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

 



Forgot your password?
typodupeerror
Linux Software

Linux 2.2.8 73

KrON writes "Linux 2.2.8 Is being upp'ed to mirrors as we speak... " A few people have noted that a 2.3 directory has appeared on some of the mirrors, although it is empty.
This discussion has been archived. No new comments can be posted.

Linux 2.2.8

Comments Filter:
  • by Anonymous Coward
    I noticed that this time they actually made it to prepatch 7 which is a long way past most of the 2.2 kernels have been. Maybe I can actually get an uptime of more then 2 weeks now.
    ftp.funet.fi and ftp.ca.kernel.org have 2.2.8 for you must-have-it-right-now-cant-wait-30-mins people
  • by Anonymous Coward
    [root@localhost linux]# make
    make: *** No rule to make target `include/linux/autoconf.h', needed by `include/config/MARKER'. Stop.
    [root@localhost linux]#


    at one stage it was compiling then it told me I didnt have 'as86' command (near the end of a make bzImage)...

    fuck
  • by Anonymous Coward
    Try: make mrproper
  • by Anonymous Coward
    as86 is part of the bin86 package (RedHat) ...

    you need a few basic things installed iwhen you try to compile stuff ... for kernels you need kernel headers and some of the development packages. It isn't so hard to figure out if you try to understand the errors
  • by Anonymous Coward
    There's an option to tie down the DMA buffer for an ISA bus soundcard so it doesn't swap. Gives me a major reduction in drive thrashing under gnome + Enlightenment.
  • by Anonymous Coward
    Actually, when the redhat kernel-headers RPM is removed, the symlinks from the include directories to the proper directories in the linux src is missing.

    My trick in de-rpming my kernel has been to remove all kernel packages but kernel-headers. Then go to /src/ and remove linux*, then unpack your new kernel src there and configure and build as desired/needed.

    Jeff
  • by Anonymous Coward
    Pop over to www.linuxhq.com [linuxhq.com]. If they don't have a change summary up yet, there should be a link to the cutting edge linux site that usually has them.
  • by Anonymous Coward
    ...to discover that Linus has taken a patch you
    sent out twelve hours before marked "please test"
    and stuck it into the official kernel, without
    saying a word.
  • by Anonymous Coward
    Is USB supported? No.

    But that doesn't mean it won't be. The code is out there, ready for nice people to test. Uncomment it from config.in and have a go at it, and be sure to check out the project info at the new Linux-usb web site [linux-usb.org].
  • Of course, after ESR's paper [tuxedo.org], marking a piece of open source code "please test" is equivalent to asking the author to have lots of people try it out. :)
  • Try either of
    make menuconfig
    make dep

    I know that one of these does the trick (or at
    least it did for my housemate)

    Can anybody improve on this?
  • by Lurker ( 1078 ) on Tuesday May 11, 1999 @05:47PM (#1898119)
    I used to play the upgrade-with-every-kernel game, too. That got old, so now I just keep the latest release kernel compiled and lilo set to boot it in case my box crashes (hasn't happend since I got a clue and stopped messing about as root), the power goes out (doesn't happen too frequently), I add/upgrade hardware, or I need to install a kernel that fixes bugs/exploits. I've been running 2.2.5 for 38 days now. I'm going to install more RAM this weekend, so 2.2.8 will go up.

    I keep my thirst for upgrades/beta code quenched by occasionally booting my Mac into the latest dev version.
  • You will see such updates and fixes in EVERY kernel release. What goads me is people who complain about just having downloaded 2.2.X and now 2.2.X+1 is out so now they have to download THAT. It all comes down to this: if there is a legitimate reason to upgrade the kernel, do so. If not and you have some fetish with running the latest kernel all the time, then don't complain about them being released so often. No one is making anyone upgrade and its nice that there is active development on the kernel. Like someone mentioned, we could have the Microsoft format - a monolithic service pack every few quarters or few years, with a bunch of hot fixes in between.
  • Do you have the binutils rpm installed? the as86 assembler is part of that package....
  • by KmArT ( 1109 ) on Tuesday May 11, 1999 @01:42PM (#1898122)
    There's an easy way to get long uptimes - don't upgrade the kernel unless you don't have to. It amazes me all the people whining about HAVING to upgrade the kernel everytime a new one comes out. You don't have to and if you've got a configuration that is working fine, doesn't contain bugs that cause it to crash or have bugs that lead to security exploits, then just leave it be for goodness sake! I played the kernel upgrade game with the 2.1.x on my own workstation but my servers here behind the firewall are running whatever they are running and they'll stay that way until someone comes out with an exploit that can be launched from a Macintosh (we have almost entirely Mac clients) or until I absolutely need one of the new kernel features to do something I need to get done.
  • I believe the scheduler updates did make it.
  • Like the scheduler fixes, that everyone needs?
  • There is a lot of new scheduler code in, certainly, and I remember all of the discussion about scheduler issues, but could someone please summarize exactly what they do? Is it just to handle high-load situations with a lot of processes (i.e. Apache), or does it fix a scheduler bug of some kind?
  • Also they added wak_up_on_one, which affects accept (waking up only a process that is waiting, not all. *think to 250 apache process :)*

