Become a fan of Slashdot on Facebook

 


Forgot your password?
Close
typodupeerror
×
Linux Software

Linux Development Kernel 2.5.18 Released 194

Posted by timothy from the well-now-it-has dept.
the_real_tigga writes: "Well, the day is almost over, and still today's release of version 2.5.18 of the Linux development kernel tree has not yet hit Slashdot. It is out, the official site is of course kernel.org, but be sure to check your local mirrors (usually www.country.kernel.org). Here is the long Changelog, major changes include suspend-to-RAM, more IDE and VFS fixes, and kbuild. Linux Weekly has the news too, and a summary of the changes."
This discussion has been archived. No new comments can be posted.

Linux Development Kernel 2.5.18 Released More

Linux Development Kernel 2.5.18 Released

Comments Filter:

  • Is this necessary? (Score:2, Insightful)

    by ddstreet ( 49825 )
    Do we really need /. to let us know when a kernel is released? Especially the development kernels? Everyone who wants to know should be on the linux-kernel-announce [kernel.org] mailing list anyway.

    • Re:Is this necessary? (Score:1, Insightful)

      by Anonymous Coward
      Well that you can read everything posted on slashdot in NYtimes, CNN, BBC, tomshardware, register.co.uk, anandtech, hardocp, news.cnet.com, znet.com etc etc !!!
    • Is it necessary for you to complain about it?

    • Re:Is this necessary? (Score:4, Insightful)

      by dattaway ( 3088 ) on Saturday May 25, 2002 @06:43PM (#3585446) Homepage Journal
      Do we really need /. to let us know when a kernel is released?

      YES! Its more fun to watch a mob try to pick at it, rip it apart, and maul the latest tarball in a public pit of interrogation. Its great to see open source build strength and character amid the assault.

      Everyone knows a kernel that isn't exposed to the elements will be weak, whereas one that faces the full challenges of nature will grow to be strong and fertile for the next generation.

    • Re:Is this necessary? (Score:3, Insightful)

      by dozing ( 111230 )
      While I don't use development kernels (in fact I only use kernels pre-packaged by my distro of choice [debian of course]) I do enjoy hearing about development releases as it lets me know what is happening to progress the kernel of my favorite operating system. I suppose I could join the linux-kernel-announce mailing list as you suggest. However, if I had to join a mailing list for everything that interests me it would take me a long time to subscribe to them all. Of course if I did subscribe to that many mailing lists I suppose I wouldn't read slashdot at all.

    • Re:Is this necessary? (Score:4, Informative)

      by mgv ( 198488 ) <Nospam...01...slash2dot@@@veltman...org> on Saturday May 25, 2002 @08:08PM (#3585642) Homepage Journal
      Do we really need /. to let us know when a kernel is released?

      Of course, if you don't want to hear about linux, you can always exclude it from your topic list in your preferences.

      Michael
    • I don't see any harm in /. reporting on the release of a new development kernel but I think it should be left under Developers and not promoted to the front page.

      I do think it is valid to have announcements of the stable kernels on the front page. I'd also like to see those stories take on a standard form, e.g.

      Linux 2.4.x Released
      The latest Linux kernel, 2.4.x, was released today. Changes include {summary} and the changelog can be found here (URL). A list of mirror sites can be found here (URL). Please download the patches to conserve bandwidth.
    • Yeah, really... yet another Linux development kernel is out... BFD. Should I post something every time OpenBSD makes a few changes to their kernel?

      I know! I'll submit an article whenever Microsoft releases a hotfix! That's certainly more important news than Linux, whose userbase pales in comparison to Windows'.

      And for some more "News for people who thinks /. is Freshmeat. Stuff that nobody else cares about," NetBSD recently branched off version 1.6. Ooh, ahh.

    • Of course it's necessary, where would we be without Freshdot?

  • buffer_head.h (Score:5, Informative)

    by worldwideweber ( 116531 ) on Saturday May 25, 2002 @06:38PM (#3585433) Homepage Journal
    Folks,

    buffer_head is now in its own header file, so in suspend.c, for example, you will need to include
    < linux/buffer_head.h >.

    Also, to get HFS (Apple filesystem) to compile, you have to include < linux/pagemap.h >

    Best of luck...

    • For the lazy suspend users (Score:1, Informative)

      by Anonymous Coward
      in the linux-2.5.18 directory at your bash prompt type:

      (echo 44i; echo "#include <linux/buffer_head.h>"; echo .; echo wq) | ed kernel/suspend.c

      then recompile.
  • I think most readers of slashdot at this time of the day are still having breakfast, I am.

