Catch up on stories from the past week (and beyond) at the Slashdot story archive

 


Forgot your password?
Close
typodupeerror
×
Linux Software

Linux 2.4.8 is Out 212

Posted by michael from the so-much-for-my-2.4.7-installation dept.
Justin was the first to note that kernel 2.4.8 is out. Check the changelog, or just hit kernel.org, you know the drill.
This discussion has been archived. No new comments can be posted.

Linux 2.4.8 is Out More

Linux 2.4.8 is Out

Comments Filter:

  • EMU10K1 fixes, still no ice1712/envy24 :-( (Score:4, Interesting)

    by fishbowl ( 7759 ) on Saturday August 11, 2001 @03:02AM (#2111195)
    I have a bone to pick with the M-Audio company.
    They go as far as to put a penguin on the box claiming Linux support. But the linux support they claim to have amounts to two choices: an
    ALSA driver that does not work, or the 4Front
    OSS driver. The price for the 4Front driver
    is quite shocking to me, on the order of $65.00.
    Where are the OSS/Free drivers for Delta cards
    (Audiophile 2496, etc.)? Had I known I'd have to
    pay for a closed source driver, I would have kept
    shopping for a pro sound card. If the 4Front
    driver wasn't closed source, I'd pay their price.
    Just can't win.

    • : an ALSA driver that does not work...

      Sorry? The ALSA driver has worked flawlessly with my SB Live! Platinum ever since I got it almost 6 months ago. Ok, the mixer settings for the Live! Drive II panel aren't all that obvious, but I've hardly used that panel anyway. What problems did you have? I'd be willing to help you out. Email me.

      • well thtas all well and good, but he is talking about delta cards, not hte sblive, there are multiple free drivers for sblive, but the poster said there are no free drivers for the delta cards that work(though a comment says the alsa drivers works now). Read the comment(not just the title), then reply.

    • Re:EMU10K1 fixes, still no ice1712/envy24 :-( (Score:1, Informative)

      by Anonymous Coward
      I just received my DIO2496 card, and it works well with the latest alsa 0.9.0beta6, and envy24control from cvs tree. I am listening to music stored on my HD (raw pcm, no mp3 thanks) on my stereo system, via the optical output of the card.
      • > I am listening to music ... via the optical >output of the card.

        Does the analog output work?
        Does it record from the analog input?
        Can you set the patchbay routing params?
        Can you toggle SCMS?
        What sampling bitrates and frequencies
        can you use?

        Like I said elsewhere in the thread, the ALSA
        driver appears to work, but no output, and the
        OSS emulation layer appears to be quite broken.
        I do not have marginal hardware, have tried many
        versions of the Alsa driver as well as kernel
        versions 2.2.19 and 2.4.[567], on both intel and
        amd based machines with generally the same results. I think it's documented that the OSS
        emulation support isn't ready for any multitrack
        cards, so that would rule out ALSA from a certain
        application perspective anyway.

        I guess I'm stuck with buying the 4Front driver,
        which wouldn't bother me that much if Midiman didn't imply that such a driver was included with
        the card. At first, I loved them, but now I think they are complete bastards.

        I don't imagine there's an OSS/Free module for
        envy24 coming anytime soon. It sucks because there are so damned few options for 24bit/96khz
        audio recording in linux. Midiman has filled the
        niche in the hardware department, the software situation is pretty good, but if the damned driver costs half again as much as the card it's not a good deal. Nothing against 4Front, who will probably get my money anyway, but
        why do I seem to be the only person who's upset here? We need an OSS/Free module for Delta cards.

  • Here we go again....

    patch, make oldconfig, make bzlilo, make modules_install, depmod -a 2.4.8, cross fingers, reboot....

    Damn. I forgot about that stupid third party kernel patch for my Tekram/Creative Labs/Nvidia/ hardware.

    Looks like I'll have to wait another month for those NDA morons to catch up again.

    • make modules_install does depmod for you now. Doesn't anyone pay attention to what the build actually does anymore? Sheesh.

