Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 


Forgot your password?
Close
typodupeerror
×
Linux Software

Linux 2.4.7 Released 161

Posted by michael from the sharp-end-of-the-spear dept.
Kazmat was one of the earlier people to write in with the news: "Linux 2.4.7 has just been released! Head on over kernel.org to download it or check the changelog." Remember to run lilo before you reboot! :)
This discussion has been archived. No new comments can be posted.

Linux 2.4.7 Released More

Linux 2.4.7 Released

Comments Filter:

  • Re:Good Christ (Score:1)

    by Anonymous Coward
    These kernel updates are very important to me, so please keep your snide comments to yourself! Thankfully I can spend my friday night updating my OS - I've been waiting for a long time for Jacek Stepniewski's fix for the nasty deadlock in rename().

  • Re:What is "linux" (Score:1)

    by Anonymous Coward
    "sic" is a term used in publishing which means that any errors in the source quote/letter were reprinted, usually to demonstrate either someone's ignorance or for language "flavor."

    Like if someone wrote a letter to the newspaper saying "you guys are a bunch of rediculous jerks," the newspaper could (if they didn't want to correct it and keep you from looking like an asshole) follow "rediculous" with a [sic] so the readers would know that it was the letter writer's error and not theirs, but they were reprinting it as-is.
  • Anyone know what is actually changed (aside from the non-human-readable changelog)? Can anyone give a quick and dirty "bug x in 2.4.6 fixed" list?
  • 2.4.7, eh? I'm still on 2.2.19 - that's where it's at. Just like I stayed in the 2.0 series long after the release of 2.2. I'm ub3r elite.
    I'm actually just too busy (lazy) to bother dealing with the new firewall system. Someone will probably reply to me to point out the fact that I don't know what I'm talking about, which just emphasizes my lazyness - I haven't really even looked into the 2.4 branch :P

  • ieee1394 appears to be horked (Score:4)

    by ultrapenguin ( 2643 ) on Friday July 20, 2001 @11:42PM (#70942)
    for those with ieee1394 devices (in particular, DV cameras), keep in mind that due to recent "updates" to the 1394 driver things stopped working. specifically, there are issues in video1394 driver, which cause a kernel oops when using "dvconnect" program to read/send raw data to the camera. On the other hand there has been some restructuring in the 1394 code and now it at least seems to detect device insertion/removal and runs a kernel thread NodeMngr for this purpose...
    anyway, hopefully 1394 stuff will be fixed by the next release (sending dv data didn't work in 2.4.6 + patches from linux1394.sourceforge.net)
  • Yes. You have to use whats called chainloading. I have 2 scsi disks, win98(sda1) & Linux(sda2) on the first, and Win2k (sdb1) exclusevely on the second.

    # For booting Windows 2000
    title Windows 2000
    map (hd0) (hd1)
    map (hd1) (hd0)
    root (hd1,0)
    makeactive
    chainloader +1

    # For booting Windows 98
    title Windows 98
    root (hd0,0)
    makeactive
    chainloader +1

    # For booting Linux
    title Debian GNU/Linux
    kernel (hd0,1)/vmlinuz root=/dev/sda2 devfs=mount

    # Previous Kernel in case that last upgrade went awry.
    title Debian GNU/Linux
    kernel (hd0,1)/vmlinuz.old root=/dev/sda2 devfs=mount

  • It never seems to fail. I install a new kernel, and a day or two later, a new version is released. I literally installed 2.4.6 last night (maybe 27 hours ago, tops). And I'll wager that, had I not, we'd all still be waiting for 2.4.7 to come out. I'll hold off on installing it for a few days to give kernel.org some time to cool off before 2.4.8 comes out...

    --
  • The humor value of a post really loses something if you have to explicitly point out it's humor, even if it's true.
    --
  • Actually, I've seen this happen. I have a system with 96MB RAM/128MB swap. It grinded to a near halt with some web browsing in Konqueror. Under 2.4.6 it eventually killed kdeinit and recovered somewhat from the near-halt. I basically just had to kill X (It was using ~80% of mem). KDE is quite the hog! It is far worse than Windows Explorer.
  • From what I've read, with 2.4.x kernels you should have twice the swap as you have physical memory.

    You have 64MB of physical RAM, so you should have a 128MB swap partition. This may be the source of your problems when memory is tight.
  • Perhaps Kazmat submitted even earlier than you did, although possibly by only a couple of seconds, and the both of you, along with everybody else who submitted it, sat in the submission queue for a while 'til they (Slashdot) finally got around to posting it.
  • Have you been paying attention to 2.4 kernel development at all?

    If you've watched linux-kernel at all, you will know that there are several potentially system-crippling bugs which were fixed in 2.4.7.

    If you NEED your 2.4.x kerneled Linux machine running, I would recommend one of the following before upgrading any mission critical servers:

    1. Stick with RedHat's official 2.4.3-12 kernel if you don't know what you're doing.
    2. Else wait a couple days for the guinea pigs to find any serious bugs and any patches you run to be ported to 2.4.7.
    3. If it all looks OK, then download, config, compile and install.

    Besides, using the newest kernel DOES make you more 1337. Ask anyone. You don't see Linus running old kernels do you, and no-one gets more 1337 than him!
  • The quick list:

    Lots of obscure deadlocks fixed.
    A number of Reiserfs patches (including one to make exporting NFS shares work properly).
    Lots of driver updates.
    Lots of platform updates (s390, sparc, mips)

    The super quick list:

    Lots of bug fixes, just upgrade damnit!
  • Not a problem: 2.4's netfilter has modules to emulate both ipchains *and* ipfwadm.