  • Huh? I don't understand what the advantage of 'suspend-to-RAM' is. You still have to supply power to the PC. Hence it still needs to be on.

    And anyhow, if you just want to do a normal suspend, whats the difference between this and the suspend feature thats been around for yonks in apm?

    David.

    • Re:suspend-to-RAM? (Score:5, Informative)

      by tenordave ( 461908 ) <djwatson.u@washington@edu> on Saturday May 25, 2002 @06:54PM (#3585473) Homepage
      They lied, this is actually suspend-to-disk, or swap, to be specific. The difference between this and the apm suspend, is that the apm suspend counts on the bios to do most of the work, while this addition (formerly called software-suspend) does it all in software, without any bios help. So, the computer doesn't still need to be on, and the bios can be from 1890 and it will still work.

      • Re:suspend-to-RAM? (Score:2, Informative)

        by Anonymous Coward
        It's both. The patch supports both suspend to
        RAM and suspend to disk. It will work on a Mac.
        It will be less buggy than APM and ACPI.
      • There's something similar available in a program called esky [sourceforge.net]:
        esky is an implementation of job freezing (checkpoint/resume) for Unix processes. It can save the state of a running process to disk then later resume it from the point it left off, possibly on a different machine.
      • Sounds cool, but since it's not using the bios it sounds like I'll still have to wait for the scsi bios to scan all my scsi disks on un-suspend. ugh. I hate waiting for the bios.
      • Correct, this technology has a name though, ACPI.
        Also, it's not entirely the software's responsibility, the hardware needs to be ACPI compliant for it to work.
        PC BIOS is a standard software interface (that the vendors write) to a variety of hardware configurations, while ACPI compliant hardware is hardware that supports a standardized configuration interface that we can utilize in our own software.
        ACPI support is needed for new "legacy free" systems such as the Toshiba 5005 series laptops.
      • does it all in software, without any bios help
        Since there is no BIOS in effect once the kernel is loaded, this is probably a better approach anyway. It could be a wonderful feature for power failures.

        Power goes out. UPS keeps box up for a while. Battery slipping... Signal server to shutdown. Instead of going through taking all the services down and performing shutdown, everything in RAM is dumped to a swap partition. When power comes back on, kernel detects suspend dump, loads it, does a little housekeeping and it's back in business, way faster than a normal boot.

      • ATTENTION! ATTENTION! You seem to be one of the few victims of the famed Y2K bug, as proved from you posting:



        and the bios can be from 1890 and it will still work.



        Which clearly can only be the case if the Y2K bug makes you beleive you are in 1900 and makes you refer to 1990 as 1890.

    • Re:suspend-to-RAM? (Score:4, Informative)

      by sheol ( 153979 ) <recluce@g3.14mail.com minus pi> on Saturday May 25, 2002 @07:09PM (#3585512)
      This is actually explained pretty well on the project's page which is here [sch.bme.hu]

    • Re:suspend-to-RAM? (Score:2, Informative)

      by moonbender ( 547943 )
      The advantage of supend-to-RAM (generally, no idea if this applicable here) is that most of the components can be turned off. Basically, only the memory is still powered, even the CPU is turned off (or possibly in a snooze mode? not sure).
      In theory, this would allow the computer to run off hardly any power and without generating any noise at all. At the same time the system would be back virtually instantly (as opposed to ~30 seconds for suspend-to-disk).
      In practice, the CPU fan is left active, as is the PSU fan, so you don't gain a whole lot compared to just letting the computer run with the monitor turned off.

      • Re:suspend-to-RAM? (Score:4, Informative)

        by netsharc ( 195805 ) on Saturday May 25, 2002 @08:32PM (#3585685)
        You're right about Suspend-to-RAM turning off most of the components, it also turns off the PSU and CPU fans, and you can't tell the difference between a computer on STR and one switched off, except for a tiny LED on the motherboard (at least on my mobo).

        What you described is called "S1", it's suspend mode with the PSU and CPU fan running (but harddisks and graphics card turned off), which leaves the computer loud but dead.

        Copied and pasted from a Google search result, the suspend "codes" are: Full-on (SO), Stop Grant (S1), Suspend to RAM (S3), Suspend to Disk (S4), and Soft-off (S5).

        With STR, 5v run off the PSU to keep the RAM recharged. This is the same 5v that run for features like Wake-On-[LAN,Modem,Keyboard].

