SUMMARY: Weird SEGV from ufsdump

From: Dan Razzell (razzell@cs.ubc.ca)
Date: Tue Jul 16 1996 - 21:35:34 CDT


> Starting a few weeks ago, ufsdump started to abort when doing incremental
> dumps on one particular filesystem. Full Level 0 dumps are no problem.
>
> # /usr/sbin/ufsdump 1f /dev/rmt/1 /ai3
> DUMP: Writing 32 Kilobyte records
> DUMP: Date of this level 1 dump: Thu Jul 11 10:45:14 1996
> DUMP: Date of last level 0 dump: Tue Jul 09 09:51:29 1996
> DUMP: Dumping /dev/rdsk/c2t12d0s0 (lcifs2.cs.ubc.ca:/ai3) to /dev/rmt/1.
> DUMP: Mapping (Pass I) [regular files]
> DUMP: Mapping (Pass II) [directories]
> DUMP: SIGSEGV() ABORTING!
> Abort
>
> This is on a Sparc 20 running Solaris 2.4 Hardware 11/94 with the following
> patches:
>
> 101945-38 SunOS 5.4: jumbo patch for kernel
> 102580-08 Solstice DiskSuite 4.0: Jumbo patch

Turns out it was caused by a corrupt directory, not detected by "fsck -p".
A full fsck was sufficient to fix it, so happily I didn't have to rebuild
the filesystem.

To be precise, fsck had to be run twice before it stopped finding problems.
However, the first pass was enough to prevent ufsdump from falling over.

First prize goes to Agnes Burgess <ecl6ab@gps.leeds.ac.uk> for the best
diagnosis. Others who suspected filesystem corruption and suggested fsck
were:

  Mark S. Anderson <anderson@neon.mitre.org>
  Michael Sullivan <mike@trdlnk.com>

Thanks all!



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:11:05 CDT