    • Re:Hurry up and wait (Score:3, Informative)

      by tzanger ( 1575 )

      Damn. I forgot about that stupid third party kernel patch for my Tekram/Creative Labs/Nvidia/ hardware.

      Not sure what you're smoking but most kernel updates don't break third-party drivers unless the source API has changed -- something which doesn't occur often in stable releases.

      My nVidia kernel module (1241) has compiled without issue on every 2.4.x release to date. The VMWare kernel modules have worked up to 2.4.7, where they did manage to change the source API and ended up breaking compatibility. However, the VMWare mailing lists provided an answer (patch) the same day that the new kernel was released.

      Perhaps you should start using the more expensive crack, since the cheap stuff seems to be affecting your thinking.

  • The Makefile for the emu10k1 drivers erroneously includes 'joystick.o' in the emu10k1.o module. That .o file should only be included in 2.2 kernels in the first place, and in those, it should be stuck into a seperate module, emu10k1-joy.o

  • Times. (Score:2, Interesting)

    by friscolr ( 124774 )
    Check the changelog, or just hit kernel.org, you know the drill.

    [frisco@eff /music]$ uptime
    8:52am up 142 days, 3:34, 2 users, load average: 3.10, 3.26, 3.05
    [frisco@eff /music]$

    no changing here. i'm waiting for a year uptime, then maybe i'll think about upgrading. i wonder what i did 142 days ago that required a reboot?

    • no changing here. i'm waiting for a year uptime, then maybe i'll think about upgrading. i wonder what i did 142 days ago that required a reboot?

      2.4.0?

  • You know what would be really useful? (Score:4, Interesting)

    by Mustang Matt ( 133426 ) on Saturday August 11, 2001 @03:15AM (#2115798)
    Included in the news clip it would be really useful if there was a link for newbies that explained how to compile. I'm not just talking about pointing to the how-to's, it needs to either be a direct link or a step by step guide.

    If linux ever is to become main stream on the PC for average joe user, we really need to step up the help for newbies. Education is the solution to ignorance.
    • Check out #kernelnewbies on Open Projects Network IRC and http://www.kernelnewbies.org [kernelnewbies.org].

      Yes this would be much better in the original article.

    • Included in the news clip it would be really useful if there was a link for newbies that explained how to compile.

      Recompiling the kernel is not exactly for newbies. Let them get comfortable with Linux first... and this may take a few months or even years for some. You have to have some idea what all those configuration switches do and why you might want to change some; otherwise you're more likely to break stuff than you are to optimize the system. Once they are comfortable enough to read a HOWTO, then the link you want will not be needed.

      Anyone who doesn't know about the HOWTOs, or refuses to read them, definately does not need to be recompiling the kernel.
      • I agree. To customize the kernel for a system, one should be prepared to read a shitload of help screens during xconfig. There are alot of options in there and thay are hardware-specific so many need to be read, at least the first few times someone learns to configure the kernel.

        That said, I think compiling the kernel is a fantastic learning experience and everyone should do it; just maybe not until they are already used to having to dig through documentation, use Google, etc.

  • Just tried a build with a configuration that worke d with 2.4.7. The build of the emu10k module failed. This is the first time a kernel build failed for me in a LONG time. Is it just me?

    • Re:Build Fails (Score:3, Informative)

      by kilrogg ( 119108 )
      The patch was rushed in at the last minute:

      From the kernel mailing list:

      "hm, edit drivers/sound/emu10k1/Makefile and remove the object reference for `joystick.o' -- i dont compile as a module so i missed this, thanks."

    • Re:Fix available (Score:2, Informative)

      by Sits ( 117492 )
      As mentioned in another post there is a fix available for this. Take a look at this post emu10k1 mailing list [creative.com].
    • Just tried a build with a configuration that worke d with 2.4.7. The build of the emu10k module failed. This is the first time a kernel build failed for me in a LONG time. Is it just me?

