Announcing DragonFly BSD! http://www.dragonflybsd.org/ Hello everyone! For the last few months I have been investigating and then working on a new approach to the BSD kernel. This has snowballed into a far more ambitious project which is now ready for wider participation. It is the intent of this project to take over development of the 4.x tree, to move kernel development along an entirely new path towards SMP, and to completely rewrite the packaging and distribution system. We eventually intend to backport many FreeBSD-5 features into the new tree, but that is not where the initial focus will be. The preliminary 'proving' work I have done is now available on the new DragonFly site. You can access it through cvsup or browse it through ftp. This proving work involved implementing much of the earlier UP->SMP converstion work that was done when 5.x first branched, but under an entirely new mutex-free light weight kernel threading infrastructure. It includes the LWKT system, interrupt threads, and pure threads for system processes amoung other things. For obvious reasons the codebase will only run on i386 for now, and ports to other platforms will not happen until the MD infrastructure is cleaned up and finalized. I considered starting with a 5.x base but it is simply too heavily mutexed, it was actually faster to start with 4.x and move forward rather then to start with 5.x and move backwards. I have both UP and SMP builds working in the current codebase. I believe it proves out the core concepts quite nicely and there is much more work coming down the pipeline. The site is: http://www.dragonflybsd.org/ Hopefully my T1 can handle the cvsup load. Eventually I'll colocate some boxes to deal with that issue. For the next few months the project is going to concentrate on low level kernel development. There are still a number of big ticket items that have to be accomplished, primarily in converting the I/O path to using VM Object/range lists, before work can branch out into other areas. I expect the project to start fairly slowly but then for momentum to build. Anyone interested in working on or discussing the project is welcome! I have created a mailing list server and newsgroup forums and I am working on web-accessibility to same for passive listeners. I will be posting periodic updates to freebsd-hackers as well. Again, the site is below. It contains a great deal of documentation and other information. I even have a mascot! And, hopefully, it will all work from outside my LAN :-) http://www.dragonflybsd.org/ -Matt
Matthew Dillon wrote:> Anyone interested in working on or discussing the project is welcome! I > have created a mailing list server and newsgroup forums and I am working > on web-accessibility to same for passive listeners. I will be posting > periodic updates to freebsd-hackers as well.I'm especially encouraged by the committment to fix the VFS subsystem so that stackable filesystems will really work, by the caching/locking discussion, and the acknowledgement that system configuration and packages need a publish-subscribe (not Matt's words) mechanism. Manuel Kasper's m0n0wall configuration system, XML-based, is really cool. You could easily extend it to signed XML for trusted packages/components. Crypto and ACL filesystems could finally be done in a modular, stackable way. Esp. if the messaging subsystem works as advertised. This announcement has made an otherwise dreary and mind-numbing day at work a little better, thanks.
On Wed, Jul 16, 2003 at 12:42:07PM -0700, Matthew Dillon wrote:> Hello everyone! For the last few months I have been > investigating and then working on a new approach to the BSD > kernel. This has snowballed into a far more ambitious project > which is now ready for wider participation.Hi Matt, I hope this project manages to explore new boundaries of performance and am pleased that you chose the FreeBSD stable development branch as a suitable starting point for your endeavors. I hope Dragonfly will produce the same kinds of synergies with the FreeBSD Project that our ongoing relationships with NetBSD, OpenBSD, and various Linux projects have. I'm especially interested in seeing you succeed with your package management goals, where many other attempts have failed in that area. For our part, we're planning to release FreeBSD 4.9 with PAE support merged at the beginning of September. We certainly haven't ruled out further 4.X releases after that time either. As usual, our release related plans are available from http://www.FreeBSD.org/releng. Good luck, - Murray -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 155 bytes Desc: not available Url : http://lists.freebsd.org/pipermail/freebsd-stable/attachments/20030717/0065cf35/attachment.bin
On Wed, 16 Jul 2003, Matthew Dillon wrote:> Announcing DragonFly BSD! > http://www.dragonflybsd.org/I will certainly track the progress with interest! I will have to finish browsing the site before forming any real opinion... It seems you have some very good goals, but why not just work to make FreeBSD (which I'd think has more visibility, and hence a chance of benefitting more people) better? :) I will admit, I can see how it could be easier to just start your own project from scratch, but I can also see downsides. Hopefully, as others mentioned, the insight you gain can be back ported to *BSD and/or other existing OSes as well.> Hello everyone! For the last few months I have been investigating > and then working on a new approach to the BSD kernel. This has snowballed > into a far more ambitious project which is now ready for wider > participation.This is my primary concern... If the ideas for improving the kernel, vm system, etc. are desireable... Why can't they be successfully implemented in the existing codebase? Was there significant resistance within the project, or did the new methodologies simply encompass too much destabalizing (at first) code change?> ftp. This proving work involved implementing much of the earlier UP->SMP > converstion work that was done when 5.x first branched, but under an > entirely new mutex-free light weight kernel threading infrastructure. > It includes the LWKT system, interrupt threads, and pure threads for > system processes amoung other things.Mmm.> Hopefully my T1 can handle the cvsup load. Eventually I'll colocate > some boxes to deal with that issue.I'm sure you will have no difficulty finding mirrors, should the need arise. Good luck, -mrh -- From: "Spam Catcher" <spam-catcher@adept.org> To: spam-catcher@adept.org Do NOT send email to the address listed above or you will be added to a blacklist!