OpenBSD Journal

SMP is now in HEAD

Contributed by Peter Hessler on from the dept.

According to this SMP is now in -current. If you havn't been testing the SMP snapshot kernels, its now in -current. Please report any breakage, since this is a BFD.

(Comments are closed)


Comments
  1. By Anonymous Coward (213.119.4.16) on

    Yay! Thanks for all the hard work!

    Comments
    1. By phessler (208.201.244.160) spambox@theapt.org on http://theapt.org/openbsd/

      Please test, and help to make sure this doesn't break anything that used to work. That is super important.

      Comments
      1. By Nate (209.162.235.146) on

        There is a box at work we setup to run a few days ago using the multi kernel, just set it to making packages of everything, it hadn't run into any problems after a day of making packages. We intend on setting up another to do the same if we can come up with enough spare hardware.

    2. By Anonymous Coward (68.147.189.74) on

      Do I read the list of changes right and see this is more then just SMP in i386, and does anyone know if it supports missmatched processor modules on sparc like the older SunOS did? I have a system that ran SunOS 5 which had a 40mhz and a 50mhz processor modules, and smp worked fine.

      Impressive work though, Seems like just yesterday development spead up on it and here already theres something.

  2. By Anonymous Coward (142.166.109.145) on

    I see files altered for many processors, but which architectures are supported (or going to be supported)? Anyone know?(I only have/need single CPU systems, but I'm as curious as the next guy).

    Comments
    1. By Brad (216.138.200.42) brad at comstyle dot com on

      Only i386 at the moment though I'm sure you'll see amd64 sometime soon.

      Comments
      1. By Anonymous Coward (194.250.174.227) on

        ... right when they're finished importing it from NetBSD.

        Comments
        1. By Philipp (53.122.34.116) pb@ on

          I dont think so.

          Comments
          1. By Anonymous Coward (194.250.174.227) on

            Where do you think OpenBSD's amd64 and SMP support is from?

            Read the licenses; and yes, this is all that 4-clauses BSD Theo has been so vocal about.

            Comments
            1. By gwyllion (134.58.253.113) on

              Yes a lof of code was written by Bill Sommerfeld, but not all of it. Look at who wrote sys/arch/i386/i386/db_mp.c and it's license.

    2. By kris (67.160.110.105) on

      Why AMD64 ... I think many would love to see it on Sparc/Sparc64. Sparc hardware is much cheaper than x86 if you know where to get it, and IMHO, OpenBSD on Sparc* runs better than x86. Thats just my $0.02

      Comments
      1. By Brad (216.138.200.42) brad at comstyle dot com on

        Availability of hardware, the fact that its our fastest architecture and code sharing with i386.

        Comments
        1. By kris (67.160.110.105) on

          Do you need multi-proc sun gear? its cheap to get, and i am sure many people will have some to spare.

          Comments
          1. By conne (213.224.83.166) conne_@pandora.be on

            mail me ;)

      2. By Anonymous Coward (195.217.242.33) on

        There are test kernels for sparc

        Comments
        1. By Anonymous Coward (213.119.4.16) on

          Yes, uniprocessor test kernels. Not SMP ones.

          Comments
          1. By Anonymous Coward (195.217.242.33) on

            Hmmmm .... it appears you are right

            bollocks !

            Comments
            1. By Anonymous Coward (134.134.136.1) on

              Bleep. You said it, I could care less about SMP support for x86 systems, as I run OpenBSD on my Sparc/UltraSparc systems. Sparc SMP systems are a lot more common that x86 ones, and cheaper to. I've got several Sparc 20/712's, an Ultra 2/2170, and an Ultra 60/2300 (I don't think I have any Sparc 10's left). I really hate having to pull a CPU to run OpenBSD on one of these boxes, so am currently using a Sparc 5/110, but am starting to feel the need for a faster system.

      3. By Anonymous Coward (81.26.253.124) on

        Care to share your sources? I'd love to buy sparc hardware at x86 prices. As long as performance is in the same ballpark, of course.

        Comments
        1. By k3k (81.17.198.221) on helzinki.net

          Me too. Nowhere to be found in Finland...

        2. By Anonymous Coward (195.217.242.33) on

          eBay

  3. By Anonymous Coward (24.19.177.166) on

    woot!

  4. By Michael van der Westhuizen (198.54.202.2) on https://skanky.homeunix.net/

    Big congrats to niklas@ and the rest of the OpenBSD developers - this is fantastic work.

  5. By Anthony (68.145.111.152) on

    I'm going to be very disapointed if, by the time I get to work, this hasn't a) been posted Slashdot, b) made it onto the front page, and c) turned into a security-vs-performance flame fest.I haven't suffered from the lack of SMP, but with the desktop dual-core chips set to arrive next year, this is well timed. Nicely done.