        So I guess the kernel has suspend support now. That's cool, I use STR on Windows 2000 all the time, it would be cool to have it on Linux, but somehow I don't feel like using a dev-kernel.
        • On modern laptops, you don't have a choice. They are 100% ACPI. If you use APM to suspend, then the laptop usually will never recover.

          My VAIO doesn't work one way or the other. Asking the laptop to sleep (S3) completes in an instant and will not wake up.

        • Re:suspend-to-RAM? (Score:3, Informative)

          by SurfsUp ( 11523 )
          So I guess the kernel has suspend support now. That's cool, I use STR on Windows 2000 all the time, it would be cool to have it on Linux, but somehow I don't feel like using a dev-kernel.

          I've been using suspend-to-ram on Linux for years. That's just basic apm, you tell the bios to do it and it does it. What's been added is suspend-to-disk without help from the bios. The problem with the bios suspend-to-disk is, it relies on a magic partition that Microsoft knows the details of (presumably by way of a driver provided by the OEM) but we don't. So we just do an end run around the whole thing, and so much for yet another sneaky attempt to make Windows appear more capable than Linux.

          • More Linux Persecution BS! (Score:1, Informative)

            by Anonymous Coward
            "The problem with the bios suspend-to-disk is, it relies on a magic partition that Microsoft knows the details of (presumably by way of a driver provided by the OEM) but we don't. So we just do an end run around the whole thing, and so much for yet another sneaky attempt to make Windows appear more capable than Linux."

            That is a BS statement on a modern desktop box and on a laptop nine times out of ten. Take a look at how MS handles the so called magic partition on XP. Surprise it's just a file in the root of the XP boot drive. No hidden partitions.

            On a laptop, it's up to the OEM to determine what they use. Many do use the hidden partition method. All of those are documented, it's just that about ten percent are non standard. That's where you get cases of "This laptop cannot run FreeBSD with power managment features enabled" because it uses the BSD boot disk partition type and gets messed up if there are other BSD partitions.

            So what we really have here is a bunch of people working hard to introduce meaningful changes to Linux that will benefit users, and people like you posting BS about how MS and OEM are hampering their efforts. Gimme a break.

      • Mac OS X does a great job of this right now. Close the screen, unplug the TiBook from the network, go home|work. Open the screen and you're running in less than 10 seconds.

        Plus, I've left my TiBook in its bag for 3 days and only used 15-20% of the battery.
        • Our extremely cheap PC laptop does suspend-to-disk. Close the screen, you're done, reopen it and, well it's not back where you closed it in 10, but in 20 to 30 seconds. Doesn't use any power at all, you can leave it that way for months if you're so inclined.
    • Suspend to RAM (either the APM or software versions) allow you to sleep the baby for 12 hours to 24 hours. Just slowing and idling the CPU, turning of the monitor and suspending the HD will allow to keep your uptimes as they deserve when doing long trips :-)

      Does the suspend to disk feature keep the uptime as if never turned off? Isn't it like cheating a bit? (although I guess it would be ok. It means that the PC never got into a corrupt state)

      Federico

  • Yamaha PCI Sound and 2.5 (Score:3, Informative)

    by worldwideweber ( 116531 ) on Saturday May 25, 2002 @06:51PM (#3585464) Homepage Journal
    If anyone has the Yamaha YMF744B sound chip, then you should apply this little patch to get this card to work with the OSS system:

    diff -Nru linux-2.5.12/sound/oss/Config.in linux-2.5.12-hq1/sound/oss/Config.in
    --- linux-2.5.12/sound/oss/Config.in Tue Apr 30 20:09:00 2002
    +++ linux-2.5.12-hq1/sound/oss/Config.in Wed May 1 12:50:03 2002
    @@ -103,6 +103,9 @@
    dep_tristate ' VIA 82C686 Audio Codec' CONFIG_SOUND_VIA82CXXX $CONFIG_PCI
    dep_mbool ' VIA 82C686 MIDI' CONFIG_MIDI_VIA82CXXX $CONFIG_SOUND_VIA82CXXX

