Jason Thorpe wrote:
> Just wanted to get tech-kern@netbsd.org on this thread, too :-)

The "thread kernel" for a user-thread implementation (in libpthread
or libc_r in FreeBSD-current) is in user space. *Nothing* in the
"real kernel" is affected => tech-user@netbsd.org?  8-)

The user-thread implementation just uses the best that the real kernel
has to offer. That way, the implementation doesn't get in the way of
a kernel thread implementation later on.

> 
> "Discuss." :-)

If NetBSD's kernel is moving towards everything being done in timespecs,
then the best that the real kernel *should* offer would be a nanopoll
syscall with a timespec argument. Then libc could provide wrappers for
poll(2) and upoll(2).


> Jason R. Thorpe                                       thorpej@nas.nasa.gov

Regards,

-- 
John Birrell                                CIMlogic Pty Ltd
jb@cimlogic.com.au; jb@netbsd.org           119 Cecil Street
Ph  +61  3 9690 6900                        South Melbourne Vic 3205
Fax +61  3 9690 6650                        Australia
Mob +61 18  353  137