quality MP implementation.

2- The Solaris synchronization APIs and implementation were thoroughly  
documented.

Andrew Doran has done a fantastic job bringing the code forward and  
improving it, and we all owe him our thanks and congratulations.  I  
sincerely appreciate his work in this area (and wish that I had more  
time to contribute to it directly myself, these days).

BTW, a nice side-effect of adopting Solaris-like APIs for this stuff  
-- makes it easier to port things like DTrace and ZFS :-)

-- thorpej