xfs
[Top] [All Lists]

[Bug 293] New: mount process crashes while replaying transaction log

To: xfs-master@xxxxxxxxxxx
Subject: [Bug 293] New: mount process crashes while replaying transaction log
From: bugzilla-daemon@xxxxxxxxxxx
Date: Fri, 28 Nov 2003 04:52:19 -0800
Sender: linux-xfs-bounce@xxxxxxxxxxx
http://oss.sgi.com/bugzilla/show_bug.cgi?id=293

           Summary: mount process crashes while replaying transaction log
           Product: Linux XFS
           Version: 1.3.x
          Platform: All
        OS/Version: Linux
            Status: NEW
          Severity: critical
          Priority: High
         Component: XFS kernel code
        AssignedTo: xfs-master@xxxxxxxxxxx
        ReportedBy: rlewczuk@xxxxxxxxx
                CC: rlewczuk@xxxxxxxxx


The following test always screws transaction log on my machine:
Filesystem mounted at /mnt/xfstest. Directory /mnt/xfstest/bonnie exists.
# bonnie++ -d /mnt/xfstest/bonnie -s 512 -n 512 -r 0 -n nobody &
# bonnie++ -d /mnt/xfstest/bonnie -s 512 -n 512 -r 0 -n nobody &
# bonnie++ -d /mnt/xfstest/bonnie -s 512 -n 512 -r 0 -n nobody &
3 processes are working.
Wait until bonnie++ finishes working on big files and creates enough small files
(directory size will exceed 2MB, many thousands files inside :-) ). Press reset
button on test machine. After reboot filesystem won't mount. Mount process
crashes with SEGV and device becomes locked somehow, so next mount will hang
forever. 

xfs_repair indicates that transaction log is broken, so it's only possible to
repair filesystem with discarding transaction log.

mkfs options: mkfs -t xfs -l size=32768b -f /dev/sdy

fstab entry:
/dev/sdy /mnt/xfstest xfs noauto,defaults,logbufs=8,grpquota   1   0

While crashing mount process kernel logs the following things:
Nov 26 17:37:56 file-comm-adm SGI XFS 1.3.1 with ACLs, debug enabled
Nov 26 17:37:56 file-comm-adm SGI XFS Quota Management subsystem
Nov 26 17:37:56 file-comm-adm XFS mounting filesystem sd(65,128)
Nov 26 17:37:56 file-comm-adm Starting XFS recovery on filesystem: sd(65,128)
(dev: 65/128)
Nov 26 17:37:59 file-comm-adm Unable to handle kernel NULL pointer dereference
at virtual address 00000058
Nov 26 17:37:59 file-comm-adm printing eip:
Nov 26 17:37:59 file-comm-adm f94a5d5e
Nov 26 17:37:59 file-comm-adm *pde = 00000000
Nov 26 17:37:59 file-comm-adm Oops: 0000
Nov 26 17:37:59 file-comm-adm CPU:    2
Nov 26 17:37:59 file-comm-adm EIP:    0010:[<f94a5d5e>]    Not tainted
Nov 26 17:37:59 file-comm-adm EFLAGS: 00010206
Nov 26 17:37:59 file-comm-adm eax: 00000000   ebx: 00000000   ecx: f69e0000  
edx: 0021f520
Nov 26 17:37:59 file-comm-adm esi: f69e0000   edi: 0021f520   ebp: 00000000  
esp: f69ddaa8
Nov 26 17:37:59 file-comm-adm ds: 0018   es: 0018   ss: 0018
Nov 26 17:37:59 file-comm-adm Process mount (pid: 1573, stackpage=f69dd000)
Nov 26 17:37:59 file-comm-adm Stack: 00000000 0000000c f948aefb f94e8286
f69e0000 00000000 0021f520 00000000
Nov 26 17:37:59 file-comm-adm 00000000 f69e0000 00000246 00000246 f69dc000
f948bd35 00000000 000001f0
Nov 26 17:37:59 file-comm-adm f948a261 f6be5924 f6b521c0 f6869e80 f6be5900
f948a2ba 00000000 f6869900
Nov 26 17:37:59 file-comm-adm Call Trace:    [<f948aefb>] [<f94e8286>]
[<f948bd35>] [<f948a261>] [<f948a2ba>]
Nov 26 17:37:59 file-comm-adm [<f948bfa2>] [<f948c116>] [<f948c32e>]
[<f948d84f>] [<f948db12>] [<f948dbb5>]
Nov 26 17:37:59 file-comm-adm [<c011a5d8>] [<f94e9c05>] [<f948ddb2>]
[<f94ddca0>] [<f9483c1d>] [<f9492ac6>]
Nov 26 17:37:59 file-comm-adm [<f94a7887>] [<f9491c4f>] [<f948e7b2>]
[<f949c6d8>] [<f94b309d>] [<f94b309d>]
Nov 26 17:37:59 file-comm-adm [<f94b2c7f>] [<c0141323>] [<f94fd42c>]
[<f94fd42c>] [<c0127447>] [<f94fd42c>]
Nov 26 17:37:59 file-comm-adm [<c014152c>] [<f94fd42c>] [<c01466b8>]
[<c0154736>] [<c01549eb>] [<c015484c>]
Nov 26 17:37:59 file-comm-adm [<c0154e14>] [<c0108c73>]
Nov 26 17:37:59 file-comm-adm
Nov 26 17:37:59 file-comm-adm Code: 8b 58 58 85 c0 8b 4c 24 1c 53 bb 0c 00 00 00
74 07 0f b7 98
Sorry for not providing symbol names here :) Need to make syslog work with
kernel symbols :) 

Machine: IBM xSeries 360 with 2 Xeons (HT, so it is seen by system as 4-way
machine) + 4GB RAM. It uses external storage array connected via two Fibre
Channel HBA cards (QLogic 2300, via FC switches, failover compiled in). 

# uname -a
Linux file-comm-adm 2.4.21-11fo1-SMP #1 SMP ¶ro lis 26 17:02:35 CET 2003 i686
unknown


Don't know how to set up rest of those bugzilla fileds, so left these
empty/default :)



------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.


<Prev in Thread] Current Thread [Next in Thread>
  • [Bug 293] New: mount process crashes while replaying transaction log, bugzilla-daemon <=