    Personally I prefer iptables though: the HOWTOish things at http://netfilter.samba.org/unreliable-guides [samba.org] are quite helpful.

    --
  • It's really weird to be reading this thread about Linux 2.4.7 when it has 247 articles at my current threshold. Now I have to upgrade to 2.4.7...

    Ooops...guess I just blew my own luck by posting this! Looks like I'll have to wait for 2.4.8. :-)


    --
  • This should not be regarded as a complaint, but 'just upgrade it' isn't always that easy, because there are other programs depending on kernel header files etc. like VMware express. I'm lucky not having only one PC, so I can move to 2.4.7 on my laptop (without VMware now, I hope those guys are a bit faster with their updates this time than with the move from kernel 2.2 to 2.4) and staying with 2.4.6 on my desktop.

  • heh (Score:1)

    by ananke ( 8417 )
    i installed 2.4.6 about 30 mins before i noticed that 2.4.7 was out on kernel.org. doh!
    mind as well, i'm gonna go with 2.4.7, simply because they mentioned some updates to pppoe, and i've been having weird problems with that.
  • yeah, what about updating that 'mac' thing you're talking about? adding new features? hehehe
  • i hope you're aware that 'make install' runs lilo.
    and && would not result in executing 'reboot', unless the exit codes would equal success from the previous command - make
  • the point is, that the 'car' as well as 'linux' works as it should from start. however, if you want a newer 'stereo system' for your car, you have that choice. nobody forces you to upgrade, it's totally up to you. i, for instance, like the opportunity to have better support for my devices, drivers for new hardware, and so forth.

    heck, being able to 'update' your car at no cost other than 5 mins of work would be nice. i'd love to be able to 'update' from carburator to fuel injection so easily.
  • My karma has been maxed-out since the cap was implemented, pal, so I'm hardly going to need to karma-whore.

    Go you big red fire engine!
  • And save yourself time and bandwidth.

    (Yes, I know it gets said every time but people still don't seem to do it).



    Go you big red fire engine!
  • I get a kick out of FreeBSD's bootloader. Not only reads the filesystem, it has a modular structure in order to add nifty bits like splash screens to the boot sequence. The kernel is itself a module it loads. Plus you gotta love the fact that it has a full-blown forth interpreter built in :)

    Now I just need to learn how to configure it someday. Thankfully it's like grub in that if I hose things up, I can make it boot arbitrary partitions and kernels from the command line.

    Why Linux is still using that abomination lilo is beyond me. Maybe ESR should take that on for his next project ... doubtful he'll be writing it in python tho :)
    --
  • Seriously though, I found Debian a bit too staid, and before I found out about sid, woody, etc., I had picked up progeny.

    I too found Debian too staid, back in the days when there were only two options: running the stable (but archaic) version or running the unstable (bleeding edge "your system may not boot today") version, but all that changed some time ago.

    Debian has improved vastly ... so much so that when Mandrake 8.0 wouldn't install on my notebook and had problems on my desktop I gave it a new look and haven't looked back yet.

    Debian now has three choices:

    1) The stable but staid release
    2) The testing release (which is more stable that most other distro's final releases, but keeps you very current nevertheless)
    3) the bleeding edge unstable release.

    (to give you an idea of the difference between #2 and #3, testing gives you X 4.0.3 with KDE 2.1.2 while #3 gives you X 4.1 and KDE 2.2-beta. Even unstable is usable, though from time to time packages break, which is why I prefer to run Testing instead).

    I am actually using "testing" in several production environments with less stress and more solidity than I had running Mandrake 7.2, Suse (forget the exact version) or red hat.

    With debian the trade off is doing the work up front (its installation is nowhere near as easy as Mandrake, for example, though it has improved dramatically) vs. doing it later (when it is time to apply those security patches or upgrade ... if you've tried using any of the other distributions' upgrade utilities you understand the pain first hand. Debian relieves that pain ... making upgrades as snappy and easy as typing two commands: "apt-get udpate" and "apt-get dist-upgrade"). Of course, on systems with diskspace enough for two installations (most these days with 40 GB IDE drives selling at less than $200) I cheat and install with Mandrake, then install debian and copy over the configuraitons I need. On my laptop I cheated with Progeny, then pointed /etc/apt/sources.list at Debian testing and did a dist-upgrade to run stock Debian testing, so even the upfront work can be eased significantly.

    The real payoff ... upgrades and maintenance over time, which with Debian has already lightened my workload immensly.

  • "GRUB, on the other hand, reads the filesystems directly"

    I can? Man I'm L33T3R than I thought.

  • Does that back up your previous known good kernel before installing the new one? If so I might try it, otherwise no way. I've burned myself enough times to not have total confidence in my ability to configure a bootable kernel correctly the first time :)

  • Kernel Build Process Usability Review (Score:4)

    by ethereal ( 13958 ) on Friday July 20, 2001 @09:49PM (#70964) Journal

    (P1): arch/i386, arch/ppc, arch/sparc? These are all very confusing

    (P2): gzip or bzip2 - this is a sort of browser, right? There are too many choices.

    (P3): I couldn't find an icon for this "lilo" thing - it wasn't under the "foot" menu.

    (P4): It said I was eaten by a grue, and I can't continue configuring it. What's a grue?