    The funny thing is this probably won't affect the benchmarks at all. In the benchmarks, the web clients are all on fast LAN links, so they will quickly download the files and get on with the next request. In the real world your clients are on the other side of modems, ISDN connections and other bottlenecks like the Altantic and it takes them an appreciable time to get the data. So for a given bandwidth you have a lot more open connections ie a lot more Apache threads/processes. to get the data.

    I wonder when there will be an Apache release that actually takes advantage of this.

  • Actually some of the most respected benchmarks like SPECcpu95 [spec.org] and TPC are done by the vendors themselves. The trick seems to be firstly to have a strong body overseeing the benchmarks (like SPEC) and secondly that everyone benchmarks only their own stuff and never anyone elses.

    Of course, you also need realistic benchmarks.

  • I haven't found a Changelog either, but I went through the diff and noted a lot of changes in the USB tree. You still can't pick it from the menu though (as far as I can tell).
  • Why waste your time getting the tarball and then doing a 'make config'. The mirrors are fast, but they aren't that fast (and I'm sure your typing isn't either). Get the patch and 'make oldconfig'.
  • at one stage it was compiling then it told me I didnt have 'as86' command (near the end of a make bzImage)...


    Ummm ... you gotta install the x86 binutils to build a kernel. That will get by THIS error.

    /dev

  • You can do that if you're truly stupid. Sensible people will get bzip2'd patches rather than gzip'd tarballs.
  • First off, simply install the bin86 package to get round your second error. For a quick and painless kernel build:


    make mrproper
    make menuconfig
    (fiddle with those options)
    make dep
    make clean
    make bzImage
    cp arch/architecture/boot/bzImage /boot/vmlinuz-new
    cp System.map /boot/System.map-new
    vi /etc/lilo.conf
    (add stanza for new kernel)
    /sbin/lilo
    reboot


    Once everything is running Ok, then you can remove the RedHat kernel RPM's (kernel, kernel-pcmcia-cs, kernel-headers, etc) using --nodeps if necessary. Edit Do make clean in your Linux source directory to recover some space, and wait in anticipation for the next kernel patch ...

    Chris Wareham

  • That is probably the best combination of having the latest kernel, and keeping uptime. I'm running 2.2.2 (with the bug/exploit/whatever that people are talking about), and having 70 days uptime now (the last time it went down was when somebody unplugged the power). But I will probably compile 2.2.8 (aka 2.3.0) any time soon, just in case...

    /* Steinar */
  • Well, I did use an AC patch... 2.2.5-ac6 has been running on this machine for about 30 days now. It's very tempting to boot a new kernel...
  • > Hmmm, if I flushed NT any more, it would sound like a toliet.

    Ya know that goofy little chimes sound Windows makes when it starts up? Well, you'll never guess the sound file I installed to replace it with... ;-)

  • The include for drivers/usb/Config.in is commented out in arch/i386/config.in. Uncomment it, and USB will show up in menuconfig. It isn't even in the config.in for other platforms.

    I still haven't gotten it running, tho, so YMMV. At boot time, it doesn't detect the USB keyboard. If you send SIGUSR1 to the uhci-control thread to dump some debugging info, it does detect the keyboard... but alas it still doesn't work.



    -- Rob
  • by Dast ( 10275 ) on Tuesday May 11, 1999 @03:06PM (#1898137)
    Could someone kindly point me to a changelog? I've looked and I can't seem to find one. :(
  • I'm having some trouble getting the 2.2.8 source to compile. I admit I've never done it before, but I think the problem is more a result of my system config than my lack of procedural knowledge. I am using a LinuxPPC system. It's currently using a 2.2.1 kernel. I get it mostly compiled, and at the very end I get this:

    ld -T arch/ppc/vmlinux.lds -Ttext 0xc0000000 -Bstatic arch/ppc/kernel/head.o init/main.o init/version.o \
    --start-group \ arch/ppc/kernel/kernel.o arch/ppc/mm/mm.o arch/ppc/lib/lib.o kernel/kernel.o mm/mm.o fs/fs.o ipc/ipc.o \
    fs/filesystems.a \
    net/network.a \
    drivers/block/block.a drivers/char/char.a drivers/misc/misc.a drivers/net/net.a drivers/scsi/scsi.a drivers/cdrom/cdrom.a drivers/sound/sound.a drivers/pci/pci.a drivers/macintosh/macintosh.a drivers/video/video.a \
    /usr/src/linux-2.3.0/lib/lib.a \
    --end-group \
    -o vmlinux
    arch/ppc/kernel/kernel.o: In function `sys_mmap':
    arch/ppc/kernel/kernel.o(.text+0x4a38): undefined reference to `fget'