      Just apply Alan Cox's 2.4.8ac1 patch, that'll fix it.

      ftp://ftp.kernel.org/pub/linux/kernel/people/alan/ 2.4/patch-2.4.8-ac1.bz2

    • Re:Build Fails (Score:3, Informative)

      by Anonymous Coward
      Nope... Its not just you... Apparently the update to the emu10k broke building it as a module. I just build it into the kernel itself (not as a module) and it seems to work fine.
      • Thanks so much for the info. I just built it into the kernel and it seems OK!
      • Apparently the update to the emu10k broke building it as a module. I just build it into the kernel itself (not as a module) and it seems to work fine.

        I have to admit I patched it with 2.4.8-ac1 (after patching my 2.4.6 with patch-2.4.7 and patch-2.4.8 from my local mirror) and it builds fine as either a module or builtin.

        Yay for Alan.

    • Re:Build Fails (Score:3, Informative)

      by Anonymous Coward
      read the changelog...
      they've changed teh emu10k module...
      merged it with some other stuff.
      the options will be different.

      bound to happen eventually.
      consider it a Good Thing (tm)
      It means the driver you rely on
      is now (hopefully) that much stabler.

  • How to Test Your New Kernel (Score:5, Interesting)

    by goingware ( 85213 ) on Saturday August 11, 2001 @03:14AM (#2118686) Homepage
    Two articles I wrote on the subject of testing kernels:

    You do test your kernel before putting it into production, don't you?

    • You do test your kernel before putting it into production, don't you?

      I thought that putting it into production was how you tested it! ;-)

    • Running tests after installing a new kernel sounds like a good idea, but it would be better if they could be automated. Hardware tests might be better run by hand, since they risk damaging cheap chop suey hardware, but for software tests (like Python, Mesa etc. mentioned on that page) it'd be possible to have the tests installed and ready to run.

      What I mean is that every package in the distribution installs its test suite in /usr/test/ or somewhere, so after doing an upgrade you can just 'test everything' with a single command. As well as kernels, that would be handy for checking that a library upgrade hasn't broken anything which uses that library, a new shell hasn't broken any shell scripts, and so on.
  • Whilst trying to compile the new kernel with the updated emu10k1 drivers I hit a a few problems (missing .o files I think). Has anyone else seen this?
    • Yes I saw it.... I'm going to wait until I see a patch =)

      ---

      make[3]: Entering directory `/usr/src/linux-2.4.8/drivers/sound/emu10k1'
      rm -f emu10k1.o
      ld -m elf_i386 -r -o emu10k1.o audio.o cardmi.o cardmo.o cardwi.o cardwo.o ecard.o efxmgr.o emuadxmg.o hwaccess.o irqmgr.o joystick.o main.o midi.o mixer.o passthrough.o recmgr.o timer.o voicemgr.o
      main.o(.modinfo+0x40): multiple definition of `__module_author'
      joystick.o(.modinfo+0x80): first defined here
      ld: Warning: size of symbol `__module_author' changed from 67 to 81 in main.o
      main.o(.modinfo+0xa0): multiple definition of `__module_description'
      joystick.o(.modinfo+0xe0): first defined here
      ld: Warning: size of symbol `__module_description' changed from 83 to 96 in main.o
      main.o: In function `init_module':
      main.o(.text+0x1880): multiple definition of `init_module'
      joystick.o(.text+0x210): first defined here
      main.o: In function `cleanup_module':
      main.o(.text+0x18c0): multiple definition of `cleanup_module'
      joystick.o(.text+0x250): first defined here
      make[3]: *** [emu10k1.o] Error 1
      make[3]: Leaving directory `/usr/src/linux-2.4.8/drivers/sound/emu10k1'
      make[2]: *** [_modsubdir_emu10k1] Error 2
      make[2]: Leaving directory `/usr/src/linux-2.4.8/drivers/sound'
      make[1]: *** [_modsubdir_sound] Error 2
      make[1]: Leaving directory `/usr/src/linux-2.4.8/drivers'
      make: *** [_mod_drivers] Error 2

  • DRM (Score:2, Interesting)

    by Elendur ( 228338 )
    I notice it says in the changelog that X 4.0x is no longer supported and you need 4.1.x.