    In all seriousness, great job to Sun for their Gnome Useability Report [slashdot.org] earlier today. This post is in honor of their hard work - hopefully Sun won't ask them to investigate kernel build process usability until they've rested up a little :)

  • Of course it should also be said that if you have a machine which you can test the new kernel on, expose it to as much different hardware, to as many different systems as possible, and check for bugs that would be useful. Making sure any bugs you do find are emailed to the kernel mailing list [tux.org]. The more people using the new kernels on systems which they don't mind the odd crash on the better. More eyes find more bugs.
  • I have to run lilo, since I use "make bzImage", rather than "make fairly-well-crafted".

    --
  • Geez. I read your post and went abso-fscking-loutely cold. What scared me is I know a few Maclots who might actually post something like this. Don't give me any more heart palpitations, K?

    BTW, my next thought was "Gooood troll, buddy."
  • Well, It *is* getting better and easier.
    If you're interested in seeing what all this nonsense is about, get your hands on a brand new distro from somewhere (Linuxiso.org or cheapbytes or a magazine) and try it again.
    I've done a *lot* of installs. Took me a year of occasionally farting with it to get X running on Slackware a few years ago. Now, RedHat and all the others set up like a dream. USB works. Sound works - Good, easy DVD - well, I guess that's about 6mos away.
    Sometimes, I try a distro and it fails on my hardware - C'est la vie. I try another distro. Right now, RedHat is running great for me - 2 months ago, it was Caldera. All the time, I've kept Windows without a problem.
    Now I've been using Linux exclusively for weeks. It has, for me, finally surpassed Windows in usability. (I *do* miss NoteTab Pro, though...)

    Cheers,
    Jim in Tokyo

    MMDC Mobile Media [mmdc.net]
  • yes, I've read that too, but I don't understand why you would want the kernel to kill your processes. Just let malloc() return NULL and let the application deal with the low-memory condition. Part of the problem is that Linux's malloc() will NEVER return NULL because Linux overcommits RAM. That is, Linux uses a lazy algorithm and does not actually reserve pages in RAM when you call malloc() until you actually touch the page. If Linux cannot allocate the page lazily when you actually touch it, then you are screwed because there is no way for the kernel to signal a low-memory error condition to the program.

    Thus your program can (uselessly) check all its malloc() calls for NULL and your program will still crash when you eventually touch the "allocated" memory.
  • FreeBSD's boot loader will load anything with no messy configuration scripts. Just install and boot. Too easy.
  • Patch if you're on 2.4.6 and use iptables.Our sysadmin even posted a cry for help on the kernel mailing list but they didn't even reply to tell him that that was not the place for that.
  • > Does this at all fix the problems when using the K7/Athlon optimziations on VIA boards like the IWILL?

    No.

    Here is the situation as best I understand it right now:
    • The specific problem is with the VIA *686B southbridge. [* = wildcard; I'm not sure what's out there.]
    • That southbridge is part of the KT133A chipset, but on some boards is used with a mixed chipset (i.e., a northbridge from another company).
    • Alan Cox firmly believes that the problem is a bug in the chipset rather than a bug in the kernel.
    • Alan reported within the last week that he now has contacts inside VIA, so he is now showing optimism that they will tell him what needs to be done to program around the bug.
    • Many people report that with various 2.4.* kernels they get a bootable system if they compile for i686 optimizations rather than with Athlon optimizations.
    • Some people get a bootable system that way, but still get random oopses after they have been running for a while.
    caveat lector: IANAExpertOnThis; it's just what I've derived from playing with a 686B-based board, following the LKML, and exchanging e-messages with several other individuals who have played with the board as well.

    Notice that Alan had a VIA-specific patch in 2.4.6-ac4, but that was for a different problem; he reports that he did not expect it to fix the Athlon-optimization problems, but will try to address those when he can find the time.

    Meanwhile, I have to agree with another replier who says buying a board with the VIA *686B southbridge is a bad risk right now. I hate to scare business away from VIA if they are in fact actually working with Alan on this, but I also hate to see people drop $US100++ on motherboards that they can't use.

    BTW, some of these boards are quite nice, so you might consider postponing a new purchase rather than shopping elsewhere, if that fits your needs.

    --
  • Now we can slow the internet down again with all of us downloading this.
    Not as much as the traffic generated by a certain OS'web server frailties...
  • More people running the latest kernel means more bugs found, faster. Folks being l33t benefits us all.
  • freenet:SSK@sUOkGXJDjktWahCNZmvg0sDkEKgQAgE/foldr. org/linux-2.4.7.tar.bz2
    freenet:SSK@sUOkGXJDjktWahCNZmvg0sDkEKgQAgE/fold r. org/linux-2.4.7.tar.bz2.sign

    What is Freenet? [sf.net]

    I still use FreeBSD [freebsd.org], though :)
  • stolz@agamemnon [10:53:23]> telnet freenet.sf.net 80
    Trying 216.136.171.201...
    Connected to usw-pr-web.sourceforge.net.
    Escape character is '^]'.
    GET index.html HTTP/1.1
    Host: freenet.sf.net

    HTTP/1.1 302 Found
    ...

  • Re:Can we dedicate a linux release? (Score:3)

    by Azghoul ( 25786 ) on Saturday July 21, 2001 @05:13AM (#70977) Homepage
    Oh give me a BREAK!

    It could be like that growing number of programs that have to come up with 'funny' release names... this one could be Linux 2.4.7: The Lunatic Fringe.
  • Why not just edit /boot/grub/menu.lst?

