#14 BUG at disk_stat.c line 517

closed-fixed
nobody
None
5
2005-05-11
2004-09-11
No

noflushd dies a few seconds after starting on FC2
(kernel 2.6.8-1.521):

[root@funghi root]# noflushd -d -v -n 1 /dev/hda
Kernel talks oldstyle dev
Probing for valid blkdev at /dev/hda
Detected oldstyle dev at /dev/hda
Kernel talks oldstyle dev
Added entry for /dev/hda (3, 0)
Added disk /dev/hda, current timeout 60, state 2
Controlling pdflush daemon
Kernel talks linux 2.5-style stat
pdflush stopped
Check interval 5
Allocated stat entry for (1, 0)
Allocated stat entry for (1, 1)
Allocated stat entry for (1, 2)
Allocated stat entry for (1, 3)
Allocated stat entry for (1, 4)
Allocated stat entry for (1, 5)
Allocated stat entry for (1, 6)
Allocated stat entry for (1, 7)
Allocated stat entry for (1, 8)
Allocated stat entry for (1, 9)
Allocated stat entry for (1, 10)
Allocated stat entry for (1, 11)
Allocated stat entry for (1, 12)
Allocated stat entry for (1, 13)
Allocated stat entry for (1, 14)
Allocated stat entry for (1, 15)
Allocated stat entry for (3, 0)
BUG at disk_stat.c line 517: No entry for (9,0)
pdflush restarted with interval 500.
Segmentation fault

Discussion

  • Darik Horn

    Darik Horn - 2004-10-01

    Logged In: YES
    user_id=560878

    I get this error when the device that noflushd is monitoring
    is part of a soft RAID array.

    Device (9,0) is /dev/md0.

     
  • Bruce Karsh

    Bruce Karsh - 2004-11-12

    Logged In: YES
    user_id=1126900

    I just ran into this bug too. The problem is that
    alloc_entry_new attempts to filter out meta devices with
    these lines:

    /* Filter out meta devices */
    if (major == MD_MAJOR || major == LVM_BLK_MAJOR)
    return NULL;

    So when it sees an md device, it returns NULL.

    The calling routine, update_io_25 then executes:

    BUG("No entry for (%d,%d)", major,
    minor);

    which never returns.

    A workaround is to comment out the BUG statement. This
    should be safe since the only way that alloc_entry_new can
    return NULL is when it has filtered out a meta device and
    those cases are not BUGs.

     
  • Daniel Kobras

    Daniel Kobras - 2005-05-11
    • status: open --> closed-fixed
     
  • Daniel Kobras

    Daniel Kobras - 2005-05-11

    Logged In: YES
    user_id=7832

    This bug is fixed in version 2.7.5 that I just released.

     

Log in to post a comment.

Get latest updates about Open Source Projects, Conferences and News.

Sign up for the SourceForge newsletter:

JavaScript is required for this form.





No, thanks