Date:   Mon, 11 Sep 2000 08:05:05 -0700 (PDT)
Message-Id: <200009111505.IAA06738@feature.engr.sgi.com>
Reply-To: sgi.bugs.xfs@fido.engr.sgi.com
From:   pv@fddi-odin.corp.sgi.com (jones@tacc.cc.utexas.edu.sgi.com)
Subject: ADD 801241 - xfsdump can cause filesystem corruption
To:     dxm@cthulhu.engr.sgi.com
Cc:     linux-xfs@oss.sgi.com
Sender: owner-linux-xfs@oss.sgi.com
Precedence: bulk

 Submitter : ivanr                     Status : open                        
 Assigned Engineer : dxm               Priority : 1                         
*Modified Date : 09/11/00             *Modified User : jones                
*Modified User Domain : tacc.cc.utexas.edu
*Description :
QA for xfsdump/xfsrestore revealed that it is possible for
xfsdump to cause filesystem corruption.  The test involves
creating a filesystem, creating a whole buncha files, xfsdumping
them to a file somewhere, then removing all the files.  xfs_check
reports lots of inodes with zero link counts, etc.  It's probably
due to xfsdump using bulkstat on the entire fs.

The following is a script which will reliably (at least on
bruce.melbourne) produce the error.  Run it in cmd/xfs/stress -
it needs the src/fill program, and make sure you fix the

.....


==========================
ADDITIONAL INFORMATION (ADD)
From: "william l. jones" <jones@tacc.cc.utexas.edu>
Date: Sep 11 2000 08:05:04AM
[pvnews version: 1.71]
==========================
At 11:15 PM 9/10/00 -0700, you wrote:
>View Incident: 
>http://co-op.engr.sgi.com/BugWorks/code/bwxquery.cgi?search=Search&wlong=1& 
>view_type=Bug&wi=801241
>
>  Status : open                         Priority : 1
>  Assigned Engineer : dxm               Submitter : ivanr
>*Modified User : dxm                  *Modified User Domain : engr
>*Description :
>QA for xfsdump/xfsrestore revealed that it is possible for
>xfsdump to cause filesystem corruption.  The test involves
>creating a filesystem, creating a whole buncha files, xfsdumping
>them to a file somewhere, then removing all the files.  xfs_check
>reports lots of inodes with zero link counts, etc.  It's probably
>due to xfsdump using bulkstat on the entire fs.
>
>The following is a script which will reliably (at least on
>bruce.melbourne) produce the error.  Run it in cmd/xfs/stress -
>it needs the src/fill program, and make sure you fix the
>
>.....
>
>
>==========================
>ADDITIONAL INFORMATION (ADD)
>From: dxm@engr (BugWorks)
>Date: Sep 10 2000 11:15:42PM
>==========================
>
>xfs_open_by_handle is causing some kind of reference leak
>which means deleted files stay in the agi_unlinked list
>when the FS is unmounted instead of being passed to
>xfs_iunlink_remove when their count hits zero
>(ie before unmount).
>
>I've added qa 034 which isolates this bug...


You should also add bulk_stat_single to the list of suspects.

Bill Jones