Follow Slashdot stories on Twitter

 



Forgot your password?
typodupeerror
×
GNU is Not Unix Software Linux

Linux 2.6.9 Released 90

An anonymous reader writes "Linux 2.6.9 has been released. Read Linus's official announcement, and go get it!" Better yet, if you hanker for the upgrade, use one of the mirrors instead.
This discussion has been archived. No new comments can be posted.

Linux 2.6.9 Released

Comments Filter:
  • Althout i may be thrillied that they released a 2.6.9 kernel, i dont see any reason to run off and update me 2.6.8 kernel with this one. Am i missing something but i dont see any major improvments (i just glanced at the notes - :) )
  • by Anonymous Coward on Monday October 18, 2004 @11:21PM (#10562056)
    Windows XP firewall says the download is not safe. Better stay away people, and wait until they distribute those on free CDs.
    • Actually it's not the Windows firewall - it's changes to MSIE security that came with XP SP2, in relation to downloading files.

      The file downloads fine thru Mozilla 1.7.3 on WinXP SP2 :-)

      I'm going to compile this after work tonight with GCC 3.4 - wish me luck ;-)

      Selecting all my settings again in make menuconfig is going to take a while :-| I might see if I can use the old .config

      mv /usr/bin/gcc /usr/bin/gcc.old
      ln -s /usr/bin/gcc34 /usr/bin/gcc
  • I wonder... (Score:3, Interesting)

    by brilinux ( 255400 ) on Monday October 18, 2004 @11:24PM (#10562073) Journal
    I have been upgrading kernels ever two or so sub release (I am currently running 2.6.8.1). But it that really necessary? It seems that there is much more upgrading than there needs to be. Obviously, some people will need the new kernel, but it seems that often, once the kernel works well for my hardware, why should I keep upgrading? Any thoughts? I am probably overlooking something obvious.
    • by Stevyn ( 691306 ) on Monday October 18, 2004 @11:44PM (#10562185)
      I'm a gentoo user. I've compiled a new kernel twice this week. I'm not joking. 2.6.8-r9 and 2.6.8-r10.

      I don't need this new kernel. It's an obsession. we need a Linux Users Anonymous
    • Re:I wonder... (Score:3, Interesting)

      by dtfinch ( 661405 ) *
      But it that really necessary?

      Not at all, but 2.6.8 is generally not as stable as say, 2.4.27. If you do something out of the ordinary or pull the plug at random times you may feel the effects of one of the many trickier bugs or misfeatures that may be fixed in newer versions.
    • If I haven't compiled one in a while, I check out kernel.org to see if a new one is out. If it's less than a month since I last upgraded, or if there are RC's out for the next version out, I wait a bit more. I am running 2.8.1 now, and if I get un-lazy enough, the timing is far enough apart that I would be willing to compile 2.6.9, but I probably won't get around to it until 2.6.10 RC1 is out, at which point I will decide it is better to just wait for 2.6.10 proper.
    • I'm wondering why you're even bothering to ask other people for their opinion... you said it yourself, it works for your hardware. Do you download the shiny new version of any other software you run whenever it comes out? I don't mean this offensively, but you've already answered the question you pose, and it seems you just need to learn to think rationally about your needs and to have faith in your own decisions. I upgrade only when I read something from the changelog that I'd like to see working (bette
  • by dtfinch ( 661405 ) * on Monday October 18, 2004 @11:31PM (#10562110) Journal
    Looks important enough for me.
  • Torrent? (Score:5, Funny)

    by schnits0r ( 633893 ) <`nathannd' `at' `sasktel.net'> on Monday October 18, 2004 @11:33PM (#10562116) Homepage Journal
    Does somebody have a torrent of this yet? If they do, coudl they posta link so we can slashdot it and downlaod it faster...wait, that sounded weird.
    • Re:Torrent? (Score:2, Interesting)

      If you /. the tracker, it doesn't help the torrent any. You'd need a multitracker solution.
    • Re:Torrent? (Score:3, Informative)

      by Anonymous Coward
      Bittorrent is only really useful if the initial distributor has very limited bandwidth and/or a poor mirroring system. Neither is true for kernel.org, so just download it from ftp.<countrycode>.kernel.org and be happy, no need for a torrent.
  • Buggy of Bug Free (Score:2, Interesting)

    by gmhowell ( 26755 )
    Lacking 'safe' vs. 'devel' kernel branches, what category does this point release fall into?
    • A while ago, it was mentioned that 2.6 is now basically a devel branch. The stabilisation of the kernel has been left up to the distributions. That said, I consider it to be fine to use as long as you're not using a production system that requires excellent stability.
      -ReK
      • I ran 2.5.70-bk12 or some similar version for over a year on my semi-production (if it does not work I do not have internet) internet-router/file-server/irc-bot-server/console -irc-client/console-icq-client/...-server at home for over a year. The only time I was forced to reboot was a power outage. Linux is not exactly unstable even in the development branch as long as you stay away from modules marked as "Experimental" or "Dangerous".
    • We don't lack 'safe' vs 'devel' branches. We have the mm-tree and the linus-tree.

