mechglue registration of gss_buffer_t pointers

[Nicolas Williams]

On Wed, Oct 31, 2007 at 04:39:26PM -0400, Tom Yu wrote:
> >>>>> "nico" == Nicolas Williams <Nicolas.Williams at sun.com> writes:
> 
> nico> On Tue, Oct 23, 2007 at 09:29:08AM -0400, Sam Hartman wrote:
> >> >>>>> "Nicolas" == Nicolas Williams <Nicolas.Williams at sun.com> writes:
> >> 
> >> What can we do to figure out how well we could do with a registration
> >> mechanism and what performance is required?
> 
> nico> Any registration mechanism would have to be thread-safe and not
> nico> materially slower than an MT-hot malloc()/free().  Alternatively any
> nico> registration mechanism needs to be build-time optional so we can someday
> nico> do code drops and still opt out of it.
> 
> To clarify, are you more concerned about lock contention than about
> CPU time consumption?

Scalability and overall performance relative to the wall clock time to
do a per-message operation.  Which I agree should translate to "lock
contention."