    FWIW, I was due to upgrade kernels anyway. When sniffing around, I noticed two things: first, there was linux-2.4.7_tar.bz2. Second, things were going a bit slow towards the end. Thank goodness for my dumb luck:)

  • If there is justice, somebody is going to get metamodded to hell. "Flamebait"?

  • Actually, I would have been editing /etc/lilo.conf if it weren't for how easy it was to install grub on progeny (or was it the default?)

    I tried to wade through the grub documentation, but it didn't make any sense until I saw the menu.lst that progeny put on my machine. It actually works pretty well. I'm not sure that it's any better than lilo, but the menu.lst does look a bit cleaner than lilo.conf.

    FWIW, I put in the comment because I felt someone had to. Looks like it cost me some karma.

    And yes, lilo vs. grub may one day become the next emacs vs. vi. But really, who cares?

  • It sure is. Ever tried Debian ? Now that's what i call upgrading.


    To what? Kernel 2.0.39? X 3.3.5? Gnome 0.8?

    Seriously though, I found Debian a bit too staid, and before I found out about sid, woody, etc., I had picked up progeny.

    And it *has* been an improvement over RedHat (6.2 was working okay for me, but I wanted too much new stuff, and didn't want to deal with the 7.x weirdness).

  • you idiot


    EVER HEARD OF SID/UNSTABLE BRANCH

    what a fucking karma whore faggot.


    Ever heard of reading? It says quite clearly in my post that I picked up progeny prior to learning exactly how sid worked.

    And, FWIW, Progeny has much that sid does, without running the risk of hosing your system.

  • Can it be confirmed this is an issue related to the Via *b Southbridge and Athlon CPUs only? I've been eyeing the Gigabyte dual Socket370 VIA board and hope I'm not stepping into sometihng slick and smelly.

    thx,

    -'fester
  • The Bastille iptables firewall in 2.4+ distributions is good to go, out of the box (so to speak). It's slick - even does IPsec VPN through NAT for one client behind the firewall. I use it to firewall my work Thinkpad into my company. Invisible to the 'net, too.

    I'm running cups and webmin, so I edited the cfg file to lock down several listening ports from external sources, but it was easy.

  • Yes:

    # First Entry: Windows 2000
    title Windows 2000 Professional
    root (hd0,0)
    makeactive
    chainloader +1

    -- iCEBaLM

  • Re:lilo? (Score:5)

    by iCEBaLM ( 34905 ) on Friday July 20, 2001 @11:55PM (#70986)
    I'm not sure that it's any better than lilo, but the menu.lst does look a bit cleaner than lilo.conf.

    Oh, it is. LILO stores the physical address of the first sector of the kernel image and boots off that, if you screw with the partitions even a little bit it'll cause the dreaded "LO" syndrome.

    GRUB, on the other hand, reads the filesystems directly, no running it after every kernel compile, hell, you can even launch the uberleet GRUB command prompt and boot stuff you haven't defined ahead of time, it's extremely flexable and saved my ass a few times.

    In conclusion kids: GRUB > LILO.

    -- iCEBaLM
    This post powered by Mozilla

  • This is going to sound like a flame but its not intended as such!



    Frankly, I'm pretty scared of it. I found a couple lock bugs in it the other day. I posted workarounds which kept the thing from randomly killing processes. The patches didn't get accepted because I'm not one of the worthy 10 kernel 'gods'. The bug is still there so not only did they reject the workaround they didn't do anything else to fix it. Yet they take bugs from the worthy 10 which often times just make the problems worse. The bugs I discovered are 100x as bad in the releases >2.4.6 because of a schedule() change in one of the vmm code paths.


    Its really beginning to get on my nerves. I guess I'm going to switch to some OS other than linux where they actually pay attention to their users/maintainers. What's the point of 10k kernel hackers if they ignore 9.99k of them? How about another example? A few weeks ago someone posted a >3x speed improvement to one of the vmm related paths. This patch was extremely conservative and couldn't have possibly broken anything. Did it get accepted? NO! A couple weeks later someone ese posted a patch that was almost exactly the same thing. Did it get accepted? No. The code is obviously right and a hell of a lot faster but it just can't get changed because the main Linux kernel maintainers don't read the LKML and filter all mail from people they don't know.


    Combine this with the standard Linux attitude of 'it can't happen here, we are better than that' and you will soon discover that Linux is about to have some pretty serious problems. The main kernel hackers are getting in over their heads. You can tell this because of the huge bug counts in systems that should be really robust and stable. You can't run a stable system when your vmm has bugs. Then there are the stupid decisions. Instead of fixing the architectural problems in linux that cause the machine to get into states that are unrecoverable they hack in changes like the OOM (out of memory) killer which kills processes when the machine runs out of RAM. For years OS experts have pointed out serious problems in linux that have been ignored because of the 'can't happen here' philosophy. Linus the fearless leader still sounds like an idiot when general OS and software quality topics come up. For instance, the whole kernel debugger issue (linux shouldn't have a kernel debugger, but Linus admits to using one himself), performance related to coding style issues (maintainability be dammed as long as the code is faster combined with a focus on making the system faster by using obscure compiler tweaks rather than writing clean code and using better algorithms), and the fact that there still isn't a 'real' kernel regression test. Every notable piece of software in the world has a regression/test program that is run to test the SW for weird operations and machine stress testing. I've worked for companies that write OS's whenever a bug was found/fixed there was pressure to create a test in a regression/test base that tested for that condition and assured that the bug fix really worked. This group of tests was combined into a huge batch that got run before release to assure that old bugs were still fixed. This sounds silly but it found a lot of problems. Sometimes these problems were 'new' bugs found because the system was stressed in a way the author of the new bug didn't think of.

