Fine grained locking at the socket level?

[Bosko Milekic]

On Sun, Jun 22, 2003 at 04:58:54PM -0700, George V. Neville-Neil wrote:
> Hi,
> 
> 	It would seem that splnet() and frieds now simply return 0,
> 	which I figure is part of making the code look like it used
> 	to.  What I'm wondering is why the Giant lock is still used in
> 	the socket layer?  I thought sockets had had fine grained
> 	locking applied to them.  Am I confused?  I'm looking at the
> 	bits du jour (-CURRENT).
> 
> Thanks,
> George

  The short answer is: we're not done.

  The long answer is: we're not done. :-)  We can't simply unwind Giant
  just anywhere yet because there is still code in other layers that
  requires Giant.

Cheers,
-- 
Bosko Milekic  *  bmilekic at technokratis.com  *  bmilekic at FreeBSD.org
TECHNOkRATIS Consulting Services  *  http://www.technokratis.com/