Slashdot Mirror

← Back to Stories (view on slashdot.org)

Linux 2.4.22 Stable Kernel Released

Posted by simoniker on from the latest-and-greatest dept.

An anonymous reader writes "Marcelo Tosatti has officially released another stable 2.4 Linux kernel. 2.4.22 was released early this morning and includes a lengthy list of fixes. It follows the last stable kernel in this tree, 2.4.21, by a little over two months."

19 of 342 comments (clear)

  1. Too fast... again! by ospirata · · Score: 1, Interesting

    Doesn't this realese seems like the 2.4.13? The RCs came out just one week after the other. Please God, hope to have no VM issue!

    1. Re:Too fast... again! by BillKaos · · Score: 3, Interesting

      There are not big core changes in this kernel, so it's believed to be very stable. Also, it includes some security fixes so you should upgrade.

  2. 2.4 VS 2.6 Performance by Hornsby · · Score: 4, Interesting

    I'm curious if anybody has experienced dramatic performance increases running X when switching from 2.4 to the 2.6 testing branch of the kernel.

    --
    A musician without the RIAA, is like a fish without a bicycle.
    1. Re:2.4 VS 2.6 Performance by 955301 · · Score: 5, Interesting

      Absolutely! I'm not sure how to qualify it other than to say that X comes up more aggressively, it's more responsive to user input, and I haven't had any nasty spills with test3 so far.

      It's probably not the smartest thing I've done running on a test kernel for my work notebook, but the added functionality including support for all of the hardware on a Sager 4760 (save the build in vid camera) makes me all warm and fuzzy inside.

      It's just a shame that the Cisco VPN client isn't out for 2.6 (that I'm aware of - please correct me if I'm wrong). Otherwise I would be all set.

      --
      You are checking your backups, aren't you?
  3. What's the point by Anonymous Coward · · Score: 0, Interesting
    And what's the point?

    I can't install a vanilla source distributed Linux kernel on my RedHat anymore because the boot process requires that specific modules and kernel services are available. To ensure that I need to download a RedHat version of the kernel.

  4. Are we ever going to get hibernate? by Malc · · Score: 4, Interesting

    Summary of changes from v2.4.22-rc2 to v2.4.22-rc3

    @lt;len.brown:intel.com>:
    o ACPI update
    o ACPI build fix
    o linux-acpi-2.4.22.patch

    1. Re:Are we ever going to get hibernate? by zenyu · · Score: 4, Interesting

      The problem from what I understand is there are so many iterations of it that the devs simply can't get the kernel to work with all of them. I can't fault them for that, but ACPI is as common as TCPIP now and this is one area where Linux has fallen way behind the curve. Having only some hardware work and only certain functions available just isn't good enough.

      As somebody who wrote an early user space cpu frequency scaling deamon, I think it's not just all the broken ACPI tables but also the fact that the intel engineers change the interface at ever minor iteration. Even with all the changes they still have very strange convetions such as numbering CPU's starting at 1. If you just want to last longer on batteries you can use the cpufreq deamon, it's a strange beast in 2.4, but makes some sense in 2.6 with the /sys filesystem interface. There is an effort to backport the /sys interface if like me your laptop doesn't like 2.6. When ACPI doesn't work there are also CPU only modules that don't rely on ACPI. The only hitch is that your laptop may not be enabled if no one has posted to the list that it works for them, cuz different motherboards require different voltages and may need longer times between frequency changes than the CPU does (memory bus lines, support chips, and power supply all effect these params.) This is one of the things ACPI is supposed to tell you but most BIOSes are very broken, which is why ACPI doesn't always work when it should.

      Once it's working though your Linux laptop can be both more responsive and last longer on batteries than Windows. Windows is very conservative in interpreting ACPI tables and also doesn't have a very flexible set of frequency scaling algorithms.