    People should take note of this...I suspect most people are still running 4.0.x. I know that's what's in debian woody as of earlier today.

    It's possible I'm mistaken about how this works, but it seemed pretty self explanatory.

    • Re:DRM (Score:5, Informative)

      by Oestergaard ( 3005 ) on Saturday August 11, 2001 @07:50AM (#2125744) Homepage
      As Linus wrote in his e-mail to LKML:

      Ok, this one has various VM niceness tweaks that have made some people
      much happier. It also does a upgrade to the XFree86-4.1.x style DRM code,
      which means that people with XFree86-4.0.x can no longer use the built-in
      kernel DRM by default.

      However, never fear. It's actually very easy to get the old DRM code too:
      if you used to use the standard kernel DRM and do not want to upgrade to a
      new XFree86 setup, just get the "drm-4.0.x" package from the same place
      you get the kernel from, and do

      - unpack the kernel
      - cd linux/drivers/char
      - unpack the "drm-4.0.x" package here
      - mv drm new-drm
      - mv drm-4.0.x drm

      and you should be all set.

  • emu10k1 update (Score:5, Informative)

    by kilrogg ( 119108 ) on Saturday August 11, 2001 @03:49AM (#2125298) Homepage
    The emu10k1 (aka the SBLive) updates were merged in at the last minute. a patch to fix compile problem was posted [helsinki.fi] by one of the maintainers.

    The user-tools can be download here [creative.com]

    This update adds support for the 5.1 cards (including the IR remote, though it's still being reversed-eng'd), AC3 passthrough, multichannel playback (for AC3 pre-decoded in software), sequencer support, dsp effects (flanger, chorus, etc), and much more.

    This is the first sync with CVS in almost a year (development was idle between Nov-April). Report any problems with the driver on the emu10k1 mailing, here: emu10k1-devel@opensource.creative.com

    • Yeah my lirc interface to the remote is coming along quickly. Probably 1-2 weeks.. Check the emu10k1-devel list for updates.

      ~GoRK
    • Hallelujah! The emu10k1 driver in 2.4.2 lacked so many basic features of the SBLive - most importantly to me was the headphone jack in the livedrive IR. DSP effects are also very important - I play guitar and can't access all the cool effect in Linux (had to reboot into windows for this, and I'm too cheap to buy 10 different effects boxes).

      I tried using ALSA, which supposedly works, but I had a very hard time getting it installed. It wouldn't compile for the longest time. When I did eventually get it compiled, I had to hack around the configuration file, and could never get it working. The people at ALSA were not helpful - it's probably the worst example of an opensource project I've seen (lack of useful documentation, difficult to configure and install).

      Can't wait to give the new kernel a try..

  • Question for the masses of linu-geeks (Score:3, Interesting)

    by Lumpy ( 12016 ) on Saturday August 11, 2001 @09:19PM (#2125445) Homepage
    I have used linux for over 5 years now, I can compile the kernel in my sleep,etc...

    I have a big question... HOW the hell do you compile and Install the kernel for a Redhat7.1 system without breaking every damned redhat tool? I kind of like the Automated hardware discovery config, and I would like to start setting up friends computers with linux (RH7.1+Ximian 1.4 is a Windows-killer.. I have 3 win-lovers converted because of that combo)

    Not even redhat support can give me the answer to this question.

      • cd /usr/src
      • rm linux
      • mkdir linux-2.4.8
      • ln -s linux-2.4.8 linux
      • bunzip2 -c linux-2.4.8.tar.bz2 | tar xv
      • cp old-redhat-linux/.config linux
      • cd linux
      • make menuconfig (config, xconfig, whatever)
      • make dep && make bzImage && make modules && make modules_install
      • cp arch/blah/boot/bzImage /boot/bzImage-2.2.18
      • emacs /etc/lilo.conf
      • lilo; sync
      • reboot
      Change as you prefer, of course.
      • BzzzT... I tried this with 2.4.7 and it broke every one of the damned startup programs.
        Modules refused to load (Yet if I insmod the module it loads happily) the new hardware detection system fails with an error also.

