#34 Kernel oops from 'gettimeofday'

closed
nobody
None
5
2005-02-18
2004-07-21
Anonymous
No

After running for a while a server daemon I got the
following kernel oops:

divide error: 0000 [#1]
Modules linked in:
CPU: 0
EIP: 0060:[<c010fd97>] Not tainted
EFLAGS: 00010283 (2.6.7-co-0.6.2)
EIP is at get_offset+0x57/0x60
eax: e65eb880 ebx: c1639fa8 ecx: ffffcbee edx:
ffffffff
esi: 06775b1c edi: 06775b1c ebp: c1639f9c esp:
c1639f88
ds: 007b es: 007b ss: 0068
Process etded.x86 (pid: 19399, threadinfo=c1638000
task=cba646b0)
Stack: c010da5a 00000000 c1639fa8 bfff7398 00000000
c1639fbc c0117b8b c1639fa8
0000004e 0000004e bfff7398 bfff7440 4c4adca8
c1638000 c0105967 bfff7398
00000000 bfff7398 bfff7440 4c4adca8 bfff7438
0000004e c010007b 0000007b
Call Trace:
[<c0106c65>] show_stack+0x75/0x90
[<c0106dbb>] show_registers+0x11b/0x170
[<c0106ef0>] die+0x50/0xb0
[<c0106fe9>] do_divide_error+0x99/0xc0
[<c01068cd>] error_code+0x2d/0x40
[<c0117b8b>] sys_gettimeofday+0x1b/0x60
[<c0105967>] syscall_call+0x7/0xb

Code: f7 35 20 05 2f c0 c3 89 f6 55 89 e5 e8 d8 fd ff
ff 5d a1 18

I got quite surprise that a simple user level
application could bring this kernel error (btw, I've
been using this application for a long time with normal
kernels a no problems so far).

It seems to be related with the system function
'gettimeofday'.

Thanx for your help.

PD. I've just read in the mailing lists about a patch
for 'gettimeofday'. Maybe it is closely related.

Related

Patches: #1

Discussion

  • Logged In: NO

    BTW, I forgot to mention that I used 20040710 snapshot.
    Thank's again.

     
  • Logged In: NO

    I forgot to mention that I'm using the 0710 snapshot.
    Thx again.

     
  • Logged In: YES
    user_id=30412

    Ok. Thanks for mentioning the version. Couple questions &
    1 thing to try:

    Does it kernel oops everytime?

    Try getting
    http://firechief.dyndns.org/colinux/coLinux-0.6.2-gpb.tar.bz2
    make a back-up of your coLinux directory,
    unarchive this file into your coLinux directory.
    Try to run the service that kernel oops.
    Let us know if this fixes your problem or not.

    Thanks,
    George

     
  • Logged In: NO

    The "oopses" don't appear everytime... but unfortunately
    quite often (specially if I run a very CPU intensive
    application).

    I'll try to test the version you mentioned and I'll try to
    give you back a report as soon as possible.

     
  • Logged In: NO

    FYI: I've also noticed that the date drifts very badly
    between coLinux and windows... enough so that that after a
    few hours of run time, the date in coLinux is several days
    into the future. This is with the stable 0.61 release.

    -- ray {_at_} arachelian [period] net

     
  • Logged In: NO

    Something strange happenned: your 0.6.2-gpb worked well once
    but after a silly colinux reinstall it iis completely broken
    now. The colinux daemon just exits with the code '80162001'.
    I can only run the buggy 0710 snapshot.

    Thx anyhow for the help.

     
  • Logged In: YES
    user_id=30412

    Yes, Seems there have been some reports of daemon reportin
    g '80162001' It's almost always solved by running
    colinux-daemon --remove-driver colinux-daemon
    --install-driver, or by reducing a little bit the memory
    settings for colinux.

    But it's encouraging to hear that the latest source solves
    this problem. Thanks for reporting back.

    Feel free to e-mail me privately if you continue to have
    trouble with

    George

     
  • Logged In: NO

    No problem so far with the 20040910 snapshot. I'll try to do
    some more intensive testing.

    Thx.

     
  • Logged In: YES
    user_id=30412

    Closing this bug as it seems the gettimeofday fixes are solid.

     
    • status: open --> closed