  • Way ahead of you! (Score:4)

    by MartyJG ( 41978 ) on Friday July 20, 2001 @10:42PM (#70988) Homepage
    Linux 2.4.7? I can beat that... I'm running on Linux 7.1!

    So there!

    (for those of you who don't know I'm joking - one of us needs help)
  • Actually, Iwill offers several mainboards with the ALi north/south bridges. I plan on buying one for my next motherboard (damned Via incompatibility...)

    If you don't want to spend the $120 for a KA266 (average price offered on pricewatch) then you can try looking for an ECS K7AMA, which also uses the ALi MaGiK chipsets (and also has support for PC133 SDRAM). The ECS K7AMA is about $70 on pricewatch.
  • Yes, you always hear it and say "yea right", but I thought that 2.4.7 wasn't far behind 2.4.6. There were too many '-pre' releases too soon after the release, and the VIA changes seemed not to bode well for myself and fellow Athlon users. I had no reason to update, and I was certainly not going to while I heard the VIA drivers were shaky.

    Anyway, glad to see another release. Seems like only yesterday that I was compiling 2.0.32...
  • Perhaps you should find someone else besides the kernel developers to judge, eh?
    Um... I wasn't 'judging' anyone, let alone the kernel developers. I merely made an observation. If I were to make a judgement, I would do so against VIA. Their handling of their chipsets is an atrocity and it should not be left to volunteer/paid developers to clean up their mess.

    I appreciate your remarks, but your aiming at the wrong person.
  • That's right. It was only beginning with 2.4.6-ac5 that my Dell PowerEdge 2400 w/ 512MB/1024MB of RAM/Swap stabilized. There were constant oopses and races while accessing the disk more aggressively. Especially Ext3 and Reiserfs were bitten by the pathetic state of the VM. The VM just couldn't handle the load journaling filesystems placed on it. But it's all right now, I can stress the VM all I want and it doesn't falter. A kernel compile with concurrent make 'make -j bzImage' does grind the machine into a halt but thankfully it doesn't go belly up.

  • Re:wghat/? (Score:3)

    by dimator ( 71399 ) on Friday July 20, 2001 @11:44PM (#70993) Homepage Journal
    You're the guy they made this [thinkgeek.com] for!


    ---
  • Hmmm... hope you're explicitly setting the umask at the beginning of every single init script. =) Otherwise, you're getting love from a rather cute bug in 2.4.3-6 that makes lots of files world-writable that shouldn't be.
  • Thank you so much for this post! I saw this article, and saw 37 of whatever.

    I immediately thought "Surely by now some self righteous asshole has posted something about being a grub user, not a lilo user by now."

    It is such an ego boost for me to be right about this sort of thing.

    -Peter

    PS: This isn't a troll as such, since this is completely true, and not made up just to piss people off.

  • Okay, I can't legally drive right now and the scary part is I currently have moderator access and I can barely type "cd /pub/linux/kernel/v2.4". But somehow I managed to patch evcerything and I am already reconfiguring. Recompiling a kernel is'nt that big a deal, is it? (Although it would be better if they released on Monday night, rather than Friday night...)
  • I think you'll find that Service Packs are free. At least they are for Windows products.

    xx Stuii!
  • Yes the VM is fix0red up. 2.4.5 was an abberation. 2.4.6 was great. and 2.4.7 is even better.

  • Oh boy, where to begin on the NFS issues in the latest kernels... I've been tearing my hear out for the last week trying to figure out how to get NFS working properly...

    First I had 2.4.5 and Reiser... that didn't work. After reading a lot of mailing list archives, I came to the conclusion that Reiser+NFS just plain didn't work with each other.

    Then someone tells me that 2.4.6 fixed it, so I patched my kernel to find that it was not, in fact, fixed. NFS wouldn't work properly even with ext2 as the underlying fileystem.

    So I went back and tried 2.4.5 with ext2 and this *appears* to work so far. Ergo, I'm not hedging my bets on ReiserFS and NFS working in Linux anytime soon.


  • Not an altogether bad idea, but remember that the kernel maintainers are programmers at heart, not activists. I can imagine Linus right now saying something along the lines of "Let the activists activate, I'll keep writing code."

  • Well, it might be that the partition I was trying to export was 40 gigs in size... I don't know if that's large enough to cause the problem.

    At first, I had 2.4.5 with Reiser. It worked, but would choke whenever I copied lage files (100+ MB) so I upgraded to 2.4.6.

    With kernel 2.4.6, I could get neither filesystem to work properly with NFS. Creating a subdir would work fine, but creating any files or directories underneat that resulted in the standard utilities telling me that the file didn't exist. Something along the lines of:

    $ cd /share
    $ mkdir foo
    $ ls
    . .. foo
    $ cd /share/foo
    $ ls
    [no output]
    $ touch bar
    bash: touch: bar: no such file or directory
    $ mkdir bar
    bash: mkdir: bar: no such file or directory

    (Note that these error messages are not exact.) I then went back to 2.4.5 and tried making the NFS partition ext2 and all of my problems seem to have gone away. Granted, it's going to be pain in the ass to fsck that 40GB partition when the time comes, but for now I guess I have to keep ext2.
  • You shouldn't need to manually run lilo, assuming that you're makeing the fairly-well-crafted install target, and have your lilo.conf set to use /boot/vmlinuz as your kernel...

