DragonFly users List (threaded) for 2007-08
DragonFly BSD
DragonFly users List (threaded) for 2007-08
[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]

Re: DRAGONFLY 1.10 RELEASED!


From: Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx>
Date: Tue, 7 Aug 2007 13:03:49 -0700 (PDT)

:Hi, I want to make a couple of questions to Matt, for a mini interview 
:for my site SoloBSD.org
:
:First of all, is obvious that when you forked from FreeBSD, you had in 
:mind a roadmap for DragonFly BSD. Where are we now in that roadmap? What 
:are the goals for the future?

    We are well over half way.  All the major rewiring of the kernel except
    perhaps the VM system is done.  The platform is ready for the component
    work needed to actually do clustering.  In a sense you could say that
    only the 'fun stuff' is left to do, which is a big relief to me because
    rewriting half the kernel was a necessary but also fairly unrewarding
    experience for me.  Now I get to do some fun stuff!

:And taking FreeBSD as comparison, what is DragonFly like? Are we like a 
:FreeBSD 4.20? or are we like FreeBSD 5? or are we better than FreeBSD 6? 
:or are we totally different from FreeBSD?
:
:Thanks!

    I probably reference FreeBSD too much.  When I refer to FreeBSD-4 I
    am actually talking about its stability, not its code base.  FreeBSD-4
    is still the benchmark for stability.  Probably the most uncompromising
    position I have taken with the DragonFly project is that stability must
    be maintained through all the work that is done, and I think we've
    accomplished that.

    After years of work the DragonFly kernel is not even remotely similar
    to any version of FreeBSD any more.  The internal infrastructure is
    totally different.  We have our own threading (LWKT), light weight
    processes, virtualization technology, our own spin locks, our own
    kernel memory allocator (The slab allocator and Objcache), a replicated
    routing table, threaded network protocol stacks, check pointing,
    cpu-localized timers and callouts, our own network driver serialization
    infrastructure, our own implementation of major TCP RFCs, multiple
    compiler support, and the list goes on.

    If any continued similarity could be argued it is almost entirely
    relegated to hardware device subsystems, and even there we generously
    borrow from all the BSDs as well as Linux.  The ATA driver is ported
    from FreeBSD, a good chunk of USB is ported from NetBSD and OpenBSD.
    The network drivers are a combination of ports and our own rollouts,
    and so on.

    Overall I would say the DragonFly is its own project these days.

					-Matt
					Matthew Dillon 
					<dillon@backplane.com>



[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]