      Now, the tricky part may be to determine which is what. But judging from the 2.4-situation - where the ac-tree was 'safe', and the linus-tree was 'devel'; I think either the linus-tree is 'devel' this time too, or Linus is getting rusty. ;)
  • Have anyone tried SATA discs with the Linux kernel lately?

    I heard it should be in progress.

    • I've only tested it briefly, but it WFM with Fedora Core 2...

      I pre-tested a box for a friend, and he's using SATA. He hasn't complained yet (after telling me that Redhat 9 didn't install due to SATA issues. Installed FC2, now no complaints.)

      So it think it's safe to say that it's well and truely "in progress"...
    • It worked for me under Fedora Core 2 (although my ATI card did not, 9600 xt)
      also, SATA worked under unbunto (sp) distro. Actually like that desktop distro...

      I have to admit I did not kick the tires too hard by running oracle or anything else like that.

      SPP
    • We've been thrashing it out on a test box. Only had one problem and I'm not sure whether or not is SATA related.. The mount magically appeared full. I couldnt anything that was taking up all the space. A reboot and its all magically back to normal. And were talking about 35gb suddenly going missing...

      Apart from that, everything on the SATA drive has been fine.

    • Im currently writing this on my machine with a Gentoo 2.6.9 kernel(had 2.6.9-rc2 on) with an SATA drive. It works great!!! Much much faster then any IDE drive. Ive had no problems with it. To enable SATA you need to do it through SCSI, just enable SCSI and in SCSI low level drivers select your chipset. Incase you wanted to know here are my system specs. System Specs CPU: AMD Athlon 64 3500+ Mobo: Asus A8V SATA Controller: Promise TX4 Soundcard: SB Audigy 2 zs Videocard: NVIDIA FX 5900 Ultra
    • I'm using the sata_promise and libata modules with my Promise SATA150 TX2Plus on 2.6.8.1. Works great with the two SATA drives I have installed (one hot-swapped in a Kingwin enclosure, one fixed in the case.) The PATA (IDE) connection doesn't work AFAIK, but that's not a big deal for me. (And the hardware isn't supposed to work with optical drives, which is a bigger deal, but not the fault of Linux.)
  • Can someone who understands the Kernel numbering scheme explain it to me?

    Is 2.4.x the stable branch and 2.6.x the development branch? Or is 2.6 stable, and there's already a 2.7.x development branch? Or how does that work?

    And are all kernel modules guaranteed to maintain strict binary compatibility across all 2.4 releases, or alternatively across all 2.6 releases? Or is it source compatibility only? Or, is it even that?
    • There are no stable and development branches anymore. For kernel 2.4 the development branche was 2.5 but there will be no 2.7 development branch for 2.6

      • Correct me if I'm wrong but didn't Linus say [licklinux.com] that they didn't really need a 2.7 series yet? My feeling was that they didn't plan on anything revolutionary that would warrant branching, and Linus thought the way 2.6 was being handled was going well. Maybe he just doesn't want to hand over the reigns to Andrew Morton (the "official" 2.6 maintainer, and proprieter of the infamous mm patches).

        I still don't know how I feel about having to have a kernel called 2.6.5-1.358.8kstacks (redhat with nvidia modific
        • If I remember correctly 2.5.0 (the start of the 2.5-development-branch) was identical to 2.4.15. The 2.6 stable branch simple needs some time. The early 2.4 were not exactly stable either.
    • by turgid ( 580780 ) on Tuesday October 19, 2004 @06:48AM (#10563538) Journal
      And are all kernel modules guaranteed to maintain strict binary compatibility across all 2.4 releases, or alternatively across all 2.6 releases? Or is it source compatibility only? Or, is it even that?

      No. Even different compiles of the same kernel can be incompatible as far as modules are concerned. It depends on the compiler (and version) you use, the kernel and patches and the configure options. When you compile yourself a new kernel, you should rebuild and reinstall all your modules. You will also have to "recompile" closed-source binary drivers. There come with an open-source "shim" layer to interface with the kernel c.f. nVidia drivers. You then need to go into /etc and frob with the scripts that load the modules at boot time.