    Now, man has references to fgetc, fgets, getc, getchar, gets, and ungetc, but there is no mention of 'fget'. Where does this come from? The code reference is apparently in syscalls.c as follows:
    file = fget(fd);

    where fd is an unsigned long in the function sys_mmap. Is there a simple answer here that I am missing? Where is the right place to ask such questions?

  • Here is the URL to the changelog

    http://edge.linuxhq.com/changelist.cgi?show=2.2.

    And for those who are really lazy:

    Notes: Due to the changes in this kernel's buffer code, you may want to disable the update (bdflush) daemon
    in your init scripts.

    Documentation updates.
    3c527 net card driver added.
    MTRR support added for Cyrix 6x86, Cyrix 6x86MX, Cyrix mII, AMD K6-2 (stepping 8 or higher)
    and K6-3.
    Netwinder Rockwell WaveArtist sound system driver added.
    #PCMCIA NE2000 net card driver added.
    Major Netwinder (ARM) architecture.
    Alpha architecture updates.
    ARM architecture updates.
    x86 IRQ code updates.
    x86 MCA support updates.
    Cyrix 6x86 coma bug workaround added.
    x86 SMP code updates.
    Large m68k architecture updates.
    MIPS architecture updates.
    Major PowerPC architecture updates.
    Macintosh architecture updates.
    SPARC architecture updates.
    SPARC64 architecture updates.
    Minor IDE driver update.
    Ramdisk driver updates.
    Generic Non-IDE CD-ROM driver updates.
    Lineprinter driver updates.
    PlanB frame grabber driver added.
    Cadet radio card driver updates.
    Zoltrix radio card driver updates: Radio strength and stereo scans now use somewhat less CPU.
    minor SpecialX multiport serial board driver updates: It appears that IO base and IRQ are now
    configurable with module parameters.
    act2000 ISDN driver updates.
    Parport driver updates.
    de4x5 net card driver updates: Workaround for buggy SROM in Motorola embedded boards, PCI/EISA
    probing order changed.
    Locking changes in various drivers: SCSI drivers, net card drivers.
    Minor EQL net device driver update.
    ethertap net card driver updates.
    PPP driver updates.
    WaveLAN net card driver updates.
    IBM MCA SCSI driver updates.
    MegaRAID SCSI driver updates.
    Generic SCSI driver updates.
    ad1816 sound driver updates.
    ESS sound driver updates.
    USB driver updates: OHCI-HCD and OHCI-HCD virtual root hub support added.
    USB mouse driver updates.
    Minor Matrox framebuffer console driver updates: CHRP and PReP fixes.
    Framebuffer console drivers updated.
    Minor ADFS filesystem driver update.
    Minor AutoFS update.
    Minor a.out binary loader updates.
    Minor ELF binary loader updates.
    Filesystem buffer code updates.
    Disk quota code updates.
    ext2 filesystem driver updates.
    VFS updates.
    Minix filesystem driver updates.
    NFS filesystem driver updates.
    /proc virtual filesystem driver updates.
    Minor QNX4 filesystem driver update.
    Minor SMB filesystem driver update.
    Memory management code updates.
    Scheduler updates.
    Net code updates.
    ipv4 updates.
    Minor ipv6 updates.
    IRDA drivers updated.
    SunRPC updates.
    Unix domain sockets code updated.
    Minor MenuConfig update.
    XConfig updates.
    ********************************************
    Superstition is a word the ignorant use to describe their ignorance. -Sifu
  • You want people to stop whining because they don't NEED to upgrade? I'm ignorant.. I'll admit it. But when you look at the changelog and see things like:

    ipv4 updates
    Filesystem buffer code updates
    Minor IDE driver update
    Locking changes in various drivers: SCSI drivers..

    ... they seem pretty important when you look at them, don't they? Or is there a difference between an update and a fix?

  • Would you rather that Linus just sat on the bug fixes, or released a "service pack" every 3 or 4 months ala windows nt. I'm glad that there have been 8 kernels since january. The more bug fixes the better.
  • Arg! I just finished going from 2.0.36 to 2.2.7 on my main server last night. (what a rush).

