Re: Documenting LWKT?

[Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx>]

:Hi all, how solid is LWKT? I'm keen to get involved in dfly, especially 
:in the SMP area. I thought that documenting some of the new kernel 
:functionality would be a good way to get a grip on the kernel code.
:
:Cheers,
:	Warren
:
:P.S Credentials: FreeBSD 1.x kernel hacking, 12 yrs of teaching OS at 
:uni, recent work on adding SMP to VSTa (www.vsta.org), maintainer of the 
:Unix Historical Society (www.tuhs.org).

    LWKT is extremely solid and well-defined now.  In fact, just today I
    will be committing LWKT messaging 'abort' support (which the network
    code is going to have an immediate use for).  This is the last major
    messaging piece that was in the original LWKT messaging abstraction.
    And no significant changes have been made to the LWKT scheduler in weeks.
    The LWKT token code is stable but might go into flux later on as we
    refine it.  The token abstraction is solid, though, and probably will
    not be changing any time soon.

    I think that this is a good time to push documentation of the LWKT
    functions.  I would dearly love to see some manual pages written up
    on them.

    Are you volunteering?  I would also be very interested in seeing how
    easily someone who has not been deeply exposed to the LWKT interface
    can understand the code, both so I can better document the code and
    so we can give general kernel@ list users some exposure to the 
    LWKT abstractions from a non-Matt-Dillon perspective.

					-Matt
					Matthew Dillon 
					<dillon@xxxxxxxxxxxxx>