Re: Kernel 2.6.0-final oopsing when using e2fsck

From: Rob Weir (rweir_at_ertius.org)
Date: 12/21/03

  • Next message: Joerg Rossdeutscher: "Re: DynIP mail blocking considered harmful" 
    Date: Sun, 21 Dec 2003 19:41:46 +1100
To: "ML Debian-User (eng)" <debian-user@lists.debian.org>

    On Sat, Dec 20, 2003 at 11:18:04AM +0100, Matthias Hentges said
    > Hello list!
    >
    > I'm currently running a heavily patched 2.4.22 kernel on my P4 system.
    > Since upgrading to 2.4.23 would be a lot of work due to all the patches
    > required to get my hardware working, i thought i would give 2.6.0-final
    > a try (which doesn't need *any* additional patches).
    >
    > I've read Documentation/Changes and upgraded the installed software to
    > meet the requirements mentioned there. I even tripple-checked all
    > versions.
    >
    > The only exceptions are jfsutils, reiserfsprogs, xfsprogs and oprofile
    > (i do not use JFS, XFS or ReiserFS. Dunno what oprofile is and apt-cache
    > doesn't know it).

    oprofile is a kernel profiling tool, one of those "if you don't know
    what it does, you don't need it" sort of deals.

    > Sadly, the kernel is reproducibly oopsing on me when i try to use e2fsck
    > on an ext3 encrypted cryptoloop file (i did *not* try to e2fsck an
    > unencrypted
    > file system because 2.6.0-test11 managed to _corrupt_ [yes,
    > reproducibly] an encrypted one).

    This is almost certainly a kernel bug (the other possibilities are bad
    memory or an angry technomage); either email lkml[0] (after checking if
    anyone else has reported similar problems) or post it to
    http://bugzilla.kernel.org/.

    > I found that cryptoloop in 2.6.0 wouldn't mount the encrypted file at
    > all but 2.6.0-test11 did. However -test11 corrupted the file so maybe
    > the feature has been disabled?

    Sounds like another related bug. Did 2.6.0 oops and fail, or just
    refuse to even try?

    > I have attached a part of my logfile containing the oops and the full
    > output of "dmesg".

    Cool.

    > I had libm???.so going crazy on me when using the tool "replace", too
    > but i didn't save the exact error message, blame on me.
    >
    > My system:
    >
    > OS: Debian Woody (lots of upgrades)
    > e2progs version: tune2fs 1.35-WIP (21-Aug-2003)
    >
    > Please note, that the NVIDIA kernel module was *not* loaded at the time.

    Excellent.

    [snip useful debugging info]

    Also, good on you for digging into the problem and gathering so much
    useful information, the kernel people will love such a clear and concise
    bug report :)

    [0] Linux kernel mailing list, http://www.lkml.org/ for archives and
    http://vger.kernel.org/ for posting information. 

    -- 
Rob Weir <rweir@ertius.org> | mlspam@ertius.org  |  Do I look like I want a CC?
Words of the day:    encryption beanpole Crowell Peking covert video revolution



    -- 
To UNSUBSCRIBE, email to debian-user-request@lists.debian.org 
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
  • Next message: Joerg Rossdeutscher: "Re: DynIP mail blocking considered harmful"

    Relevant Pages