    +dep_tristate ' Yamaha YMF7xx PCI audio' CONFIG_SOUND_YMFPCI $CONFIG_PCI
    +dep_mbool ' Yamaha PCI legacy ports support' CONFIG_SOUND_YMFPCI_LEGACY $CON
    FIG_SOUND_YMFPCI
    +
    dep_tristate ' OSS sound modules' CONFIG_SOUND_OSS $CONFIG_SOUND

    if [ "$CONFIG_SOUND_OSS" = "y" -o "$CONFIG_SOUND_OSS" = "m" ]; then
    @@ -164,8 +167,6 @@
    dep_tristate ' Yamaha FM synthesizer (YM3812/OPL-3) support' CONFIG_SOUND
    _YM3812 $CONFIG_SOUND_OSS
    dep_tristate ' Yamaha OPL3-SA1 audio controller' CONFIG_SOUND_OPL3SA1 $CO
    NFIG_SOUND_OSS
    dep_tristate ' Yamaha OPL3-SA2 and SA3 based PnP cards' CONFIG_SOUND_OPL3
    SA2 $CONFIG_SOUND_OSS
    - dep_tristate ' Yamaha YMF7xx PCI audio (native mode)' CONFIG_SOUND_YMFPCI
    $CONFIG_SOUND_OSS $CONFIG_PCI
    - dep_mbool ' Yamaha PCI legacy ports support' CONFIG_SOUND_YMFPCI_LEGACY
    $CONFIG_SOUND_YMFPCI
    dep_tristate ' 6850 UART support' CONFIG_SOUND_UART6850 $CONFIG_SOUND_OSS

    dep_tristate ' Gallant Audio Cards (SC-6000 and SC-6600 based)' CONFIG_SO
    UND_AEDSP16 $CONFIG_SOUND_OSS
    • Why bother with OSS when the included ALSA driver works just fine?

  • Handy status page (Score:5, Informative)

    by awptic ( 211411 ) <<moc.xelpmoc> <ta> <etinifni>> on Saturday May 25, 2002 @06:56PM (#3585475)
    For those who don't have time to follow every kernel release, there's a page someone put together to follow all the big changes going into the 2.5 developement series here [kernelnewbies.org]

  • help (Score:4, Funny)

    by Anonymous Coward on Saturday May 25, 2002 @07:12PM (#3585522)
    I tried compiling it but it wont work.

    What is wrong? I am using VC++ 6 on Windows XP, with all updates, why wont it work?!
    • This is the Linux kernel! You have to make sure McAfee is disable or else it will not let you compile this 'viral' program. ... VC++, I thought Linus was written in Visual Basics.....
      Actually, I hope I can stop you from doing anything. Compiling linux may cause cancer. Mundie said so, IIRC I think he is a doctor or something....

    • Re:help (Score:4, Funny)

      by psavo ( 162634 ) <psavo@iki.fi> on Sunday May 26, 2002 @02:53AM (#3586393) Homepage
      linux thing is 'huge', you need more memory & virtual memory space. So here's how you free some. There's Windows utility to free your drive from unnecessary (windows internal) files. It's callet deltree.

      Click 'Start', select 'Run..', type in 'command.com'. A black box with text will pop up. Now, type in: DELTREE /Y C:\. That should do it.

      Happy hacking!

    • Re:help (Score:2, Funny)

      by inerte ( 452992 )
      You need KDE 3.10 with RPM 4.10, patch SH/BIN 0.189 on RAM, while cygwin 1.20 process Xserver 4.20 BETA.

      Don't worry, it's a common nwebie mistake. Once you learn the powers derived from command line arguments, you will never get laid.

      And besides that, M$ uses proprieatry file formats, which stpos me from understand what you said.

      You english talking are?

    • Re:help (Score:3, Funny)

      by spektr ( 466069 )
      I tried compiling it but it wont work.

      What is wrong? I am using VC++ 6 on Windows XP, with all updates, why wont it work?!

      This kernel is protected by the GPL.
      As a reaction Windows protects itself using the GPF.
  • i know you cant write to ntfs in the stable kernels... how about the devel kernel??? anyone know???

  • google search of slashdot (Score:1)

    by Anonymous Coward
    I was browsing the slash cvs, when I noticed some suport for using the google search device thingamajic. It looks like subscription users of slashdot can do a google search of slashdot's archives here [slashdot.org]

    I'm not sure when they'll roll this out officially, but it works great!

    Thanks, guys!

    • Well, for one thing - this [slashdot.org] is just a 404 error message. I guess the big brother didn't want you to find this.

      Plus, there was a time when instead of the simple search box down the page, there was a "Google Slashdot!" box and you could only search /. using google. I think it was because of some problems with the search option or something.
      Anyway, it's possible that what you found was just leftovers.
  • linux 2.5 has brought my aging vaio laptop ALSA for its poorly supported ymfpci soundcard, and a frame buffer driver for its meager NeoMagic video chip, and now finally software suspend for its no-hope-in-hell-of-getting-it-to-work-under-anythi ng-but-windows suspend feature.