        I understand that that is a standard way to install a kernel. but from what I have seen so far, Redhat is not standard in any way anymore (was it ever?).

        • Hmm, works just fine on mine; I've run 2.4.5, 2.4.7 and now 2.4.8. Although, I had to remove the soundblaster live stuff from 2.4.8 and the buz driver from >2.4.3 (but that's not RedHat's fault). You must be changing some sort of configuration option.
      • works wonders
    • Use RedHat rpms... If you want to do it your way, then you have to remove installed kernel rpms from the system... or you'll get screwed :)

      You can always use srpms, to make your own packages, which are easier then to distribute if you have more then 1-2 machines...

      • where do I get the kernel 2.4.8 rpm then?
        redhat only supports 2.4.3 :-)
        how about adding hardware support that they didn't compile in? recompile from their srpms? Nope it blows up all the apps too.

        I have yet to find anyone that can tell me the correct redhat kernel install commands. as a normal kernel install does not work.

  • Yay (Score:1, Redundant)

    by Gnight ( 163400 )
    Can you feel the joy!?!?

    ... I can.

  • What I Want To Know Is... (Score:4, Interesting)

    by Anonymous Coward on Saturday August 11, 2001 @12:53AM (#2128839)
    When will the cryptography patches [kerneli.org] be updated?! I'm still stuck on 2.4.5pre1 for this reason!
    • If all that you want is FS encryption use loop-AES (use the google, luke!). It's kernel version independent and works great for me

      BTW, I don't think my parent post should be marked as a troll. Maybe offtopic

    • You're talking about the patch failing on Makefile, right? It's really a very minor problem, I've been editing it by hand. You just have to add the CRYPTO=crypto/crypto.o line and add crypto to the SUBDIRS. Running 2.4.8 now, mine looks like this:

      CORE_FILES=ker nel/kernel.o mm/mm.o fs/fs.o ipc/ipc.o
      NETWORKS =net/network.o
      CRYPTO&n bsp;=crypto/crypto.o

      LIBS&nbs p;=$(TOPDIR)/lib/lib.a
      SUBDIRS& nbsp;=kernel drivers mm fs net ipc lib crypto

      ...hey guys? What the fuck happened to the <CODE> tag? This was a pain to type up. geez.

      ...AAAGGH!!! THE SPACES!!! I give up. Slash is fucking broken.

  • Kernel (Score:1, Redundant)

    by Anonymous Coward
    Wowzers, so where is the setup.exe? Do I have to edit my registry??? SuperZike.com
  • Is there a cause for updating a 2.4.6 kernel? Are there important bugfixes? The 2.4.6 works very well for me I think I will wait for the 2.5.

  • I can see the future... (Score:5, Funny)

    by Anonymous Coward on Saturday August 11, 2001 @01:00AM (#2144888)
    Three guys found themselves in Hell: CmdrTaco, Hemos, and Jon Katz. A little
    confused at their present situation, they were startled to see a door
    in the wall (which they hadn't noticed before) open, and behind the
    door was perhaps the Ugliest woman they had ever seen. She was 3'4",
    dirty, and you could smell her even over the Brimstone.

    The voice of the Devil was heard, "CmdrTaco, you have sinned!!! You are
    condemned to spend the rest of eternity in bed with this woman!" And
    CmdrTaco was whisked through the door by a group of lesser demons to his
    doom.

    This understandably shook up the other two, and so they both jumped
    when a second door opened, and lo! an even more disgusting example of
    womanhood gone wrong. She was over 7', covered in thick black hair,
    and flies circled her.