    Comments
    1. By Anthony (68.145.111.152) on

  6. By Evren (212.156.230.82) evbim at gmx dot com dot tr on

    This is a good work for community.. After stability of SMP what will happen in the OS world? Let's wait and see the rise of OpenBSD..

  7. By Anonymous Coward (4.61.199.192) on

    And now, everyone who has been collectively bitching to get it for years now, have a moment of solitude, and realize that they don't even have any SMP hardware at all...

    :-)

    Comments
    1. By Philipp (53.122.34.116) pb@ on

      Anyway this was a big step. And several patches now in HEAD will help other archs reaching SMP a lot, too.

    2. By Anonymous Coward (64.229.125.92) on

      ROFLMAO! Good one!

    3. By Anonymous Coward (141.213.74.30) on

      (Except of course for everyone who bought a Pentium 4 in the past year)

  8. By Xardion (65.117.155.94) on

    HOOAH! My little dual proc ss20 has been waiting for this day to come. It will finally bear its biscuit-filthy fangs against its oppressive masters and rise above them to rule the UNIVERSE!! MWUAHAHAHAH.... MWUUAAAAHAHAHHAHAHAHA!!! </megalomaniacal rant>

  9. By Anonymous Coward (195.217.242.33) on

    Now the trolls can rest ... but ... for how long ?

    Comments
    1. By Anonymous Coward (82.4.233.223) on

      There's plenty of other things for the trolls, to troll about.

      Comments
      1. By X (80.15.150.98) on

        Like 802.11g wifi card ?

        Comments
        1. By Eduardo Alvarenga (200.242.254.7) eduardo@thrx.org on

          It does not depends on OpenBSD but on the documentation vendors may provide. If no docs, no drivers.

  10. By Anonymous Coward (80.65.225.73) on

    Would tests on uniprocessor hardware be helpfull (eg to see if something is broken) ? I sadly don't have spare smp machines but want help

    Comments
    1. By Sean (24.191.38.89) on

      I believe one of the main points Theo tried to make was the nessesity that the smp kernels work on uniprocessor systems.

  11. By grey (207.215.223.2) on

    As an aside, it irks me a bit to see this story posted before there was enough to report, especially since it was requested that this information be kept close to the breast until it was in a state for wider release. I even made a point of this in other thread here. on undeadly.

    This isn't to chew out phessler too much (he's already got an earful elsewhere), but undeadly readers should be aware that this work is in preparation for something bigger heavy lifting exercises (for those who aren't aware c2k4 is almost upon us). We have known about these efforts for a while, but had not posted them intentionally and I wish it hadn't slipped out in this manner. If you want to see where efforts really started to become public on the mailing lists, check: http://marc.theaimsgroup.com/?l=openbsd-tech&m=108672757809273&w=2.

    Comments
    1. By jose (12.7.85.10) on http://monkey.org/~jose/

      a few people always bitch about this, but let's be honest here. it was checked into CVS. people saw it.

      the whiners can deal, they always live after such "catastrophic" events.

  12. By Bayu Krisna (2001:470:1f01:ffff::7b) moc.tenidap@ansirk on http://www.versalite.com

    Congrate!!!! .... Great job bro.

  13. By Peter (204.38.166.3) peter@pbw.us on http://pbw.us

    Even though I don't have many SMP computers (only 1 to be exact!), this is great to hear. I think a lot of people have been looking for this support in OpenBSD. Thank you very much for all the time and hard work OpenBSD! ~ Peter

  14. By Anonymous Coward (65.60.138.126) on

    Can anyone point to an article or other technical overview that includes some discussion of the current state of the NetBSD smp and how it compares to FreeBSD, Linux, Win32 etc?thanks

  15. By Zerash (24.73.230.118) zerash[at]metawire.org on http://www.metawire.org

    Absolutely fantastic. Alot of us have been waiting for this; say hello to OpenBSD in the enterprise. :-) Regards, -Daniel

  16. By Anonymous Coward (211.30.145.52) on

    Well, I have a pair of Dual PIII (one is 1Ghz, the other 933Mhz)...I could give it go once my exams are done...

  17. By BSDero (200.33.72.80) bsdero at gulags dot org on http://bsdero.gulags.org

    What's up with the sparc/smp support? Will there be Sparc/SMP kernels for testing soon? I'd like to test this new feature...

    BSDero

    Comments
    1. By Nate (209.162.224.62) on

      No, as Theo has already said on the misc list, there is no code for anything but i386.

Latest Articles

Credits

Copyright © - Daniel Hartmeier. All rights reserved. Articles and comments are copyright their respective authors, submission implies license to publish on this web site. Contents of the archive prior to as well as images and HTML templates were copied from the fabulous original deadly.org with Jose's and Jim's kind permission. This journal runs as CGI with httpd(8) on OpenBSD, the source code is BSD licensed. undeadly \Un*dead"ly\, a. Not subject to death; immortal. [Obs.]