      The rationale for this design decision was to force vendors to either provide GPL'd drivers for their hardware or at least to open the specs. so that volunteers could implement them themsleves. This has been largely successful, but there are a number of significant instances where this has been a problem, for example accelerated 3D graphics drivers. nVidia has been providing binary modules with a shim for years now, and recently ATI has started doing the same. There are various reasons why in the "real world" drivers can not be open-sourced and specs. can not be divulged. Sadly, we do not live in a GNU utopia.

      So, for idealogical reasons, we have this dreadful system of driver modules in Linux.

      Please note, I'm actually quite a Linux fan. I've been using it exxclusively at home since 1996 (Slackware all the way). I just think it's maybe time for Linux to grow up and take a leaf out of Solaris' book, for example, and to provide a _stable_ (i.e. unchanging between kernel minor versions at least) binary interface for device drivers and other kernel modules. This would make my life a lot easier and cut down on the recompiles.

      Sadly, I fear the ideology might get in the way...

      • Sadly, I fear the ideology might get in the way...

        It's a difficult trade-off. If there was a stable way for hardware manufacturers to release binary Linux drivers, there would be a whole lot less incentive to give out specs / driver code among those who are half-hearted about supporting OSS. Sure, we might see almost all hardware supported in Linux, but almost none of the drivers would be open source.

        I think the current scenario is the best balance for now. Those who play nice with the community have
    • 2.6 is SUPPOSED to be the stable branch, but there are no guarantees, especially since the distros are supposedly responsible for stabilizing it. However when/if 2.7 comes out, I'm SURE it will be unstable. You can still count on that.
  • I've heard that there were some changes in 2.6.8 to disallow non-root users access to recorders (something related to firmware, was it?).

    Could someone comment on this, and has the behavior been changed in 2.6.9?
    • With the 2.6.8 kernel and K3b the very nice KDE cdburning suite, non root users could not burn CDs. I think this was fixed with 2.6.8.2 Most distro's put in their own patch also.

      I don't know that it affected other cd burning tools.
  • 2.6.9 and Nvidia (Score:2, Informative)

    by zenengnr ( 776098 )
    If you use the propietary nvidia drivers.... Forget it, DONT UPGRADE to 2.6.9 ! Its very very broken :(
    • Video, network, or both?
      • The fix for the nvidia "graphics" driver is here http://ngc891.blogdns.net/kernel/patches/NVIDIA-Li nux-x86-1.0-6111-jp1.tar.bz2 but now my vmware modules wont compile!! Apparently the kernel developers have broken things that effect most external modules. Maybe the should fork 2.7 branch before they start breaking shit !!!! Just one asshole's oppinion.
  • by tuxq ( 703148 ) on Tuesday October 19, 2004 @02:03PM (#10568010) Journal
    This works ... I grabbed it out of a patch file...
    Add this to arch/i386/mm/init.c
    On line 43 right below unsigned int __VMALLOC_RESERVE = 128 20;
    add this line...
    ----- Begin -----

    EXPORT_SYMBOL(__VMALLOC_RESERVE);

    ----- End -----

    nVidia drivers WILL work. I'm using 2.6.9 and 6111 nvidia drivers right now.

    Btw... http://tuxq.com/~tuxq/wtf.jpg ... Why the hell on kernel.org is 2.2.19 the latest stable... lol
  • Or should I wait for 2.6.9.1.XYZ?
  • Updated UML Support (Score:3, Interesting)

    by rimu guy ( 665008 ) on Tuesday October 19, 2004 @11:57PM (#10572693) Homepage

    UML support was added to the 2.6 kernel a while back (2.5.34 [kerneltrap.org] in Sep 2002).

    Since then the mainline kernel has lagged behind the latest UML releases on user-mode-linux.sf.net [slashdot.org].

    Over the 2.6.8 to 2.6.9 timeframe BlaisorBlade [user-mode-linux.org] (aka Paolo Giarrusso) has worked with Andrew Morton and Jeff Dike to bring the mainline kernel up to date with the latest UML changes. (To the point where the 2.6.9 kernel is more current than the latest 'official' UML release). I would guess this was the biggest, in terms of lines of code, change in 2.6.9. Most of the changes just touched the 'um' architecture though. So changes are pretty isolated from other arch-es.

    This may be of interest to you if you run chrooted systems anywhere (UML may be more secure). Or if you are a kernel hacker (so much easier to debug things that run in a user process).

    --
    (UML-based) VPS Hosting [rimuhosting.com]

  • 2.6.8 introduced a bug in which it makes k3b unable to detect the cd writers as normal user and would require root only.

    Has that been fixed in 2.6.9?

To be is to program.

Working...