    The Voice of the Devil was heard, "Hemos, you have sinned!!! You are
    condemned to spend the rest of eternity in bed with this woman!!!" And
    Hemos, like CmdrTaco, was whisked off.

    Jon Katz, now alone, felt understandably anxious, and feared the worst
    when the third door opened. And as the door inched open, he strained
    to see the figure of... Natalie Portman!!!!!!!!!!!!

    Delighted, Jon Katz jumped up, taking in the sight of this beautiful
    woman, barely dressed in a shiny metallic NASA-space-age material
    bikini. Then he heard the voice of the Devil say :

    "Natalie, you have sinned ........"
    • Yikes, this thing was at Score: 5 a few minutes ago, now it's at Score: 2. I don't want to think about how many valuable mod points this troll just stoll from the system!!!
      • Moderation Totals:Offtopic=2, Troll=5, Insightful=1, Funny=17, Overrated=6, Total=31.

        Funny, sure.
        Overrated, yes.
        Troll, no doubt.
        Offtopic, of course it is...
        Insightful.... Insightful?!? C'mon... What kinda crack was that moderator smoking?
      • I've seen it at scores 5,2,4,0,and 3 in that order.

        Particularly noteworthy is the person that moderated it "Insightful".....

        Will it get 30 or more mods? Only time will tell!
      • but granted, it IS the funniest troll I have *ever* seen!!!!!!!! I'm still ROTFL an hour later....

    • There are so many [google.com] versions out there that I'm not going to guess which is original.

      This troll seems to be based on this version [ucsf.edu] because it mentions space-age material.

      Amusingly the identities of the three guys keep changing, but it's always Cindy Crawford who comes for the last one. ('Cept here, of course.)

  • Just hit kernel.org? (Score:5, Informative)

    by qslack ( 239825 ) <qslack@NOspAM.pobox.com> on Saturday August 11, 2001 @01:02AM (#2146080) Homepage Journal
    I wonder why the /. editors are telling us to "Just hit Kernel.org" when everyone SHOULD use the mirrors.

    Wait a few hours for the changes to propogate, then find a local mirror at http://kernel.org/mirrors/ [kernel.org]. It helps everyone: Kernel.org still has bandwidth, and you get faster download speeds!
    • One thing you forget to mention... unlike many sites mentioned on /., kernel.org is on a 100Mbit connection. In other words, they can serve a whole heck of a lot of people, and still have bandwidth left over. I don't think I've ever seen them useing over 80Mbit or so at any one time, even when 2.4.0 first came out. At the moment (1:38am, PDT), the bandwidth utilization is at 32.84 Mbit/s. In other words, they've still got almost 70Mbit/s not doing anything. While I generally agree that mirrors should be used, in the case of kernel.org it really doesn't make all that much difference.
      • You make a good point, but you should still choose the mirror nearest you. If you're in Saudi Arabia and you download the latest kernel from kernel.org, it is not only going to be much slower for yourself but it will also cause needless international Internet traffic.
    • Please kids, it's simple, I got my PATCH within 25 seconds of typying:

      src@primal:~$ ftp ftp.ca.kernel.org
      ftp
      me@myaddress.com
      cd pub/linux/kernel/v2.4
      get patch-2.4.8.bz2
      get patch-2.4.8.bz2.sign
      bye

      KeepItSimpleStupid

      Use a MIRROR!

      ftp.YOURCOUNTRYABBREV.kernel.org

  • Movin' On Up (Score:3, Funny)

    by malus ( 6786 ) on Saturday August 11, 2001 @08:35AM (#2146141) Journal

    Sadly, I cannot resist the crack-like addiction to having, holding and cherishing the latest greatest kernel release. It's been that way for me since 2.0.10

    Even now, I have my box doing everything I need it to, and it does it well. So why do I tempt breakage by going to 2.4.8?

    +++

    Now for little Microsoft Rant.

    You see, I can't wait to move to a newer Linux kernel release. How many MS OS users can honestly tell me they enjoy moving to a new release of MS system software?