  5. New Kernel doesn't matter here. by deathcow · · Score: 4, Interesting

    Mod me off-topic if you wish, I for one welcome our new off-topic overlords.

    We don't need a new kernel for now. The existing kernel has been highly stable in the types of jobs we throw at it. But that's not why it doesn't matter. Our director has finally handed down the "No more Linux installs" message. Here is the message:

    As some of you may know. SCO has a lawsuit filed against IBM for breach of contract, misappropriation of trade secrets, etc. The bottom line of this disagreement is that SCO is accusing IBM of including some of SCO's trade secrets (proprietary code) in its Linux kernel and that this source code has also been included in the Linux kernel available off the Internet.

    To top all of this off, the US Copyright Office awarded SCO a copyright to System IV Unix last month. SCO, through a series of mergers and acquisitions, acquired the patents, copyrights, etc to System IV Unix that was originally developed by Bell Labs in 1969. About this same time, SCO created a new division whose only purpose in life was to license Linux and Unix to end users.

    Because of all of this confusion, I have asked our legal eagles to give me an opinion as to whether or not our recent installation of Linux systems in the (our department name) places (our company name) in any sort of jeopardy. Frankly, I think that SCO is simply struggling for survival as it is in severe financial trouble. I also think that any rights SCO may have think it had at one time has been abrogated since SCO was (until last month) freely distributing Linux on its web site under the GNU General Public License.

    But until I receive an opinion from Legal, we will not deploy any more Linux systems.

    p.s. We are a 1200 employee telecommunications company, ISP, cable TV, long distance telephone, etc.

    1. Re:New Kernel doesn't matter here. by Lumpy · · Score: 4, Interesting

      so you are one of the small fish.

      I'm with a 2.2 million employee Telecommunications/ CABLE TV etc.. company and we are accelerating the deployment and installation of linux in the enterprise.

      the last 2 weeks have put the Top of the company in a position that they no longer can trust microsoft systems for the backbone of the company. Linux is being given the green light all over the place.

      SCO's lawsuit has zero effect on companies that have sharp upper management and employees. in fact the District vice president of sales mentioned to us in a meeting that the SCO lawsuit is to be considered a joke and ignored in our deployment research and plans.

      Sorry that your company is ran by people that do not get a complete picture before making business decisions, and best of luck to you.

      --
      Do not look at laser with remaining good eye.
  6. How will this change my life? by 2toise · · Score: 2, Interesting

    I run Debian, what practical difference will this make if I install it? I use the system for office type tasks.
    Thanks

  7. Slowed to a Crawl by Anonymous Coward · · Score: 2, Interesting

    Well, it's happened again. kernel.org just slowed to a crawl.

    Remember to use those mirrors, folks!

  8. For that desktop box, try the 2.6.0 kernels... by sonicattack · · Score: 4, Interesting

    I've been running the -mm series of the 2.6.0-testX kernel since they started coming out, and I'd recommend anyone not running a critical system (where stability is of high importance) to try them out. Some of the main improvements easily noticeable are in the scheduling / Disk I/O areas, which makes interactive (desktop) usage much smoother. For instance, problems I've had with the 2.4 series where a system under high disk usage made the desktop jerky, and sometimes slowed down to a crawl, I have yet to see with these new kernels.

    Sure, I've stumbled upon an occasional bug in the process, and once after upgrading found the kernel unbootable, but with help from the kernel newsgroups, a quick fix could always be found. Besides that, I've found these kernels to be as stable as the 2.4-series..

    And, don't forget, the more people running the soon-to-be-stable kernel, reporting bugs as they are found, the ... sooner it'll be stable! :)

    1. Re:For that desktop box, try the 2.6.0 kernels... by nvrrobx · · Score: 3, Interesting

      I have to agree with you fully on this one.