    Its real depressing when the frequency of reboots to an OS is dictated by the frequency of new features to it. :)

    Though, I'd rather reboot my linux box every few weeks for an upgrade than reboot it for the same reason we reboot the NT server at work.
    "Its recommended that NT be rebooted each week to flush out any crap that gets built up in it"... Hmmm, if I flushed NT any more, it would sound like a toliet.
  • What's the point of dpkg'ing (or RPM'ing) a kernel? Aren't they MEANT to be customised for your computer? (Yes... i'm not talking about .src.rpm's :P
    Isn't customised kernels one of linux's main advantages over the Win32 platform?
  • by jelle ( 14827 ) on Tuesday May 11, 1999 @12:54PM (#1898144) Homepage
    Rumors say that this is the first version in the 2.2.x series where some problems with the scheduler (that existed since the late 2.1.1xx kernels) are finally ironed out.

    I'd like to see what difference this one makes on the 'dreaded' Quad PIII/Quad 100mbit/Hardware RAID/>1GB RAM configuration that is soo popular lately in MS-sponsored 'benchmarks'...

    Btw: Who thinks like me that it may be time for one or more larger Linux organisation (like Linux International, RedHat labs, and/or Suse) to setup a high-performance benchmark lab that can be used for in-house benchmarking, and kernel development and optimization of Linux by our cutting-edge Linux developers (linus, alan, david, andrea, donald, stephen, andre, richard, rick, and the o-so many others, including the samba, apache, and squid guys).

  • There's a file in /etc on Redhat that specifies the name of the kernel release. You need to change this to 2.2.8, or it will continue to search in the old directory.
  • by mlc ( 16290 )
    On the Cutting Edge Linux [linuxhq.com] page, it says to disable bdflush when installing the new kernel. Could someone explain exactly what this does, and why I have to disable it now?
  • by mlc ( 16290 )
    For those of us who don't have the disk space needed to recompile the kernel, debs are nice.
  • I just compiled 2.2.8 not too long ago, and I don't recall seeing anything in "make menuconfig" related to USB. Kernel 2.2.7-ac4 had it, though.
  • I put some debs of the new kernel up at my ftp server [150.135.194.252]. They don't have SCSI, SMP, or ISDN, but they've got most everything else.
  • You might be able to de-politicize it by just benchmarking Linux rather than Linux vs. other.

    It would still be useful to see absolute ratings about what performance levels you could get for different levels of investment in hardware, and to document what kernel versions to favor/avoid, and what tuning tweaks were needed to get the various performance levels.

    And besides... who said we were mature adults, anyway?

  • Consider it as confirmation that someone else agrees that you had a good idea. If your patch was included unaltered, then they agree that you also phrased it correctly.
  • I saw a couple threads not long ago that there was a couple different USB efforts going on and that 2.2.7 had some code disabled.

    I've looked (not very hard I'll admit) for a changelog or something to see whats in 2.2.8

  • from

    cutting edge change listing [linuxhq.com]


    Simple USB driver added: You can't enable this option from the configuration yet. Maybe in 2.2.8.



    but there is no changelog up yet for 2.2.8?

  • its not up yet but I guess it will be here [linuxhq.com] when it gets posted

  • Better yet is to just un-rpm kernel and kernel-headers. Ignore the errors. Then, cd to /usr/src and unpack the kernel source there, mv the linux directory to one that is aptly named (linux-2.2.8, in this case), ln -s linux-2.2.28 linux, then unpack the kernel sources and then make the symlinks from /usr/include for linux and asm to /usr/src/linux/include/linux and /usr/src/linux/include/asm, respectively. Finally, go to /usr/src/linux/include and ln -s asm- asm. In my case, I use asm-i386 for the architecture-specific asm headers.

    After that, NEVER use another RedHat kernel RPM. :) Just use the tar.gz's (or bz2 if you prefer them).
  • Im running 2.2.5 on my SuSE 6.1 system, which also has 2 iP2-350 processors. The SMP performance is actually quite nice and lacks nothing in comparison to NTs SMP support.
  • If you read again, you will see that the idea of an in-house benchmarking lab is not to make good PR, but to do some real tests that provide the developers with data that helps them in optimizing Linux.

    There's no conflict of interest at all.
  • How is SMP support in these newer kernels?
    I am about to put together a dual proc pent II
    and I want to make sure I get a rock solid
    SMP kernel on it. Would 2.2.8 be the
    way to go? or would another rev be more appropriate? How is smp support in 2.2.5..the
    one that comes with redhat 6.0?

    Thanks.. Mike
  • Use the file drivers/net/ethertap.c from 2.2.7
    and then recompile... it should work fine then.

  • It's parody, not politics, my anonymously cowardly little friend! Don't hate me, I'm the ever-lovable gnulix guy!

Solutions are obvious if one only has the optical power to observe them over the horizon. -- K.A. Arsdall

Working...