    With Linux, nobody is puting a gun to my head and forcing me to make this upgrade, nobody is telling me "Update your system software for More Security! More Configurability! More Stability!".

    No, quite the opposite. The Linux kernel developers give me a gift every few months, "Here, lookut' we made. Give it a try if you want."

    Microsoft takes such a different, unfortunate attitude towards their system software evolution

    "Here, buy this new update or you will die. We cannot and will not be held responsible for any cancers, distended bowels, hypertension, or any other fate which might befall you if you do not upgrade"

    +++

    Such a pity that MS will never, ever, due to clauses in the EULA, face massive fines and class action lawsuits for the instability of their product (Specifically WIndows operatiting systems)

    If Toyota released a new Car which caught fire at every 5th stoplight, it'd be recalled, and Toyota would in the least settle some out of court cases to familys of burnt owners.

    MS has hundreds of these type of bugs. To think that the damage these faults cause is unlike the previous burning example, you're mistaken.

    Yet nothing happens? No Gov't investigation? I just don't get it.

    • Right on man!

      My windows box caught fire, killed my girlfriend, and burnt off the left side of my face the other day!

      Is there no justice in this world?!
    • The reason no one _can_ sue them is because (a) no one has (to my knowledge) been physically injured or killed (yet) due to bugs in Windows, and (b) Microsoft disclaims thee everliving fuck out of all their products in their EULA (This product is not guaranteed for any purpose, even the purpose it was intended for; no warranties, express or implied; we cannot be held legally liable for this product's failure to perform under any circumstances; if it breaks, you keep all the pieces; etc.), so unless it _really_ screws up, you can't do anything about it. (And of course, the fact that if you did try to use it in a mission-critical application of any kind, they'll point right at the EULA and say "Gee, we said not to do that, didn't we!?!?")
  • 2.4.6 broke VMware's modules. 2.4.7 broke them even worse. I wouldn't mention this, but 2.4 is supposed to be the STABLE tree. If the interface is changing, that's what the 2.5 series is for. Something as big as compatability with VMWare ought to be checked before releasing. If no-one is doing any QA on the big easy obvious things, what little things are breaking that nobody knows about? Professional users of Linux are still hurting for credibility, and this kind of thing just does not help.

    • The interfaces that 2.4.x exposed to the world have been quite stable during the branch's lifetime. However, the internal interfaces have and always willbe subject to change, and this break VMWare's kernel modules.

      It is not the role of the kernel maintainers to cater to commerical software vendors. The kernel hackers want to push out the best standalone product possible, and this involves making modifications to the internal architecture when nessicary. These sorts of changes will break compatibility with kernel modules external to the kernel tree. They are quite aware of this, and are knowingly making the comprimise

      Anyways, if you want the kernel evolution to standstill just so your stupid commerical apps will always run the way they always have, why don't you just refrain from upgrading your kernel?

      -Bodnar42


      • The kernel hackers want to push out the best standalone product possible....
        I thought that if there was anything in this world that was not a "standalone product," it was an operating system. If an "improvement" breaks important and/or popular applications, just how much of an "improvement" is it really?

        Then again, it is not like VMware is either popular or important. Who cares if people can run their "stupid" commercial apps? Particuarly if they are those stupid Windows productivity applications that no one uses anway.

        Far better for the Linux world for those VMware lusers to simply save their data, reformat, and install Windows XP.

    • Re:2.4.6+ broke vmware (Score:5, Informative)

      by mrossbrown ( 70015 ) on Saturday August 11, 2001 @03:40AM (#2146425) Homepage
      I wouldn't mention this, but 2.4 is supposed to be the STABLE tree. If the interface is changing, that's what the 2.5 series is for.
      You shouldn't have. 2.4.x is the stable tree, which means that bugfixes, optimizations, and suggestions of improvement go into these kernels. Massive changes to the entire tree (like a rewrite of the MM, CML2) do not, as it would break kernel builds for a number of releases.

