Subject: Re: Please Revert newlock2
To: Bucky Katz <>
From: Daniel Carosone <>
List: tech-kern
Date: 02/21/2007 17:01:58
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline

On Tue, Feb 20, 2007 at 08:21:05PM -0800, Bucky Katz wrote:
> If I haven't used up too much of my "cred", would it be possible to
> get all of our fixes pulled up into 4.0,?

We're very interested in fixes for SA (and related pthread, and
anything else). These will be in the upcoming 4.x release, which we'll
be supporting for some considerable time yet (until 6.x, according to
our usual maintenance policy to date).

Let me emphasise, in case it got lost along the way, that SA pthreads
are not being abandoned, even if they turned out to get in the way of
other more extensive and radical kernel infrastructure redevelopment.
This is the threading implementation that the majority of users will
be using for the reasonable future, and your efforts and contributions
towards its improvement are most valuable indeed.



PS: It's also worth noting that this is one of the few cases where the
usual kernel-interface stability and back-compatibility we normally
strive very hard to maintain (eg, binaries from 0.8 still run) simply
wasn't feasible.  The fact that the stable interface had to be moved
out to libpthread, and the tight coupling of that with the kernel
internals, was just another aspect of the problem, much as SA seemed
attractive and clever in the first instance.  Solaris takes a
different approach and puts the boundary at the .so; one can argue the
merits of either choice, but the fact remains that SA pthreads was a
big mismatch with the stable boundary NetBSD uses for the rest of the

Content-Type: application/pgp-signature
Content-Disposition: inline

Version: GnuPG v1.4.6 (NetBSD)