    (e.g. I usually just do a quick "make oldconfig all modules install modules_install && reboot" and walk away from the system for a while.)
  • How so? Does the bzimage kernel output selection have some size limit I don't know about?
  • It moves the old /boot/vmlinuz before putting the new one in... Thus you can configure an "Old" boot selection which points to the old kernel to be used in case something goes wrong. (However, be careful not to accidentally cycle your only working kernel out of "Old" and into oblivion...)

  • ROFL!!

    Well, the Sun GNOME usability study that this post is satirizing sure demonstrates one thing: Sun usability engineers need to get out more. =)

    [...] Perhaps people are more accustomed to a literal hardware icon used to represent their home directory. For example, the Macintosh has an image of a hard drive and Windows has an image of a computer.

    I mean, come on! How is a UNIX home directory akin to "My Computer"? If so, what is the user's personal "My Documents" folder for if not for holding documents? =) [Unless... But, you expect me to believe that there are people at Sun who don't know what a UNIX home directory is for, do you?!]

  • Given the nature of the pragmatist we all [know|don't know] and [love|feel neutrally about|hate], I wouldn't be surprised to see a comment along these lines [e.g. perhaps with the insult to the intelligence of the poster removed, and with a slight rearrangement of words =) ] come from his general direction.
  • Oh, yeah. You'll need to slip a "dep" in between that "oldconfig" and the "all".
  • and I just upgraded to 2.4.6 on two machines yesterday.

    Well, that's progress for you. At least it keeps on coming.
    -----

  • Can we dedicate a linux release? (Score:4)

    by molo ( 94384 ) on Friday July 20, 2001 @09:20PM (#71009) Journal
    I wonder if Linus, Alan and the whole linux-kernel crew would consider dedicating this version of the kernel to Dmitri Sklyarov. It would be a great way of bringing attention to a subject that we know hits home at least to Alan Cox.

    http://www.newsforge.com/article.pl?sid=01/07/20 /1 228200
  • gah, time for a new processor in the lunix box. it shouldnt take 34 minutes to compile on a 262mhz proc, but it does...
  • i'm talking about this one [zork.net]

    personally i've been running 2.4 for a few months now without any crashs. however (coming from freebsd) i really dislike the ipfw/ipchains/iptables crap. i would just like to be presented with one that is the standard.

    -Jon
  • (why the fuck did this get modded up?)

    i read the fucking "manuals" what ever the fuck manuals are; search for google and you'll find lots of little write ups, some FAQ's some half baked guides, no "manual".

    if redhat linux, or any linux for that matter is a product i would like them so state clealy on what the deal is with three competing firewalls shipping with their product.

    -Jon

  • Ergo, I'm not hedging my bets on ReiserFS and NFS working in Linux anytime soon.

    No, I've been happily serving up NFS off a reiserfs (over IDE raid-0), with no problems! I did have to reiserfsck once, but then found FAQ #3 [namesys.com] ("What's up with NFS and ReiserFS?") at namesys. It addresses your problems and provides patches for older kernels. I've been using 2.4.3 + reiser+nfs_fixes stably for a long time now. They say in the FAQ that these patches have been incorporated since 2.4.6pre3. Go forth and enjoy, I think this combo (IDE RAID + reiser + NFS ) rox0rs!

  • wghat/? (Score:4)

    by acm ( 107375 ) on Friday July 20, 2001 @09:23PM (#71016) Homepage
    hrmm... recompile the kerknel while i'm drunjk or wailt till tomoror moringnig..? Ahh wtf, whats teh worst that coudl happen?

  • Informal test suggests improvement (Score:3)

    by knutroy ( 111950 ) on Saturday July 21, 2001 @04:35AM (#71017)
    I have experienced random freezes and semi-freezes (that defrosts after a few seconds) in the last three releases (2.4.4 - 2.4.6) when memory gets tight. I guess it's too early for me to tell whether this situation has improved in 2.4.7, after just a few hours of uptime. However, my informal test suggests an improvement.

    The test was, on a P200MMX 64MB RAM 64MB swap, to launch Netscape Communicator 4.77, Mozilla 0.9.2, Galeon 0.11.1, Opera 5.0 and Limewire 1.6 ("heavy" Java application) simultaneously. This was done from within the already voluminous XFree 4.1 and Ximian GNOME (not running Nautilus). I also monitored the happening with a GNOME Terminal running "top". Everything was fine, although there was quite a bit of swapping going on. I therefore launched XMMS 1.2.5. The sound was smooth as long as I did not return to the screen with the four browsers. But I had to return to open a few web pages as simultaneously as possible in the different browsers.

    I probably should have ended the test at this point - my HD LED was lighting up the entire neighborhood. However, I launched the GIMP 1.2.1. Confident that this was a manageable task, having no more swap and 1700 kB of memory, Linux 2.4.7 went on starting the GIMP. I admit there were quite a few glitches in the XMMS MP3 playback by this time and eventually, the system came to an effective halt, thrashing like crazy.

    So freezes are gone, good old thrashing is back... I hope.

    Cheers!
  • I'd be surprised to hear about any problems w/ ReiserFS and NFS just not plain working together. While it certainly has (had?) some problems w/ NFS , they're not problems that prevent NFS from working at all.