      Something as big as compatability with VMWare ought to be checked before releasing.
      Hmm, I think you have it the other way around. The VMWare developers are responsible for any breakage to their source-only modules, that's the reason they distribute them in source form, so that they can be in tune with the latest kernel release.

      Anyway, get all your 2.4.6+ (and older) VMWare kernel fixes here:

      I'm currently using VMWare with 2.4.7.

      Professional users of Linux are still hurting for credibility, and this kind of thing just does not help.
      Professional Linux users already know that when you update a kernel, anything that's closely tied to the kernel (ALSA, VMWare, etc.) gets recompiled too. It comes with the territory.

      M. R.

      • Now, I wonder when Vmware will fully support XFree86 V4.x.x - without it running WinNT under Vmware is really slow on any graphics operation.

      • >Professional Linux users already know that when >you update a kernel, anything that's closely >tied to the kernel (ALSA, VMWare, etc.) gets >recompiled too. It comes with the territory.

        If you have a reasonable expectation that those
        things will compile, then that's great. If you
        can't have that expectation, it's an indication that the QA process consists of "releasing it into the wild and hoping for the best."

        I've been a linux user since 1.0.9, and I'm not
        trying to imply that I'd put a 2.4.8 kernel into production on monday, or even change a working
        2.4.5 system to anything later without a good reason. But if the devteam want
        to break things, it's high time to start breaking them in a 2.5.0.

        I pointed out vmware because it was what broke for me, and because it is something that is available to anyone testing the i386 kernel.
        Probably it's a bad example because we should
        expect things to be very breakable at that compatability level, and we should know that
        very different results can be obtained from
        different versions of the kernel, and even from
        slight differences in kernel config. The more
        adept ones go as far as to tweak the kernel beyond what's in make config, especially in the
        graphics and networking departments. They know
        that they have to do their own regression tests for every change and it's hard, expensive work.

        I say that vmmon/net and alsa with at least the loopback soundcard should be part of the what's tested before a kernel is packaged for the stable tree. Yes it's not Linus' Alan's, or anyone else's responsibility to test this except the end users'. It would just be a nice, helpful, and professional thing to do.

        "You should have known it would be broken" is
        not an acceptable response to a compatability
        bug report from the field. 2.4.[678] are updates to a stable product, not incremental beta releases.

        Well, on a positive note, linux 2.4.8 seems to run okay under VMWare under 2.4.5. Where can
        I get my hands on the testing framework Alan uses?

        • I've been a linux user since 1.0.9, and I'm not trying to imply that I'd put a 2.4.8 kernel into production on monday, or even change a working 2.4.5 system to anything later without a good reason. But if the devteam want to break things, it's high time to start breaking them in a 2.5.0.
          and
          You should have known it would be broken" is not an acceptable response to a compatability bug report from the field. 2.4.[678] are updates to a stable product, not incremental beta releases.
          You should limit your use of blanket statements to avoid confusing newbies into thinking the wrong thing.

          What's broken? Your compilation of third party modules (where it's always been accepted practice that the onus is on the third-party developer to stay in sync with the mainline). The kernel isn't broken, the point releases are bugfixes and improvements. If a header or API changes slightly (which AFAIK is the only reason why VMWare fails on recent kernels) it should be viewed as a bugfix or cleanup, not an attempt to radically change the way the kernel works.

          Again, it's the responsibility of third-party developers (i.e. VMWare) to insure that their product is compatible with the kernel, not vice-versa. Let the kernel developers worry about stability.

          Case-in-point: The CVS of SGI's XFS is constantly kept in sync with stable kernel releases.

          M. R.

      • I'm using vmware with 2.4.7 and it works fine.

        What basically failed were the vmmon and vmnet modules. You can download them here [vc.cvut.cz]. But remember not to follow the instructions this time. Ungzip the files, rename them to vmmon.tar and vmnet.tar, copy them to your vmwareinstalldir/lib/modules/source directory, and then do a normal "vmware-install" or "vmware-config".

Slashdot Top Deals

UNIX enhancements aren't.

Close