      I'm running 2.6.0-test3-mm on my dual PIII 600 box. It's not a fast machine, but with 2.6.0 it feels downright snappy. (The machine is a Dell Precision 210M Workstation - 384 megs of PC100 RAM, a 5400 RPM hard drive, GeForce4 Ti 4200 card)

      The system as a whole feels a lot faster than it did with 2.4.x.. Of course, I have no real perf numbers to throw at you, but I'm really impressed with 2.6. VMware and WineX has some issues with 2.6 (CD-ROM access doesn't like to work correctly)

      Also, the new kernel build process is much more streamlined, and building a kernel doesn't seem to take as long. The output is even prettier. :)

  9. Crypto API by njchick · · Score: 5, Interesting

    For the first time we have cryptographic API in a stable released version of the Linux kernel. Until recently, cryptographic software could not be exported from the U.S. without a special permission. It took some time to explain to the government that the "bad guys" already have access to strong encryption. We have succeeded. Cryptographers, rejoice!

    1. Re:Crypto API by Vantage13 · · Score: 2, Interesting

      yeah, but is the cryptoloop module included or is that still a separate patch?

  10. Cmedia Bug Fix by ciroknight · · Score: 3, Interesting

    Anyone know if the new kernel has been fixed for the C-Media 9738 AC97 Codec? Last time I checked Alan Cox "patched" this in 2.4.21 and my integrated sound didn't work. I know this has gotta be a pretty big problem.. my board was pretty popular (K7S6A)...

    --
    "Victory means exit strategy, and it's important for the President to explain to us what the exit strategy is." G.W.Bush
  11. For those of us unenlightened by Matey-O · · Score: 3, Interesting

    Okay, we've got a new kernel. That's cool. I've built kernels before in the past...in the past.

    Now with RH9, every time I try to make a custom kernal it pooches the system mightily. I'll lose video, or I won't get power, or it'll give LOTS of errors when starting up because I didn't have a Makefile just like Redhat's magic one.

    So how would a person compile 2.4.22 and still have the same support as whatever RH9 based build they're currently running?

    --
    "Draco dormiens nunquam titillandus."
  12. 2.6.0-test4 out too by evilned · · Score: 2, Interesting

    Just saw that 2.6.0-test4 is out. Been using test-3 with a patch akmp patch for the APIC on the nforce2, and have loved the results. X is more responsive, sound is much less likely to skip when the disk is in use. Havent found out if they included that patch in test-4, so if anyone knows, please reply.

    --

    "My head hurts, My feet stink, and I dont love Jesus." -Jimmy Buffett

  13. Hmm, Here's a better reason. by Bruha · · Score: 2, Interesting

    You show me one average user that can update their kernel.

    Can it pass the mom test? No, then point made.

    Kernel updates should be seamless and not break anything that you had compiled for the kernel..
    EG Nvidia drivers. I'd love to have my parents use Linux but if every time the updates are needed (although rare) I'd have to either ssh into the box or go there physically. Trust me you cant tell them how to do it over the phone (hard of hearing)

    Maybe that's the whole problem. Upgrades in windows have rarely caused problems. But on the other hand a simple update to Redhat 8 on a production server completely broke it requiring a backup to a older version. The system in question was a webserver and some security update rendered RPM unusable and some subset of RPM has to work for apache to work properly so we were screwed.

    I think it's time to get the devs of RPM, apt and all other major packaging programs and make them form a standard and stick with it. And whatever standard should work to where 1 package will work for all systems instead of such things like RPM where you have a rpm for every flavor and version of processor and distribution. EG Redhat RPM's and Mandrake RPM's.

    As processor speeds and faster subsystems make code compling a short trip instead of the long times it can take for bigger projects maybe source packages that compile on the target system into a correct rpm for that system/distro and installs itself. But for it to be tolerable you're looking at least 6ghz with solid state HD/PCI-X/XDDR-RAM wihtout any bottlenecks.

    Anyways I think I'm rambling now :) but many see the same problem and that's why Linux will work in corporations fine with correctly trained IT staff but not for the average home users.