    The problems all relate to big and busy file servers. We've been running it on our 50-client fileserver at work w/ a 2.4.2 kernel for 4 or 5 months now with no problem.

    Have you actually pinpointed the problem to be Reiser? In other words, does your NFS setup work properly w/ ext2?


    signature smigmature

  • Re:vmware breaks (Score:3)

    by TheGratefulNet ( 143330 ) on Saturday July 21, 2001 @09:59AM (#71030)
    here's a fix: vmware patch for linux 2.4.7 [google.com]

    --

  • Farg, and I just bought a mobo with VIA for my new Athlon CPU...

    Well, it's pretty hard not to if you're gonna use AMD. Notice 2.2 is quite stable with these boards. In fact someone on the LKLM told me there were no problems with 2.2. Obviously Windows users are getting away with it (Then again, reporting blue screens would be pointless).

  • Re:Athlon Problems (Score:3)

    by KidSock ( 150684 ) on Friday July 20, 2001 @11:08PM (#71037)

    Does this at all fix the problems when using the K7/Athlon optimziations on VIA boards

    No. The VIA problems are believed to be a harware issue. The VIA chipset is suspect. New buyers should beware not to buy boards with the VIA chipsets. Does someone have an accurate list of the chipsets believed to be errant?

    There was a post from the 2.4.6 release announcement that I found interesting. Strangely the "drivers" mentioned are obviously Windows drivers but the fact that software claims to fix the problem and the bios update are worth investigating.

    Re:Troubles (Score:1)
    by dlapine (lapine @ uiuc . edu)

    The fix is simple. Grab the latest the via drivers set, 4in132 and install it. There may also be a bios update that fixes this problem as well.

    The problem was: copy 100 megs in 1 or more files at a time from one ide drive to another drive. system locks hard, requireing a reset at least, and sometimes a power cycle. You know that its fixed when it doesn't do this again.

    Small tip: grab and install the via busmaster drivers 3011 as well, selecting the miniport option. This lets windows "see" the correct info about your harddrives (i.e. IBM DTLA 305020 is reported as such and not "drive type 47") without any performance hits.

    These files are available at:

    http://www.viahardware.com [viahardware.com]

  • The VM (Score:5)

    by KidSock ( 150684 ) on Friday July 20, 2001 @09:35PM (#71038)

    All I want to know is wheather or not the VM is stable. From what I understand it has been the source of instability and deadlock. I know Linus, Rik van Riel, Andrea Arcangeli, and others have been hunting for the source of the reported problems and did find some potentialially serious bugs but it's been difficult to reproduce the problem which as the programmers here know greatly complicates finding a fix. I'm getting some of this from the last paragraph of the Kernel page at lwn.net [lwn.net] and there's an intersting thread in the lk mailing list here: Re: VM in 2.4.7-pre hurts... [indiana.edu]. Anyone have any insight into this particular problem? And I wonder if Linus will drop the issue and turn his attention to the imminent 2.5. I think that would be a mistake.
  • I was wondering what the reasons were behind not
    including a low-latency patch; either ingo's or
    andrew's. I always apply Andrew's patch before I
    try a new kernel. It does help quite a bit. Newer
    versions also include support for reiserfs and are
    configurable via the /proc fs. It seems like it
    would be a good thing to include nowadays.

    my 2% of $1.00 for today
  • I reinstalled Red Hat 7.1 and as soon as I get it to compile, I'll be running kernel 2.4.7!!! Gotta love linux, no paying or waiting for service packs! Woohoo!!
  • I once saw a presentation on an IRC channel from Rik van Riel (I think) and he said that 2.4 would have an algorithm that when memory was exhausted it would choose a process to kill based on the time it was running, the memory it was consuming and some other things. From what you're saying, it seems that it isn't working or it hasn't been made yet.

    Does anyone know anything about this or did I just dream that?

  • vmware breaks (Score:5)

    by willamowius ( 193393 ) on Friday July 20, 2001 @11:37PM (#71053) Homepage
    2.47 breaks the compilation of the vmware modules. If you need them, wait for an update from vmware until upgrading.

  • XOSL (Score:3)

    by cbr372 ( 193706 ) on Saturday July 21, 2001 @01:53AM (#71054)

    You might want to check out XOSL [xosl.org] (Extended Operating System Loader). It pretty much rules over GRUB and LILO. It's GPLd, has a graphical interface that reminds one of Norton System Commander, yet the menu-driven options reveal features that System Commander didn't have.

    It's a breeze to use and allows all the regular stuff - password protection, multiple partitions, as many OS's as you want, different OS labels, plus, there's no need to reboot or restart (ie lilo -s) to add new OS entries. You can do it all from the system menu and it will take effect immediately. A very good system.


    Cedric Balthazar Rotherwood
    Sun Certified Programmer for the Java Platform +

  • something else that should be said (Score:3)

    by gtx ( 204552 ) on Friday July 20, 2001 @09:22PM (#71059) Homepage
    if you're system is working fine, and nothing is broken, and you cannot really think of any good reason to update your kernel... you probably should just leave it alone. using the newest kernel won't make you any more 1337. don't upgrade until you can see a good reason to.

    (although, i'm guessing that 90% of you already know this, there's probably hundreds of people rushing to download the new kernel just for the sake of doing so. give a hoot. don't pollute.)