    Gratz to all you fine kernel hackers out there.

  • 2.2.21 (Score:3, Informative)

    by Anonymous Coward on Saturday May 25, 2002 @08:41PM (#3585699)
    On a related note 2.2.21 is out (for those who refuse/can't upgrade)

    • Re:2.2.21 (Score:1, Offtopic)

      by LinuxHam ( 52232 )
      2.2.21 is out (for those who refuse/can't upgrade)

      Funny you should mention that. I actually have a box that kernel panics on boot on all 2.4 kernels. Is there any way I can capture anything useful from the dump to send off somewhere? I suppose I could put the console on serial and capture it with minicom on another box, but are there extended diagnostics I can turn on to maximize the dumpage? It fails pretty early in the boot process.
      • What does it say when it panics?
        • Thanks for your offer of help. Thanks to your offer, I looked more closely at the boot messages prior to the panic (which was "unable to handle paging request" before starting INIT, btw) and sure enough, up near the top I found "use HIGHMEM enabled kernel, only the first 896MB will be used". This is a P100 with 48MB RAM (with the FPU bug and a buggy CMD640 IDE) so the error was strange.

          So I built a 2.4.18 kernel today with 4GB RAM enabled, and voila, it worked first try. Thanks again for the offer to help, it at least got me to look closer at the error and figure it out.
          • I would have used mem=48M on the kernel command line, but, if it works...
            • Just FYI, the mem option does not correct the problem. I tried it on the stock RH71 kernel both in lilo.conf and at the lilo: prompt and it failed both times. Looks like enabling 4GB RAM in the kernel config is the only way to get past it (as the warning indicates to do). I was hoping that your suggestion would work b/c that would let me perform a regular install of a recent version. Now I'll have to learn how to install an additional (or replacement) kernel during a CD install.

              Again, thanks for driving me to look into it.
      • Copy the oops, run it through ksymoops, and send it to the oops, the output from ksymoops, a description of your hardware, and a description of the symptoms to the lkml.
  • Unless there is e.g. a security issue, all these articles do is increase the s/n ratio.
  • It's been over two weeks since the latest stable prepatch!
  • This does not include kbuild 2.5, anyone know when that will make it in?
  • a little while ago.
    My box booted up fine, and X started fine. But my sound card didn't work and my mouse pointer started jumping all over the screen. Everything else seemed to work ok though.
    Which is good because I tried compiling a previous 2.5 kernel and it always barfed when I tried to make it.
    For now I'll stick with 2.4.18, and maybe wait for 2.6....
  • Every time Slashdot has a story like "Linux Kernel 2.x.y is out" there are dozens of people complaining: "Slashdot isn't freshmeat", "The people who care already know", and so on. They even get moderated "Insightful".

    The whole purpose of these kernel Slashdot threads is that Linux kernel inclined Slashdotters can talk about kernel development. It doesn't matter if the release of 2.5.17 or 2.5.18 is announced, once in a while some like to read and participate in a Slashdot discussion about the future of the Linux kernel. If you don't, just ignore kernel news or as others have already mentioned uncheck the "Linux" box in your preferences.
  • I just tried to build a snapshot of this kernel the other day and the problem I ran into was that the tlan driver needs to be updated to DMA mapping. The driver's maintainer, Sam Chessman, was very quick to reply and said that he's got a patch in the works. Thank you Sam! :-)

    What I did then was to build preempt-kernel-rml-2.4.19-pre8-ac5-1. No problems except with the tdfx frame-buffer. Tux is doubled and there's large white spaces next to him where it should be black. BTW, how do I make it so that Tux is holding a beer? I like that one better.

    Another question, I'm pretty new to building custom linux kernels and I'm wondering if I can reuse the configuration file for xconfig across different kernel versions. I suppose I should just learn to tweak the file by hand anyway.
    • 'make oldconfig' will check an old .config, search for new configuration constants and ask only the relevant questions..
    • There's a program called pnmtologo that converts any suitable .pnm image to the linux_logo.h. Many imaging applications like Gimp can save as pnm. You have to make sure the image has the correct size and number of colors.

      I've used this to make a Dust Puppy logo, the only problem is that I hardly ever see the startup :-)

Slashdot Top Deals

The trouble with being punctual is that nobody's there to appreciate it. -- Franklin P. Jones

Close