    "I hope I don't make a mistake and manage to remain a virgin." - Britney Spears

  • Re:Umm... Do you EVEN know what you're talking abo (Score:4)

    by gtx ( 204552 ) on Saturday July 21, 2001 @05:29AM (#71060) Homepage
    being a long time subscriber to 'if it ain't broke, don't fix it' i just thought i'd point out that i haven't upgraded to 2.4.7, and my machine still works just the same as it did yesterday... beautifully. it's been running beautifully for... 323d 10:09.45 now.

    and even if using the newest kernel makes you 1337 TODAY, just think 10 years down the line when i pull out my crusty AMD and show the kids my 2.4.6 kernel. then they'll all be like "whoa, that guy's old school, he must be 1337 as fuck!" ;


    "I hope I don't make a mistake and manage to remain a virgin." - Britney Spears
  • i read the fucking "manuals" what ever the fuck manuals are; search for google and you'll find lots of little write ups, some FAQ's some half baked guides, no "manual".

    Why so hostile?

    Personally, I can't see what the problem is. I'd never performed firewalling of any kind until I got a decent network connection 2 weeks ago. Coming from zero knowledge about firewalls to writing some effective iptables rules really wasn't difficult at all. As far as I saw, the documentation I read made it very clear that ipfw/ipchains was legacy support, and iptables was the firewalling system. I even managed to get NAT working despite not knowing what NAT was 2.5 weeks ago.

    It just ain't rocket science, and I can't see why it's so confusing. If you don't like Linux, instead of complaining about it - why not go back to using FreeBSD?

  • i really dislike the ipfw/ipchains/iptables crap. i would just like to be presented with one that is the standard.

    Reading the Fine Manuals might have cleared up your confusion on that score...

    In any case, here's the short answer: iptables is the standard on 2.4.x and henceforth. Period, end of discussion. The ipchains/ipfw user and api interfaces are provided for the usual legacy support reasons, but the iptables code is still doing the lifting.

    Ugly, but you can only change so many things before the major distribution vendors start sending you dead squirrels by surface post in appreciation for making their lives more difficult.

  • Life was going nowhere for me, until one day a friend directed me to get help. "Who should I see" I asked. Rather than suggesting a doctor, analyst or other orthodox methods, he directed me to the Linux PrePatch Cycle of happiness and relief. An unbeliever, I doubted him until, that is, I patched my kernel to 2.4.7-pre5 and then:
    -pre5: - Andrea Arkangeli: softirq cleanups and fixes, and everybody is happy again (ie I changed some details to make me happy ;)
    Yes! Happiness entered my life! The sun was shining, birds were singing. And then came -pre6 and my life was forever changed, I became a firm believer:
    -pre6: - Tim Waugh: parport drievr documentation, init sanity
    My sanity was restored, initialized like I had never been able to do. Yes friends, the Linux Prepatch Cycle changed my life, it can change yours -

  • Ahhhh! (Score:3)

    by V50 ( 248015 ) on Friday July 20, 2001 @09:11PM (#71073) Journal

    I downloaded this 'Linux' 2.4.7 like you told me to and after figuring how to unstuff a 'tar.gz' file all that I have are about 8000 files! And none of them seem to be installers!

    How am I supposed to use this 'Linux' everyone here seems to be talking about if I can't install it on my PowerMac 6100? And now you are telling me to run 'LILO' before rebooting? I don't have a 'LILO'! And why do I have to reboot to run this 'Linux'?

    Is this some evil Wintel only program? Can someone find me the Mac version of 'Linux'? For that matter can someone tell me what Linux does? Is it a Word Processor? A browser? A Graphics program? HELP!!




    Stupidity Disclaimer: That was an attempt to be funny. I know what Linux is. I've run it before. I use a Mac. I was making fun of some computer newbie's posts I've seen on some message boards.

    --Volrath50

  • Well Athlon users don't have much choice as the VIA chipsets are used in most mobo's Even new boards with teh AMD 760 are not using the AMD southbridge - they are using the VIA one...
  • Although every reader can figure out that the parent was not written by me, I still do not like the imposterer.

    FWIW, Please refrain from this in the future.

    Now, all go and download 2.4.7! I didn't work so hard on it for nothing. :)

    Linus
  • -final:
    - me: fix ptrace and /proc issues with new core-dump permissions
    - numerous: IDE tape driver update for completion handlers
    - Ben Collins: ieee1394 GUID cleanups
    - Jacek Stepniewski: nasty deadlock in rename()


    Whew, good thing Jacek was able to get that deadlock into this patch!


    ...ducks! :)

  • Re:wghat/? (Score:4)

    by MrRudeDude ( 450053 ) <mr_rude_dude@yahoo.com> on Friday July 20, 2001 @09:25PM (#71089)
    I have never successfully built a kernel sober. Odd but true fact.

  • Kernel Review service (Score:5)

    by The Ultimate Badass ( 450974 ) on Friday July 20, 2001 @09:33PM (#71091) Homepage

    2.4.7 is not only a dramatic improvement over 2.4.6, it is a whole new paradigm in linux computing. I urge any linux users to update with all haste. You simply must not miss out on the transition, or you will be left behind. From now on, we won't be asking "Got 2.4.6?" No, now it's "Got 2.4.7?"

    The minor changes to USB will definitely rock your world, and you'll love what "initialize page->age" will do to your swap cache. I think Linus said it best, when he said "semaphores are not good swap handlers."

    Remember: 2.4.7, a quantum leap forward.

  • But has anyone gotten GRUB to boot NT partitions?

Slashdot Top Deals

Seen on a button at an SF Convention: Veteran of the Bermuda Triangle Expeditionary Force. 1990-1951.

Close